Gesellschaft für Informatik Fachgruppe Angewandte Kryptographie. Book of Abstracts Krypto-Tage

Größe: px
Ab Seite anzeigen:

Download "Gesellschaft für Informatik Fachgruppe Angewandte Kryptographie. Book of Abstracts Krypto-Tage 2004 2012"

Transkript

1 Gesellschaft für Informatik Fachgruppe Angewandte Kryptographie Book of Abstracts Krypto-Tage Herausgeber: Stefan Katzenbeisser, TU Darmstadt Dezember 2012

2 Der Krypto-Tag ist eine zentrale Aktivität der Fachgruppe Angewandte Kryptologie der Gesellschaft für Informatik und soll insbesondere Nachwuchswissenschaftlern auf dem Gebiet der Kryptographie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert werden oder einen Schwerpunkt der eigenen Masterarbeit oder Dissertation bilden. Dieser Band enthält die Zusammenfassungen aller Beiträge, die zwischen 2004 und 2012 auf den Krypto-Tagen sowie auf den beiden Krypto-Wochenenden in 2006 und 2008 präsentiert wurden: 1. Krypto-Tag am 1. Dezember 2004 Universität Mannheim, Lehrstuhl für Theoretische Informatik Organisation: Stefan Lucks und Christopher Wolf 2. Krypto-Tag am 31. März 2005 Universität Ulm, Abteilung für Theoretische Informatik Organisation: Wolfgang Lindner und Christopher Wolf 3. Krypto-Tag am 15. September 2005 Technische Universität Darmstadt, Theoretische Informatik Organisation: Ralf-Philipp Weinmann 4. Krypto-Tag am 11. Mai 2006 Ruhr Universität Bochum, Horst-Görtz Institut Organisation: Ulrich Greveler 1. Krypto-Wochenende am 1. und 2. Juli 2006 Tagungszentrum Kloster Bronnbach der Universität Mannheim Organisation: Frederik Armknecht und Dirk Stegemann 5. Krypto-Tag am 11. September 2006 Universität Kassel, Arbeitsgruppe Theoretische Informatik Organisation: Heiko Stamer 6. Krypto-Tag am 19. Februar 2007 Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG Organisation: Michael Backes und Ammar Alkassar 7. Krypto-Tag am 9. November 2007 Bonn-Aachen International Center for Information Technology Organisation: Michael Nüsken und Daniel Loebenberger 8. Krypto-Tag am 11. April 2008 Diskrete Mathematik der Universität Tübingen Organisation: Michael Beiter, Claudia Kaesser und Anja Korsten 2. Krypto-Wochenende Juli 2008 Tagungshotel Am Burgholz in Tabarz (Thüringen) Organisation: Stefan Lucks

3 9. Krypto-Tag am 10. November 2008 Institut für Internet-Sicherheit der Fachhochschule Gelsenkirchen Organisation: Markus Linnemann 10. Krypto-Tag am 20. März 2009 Institut für Mathematik der Technischen Universität Berlin Organisation: Florian Heß 11. Krypto-Tag am 30. November 2009 Lehrstuhl für Informationssicherheit und Kryptographie der Universität Trier Organisation: Ralf Küsters 12. Krypto-Tag am 9. April 2010 Institut für Kryptographie und Sicherheit des KIT Organisation: Willi Geiselmann und Jörn Müller-Quade 13. Krypto-Tag am 5. November 2010 CASED, Security Engineering Group der Technischen Universität Darmstadt Organisation: Andreas Peter 14. Krypto-Tag am 21. und 22. März 2011 Ruhr-Universität Bochum, Horst Görtz-Institut für IT-Sicherheit Organisation: Christopher Wolf 15. Krypto-Tag am 1. und 2. Dezember 2011 Carl von Ossietzky Universität Oldenburg Organisation: Florian Hess 16. Krypto-Tag am 15. Juni 2012 NXP Semiconductors, Hamburg Organisation: Frederik Armknecht 17. Krypto-Tag am 7. Dezember 2012 NEC Research, Heidelberg Organisation: Frederik Armknecht

4 Universität Mannheim, Reihe Informatik, TR Katholieke Universiteit Leuven, Technical Report ESAT-COSIC, 2004-CW-1 1. Kryptotag Workshop über Kryptographie Universität Mannheim Stefan Lucks Universität Mannheim, Deutschland Christopher Wolf KU Leuven, Belgien 1. Dezember

5 Beiträge 1. Sebastian Faust und Stefan Lucks: Neue Wege des anonymen Datenaustausches in Peer-2-Peer-Netzwerken 2. Stefan Lucks: Ein neues Designprinzip für kryptographische Hashfunctionen 3. Jörg Lässig, Stefanie Thiem and Matthias Baumgart: The Deterministic AKS Primality Method in the PRAM-Model and a Parallel Implementation on a High Performance Cluster 4. Christof Paar and Jan Pelzl and Thomas Wollinger: Hyperelliptic Cryptosystems in Practice 5. Heiko Stamer: Oleshchuk s Public Key Cryptosystem 6. Christopher Wolf: Äquivalente Private Schlüssel in Systemen mit Multivariaten Quadratischen Öffentlichen Polynomgleichungen 7. Frederik Armknecht: Algebraische Angriffe auf LFSR-basierte Stromchiffren 8. Magnus Daum: Solving Systems of Equations with incompatible operations 9. Dirk Stegemann: BDD-basierte Kryptanalyse von Flusschiffren am Beispiel des A5/1 Schlüsselstromgenerators 10. Philipp A. Baer: Group Authentication and Encryption in Distributed Environments 11. André Schaumburg: Authentification within Tree Parity Machine Rekeying 12. Elena Ivanova Andreeva : Overview of Authenticated Encryption Modes of Operation 13. André Adelsbach, Markus Rohe and Ahmad-Reza Sadeghi: Improving The Security Of Watermarking Schemes With Cryptographic Techniques 14. Stefan Katzenbeisser: Cryptographic Watermarking 15. Tobias Straub: Bridging the Usability Gap of PKI 2

6 Neue Wege des anonymen Datenaustausches in Peer-2-Peer-Netzwerken Sebastian Faust und Stefan Lucks Universität Mannheim Deutschland P2P-Netzwerke erfreuen sich zunehmender Beliebtheit und sind bisher besonders im Bereich des direkten Austausches von Dateien und Dokumenten, dem so genannten File-Sharing, weit verbreitet. Aufgrund der hohen Flexibilität durch die Verwendung einer dezentralen Netzwerk-Infrastruktur, lassen sich jedoch auch viele weitere Probleme mit Hilfe von P2P-Netzwerken lösen, beispielsweise aus dem Gebiet des Distributed Computing oder dem Instant Messaging. Diese Arbeit beschäftigt sich mit dem sicheren Datenaustausch in Peerto-Peer-Netzwerken. Hierbei werden wir uns auf die Anonymität der Teilnehmer konzentrieren. Wir werden die Sicherheit der Teilnehmer in einem formalen Modell spezifizieren und, darauf aufbauend, zwei Bedrohungsmodelle entwickeln. Wir werden die Begriffe der ( absoluten ) Sicherheit und der ( t-sicherheit ) einführen. Bei ersterer kontrolliert ein aktiver, dynamischer, arbeitsamer Angreifer mit Ausnahme einer Partei das Verhalten aller am Datenaustausch teilnehmenden Benutzer. Wir weisen nach, dass diese größtmögliche ( absolute ) Sicherheit in einem Peer-to-Peer Netzwerk nicht erreichbar ist. Im Sinne eines etwas abgeschwächten Bedrohungsmodells, der ( t-sicherheit ), bei dem maximal t Teilnehmer vom Angreifer korrumpiert werden dürfen, kann jedoch die Sicherheit aller ehrlichen Teilnehmer gewährleistet werden. Wir geben ein Protokoll an, das die t-sicherheit garantiert. Literatur [ASF01] Dmitri Asonov, Markus Schaal and Johann-Christoph Freytag. Absolute Privacy in Voting. LNCS ISC2001, [DJN03] I. Damgard, M. Jurik and J. Nielsen. A generalization of paillier s public-key system with applications to electronic voting, [Ste98] Julien P. Stern. A New Efficient All-Or-Nothing Disclosure of Secrets Protocol. LNCS vol / 1998 ASIACRYPT 98,

7 Ein neues Designprinzip für kryptographische Hashfunctionen Stefan Lucks Universität Mannheim Die Grundstruktur aller aktuell in der Praxis verwendeten Hashfunktionen stammt von Merkle [3] und Damgård [1]. Eine Kompressionsfunktion fester Eingabelänge wird so lange iteriert, bis der gesamte Input für die Hashfunktion verarbeitet wurde. Um das Sicherheitsziel Kollisionsresistenz für die Hashfunktion zu erreichen, genügt schon die Kollisionsresistenz der Kompressionsfunktion. Verfehlt die Kompressionsfunktion dieses Ziel allerdings, sind auf die Hashfunktion sogar verheerende K-fache Kollisionsangriffe möglich. Dies zeigte Joux [2] auf der Crypto Der Vortrag gibt eine Übersicht über verschiedene Sicherheitsziele für Hashfunktionen, das Designprinzip von Merkle und Damgård und den Angriff von Joux. Des weiteren wird eine Verbesserung des Konstruktionsprinzips von Merkle und Damgård eingeführt, die nachweisbar Sicherheit gegen Angriffe wie den von Joux gewährleistet. Literatur [1] I. Damgård. A design principle for hash functions. Crypto 89. [2] A. Joux. Multicollisions in iterated hash functions, application to cascaded constructions. Crypto 04. [3] R. Merkle. One-way hash functions and DES. Crypto 89. 4

8 The Deterministic AKS Primality Method in the PRAM-Model and a Parallel Implementation on a High Performance Cluster Jörg Lässig, Stefanie Thiem and Matthias Baumgart Chemnitz University of Technology Chemnitz Germany In August 2002 the three Indian researchers Manindra Agrawal, Neeraj Kayal & Nitin Saxena at the Indian Institute of Technology in Kanpur published the manuscript PRIMES is in P [Ag02] and presented a deterministic, polynomial time algorithm to determine whether a given integer is prime or composite. Later on a series of variations of the original algorithm, has been published (e.g. [Br02]), the so-called AKS-class of algorithms. The improvements over the original AKS breakthrough are orders of magnitude, referring to Daniel J. Bernstein, one of the major protagonists [Be03]. The objective of these studies is to present a completely deterministic AKS instance with conjectured running time T (n) = Õ(log4 n) in the PRAM Model, including all available speedups. Additionally, we present its parallel implementation and some experimental results besides a practical comparison to standard methods. Key Words and Phrases: PRIMES, AKS-class, PRAM-model, Deterministic primality method References [Ag02] [Br02] M. Agrawal and N. Kayal and N. Saxena. PRIMES is in P (revised version). Preprint, Indian Institute of Technology Kanpur. December P. Berrizbeitia. Sharpening Primes is in P for a Large Family of Numbers. Preprint, Universidad Simón Bolívar. November [Be03] D. J. Bernstein. Proving Primality after Agrawal-Kayal-Saxena. Preprint, University of Illinois at Chicago. January

9 Hyperelliptic Cryptosystems in Practice Christof Paar and Jan Pelzl and Thomas Wollinger Communication Security Group COSY Ruhr Universität Bochum Germany The Hyperelliptic curve cryptosystem is one of the emerging cryptographic primitives of the last years. This system offers the same security as established public-key cryptosystems, such as those based on RSA or elliptic curves, with much shorter operand length. However, until recently the common belief in industry and in the research community was that hyperelliptic curves are out of scope for any practical application. We were able to show the practical use of hyperelliptic curve cryptosystems (HECC) by narrowing the performance gap between elliptic curve (EC) and hyperelliptic curve cryptosystems. The complexity of the group operation for small genus hyperelliptic curves was reduced and efficient algorithms have been proposed [PWGP03, PWP03]. We developed a new metric to compare different cryptographic primitives based on the atomic operations of a processor and our theoretical comparison between elliptic curve and hyperelliptic curve cryptosystems, as well as our software and hardware implementations show that the performance of both cryptographic primitives are in the same range [P02]. Surprisingly, the hyperelliptic curve cryptosystems even outperform elliptic curves using certain curve parameters. We implemented these cryptosystems on general purpose processor and on a variety of different embedded processors, and build even a prototype implementation of a hyperelliptic curve coprocessor on FPGAs [WPWPSK04, Wol04]. References [Wol04] [PWGP03] Thomas Wollinger. Software and Hardware Implementation of Hyperelliptic Curve Cryptosystems Ph.D., Ruhr-University Bochum, Bochum, Germany July Jan Pelzl and Thomas Wollinger and Jorge Guajardo and Christof Paar. Hyperelliptic Curve Cryptosystems: Closing the Performance Gap to Elliptic Curves Workshop on Cryptographic Hardware and Embedded Systems - CHES 2003 September [PWP03] Jan Pelzl and Thomas Wollinger and Christof Paar. Low Cost Security: Explicit Formulae for Genus-4 Hyperelliptic Curves Tenth Annual Workshop on Selected Areas in Cryptography - SAC 2003 August [P02] Jan Pelzl. Hyperelliptic Cryptosystems on Embedded Microprocessors Diplomarbeit, Ruhr- Universitt Bochum September [WPWPSK04] Thomas Wollinger and Jan Pelzl and Volker Wittelsberger and Christof Paar and Gokay Saldamli and Cetin Koc. Elliptic and Hyperelliptic Curves on Embedded up Special issue on Embedded Systems and Security of the ACM Transactions in Embedded Computing Systems (TECS) 6

10 Oleshchuk s Public Key Cryptosystem Heiko Stamer Universität Kassel, Fachbereich Mathematik/Informatik Heinrich-Plett-Straße 40, D Kassel, Germany stamer@theory.informatik.uni-kassel.de Vor etwa zehn Jahren stellte Vladimir A. Oleshchuk [Olk95] ein neues asymmetrisches Verschlüsselungsverfahren vor, dessen Sicherheit auf der Unentscheidbarkeit des Wortproblems für endliche Wortersetzungssysteme basiert. Zur Konstruktion der Falltürfunktion verwendet er dabei eingeschränkte Systeme mit Church-Rosser Eigenschaft, für die das Wortproblem in linearer Zeit lösbar ist. Diese grundsätzliche Strategie wurde später auch auf Baumersetzungssysteme [Sa03] übertragen. Ein motivierender Entwicklungsaspekt solcher Verfahren ist sicherlich die beweisbare Schwierigkeit des zugrundeliegenden Problems. Leider ist damit noch keinerlei Aussage über die praktische Sicherheit getroffen, denn alle bekannten Ansätze leiden unter anderen Verwundbarkeiten. Insbesondere spielt hier die Problematik schwacher Schlüssel eine große Rolle, d. h. die bekannte Lücke zwischen average- und worst-case Komplexität von Instanzen tritt hier verstärkt zutage. Hinsichtlich Oleshchuks Verfahren hat sich beispielsweise herausgestellt, daß viele der möglichen Schlüssel (Wortersetzungssysteme) durch Vervollständigung [StO04] angreifbar sind. Der Vortrag führt anfangs die notwendigen theoretischen Grundlagen von Wortersetzungssystemen ein. Im Hauptteil wird das oben erwähnte Kryptosystem vorgestellt. Schließlich betrachten wir einige Ideen zur Kryptoanalyse und untersuchen praktische Implementierungsfragen. Literatur [Olk95] Vladimir A. Oleshchuk: On Public-Key Cryptosystem Based on Church-Rosser String-Rewriting Systems, Proc. COCOON 95, Lecture Notes in Computer Science 959, pp , 1995 [Sa03] S.C. Samuel, D.G. Thomas, P.J. Abisha, K.G. Subramanian: Tree Replacement and Public Key Cryptosystem, INDOCRYPT 2002, Lecture Notes in Computer Science 2551, pp , 2002 [StO04] Heiko Stamer, Friedrich Otto: On Oleshchuk s Public Key Cryptosystem, Cryptology eprint Archive, Report 2004/220, 2004, 7

11 Äquivalente Private Schlüssel in Systemen mit Multivariaten Quadratischen Öffentlichen Polynomgleichungen Christopher Wolf ESAT-COSIC, K.U. Leuven, Belgien oder Seit dem Artikel von Matsumoto und Imai [MI88] über asymmetrische Systeme mit multivariaten Polynomgleichungen als öffentlichem Schlüssel wurden eine Reihe solcher Systeme vorgeschlagen, z.b. [Pat96] und [KPG99]. Die Variante HFE- bildet die Grundlage einer verbesserten Version des Signaturverfahrens Quartz während C im Signaturschema Sflash v3 Verwendung findet. In diesem Vortrag beleuchten wir die Struktur des privaten Schlüsselraums in diesen Systemen. Insbesondere zeigen wir, dass sowohl in HFE- als auch in C für jeden privaten Schlüssel eine große Anzahl äquivalenter Schlüssel existiert, die alle auf den selben öffentlichen Schlüssel abgebildet werden. Des weiteren stellen wir einen effizienten Algorithmus vor, mit dessen Hilfe eine Normalform für private Schlüssel in diesen Systemen berechnet werden kann. Literatur [KPG99] Aviad Kipnis, Jacques Patarin, and Louis Goubin. Unbalanced oil and vinegar signature schemes. In EUROCRYPT 1999, volume 1592 of LNCS, pages [MI88] Tsutomu Matsumoto and Hideki Imai. Public quadratic polynomial-tuples for efficient signature verification and messageencryption. In EUROCRYPT 1988, volume 330 of LNCS, pages [Pat96] Jacques Patarin. Hidden Field Equations (HFE) and Isomorphisms of Polynomials (IP): two new families of asymmetric algorithms. In EUROCRYPT 1996, volume 1070 of LNCS, pages Extended Version: 8

12 Algebraische Angriffe auf LFSR-basierte Stromchiffren Frederik Armknecht Universität Mannheim Durch ihre hohen Effizienz haben LFSR 1 -basierte Stromchiffren ihre Anwendung in der Praxis gefunden (bspw. im Bluetooth- oder GSM-Standard). LFSRs sind endliche Automaten, die auf einfache Art und Weise Bitstrme beliebiger Lnge produzieren knnen, welche sehr gute statistische Eigenschaften besitzt. Mit der Zeit entwickelten Kryptographen Techniken, um die Resistenz solcher Stromchiffren gegen eine Reihe von Angriffen (bspw. Fast Correlation Attacks) zu erhhen. Vor knapp zwei Jahren wurde eine neue Angriffsmethode vorgestellt, die sogenannten algebraischen Angriffe. Die Idee ist es, den geheimen Schlssel durch Lsen eines (ggf. nichtlinearen) Gleichungssystems zu rekonstruieren. Whrend das Lsen von nichtlinearen Gleichungssystemen im Allgemeinen schwierig ist, ist es aus mancherlei Grnden in diesem speziellen Fall einfacher. Es stellte sich heraus, dass die algebraischen Angriffe in vielen Fllen die schnellsten (theoretischen) Angriffe darstellen. Der Vortrag gibt eine Einfhrung in die obige Thematik und bietet einen Einblick in den aktuellen Stand der Forschung. Literatur [1] Frederik Armknecht, Matthias Krause: Algebraic attacks on Combiners with Memory, Proceedings of Crypto 2003, LNCS 2729, pp , Springer, [2] Nicolas Courtois, Willi Meier: Algebraic attacks on Stream Ciphers with Linear Feedback, Proceedings of Eurocrypt 2003, LNCS 2656, pp , Springer, An extended version is available at [3] Nicolas Courtois: Fast Algebraic Attacks on Stream Ciphers with Linear Feedback, Proceedings of Crypto 2003, LNCS 2729, pp , Springer, LFSR = Linear Feedback Shift Register 9

13 Solving Systems of Equations with incompatible operations Magnus Daum CITS Research Group, Ruhr University Bochum In many cryptographic algorithms, which are aimed at being efficiently implementable rather than having some formal proof of security (e.g. in many dedicated hash functions or block ciphers), a mixture of very different kinds of operations is used. These operations include GF(2)-linear operations, additions modulo 2 n, bitwise applied Boolean functions and bit shifts and rotations. As these operations are not very compatible from a mathematical point of view, it is hard to analyse these structures theoretically and you need sophisticated algorithms to solve equations in which some different of these operations are involved. In his attacks on various hash functions (see for example [Do97]) Dobbertin had to solve large systems of equations of this kind. We analyse the algorithms used by Dobbertin and describe improvements which lead to directed graphs, which represent the sets of solutions of such equations quite efficiently. These graphs are related very closely to binary decision diagrams (see [We03]). Thus from the theory of decision diagrams many algorithms can be adopted. For example, it is possible to efficiently compute the number of solutions or to combine two such graphs to compute the intersection of two sets of solutions. T-functions, as proposed by Klimov and Shamir (see [Kl04]), are another topic for which these algorithms could be of some interest, because due to their defining property, equations which only include T-functions should be solvable quite efficiently with such algorithms. References [Do97] H. Dobbertin (1997). RIPEMD with two-round compress function is not collision-free. Journal of Cryptology 10, pp [We03] I. Wegener (2003). Branching Programs and Binary Decision Diagrams Theory and Applications. SIAM. [Kl04] A. Klimov (2004). Applications of T-functions in Cryptography. PhD Thesis, Weizmann Institute of Science. (available from ask/) 10

14 BDD-basierte Kryptanalyse von Flusschiffren am Beispiel des A5/1 Schlüsselstromgenerators Dirk Stegemann Universität Mannheim Viele praktisch eingesetzte Flusschiffren basieren auf einer kleinen Zahl linear rückgekoppelter Schieberegister (Linear Feedback Shift Registers, kurz LFSRs), deren Ausgabebitströme mit Hilfe einer nichtlinearen Kompressionsfunktion zu einem Schlüsselstrom y {0, 1} verdichtet werden. [Kra02] identifiziert die best case Kompressionsrate γ und die durchschnittliche Informationsrate α der Kompressionsfunktion als entscheidende Sicherheitsparameter und beschreibt einen Angriff auf LFSR-basierte Flusschiffren, der den geheimen Initialzustand x {0, 1} n in einer Laufzeit von n O(1) 2 1 α 1+α n aus den ersten γα 1 n aufeinanderfolgenden Bits des Schlüsselstroms y rekonstruiert. Die Grundlage dieses Angriffs bildet die Repräsentation der Zwischenergebnisse in Binären Entscheidungsdiagrammen (Binary Decision Diagrams, kurz BDDs), die bisher vor allem zur formalen Schaltkreisverifikation im VLSI-Design eingesetzt wurden. Von [Sch02] und [Ste04] durchgeführte Experimente mit verschiedenen Flusschiffren scheinen die theoretischen Resultate zu bestätigen, zeigen jedoch gleichzeitig die durch hohen Speicherbedarf bedingten Grenzen der praktischen Durchführbarkeit von BDD-basierten Angriffen auf. Am Beispiel des A5/1 Schlüsselstromgenerators [BGW99] aus dem GSM-Standard soll die BDD-basierte Kryptanalyse vorgestellt und auf Implementationsaspekte und experimentelle Resultate eingegangen werden. Literatur [Kra02] M. Krause. BDD-based cryptanalysis of keystream generators. In EUROCRYPT 2002, Band 2332 der Reihe Lecture Notes in Computer Science, Seiten Springer Verlag, Heidelberg, [BGW99] M. Briceno, I. Goldberg, and D. Wagner. A pedagogical implementation of A5/1, May [Sch02] F. Schleer. Einsatz von OBDDs zur Kryptanalyse von Flusschiffren. Diplomarbeit, Universität Mannheim, [Ste04] D. Stegemann. Fbdd-basierte Kryptanalyse des A5/1 Schlüsselstromgenerators. Diplomarbeit, Universität Mannheim,

15 Group Authentication and Encryption in Distributed Environments Philipp A. Baer University of Ulm Department of Theoretical Computer Science Germany Security is not always considered an important issue for groups of distributed systems. Message authentication and encryption are often disregarded, sometimes just because of missing implementations. Nevertheless, especially in the context of communication, control and monitoring, security is an extremely important issue. This paper discusses techniques that address some of the basic security requirements for unreliable group communication scenarios. It combines existing security technologies (DH, GDH [STW96], DSA, AES) and communication protocols/schemes (IPv6, Multicast) for group collaboration scenarios in unreliable environments. Message authentication and message stream encryption for groups, to only mention the most important ones, are considered exemplary. The architecture and its communication primitives are tailored to the needs of unreliable environments. This is mostly due to its intended field of application: groups of autonomous mobile systems. The architecture is designed for a wide variety of systems and open in the sense of extensibility. For the proposed techniques, i.e. key agreement, authentication and encryption, very simple yet extensible protocols are used. Because of the many unsolved problems in the area of secure ad-hoc communication, and due to the wide variety of involved scientific subjects, only a superficial solution can be presented. References [Bae04] Philipp A. Baer. Group Authentication and Encryption in Distributed Environments. University of Ulm, July [STW96] Michael Steiner, Gene Tsudik, and Michael Waidner. Diffie- Hellman key distribution extended to group communication. In ACM Conference on Computer and Communications Security, pages 31 37, March

16 Authentification within Tree Parity Machine Rekeying André Schaumburg Hamburg University of Science and Technology, Distributed Systems Schwarzenbergstraße 95, Hamburg - Germany Interaction of Tree Parity Machines (TPMs) has been discussed as an alternative secure key exchange concept and attacks have been proposed [KKK02]. Authentication is at least as important as a secure exchange of keys. Adding an authentication e.g. via hashing is straightforward but outside the concept named Neural Cryptography. The here presented is a consequent formulation of an implicit Zero-Knowledge authentication from within the key exchange concept and another alternative, integrating an explicit Zero- Knowledge authentication into the already interactive protocol. A Man-In- The-Middle attack and even all currently known attacks can so be averted. This in turn allows to securely exploit the trajectory in key space along with rapid key exchange and an efficient increase of key length. Another benefit of the here presented authentication method is that all currently known findings concerning Neural Cryptography are untouched and still valid - even with the extension of authentication. Further on there is no need to reimplement the interface, it only gets extended by an authentication control unit. The general trade-off in applied cryptography between available resources and the required level of security also applies using the TPM principle. In many practical embedded security solutions it is often admissive to provide a system safe enough for the particular application, and given certain attack scenarios. The TPM principle extended with the proposed authentication is very attractive for such embedded applications due to its hardware-friendly basic operations, particularly not operating on large numbers. References [KKK02] Kanter, I., Kinzel, W. and Kanter, E.: Secure exchange of information by synchronization of neural networks Europhysics Letters 57 (2002), pp [VS04] Volkmer, M. and Schaumburg, A.: Authenticated tree parity machine key exchange Preprint cs.cr/ (2004), submitted to Europhysics Letters 13

17 Overview of Authenticated Encryption Modes of Operation Elena Ivanova Andreeva Saarland University Traditional block cipher modes of operation, namely CBC, CFB, OFB and CTR, provide encryption with achieving the confidentiality goal without any integrity guarantees. On the other hand, there exist authentication modes that are custom-made to ensure integrity. However, they do not provide a secure encryption. A conventional way to satisfy both security properties, confidentiality and integrity, is to make two separate passes on the data. One encryption pass for the encrypting of the data block cipher-wise, and a second authentication pass for checking the data integrity. Recently, new unconventional integrity-aware modes of operation for block ciphers have been proposed. They provide confidentiality and integrity by combining authentication modes with the traditional block ciphers making only a single pass on the data. These modes are called authenticated encryption modes. In our presentation we give an overview on some of these newly proposed authenticated encryption modes, like IACBC, IAPM[JU01], XCBC[GD01] and OCB[RBBK01], their properties and advantages for a future use. References [GD01] V. Gligor, P. Donescu. Fast encryption and authentication: XCBC Encryption and XECB Authentication modes. Proceedings of the Fast Software Encryption Workshop - FSE 01. Springer-Verlag, October [JU01] C. Jutla. Encryption Modes with Almost Free Message Integrity. Advances in Cryptology-EUROCRYPT [RBBK01] P. Rogaway, M. Bellare, J. Black and T. Krovetz. OCB: A Block- Cipher Mode of Operation for Efficient Authenticated Encryption ACM Transactions on Information and System Security (TISSEC), vol. 6, no. 3, pp , August

18 Improving The Security Of Watermarking Schemes With Cryptographic Techniques André Adelsbach, Markus Rohe and Ahmad-Reza Sadeghi Universität des Saarlandes Ruhr-Universität Bochum Saarbrücken Standard information hiding schemes, such as watermarking schemes suffer from a major problem: They require to reveal security critical information to potentially untrusted parties, when proving the presence of a watermark to these parties. Zero-knowledge watermark detection is a promising means to overcome this problem: The embedded information is concealed from the verifying party while the proving party is committed to it. A prover and a verifier perform a zeroknowledge detection protocol after which the verifier is convinced that the committed information is imperceptibly present in the given digital content without gaining any new knowledge on the security critical information. However, concealing the embedded information prevents the verifying party from performing additional checks on this data, e.g., on its probability distribution, which may be required for certain applications. We overcome this limitation by proposing concrete and practical protocols, which pursue two promising strategies: The first one is to prove in zero-knowledge that a concealed information suffices a certain predicate, whereas the second strategy is to interactively and verifiably generate committed information that suffices the desired predicate, e.g., matches a certain probability distribution. Moreover, we have designed an efficient implementation of a zero-knowledge watermark detection protocol to demonstrate its applicability in practice. 15

19 Cryptographic Watermarking Stefan Katzenbeisser Institut für Informatik, Technische Universität München D Garching bei München The rapid growth of the Internet as a distribution medium for digital goods increased the risk of copyright infringements. From an economic point of view, this risk makes the commercialization of digital works difficult, if not impossible. Therefore, the need for technical copyright protection solutions has increased steadily over the last years. Robust digital watermarking became a promising technology in the context of copyright protection and was proposed as a central building block in various e-commerce protocols (such as dispute-resolving, copy protection and traitor tracing schemes or DRM applications). Traditionally, the design of watermarking schemes was seen as a signal-processing problem and concentrated on issues such as the imperceptibility of the watermark or its resistance against unauthorized removal. However, when a watermark is to be used in an e-commerce system, its properties may become critical to the security of the overall scheme. It is therefore necessary to gain a thorough and mathematically precise understanding of the essential security properties of watermarks. In this talk, I review recent results [1, 2] that establish the security of two cryptographic protocols that employ watermarking operations as basic primitives. The first protocol can be used in dispute-resolving schemes in order to assure their resistance against an important class of attacks. The second protocol allows to detect forgeries in image files or video streams by embedding a watermark carrying a cryptographic signature. Both constructions are provably secure under standard cryptographic assumptions. References [1] A. Adelsbach, S. Katzenbeisser, H. Veith, Watermarking Schemes Provably Secure Against Copy and Ambiguity Attacks, in ACM Workshop on Digital Rights Management (DRM 2003), Proceedings, Washington DC, 2003, pp [2] J. Dittmann, S. Katzenbeisser, C. Schallhart, H. Veith, Provably Secure Authentication of Digital Media Through Invertible Watermarks, to appear as IACR eprint report,

20 Bridging the Usability Gap of PKI Tobias Straub Computer Science Department Technische Universitaet Darmstadt From a user s viewpoint, security in general and PKI (public key infrastructure) in particular are complex matters. According to a recent study [kes04], the majority of security incidents still can be traced back to user errors or carelessness, although users troubles with PKI-enabled applications are known since the study of Whitten and Tygar investigating the handling of secure [WT99]. Unfortunately, due to the nature of PKI, user interaction cannot be avoided entirely, e.g. when an unknown certificate has to be imported as a trust anchor. In my work, I am proposing several measures to face the usability challenges of PKI. One of them is a generic framework to comprehensively evaluate usability and utility of PKI-enabled applications [SB04]. Apart from being a tool to detect usability problems and assess applications, it may as well serve as a requirements specification for application designers. Another idea is to delegate complex and security-critical tasks to skilled personnel or a service provider. For instance, the protocol described in [Str04] allows an organization to outsource the task of maintaining a PKI in a way that it retains full control and does not have to trust the service provider. References [kes04] Lagebericht zur Informationssicherheit. In: kes 4/2004 and 5/2004. [Str04] T. Straub. A Method for Strengthening Certificate Enrollment. WartaCrypt, [SB04] T. Straub and H. Baier. A Framework for Evaluating the Usability and the Usefulness of PKI-enabled Applications. European PKI- Workshop, (Springer LNCS 3093) [WT99] A. Whitten, J.D. Tygar. Why Johnny Can t Encrpyt: A Usability Evaluation of PGP 5.0. USENIX Security Symposium,

21 CURANDO 2. Krypto-Tag Workshop über Kryptographie Universität Ulm Wolfgang Lindner (Hg.) Universität Ulm Christopher Wolf (Hg.) K.U.Leuven 31. März 2005 UNIVERSITÄT ULM DOCENDO SCIENDO Ulmer Informatik-Berichte Nr K.U.Leuven, Technical Report ESAT-COSIC 2005-CW-1 1

22 Inhaltsverzeichnis Multicast-Security Simulation logischer Schlüsselbäume unter Berücksichtigung unterschiedlichen Benutzerverhaltens Alexander Heese, Luigi Lo Iacono, Christoph Ruland Cryptographic Watermarking Stefan Katzenbeisser Mental Poker in practice: An extended implementation of Schindelhauer s Toolbox for Mental Card Games Heiko Stamer Mystery-Twister Hans Dobbertin, Magnus Daum, Patrick Felke, Gregor Leander 6 Erweiterte visuelle Kryptographie Andreas Klein Fair DRM Fair Use by Secret Sharing André Adelsbach, Ulrich Greveler, Jörg Schwenk Klassifizierung von Public Key Systemen mit Multivariaten Quadratischen Polynomen Christopher Wolf Immediate Rekeying by Tree Parity Machines in a WLAN-System Nazita Behroozi Security Challenges of Location-Aware Mobile Business Emin Islam Tatli, Dirk Stegemann, Stefan Lucks Secure Group Communication in WLAN Ad-Hoc Networks with Tree Parity Machines Björn Saballus

23 Multicast-Security Simulation logischer Schlüsselbäume unter Berücksichtigung unterschiedlichen Benutzerverhaltens Alexander Heese, Luigi Lo Iacono, Christoph Ruland Institut für Digitale Kommunikationssysteme Universität Siegen In Multicast-Umgebungen werden Nachrichten von einem Sender an mehrere Empfänger geschickt. Durch Verschlüsselung der Kommunikation und durch kontrolliertes Verteilen der kryptographischen Schlüssel zur Entschlüsselung wird der Zugang zu den Kommunikationsinhalten kontrolliert. Die Verteilung und Verwaltung der kryptographischen Schlüssel ist hierbei von grundlegender Bedeutung und stellt eines der Hauptprobleme bei der Realisierung vertraulicher Multicast-Kommunikation dar. Insbesondere große Gruppen mit häufigen Ein- und Austritten der Mitglieder sind hinsichtlich der Skalierbarkeit und Sicherheit in diesem Zusammenhang zu betrachten. Zur Lösung dieser Problemstellung wurden logische Schlüsselbäume vorgeschlagen, die zur Verwaltung des Gruppenschlüssels dienen [WGL98, WHA99]. Die Vorteile dieser Verfahren gehen offensichtlich verloren, sobald die zugrundeliegende Baumstruktur entartet. Gegenstand dieser Arbeit ist die Veränderungen der Schlüsselbäume anhand des Verhaltens von Benutzergruppen zu simulieren und hieraus den durchschnittlichen Aufwand für die verschiedenen Schlüsselverteilstrategien zu ermitteln. Als Grundlage für die Simulationen werden Untersuchungsergebnisse zum Benutzerverhalten in Multicast-Gruppen herangezogen [Al00, HB01]. Aus dem beobachteten Benutzerverhalten sowohl im MBone als auch bei Netzwerk-Spielen werden Verhaltensmuster herausgearbeitet und darauf aufbauend Szenarien definiert, die die Basis für die Simulationen bilden. Literatur [WGL98] C. K. Wong, M. Gouda, and S. S. Lam. Secure Group Communication using Key Graphs. In Proceedings of ACM SIGCOMM 98, [WHA99] D. M. Wallner, E. G. Harder, and R. C. Agee. Key Management for Multicast: Issues and Architectures. IETF RFC 2627, [Al00] [HB01] Kevin C. Almeroth. A Long-Term Analysis of Growth and Usage Patterns in the Multicast Backbone (MBone). IEEE Infocom, Tristan Henderson and Saleem Bhatti. Modelling User Behaviour in Networked Games. In Proceedings of the 9th ACM international conference on Multimedia, Ottawa, Canada, pp ,

24 Cryptographic Watermarking Stefan Katzenbeisser Institut für Informatik, Technische Universität München D Garching bei München The rapid growth of the Internet as a distribution medium for digital goods increased the risk of copyright infringements. From an economic point of view, this risk makes the commercialization of digital works difficult, if not impossible. Therefore, the need for technical copyright protection solutions has increased steadily over the last years. Robust digital watermarking became a promising technology in the context of copyright protection and was proposed as a central building block in various e-commerce protocols (such as dispute-resolving, copy protection and traitor tracing schemes or DRM applications). Traditionally, the design of watermarking schemes was seen as a signal-processing problem and concentrated on issues such as the imperceptibility of the watermark or its resistance against unauthorized removal. However, when a watermark is to be used in an e-commerce system, its properties may become critical to the security of the overall scheme. It is therefore necessary to gain a thorough and mathematically precise understanding of the essential security properties of watermarks. In this talk, I review recent results [1, 2] that establish the security of two cryptographic protocols that employ watermarking operations as basic primitives. The first protocol can be used in dispute-resolving schemes in order to assure their resistance against an important class of attacks. The second protocol allows to detect forgeries in image files or video streams by embedding a watermark carrying a cryptographic signature. Both constructions are provably secure under standard cryptographic assumptions. References [1] A. Adelsbach, S. Katzenbeisser, H. Veith, Watermarking Schemes Provably Secure Against Copy and Ambiguity Attacks, in ACM Workshop on Digital Rights Management (DRM 2003), Proceedings, Washington DC, 2003, pp [2] J. Dittmann, S. Katzenbeisser, C. Schallhart, H. Veith, Provably Secure Authentication of Digital Media Through Invertible Watermarks, to appear as IACR eprint report,

25 Mental Poker in practice: An extended implementation of Schindelhauer s Toolbox for Mental Card Games Heiko Stamer University of Kassel, Department of Mathematics/Computer Science Heinrich-Plett-Straße 40, D Kassel, Germany stamer@theory.informatik.uni-kassel.de A lot of cryptographic research has been carried out on Mental Poker during the last decades. But efficient implementations are still very rarely. Few years ago Schindelhauer [Sc98] introduced a general toolbox which extends previous work of Crépeau [Cr87]. Roughly speaking, the type of a card is shared among the players through bitwise representation by quadratic (non-)residues. Thus the security relies on the well known Quadratic Residuosity Assumption (QRA). Unfortunately, the size of a card grows linearly in the number of players and logarithmically in the number of different cards. Recently a more efficient solution [BS03] was proposed, whose security can be based on the Decisional Diffie-Hellman Assumption (DDH). Moreover, the encoding is independent of the number of players respectively cards. My talk presents the technical details of a extended implementation for the open source library libtmcg [St05]. Further we discuss the practicability while considering as example the german card game Skat [St04]. References [Sc98] [Cr87] [BS03] Christian Schindelhauer. Toolbox for Mental Card Games. Technical Report A-98-14, University of Lübeck, Claude Crépeau. A zero-knowledge poker protocol that achieves confidentiality of the players strategy or how to achieve an electronic poker face. In Advances in Cryptology: CRYPTO 86 Proceedings, Lecture Notes in Computer Science 263, pp , Adam Barnett and Nigel P. Smart. Mental Poker Revisited. In K.G. Paterson (Ed.): Cryptography and Coding 2003, Lecture Notes in Computer Science 2898, pp , [St04] Heiko Stamer. Kryptographische Skatrunde. In Offene Systeme 4, pp , ISSN [St05] Heiko Stamer. 5

26 Mystery-Twister Hans Dobbertin, Magnus Daum, Patrick Felke, Gregor Leander CITS Research Group Ruhr Universität Bochum Die Forschungsgruppe Kryptologie und Informationssicherheit (CITS) der Universität Bochum organisiert im Jahr 2005 den internationalen Kryptographie-Wettbewerb Mystery-Twister. Es geht um Verschlüsselung, die jedem jeden Tag begegnet, meist unbemerkt. Der Wettbewerb richtet sich sowohl an Jedermann, der gerne verstehen möchte wie Verschlüsselung bei Online Banking, im Handy oder im Bankautomaten funktioniert, als auch an führende Forscher auf dem Gebiet der Kryptographie. Im Laufe des Jahres 2005 werden 13 CryptoChallenges (CC1 bis CC13) mit steigendem Schwierigkeitsgrad veröffentlicht. Es geht unter anderem um das Entschlüsseln von Nachrichten, das Klonen von Mobil-Telefonen und das Fälschen von digitalen Signaturen. Die Auswahl der Themen für die 13 Challenges werden dabei einen Überblick über viele Bereiche moderner Kryptographie liefern. Einer der wichtigsten Gründe für Mystery-Twister ist es, auch Laien die nötigen Informationen und Einsichten zu vermitteln, die nötig sind, um die Prinzipien moderner kryptographischer Verfahren zu verstehen. Wichtig ist daß Mystery-Twister diese Einsichten spielerisch vermittelt. Für Forscher auf dem Gebiet der Kryptographie sind die CryptoChallenges 9 bis 13 gedacht. Hier sind Profis gefragt und in den Aufgaben geht es zum Beispiel um AES und RSA, also die defacto Verschlüsselungs-Standards schlechthin. Wer hier Erfolg haben will, muss einen echten Beitrag zur aktuellen Forschung leisten. Und nicht zuletzt lohnt sich die Teilnahme bei Mystery-Twister, da......es Spaß macht Codes zu knacken und Geheimnisse zu lüften. 6

27 Erweiterte visuelle Kryptographie Andreas Klein Fachbereich für Mathematik/Informatik, Universität Kassel Visuelle Kryptographie wurde 1994 von Naor und Shamir [3] erfunden. Dabei wird ein Schwarz-Weiß-Bild so in zwei Bilder codiert, daß jedes der beiden Bilden wie ein zufälliges Punktmuster wirkt. Kopiert man jedoch diese Bilder auf Folien und legt diese übereinander, so erscheint wieder das ursprüngliche Bild. Allgemeiner kann man visuelle Kryptographie als ein Shared-Secret- System auffassen, dabei wir das geheime Bild so auf n Folien verteilt, daß man es nur durch das Übereinanderlegen von mindestens k Folien rekonstruieren kann. Eine weitere Verallgemeinerung erzeugt n Folien, so daß man bei jeder der 2 n 1 möglichen Kombinationen von Folien ein anderes Bild sieht. Dabei sind alle Bilder voneinander unabhängig, d.h. man kann durch Untersuchen eines Teils der Bilder keine Rückschlüsse auf die anderen Bilder ziehen. Die folgende Abbildung zeigt das Prinzip mit zwei Folien. = Das dies für jede natürliche Zahl n möglich ist wurde zu erst in [1] bewiesen. In meinem Vortrag werde ich nicht nur die Existenz dieser System beweisen, sondern auch das jeweils beste System charakterisieren. Literatur [1] S. Droste: New results in visual cryptography. In: Advances in cryptology CRYPTO 96, Band 1109 von Lect. Notes Comput. Sci., S Springer, Berlin, [2] A. Klein und M. Wessler: Extended visual cryptography schemes. Information and Computation to appear. [3] M. Naor und A. Shamir: Visual cryptography. In: Advances in Cryptology EUROCRYPT 94 (Herausgegeben von A. D. Santis), Band 950 von Lect. Notes Comput. Sci., S Springer, Berlin,

28 Fair DRM Fair Use by Secret Sharing André Adelsbach, Ulrich Greveler, Jörg Schwenk Horst Görtz Institute for IT security Ruhr-Universität Bochum, Germany We present a fair DRM environment, where each user can act pseudonymously and is entitled to make a fixed number of copies for private purposes (e.g., a maximum of 7 copies) but where the user can be identified and prosecuted when more copies are produced. Our system constitutes three central authorities (pseudonymization, licensing, issuer) in order to protect the user agains malicious providers and uses secret-sharing schemes in a way that a central authority can only reconstruct the secret (the user s identity) when a number of shares is transmitted by the media players. The user s identity will not be known to this authority as long as the user limits the number of private copies to the maximum (being a parameter in the system). The set-up phase of the DRM system let each user choose a pseudonym that is a key pair (sk P, pk P ) and let him send a request for a certificate on this pseudonym Req Pseud = Sign(sk B, pk P terms) to a pseudonymization authority, where terms is a description of the contractual conditions regarding liability and depseudonymization. The users may obtain a licence for an object identified by DOI when he sends a signed request Req = Sign(sk P, < DOI cert P terms >) to the licensing agency. This licence can then be of the form Licence = Sign(sk L, DOI P ol Rights Enc(pk P, key)) where P ol is a random polynom of the same degree as the number of allowed copies so that Shamir s secret-sharing scheme may be used here. We will also present other forms with different performance characteristics. The media player evaluates Pol at position i := Hash(Licence player ID) and transmit share S := DOI Hash(Licence) i P ol(i) so each time the object is played the same share is computed but different players will compute different shares with high probability. 8

29 Klassifizierung von Public Key Systemen mit Multivariaten Quadratischen Polynomen Christopher Wolf ESAT-COSIC, K.U. Leuven, Belgien oder Seit den frühen 1980er Jahren werden multivariate quadratische Gleichungen für die Konstruktion von Public Key Kryptosystemen eingesetzt. Diese Gleichungen sind im Gegensatz zu linearen Gleichungen schwer zu lösen Stichwort N P-Vollständigkeit. Bisher wurden eine Reihe von grundlegenden Klassen vorgeschlagen 1988 das System MIA von Matsumoto-Imai (Schema A), 1996 die HFE (Hidden Field Equations) von Jacques Patarin, und 1997 die Essig und Öl Systeme von Patarin, die 1999 von Kipnis, Goubin und Patarin zu den UOV (Unbalenced Oil and Vinegar) Systemen verallgemeinert wurden. Als letzte Klasse seien noch die 1993 von Shamir als Birational Permutations vorgeschlagenen Systeme genannt, die 2000 von Goubin und Courtois zu TPM (Triangular Plus Minus) spezialisiert wurden und letztere 2004 von Wolf, Braeken und Preneel zu STS (Stepwise Trianguar Systems) in eine andere Richtung verallgemeinert wurden. Diese Grundklassen können in eine einfache Taxonomie gebracht werden. Neben diesen Grundklassen sind auch eine Reihe von generischen Modifikatoren bekannt, wie z.b. die Minus Modifikation, die Plus Modifikation, das Fixieren von Variablen, das Aufspalten in verschiedene Äste, dünn besetzte Polynome für den geheimen Schlüssel sowie das interne Stören der Gleichungen. Darüber hinaus wurden seit Beginn 2000 verschiedene Mischsysteme vorgeschlagen, die jeweils versuchen aus bekannten Bausteinen neue Systeme zu entwickeln. Analog dem Design-Prinzip von Blockchiffren werden dabei (einzeln) schwache Komponenten verbunden, um damit ein stärkeres Gesamtsystem zu erhalten. Vertreter sind die enhanced Tame Transformation Signatures von Chen und Yang sowie die Tractable Rational Maps von Wang und Chang. Beide benutzen als Grundgerüst STS. Während Chen und Yang hier dünn besetzte UOV Systeme einbetten, verwenden Wang und Chang dünn besetzte Monome über einem Erweiterungskörper und sind damit relativ nahe an der MIA Konstruktion. 9

30 Immediate Rekeying by Tree Parity Machines in a WLAN-System Nazita Behroozi Hamburg University of Technology, Distributed Systems, Schwarzenbergstraße 95, D Hamburg Germany nazita behroozi@hotmail.com WPA and IEEE802.11i provide the Pre-Shared-Key (PSK) for key establishment in Wireless LAN (WLAN) in the case that no 802.1x authentication server (RADIUS) is available. Recent research shows the risk of using PSK because the security depends on the length and quality of the pass phrase used [2]. We investigate secure key exchange in WLAN via Tree Parity Machines (TPMs) [1]. In order to evaluate the capability and suitability of an integrated key exchange via TPM, a system was developed which allows the transfer of encrypted data between two WLAN clients. One WLAN client is an embedded system that includes a hardware- and a software-part. It was developed on an embedded ARM processor based development kit with a FPGA device. The other WLAN client was implemented in software running on a Personal Computer. In this work an immediate rekeying scenario was implemented to study frequent key exchange via TPMs in parallel to the encrypted data transmission in WLAN. The immediate rekeying technique allows exchange a new key as soon as the previous key has been exchanged. We measured the number of keys exchanged and used for encryption of different amounts of data to be transferred. The evaluation shows that each Ethernet packet with a maximum length of 1500 byte can be encrypted with a new key in the best case. In Temporal Key Integrity Protocol which is used in WPA and IEEE802.11i, a rekeying message will be sent around every packets to request for a new key [3]. References [1] Kanter, I., Kinzel, W. and Kanter, E.: Secure exchange of information by synchronisation of neural networks Europhysics Letters 57 (1), pp, (2002) [2] Moskowitz, R.: WLAN Testing Reports, PSK as the Key Establishment Method, ICSA Labs, 2003 [3] Walker, J.: Security Series, PartII: The Temporal Integrity Protocol (TKIP), Network Security Intel Corporation,

31 Security Challenges of Location-Aware Mobile Business Emin Islam Tatli, Dirk Stegemann, Stefan Lucks Department of Computer Science, University of Mannheim In addition to mobility, the ability of context awareness and especially location awareness has enhanced mobile businesses to support context-aware services. Today, many different kinds of context-aware services, ranging from finding nearby restaurants [FR] to sending ambulances to people in emergency [LP], have already taken their places in the business. The m-business research group at the University of Mannheim [MB04] aims at building a generic framework that is able to execute any kind of context-aware service. Our talk presents the security challenges of this mobile business framework with special focus on location as a context property. Our analysis shows that, in addition to privacy and confidentiality, other security challenges, especially anonymous and unlinkable services, usability with security, integrity and authenticity of services, secure payment, fair exchange, location-based spamming, rogue access points and forged GPSsignals would directly affect the user acceptance of the m-business framework. Having specified the challenges and possible solutions, our next step is to design an open and flexible security architecture that can be integrated into the application framework. References [MB04] [LP] [FR] The Mobile Business Research Group URL: Locating people in emergency URL: Location-based services for mobile communities URL: 11

32 Secure Group Communication in WLAN Ad-Hoc Networks with Tree Parity Machines Björn Saballus Hamburg University of Technology, Distributed Systems Schwarzenbergstraße 95, Hamburg - Germany Most portable customer devices, such as laptops and PDA s, have the ability to communicate via Wireless Local Area Networks (WLAN). These are described in the IEEE standard which also allows to set up selfconfiguring and self- organising ad-hoc networks without a central server. This work investigates a method to allow secure group communication in ad-hoc networks using symmetric key exchange by Tree Parity Machines (TPMs) [KKK02]. A main problem in secure group communication is the need to establish and distribute a shared, secret key between the members in the group, especially on join- or leave-actions [HD03]. With TPMs, multiparty key exchange is inherently possible based on multiparty synchronisation. A group of TPMs, all sharing one key, can perform a synchronisation with another group of TPMs sharing another key. Once the synchronisation process is finished, all TPMs will share the same key. The suggested chained synchronisation has a runtime-complexity of O(n), with n being the group size. Another concurrent synchronisation has a runtime-complexity of O(log(n)). This Work-in-Progress presents some preliminary experimental results on the usability of TPMs in secure group communication. References [KKK02] Kanter, I., Kinzel, W. and Kanter, E.: Secure exchange of information by synchronisation of neural networks, Europhysics Letters 57 (1), pp, , 2002 [HD03] Hardjono, T. and Dondeti, L.R.: Multicast and Group Security, Artech House Inc,

33 3. Krypto-Tag Workshop über Kryptographie Technische Universität Darmstadt Ralf-Philipp Weinmann (Hg.) Technische Universität Darmstadt, Fachbereich Informatik Hochschulstr. 10, D Darmstadt 15. September 2005 Technical Report No. TI-1/05 1

34 Inhaltsverzeichnis Fault Attacks on Combiners with Memory Frederik Armknecht and Willi Meier Some Thoughts about Block Ciphers and Stream Ciphers Erik Zenner Runners, Starting Lines and Mutual Distances: On the Security of Tree Parity Machine Key Exchange Markus Volkmer and Florian Grewe A Framework for Computer Proofs in Probability Theory for Use in Cryptography Markus Kaiser, Johannes Buchmann, and Tsuyoshi Takagi. 6 Hocheffiziente modulare Multiplikation für GF(P ) Rainer Blümel, Ralf Laue und Sorin A. Huss Angriffe auf RC4 Andreas Klein Verifizierbares geheimes Mischen Heiko Stamer Designing Secure Protocol Implementations Philipp A. Baer Improved Boomerang Attack on Eight-Round-Serpent Anne Schwalb Wiedererkennung anonymer Knoten Stefan Schlott und Frank Kargl Strengthening the E 0 Keystream Generator against Correlation Attacks and Algebraic Attacks Frederik Armknecht and Matthias Krause and Dirk Stegemann 13 Elektronische Wahlen mit Observer Jörn Schweisgut Kryptographisch t-private Auktionen Nina Moebius

35 Fault Attacks on Combiners with Memory Frederik Armknecht and Willi Meier Universität Mannheim FH Aargau Germany Switzerland Fault attacks are powerful cryptanalytic tools that are applicable to many types of cryptosystems. Recently, general techniques have been developed which can be used to attack many standard constructions of stream ciphers based on LFSR s. Some more elaborated methods have been invented to attack RC4. These fault attacks are not applicable in general to combiners with memory. In this paper, techniques are developed that specifically allow to attack this class of stream ciphers. These methods are expected to work against any LFSR-based construction that uses only a small memory and few input bits in its output function. In particular, efficient attacks are described against the stream cipher E0 used in Bluetooth, either by inducing faults in the memory or in one of its LFSR s. In both cases, the outputs derived from the faulty runs finally allow to describe the secret key by a system of linear equations. Computer simulations showed that inducing 12 faults sufficed in most cases if about 2500 output bits were available. Another specific fault attack is developed against the stream cipher SNOW 2.0, whose output function has a 64-bit memory. Similar to E 0, the secret key is finally the solution of a system of linear equations. We expect that one fault is enough if about 2 12 output words are known. References [1] Biham, Granboulan, Nguyen: Impossible Fault Analysis of RC4 and Differential Fault Analysis of RC4, FSE 2005, Springer, [2] Boneh, DeMillo, R.J. Lipton: On the Importance of Checking Cryptographic Protocols for Faults, Eurocrypt 1997, LNCS 1233, pp , Springer, [3] Biham, Shamir: Differential fault analysis of secret key cryptosystems, Crypto 1997, LNCS 1294, pp , Springer, [4] Hoch, Shamir: Fault Analysis of Stream Ciphers, CHES 2004, LNCS 3156, pp , Springer,

36 Some Thoughts about Block Ciphers and Stream Ciphers Erik Zenner Cryptico A/S Block ciphers and stream ciphers are well-known concepts in cryptography. They are encountered in almost all major textbooks, and they form the basis of all known symmetric encryption algorithms. Nonetheless, many misconceptions surround these concepts. Adi Shamir announced in early 2004 that stream ciphers were dead, just to revoke that statement on the SASC workshop and on Asiacrypt At the SKEW workshop that was held in May 2005 in Aarhus, it turned out that the specialists in stream cipher cryptography do not even agree on what a stream cipher actually is. Security-wise, things do not look much better. The common perception is that block ciphers (like AES) are more secure than stream ciphers - but are they? In fact, such a statement is comparing apples with oranges, since block ciphers are not used directly for encryption. Instead, they are used in a mode of operation, which is usually a stream cipher that often turns out to be cryptographically weaker than the dedicated stream ciphers themselves. In this talk, we attempt to clear up some of the conceptional muddle around block and stream ciphers. We will review some definitions and notions of security, point out the true advantages of block and stream ciphers, and advocate a clear use of terminology. 4

37 Runners, Starting Lines and Mutual Distances: On the Security of Tree Parity Machine Key Exchange Markus Volkmer and Florian Grewe Hamburg University of Technology, Computer Engineering VI Schwarzenbergstraße 95, D Hamburg, Germany Attacks on key exchange by Tree Parity Machines (TPMs) [1] exist that also employ one or more TPMs (e.g. [2]). An attacker tries to learn the internal state of the interacting parties from observing the publicly communicated outputs of their synchronisation process. The security of the principle has so far only been accessed experimentally in terms of success probabilities of an attacker with respect to the chosen TPM parameters. This contribution suggests to take a different view on the key exchange and the related attacks. The interacting as well as the attacking TPMs are considered to be runners that start a race at different starting lines chosen at random. Although the finish is determined by the choice of the interacting runners, it is unknown to all the runners. The attacking runner has the disadvantage of being slower than the two interacting runners, because he can only chase them and does not interact. In this unusual race, the starting lines can be chosen freely and neither runner knows the starting line of the other runners. A slower attacking runner can thus win by chance, if he picks an advantageous starting line relative to one of the other two runners. In particular, the initial mutual distances between the runners and the attacking runner determine who will win. This perspective still leaves one with success probabilities, after all. Yet it allows for fundamental insights in terms of the relation between the initial mutual distances, synchronisation times, success probabilities and thus the discussion of security. Countermeasures against attacks are also motivated: increase the interaction of your runners, slow down the attacking runner or choose close starting lines. References [1] Kanter, I., Kinzel, W. and Kanter, E.: Secure exchange of information by synchronisation of neural networks, Europhysics Letters 57 (1), pp, , [2] Klimov, A., Mityagin, A., and Shamir, A.: Analysis of neural cryptography, In Proc. of AsiaCrypt 2002, volume 2501 of LNCS, pages , Queenstown, New Zealand, December , Springer Verlag. 5

38 A Framework for Computer Proofs in Probability Theory for Use in Cryptography 1 Markus Kaiser, Johannes Buchmann, and Tsuyoshi Takagi Darmstadt University of Technology Germany Future University - Hakodate Japan Mathematical proofs are often complex and hard to verify by their readers. Consequently, the application of formal proof systems are a useful approach in the area of verification. We present a framework for computer proofs in probability theory. Therefore we describe formalized probability distributions and fundamental lemmata concerning σ-algebras, probability spaces and conditional probabilities. These are given in the formal language of the formal proof system Isabelle/HOL. Besides we describe an application of the presented formalized probability distributions and fundamental lemmata to cryptography. Our achievements are a step towards computer verification of cryptographic primitives. They describe a basis for computer verification in probability theory for interactive proof constructions within the formal proof system mentioned above. Computer verification can be applied to further problems in cryptographic research, if the corresponding basic mathematical knowledge is available in a database. References [BR93] M. Bellare and P. Rogaway. Random oracles are practical: a paradigm for designing efficient protocols. In proceedings of the First ACM Conference on Computer and Communication Security, [Hur01] Joe Hurd. Formal Verification of Probabilistic Algorithms. PhD thesis, Trinity College University of Cambridge, [Ric03] [Sho01] Stefan Richter. Formalizing Integration theory, with an Application to Probabilistic Algorithms. Diplomarbeit, Technische Universität München, V. Shoup. OAEP Reconsidered. In Advances in Cryptology CRYPTO 2001, volume 2139 of Lecture Notes of Computer Science, pages Springer-Verlag, This work was partially funded by the German Federal Ministry of Education and Technology (BMBF) in the framework of the Verisoft project under grant 01 IS C38. The responsibility for this article lies with the authors. 6

39 Hocheffiziente modulare Multiplikation für GF(P ) Rainer Blümel, Ralf Laue und Sorin A. Huss cv cryptovision GmbH Gelsenkirchen Integrierte Schaltungen und Systeme TU Darmstadt Die Leistungsfähigkeit heutiger Public-Key-Systeme beruht hauptsächlich auf der Effizienz der modularen Körper-Multiplikation. Wir stellen einen Algorithmus für GF(P ) mit nur n 2 + 7n Wort-Multiplikationen vor. Die grundlegende Idee ist von Karatsuba abgeleitet: Je zwei Wort-Multiplikationen werden zu einer Wort-Multiplikation zusammengefasst, allerdings ohne Einsatz von Rekursion. Die beschleunigte Multiplikationsformel ermöglicht es, eine Mehr-Wort-Multiplikation mit nur n (n+1) 2 Wort-Multiplikationen zu berechnen. Die Reduktion lässt sich auch als Multiplikation darstellen: 0 X Y P Z < P. So kann die beschleunigte Multiplikation auch für die Reduktionsphase eingesetzt werden. Sortiert man die Terme in Reihenfolge absteigender Wertigkeit, ergibt sich X Y P Z = 0 i 1 { ((x i + x j ) (y i + y j ) + p i z i x i y i ) 2 (i+j)b i=n 1 j=0 (p i z i x i y i )2 2ib + n 1 j=i+1 (p i z i x i y i (p i + p j ) (z i + z j )) 2 (i+j)b }. In jedem Schritt wird nur je ein neues z i benötigt, welches zu Beginn jedes Schrittes mit einem Look Ahead-Mechanismus abgeschätzt wird: Die höchstwertigen Terme des i-ten Schrittes mit Ausnahme der z i enthaltenden Teilterme werden ausgewertet. Aus dem Ergebnis lässt sich z i mit einer Division durch P errechnen. Als Ersatz für die Division wird eine Multiplikation mit dem Kehrwert der führenden Wörter von P verwendet (ähnlich zu Barrett). Ungenaue z i werden durch Korrekturberechnungen ausgeglichen. Geschickte Parameterwahl ermöglicht es, die Genauigkeit soweit zu erhöhen, dass der Aufwand zur Korrektur ignoriert werden kann (Fehlerwahrscheinlichkeit 10 5 ). Für jeden Schritt werden 2 Wort-Multiplikationen für (p i z i x i y i ) und weitere n 1 für die Summen benötigt. Unsere Implementierung benötigt weiterhin 3 Wort-Multiplikationen für die Auswertung der höchstwertigen Terme und 3 für die Multiplikation mit dem Kehrwert. Als Summe ergibt sich die oben erwähnte Komplexität von n 2 + 7n. 7

40 Angriffe auf RC4 Andreas Klein Arbeitsgruppe Computational Mathematics, Universität Kassel RC4 ist eine Stromchiffre die 1987 von Ron Rivest erfunden wurde. Die Chiffre ist sehr schnell und wird daher in Anwendungen wie SSL und WEP eingesetzt. Damit gehört sie zu den populärsten Stromchiffren. Der Algorithmus wurde als Firmengeheimnis von RSA Data Security Inc. behandelt. Die Beschreibung des Algorithmus wurde erst 1994 anonym auf der Mailing- Liste Cypherpunks veröffentlicht. Die Struktur des Verfahrens ist recht einfach. Der Algorithmus arbeitet mit einer Permutation S 0,..., S 255 der Zahlen von 0 bis 255. In jedem Schritt wird auf folgende Weise eine Pseudozufallszahl erzeugt. i = (i + 1) mod 256 j = (j + S i ) mod 256 vertausche S i und S j t = (S i + S j ) mod 256 gebe S t aus In meinem Vortrag möchte ich eine Schwäche der von RC4 erzeugten Pseudozufallsfolge vorstellen und zeigen wie diese zu einem Angriff ausgenutzt werden kann. In Gegensatz zur bekannten FMS-Attacke [1] braucht der neue Angriff keine schwachen Schlüssel. Daher reichen jetzt bereits etwa statt abgefangener Sitzungen zur Analyse aus. Mit der neuen Technik ist es auch dann möglich den Schlüssel zu rekonstruieren, wenn die ersten 256 Byte des Schlüsselstroms für eine Analyse nicht zu Verfügung stehen. Literatur [1] S. Fluhrer, I. Mantin, and A. Shamir. Weakness in the Key Scheduling Algorithm of RC4. In Selected areas in cryptography, volume 2259 of LNCS, pages 1 24, Berlin, Springer. 8

41 Verifizierbares geheimes Mischen Heiko Stamer Universität Kassel, Fachbereich Mathematik/Informatik Heinrich-Plett-Straße 40, D Kassel stamer Geheimes Mischen hat vielfältige Anwendungsmöglichkeiten. Beispielsweise ist es ein wichtiger Bestandteil von sicheren Mix-Netzen, welche wiederum für elektronische Wahlen oder Online-Auktionen benötigt werden. Weitere Einsatzgebiete sind Anoymisierungsdienste, datenschutzfreundliche Statistikabfragen sowie elektronische Bezahlsysteme. Fast alle der genannten Anwendungen erfordern jedoch neben einer unbedingten Geheimnisbewahrung der Permutation zusätzlich deren Verifizierbarkeit, d. h. unter vernünftigen kryptographischen Annahmen soll die Korrektheit des Mischens beweisbar sein. Seit kurzem wird dieser Bereich intensiv untersucht [Ne01, Gr03, Fu04, Wi05, Gr05]. Infolgedessen gibt es mittlerweile auch effiziente und praxistaugliche Protokolle. Der Vortrag stellt kurz die bekannten Techniken für verifizierbares geheimes Mischen gegenüber und vergleicht sie hinsichtlich ihrer Berechnungsund Kommunikationskomplexität. Weiterhin werden die Grundlagen sowie die praktische Umsetzung eines aktuellen Verfahrens [Gr05] diskutiert. Literatur [Fu04] [Gr03] [Gr05] [Ne01] [Wi05] J. Furukawa. Efficient, Verifiable Shuffle Decryption and Its Requirement of Unlinkability. In Public Key Cryptography - PKC 2004 Proceedings, Lecture Notes in Computer Science 2947, pp , J. Groth. A Verifiable Secret Shuffle of Homomorphic Encryptions. In Public Key Cryptography - PKC 2003: Proceedings, Lecture Notes in Computer Science 2567, pp , J. Groth. A Verifiable Secret Shuffle of Homomorphic Encryptions. Cryptology eprint Archive, Report 2005/246. C.A. Neff. A Verifiable Secret Shuffle and its Application to E-Voting. 8th ACM Conference on Computer and Communications Security, pp , D. Wikström. A Sender Verifiable Mix-Net and a New Proof of a Shuffle. Cryptology eprint Archive, Report 2005/137. 9

42 Designing Secure Protocol Implementations Philipp A. Baer University of Kassel, FB 16, FG Distributed Systems, Germany Security network protocols specified in only a formal language normally cannot be translated into software right away, mostly due to missing implementation details. Furthermore, a naïve implementation is often error-prone because of the variety of environmental configurations. We propose the interactive assisted modeling (IAM) architecture for security protocol specification. Its objective is to improve the quality of protocol implementations and portability. The IAM architecture offers detail level-filtered modeling, support for group communication, and optimized code generation. An abstract and platform-independent representation language is introduced to guarantee portability of protocol specifications. The AIM modeling interface provides an abstract view on the communication scenario and the environment. It furthermore supports specification of environmental properties such as characteristics of the communication media. Third-party tools for protocol and security analysis will also be supported. Projects like [1] follow a similar approach. Cryptographic or communication primitives and common networking parameters are directly mapped into our representation language. It is similar to MuCAPSL [2] which is primarily targeted towards specification of multicast authentication protocols. The objective of MuCAPSL is protocol analysis whereas our language was explicitly designed for automatic code generation. In another transformation process a protocol specification is translated into intermediate or native code. The intermediate code target is similar to Microsoft s Intermediate Language (MSIL). An optimized interpreter executes this code (communication sandboxing). Literatur [1] E. Saul, A.C.M. Hutchison. SPEAR II: The Security Protocol Engineering and Analysis Resource. 2nd Annual South African Telecommunications, Networks and Applications Conference, [2] J. Millen, G. Denker. CAPSL and MuCAPSL. Journal of Telecommunications and Information Technology 4,

43 Improved Boomerang Attack on Eight-Round-Serpent Anne Schwalb Mathematisches Institut Justus-Liebig-Universität Giessen, Germany One of the five AES finalists is the block cipher Serpent (see [ABK98]) which is a 32-round SP-network. In the beginning of this talk a short introduction to this cipher is given. Then the boomerang attack on 8-round-Serpent is presented as an extension of the differential cryptanalysis. The boomerang attack is a key-recovery attack which needs chosen-plaintexts and adaptive-chosen-ciphertexts. Both, the differential cryptanalysis and the boomerang attack, use characteristics. Since for the efficiency of the attack it is important that the used characteristics have a probability as high as possible, an introduction to differential characteristics is also given as a component of the boomerang attack. As a novel contribution, the attack on the 8-round-Serpent as given in [KKS00] is improved by using a characteristic with a probability higher than the one used there. This better characteristic is taken from [BDK01]. The attack presented in [KKS00] requires chosen plaintexts and ciphertexts (which means the entire codebook), bytes of memory and time equivalent to approximately round-Serpent-encryptions. The new attack which uses the better characteristic from [BDK01] also works with the entire codebook, which means it also requires chosen plaintexts and ciphertexts and bytes of memory but it decreases the required time to approximately round-Serpent-encryptions. To the best of our knowledge this new attack is the best published attack on 8-round-Serpent. References [ABK98] R. J. Anderson, E. Biham, L. R. Knudsen. Serpent: A Proposal for the Advanced Encryption Standard. NIST AES Proposal, [KKS00] T. Kohno, J. Kelsey, B. Schneier. Preliminary Cryptanalysis of Reduced- Round Serpent. Third AES Candidate Conference, [BDK01] E. Biham, O. Dunkelman, N. Keller. The Rectangle Attack - Rectangling the Serpent. Proceedings of EUROCRYPT 2001, Advances in Cryptology, LNCS 2045, Springer 2001, S

44 Wiedererkennung anonymer Knoten Stefan Schlott und Frank Kargl Universität Ulm {stefan.schlott Die Privatsphäre ist ein wichtiger Aspekt des Ubiquitous Computings. Um das Erstellen von Personenprofilen zu verhindern, werden die Knoten solcher Netze anonymisiert. Dies kann beispielsweise über ständig wechselnde Identitäten oder Kommunikation über Mixkaskaden geschehen. In bestimmten Situationen soll es jedoch möglich sein, einen Knoten wiederzuerkennen. Beim erfolgreichen Wiedererkennen soll ein Knoten auf entsprechende gespeicherte Werte (z.b. der letzte Sitzungsschlüssel) zurückgreifen können; falls jedoch keine passende Kommunikationsbeziehung besteht, soll durch den Erkennungsversuch keiner der Knoten zusätzliche Informationen erlangen. Das Abhören der Kommunikation darf ebenfalls keine Informationen preisgeben. Wiedererkennung anhand des ausgetauschten Schlüssels: Bei einer ersten Verbindung werden ein gemeinsamer Schlüssel k sowie ein Token T (String) vereinbart. Will A nun B wiedererkennen, so sendet A zunächst eine Nonce N A. B sendet nun für alle gespeicherten Verbindungen E k (T, N A, N B ). A versucht nun, jedes der Pakete mit allen ihm bekannten Schlüsseln zu dechiffrieren. Erhält er hierbei T und N A, so war die Erkennung erfolgreich. Um die Suche nach dem richtigen Schlüssel bei A einzugrenzen, kann zusätzlich ein Index ID übertragen werden; der Indexraum darf aber nicht so groß sein, daß anhand von ID eine Identifizierung möglich ist. Bei Verwendung des Indexes ist die Reihenfolge, in der B seine Pakete überträgt, zufällig zu wählen; sonst wäre eine u.u. Identifizierung anhand der ID-Reihenfolge möglich. Nach einer erfolgreichen Wiedererkennung wird ein neues ID vereinbart. Wiedererkennung mit Hash Chains: Bei der ersten Verbindung wird neben dem Schlüssel k der Start von zwei Hashchains h A, h B ausgetauscht. Zum Wiedererkennen sendet A an B eine Nonce N A. B sendet an A N B, H k (h B, N A, N B ) (mit H k dem k-ten Element der Hashchain und k = k 1 bei jedem Versuch), k, N B. A berechnet nun den kten Wert von jedem gespeicherten h B ; stimmt dieser mit dem übertragenen Wert überein, war die Erkennung erfolgreich. A sendet nun seinerseits H j (h A, N A, N B ) (mit j = j 1 bei jedem Versuch), B erkennt A nach dem identischen Verfahren. 12

45 Strengthening the E 0 Keystream Generator against Correlation Attacks and Algebraic Attacks Frederik Armknecht and Matthias Krause and Dirk Stegemann University of Mannheim Germany Stream ciphers are widely used for online-encryption of arbitrarily long data. An important class of stream ciphers are combiners with memory, with the E 0 generator from the Bluetooth standard for wireless communication [2] being their most prominent example. E 0 consists of 4 driving devices, a finite state machine (FSM) C with a 4 bit state, an output function f and a memory update function δ. At each clock, one keystream bit z t is produced from the output X t {0, 1} 4 of the driving devices and the current state C t {0, 1} 4 of the FSM according to z t = f(c t, X t ), and the state of the FSM is updated to C t+1 := δ(c t, X t ). So far, the best publicly known attacks against combiners with memory are correlation attacks [4] and algebraic attacks [1]. Correlation attacks exploit linear equations L(X t,..., X t+r 1, z t,..., z t+r 1 ) = 0 that are true with some probability λ with λ 0. Algebraic attacks use valid nonlinear equations of preferably low degree to describe the secret key by a system of equations. We show how to avert a special class of correlation attacks [3] that is currently the most effective against E 0 and introduce a general design principle which guarantees that all valid equations have a degree not smaller than a certain lower bound. Combining these results, we construct a slightly modified version of E 0 with significantly improved resistance against correlation attacks and algebraic attacks. References [1] Armknecht, Krause: Algebraic Attacks on Combiners with Memory, Crypto [2] Bluetooth specification Version [3] Lu, Vaudenay: Faster Correlation Attack on the Bluetooth Keystream Generator, Crypto [4] Salmasizadeh, Golić, Dawson, Simpson: A Systematic Procedure for Applying Fast Correlation Attacks to Combiners with Memory, SAC

46 Elektronische Wahlen mit Observer Jörn Schweisgut Mathematisches Institut, Justus-Liebig-Universität Giessen Im Herbst 2005 wird in Estland die Stimmabgabe zu den Kommunalwahlen über Internet möglich sein. Die Entscheidung des estnischen Parlamentes verdeutlicht den immer breiter werdenden Trend zu elektronischen Wahlen. Ein großes Problem elektronischer Wahlen ist die Unüberprüfbarkeit (receipt-freeness). Selbst wenn ein Wähler mit einem Angreifer kooperiert, darf es ihm nicht möglich sein, beweisen zu können, wie er gewählt hat, da er sonst erpressbar oder bestechlich wird. Das System, das bisher die Unüberprüfbarkeit (am besten) realisiert hat, ist das von Hirt und Sako [HS00]. Es hat sich jedoch herausgestellt, das hier ein Angreifer einen Wähler zwingen kann zufällig oder gar nicht zu wählen. In diesem Vortrag wird zunächst kurz ein Wahlsystem erläutert, das zur Stimmabgabe eine manipulationssichere Hardware, einen Observer, verwendet, der Unüberprüfbarkeit wie in [HS00] gewährleistet, jedoch effizienter ist. Das System wurde 2001 von Magkos et al. vorgestellt [MBC01]. Innerhalb des Protokolls wird ein honest-verifier Zero-Knowledge Beweis vom Observer gegenüber dem Wähler durchgeführt. Im Gegensatz zur Meinung der Autoren ist dieser Beweis übertragbar, wenn der Wähler eine Challenge wählt, die der Angreifer vorher festgelegt hat. Im Rahmen des Vortrags wird daher beschrieben, wie dieses Problem mittels eines Designated-Verifier Beweises (siehe [JSI96]) gelöst werden kann. Literatur [HS00] Martin Hirt, Kazue Sako. Efficient Receipt-Free Voting Based on Homomorphic Encryption. Band 1807 der LNCS, Springer-Verlag, Berlin. Eurocrypt Advances in Cryptology, Seiten , [JSI96] Markus Jakobsson, Kazue Sako, Russel Impagliazzo. Designated-verifier proofs and their applications. Band 1070 der LNCS, Springer-Verlag, Berlin. Eurocrypt Advances in Cryptology, Seiten , [MBC01] Emmanouil Magkos and Mike Burmester and Vassilis Chrissikopoulos. Receipt-freeness in Large-scale Elections without Untappable Channels. In B. Schmid et al., editor. First IFIP Conference on E-Commerce, E- Business, E-Government (IEEE), Seiten ,

47 Kryptographisch t-private Auktionen Nina Moebius Universität zu Lübeck Bei einer Auktion haben die Bieter, die ein Produkt ersteigern wollen, und der Verkäufer des Artikels gegensätzliche Interessen. Ein Bieter ist daran interessiert, das Produkt für einen möglichst geringen Preis zu erhalten, während der Verkäufer seinen Erlös maximieren möchte. Eine Gemeinsamkeit ist jedoch, dass beide Parteien sich einen für sie fairen Verlauf der Auktion wünschen. Das heißt, die Preisfindung muss so gestaltet sein, dass jeder Bieter den für ihn wahren Wert des Produktes bietet. Das in dieser Arbeit präsentierte Protokoll fördert die Fairness einer Auktion dadurch, dass kein Wissen im kryptographischen Sinn über die Gebote der Bieter preisgegeben wird. Nach dem Ende der Auktion werden lediglich der Höchstbietende und sein Gebot veröffentlicht. Dies gilt auch, wenn sich t Angreifer zu einer Koalition zusammenschließen und gemeinsam versuchen Wissen zu gewinnen. Die von uns betrachteten Angreifer sind passiv, d.h. sie folgen der Protokollspezifikation. Weiterhin ist es möglich, das Ergebnis der Auktion zu verifizieren. Wir erweitern das in [NPS99] entwickelte 1-private Auktionsprotokoll unter Verwendung von Yaos garbled circuits ([Y82]). Dabei stützen wir uns auf den in [HJS05, S05] vorgestellten Auktionsmechanismus. Dieser ist informationstheoretisch sicher, benötigt aber eine sehr große Zahl an zufälligen Bits, sodass eine praktische Umsetzung nicht möglich ist. In unserem Protokoll wird durch Zuhilfenahme von kryptographischen Primitiven die Anzahl der Zufallsbits eingeschränkt, sodass eine Implementierung möglich ist. Literatur [HJS05] M. Hinkelmann, A. Jakoby, P. Stechert, Dynamic t-private Auctions, Technical Report SIIM-TR-A-05-11, Universität zu Lübeck, 2005 [NPS99] M. Naor, B. Pinkas, R. Sumner, Privacy Preserving Auctions and Mechanism Design, 1st ACM Conference on Electronic Commerce, pp , [S05] P. Stechert, Dynamic Private Auctions, Diplomarbeit, Institut für Theoretische Informatik, Universität zu Lübeck, 2005 [Y82] A. C. Yao. Protocols for secure computations, 23rd FOCS, pp ,

48 4. Krypto-Tag Workshop über Kryptographie Ruhr-Universität Bochum Ulrich Greveler (Hg.) Ruhr-Universität Bochum, Horst-Görtz-Institut für IT-Sicherheit Universitätsstr. 150, D Bochum 11. Mai 2006 Technical Report No. NDS-1/06

49 Inhaltsverzeichnis New Paradigms for Group Signatures in Federated Systems Mark Manulis Über Hash-Funktionen Christopher Wolf Härten von RC4-artigen Kryptosystemen Andreas Klein Cryptographically Sound Automatic Analysis of Security Protocols Ralf Küsters Implementing Direct Anonymous Attestation for the TPM-Emulator Project Heiko Stamer Encrypted and Authenticated Communication via Tree-Parity Machines in AMBA Bus Systems Sascha Mühlbach, Markus Volkmer and Sebastian Wallner Reducing the Memory Requirements of BDD-Attacks on LFSR-based Stream Ciphers Matthias Krause and Dirk Stegemann On proving Completeness, Soundness and Security of Authenticated Tree Parity Machine Key Exchange Markus Volkmer Generische Modifikationen von codebasierten Kryptosystemen Christian Wieschebrink Kryptographische Verfahren zur elektronischen Signatur Franz Scheerer

50 New Paradigms for Group Signatures in Federated Systems Mark Manulis Horst-Görtz Institute for IT-Security Ruhr-University of Bochum, Germany For many mutli-party applications group signatures are important cryptographic primitives that can be used for the purpose of anonymity and privacy. Group signatures can be used by employees of a company to sign documents on behalf of the company, or in electronic voting and bidding scenarios. In classical group signatures members of a group are able to sign messages anonymously on behalf of the group. However, there exists a designated authority, called group manager, that initializes the scheme, adds new group members, and is able to open group signatures, i.e., identify the signer. Some group signature schemes distinguish between two management authorities: a membership manager that sets up the scheme and controls admission to the group, and a revocation manager that opens the signatures. Obviously, in classical group signatures the group manager is given enormous power compared to other group members and is required to be trusted to act as predestinated. On the other hand there exist multi-party applications where such centralized control (trust) is undesirable, e.g., distributed or federated systems. For this kind of applications it is desirable to have a group signature scheme which provides similar properties but is independent of any centralized control. In this talk we summarize research results concerning this issue. In particular we have proposed a novel group-oriented signature scheme called democratic group signatures [Ma06] which can be seen as a variant of classical group signatures with group manager s rights equally distributed between all members of the group. In democratic group signatures each group member can sign on behalf of the group and is also able to identify the signer of a given group signature. The signer s anonymity is provided only against non-members who are only able to verify group signatures. The group membership is controlled jointly by all group members. We also consider dynamic groups where group membership may vary other time. Obviously, for security reasons in this case relevant group secrets have to be changed. In a subsequent work [MaSaSc06] we have described linkable democratic group signatures where anonymity requirement has been relaxed to allow linkability of issued group signatures. Linkability is useful in some application scenarios which subsume communication of group members and non-members. By allowing linkability we were also able to obtain more efficient constructions. References [Ma06] Mark Manulis. Democratic Group Signatures - On an Example of Joint Ventures - Fast Abstract. In Proceedings of ACM Symposium on Information, Computer and Communications Security (ASI- ACCS 06), pp. 365, ACM Press, Full version at: [MaSaSc06] Mark Manulis and Ahmad-Reza Sadeghi and Jörg Schwenk. Linkable Democratic Group Signatures. In Proceedings of the 2nd Information Security Practice and Experience Conference (ISPEC 2006), LNCS 3903, pp , Springer,

51 Über Hash-Funktionen Christopher Wolf École Normale Supérieure, Département d Informatique 45 rue d Ulm, F Paris Cedex 05, France Christopher.Wolf@ens.fr or chris@christopher-wolf.de Unter eine Hashfunktion verstehen wir eine Abbildung H : {0, 1} {0, 1} k für fixes k N. Umgangssprachlich setzt eine Hashfunktion einen Binärstring beliebiger Eingabelänge in einen Binärstring fester Eingabelänge um. Die obige Definition lässt sich natürlich auch auf nichtbinäre Eingaben verallgemeinern. Wir behalten im folgenden jedoch die Einschränkung auf Binärstrings bei, da dies der in der Praxis am häufigsten vorkommende Fall ist. Hashfunktionen werden in vielen Bereichen eingesetzt, z.b. zur Suche, aber auch in der Kryptographie. Um von kryptographischem Interesse zu sein, müssen Hashfunktionen allerdings noch über folgende weitere Eigenschaften verfügen: 1. Einweg-Eigenschaft: es ist für eine gegebene Ausgabe y {0, 1} k schwierig, ein x {0, 1} zu finden, so dass y = H(x) gilt ( one-wayness, preimage resistance ) 2. Schwache Kollisionsbeständigkeit: es ist für gegebenes x {0, 1} schwierig ein x {0, 1} zu finden mit x x und H(x) = H(x ) ( 2 nd preimage resistance, weak collision resistance ) 3. Starke Kollisionsbeständigkeit: es ist schwierig, zwei x, x {0, 1} zu finden mit x x und H(x) = H(x ) ( strong collision resistance, collision resistance ) 4. Null-Beständigkeit: es ist schwierig, ein x {0, 1} zu finden mit H(x) = 0 für 0 dem Null- Vektor in {0, 1} k ( zero resistance ) Die letzte Eigenschaft wird in manchen Protokollen benötigt und ist nicht allgemein anerkannt. Sie ist insbesondere ein Spezialfall der Einweg-Eigenschaft. Beispiele solcher Funktionen sind MD5, SHA-1, RIPEMD-128, RIPEMD-160 sowie die SHA-2 Familie SHA-224, SHA-256, SHA-384 und SHA-512. Des weiteren seien noch neuere Konstruktionen wie Tiger, Whirlpool oder VSH (Very Smooth Hash) erwähnt. In den letzten Jahren gab es eine Reihe von Angriffen gegen die bisher als sicher geltenden Funktionen der Merkle-Damgaard Familie wie MD5 oder SHA-1. Ein wichtiger Angriff in diesem Kontext ist der generische Angriff von Joux (Crypto 2004). Dieser besagt in etwa dass die parallele Ausführung von zwei Hash-Funktionen nur höchstens so stark ist wie die stärkere der beiden beteiligten Funktionen. Dieses Ergebnis war sehr überraschend, da es vorher zumindest stillschweigenden Konsens darüber gab dass eine solche doppelte angewandte Hash-Funktion sehr viel sicherer sei als die beiden beteiligten Funktionen. Eine Auswirkung dieses Angriffs ist dass das Merkle-Damgaard Design deutlich schwächer ist als zunächst unterstellt. Neben diesem Angriff werden wir noch kurz auf die Auswirkungen der aktuellen Angriffe gegen Hash-Funktionen eingehen. 4

52 Härten von RC4-artigen Kryptosystemen Andreas Klein No Affiliation RC4 ist eine populäre Stromchiffre, die sich gut in Software implementieren lässt. Algorithmus 1 Der RC4 Pseudozufallsgenerator 1: {Initialisierung} 2: i := 0 3: j := 0 4: {Pseudozufallszahlen erzeugen} 5: loop 6: i := (i + 1) mod n 7: j := (j + S[i]) mod n 8: Vertausche S[i] und S[j] 9: k := (S[i] + S[j]) mod n 10: print S[k] 11: end loop In Zeile 7 soll j pseudozufällig geändert werden und in Zeile 9 soll ein pseudozufälliges k ausgewählt werden. Es stellt sich die Frage, warum man an diesen Stellen die Addition modulo 256 als Verknüpfung benutzt. Könnte eine andere Verknüpfung wie z.b. XOR nicht genau so gut oder vielleicht sogar besser geeignet sein? Die überraschende Antwort lautet, dass RC4 massiv schwächer wird falls man die Addition durch XOR ersetzt. In meinem Vortrag möchte ich erläutern, warum dies der Fall ist, welche Operationen statt dessen geeignet sind, und wie man gute Operationen erkennt und findet. 5

53 Cryptographically Sound Automatic Analysis of Security Protocols Ralf Küsters Institut für Informatik Chrstian-Albrechts-Universität zu Kiel Kiel, Germany Two distinct approaches for the rigorous design and analysis of cryptographic protocols have been pursued in the literature: the so-called Dolev-Yao, symbolic, or formal approach on the one hand and the cryptographic, computational, or concrete approach on the other hand. In the symbolic approach, messages are considered as formal terms and the adversary can manipulate these terms based on a fixed set of operations. The main advantage of this approach is its relative simplicity, which makes it amenable to automated analysis. In the cryptographic approach, messages are bit strings and the adversary is an arbitrary probabilistic polynomial-time (ppt) Turing machine. While proofs in this model yield strong security guarantees, the proofs are often quite involved and only rarely suitable for automation. Starting with the seminal work of Abadi and Rogaway (2000), a significant amount of research has been directed at bridging the gap between the two approaches. The goal is to obtain the best of both worlds: simple, automated security proofs that entail strong security guarantees. In this talk, I present results for the automatic analysis of cryptographic protocols and report on current research on linking the symbolic and computational approach. 6

54 Implementing Direct Anonymous Attestation for the TPM-Emulator Project Heiko Stamer University of Kassel, Department of Mathematics/Computer Science Heinrich-Plett-Straße 40, D Kassel, Germany The Trusted Computing Group is an industry organization that aims to develop and promote several standard specifications for trusted computing building blocks and the corresponding software interfaces, e.g. the Trusted Platform Module [TCGa] and the TPM Software Stack [TCGb]. Direct Anonymous Attestation [BCC04] was recently proposed by Brickell, Camenisch, and Chen as successor for the Privacy CA Remote Attestation. The former solution was faced with the disadvantage of a required trusted third party that needs to be involved in every transaction and thus must be highly available. The new approach overcome these obstacles by using advanced cryptographic techniques, i.e. the combination of group signatures with the Camenisch-Lysyanskaya anonymous credential system. We have implemented the TPM DAA commands (cf. Part 3 [TCGa]) in the software-based TPM emulator [Str04, Str06]. The implementation was carefully tested with the IBM DAA Test Suite [IBM]. Our next goal is the support of DAA in the open-source TPM Software Stack [Yod06, TCGb]. References [BCC04] Ernie Brickell, Jan Camenisch and Liqun Chen. Direct Anonymous Attestation. Proceedings of 11th ACM Conference on Computer and Communications Security, ACM Press, [TCGa] Trusted Computing Group. TPM Specification. Version 1.2rev85, [TCGb] Trusted Computing Group. TSS Specification. Version 1.2, [IBM] IBM Direct Anonymous Attestation Tools TPM Test Suite [Str04] Mario Strasser. Software-based TPM Emulator for Linux. Semester Thesis, ETH Zurich, [Str06] Mario Strasser, et al. TPM-Emulator Project. Release 0.3, [Yod06] Kent Yoder, et al. TrouSerS Project. Release 0.2.5,

55 Encrypted and Authenticated Communication via Tree-Parity Machines in AMBA Bus Systems Sascha Mühlbach, Markus Volkmer and Sebastian Wallner Hamburg University of Technology, Institute for Computer Technology, Schwarzenbergstrasse 95, D Hamburg, Germany Security is one of the most important factors in embedded systems such as gaming consoles, identification, pay TV, SIM cards in mobile phones, USB keys for network access or secure devices that handle Digital Rights Management (DRM) for audio and video content. The protection of chip-level microcomputer bus systems in these devices is essential to prevent the growing number of hardware hacking attacks such as presented in [Hu02]. This contribution presents a low-cost encryption and key exchange solution in order to secure chip-level microcomputer bus systems with tree parity machines [VW05]. Due to this intention, a parameterizable Tree Parity Machine Rekeying Architecture (TPMRA) IP-core was designed and implemented. It matches different performance requirements and allows the authentication of bus participants and the encryption of external chip-to-chip buses. We use a stream cipher unit for which the TPM Trajectory Mode together with a universal hash algorithm act as keystream generator. The designer can choose between encrypted or plain bus communication seperately for every bus participant in an early design phase. The solution is transparent and easy applicable to an arbitrary microcomputer bus system for embedded devices. A proof of concept implementation shows the applicability of the TPMRA in the standardized AMBA I/O 1 bus system by implementing the IP-core into the peripheral busto-bus interface (AHB-APB-bridge) without lowering the bus throughput. It will be shown that TPMs can be used in order to protect the ARM bus system considering all AMBA bus features. References [Hu02] A. Huang, Keeping Secrets in Hardware: The Microsoft Xbox Case Study, Proceedings of the Workshop on Cryptographic Hardware and Embedded Systems (CHES 2002), pp , Aug , LNCS, Springer 2002 [VW05] Markus Volkmer, Sebastian Wallner, Lightweight Key Exchange and Stream Cipher based solely on Tree Parity Machines, ECRYPT (European Network of Excellence for Cryptology), pp , July , Graz, Austria 1 8

56 Reducing the Memory Requirements of BDD-Attacks on LFSR-based Stream Ciphers Matthias Krause and Dirk Stegemann Theoretical Computer Science University of Mannheim, Germany The main application of stream ciphers is online-encryption of arbitrarily long data, for example when transmitting speech data between a Bluetooth-headset and a mobile GSM-phone or between the phone and a GSM base station. Examples for practically used and intensively discussed stream ciphers are the E 0 generator used in Bluetooth [1], the GSM cipher A5/1 [2], and the self-shrinking generator [4]. These ciphers consist of a small number of linear feedback shift registers (LFSRs) and a non-linear compression function C : {0, 1} {0, 1}. Based on a secret key k {0, 1} n, the LFSRs produce an internal bitstream z {0, 1} which is then transformed into an output keystream y {0, 1} via y = C(z). For a given plaintext stream p, the ciphertext stream c is computed by bitwise XORing the plaintext and the keystream, i.e., c i = p i y i for all i. Any receiver who knows the secret key k can produce the keystream y himself and compute the plaintext bits as p i = c i y i. In 2002, Krause proposed a generic, Binary Decision Diagram (BDD) based attack [3] on this type of ciphers that reconstructs the internal bitstream z and thereby the secret key k from a short prefix of a given output keystream y. Currently, the BDD-attack is the best known short-keystream attack against E 0 and one of the best generic attacks against A5/1. However, BDD-attacks require a large amount of memory. We approach this problem by presenting various efficiently parallelizable divide-and-conquer strategies (DCS) for E 0 and A5/1 that substantially reduce the memory requirements and allow us to tackle much larger keylengths with fixed computational resources. In the case of E 0, our DCS lowers the attack s memory requirements by a factor of 2 25 and slightly improves its runtime. In [3], the application of the basic BDD-based attack to E 0, A5/1 and the self-shrinking generator were theoretically described, but with rather pessimistic assumptions on the time and memory requirements. We present comprehensive experimental results for the BDD-attack on reduced versions of these ciphers, showing that the performance in practice does not substantially deviate from the theoretical figures. References [1] The Bluetooth SIG. Specification of the Bluetooth System, February [2] M. Briceno, I. Goldberg, and D. Wagner. A pedagogical implementation of A5/1, May [3] M. Krause. BDD-based cryptanalysis of keystream generators. In Proceedings of EUROCRYPT 2002, volume 2332 of LNCS, pages Springer, [4] W. Meier and O. Staffelbach. The self-shrinking generator. In Proceedings of EUROCRYPT 1994, volume 950 of LNCS, pages Springer,

57 On proving Completeness, Soundness and Security of Authenticated Tree Parity Machine Key Exchange Markus Volkmer Hamburg University of Technology, Institute for Computer Technology Schwarzenbergstraße 95, D Hamburg, Germany Attacks on non-authenticated key exchange by Tree Parity Machines (TPMs) also employ one or more TPMs. They try to learn the internal state of the interacting parties from eavesdropping on the publicly communicated outputs of the synchronization process. In this contribution, proofs with regard to Completeness, Soundness and the Security of an authenticated variant of the protocol are sketched. The concept and early practical considerations for an entity authentication and authenticated key exchange in the framework of TPMs were already presented in [1]. The Soundness of the protocol can be proven by showing that a TPM B that has an identical structure to TPM A, as well as an identical output generation, cannot become synchronous by updating its weights according to the learning rule, when having different inputs. The Security with regard to attacks using TPMs can be proven by showing that an attacker E using a TPM with identical structure to the TPMs of parties A and B, as well as with identical output generation, can never remain (and become) synchronous (with A or B) having different inputs from the synchronizing parties. These proofs are available as preprint [2]. Additionally, the proof of Completeness is sketched. It is much more difficult (and lengthy) than the proof of soundness, as (not surprisingly) the relative probabilities for neutral, repulsive and attractive steps in the protocol have to be considered. Next to averting a Man-In-The-Middle attack, the currently known attacks on the symmetric key exchange principle using TPMs can provably be averted for the authenticated variant. References [1] Schaumburg, A.: Authentication within Tree Parity Machine Rekeying, Technical Report TR , Reihe Informatik, Universität Mannheim, Stefan Lucks and Christopher Wolf eds., October [2] Volkmer, M.: Entity Authentication and Authenticated Key Exchange with Tree Parity Machines, IACR Cryptology eprint Archive, Report 2006/112, March

58 Generische Modifikationen von codebasierten Kryptosystemen Christian Wieschebrink Bundesamt für Sicherheit in der Informationstechnik (BSI), Godesberger Allee , Bonn Codebasierte Public-Key-Verfahren stellen eine interessante Alternative zu gängigen PK-Systemen wie RSA und (EC)DSA dar. Ihre Bedeutung könnte zukünftig wachsen, wenn die Entwicklung eines skalierbaren Quantencomputers möglich werden sollte. Zu den bekanntesten codebasierten Systemen zählen die Verfahren von McEliece, Niederreiter und Gabidulin. Diese unterscheiden sich voneinander im Wesentlichen durch die zum Einsatz kommenden Codes. Während das McEliece- System, das auf Goppa-Codes basiert, bis heute ungebrochen ist, sind die beiden anderen genannten Verfahren in ihrer urpsrünglichen Fassung unsicher. Um diese Verfahren zu reparieren, wurden einige generische Modifikationen vorgeschlagen, d.h. solche, die unabhängig vom verwendeten Code sind. Beispielsweise können die bei Niederreiter und Gabidulin zum Einsatz kommenden GRS- bzw. Rangdistanz-Codes durch entsprechende Teilcodes ersetzt werden [1]. Wie sich herausstellt, ist eine solche Modifikation bei unvorsichtiger Parameterwahl allerdings immer noch angreifbar. Dies ist insbesondere dann der Fall, wenn die Dimension des gewählten Teilcodes zu groß ist. Im Vortrag wird ein möglicher Angriff auf das genannte Verfahren skizziert. Zudem sollen weitere denkbare Modifikationen vorgestellt und diskutiert werden. Literatur [1] Berger, T., Loidreau, P.: How to mask the structure of codes for a cryptographic use. Designs, Codes and Cryptography 35(1) (2005) [2] Gabidulin, E., Ourivski, A., Pavlouchkov, V.: On the modified Niederreiter cryptosystem. Proceedings of the Information Theory and Networking Workshop (ITW), Metsovo, Greece (1999) [3] Overbeck, R.: A new structural attack for GPT and variants. In: Mycrypt Number 3715 in Lecture Notes in Computer Science, Springer-Verlag (2005) [4] Wieschebrink, C.: An attack on a modified Niederreiter encryption scheme. In: PKC Number 3958 in Lecture Notes in Computer Science, Springer-Verlag (2006) 11

59 Kryptographische Verfahren zur elektronischen Signatur Franz Scheerer Scheerer Software Fritz-Philippi-Straße Wiesbaden Die elektronische Signatur kann als das elektronische Äquivalent zur eigenhändigen Unterschrift angesehen werden. Verfahren zur elektronischen Signatur beruhen auf mathematischen Algorithmen und eventuell spezieller Hardware wie Chipkarten. Die Algorithmen können grob in symmetrische und asymmetrische [DH76] unterteilt werden. Die Vor- und Nachteile dieser Verfahren werden insbesondere unter dem Gesichtspunkt der langfristigen Sicherheit und der praktischen Anwendung diskutiert. Symmetrische Verfahren können auf Basis eines Message Authentication Code, kurz als MAC bezeichnet, realisiert werden. Derartige Algorithmen können auf Basis einer Hashfunktion oder einer Blockchiffre wie dem DES oder AES realisiert werden. Die Prüfung der Signatur erfolgt dabei mit dem gleichen Schlüssel wie die Signaturerstellung. Für die Wahl des Schlüssels gibt es je nach Anforderungen an das Verfahren mehrere Möglichkeiten. Ein öffentlicher Schlüssel ermöglicht es Übertragungsfehler aufzudecken. Er kann jedoch, zumindest bei symmetrischen Verfahren, eine Fälschung durch Dritte nicht ausschließen. Falls sich die Vertragsparteien gegenseitig vertrauen, können sie einen geheimen Schlüssel vereinbaren. Ein Dienstleister kann mit den Vertragsparteien jeweils einen geheimen Schlüssel vereinbaren. Die Vertragsparteien können dann die mit einem MAC berechnete Signatur verschlüsselt beim Dienstleister einreichen. Die Signatur ist damit verbindlich und nicht ableugbar. Die langfristige Sicherheit des asymmetrischen Verfahrens RSA ist fraglich. Ein hier vorgestelltes, auf dem Quadratischen Sieb [CP96] basierendes Verfahren, arbeitet mit einer dynamischen Faktorbasis. Diese wird mit einem effizienten Algorithmus in einem ersten Siebschritt erstellt. Das Verfahren beruht wie das Quadratische Sieb auf der Primfaktorzerlegung von Relationen ( k, Q(k) = k 2 n ) der kleinsten Zahlen k größer als n, wobei n die zu faktorisierende Zahl ist. In dem hier vorgestellten Verfahren werden nur Faktoren selektiert, die in der Primfaktorzerlegung der Q(k) auftreten. Das Vorgehen entspricht der Pollard-Rho-Methode [JP75] und nutzt den Euklidischen Algorithmus zur Berechnung des größten gemeinsamen Teilers von Q( n + m) und Q( n + 2m), wobei m kleine natürliche Zahlen durchläuft. Es wird dabei ausgenutzt, dass sich die Primfaktoren in den Relationen zyklisch wiederholen. Nach diesem ersten Siebschritt können die Primfaktorzerlegungen der Q(k) effzient bestimmt werden, die sich vollständig durch Faktoren der Faktorbasis und Quadratzahlen darstellen lassen. Literatur [CP96] Carl Pommerance. A Tale of Two Sieves Notices of the AMS (43): , Dezember 1996 [DH76] Whitfield Diffie and Martin E. Hellman. New directions in cryptography. IEEE Transactions on Information Theory, IT-22(6): , November [JP75] J.M. Pollard. A Monte Carlo Method for Factorization. BIT 15: ,

60 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie Ėr ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 1. Kryptowochenende am 1.&2. Juli 2006 (Einreichung/Anmeldung: 26. Mai 2006). Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 5. Kryptotag am 11. September 2006 (Einreichung: 11. August 2006, Anmeldung: 7. September 2006). Universität Kassel, Arbeitsgruppe Theoretische Informatik. Kontakt: Heiko Stamer. 6. Kryptotag im Februar Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. Bisherige Kryptotage 3. Kryptotag am 15. September Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldeten Teilnehmer. 2. Kryptotag am 31. Mrz Universität Ulm, Abteilung für Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldeten Teilnehmer. 1. Kryptotag am 1. Dezember Universität Mannheim, Lehrstuhl für Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldeten Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (Ecole Normale Superieure, Paris) verantwortlich für die Organisation der Kryptotage. Für evtl. Rückfragen bitte an sie wenden.

61 Universität Mannheim, Reihe Informatik, TR-2006-xxx Kryptowochenende 2006 Workshop über Kryptographie Universität Mannheim Frederik Armknecht NEC Europe Ltd. Dirk Stegemann Universität Mannheim Juli 2006

62 Inhaltsverzeichnis Public-Key-Kryptographie mit Halbgruppen-Aktionen und Halbringen Jens Zumbrägel Äquivalente Schlüssel in Multivariate Quadratic Public Key Systemen Aktueller Stand Christopher Wolf Effiziente Bestimmung der Algebraischen Immunität Simon Künzli Konstruktion von booleschen Funktionen mit maximaler algebraischer Immunität Hellen Altendorf Erste Erfahrungen zu meinem Post-doc-Leben in der Industrie Frederik Armknecht Analyse der Entwicklung von Malware Oliver Schmid Jointly Generating Random Keys for the Fully Distributed Environment Sebastian Faust and Stefan Lucks Theorie und Anwendungen von Tree Parity Machines für die Kryptographie Andreas Ruttor und Markus Volkmer Opportunistische -Sicherheit Alexander Naumann und Tobias Straub Sicherheitsbeweise für zertifikatlose Public-Key Schemata Ewan Fleischmann Universelle Message Authentication Codes Christian Forler

63 Privacy Friendly Location Based Service Protocols using Efficient Oblivious Transfer Markulf Kohlweiss and Bartek Gedrojc Google Reveals Cryptographic Secrets Emin Islam Tatlı Trusted Computing mit Open Source Software Heiko Stamer

64 Public-Key-Kryptographie mit Halbgruppen-Aktionen und Halbringen Jens Zumbrägel Institut für Mathematik, Universität Zürich Winterthurerstr. 190, CH Zürich jzumbr@math.unizh.ch Das klassische Diffie-Hellman-Protokoll zum Schlüsselaustausch kann im Kontext von Halbgruppen- Aktionen verallgemeinert werden [MMR05]: Öffentlich gegeben sei hierzu eine Halbgruppen-Aktion, d.h. eine (effizient berechenbare) Abbildung G S S, (g, s) g.s einer endlichen abelschen Halbgruppe (G, ) auf eine endliche Menge S, so dass die Identität g.(h.s) = (g h).s für alle g, h G und s S erfüllt ist. Desweiteren soll für s S die Bahnen- Abbildung G G.s S, g g.s eine Einweg-Funktion sein, so dass also die Schwierigkeit des folgenden Problems besteht: gegeben s S und t G.s, finde ein g G mit t = g.s. Wir haben dann die folgende Abwandlung des Diffie-Hellman-Protokolls: Alice öffentlich Bob wähle s S wähle a G a.s S b.s S wähle b G berechne a.(b.s) berechne b.(a.s) Alice und Bob besitzen somit den gemeinsamen Schlüssel (a b).s. Besitzt S zusätzlich die Struktur einer Gruppe, so ergibt sich außerdem eine Verallgemeinerung des ElGamal-Protokolls zur Chiffrierung. Im klassischen Fall haben wir hierbei als Aktion die Potenz-Abbildung Z H H, (n, a) a n der ganzen Zahlen (Z, ) auf eine Gruppe (H, ), wobei H z.b. die multiplikative Gruppe F eines endlichen Körpers F oder die Gruppe E(F) der F-rationalen Punkte einer elliptischen Kurve E ist. Aus verschiedenen Gründen ist es interessant, verallgemeinerte Halbgruppen-Aktionen zu konstruieren und zu analysieren. Als wichtige Bausteine für die Konstruktion von Halbgruppen- Aktionen erweisen sich sogenannte Halbringe. Unter einem Halbring versteht man einen verallgemeinerten Ring in denen keine negativen Elemente zu existieren brauchen, d.h. (R, +, ) ist Halbring (R, +) ist abelsche Halbgruppe, (R, ) ist Halbgruppe, Distributivgesetze gelten. Elemente 0, 1 R mit 0 + x = x, 0x = 0 = x0 bzw. 1x = x = x1 für alle x R werden Null bzw. Eins genannt. Beispiele für Halbringe mit Null und Eins sind die natürlichen Zahlen {0, 1, 2,... } mit gewöhnlicher Addition und Multiplikation, sowie der boolesche Halbring (S, or, and), wobei S := {0, 1} und or bzw. and die entsprechenden logischen Operationen sind. 4

65 Um eine Pohlig-Hellman-artige Reduktion auf kleinere Instanzen zu vermeiden, ist es außerdem wichtig, dass der Halbring R im folgenden Sinne einfach bzw. kongruenz-frei ist: Jede Äquivalenzrelation auf R, die { a + c b + d a b, c d a c b d für alle a, b, c, d R erfüllt, ist trivial, d.h. = R R oder = id R. Sicherlich sind alle Halbringe der Größe 2, wie der boolsche Halbring, einfach. Die Klassifikation endlicher, kongruenz-freier Halbringe ist ein interessantes Problem, welches weiterhin offen ist (siehe [Mo04] für eine Teilklassifikation). Jedoch lässt sich eine unendliche Familie solcher Halbringe angeben, denn für jeden einfachen Halbring R mit Null und Eins ist der Matrix-Halbring Mat n n (R) ebenfalls einfach. Eine computer-unterstützte Suche meinerseits nach einfachen Halbringen brachte folgendes Ergebnis: Unter allen Halbringen R mit Null, die keine Ringe sind, und eine Größe #R {3,..., 12} besitzen, gibt es bis auf Isomorphie nur einen einzigen einfachen Halbring. Dieser hat 6 Elemente R = {0, 1, a, b, c, d} und die folgenden Operationstafeln: + 0 a b c 1 d 0 0 a b c 1 d a a a b c 1 d b b b b 1 1 d c c c 1 c 1 d d d d d d d d d 0 a b c 1 d a a a b b 0 a b a b b c c c d 1 0 a b c 1 d d 0 c d c d d Es gibt also Indizien dafür, dass einfache Halbringe mit Null überaus selten sind. Hat man nun einen einfachen Halbring R und zusätzlich einen R-Halbmodul M gegeben, so ergibt sich eine Halbgruppen-Aktion aus der Matrix-Multiplikation Mat n n (R) M n M n ; hierbei wird dann eine abelsche Unter-Halbgruppe G Mat n n (R) gewählt. In diesem Kontext ergibt sich das klasssiche Diffie-Hellman-System als Spezialfall (n = 1). Ich werde Angriffsmöglichkeiten auf solche Systeme diskutieren, sowie eine weitere Halbgruppen-Aktion vorstellen, die Halbringe verwendet. Literatur [MMR05] Gérard Maze, Chris Monico, Joachim Rosenthal. Public Key Cryptography based on Semigroup Actions Preprint, arxiv:cs.cr/ v2 28. Januar 2005 [Mo04] C. Monico. On finite congruence-simple semirings Journal of Algebra 271(2), ,

66 Äquivalente Schlüssel in Multivariate Quadratic Public Key Systemen Aktueller Stand Christopher Wolf École Normale Supérieure, Département d Informatique 45 rue d Ulm, F Paris Cedex 05, France Christopher.Wolf@ens.fr or chris@christopher-wolf.de 1 Initial Considerations In the last 20 years, several schemes based on the problem of Multivariate Quadratic equations (or MQ for short) have been proposed. The most important ones certainly are MIA / C and Hidden Field Equations (HFE) plus their variations MIA- / C, HFE-, HFEv, and HFEv-. Both classes have been used to construct signature schemes for the European cryptography project NESSIE, namely the MIA- variation in Sflash, the HFEv- variation in Quartz and the HFE- variation in the tweaked version Quartz-7m. Unbalanced Oil and Vinegar schemes and Stepwise Triangular Schemes are also important in practice. While the first is secure with the correct choice of parameters, the second forms the basis of nested constructions like the enhanced TTM, Tractable Rational Maps, or Rainbow. An overview of all these systems can be found in the taxonomy article [WPc]. In this talk, we give an overview on the question of equivalent keys of MQ-schemes. At first glance, this question seems to be purely theoretical. But for practical applications, we need memory and time efficient instances of Multivariate Quadratic public key systems. One important point in this context is the overall size of the private key: in restricted environments such as smart cards, we want it as small as possible. Hence, if we can show that a given private key is only a representative of a much larger class of equivalent private keys, it makes sense to compute (and store) only a normal form of this key. Similar, we should construct new Multivariate Quadratic schemes such that they do not have a large number of equivalent private keys but only a small number, preferably only one per equivalence class. This way, we make optimal use of the randomness in the private key space and neither waste computation time nor storage space without any security benefit. All systems based on MQ-equations use a public key of the form p i (x 1,..., x n ) := 1 j k n γ i,j,k x j x k + n β i,j x j + α i, with n Z + variables and m Z + equations. Moreover, we have 1 i m; 1 j k n and α i, β i,j, γ i,j,k F (constant, linear, and quadratic terms). We write the set of all such systems of polynomials as MQ(F n, F m ). Moreover, the private key consists of the triple (S, P, T ) where S Aff 1 (F n ), T Aff 1 (F m ) are bijective affine transformations. Moreover, we have P MQ(F n, F m ) is a polynomial-vector P := (p 1,..., p m) with m components; each component is a polynomial in n variables x 1,..., x n. Throughout this paper, we will denote components of this private vector P by a prime. In contrast to the public polynomial vector P MQ(F n, F m ), the private polynomial vector P does allow an efficient computation of x 1,..., x n for given y 1,..., y m. Still, the goal of MQ-schemes is that this inversion should be hard if the public key P alone is given. The main difference between MQ-schemes lies in their special construction of the central j=1 6

67 equations P and consequently the trapdoor they embed into a specific class of MQ-problems. An introduction to Multivariate Quadratic public key systems is given in [WPc]. This talk is based on the two conference papers [WPa,WPb], which deal with the classes MIA, HFE, and UOV. An extended version which also includes STS and shows that the reduction for MIA/MIA for q 2 is tight is [WPd]. 2 Mathematical Considerations Before discussing concrete schemes, we start with some general observations and definitions. Obviously, the most important term in this article is equivalent private keys. We give a graphical input x x = (x 1,..., x n ) private: S σ 1 S x private: P τ P σ y private: T T τ 1 output y public: (p 1,..., p n ) Figure 1: Equivalent private keys using affine transformations σ, τ representation of this idea in Figure 1. We can also express this idea in the following definition: Definition 2.1 We call two private keys (S, P, T ), ( S, P, T ) Aff 1 (F n ) MQ(F n, F m ) Aff 1 (F m ) equivalent if they lead to the same public key, i.e., if we have T P S = P = T P S. In the above definition, Aff 1 ( ) denotes the class of bijective affine transformations. In order to find equivalent keys, we consider the following transformations: Definition 2.2 Let (S, P, T ) Aff 1 (F n ) MQ(F n, F m ) Aff 1 (F m ), and consider the four transformations σ, σ 1 Aff 1 (F n ) and τ, τ 1 Aff 1 (F m ). Moreover, let P = T τ 1 τ P σ σ 1 S. (1) We call the pair (σ, τ) Aff 1 (F n ) Aff 1 (F m ) sustaining transformations for an MQ-system if the shape of P is invariant under the transformations σ and τ. For short, we write (σ, τ) (S, P, T ) for (2.2) and (σ, τ) sustaining transformations. This idea has already been outlined in Figure 1. 7

68 3 Sustaining Transformations We have several sustainers which can be used with different multivariate quadratic public key systems. Additive Sustainer: Add a constant A E or a F n, b F m. Big Sustainer: Multiply with a non-zero constant B E. Small Sustainer: Multiply with a diagonal matrix with non-zero coefficients b 1,..., b n, b 1,..., b m F, respectively. Permutation Sustainer: Permute the input variables / the equations. Gauss Sustainer: Perform Gauss operations. Frobenius Sustainer: Perform the operation X X qi for 1 i n and i N. Reduction Sustainer: Observe that the last r rows have no effect with r N being the number of equations missing. These sustainers can now be combined with different multivariate quadratic public key systems. We summarise their effects in the next section. 4 Results The sustainers outlined above can be applied to several basic classes, such as Hidden Field Equations (HFE), Matsumoto-Imai Scheme A (MIA), Unbalanced Oil and Vinegar schemes (UOV), and Scheme UOV STS Table 1: Summary of the reduction results of this article Reduction q n+mn n m 1 i=0 (q n m q i ) m 1 i=0 (qm q i q m+n L i=1 (q n i(n P i j=1 n j) ) n i 1 j=0 (qn i q j ) L i=1 (q m i(n P i j=1 m j) ) m i 1 j=0 (qm i q j ) MIA n(q n 1) MIA- n(q n 1)q r n 1 i=n r 1 (qn q i ) HFE nq 2n (q n 1) 2 HFE- HFEv HFEv- nq 2n (q n 1)(q n r 1) n 1 i=n r 1 (qn q i ) n q n+n +vm (q n 1) 2 v 1 i=0 (qv q i ) n q r+2n vn (q n 1) 2 v 1 i=0 (qv q i ) n 1 i=n r 1 (qn q i ) Stepwise-Triangular Systems (STS). We have summarised our results in tables 1 and 2. The first gives an overview on the formulae achieved while the latter features some numerical examples. The 8

69 symbols used in Table 1 are defined as follows: n Z + denotes the number of variables, m Z + is the number of equations, q := F is the number of elements in the ground field F, L the number of layers for STS, and n l, m l for 1 l L the number of new variables and equations, respectively. Table 2: Numerical examples for the reduction results of this article Scheme Parameters Choices for S, T Reduction (in log 2 ) (in log 2 ) UOV q = 2, m = 64, n = ,054 32,956 q = 2, m = 64, n = ,790 57,596 STS q = 2, r = 4, L = 25, n = ,096 11,315 q = 2, r = 5, L = 20, n = ,096 11,630 HFE q = 2, n = 80 12, HFE- q = 2, r = 7, n = , HFEv q = 2, v = 7, n = , HFEv- q = 2, r = 3, v = 4, n = , MIA q = 128, n = 67 63, MIA- q = 128, r = 11, n = 67 63, We see applications of our results in different contexts. First, they can be used for memory efficient implementations of the above schemes: instead of saving the whole private key, we can only save a normal form. Second, they apply to cryptanalysis as they allow to concentrate on special forms of the private key. Third, constructors of new schemes should keep these sustaining transformations in mind: there is no point in having a large private key space if it can be reduced immediately by an attacker who can just apply some sustainers. Moreover, the results obtained in this talk shine new light on cryptanalytic results, in particular key recovery attacks: as each private key is only a representative of a larger class of equivalent private keys, each key recovery attack can only recover it up to these equivalences as the public key P cannot contain information about individual private keys but the equivalence class used to construct P. References [WPa] Christopher Wolf and Bart Preneel. Superfluous keys in Multivariate Quadratic asymmetric systems. In Public Key Cryptography PKC 2005, volume 3386 of Lecture Notes in Computer Science. pages Serge Vaudenay, editor, Springer, Extended version http: //eprint.iacr.org/2004/361/. [WPb] Christopher Wolf and Bart Preneel. Equivalent keys in HFE, C, and variations. In Proceedings of Mycrypt 2005, volume 3715 of Lecture Notes in Computer Science, pages Serge Vaudenay, editor, Springer, Extended version 15 pages. [WPc] Christopher Wolf and Bart Preneel. Taxonomy of public key schemes based on the problem of multivariate quadratic equations. Cryptology eprint Archive, Report 2005/077, 12 th of May pages. [WPd] Christopher Wolf and Bart Preneel. Equivalent Keys in Multivariate Quadratic Public Key Systems. Cryptology eprint Archive, Report 2005/464, 22 nd of December /464/, 19 pages. 9

70 1 Einleitung Effiziente Bestimmung der Algebraischen Immunität Simon Künzli Fachhochschule Nordwestschweiz CH-5210 Windisch (Schweiz) Stromchiffren sind grundlegende kryptografische Algorithmen, um die Vertraulichkeit der elektronisch übermittelten Daten zu gewährleisten. Verglichen mit anderen Verfahren sind Stromchiffren sehr schnell, und eine Implementierung ist oft mit sehr wenig Hardware möglich. Daher sind Stromchiffren besonders geeignet in einer Umgebung mit wenig Ressourcen, etwa in kabellosen mobilen Netzwerken. Wir betrachten eine Klasse von Stromchiffren, die auf einem linearen Schieberegister (LFSR) und einer Boolschen Filterfunktion basieren. Solche Verfahren haben einen geheimen N-bit Zustand x t zur Zeit t. Der Zustand wird mit einem geheimen Schlüssel K initialisiert und mit einem LFSR L aktualisiert entsprechend x t+1 = L(x t ). Eine nichtlineare boolesche Filterfunktion f mit n Variablen und algebraischem Grad k wird auf den Zustand angewendet, um ein Schlüsselstrom-bit z t = f(x t ) zu erzeugen. Ein Klartext-bit p t wird dann zum Chiffretext-bit c t = p t z t. 2 Angriffe Bei einem Angriff auf LFSR-basierte Stromchiffren wird angenommen, dass L und f bekannt sind, dass der Angreifer viele Schlüsselstrom-bits z t kennt (Angriff mit bekanntem Klartext), und dass sein Ziel ist die Rekonstruktion von x t (und damit K) ist. Mit dem bekannten Algorithmus von Berlekamp und Massey kann der Schlüsselstrom synthetisiert werden, und dies mit einer Komplexität von etwa ( ) N 2. k Inzwischen sind aber viele spezifische Verfahren beweisbar resistent gegen diese Berlekamp-Massey-Synthese. In einem anderen Ansatz kann der Angreifer ein multivariates, nichtlineares und überbestimmtes Gleichungssystem z t = f(x t ) für verschiedene Zeiten t aufstellen. Das System kann etwa durch Linearisierung gelöst werden, wobei jeder nichtlineare Term durch eine neue Variable ersetzt wird. Schließlich wird das lineare Gleichungssystem mit bekannten Methoden gelöst (etwa durch Gauss sche Elimination). Abhängig vom algebraischen Grad der Gleichungen werden durch Linearisierung viele zusätzliche Variablen eingeführt. Dabei ist der Grad der Gleichungen durch den Grad von f bestimmt, und der Aufwand für das Lösungsverfahren beträgt etwa ( ) N 3. k Das ist natürlich nicht effizient, aber ein geschickter Angreifer kann Gleichungen mit reduziertem Grad aufstellen. Solche leistungsfähigen Angriffe sind seit wenigen Jahren bekannt als algebraische Angriffe. 3 Algebraische Angriffe Wie ist es möglich, den Grad der Gleichungen zu reduzieren? Eine Idee ist es, eine geeignete Funktion g von tiefem Grad d zu suchen, so dass f g = 0 in GF(2). Eine solche Funktion g nennt man Annihilator von f. Die Gleichung z t = f(x t ) kann mit dem Annihilator g(x t ) multipliziert 10

71 werden, und man erhält für z t = 1 die Gleichung g(x t ) = 0 von Grad d. Ähnliche Gleichungen können mit einem Annihilator von f + 1 aufgestellt werden. Daher definiert man die algebraische Immunität einer Funktion f als minimalen Wert d, so dass f g = 0 oder (f + 1) g = 0 für eine Funktion g vom Grad d. Es ist bekannt, dass die algebraische Immunität einer Funktion f mit n Variablen höchstens n/2 sein kann [CM03]. Mit bisherigen Algorithmen kann die algebraische Immunität einer beliebigen Funktion in O(D 3 ) bestimmt werden, wobei D ( n d). Mit einem neuen Algorithmus, der auf multivariater Polynominterpolation beruht, können wir die algebraische Immunität in nur O(D 2 ) bestimmen [ACGKMR06]. 4 Schnelle Algebraische Angriffe Als Reaktion auf diese Angriffe sind Klassen von Filterfunktionen konstruiert worden, die große (oder sogar maximale) algebraische Immunität aufweisen (sowie weitere kryptografisch wünschenswerten Eigenschaften). Allerdings haben sich auch die Angriffe weiter entwickelt und sind bekannt als schnelle algebraische Angriffe [C03]. Dabei werden Funktionen g und h von tiefem Grad gesucht, so dass gilt f g = h. Mit einem Aufwand, der von d := deg h abhängig ist, kann man Terme von Grad größer als e := deg g eliminieren, und Gleichungen von Grad e aufstellen. Dabei ist d mindestens so groß wie die algebraische Immunität von f, und e kann viel kleiner sein. Es war eine offene Frage, ob Funktionen f mit guter (konventioneller) algebraischer Immunität auch gegenüber diesen neuen Angriffen resistent sind. Bekannte Algorithmen können so angepasst werden, dass die Immunität einer Funktion gegen schnelle algebraische Angriffe in O(D 3 ) bestimmt werden kann. Wir haben einen neuen, effizienten (und theoretisch fundierten) Algorithmus konstruiert, der diese Aufgabe für beliebige Funktionen in O(DE 2 ) durchführt, wobei E ( n e). Der Algorithmus basiert darauf, dass Gleichungen für g und h separiert werden können. Dann haben wir symmetrische Funktionen untersucht, da diese von speziellem Interesse für Hardware- Implementierungen sind. Die symmetrische Struktur konnte in einem noch effizienteren Algorithmus ausgeschöpft werden. Nebst diesen allgemeinen Algorithmen zur effizienten Bestimmung der algebraischen Immunität einer Funktion haben wir auch spezifische Funktionen theoretisch und experimentell untersucht. Die Resultate lassen uns schließen, dass viele Klassen von Funktionen sehr verwundbar sind gegenüber schnellen algebraischen Angriffen, und dies trotz ihrer optimalen (konventionellen) algebraischen Immunität [ACGKMR06]. Diese Arbeit wurde zusammen mit den Universitäten Mannheim und Limoges und mit INRIA durchgeführt. Literatur [CM03] N. Courtois, and W. Meier. Algebraic Attacks on Stream Ciphers with Linear Feedback. In Advances in Cryptology - EUROCRYPT 2003, LNCS 2656, pages Springer Verlag, [C03] N. Courtois. Fast Algebraic Attacks on Stream Ciphers with Linear Feedback. In Advances in Cryptology - CRYPTO 2003, LNCS 2729, pages Springer Verlag, [ACGKMR06] F. Armknecht, C. Carlet, P. Gaborit, S. Künzli, W. Meier, and O. Ruatta. Efficient Computation of Algebraic Immunity for Algebraic and Fast Algebraic Attacks. In Advances in Cryptology - EUROCRYPT 2006, LNCS 4004, pages Springer Verlag,

72 Konstruktion von booleschen Funktionen mit maximaler algebraischer Immunität Hellen Altendorf Universität Mannheim Fakultät für Mathematik und Informatik Lehrstuhl für Theoretische Informatik Das Ziel dieser Arbeit ist es boolesche Funktionen f : {0, 1} n {0, 1} m zu konstruieren, deren Graph gr(f) = {(x, f(x)) x {0, 1} n } {0, 1} n+m maximale algebraische Immunität besitzt. Wir sagen die Funktion p : {0, 1} n {0, 1} annuliert eine Teilmenge S {0, 1} n (oder p ist Annihilator der Menge S), falls gilt p(x) = 0 x S. Die Algebraische Immunität AI(S) einer Menge S ist definiert als das kleinstes d N für das ein nichttrivialer Annihilator p vom Grad d existiert. Die konstruierten Funktionen maximaler Immunität sollen als Grundbausteine für symmetrische Blockchiffren dienen und so algebraische Angriffe auf das Kryptosystem erschweren. Algebraische Angriffe auf Secret-Key Kryptosysteme bestehen daraus, nichttrivialen Annihilatoren von kleinem Grad für die Beziehung zwischen Input- und Outputbits zu finden und so ein Gleichungssystem von möglichst kleinem Grad aufzustellen und effizient zu lösen. Spätestens seit dem hypothetischen Angriff auf den Advanced Encryption Standard (AES) von Courtois und Pieprzyk (2002) verlangt die Kryptographie nach besseren algebraischen Funktionen. Dieser Angriff ist zurückzuführen auf die geringe algebraischer Immunität der S-Boxen. Die Gleichungen sind hier quadratisch, obwohl die Formate der S-Boxen ({0, 1} 8 {0, 1} 8 ) eine Immunität von 3 gewähren könnten. Konkrete Angriffe wurden erstmals 2003 von Courtois und Meier für eine spezielle Klasse von Schlüsselstromgeneratoren beschrieben. Dieser Angriff wurde von Armknecht und Krause (2003) auf eine allgemeinere Klasse erweitert, welche den E 0 -Generator beinhaltet, der im Bluetooth Standard verwendet wird. In dem Vortrag wird erläutert, wie man die maximale Immunität einer Funktion f : {0, 1} n {0, 1} m bestimmt und wie man letztendlich mithilfe der Matroidtheorie Funktionen von maximaler Immunität findet. Literatur [AK] F. Armknecht and M. Krause. Constructing Single- and Multi-Output Boolean Functions with Maximal Algebraic Immunity, Proceedings of ICALP

73 Erste Erfahrungen zu meinem Post-doc-Leben in der Industrie Frederik Armknecht Network Laboratories NEC Europe Ltd. Wie bei so vielen vor mir drängte sich im Laufe meiner Promotion die Frage auf, wie es danach weitergehen sollte. Klar war nur, dass ich immer viel Spass am Forschen hatte und diesem auch weiterhin nachgehen wollte. Dies impliziert normalerweise einen Berufsweg im akademischen Bereich, was auch lange Zeit meine Planung war. Dennoch bin ich schlussendlich in die Industrie gegangen, und zwar zum Forschunglabor von NEC in Heidelberg. In meinem Vortrag möchte ich kurz darlegen, wie und warum ich meine Entscheidung gefällt habe, und wie meine bisherigen Eindrücke bezüglich Industrieforschung sind. Damit möchte ich allen Mut machen, auch alternative Berufswege in Betracht zu ziehen. 13

74 Analyse der Entwicklung von Malware Oliver Schmid Technische Universität Darmstadt Als Fred Cohen im Jahr 1983 ein sich selbst reproduzierendes Programm namens VD vorstellte, wurde zum ersten Mal von einem Computervirus gesprochen. In den ersten Jahren entwickelte sich dieses Phänomen noch recht langsam und wurde erst 1987 zu einem echten Problem für die Computerwelt, als mit Lehigh der erste Virus gespeicherte Daten löschte. In den kommenden Jahren konnten die Programmierer mit nur wenigen innovativen Ideen aufwarten, und auch quantitativ hielt sich das Problem in Grenzen, so waren 1991 lediglich 300 verschiedene Exemplare bekannt. Makroviren, die sich wie Concept durch den Befall von Word-Dateien vermehrten, gewannen ab 1995 an Bedeutung, besonders als sie drei Jahre später erstmals wie Tristate beliebige Office- Dokumente infizieren konnten. Das Internetzeitalter begann 1997, als sich die Schadprogramme erstmals über , FTP und IRC replizierten und sich dadurch erheblich schneller verbreiten konnten, als es bisher durch die Infizierung von Disketten möglich war. Die ersten Epidemien verursachten 1999 der Makrovirus Melissa und im Folgejahr der VBS-Wurm LoveLetter. In den vergangenen Jahren haben sich Viren und Würmer nun immer schneller entwickelt und zeichnen sich dabei durch immer neue Eigenschaften aus. So haben sie beispielsweise ihre Tarnung verbessert, indem sie ihre Prozessnamen denen von Systemprogrammen und Antiviren-Software angleichen, außerdem sind sie in der Lage, Antiviren-Programme und Personal Firewalls zu deaktivieren sowie den Zugriff auf Webseiten für Sicherheitsupdates zu unterbinden. Die Entwicklung des Phänomens Malware wird nun in zwei Analysen einer näheren Betrachtung unterzogen. Die Daten, welche den beiden Analysen zu Grunde liegen, wurden durch Auswertung der Virendatenbanken zweier Hersteller von Antiviren-Software gewonnen [SoA][SyA]. Dabei wurden nur die Schadprogramme in den Datensatz aufgenommen, die in einem der Monate des Beobachtungszeitraumes seit Januar 1998 zu den Top-Ten der gemeldeten Malware gehörten [SoT][SyT]. Die unterschiedlichen Eigenschaften der Viren und Würmer, wie beispielsweise Verbreitungsweg, Schadwirkung oder Methoden zur Tarnung, wurden dabei in 41 Variablen erfasst, ebenso wie die relative Auftrittshäufigkeit der erfassten Exemplare am Gesamtaufkommen an Malware. Zunächst wurde eine Clusteranalyse durchgeführt, um festzustellen, wie schnell sich die Entwicklung der Malware vollzogen hat. Hierfür wurden für jedes Jahr die Eigenschaften der Schadprogramme nach ihrer relativen Auftrittshäufigkeit gewichtet und aufsummiert. Je kleiner nun der Abstand zweier Jahre in der Distanzmatrix ist, desto ähnlicher sind sich die in diesen beiden Jahren aufgetretenen Viren und Würmer. Das Ergebnis zeigt, dass sich zwischen den Jahren 1998 und 1999, in denen Viren und Makroviren das Geschehen beherrschten, wenig veränderte. Es gab nur wenige neue Entwicklungen, wie beispielsweise den Makrovirus Melissa, der sich selbst über MS Outlook verschickte und damit eine neue Ära der Verbreitung einläutete. In den folgenden Jahren wurden die Schadprogramme immer vielseitiger, ihren größten Entwicklungssprung erlebten sie im Jahr 2001, als die Makroviren in die Bedeutungslosigkeit fielen und die -Würmer gleich drei entscheidende Neuerungen vorweisen konnten: Erstmals fälschten sie ihren Absender und verbreiteten sich mittels einer eigenen SMTP-Engine an Adressen, welche sie auf der Festplatte ihres Opfers finden konnten. 14

75 Auch setzten sich in diesem Jahr zwei Entwicklungen des Vorjahres erst richtig durch, und zwar die Änderung von Registry-Einträgen durch die Malware und die Einrichtung von so genannten Backdoors, welche den Remote-Zugriff auf einen infizierten Rechner ermöglichen. Seit dem Jahr 2004 hingegen hat sich nicht mehr viel getan, zwar steigt die Zahl der Virendefinitionen weiterhin unbeirrt auf inzwischen über , es gab jedoch seither keine wirklichen Innovationen. Im Anschluss erfolgte eine Faktorenanalyse, um zusammenhängende Variablen auf wenige Faktoren zu reduzieren, in diesem Fall konnten die 41 Variablen durch 7 Faktoren erklärt werden. Der Faktor mit dem größten Erklärungsgehalt fasst dabei die typischen Eigenschaften moderner - Würmer zusammen, welche sich mit einer eigenen SMTP-Engine an Adressen verbreiten, die sie auf der Festplatte ihres Opfers finden und die Sprache ihres Mailtextes anhand des Domain-Kürzels des Empfängers auswählen. Diese Würmer zeichnen sich weiterhin dadurch aus, dass sie Einträge in der Registry ändern, eine Backdoor zum infizierten Rechner öffnen, und in ihrer Auftrittshäufigkeit über die Zeit hinweg kontinuierlich zugenommen haben. Der zweite Faktor bildet die Gruppe der Trojaner ab, die gezielt Sicherheitslücken der Betriebssysteme von Microsoft ausnutzen, um auf den Rechner zu gelangen und Daten wie Kreditkartennummern oder Passwörter auszuspionieren. Schadprogramme, deren vornehmliches Ziel die Manipulation von Daten ist, werden durch den dritten Faktor erklärt, weitere Faktoren sind beispielsweise die Hardwareorientierung älterer Exemplare, die einen Interrupt im BIOS nutzen und sich im MBR der Festplatte einnisten, oder die unauffällige Verbreitung über Filesharing- Verzeichnisse und durch doppelte Dateiendungen. Aus den Analysen ergibt sich eine deutliche Verbindung zwischen dem zeitlichen Verlauf und bestimmten Malware-Typen, so waren Viren und Makroviren bis 2000 stark vertreten, und wurden in den folgenden zwei Jahren von den Würmern abgelöst, die seit 2002 das Geschehen dominieren. Des Weiteren hat sich eine Entwicklung von der Vernichtung von Daten hin zu Datendiebstahl und dem Missbrauch der befallenen PCs als so genannte Bots gezeigt. Unmittelbares Phishing durch Malware ist jedoch entgegen verschiedener Pressemeldungen nicht verbreitet, vielmehr tritt eine Verbindung zwischen Malware und Phishing meist im Zusammenhang mit Spambots auf, die neben Werbeinhalten natürlich auch Phishing-Mails verschicken können. Literatur [SoA] [SyA] [SoT] [SyT]

76 Jointly Generating Random Keys for the Fully Distributed Environment Sebastian Faust and Stefan Lucks K.U.Leuven ESAT-COSIC Universität Mannheim Kasteelpark Arenberg 10, Lehrstuhl fuer Theoretische Informatik B-3001 Leuven-Heverlee, Belgium Mannheim, Germany Abstract In this paper we introduce a new efficient method to jointly generate and share k random secret keys for discrete log based cryptosystems in a fully distributed environment between a group of parties P = {P 1,..., P n }. We call such a scheme a k joint random key generation (k-jrkg) protocol. Compared with the well-known technique of distributed key generation, where the shared key is not known by any one party, the intention of a JRKG protocol is slightly different: every random key is known and shared by only one party. Here, our protocol guarantees the randomness of the keys under the DDH assumption. In particular, this applies to the keys of the corrupted parties. Hence, they do not have a chance to bias their keys to a non-uniform distribution. Our protocol reduces the dominating factor for the communication complexity, the number of reliable broadcasts, by a factor of n compared with other approaches to this problem. The security of our protocol can be proven for less than ( n 2 )-corrupted parties under the DL-assumption in the random oracle model. 1 Introduction In the current literature, protocols for jointly generating random keys are frequently used as building blocks in various protocols designed for a fully distributed environment like the internet [GJ04, GJKR96]. In particular the well-analyzed technique of distributed key generation [GJKR99, Ped91] has a wide area of application and allows often in the first place to formally prove the protocols security. Unfortunately, most distributed key generation protocols do suffer from high communication and computation complexity, hence limiting their useability in practice to small and static networks. A first approach to overcome these drawbacks was presented by John Canny and Stephen Sorkin in 2004 [CS04]. Their idea relies on the fragmentation of the set of parties P in a network to reduce the size of the broadcast groups. However, this technique is only probabilistic, i.e., has a failure probability, needs a dealer to build up and manage the broadcast groups and finally requires that honest and corrupted parties are randomly distributed in P. In our work we take a different stance. Rather than trying to develop an efficient generally applicable solution for doing distributed key generation, we focus on specific variants of DKG to reduce their communication complexity. Such a specific variant is presented in this paper by the k-jrkg protocol as an efficient solution for the generation of k random secrets, where each secret is known by only one party. This technique is applicable, for example, in the setup phase of the protocols described by Golle and Juels in [GJ04]. In particular, it decreases their communication costs compared to a trivial solution and altogether provides a more natural approach to fulfill the needed requirements. At first glance the generation of k 2 uniformly distributed secrets x 1,..., x k and corresponding public values y i = g x i, where each x i is known by only one party, seems to be an application 16

77 for common verifiable secret sharing (VSS) schemes. There, a dealer chooses a private key x and shares it in a verifiable manner with the participants in P according to the mechanisms of a threshold scheme. Although this method is very efficient, it cannot be used for our purposes, because corrupted dealers can choose their private keys non-randomly, hence, contradicting the proof of security in [GJ04]. Therefore the authors propose to use the DKG protocol of Gennaro et al. (GJKR-DKG), which indeed guarantees that all keys are uniformly distributed, but unfortunately decreases the efficiency. In contrast to this, our protocol is more efficient in terms of communication complexity in that it reduces the number of broadcasts by a factor of n, but still guarantees the randomness of the keys. In particular, we are able to show that the randomness of all keys, including the keys of corrupted parties, is guaranteed under the DDH assumption. Furthermore, we prove the security of the protocol for less than ( n 2 )-corrupted parties under the DL-assumption in the random oracle model. 2 The k-jrkg protocol In general, the protocol is structured into two phases: An initial phase which is executed only once at the beginning of the protocol and a key generation (KG) phase which has to be repeated for every needed key. In particular, in the initial phase the parties P = {P 1,..., P n } perform two instances of the GJKR-DKG scheme and use in the KG phase the ElGamal encryption and Feldman VSS. In the following let p, q be two large, odd primes with p = 2q + 1. Let g Z p be an element with order q and <g>= G Z p denote the subgroup of quadratic residues in Z p generated by g for which the DDH assumption holds. Finally, let F : Z q G be an efficiently invertible bijection. I. Initialization (executed only once): 1. The parties in P execute an instance of the GJKR-DKG scheme to generate a secret key X Z q and the corresponding public key Y = g X mod p. This key pair will be used for ElGamal threshold encryption. 2. The parties in P execute a second instance of the GJKR-DKG scheme to generate the value h = Y bx mod p. II. KG phase (k-iterations): 1. Generate random x = f(0) for dealer P D : (a) Each party P i P chooses randomly r i R Z q and broadcasts the commitment commit i = g r i h r i mod p. If one party P j doesn t broadcast her commitment she is disqualified. We denote by QUAL P the set of non-disqualified parties. (b) P i QUAL chooses randomly z i R Z q and computes the ElGamal encryption of F i = F (z i ) G: C i = (D i, E i ) = (g r i, Y ri F i ). (2) 17

78 We set: H i = commit i D i = h r i. Besides, the following non-interactive zero-knowledge proof is generated: NIZK i = P ok{r i : H i = h r i D i = g r i }. P i broadcasts C i and NIZK i in QUAL. Obviously each party can easily verify the correctness of the zk-proof. Incorrect behavior leads to disqualification. By using the multiplicative homomorphic property of the ElGamal encryption, each party can now compute the ciphertext of F ( z) := P j QUAL F (z j), i.e., C = ( P j QUAL g r j, P j QUAL Y r j F (z j )). (3) (c) P D chooses a uniformly distributed value ã 0 Z q and broadcasts Ã0 = g ea 0 in QUAL. (d) P D chooses a set T of t + 1 parties for publishing their decryption shares of C. Hence, each party in P can decrypt the value C and finally knows the unique value F all = Decrypt(C). It follows that each party can easily compute the value z = F 1 (F all ) by inverting F all. Hence, P i can calculate the unique verification value: A 0 = Ã0 g ez mod p = g ea 0+ez mod p = g a 0 mod p. In particular, only P D knows a 0 = z + ã Generate a polynomial to share the secret (Feldman-VSS): (a) P D creates a random polynomial f(z) over Z q with degree t: f(z) = a 1 z a t z t. P D chooses the value a 0 generated in II.1 as the constant coefficient. Hence, the polynomial to compute the parties shares is composed of f(z) = a 0 + f(z) = a 0 + a 1 z a t z t. P D broadcasts the following verification values in QUAL: A k = g a k, with k = 1,..., t. P D computes s i = f(i) mod q and sends it securely to P i. (b) Each party verifies the shares she received from other participants by using the following equation: g s i = t (A k ) ik (4) k=0 If the check fails P i broadcasts a complaint against P D. 18

79 (c) P D can answer this accusal by publishing valid shares s i, which satisfy equation 4. (d) The dealer P D is disqualified by each honest party P i QUAL, if either: P i receives more than t complaints, or P D wasn t able to answer with valid shares s i in II.2c. The k-jrkg scheme is called t-secure if in the presence of an attacker that corrupts at most t parties, the following requirements for correctness and secrecy are satisfied: Definition 2.1 The k-jrkg protocol is t-correct, if for all qualified dealers P D private key x and public key y = g x the following conditions hold: with shared (C1) All subsets of t + 1 shares provided by honest parties define the same unique secret x. (C2) All honest parties can compute the dealer s unique public key y = g x mod p. (C3) x is uniformly distributed in Z q and hence y is uniformly distributed in the subgroup G. (C4) Cheating leads to disqualification. Theorem 2.1 For every polynomial-time bounded adversary which corrupts at most t < n/2 parties the following holds: If the DDH-assumption is true, then for each secret x distributed in the KG-phase the correctness properties of definition 2.1 hold. The following theorem states that k-jrkg is t-secure: Theorem 2.2 For every polynomial-time bounded adversary which corrupts at most t < n 2 parties the following holds: If the DL-assumption is true, then for each secret x distributed in the KG-phase by an honest dealer P D no information on x can be learned by the adversary except for what is implied by the publicly known value y = g x. References [CS04] Canny, J., Sorkin, S.: Practical Large-Scale Distributed Key Generation. Lecture Notes in Computer Science 3027 (2004), [GJ04] Golle, P., Juels, A.: Dining Cryptographers Revisited. Lecture Notes in Computer Science 3027 (2004), [GJKR96] Gennaro, R., Jarecki, S., Krawczyk, H., Rabin, T.: Robust Threshold DSS Signatures. Lecture Notes in Computer Science 1070 (1996), [GJKR99] Gennaro, R., Jarecki, S., Krawczyk, H., Rabin, T.: Secure Distributed Key Generation for Discrete-Log Based Cryptosystems. Lecture Notes in Computer Science 1592 (1999), [Ped91] Pedersen, T.: A threshold cryptosystem without a trusted party. Lecture Notes in Computer Science 547 (1991),

80 Theorie und Anwendungen von Tree Parity Machines für die Kryptographie Andreas Ruttor und Markus Volkmer Universität Würzburg, Institut für Theoretische Physik und Astrophysik Am Hubland, D Würzburg Technische Universität Hamburg Harburg, Institut für Rechnertechnologie Schwarzenbergstraße 95, D Hamburg Schlüsselaustauschprotokolle werden immer dann benötigt, wenn zwei Partner A und B einen geheimen Schlüssel vereinbaren möchten, aber nur einen öffentlichen Kommunikationskanal zur Verfügung haben. Zur Konstruktion eines solchen Algorithmus kann man auf einen Effekt zurückgreifen, der bei der Untersuchung neuronaler Netze gefunden wurde: zwei Tree Parity Machines (TPMs), die voneinander lernen, synchronisieren schneller als ein drittes Netzwerk, das nur passiv an der Kommunikation teilnimmt [1]. Ein erfolgreicher Angriff mit den bisher bekannten Methoden erfordert deshalb einen erheblich höheren Aufwand als der Schlüsselaustausch selbst. Beide Partner verwenden für den neuronalen Schlüsselaustausch je eine aus K hidden units mit jeweils N Eingabeneuronen bestehende TPM, deren Anfangszustände zufällig und unabhängig voneinander gewählt werden. Jedes Netz definiert so eine Abbildung der KN binären Eingaben x ij { 1, +1} auf eine binäre Ausgabe τ { 1, +1}, K K N τ = σ i = sgn w ij x ij, i=1 i=1 die durch ganzzahlige Gewichte w ij { L, L+1,..., L} parametrisiert ist. Die TPMs von A und B erhalten in jedem Schritt einen zufällig erzeugten Satz der x ij und lernen die Ausgabe τ ihres Partners. Dabei werden die Gewichte gemäß der Hebbschen Lernregel, w ij + = w ij +τ x ij Θ(σ i τ)θ(τ A τ B ), oder einer ihrer Varianten angepasst. Dies führt nach t sync Schritten zu übereinstimmenden Gewichten (wij A = wb ij ), die anschließend von den Partnern als gemeinsamer Schlüssel eingesetzt werden können. Wegen der stochastischen Natur dieses Prozesses ist t sync eine Zufallsvariable, deren Mittelwert für K 3 proportional zu L 2 log N anwächst [2]. Sowohl bei der Synchronisation als auch beim Training einer weiteren TPM durch einen Angreifer E können zwei Arten von Lernschritten auftreten: attraktive und repulsive. Wenn die Ausgaben korrespondierender hidden units übereinstimmen (σi A = σi B ), dann bewirkt die Lernregel gleiche Änderungen in beiden neuronalen Netzen. Wird dabei eines der Gewichte am Rand bei L oder +L reflektiert, so verringert sich der Abstand wij A wb ij durch diesen attraktiven Schritt. Repulsive Schritte treten dagegen bei abweichenden Ausgaben auf (σi A σi B ) und führen zu einer Desynchronisation der TPMs, weil nur in einer der beiden hidden units die Gewichte angepasst werden. Folglich bestimmen die Häufigkeiten attraktiver und repulsiver Schritte wesentlich die Geschwindigkeit der Synchronisation. Da als Gesamtausgabe τ die Parität der σ i verwendet wird, lassen sich repulsive Schritte an τ A τ B erkennen, wenn eine ungerade Anzahl von hidden units betroffen ist. Hier haben A und B einen für die Sicherheit des neuronalen Schlüsselaustauschs entscheidenden Vorteil. Sie können sich nämlich gegenseitig beeinflussen und überspringen auf diese Weise einen Teil der repulsiven Schritte. Der Angreifer dagegen hat diese Möglichkeit nicht und kommt deshalb bei der Synchronisation im 20 j=1

81 Mittel langsamer voran als die Partner. Auf Grund dieses Nachteils gelingt E nur mit geringer Wahrscheinlichkeit P E eine Synchronisation mit A oder B, bevor diese einen gemeinsamen Schlüssel erzeugt haben und den Schlüsselaustausch beenden. Die Erfolgswahrscheinlichkeit P E wurde für viele verschiedene Angriffsmethoden [3] in Abhängigkeit von der synaptischen Tiefe L der TPMs untersucht. Dabei zeigt sich praktisch immer das gleiche Verhalten: P E fällt exponentiell mit zunehmendem L ab, während t sync nur proportional zu L 2 ansteigt [4, 5]. Die synaptische Tiefe hat also eine ähnliche Einfluss auf die Sicherheit des neuronalen Schlüsselaustauschs wie die Länge des Schlüssels bei einem Verschlüsselungsalgorithmus. Mit zunehmendem L steigt die Komplexität für einen erfolgreichen Angriff exponentiell an, während der Aufwand für die Erzeugung des gemeinsamen Schlüssels nur moderat wächst. Folglich kann man durch Erhöhen der synaptischen Tiefe L prinzipiell jedes gewünschte Sicherheitsniveau einstellen. Aufgrund der einfachen Arithmetik auf vergleichsweise kleinen ganzen Zahlen können TPMs vorteilhaft in Hardware realisiert werden. Im Fachgebiet der sog. Embedded Security wird neben der effizienten Realisierung etablierter kryptographischer Verfahren auch an alternativen kryptographischen Verfahren geforscht. Man versucht Sicherheitslösungen auch für solche Systeme zu finden, die zum Teil extreme Beschränkungen hinsichtlich der zur Verfügung stehenden Rechenleistung, Stromverbrauch und oder Logik-Fläche aufweisen. Im Rahmen des Projektes Tree Parity Machine Rekeying Architectures (TPMRA), an der TUHH von , wurde die Anwendung von TPMs in verschiedenen Systemen zur Absicherung von Kommunikation untersucht. Dabei wurden sowohl Software- als auch Hardware-Realisierungen [6] betrachtet. Im Unterschied zu den theoretisch erzielbaren Sicherheitsniveaus durch Skalierung der TPM-Parameter lag in diesem Projekt der Fokus auf möglichst hoher Sicherheit bei vergleichsweise kleinen Systemparametern, d.h. bei eher kleinen TPMs. Ein (bis dato) ausreichendes Sicherheitsniveau kann bei kleinen Systemen nur mit dem authentifizierten TPM-Schlüsselaustausch erzielt werden [8]. Neben der algorithmischen Variante mit Bit-Paket-Lernen wurden auch Varianten der Lernregel sowie Strategien des wiederholten Auffrischen von Sitzungsschlüsseln (sog. Rekeying-Strategien) mit dem Ziel der praktischen Anwendung untersucht. Aus dem Trajektorie-Modus der TPM wurde eine Stromchiffre abgeleitet, die sowohl OFB als auch CFB ermöglicht [7]. Die Eigenschaften und die Sicherheit der Chiffre befinden sich noch in Untersuchung. Die voll-serielle (Hardware-)IP-Core-Realisierung einer TPMRA benötigt etwa 2400 Gatteräquivalente inklusive Speicher. Aufgrund dieser geringen Logik-Fläche eignet sie sich zum Einsatz auf ressourcenbeschränkten Geräten, wie z.b. RFID-Tags [7]. Eine weitere Anwendung ist die Einbettung der (semi-parallelen) TPMRA in Bus-Systeme. Hierbei kommt sowohl die geringe Fläche (als kostengünstige Hardware-Komponente) als auch die Multifunktionalität des Verfahrens zugute. Identifikation, Schlüsselaustausch und Verschlüsselung per Stromchiffre können aus einem Prinzip abgeleitet werden [9]. Die Verwendung einer voll-parallelen TPMRA im Trajektorie-Modus erlaubte es, in einem PCI-Bus-System für jeden PCI-Buszugriff einen neuen Schlüssel bereitzustellen. TPMs erlauben es zudem einen Mehrparteien-Schlüsselaustausch über denselben Synchronisationseffekt zu realisieren. Diese Variante wurde prototypisch in einem WLAN-AdHoc Netzwerk zur Absicherung von Gruppenkommunikation auf Laptops implementiert. Hierbei können sowohl parallele als auch sequentielle Synchronisationsprozesse genutzt werden. 21

82 Literatur [1] Ido Kanter, Wolfgang Kinzel, and Eran Kanter. Secure exchange of information by synchronization of neural networks. Europhys. Lett., 57(1): , January [2] Andreas Ruttor, Georg Reents, and Wolfgang Kinzel. Synchronization of random walks with reflecting boundaries. J. Phys. A, 37: , August [3] Alexander Klimov, Anton Mityaguine, and Adi Shamir. Analysis of Neural Cryptography. In Yuliang Zheng, editor, Advances in Cryptology ASIACRYPT 2002, page 288, Heidelberg, February Springer. [4] Rachel Mislovaty, Einat Klein, Ido Kanter, and Wolfgang Kinzel. Public channel cryptography by synchronization of neural networks and chaotic maps. Phys. Rev. Lett., 91(11):118701, [5] Andreas Ruttor, Wolfgang Kinzel, Rivka Naeh, and Ido Kanter. Genetic attack on neural cryptography. Phys. Rev. E, 73(3):036121, March [6] Markus Volkmer and Sebastian Wallner. Tree Parity Machine Rekeying Architectures. IEEE Transactions on Computers, 54(4) [7] Markus Volkmer and Sebastian Wallner. Lightweight Key Exchange and Stream Cipher based solely on Tree Parity Machines. ECRYPT Workshop on RFID and Lightweight Crypto, 2005, Graz University of Technology, Graz, Austria, IACR Cryptology eprint Archive, Report 2005/232. [8] Markus Volkmer. Entity Authentication and Authenticated Key Exchange with Tree Parity Machines. IACR Cryptology eprint Archive, Report 2006/112, March [9] Sascha Mühlbach, Markus Volkmer, and Sebastian Wallner. Encrypted and Authenticated Communication via Tree-Parity Machines in AMBA Bus Systems. 4. Krypto-Tag Workshop über Kryptographie 2006, Ruhr-Universität Bochum, Horst-Görtz-Institut für IT-Sicherheit, Technical Report No. NDS-1/06, Ulrich Greveler (Hrsg.). 22

83 Opportunistische -Sicherheit Alexander Naumann und Tobias Straub Technische Universität Darmstadt, Fraunhofer-Institut für Sichere Informationstechnologie SIT, Darmstadt 1 Einleitung Die meisten der aktuellen -Programme unterstützen sichere mittels S/MIME oder PGP. Allerdings setzen Benutzer diese Funktionen aufgrund der aus ihrer Sicht ungünstigen Kosten- Nutzen-Relation sehr selten in der Praxis ein [1]. Ein Grund dafür ist die hohe technische Komplexität sowie das für ein korrektes Funktionieren der Sicherheitsmechanismen erforderliche Fachwissen. Insbesondere für Technik-Laien ist der Aufwand für die Konfiguration eines -Accounts mit Zertifikaten und für die Benutzung im täglichen Einsatz zu hoch. In dieser Arbeit wird ein opportunistischer Ansatz vorgestellt, der durch neue Interaktionsmuster und Heuristiken den Aufwand für den Benutzer senken, gleichzeitig aber ein möglichst hohes Maß an Sicherheit für s gewährleisten soll. Als Ziel wird ein sinnvoller und praxisnaher Ausgleich zwischen Benutzbarkeit und Sicherheit angestrebt. Im nächsten Abschnitt stellen wir unseren Ansatz von opportunistischer -Sicherheit vor, wie er bereits in Teilen prototypisch umgesetzt wurde (siehe Abschnitt 3). In Abschnitt 4 gehen wir auf Erweiterungen ein, die eine Integration in das Nutzer-Interface von Mozilla Thunderbird vorsehen. 2 Lösungsansatz Unser Lösungsansatz lässt sich von folgenden Kriterien leiten: Sicherheitskritische Routine- und Management-Aufgaben werden soweit wie möglich vom -Programm übernommen. Fehlerfälle oder mögliche Angriffe und entsprechende Reaktionen werden bereits frühzeitig beim Systemdesign berücksichtigt. Die notwendigen Vorkenntnisse werden als minimal angenommen, so dass auch Laien hinreichend verlässlich sichere verwenden können. Der Benutzer wird nicht mit technischen Details konfrontiert, sofern er dies nicht ausdrücklich wünscht. Die ersten beiden Kriterien führen zu einer Reihe von Heuristiken für die Anwendung von Security Policies sowie die Behandlung von Fehlern und Signalisierung möglicher Angriffe. Aufgrund des dritten und vierten Kriteriums sehen wir ein Konzept verschiedener User Levels, d.h. Kenntnisstufen vor, die den Grad an jeweils vorhandener Transparenz und Automatisierung bestimmen (vgl. [2]). 23

84 Typischerweise erfordern heutige -Programme häufige Benutzerinteraktionen für sicherheitsrelevante Einstellungen und Vorgänge. Demgegenüber besteht die Idee opportunistischer Sicherheit in einem Best Effort-Prinzip, d.h. s werden immer automatisch und ohne Benutzerinteraktion verschlüsselt und signiert, sobald sich dazu eine Gelegenheit ergibt. Signiert werden s, sofern bekannt ist, dass der Kommunikationspartner das System ebenfalls nutzt, was sich aus vorherigen s schließen lässt. Verschlüsselt wird, sofern ein Schlüssels des Kommunikationspartners bekannt ist, ansonsten erfolgt der Versand unverschlüsselt. Eine weitere Eigenheit opportunistischer Sicherheit betrifft die Gültigkeitsprüfung von Zertifikaten. Im Gegensatz zu sonst gängiger Praxis ist die Verifikation eine Option bzw. es wird ganz auf Zertifikate verzichtet. Die genannten Prinzipien sind eine Verallgemeinerung des Konzepts der opportunistic encryption, wie sie etwa auch schon für sichere eingesetzt wird [3]. Weiter werden der Sicherheitsstatus der Kommunikationspartner überwacht und Auffälligkeiten behandelt. Dazu gehört zum Beispiel eine nicht signiert empfangene , obwohl der Sender dazu eigentlich technisch in der Lage sein sollte. Veränderungen bei der Sicherheit des - Verkehrs oder Angaben zum Sicherheitsstatus einer werden dem Nutzer deutlich gemacht. 3 Prototyp Im Rahmen eines Praktikums an der TU Darmstadt wurde prototypisch ein Opportunistic Security System (OESS) in Form eines eigenständigen POP3-/SMTP-Proxy entwickelt. Da hierbei die Interaktion mit dem Benutzer über das -Programm stark eingeschränkt ist, soll das System um eine graphische Benutzerschnittstelle erweitert werden, über die der Sicherheitsstatus und eventuelle Probleme besser wiedergegeben werden können. Der in Java geschriebene Prototyp übernimmt die kryptographischen Operationen transparent für den Benutzer und dessen -Programm. Der Proxy erhält die s vom Mail User Agent, bearbeitet diese in der oben beschriebenen Weise und leitet sie, mit nur geringer Verzögerung, an den Mail Transfer Agent weiter. Das -Programm ist dazu für jeden vom Proxy zu verwaltenden Account einmalig so zu konfigurieren, dass der Datenverkehr über den Proxy geleitet wird. Vom -Programm wird eine Nachricht zum Versand vom Proxy in eine neue eingebettet, deren Header nur die Felder From, To und Subject umfasst. Die Authentizität und Integrität dieser Header wird durch eine Signatur geschützt. Außerdem ist die Vertraulichkeit der Betreffzeile dadurch gegeben, dass das Programm in der äußeren nur einen Hinweis auf die Verwendung von OESS enthält und die ursprüngliche Betreffzeile mit der inneren verschlüsselt übertragen wird. Diese Sicherheitseigenschaften der -Header bieten Ansätzen wie S/MIME und PGP nicht. Der Prototyp nutzt derzeit ein proprietäres Datenformat, wobei jedoch noch geprüft wird, ob dieses nicht auch kompatibel zu S/MIME oder PGP gemacht werden kann (vgl. [4]). Eine Idee ist dabei, die komplette gesicherte in ein S/MIME-kompatibles Format zu verpacken und zu verschicken. Gemäß kryptographischer Best Practice wird zuerst signiert, anschließend gegebenenfalls verschlüsselt. Zum Key Management haben wir einen Mechanismus entwickelt, der den Austausch der Schlüssel regelt, für Kontinuität beim Key-Rollover sorgt und somit durchgängige Sicherheit leistet. Bei gängigen -Programmen kann eine nur dann verschlüsselt an mehrere Empfänger verschickt werden, wenn für jeden dieser Empfänger ein Zertifikat vorliegt. Andernfalls ist die Op- 24

85 tion Verschlüsselung deaktiviert und die Nachricht wird an alle unverschlüsselt versendet. Dagegen ist der Proxy in der Lage, eine an verschiedene Empfänger so zu splitten, dass Empfänger, die OESS-fähig sind, ihre geschützt und alle anderen eine ungeschützte Version erhalten. Durch den Best Effort-Ansatz ist somit unter dem Strich ein höherer Anteil an verschlüsselten Nachrichten erreicht worden. 4 User Interface-Integration Zwar lässt sich der Proxy auch stand-alone betreiben, jedoch entfällt dabei die Möglichkeit, dem Benutzer Feedback über die Sicherheit seines -Verkehrs zu geben. Außerdem lassen sich keine qualitativen oder quantitativen Aussagen über den -Verkehr mit einem Kommunikationspartner machen, etwa in Bezug auf die bisherige Absicherung der s oder den Einsatz anderer -Verschlüsselungssoftware. Ziel der Visualisierung ist es nun, dem Nutzer wichtige Informationen über die darzustellen und gegebenenfalls Meldungen auszugeben oder Interaktionen zu ermöglichen. Der Nutzer muss wissen, welche der möglichen Aktionen potenziell gefährlich sind, wie sie zu vermeiden sind und wie sein aktueller Sicherheitstatus insgesamt und der einer einzelnen ist. Dafür wird als Machbarkeitsstudie eine Extension für Mozilla Thunderbird [5] implementiert. Dieses Visualisierungs- Plugin soll demonstrieren, wie entsprechende Informationen sinnvoll dargestellt werden können. Je nach User Level werden mehr oder weniger Informationen angezeigt, bzw. finden mehr oder weniger Interaktionen statt: User Level Visualisierung Anfänger Fortgeschrittener Experte Status der Mail (signiert, verschlüsselt) ja ja ja OESS-fähige Partner ja ja ja Zeitliche Entwicklung des Sicherheitsstatus nein ja ja Fehlermeldungen nein ja ja Abfrage bei an mehrere Empfänger nein ja ja Sender verwendet abgelaufene Schlüssel nein nein ja Die Evaluation des Systems soll unter möglichst realitätsnahen Bedingungen stattfinden, wobei zunächst untersucht wird, welche Usability-Methoden sich dazu am besten eignen. Die Evaluation selbst misst die Größen Nutzer-Akzeptanz (durch Interviews oder Umfragen), Effektivität der Sicherheitsmechanismen (durch Angriffsszenarien und Erkennung des Angriffe) sowie den Zeitaufwand für die Installation und Bedienung. 5 Ausblick Denkbar ist, das Prinzip opportunistischer Sicherheit auch auf andere Anwendungsbereiche von PKI zu übertragen. Wo und in wieweit dies möglich ist, soll untersucht werden. Einsatzgebiete könnten beispielsweise File Sharing- und Chat-Anwendungen oder die Absicherung von Datenverbindungen über HTTPS sein. Ähnliche Ansätze werden etwa bereits bei SSH oder IPv6 [6] verfolgt. 25

86 Literatur [1] Tobias Straub. Usability Challenges of PKI. Dissertation TU Darmstadt, [2] A. Whitten und J.D. Tygar. Safe Staging for Computer Security. In Proc. of the Workshop on Human-Computer Interaction and Security Systems, 2003 [3] Simson L. Garfinkel. Enabling confidentiality through the use of opportunistic encryption. In Proc. National Conference on Digital Government Research, 2003 [4] Simson L. Garfinkel, Robert C. Miller. Johnny 2: a user test of key continuity management with S/MIME and Outlook Express. In Proc. of the 2005 Symposium on usable privacy and security, 2005 [5] Mozilla Thunderbird Homepage. (Abgerufen am ) [6] Claude Castelluccia, Gabriel Montenegro, Julien Laganier, Christoph Neumann. Hindering Eavesdropping via IPv6 Opportunistic Encryption. In ESORICS, pages ,

87 Sicherheitsbeweise für zertifikatlose Public-Key Schemata Ewan Fleischmann Universität Mannheim Fakultät für Mathematik und Informatik Lehrstuhl für Theoretische Informatik Das Konzept der zertifikatlosen Public-Key Kryptographie (PKC) wurde im Jahre 2003 auf der AsiaCrypt erstmalig von Al-Riyami vorgestellt [AlR]. Einzuordnen ist sie zwischen der klassischen zertifikatbasierten PKC und der identitätsbasierten PKC. Ziel war es, von den Vorteilen der identitätsbasierten PKC zu profitieren (bspw. keine Public-Key Infrastruktur nötig, komplexer kryptographischer Workflow möglich) ohne jedoch deren oftmals inakzeptable Nachteile (zwangsweise Schlüsselhinterlegung) in Kauf nehmen zu müssen. Technisch basiert die zertifikatlose PKC jedoch weiterhin auf einer vertrauenswürdigen dritten Partei (genannt Key Generation Center, KGC). Diese kann jedoch keine Chiffretexte entschlüsseln, wie dies im identitätsbasierten Fall möglich ist. Bei einem Sicherheitsbeweis muss hierbei von zwei möglichen Angreifern ausgegangen werden: Ein Angreifer (Typ-I) entspricht einem normalen Angreifer einer beliebigen dritten Partei (ähnlich dem eines Public-Key Schemas). Der zweite Angreifer (Typ-II) modelliert ein KGC, welches versucht Zugriff auf den Klartext einer verschlüsselten Nachricht zu bekommen. Dieses hat jedoch Zusatzinformationen, über welche ein Typ-I Angreifer nicht verfügt. Damit ein Schema sicher ist, muss es sowohl sicher sein gegenüber einem Angreifer vom Typ-I wie gegenüber einem Angreifer vom Typ-II. Bis jetzt wurden nur zertifikatlose Schemata vorgestellt, welche entweder in einem eingeschränkten Angriffsmodell oder im Zufallsorakelmodell sicher sind. Als das zentrale Problem bei der Beweisführung im Standardmodell mit der üblichen (reduktionistischen) Vorgehensweise stellt sich heraus, dass ein Schema, welches beweisbar sicher gegenüber einem Typ-I Angreifer ist, zwangsläufig unsicher gegenüber einem Typ-II Angreifer ist. Aufgrund dieses sehr harten Resultats gingen einige Kryptographen davon aus, dass es prinzipiell nicht möglich sein kann, die Sicherheit von zertifikatlosen Schemata im Standardmodell (ohne weitere Einschränkungen vorzunehmen) zu beweisen. Dass dies nicht zwangsläufig daraus folgt, wird in [DK] genauer untersucht. Auch werden mögliche Lösungsansätze aufgezeigt. Inhalt des Vortrages ist eine kurze Darstellung der zentralen Ideen der identitätsbasierten und zertifikatlosen PKC. Dabei wird auch ein vereinfachtes Angriffsmodell auf diese Schemata vorgestellt, welches im Rahmen der Diplomarbeit entwickelt wurde. Daran anschließend wird die Problematik der Beweisführung im Standardmodell bei zertifikatlosen Schemata erläutert und auf einige konkrete Lösungsansätze eingegangen. Literatur [AlR] Sattam S. Al-Riyami. Cryptographic Schemes based on Elliptic Curve Pairings. PhD Thesis, 2004, University of London, Department of Mathematics [DK] Alexander Dent and Caroline Kudla. On Proofs of Security for Certificatelesss Cryptosystems, Information Security Group, Royal Holloway, Cryptology eprint Archive, Report 2005/348 27

88 Universelle Message Authentication Codes Christian Forler HORNBACH Baumarkt AG, Hornbachstraße, Bornheim In den letzten Jahren wurden Schwächen in den gängigen kryptographischen Hashfunktionen MD5 oder SHA-1 gefunden. Standardisierte Authentifikationsverfahren wie DSS/DSA oder HMAC basieren auf diesen Hashfunktionen. Dieser Umstand macht diese Verfahren unsicher, da sich aus unsicheren Bausteinen kein sicheres Authentifikationsverfahren konstruieren lässt. Mikle beschreibt beispielsweise, wie Authentifikationsverfahren, die auf MD5 basieren, gebrochen werden können [MIK04]. Authentifikationsverfahren wie der XOR-MAC, die auf Blockchiffren beruhen, sind sicher, wenn die verwendete Blockchiffre sicher ist. Leider gibt es noch keinen Sicherheitsbeweis für moderne Blockchiffren wie AES. Weiterhin sind solche Authentifikationsverfahren signifikant langsamer als solche, die auf kryptographischen Hashfunktionen basieren. Dies liegt daran, dass kryptographische Hashfunktionen in der Regel schneller als Blockchiffren sind. Wegman und Carter haben 1981 gezeigt, dass sich (informationstheoretisch) sichere MACs auf Basis von Hashfunktionen, die paarweise kollisionsresistent sind, konstruieren lassen [CW81]. Diese Hashfunktionen lassen sich nicht nur für die Kryptographie sondern auch für andere Bereiche der Informatik (Datenbanken, Filesysteme,...) nutzen. Daher werden sie universelle Hashfunktionen genannt. MACs, die darauf basieren, sind nicht nur beweisbar sicher sondern auch hoch performant. Wie Performancemessungen zeigen, lassen sich auf Standardhardware damit Durchsatzraten von unter einem cpb (clock cycle per byte) erreichen [KRO00]. Es spricht daher nichts dagegen, sich intensiver mit solchen universellen MACs zu beschäftigen. Literatur [MIK04] Ondrej Mikle. Practical Attacks on Digital Signatures Using MD5 Message Digest. December [CW81] J. Lawrence Carter and Mark N. Wegman. New Hash Functions and Their Use in Authentication and Set Equality. Journal of Computer and System Sciences, [KRO00] Theodore D. Krovetz. Software-Optimized Universal Hashing and Message Authentication. September

89 Privacy Friendly Location Based Service Protocols using Efficient Oblivious Transfer Markulf Kohlweiss and Bartek Gedrojc KU-Leuven Leuven Belgium TU-Delft Delft Nederlands Mobile devices add an additional dimension to context-based services: location. Bob, providing a location-based service (LBS), uses the location of Alice to answer her request, e.g., to find the next Italian restaurant. From a security standpoint the two main assets to be protected are Bob s database, and the location of Alice. Cryptographically this problem corresponds to an oblivious transfer (OT) of Bob s location specific data, where the index of the 1-out-of-n OT is the location σ of Alice, 1 σ n. By the properties of the OT, Alice learns only the information of the single map cell σ, while Bob is oblivious of Alice s location. In our work we investigate the specific needs of privacy friendly LBSs, and we design solutions based on efficient OT that take them into considerations. For instance, service providers have an interest in reducing the costs of the OT through economies of scale. Adaptive OT, where the same database is queried with little additional cost, provides a natural starting point. Similarly, the restricted capabilities of mobile users require a careful design of the system. In many of today s mobile networks there exists a dedicated party, the operator, that knows Alice s location. We investigate the role of this party to act as a proxy that inputs the user s location to the protocol and helps with doing the computation, but which otherwise remains oblivious of the protocols result. Finally we investigate the use of homomorphic encryption in order to support the access of multiple LBSs by the same user. This can be seen as a split oblivious transfer involving up to l location-based services simultaneously. Each LBS is handling a different database. Again the privacy sensitive information, i.e., which services Alice subscribed to, remains hidden from everyone else. Moreover the homomorphic property is utilized to facilitate the privacy preserving payment of the services. Privacy friendly LBS. Privacy is an enormous topic [12]. It is a sociological phenomenon which has many legal and commercial implications. The different ways location is used by an LBS greatly influences Alice s privacy experience. Is she interacting only with the service or with other users of the service? Is her location only used at the time of her request, or is she constantly tracked and notified upon certain events? Rather than covering all of this topics, we focus on a very specific sub-problem. Some of the techniques employed can however also be use for improving the privacy properties of other types of LBS protocols as surveyed in [11]. We do not consider solutions involving anonymity or service-side location specific privacy policies. Moreover we consider only solutions where no information at all about Alice s location is revealed to the LBS. The goal is to base the security of the system only on information theory and complexity theoretic assumptions. After the execution of the protocol a malicious LBS (even if collaborating with the operator) cannot compute anything, he (they) could not have computed before. It is easy to see that in this setting a notification service that contacts the user only upon events is impossible. The knowledge that the event occurred would already reveal information about Alice s location to the LBS. 29

90 Sender (d, e) $ Kg For i = 1... n C i Enc(m i ; H(i) d ) Chooser C 1,..., C n H(σ)b e H(σ) d b m σ Dec(C σ ; H(σ) d ) Figure 2: Adaptive OT based on Chaum blind signatures Oblivious transfer. We model a privacy friendly LBS as a database that maps every location i to some information m i. The number of different locations is restricted to n. A location can for instance be the name of a region, or a cell of a certain size on a map. Now the provisioning of a service corresponds to the retrieval of m σ for a hidden σ from a database m 1,..., m n. We call σ also the index into the database. The privacy requirements of the user imply the need for private information retrieval (PIR) [5]. Symmetric PIR (SPIR) is required if the LBS wants to avoid leakage of information about locations that have not been queried. It was shown that for the case where there is only one copy of the database there exists a communication-efficient reduction from any PIR protocol to a 1-out-of-n OT. Moreover for the single copy case SPIR corresponds to 1-out-of-n OT (OT 1 n) [7]. Oblivious transfer was first introduced by Rabin [18]. It captures the on first sight paradoxical notion of a protocol by which a sender sends some information to the receiver, but remains oblivious as to what is sent. The paradox is resolved by recognizing that it are the actions of the receiver and the sender that determine the outcome of the protocol. Even [8] generalized it to 1-out-of-2 oblivious transfer (OT 1 2). The receiver determines which message out of two possible messages she is going to receive. In turn it was shown how to construct OT 1 n from n [2] and even log n [13] applications of OT 1 2. [15, 1, 10] provided direct constructions for OT 1 n based on the decisional Diffie-Hellman and quadratic residuosity assumptions. Adaptive OT For location-based services, we are not so much interested in single executions of oblivious transfer, but want to query the same database multiple times at different indexes. This can be achieved by letting the sender commit to the database and running OT 1 n multiple times. However this is not the most efficient solution. Moreover the security requirements of such a system differ from those of normal oblivious transfer, as the protocol keeps internal state and queries can be chosen adaptively based on the results of previous queries. The first adaptive oblivious transfer protocol was proposed in [14]. Recently more efficient schemes were proposed by [16, 6]. [4] recognized that the last two schemes are based on a common principle to construct adaptive oblivious transfer from unique blind signature schemes. We briefly sketch the basic idea of the scheme using an example based on Chaum blind signatures (cf. Fig. 2). First, all messages are symmetrically encrypted using the RSA signature of the index. H( ) is a full domain cryptographic hash function. The encrypted database is transferred to Alice. When Alice wants to obtain the information for location σ, she runs a Chaum blind signature protocol with the sender to obtain the key. 30

91 Dynamic OT For practicality reasons we are also interested in dynamic databases that can shrink and grow during the execution of the adaptive oblivious transfer. This allows us to update parts of the database. For an update a new message is added to the database. Instead of accessing an old index, the user now has to access the new index. We require an additional table, that maps locations to their current indices. This update procedure reveals information about the database as Alice learns which entries have changed. It is an open research question whether we can do updates which don t reveal any information but are still substantially more efficient than running the whole protocol with a new database. Increasing the size of the database is straightforward. The sender just transfers a new cipher text C n+1 = Enc(m n+1 ; H(n + 1) d ) and transfers it to the sender. The sender can now also ask for blind signatures on n + 1. And decrypt C n+1. Deletion is more complicated. Our approach is to let the receiver prove that the requested σ is in the set of still valid indices V, e.g., by using a dynamic accumulator [3]. Together with every C i the sender now sends a witness w i = v (p 1 i ), with p i a prime. Before obtaining the signature the receiver now needs to prove that she knows a valid witness that corresponds to the blind signature request for index σ. For efficient protocols, it is now no longer possible to use a full domain hash RSA signature. Proxy OT and multi-database extensions For today s mobile networks it is natural to assume a third party, the operator, that knows Alice s location and can help her in executing the OT despite of limited device capabilities. This party executes most of the receiver s part of the OT protocol, but only Alice obtains the final result that allows her to decrypt C σ. We call the third party a proxy and the new protocol a proxy OT protocol. In location-based services, not only Alice s location, but also the type of service she is accessing is privacy sensitive information, which we do not want to reveal to the operator, or even the service himself. Thus a solution to this problem is of particular importance for LBS which use proxy OT, but may be of individual interest as an independent primitive. The selection of up to k services can be interpreted as an additional k-out-of-l OT, which is run independently but interleaving the proxy OT. Only Alice knows which of the l services she is accessing. Additional extensions are needed to facilitate payment for such hidden service usage. Preliminary ideas for a comprehensive solution are based on the use of homomorphic encryption in PIR [17], payment [1], and voting schemes [9]. References [1] William Aiello, Yuval Ishai, and Omer Reingold. Priced oblivious transfer: How to sell digital goods. In Birgit Pfitzmann, editor, EUROCRYPT 2001, volume 2045 of LNCS, pages , Innsbruck, Austria, May 6 10, Springer-Verlag, Berlin, Germany. [2] Gilles Brassard, Claude Crépeau, and Jean-Marc Robert. All-or-nothing disclosure of secrets. In Andrew M. Odlyzko, editor, CRYPTO 86, volume 263 of LNCS, pages , Santa Barbara, CA, USA, August Springer-Verlag, Berlin, Germany. [3] Jan Camenisch and Anna Lysyanskaya. Dynamic accumulators and application to efficient revocation of anonymous credentials. In Moti Yung, editor, CRYPTO, volume 2442 of Lecture Notes in Computer Science, pages Springer,

92 [4] Jan Camenisch, Gregory Neven, and abi shelat. Adaptive oblivious transfer from blind signatures. unpublished manuscript through personal communication, [5] Benny Chor, Eyal Kushilevitz, Oded Goldreich, and Madhu Sudan. Private information retrieval. J. ACM, 45(6): , [6] Cheng-Kang Chu and Wen-Guey Tzeng. Efficient k-out-of-n oblivious transfer schemes with adaptive and non-adaptive queries. In Serge Vaudenay, editor, PKC 2005, volume 3386 of LNCS, pages , Les Diablerets, Switzerland, January 23 26, Springer-Verlag, Berlin, Germany. [7] Giovanni Di Crescenzo, Tal Malkin, and Rafail Ostrovsky. Single database private information retrieval implies oblivious transfer. In EUROCRYPT, pages , [8] Shimon Even, Oded Goldreich, and Abraham Lempel. A randomized protocol for signing contracts. Communications of the Association for Computing Machinery, 28(6): , [9] Martin Hirt and Kazue Sako. Efficient receipt-free voting based on homomorphic encryption. In EUROCRYPT, pages , [10] Yael Tauman Kalai. Smooth projective hashing and two-message oblivious transfer. In Ronald Cramer, editor, EUROCRYPT 2005, volume 3494 of LNCS, pages 78 95, Aarhus, Denmark, May 22 26, Springer-Verlag, Berlin, Germany. [11] Tobias Kölsch, Lothar Fritsch, Markulf Kohlweiss, and Dogan Kesdogan. Privacy for profitable location based services. In Dieter Hutter and Markus Ullmann, editors, SPC, volume 3450 of Lecture Notes in Computer Science, pages Springer, [12] Scott Lederer, Jennifer Mankoff, and Anind K. Dey. Towards a deconstruction of the privacy space. Featured at Ubicomp 2003, [13] Moni Naor and Benny Pinkas. Oblivious transfer and polynomial evaluation. In 31st ACM STOC, pages , Atlanta, Georgia, USA, May 1 4, ACM Press. [14] Moni Naor and Benny Pinkas. Oblivious transfer with adaptive queries. In Michael J. Wiener, editor, CRYPTO 99, volume 1666 of LNCS, pages , Santa Barbara, CA, USA, August 15 19, Springer-Verlag, Berlin, Germany. [15] Moni Naor and Benny Pinkas. Efficient oblivious transfer protocols. In 12th SODA, pages , Washington, DC, USA, January 7 9, ACM-SIAM. [16] Wakaha Ogata and Kaoru Kurosawa. Oblivious keyword search. J. Complexity, 20(2-3): , [17] Rafail Ostrovsky and William E. Skeith III. Private searching on streaming data. In Victor Shoup, editor, CRYPTO, volume 3621 of Lecture Notes in Computer Science, pages Springer, [18] Michael O. Rabin. How to exchange secrets by oblivious transfer. Technical Report TR-81, Harvard Aiken Computation Laboratory,

93 Google Reveals Cryptographic Secrets Emin Islam Tatlı Department of Computer Science, University of Mannheim Google hacking is a term to describe the search queries that find out security and privacy flaws. Finding vulnerable servers and web applications, server fingerprinting, accessing to admin and user login pages and revealing username-passwords are all possible in Google with a single click. Google can also reveal secrets of cryptography applications, i.e., clear text and hashed passwords, secret and private keys, encrypted messages, signed messages etc. In this paper, advanced search techniques in Google and the search queries that reveal cryptographic secrets are explained with examples in details. 1 Motivation Having an index with over 25 billion entries, Google is the most popular web search engine. It indexes any information from web servers thanks to its hardworking web crawlers. But many sensitive data that should be kept secret and confidential are indexed by Google, too. Vulnerable servers and web applications, username-passwords for login sites, admin interfaces of database servers and online devices like web cameras without any access control, reports of security scanners and many more private information are available to hackers via Google. This paper focuses on the advanced search queries that enable users to search different cryptographic values which are expected to stay private and safe. The paper is organized as follows: Section 2 summarizes the useful parameters for the advanced search in Google. In Section 3, examples of search queries for each type of cryptographic secret are illustrated. Finally, Section 4 explains possible security measures against Google hacking. 2 Advanced Parameters Google supports many parameters for the advanced search and filters its results according to the parameters given by the user. The [all]inurl parameter is used to filter out the results according to if the url contains a certain keyword or not. If more keywords are needed, the allinurl parameter should be used. [all]intitle filters the results according to the title of web pages. [all]intext searches keywords in the body of web pages. With the parameter site you can do host-specific search. filetype and ext parameters have the same functionality and are needed to filter out the results based on the file extensions like html, php, asp etc. The minus sign (-) can be put before any advanced parameter and reverses its behavior. As an example, a search containing the parameter -site: will not list the results from The sign " " stands for the logical OR operation. 33

94 3 Google Search for Cryptographic Values From the cryptographic perspective, Google reveals also cryptographic secrets. Google can find out hashed passwords, secret keys, public and private keys, encrypted and signed files. What you need to do is only to enter the relevant search terms as explained in the following sections and click the search button. 3.1 Hashed Passwords Database structures and contents can be backed up in dump files. The following query searches for SQL clauses that may contain usernames and passwords in cleartext or in hashed values within dump files. Hash and encryption relevant keywords can also be searched within files. "create table" "insert into""pass passwd password"(ext:sql ext:dump ext:dmp) intext:"password pass passwd" intext:"md5 sha1 crypt" (ext:sql ext:dump ext:dmp) 3.2 Secret Keys Since the secret keys are generated mostly as session keys and destroyed after the session is closed, they are not stored on disks permanently. But there are still some applications that need to store secret keys, e.g., Kerberos [9] shares a secret key with each registered principal for authentication purposes. The following query lists the configuration files of a key distribution center (KDC) in Kerberos. Within the configuration files, the path of principal databases which contain principal ids and their secret keys is specified. inurl:"kdc.conf" ext:conf To find dumped Kerberos principal databases: inurl:"slave datatrans" OR inurl:"from master" Java provides a tool named keytool to create and manage secret keys in keystores. The extension of such keystores is ks. The following query searches for java keystores that may contain secret keys. Note that keytool can also manage private keys and certificate chains. keystore ext:ks 3.3 Public Keys Public keys, as the name implies, are public information and not secret. But for the sake of completeness, the search queries that list public keys are also written in this section. To list PGP public key files: 34

95 "BEGIN PGP PUBLIC KEY BLOCK" (ext:txt ext:asc ext:key) To list public keys in certificate files: "Certificate:Data:Version" "BEGIN CERTIFICATE" (ext:crt ext:asc ext:txt) 3.4 Private Keys Private keys should be kept secret for personal use but the following search queries show that people do not care about it and make it publicly accessible. "BEGIN (DSA RSA)" ext:key "BEGIN PGP PRIVATE KEY BLOCK" inurl:txt asc Gnupg [5] encodes the private key in secring.gpg. The following search reveals secring.gpg files: "index of" "secring.gpg" 3.5 Encrypted Files For confidentiality, cryptography provides encryption of data. By encrypting, one can store sensitive files and s securely on local storage devices. The following queries search for encrypted files and s. It is sure that you need to know the relevant keys to decrypt but as shown in the previous examples, it is also possible to find secret keys and private keys. Besides, other crypto analysis techniques can help to decrypt the encrypted files. The files that are encrypted with Gnupg get the extension gpg for binary encoding and the extension asc for ASCII encoding. The following first query searches files with gpg extension and tries to eliminate signed and public key files from the results. The second query lists ASCII encoded encrypted files. But note that signed files have also the same pattern and can be returned with the second query: -"public pubring pubkey signature pgp and or release" ext:gpg -"BEGIN PGP MESSAGE" ext:asc Many encryption applications use the extension enc for the encrypted files. There are some exceptions like AxCrypt File Encryption Software [6] which uses the extension axx for encrypted files: -intext:"and" (ext:enc ext:axx) In XML Security, the encrypted parts of messages are encoded under CipherValue element: "ciphervalue" ext:xml 35

96 3.6 Signed Messages Digital signatures provide integrity, authenticity and non-repudiation in cryptography. The following searches list some signed messages, signed s and file signatures. To list pgp signed messages ( s excluded): "BEGIN PGP SIGNED MESSAGE" -"From" (ext:txt ext:asc ext:xml) To list signed s: "BEGIN PGP SIGNED MESSAGE" "From" "Date" "Subject" (ext:eml ext:txt ext:asc) To list file signatures: -"and or" "BEGIN PGP SIGNATURE" ext:asc 4 Countermeasures Google hacking can be very harmful and therefore the required security measures should be taken against it. One method is using automatic scan tools [2, 3, 4] that search possible Google hacks for a given host. You can use the tools to search for the available flaws and risks in your system. The tools mostly use the hack database [1] when they do scan. Another solution is integration of robots.txt (robots exclusion standard) [7] files in your system. Web crawlers (hopefully) respect the directives specified in robots.txt. Providing this, you can prevent the crawlers from indexing your sensitive files and directories. The last and the most advanced suggestion is installing and managing Google honeypots [8] in your system and trying to figure out the behaviour of attackers before they deal with your real system. References [1] Google Hacking Database. [2] GooLink- Google Hacking Scanner. [3] SiteDigger v2.0 - Information Gathering Tool. [4] Johnny Long. Gooscan: Google Security Scanner. &file=index&req=getit&lid=33. [5] The GNU Privacy Guard. [6] AxCrypt File Encryption Software for Windows. [7] Robots Exclusion Standard. [8] Google Hack Honeypot Project. [9] Kerberos:The Network Authentication Protocol. 36

97 Trusted Computing mit Open Source Software Heiko Stamer Universität Kassel, Fachbereich Mathematik/Informatik Heinrich-Plett-Straße 40, D Kassel Das Thema Trusted Computing wird meist sehr kontrovers diskutiert. Häufig sind Fakten, Vermutungen und persönliche Meinungen stark miteinander vermengt, so dass sich kaum eine neutrale Sichtweise ausmachen läßt. Trotz aller Bedenken, beispielsweise hinsichtlich einschränkender Techniken wie Digital Rights Management (DRM) o. ä., gibt es mittlerweile im Open Source Bereich einige interessante Projekte, welche oft von Industrie oder Wissenschaft (FP6) gefördert werden: TPMDD [Hal06] Gerätetreiber für TPM-Hardware im Linux-Kern (IBM) TrouSerS [Yod06] Softwareschnittstelle für Anwendungsprogramme (IBM) jtss Wrapper [Win06] Java-Schnittstelle für Anwendungsprogramme (TU Graz) Trusted GRUB [Stü06] Sicheres Booten (Ruhr-Uni Bochum) TPM Emulator [Str06] Softwarebasierter Emulator für TPM-Hardware (ETH Zürich) Der Vortrag soll einen groben Überblick zur Trusted Computing Initiative der TCG geben und etwas genauer die Zielsetzung, den momentanen Status sowie weiterführende Anwendungsmöglichkeiten der oben genannten Open Source Projekte beschreiben. Zuerst werden wir die kryptographischen Komponenten der TPM-Hardware kennenlernen und einige Anwendungsszenarien (z. B. Remote Attestation) skizzieren. Danach schauen wir uns das Zusammenspiel von Hardware (TPM [TCGa]), Software (TSS [TCGb], Applikation) und Benutzer (Owner/User) anhand einiger konkreter Beispiele an. Weiterhin interessiert uns dabei, welche fortgeschrittenen kryptographischen Anwendungen sich mithilfe von Trusted Computing realisieren lassen. Zum Abschluß ist eine kurze Demonstration aktueller Software (TPM-Tools, TrouSerS, TPM- Emulator) und eine Diskussion geplant. Literatur [TCGa] Trusted Computing Group. TPM Specification. Version 1.2rev94, [TCGb] Trusted Computing Group. TSS Specification. Version 1.2, [DW06] Wilhelm Dolle, Christoph Wegener. Trusted Computing für Linux: Stand der Dinge. Linux-Magazin 04/2006. [Str06] Mario Strasser, et al. TPM-Emulator Project. Release 0.4,

98 [Hal06] Kylene Hall, et al. Linux TPM Device Driver. Kernel , [Yod06] Kent Yoder, et al. TrouSerS Project. Release 0.2.6, [Win06] Thomas Winkler, et al. IAIK/OpenTC jtss Wrapper. Release 0.2, [Stü06] Christian Stüble, et al. TrustedGRUB. Release 0.8.1,

99 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie. Er ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 5. Kryptotag am 11. September 2006 (Einreichung: 11. August 2006, Anmeldung: 7. September 2006). Universität Kassel, Arbeitsgruppe Theoretische Informatik. Kontakt: Heiko Stamer. 6. Kryptotag im Februar Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. Bisherige Kryptotage 1. Kryptowochenende am Juli 2006 Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 14 Einreichungen und 21 angemeldete Teilnehmer. 4. Kryptotag am 11. Mai Ruhr Universität Bochum, Horst-Görtz Institut. Kontakt: Ulrich Greveler. 10 Einreichungen und 32 angemeldete Teilnehmer. 3. Kryptotag am 15. September Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldeten Teilnehmer. 2. Kryptotag am 31. März Universität Ulm, Abteilung für Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldeten Teilnehmer. 1. Kryptotag am 1. Dezember Universität Mannheim, Lehrstuhl für Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldeten Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (École Normale Supérieure, Paris) verantwortlich für die Organisation der Kryptotage. Für evtl. Rückfragen bitte an sie wenden.

100 5. Krypto-Tag Workshop über Kryptographie Universität Kassel Heiko Stamer (Hg.) Universität Kassel, Fachbereich Mathematik/Informatik Heinrich-Plett-Straße 40, D Kassel 11. September 2006 Technical Report No. 06/06

101 Inhaltsverzeichnis Performanter Krypto-CoProzessor für unterschiedliche Verfahren Ralf Laue und Sorin A. Huss Generatoren für echte Zufallszahlen auf FPGAs für eingebettete Systeme Karl Tyss Starke kryptographische Algorithmen eine hinreichende Sicherheitsanforderung? Werner Schindler The estream Project Erik Zenner Neue Algorithmen für die Modularmultiplikation Viktor Bunimov The SMS4 Block Cipher Ralf-Philipp Weinmann On the Application of Merkle s Puzzle for Telemedicine and M-Health Frederik Armknecht and Dirk Westhoff Dissecting Apple s FileVault Ralf-Philipp Weinmann

102 Performanter Krypto-CoProzessor für unterschiedliche Verfahren Ralf Laue und Sorin A. Huss Fachgebiet Integrierte Schaltungen und Systeme, Fachbereich Informatik Technische Universität Darmstadt Ziel unserer Planung ist ein PKC-CoProzessor für den Servereinsatz, der verschiedene Verfahren (RSA, ECC) mit unterschiedlichen Bitlängen beherrscht. Für einen möglichst hohen Durchsatz müssen die zunehmend parallelen Strukturen heutiger Hardware kontinuierlich ausgelastet werden. Wir unterscheiden folgende Abstraktionsebenen: Die niedrigste ist die Ebene der modularen Arithmetik, welche für RSA und ECC existiert. Auf diese baut ausschließlich im ECC-Fall die Ebene der EC-Arithmetik auf. Darüber wiederum folgt die Ebene des Diskreten Logarithmus (ECC) bzw. der Integer Faktorisierung (RSA), wobei ECC die elliptischen Kurven als Gruppe nutzt, während für RSA die modularen Arithmetik diese Rolle übernimmt. Darauf gründet sich schließlich die Abstraktionsebene der kryptographischen Verfahren (z. B. ECDSA Sign). Wichtigste Elementaroperation ist die modulare Multiplikation, weshalb sie sich für Parallelisierung anbietet. Die erste Möglichkeit dafür ist in voller Bitbreite zu rechnen. Dazu muss die Hardware aber für die maximale Bitbreite ausgelegt sein und wegen der variablen Bitbreiten (> 100 bis mehrere 1000 Bit) bleiben Ressourcen oft ungenutzt. Die zweite Möglichkeit ist Pipelining, das dank frei wählbarer Anzahl der Stufen flexibler ist. Eine hohe Stufenzahl führt aber ebenfalls zu ungenutzten Ressourcen, da die Pipeline immer mit Bitlängen arbeitet, die einem Vielfachen ihrer Stufen in Wörtern entspricht. Einen guten Kompromiss stellen bei einer Wortlänge von 16 Bit Pipelines mit 2 oder 4 Stufen dar. Oberhalb der modularen Arithmetik nehmen die Verfahren hauptsächlich Steueraufgaben wahr. Außerdem sind die Algorithmen (z. B. RSA, ECDSA, usw.) stark unterschiedlich. Deshalb bietet sich eine Implementierung in Software an, die für Steueraufgaben gut geeignet ist und durch die sich der Prozessor für die verschiedenen Algorithmen wiederverwenden lässt. Auf heutigen FPGAs können dazu Soft-Cores verwendet werden: Diese sind einfache Prozessoren, die aus den konfigurierbaren Elementen des FPGAs gebildet werden. Wie oben beschrieben, ist Parallelisierung in der modularen Arithmetik nur beschränkt möglich, weshalb auch höhere Ebenen betrachtet werden sollten. Dabei beschränken jedoch Datenabhängigkeiten den Grad der Parallelisierung: Die EC Arithmetik lässt sinnvoll nur 2-3 parallele Multiplizierer zu, während die Exponentiation von RSA bis zu 2 nutzen kann. Um ungenutzte Ressourcen zu vermeiden, ist man also auf 2 Multiplizierer beschränkt und die eigentlich mögliche Parallelisierung innerhalb der EC Multiplikation entfällt. Das kryptographische Verfahren schließlich lässt sich dank fehlender Datenabhängigkeiten beliebig oft parallelisieren. PRNG HASH Soft Fkt. AES Core Mod Arith Soft Core Mod Arith... Soft Core Mod Arith IO Modul Die Abbildung zeigt vereinfacht eine mögliche Struktur für einen CoProzessor: Das IO-Modul kommuniziert mit der Außenwelt und überlässt die Ausführung der kryptographischen Verfahren den Soft-Cores. Diese verfügen je über ein Modul für modulare Arithmetik und teilen sich den Zugriff auf die weiteren Module wie AES- oder RNG-Core, da diese nicht so häufig benötigt werden. 3

103 Generatoren für echte Zufallszahlen auf FPGAs für eingebettete Systeme Karl Tyss Technische Universität Hamburg-Harburg, Institut für Rechnertechnologie Schwarzenbergstrasse 95, D Hamburg, Germany Die Entwicklung von ressourcenbeschränkten eingebetteten Systemen, wie z. B. RFID, hat dazu geführt, dass kryptografische Anwendungen immer mehr an Bedeutung gewinnen. Diese benötigen, zum generieren von Session-Keys, echte Zufallszahlen. Die Gewinnung von echten Zufallszahlen ist jedoch in der Regel mit einem hohen Aufwand verbunden, da die als Rauschquellen benutzte Hardware meist extern an die Andwendung gekoppelt oder als ein Full-Custom-Design realisiert wird. Die bei der Generierung von echten Zufallszahlen ausgenutzten physikalischen Phänomene verfügen nicht immer über die gewünschte Qualität und Stabilität. Diese Eigenschaften führen dazu, dass der Hardwareaufwand, um qualitativ hochwertige Zufallszahlen zu erzeugen, für solche Generatoren relativ hoch ausfällt. Im Fall von z. B. RFID-Anwendungen stehen aber nur wenige Hardwareressourcen zur Verfügung. Gebraucht werden deswegen vor allem Generatoren für echte Zufallszahlen, die mit wenig Hardwareressourcen auskommen und keine externe Hardware benötigen. In meinem Vortrag werde ich Generatoren für echte Zufallszahlen vorstellen, die sich durch eine geringe Komplexität auszeichnen. Es werden eine Reihe von Zufallszahlengeneratoren, basierend auf dem Phasenrauschen von oszillierenden Schaltkreisen [1] und ihre Realisierung auf einem FPGA vorgestellt. Da diese Klasse der Zufallszahlengeneratoren sich jedoch unter gewissen Umständen als anfällig für Manipulation erweist, werden im Vortrag einige bewährte [2], [3] und neu entwickelte Methoden zur Stabilisierung der Zufallszahlengeneratoren vorgestellt und diskutiert. Als Abschluss werde ich eine Lösung vorstellen, die mit wenig Hardware auskommt, auf FPGAs realisierbar und stabil gegenüber Manipulationen ist. Literatur [1] Michael Epstein, Laszlo Hars, Raymond Krasinski, Martin Rosner, Hao Zheng. Design and Implementation of a True Random Number Generator Based on Digital Circuit Artefacts. Cryptographic Hardware and Embedded Systems CHES th International Workshop, Cologne, Germany, September 8-10, 2003, Proceedings Series: Lecture Notes in Computer Science, Vol. 2779, Seiten , Springer Verlag. [2] Viktor Fischer, Miloś Drutarowský. The True Random Number Generator Embedded in Reconfigurable Hardware. Cryptographic Hardware and Embedded Systems CHES th International Workshop, Redwood Shores, CA, USA, August 13-15, 2002, Revised Papers Series: Lecture Notes in Computer Science, Vol. 2523, Seiten , Springer Verlag. [3] Werner Schindler. Efficient Online Tests for True Random Number Generators. Cryptographic Hardware and Embedded Systems CHES 2001 Third International Workshop, Paris, France, May 14-16, 2001, Proceedings Series: Lecture Notes in Computer Science, Vol. 2162, Seiten , Springer Verlag. 4

104 Starke kryptographische Algorithmen eine hinreichende Sicherheitsanforderung? Werner Schindler Bundesamt für Sicherheit in der Informationstechnik (BSI) Godesberger Allee Bonn Kryptographische Algorithmen und Protokolle bilden den Kern zahlloser IT-Sicherheitsmechanismen. Die Sicherheit kryptographischer Mechanismen beruht normalerweise darauf, dass ein potentieller Angreifer nicht in der Lage ist, einen geheimen Schlüssel zu bestimmen. Es liegt unmittelbar auf der Hand, dass die Resistenz der verwendeten kryptographischen Algorithmen und Protokolle gegen algorithmische Angriffe unverzichtbar ist. Andererseits geht es einem Angreifer nur um die Erlangung des gesuchten Schlüssels, und da ist ihm letztlich jedes Mittel recht, das zum Erfolg führt. So kann ein potentieller Angreifer den Speicherbereich eines PCs nach Schlüsseln durchsuchen, oder er könnte probieren, einen in einer Chipkarte gespeicherten Schlüssel durch einen direkten Angriff gegen die Hardware zu bestimmen oder nach einer Schwachstelle im Betriebssystem zu suchen. Wird zur Schlüsselerzeugung ein schwacher Zufallszahlengenerator verwendet, ermöglicht dies vielleicht das Errechnen oder das Erraten von Sessionkeys oder Signaturparametern, insbesondere wenn Vorgänger- oder Nachfolgerzufallszahlen bekannt sind. Seitenkanalangriffe sind kombinierte Angriffe, die den kryptographischen Algorithmus nicht direkt attackieren, sondern zudem Eigenschaften der Implementierung ausnutzen. Seitenkanalangriffe werden seit etwa 10 Jahren sowohl in der Wissenschaft als auch von der Halbleiterindustrie mit großer Aufmerksamkeit verfolgt. Besonderes Interesse gilt der Entwicklung wirkungsvoller Gegenmaßnahmen. Seitenkanalangriffe richten sich in aller Regel gegen Chipkarten, aber selbst Remote- Angriffe gegen Server sind möglich. Abhängig von ihrer Implementierung kann die Laufzeit kryptographischer Operationen von deren Input abhängen. Laufzeitangriffe versuchen, diesen Umstand auszunutzen, um aus einer Stichprobe von gemessenen Laufzeiten auf den gesuchten Schlüssel zu schließen. Die Ursache für Laufzeitunterschiede sind normalerweise laufzeitoptimierte arithmetische Algorithmen oder das Cacheverhalten. Bei Powerattacken misst der Angreifer den Stromverbrauch der Chipkarte, um hieraus sukzessiv Teile des Schlüssels zu bestimmen. Ziel eines Abstrahlangriffs ist es, an Hand der elektromagnetischen Abstrahlung zunächst die Aktivitäten auf dem Chip zu lokalisieren. Die schlüsselabhängige, d. h. die für einen Angriff nutzbare Information, ist durch Rauschen überlagert. Das gemeinsame Ziel von potentiellen Angreifern und Designern besteht darin, auf Basis von Messwerten optimale Entscheidungen zu treffen, wenngleich aus völlig unterschiedlichen Gründen: Angreifer wollen ihre Erfolgsaussichten optimieren, während ein Designer das Risikopotential eines Angriffs bzw. einer Klasse von Angriffen ausloten möchte, um zuverlässige Gegenmaßnahmen implementieren zu können. Zur Optimierung von Seitenkanalangriffen haben sich insbesondere stochastische Methoden als geeignet herausgestellt. 5

105 The estream Project Erik Zenner Cryptico A/S estream is an EU-funded project on stream cipher cryptography. After its predecessor NESSIE was unable to recommend a secure stream cipher for public use, the need for additional research in the area was recognized, and the estream project was born. Its purpose is to improve the understanding of stream cipher security and to identify a portfolio of ciphers that are both secure and resource-effective. In May 2005, the surprisingly large number of 34 stream cipher candidates were submitted to the project. After one year of public evaluation (and the submission of 128 academic papers to the project), the organisers announced a first selection in March Currently, the project is in evaluation phase 2, where the cryptographic community is again requested to continue the demolition derby and help in identifying the most suitable cipher candidates. In this talk, we will give an overview over the estream project. We will point out new trends in stream cipher cryptography and discuss some of the most interesting cipher candidates. Finally, we will go into some of the more controversial selection criteria, like the comparison of security features or resource consumption. 6

106 Neue Algorithmen für die Modularmultiplikation Viktor Bunimov Institut für Informatik, Uni Kiel In diesem Vortrag werden die wichtigsten Ergebnisse der Dissertation [1] aus dem Bereich der ganzzahligen Langzahlcomputerarithmetik für die Anwendungen vor allem aus dem Bereich der modernen Kryptographie vorgestellt. Alle hier behandelten Verfahren wurden weiterhin in Bezug auf eine Realisierung in Hardware optimiert. Bei den neuen Algorithmen handelt es sich um drei neue Methoden zur Berechnung der Modularmultiplikation, die sich durch ein besonders günstiges Flächen-Zeit-Produkt auszeichnen. Dabei wird zunächst eine neue Version der Modularmultiplikation von Montgomery entwickelt, bei der die Anzahl der Additionseinheiten gegenüber den besten bisher bekannten Verfahren [2] halbiert werden kann. Danach wird die verschränkte Modularmultiplikation behandelt. Bezogen auf die derzeit als bester geltenden Version [3] kann hier die Anzahl der Additionseinheiten um einen Faktor drei reduziert werden. Der dritte Algorithmus ist ein neu entwickeltes Verfahren [4], bei dem nicht nur die Berechnungen, sondern auch die Eingaben in redundanter Form erfolgen. Dadurch kann die Umrechnung von redundanter in nichtredundante Form bei der mehrfachen Anwendung der Modularmultiplikation, wie z. B. bei der modularen Exponentiation, eingespart werden, ohne zusätzlichen Zeit- bzw. Flächenaufwand zu benötigen. Alle drei Algorithmen werden in Hinsicht auf ihre Komplexität evaluiert. Literatur [1] Bunimov, V.: Entwicklung von neuen Algorithmen der Computerarithmetik in Hinsicht auf ihre Nutzung in der Kryptographie, Dissertation, Kiel [2] Kim, Y. S. Kang, W. S. Choi, J. R.: Implementation of 1024-bit modular processor for RSA cryptosystem, School of Electronic and Electrical Engineering, Kyungpook National University, 1370 Sankyok-Dong, Book-Gu, Taegu, Korea, pdf. [3] Koc, C. K.: RSA Hardware Implementation, RSA Laboratories, RSA Data Security, Inc. August 1995, [4] Schimmler, M., Bunimov, V.: Fast Modular Multiplication by Operand Changing. The International Conference on Information Technology ITCC 2004, pp , April 5-7,

107 The SMS4 Block Cipher Ralf-Philipp Weinmann Technische Universität Darmstadt The Wired Authentication and Privacy Infrastructure (WAPI) standard is a Chinese National Standard for securing Wireless LANs. It is an alternative to IEEE i which has become mandatory in China. Originally the block cipher SMS4 that is exclusively used in WAPI has been secret; however, due to non-acceptance of WAPI by the ISO standards organization, the Chinese government published the block cipher in January 2006 [1]. SMS4 is a 32 round unbalanced Feistel network [2] with a block and key size of 128 bits. It is source heavy, complete and homogeneous and uses a single 8-bit S-Box that has good differential and linear properties. The specification of the cipher is simple and clean: A C implementation of the cipher was finished in less than an hour. In this talk we will demonstrate that the design of SMS4 is somewhat brittle: A small change in the key schedule (different key constants) yields a variant that exhibits a large class of weak keys. Furthermore we will show differential and linear attacks against reduced-round versions of this cipher. References [1] Specification of SMS4 (in Chinese) [2] B. Schneier and J. Kelsey. Unbalanced Feistel Networks and Block Cipher Design. Fast Software Encryption 1996, Third International Workshop Proceedings (February 1996), Springer-Verlag, 1996, pp

108 On the Application of Merkle s Puzzle for Telemedicine and M-Health Frederik Armknecht and Dirk Westhoff NEC Europe, Network Laboratories, Heidelberg, Germany Cryptography for networks with low-end devices is usually designed to meet the capabilities of the weakest party. At this we are aiming at low-priced, unprotected hardware with extremely limited computing and storage capabilities which makes modular arithmetic with large numbers unsuitable or even impossible. However, in several cases, the network has an asymmetric topology with a full functioning powerful device and an extremely limited device with only reduced functioning. This is for example true in telemedicine, or, more concretely, in an m-health scenario where the patient s biosensors form a network with a more powerful control node. We argue that in these cases the well known Merkle s Puzzle [1] has its practical application and provide concrete parameter settings for specific device characteristics. As opposed to other mechanisms it takes advantage of the asymmetric topology by shifting most of the workload toward the more powerful device. The proposed solution has particular value in scenarios where security associations are required for a relatively short, but well-defined duration. Furthermore, no preinstalled secrets are required, making (often expensive) tamper resistance superfluous. References [1] R. C. Merkle. Secure communications over insecure channels. Communications of the ACM 21(4), pp , ACM Press,

109 Dissecting Apple s FileVault Ralf-Philipp Weinmann Technische Universität Darmstadt FileVault [1] is a security feature of Mac OS X that allows users to encrypt their home directories using their login passwords. Although Apple claims that security is achieved by encrypting its entire contents using the Advanced Encryption Standard with 128-bit keys, publically available technical information publically available beyond that statement is scarce: the only other thing known is that encrypted volumes (so-called disk images) are employed. Unfortunately the source code for this part of the operating system (the DiskImages framework) is not available for inspection. This makes accessing the contents of such encrypted volumes from other operating systems such as Linux or *BSD a impossible at the moment. This talk will show work in progress made by the author in reverse-engineering the the programs surrounding the Apple FileVault technology with the aim of creating a compatible driver for Linux. Being a cryptographer, the author also likes to second-guess the designers choices: however, no glaring holes in the design have been discovered yet. Blocks are encrypted in CBC mode, the IV for each block is computed using an HMAC-SHA1 variant. This raises an obvious question: Why has the cryptographic design of FileVault not been opened up for peer review? References [1] Apple Mac OS X FileVault 10

110 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie. Er ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 6. Kryptotag am 19. Februar 2007 (Einreichung: 19. Januar 2007, Anmeldung: 11. Februar 2007). Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. Bisherige Kryptotage 5. Kryptotag am 11. September Universität Kassel, Fachbereich Mathematik/Informatik, Fachgebiet Theoretische Informatik. Kontakt: Heiko Stamer. 8 Einreichungen. 1. Kryptowochenende am Juli Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 14 Einreichungen und 21 angemeldete Teilnehmer. 4. Kryptotag am 11. Mai Ruhr Universität Bochum, Horst-Görtz Institut. Kontakt: Ulrich Greveler. 10 Einreichungen und 32 angemeldete Teilnehmer. 3. Kryptotag am 15. September Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldeten Teilnehmer. 2. Kryptotag am 31. März Universität Ulm, Abteilung für Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldeten Teilnehmer. 1. Kryptotag am 1. Dezember Universität Mannheim, Lehrstuhl für Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldeten Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (Ecole Normale Superieure, Paris) verantwortlich für die Organisation der Kryptotage. Für eventuelle Rückfragen bitte an sie wenden.

111 Book of Abstracts 6. KRYPTOTAG Workshop der Fachgruppe Kryptografie in der Gesellschaft für Informatik Ammar Alkassar, Michael Backes (Hrsg.) Sirrix AG security technologies U N S A R I V E R S A V I E I T A S N I S S Sirrix AG security technologies, Im Stadtwald, Geb. D3 2, Saarbrücken Sirrix Technical Report 07/342 Classification: PUBLIC

112 Inhaltsverzeichnis Automated derivation of public key authenticity using a formal model of certificate based security infrastructures Thomas Wölfl... 3 Untersuchung von NTRUsign Richard Lindner...4 A Cryptographic model for branching time security properties the case of contract signing Ralf Küsters...5 Concealed data aggregation in wireless sensor networks Frederik Armknecht...6 Netzwerksicherheit: Mit Kryptographie gegen DDoS Ulrich Kühn und Roger Karrer...7 Neuste Ergebnisse des SASC Stream Cipher Workshops/eStream Dirk Stegemann...8 Sicherheit der Random-Oracle-Methodik: Ein neuer Ansatz Björn Fay...9 Beweisbare Sicherheit in der physikalischen Welt Sebastian Faust...10

113 Automated derivation of public key authenticity using a formal model of certificate based security infrastructures Thomas Wölfl University of Regensburg The authenticity of public keys is a well-known prerequisite for the applicability of asymmetric cryptography. Formal models of public key infrastructures (PKI) provide a theoretical foundation for the validation of public key certificates with the goal to derive the required key authenticity. A significant example is Maurer s PKI model [2]. However, existing models neglect validity periods and revocations of public key certificates. This work presents the formal model [1] which covers these temporal aspects. It allows the derivation of public key authenticity for a certain point in time. Additionally, it enables the authentication of attributes different from public keys, such as biometric reference templates, access privileges or liability commitments. The core of the model consists of eight axioms formulated in first order logic. It takes the perspective of the user Alice who represents her knowledge about digital certificates by means of two logical statements. After this, Alice verifies if public key (or attribute) authenticity is a logical consequence of the completion [3] of her knowledge and the model s axioms. Because of the complexity of the modeled scenarios, a manual derivation can be long and time consuming. Therefore, a PROLOG program is presented as an automated derivation method. It can be used in an interactive way (using e.g. SWI-PROLOG [4]) or it can be encapsulated in software and serve as module for the decision about public key (or attribute) authenticity. The latter has the advantage that soundness, completeness and termination of the logic program are formally shown which are crucial aspects for the software system or a cryptographic algorithm relying on the decision about authenticity. Another task of the program is the detection of revocation cycles. Consider the following paradox situation: A revocation r for digital certificate c exists which is used for the authentication or the authorization of the revocation r. It cannot be decided if either c or r is valid. For example, the validity of r implies the invalidity of c which implies the invalidity of r. The logic program detects these revocation cycles and delivers a warning. All in all, the formal model and the logic program can be used for the authentication of public keys and other attributes. This allows the reliable application of asymmetric cryptographic methods. References [1] T. Wölfl, Formale Modellierung von Authentifizierungs- und Autorisierungsinfrastrukturen, Deutscher Universitäts-Verlag, [2] U. Maurer, Modelling a public-key infrastructure, in: E. Bertino (Ed.), Proceedings of 1996 European Symposium on Research in Computer Security (ESORICS96), no in Lecture Notes in Computer Science, Springer, 1996, pp [3] U. Nilsson, J. Maluszynski, Logic, Programming and PROLOG, 2nd Edition, John Wiley and Sons, [4] 1

114 Untersuchung von NTRUsign Richard Lindner Technische Universität Darmstadt Kryptographie und Computeralgebra Hochschulstr. 10, Darmstadt Das neue Signaturverfahren NTRUsign[1] basierend auf NTRU[3], dem bekanntesten gitterbasierten Public-Key Verfahren von der gleichnamigen Firma aus Amerika, wurde in der Diplomarbeit Current Attacks on NTRU kritisch unter die Lupe genommen und auf schwächen geprüft. Dabei geht es inbesondere um 4 Angriffsarten, die bei der NTRUsign Variante mit Störung (engl. perturbations) beachtet werden müssen. Die Variante von NTRUsign ohne Störungen wurde ja sowohl von NTRU selber[1], als auch etwas solider von Nguyen und Regev als sehr unsicher entlarvt[5]. In der Diplomarbeit ging es hauptsächlich darum einen Algorithmus der sichere Parametervorgaben für NTRUsign bestimmt[2] zu prüfen. Dabei sind einige behebare Mängel aufgefallen. Und diese werden zusammen mit einem überraschenden und nicht so leicht behebaren Mangel vorgestellt. Literatur [1] Jeffrey Hoffstein, Nick Howgrave-Graham, Jill Pipher, Joseph H. Silverman, and William Whyte. NTRUsign: Digital signatures using the NTRU lattice. Lecture Notes in Computer Science, 2612: , [2] Jeffrey Hoffstein, Nick Howgrave-Graham, Jill Pipher, Joseph H. Silverman, and William Whyte. Performance improvements and a baseline parameter generation algorithm for NTRUsign. http: //grouper.ieee.org/groups/1363/lattpk/submissions.html, [3] Jeffrey Hoffstein, Jill Pipher, and Joseph H. Silverman. NTRU: A ring-based public key cryptosystem. In Buhler [4], pages [4] Joe P. Buhler, editor. Algorithmic Number Theory, ANTS-III, volume 1423 of Lecture Notes in Computer Science. Springer-Verlag, [5] Phong Q. Nguyen. A note on the security of NTRUsign. Cryptology eprint Archive, Report 2006/387,

115 A Cryptographic Model for Branching Time Security Properties the Case of Contract Signing Protocols Ralf Küsters ETH Zürich Some cryptographic tasks, such as contract signing and other related tasks, need to ensure complex, branching time security properties. When defining such properties one needs to deal with subtle problems regarding the scheduling of non-deterministic decisions, the delivery of messages sent on resilient (non-adversarially controlled) channels, fair executions (executions where no party, both honest and dishonest, is unreasonably precluded to perform its actions), and defining strategies of adversaries against all possible non-deterministic choices of parties and arbitrary delivery of messages via resilient channels. These problems are typically not, or not all, addressed in cryptographic models and these models therefore do not suffice to formalize branching time properties, such as those required of contract signing protocols. In this talk, a cryptographic model that deals with all of the above problems is proposed. One central feature of this model is a general definition of fair scheduling which not only formalizes fair scheduling of resilient channels but also fair scheduling of actions of honest and dishonest principals. Based on this model and the notion of fair scheduling, a definition of a prominent branching time property of contract signing protocols, namely balance, is provided, along with the first cryptographic proof that the Asokan-Shoup-Waidner two-party contract signing protocol is balanced. The cryptographic models and notions proprosed here provide a basis for relating cryptographic and formal definitions of branching time security properties. Joint work with Veronique Cortier and Bogdan Warinschi. 1

116 Netzwerksicherheit: Mit Kryptographie gegen DDoS Roger Karrer und Ulrich Kühn Deutsche Telekom Laboratories Technische Universität Berlin Distributed Denial of Service (DDoS) diese Art von Angriffen stellt im Internet ein großes Problem für ISPs und Serverbetreiber dar. Diese Angriffe existieren in einer Reihe von Varianten, z.b. kann man nach dem Angriffsziel unterscheiden: Endsysteme oder Netzwerke. So unterschiedlich die Angriffsziele, so unterschiedlich sind auch die nötigen Verteidigungsstrategien. Hier betrachten wir den Fall von DDoS-Angriffen auf die Bandbreite im Zugangsnetz von Rechnern. Typischerweise schickt dabei eine große Zahl von Bots gleichzeitig und koordiniert Daten an einen Server, so dass letztlich legitime Anfragen nicht mehr durchkommen. Diese Art von Angriffen führt in der Praxis zu erheblichen Schäden. In diesem Vortrag befassen wir uns mit der Frage, wie man dieses Problem angehen kann, und wie kryptographische Methoden dabei helfen können. Die Herausforderung besteht hier darin, dass Netzwerkelemente die Pakete des Datenstroms weiterleiten, während das Endsystem über die Legitimität urteilen kann. Auf der anderen Seite besitzen Netzelemente aber keine Möglichkeit der Unterscheidung von legitimen und unerwünschten Daten, während ein Endsystem keine Chance mehr hat, unerwünschten Datenverkehr und die dadurch entstehenden Engpässe zu verhindern. Wir werden hier einen Architektur-Ansatz aufzeigen und die zugehörigen Protokolle beschreiben, mit denen diese Herausforderung gelöst werden kann. Der Ansatz vereint dabei Sicherheit gegen Angriffe, Effizient und auch Anreize zum Einsatz. Die zentrale Idee ist, dass legitimer Datenverkehr durch den Sender mit einer kryptographischen Marke versehen wird. Diese Marke kann effizient durch ein neuartiges Netzelement Gate an geeigneter Stelle im Netzwerk überprüft werden kann. Aufbauend auf dieser Prüfung kann dann eine Entscheidung zur Ausfilterung getroffen werden. Darüber hinaus zeigen wir, wie ein effizienter und sicherer Informationsaustausch in dem sich ergebenden verteilten System realisiert werden kann. Bei den verwendeten Kommunikationsprotokollen ergeben sich eine Reihe von Herausforderungen, bei deren Lösung kryptographische Methoden zum Einsatz kommen: Sicheres Schlüsselmanagement, effiziente kryptographische Markierung, und nicht zuletzt ein Schlüsseltransport, der mit hoher Wahrscheinlichkeit menschliche Nutzer von Bots unterscheiden kann. 1

117 Sicherheit der Random-Oracle-Methodik: Ein neuer Ansatz Björn Fay Justus-Liebig-Universität Gießen Mathematisches Institut Arndstraße Gießen Spätestens seit dem Artikel [BR93] von Bellare und Rogaway gehört die Random-Oracle-Methodik zu den Standard-Beweistechniken in der Kryptologie. Dabei behandelt man während des Sicherheitsbeweises eines Protokolls auftretende Hashfunktionen als wären sie echte Zufallsfunktionen, sogenannte Random-Oracles oder kurz RO. Man hofft dabei, dass beim Austausch des Random- Oracles durch eine Hashfunktion der Sicherheitsbeweis intakt bleibt. Hierzu gab es allerdings schon mehrere Gegenbeispiele, u. a. der Artikel [MRH04] von Maurer, Renner und Holenstein, der mit Hilfe eines allgemeinen Modells zeigte, dass sich ein Random-Oracle im Allgemeinen nicht durch eine Hashfunktion ersetzen lässt. Da die Gegenbeispiele sehr künstlich sind, geht man auch weiterhin davon aus, dass bei Anwendung der Random-Oracle-Methodik auf normale Protokolle diese sicher sind, auch wenn dies kein Sicherheitsbeweis im Standard-Modell ist. Protokolle bei denen die Random-Oracle-Methodik häufig benutzt wird, sind z. B. digitale Signaturen. Oft wird dabei erst ein Hashwert von der zu signierenden Nachricht gebildet und dann dieser Hashwert anstatt der ganzen Nachricht signiert, sogenanntes hash & sign. Es gibt bisher allerdings keine allgemeinen Bedingungen unter denen sich die Random-Oracle- Methodik sicher anwenden lässt. Mittels einer neuen Art von Hashfunktionen, den unvorhersehbaren Hashfunktionen (für eine genaue Definition fehlt hier leider der Platz), lässt sich das zumindest ein wenig ändern. Es ist allerdings noch unklar, ob sich solche Hashfunktionen durch Standard- Annahmen konstruieren lassen, man die Existenz einfach annehmen muss oder vielleicht solche Hashfunktionen gar nicht existieren können. Bei diesen neuen Hashfunktionen hat man auf den Hashwert nicht mehr Einfluss als auf den Rückgabewert eines Random-Oracles. D. h. für einen Algorithmus A, der Urbilder für die Hashfunktion erzeugt, gibt es einen Algorithmus B, der Urbilder für das Random-Oracle erzeugen kann, so dass es einem dritten Algorithmus D nicht möglich ist, zwischen den beiden Ausgaben zu unterscheiden, auch wenn er noch weitere (genau spezifizierte) Zusatzinformationen hat. Es lässt sich damit eine Bedingung an ein hash & sign -Protokoll stellen unter der das Protokoll mit einer unvorhersehbaren Hashfunktion unter einem Angriff ohne Nachrichten nicht existentiell fälschbar ist, wenn das Protokoll mit einem Random-Oracale unter dem gleichen Angriff nicht existentiell fälschbar ist. Für Angriffe mit gewählten Nachrichten lässt sich eine solche Aussage noch nicht treffen. Allerdings lassen die bisherigen Ergebnisse hoffen, dass auch dies im Bereich des Möglichen liegen könnte. Evtl. lassen sich die Ergebnisse auch auf andere Protokolle anwenden. Literatur [BR93] Mihir Bellare, Phillip Rogaway. Random oracles are practical: A paradigm for designing efficient protocols. In ACM Conference on Computer and Communications Security, S , [MRH04] Ueli M. Maurer, Renato Renner, Clemens Holenstein. Indifferentiability, impossibility results on reductions, and applications to the random oracle methodology. In TCC 2004, Proceedings, Band 2951 von Lecture Notes in Computer Science, S , Springer,

118 Beweisbare Sicherheit in der physikalischen Welt Sebastian Faust KU Leuven, ESAT/COSIC Kateelpark Arenberg Heverlee Seit den frühen 80ern beschäftigt sich die Kryptographie mit dem Themengebiet der beweisbaren Sicherheit [GM84, BR93]. Typischerweise wird hierbei zunächst ein allgemeines Angreifermodell definiert unter dem dann Protokolle auf ihre Sicherheitseigenschaften hin formal untersucht werden können. Für den formalen Beweis wird dann gezeigt, dass unter bestimmten Annahmen, z.b. die Schwierigkeit von zahlentheoretischen Problemen, kein effizienter Angreifer auf das Protokoll existieren kann. Obwohl mittlerweile selbst für komplizierte Protokolle mathematische Sicherheitsbeweise existieren, werden immer häufiger praktische Angriffe auf Implementierungen der beweisbar sicheren Systeme gefunden. Nur sehr selten ist dies in einem Fehler des Sicherheitsbeweises begründet. Weitaus häufiger ist die Ursache für die Anfälligkeit des Systems im Modell zu suchen. Ein grundsätzliches Problem ist hierbei, dass Sicherheitsbeweise gegenwärtig im so genannten Black-Box Modell erbracht werden. Hier wird ein kryptographischer Algorithmus als Black-Box definiert, d.h. ein Angreifer erfährt nur eine Menge an Ein-/Ausgabepaaren, lernt jedoch nichts über interne Zustände während der Verarbeitung. In der Praxis kann ein Angreifer jedoch durchaus durch Beobachtung der physikalischen Welt, z.b. durch Messung des Stromverbrauchs oder des elektromagnetischen Feldes eines Chips, zusätzliche, so genannte Seiten-Kanal, Informationen sammeln. Diese Informationen werden im herkömmlichen Black-Box Modell nicht berücksichtigt, führen jedoch sehr häufig zum vollständigen Verlust der Sicherheit. In diesem Vortrag wird das Modell von Silvio Micali und Leonid Reyzin [MR04] vorgestellt, das genau an dieser Stelle ansetzt und versucht, die offensichtliche Lücke zwischen Theorie und Praxis zu schließen. Hierzu führen die Autoren ein abstraktes Berechnungsmodell ein, in dem kryptographische Algorithmen durch eine Kombination von abstrakten Turing Maschinen und einer Leakage- Funktion modelliert werden. Erste Resultate zeigen, dass sich die physikalische Welt bereits bei einfachen Ergebnissen deutlich von dem traditionellen Modell unterscheidet. So ist beispielsweise die Äquivalenz von Nicht-Unterscheidbarkeit und Nicht-Vorhersagbarkeit nicht mehr gewährleistet. Abschließend werden wir einige offene Fragen diskutieren, u.a. ob das MR-Modell tatsächlich die Realität abbildet und inwieweit man durch eine Anpassung des Modells für die Praxis relevante Resultate erzielen kann. Literatur [BR93] Mihir Bellare and Phillip Rogaway. Random Oracles are Practical: A Paradigm for Designing Efficient Protocols. ACM Conference on Computer and Communications Security, pages 62-73, [GM84] Shafi Goldwasser and Silvio Micali. Probabilistic Encryption. J. Comput. Syst. Sci. (28), pages , [MR04] Silvio Micali and Leonid Reyzin. Physically Observable Cryptography. Theory of Cryptography Conference, TCC LNCS 2951, pages ,

119 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie. Er ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 7. Kryptotag am 9. November 2007 (Einreichung: 2. Oktober 2007, Anmeldung: 1. November 2007). Bonn-Aachen International Center for Information Technology. Kontakt: Michael Nsken und Daniel Loebenberger Bisherige Kryptotage 6. Kryptotag am 19. Februar Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. 8 Einreichungen. 5. Kryptotag am 11. September Universität Kassel, Fachbereich Mathematik/Informatik, Fachgebiet Theoretische Informatik. Kontakt: Heiko Stamer. 8 Einreichungen. 1. Kryptowochenende am Juli Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 14 Einreichungen und 21 angemeldete Teilnehmer. 4. Kryptotag am 11. Mai Ruhr Universität Bochum, Horst-Görtz Institut. Kontakt: Ulrich Greveler. 10 Einreichungen und 32 angemeldete Teilnehmer. 3. Kryptotag am 15. September Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldeten Teilnehmer. 2. Kryptotag am 31. März Universität Ulm, Abteilung für Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldeten Teilnehmer. 1. Kryptotag am 1. Dezember Universität Mannheim, Lehrstuhl für Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldeten Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (K.U.Leuven, Belgien) verantwortlich für die Organisation der Kryptotage. Für eventuelle Rückfragen bitte an sie wenden.

120 Tagungsband zum 7. Kryptotag Workshop der Fachgruppe Kryptographie in der Gesellschaft für Informatik C Computer osec b-it SeCurity Bonn-Aachen International Center for Information Technology c 2007 Daniel Loebenberger, Michael Nüsken Version: 8. November 2007

121

122 09 30 Kaffee und Gebäck. Programm Begrüßung (Joachim von zur Gathen, Daniel Loebenberger, Michael Nüsken) Session I: Kryptographische Primitive Jens Zumbrägel (Universität Zürich). Einfache Halbringe und ihre kryptographischen Anwendungen Anna Perschl (FH Hagenberg). Attacken auf RSA durch Lösen von univariaten modularen Polynomgleichungen Jeremie Detrey (b-it). Arithmetic operators for pairing-based cryptography Kaffeepause Session II: Datenschutz Heiko Stamer (Universität Kassel). Kryptographische Missbrauchssicherung für die Voratsdatenspeicherung Martin Keßler, Sefan Lucks, Emin Islam Tatli (Bauhaus-Universität Weimar). Tracking Dog: A Privacy Tool against Google Hacking Denise Doberitz, Sebastian Gajek (Ruhr-Universität Bochum). Visual Mutual Authentication - an approach to secure online banking Mittagessen in der Kantine der Deutschen Welle Rump Session (Ankündigung anderer Tagungen etc) Gastvortrag: Werner Schindler (BSI): Für jedes Problem das richtige Werkzeug: Stochastische Methoden für Seitenkanalangriffe Kaffeepause Session III: Kryptanalyse von Stromchiffren Andrea Röck (INRIA). Entropieberechnung des inneren Zustandes eines FCSR s Tobias Eibach, Enrico Pilz (Universität Ulm). Ein Angriff auf Bivium mittels SAT Solver Deike Priemuth-Schmid, Alex Biryukov (University of Luxembourg). Auf der Suche nach schwachen Schlüsseln bei Trivium Verabschiedung Fachgruppensitzung. ab Ausklang des Kryptotages. 1

123 Teilnehmer Martin Goldack Mathias Herrmann.... Tina Lößle Deike Priemuth-Schmid.. Daniel Reinert Patrick Renz Maike Ritzenhofen.... Christopher Wolf

124 Für jedes Problem das richtige Werkzeug: Stochastik bei Seitenkanalangriffen Werner Schindler Bundesamt für Sicherheit in der Informationstechnik (BSI) Godesberger Allee Bonn Die Evaluierung kryptographischer Algorithmen und Protokolle erfordert den Einsatz zum Teil tiefliegender Techniken aus unterschiedlichen mathematischen Gebieten. Besondere Bedeutung besitzen u.a. die Zahlentheorie wie auch algebraische und komplexitätstheoretische Ansätze. Stochastik und Statistik spielen dagegen eine eher untergeordnete Rolle. Starke kryptographische Algorithmen und Protokolle sind jedoch nur eine notwendige, aber keineswegs eine hinreichende Voraussetzung für sichere IT-Systeme. Ebenso wichtig ist ihre sichere Implementierung und dabei insbesondere das Zusammenspiel mit den kryptographischen Mechanismen und ihrer algorithmischen Umsetzung. Die Chipkartenindustrie und universitären Forschungsgruppen verfolgen daher vor allem Seitenkanalangriffe mit großer Aufmerksamkeit und entwickeln sie weiter. Seitenkanalangriffe richten sich meist gegen Chipkarten, aber in den letzten Jahren sind auch Softwareimplementierungen verstärkt in den Fokus geraten. Vereinfacht gesagt, versuchen Seitenkanalangriffe, aus physikalischen Messwerte (z.b. Laufzeiten, Stromaufnahme oder elektromagnetische Abstrahlung) teilschlüsselabhängige Informationen zu extrahieren, um den gesuchten Schlüssel in kleinen Portionen (z.b. bit- oder byteweise) zu bestimmen. Diese teilschlüsselabhängigen Nutzinformationen werden jedoch in aller Regel durch weitere arithmetische Operationen und / oder gezielte algorithmische oder Hardware-Gegenmaßnahmen überdeckt und damit verschleiert. Ein gemeinsames Ziel von Angreifern und Designern von Kryptosystemen besteht darin, einen Angriff bzw. einen Klasse von Angriffen zu optimieren, wenngleich aus völlig unterschiedlichen Gründen. Während ein Angreifer schlicht seine Erfolgsaussichten optimieren möchte, ermöglicht dies dem Designer, das wahre Risikopotential auszuloten und zuverlässige und vertrauenswürdige Gegenmaßnahmen zu entwickeln. Fortgeschrittene stochastische und statistische Ansätze haben sich bei vielen unterschiedlichen Problemstellungen als die geeigneten Werkzeuge erwiesen, dieses Ziel zu erreichen. Gerade bei hardwarenahen Kryptographietagungen besitzen konkrete experimentelle Resultate, etwa das Brechen einer Testimplementierung samt einer ad-hoc Gegenmaßnahme, häufig einen höheren Stellenwert als die verwendeten mathematischen Methoden und eine tiefer gehende Analyse der Situation, um die eigentliche Ursache für den Angriff herauszuarbeiten und in einem zweiten Schritt zuverlässige Gegenmaßnahmen auch für modifizierte Angriffsvarianten zu entwickeln. Zudem erreichen Erkenntisse von hardwareorientierten Fachtagungen Softwareentwickler oft nur mit großer zeitlicher Verzögerung, obwohl ähnliche stochastische Ansätze teilweise auch Seitenkanalangriffe gegen Softwareimplementierungen ermöglichen (auch wenn die Angriffe im Detail sehr verschieden sein können). Diese beiden Effekte haben in der Vergangenheit verschiedentlich (z.b. bei OpenSSL) dazu geführt, dass Schutzmaßnahmen schrittweise nachgebessert werden mussten, nämlich als Antwort auf neue Angriffe. Im Vortrag werden verschiedene Angriffe gegen Chipkarten und Softwareimplementierungen besprochen. Gemeinsamkeiten und Unterschiede werden herausgearbeitet, die angewandten stochastischen Methoden dargestellt und Implikationen aufgezeigt. 1 3

125 Einfache Halbringe und ihre kryptographische Anwendung Jens Zumbrägel Institut für Mathematik, Universität Zürich Winterthurerstr. 190, CH Zürich Halbringe sind verallgemeinerte Ringe, in denen negative Elemente nicht zu existieren brauchen. Formal ist ein Halbring eine Struktur (R, +, ) bestehend aus einer Menge R und zwei Verknüpfungen +,, so dass (R, +) eine kommutative Halbgruppe ist, (R, ) eine Halbgruppe, und dass beide Distributivgesetze gelten. Wir nehmen weiterhin die Existenz eines Elements 0 R an, welches sich neutral in (R, +) und absorbierend in (R, ) verhält. Halbringe sind beispielsweise die natürlichen Zahlen (N = {0, 1, 2, 3,... }, +, ), der boolesche Halbring ({0, 1}, or, and), und der Endomorphismenhalbring (End(M), +, ) eines kommutativen Monoids (M, +, 0). Die Halbringstruktur ist im Wesentlichen die schwächste, für die noch eine sinnvolle Matrixmultiplikation definiert werden kann. Endliche Halbringe und Matrizen über diesen können in der Public-Key Kryptographie angewendet werden, um Operationen von kommutativen Halbgruppen auf Mengen zu konstruieren, die effizient berechenbar sind, jedoch im Allgemeinen schwierig zu invertierende Bahnenabbildungen besitzen. Solche Operationen bieten einen natürlichen Rahmen, um DL-basierte Kryptosysteme zu verallgemeinern, bei denen die Halbgruppe (Z, ) durch Exponentiation auf einer zyklischen Gruppe operiert, siehe [MMR05]. Wenn man nun Operationen unter Verwendung von Halbringen konstruiert, welche sich homomorph auf kleinere Halbringe abbilden lassen, dann ist das Kryptosystem anfällig für Angriffe, die vom verkleinerten homomorphen Bild ausgehen. Diese Angriffe sind vergleichbar mit dem Pohlig-Hellman-Algorithmus zur Berechnung diskreter Logarithmen in zyklischen Gruppen von Nichtprimzahl-Ordnung. Daher ist es für kryptographische Zwecke wichtig, dass der Halbring einfach d. h. kongruenz-frei ist, da in diesem Fall kein kleineres homomorphes Bild existiert. Die vollständige Klassifikation von endlichen einfachen Halbringen in [Zu07] (siehe auch [Mo04]) liefert zahlreiche Beispiele, nämlich gewisse Teilhalbringe von Endomorphismenhalbringen endlicher kommutativer idempotenter Monoide, welche bijektiv den endlichen Verbänden entsprechen. Es ist dadurch mit wenig Computer-Ressourcen möglich, auch sehr große Halbringe und deren Verknüpfungen zu erzeugen. Diese Vielfalt von einfachen Endomorphismenhalbringen bieten zahlreiche Möglichkeiten, Halbgruppenoperationen zu konstruieren, die unter anderem aufgrund der Effizienz für Public-Key Kryptographie potenziell interessant sind. Die Konstruktion von kryptographisch geeigneten Operationen ist Gegenstand meiner aktuellen Forschung. Literatur [Mo04] Chris Monico. On finite congruence-simple semirings. J. Algebra 271 (2004), no. 2, arxiv:math.ra/ , Mai [MMR05] Gérard Maze, Chris Monico, Joachim Rosenthal. Public Key Cryptography based on Semigroup Actions. arxiv:cs.cr/ v2, Januar 2005, erscheint in Adv. Math. Commun.. [Zu07] Jens Zumbrägel. Classification of finite congruence-simple semirings with zero. arxiv:math.ra/ , Feburar

126 Attacken auf RSA durch Lösen von univariaten modularen Polynomgleichungen Anna Perschl FH Oberösterreich - Campus Hagenberg Softwarepark 11, 4232 Hagenberg, Österreich Don Coppersmith stellte in [1] eine Methode vor, die eine univariate Polynomgleichung mit Modul N und Grad δ in polynomialer Zeit löst, wenn die gesuchte Lösung kleiner als N 1/δ ist. Auf diesen Ergebnissen aufbauend, präsentierte Nicholas Howgrave-Graham in [2] eine weitere Methode. Mit Hilfe dieser Methoden kann RSA kompromittiert werden, z.b. wenn ein Teil der Nachricht bekannt ist, die Nachricht zwei Mal mit gleichem Schlüssel und unterschiedlichem Padding oder die Nachricht mehrmals mit unterschiedlichen Schlüsseln und bekannten Paddings verschlüsselt wird. In diesen Fällen sind Angriffe dann möglich, wenn ein kleiner öffentlicher Exponent e verwendet wird und die Unbekannte kleiner als N 1/e ist. Die Methoden ermöglichen auch das Faktorisieren von N, wenn genügend höchstwertige Bits eines Teilers bekannt sind. Eine aktuelle Übersicht der Anwendungen findet sich in [3], ein aktueller Faktorisierungsalgorithmus in [4]. Die beiden Methoden lösen univariate modulare Polynomgleichungen in Zeit polynomiell in log N und δ. Da dies jedoch keine konkrete Aussage über die tatsächlichen Laufzeiten zulässt, haben wir beide Methoden im Rahmen der Diplomarbeit in Java implementiert und auf ihren Einsatz in drei Attacken getestet. Eine Gleichung dritten Grades mit einer Lösung von höchstens 256 Bit und einem Modul von 1024 Bit löst die implementierte Coppersmith-Methode in 55 Sekunden, die implementierte Howgrave-Graham-Methode in 22 Sekunden. Bei einem Modul von 2048 Bit benötigen die Methoden 340 bzw. 150 Sekunden. Wird also eine Nachricht mit öffentlichem Exponent 3 und einem 1024-Bit-Modul verschlüsselt, so kann ein Angreifer, der 3/4 der Nachricht kennt, das unbekannte Viertel der Nachricht in 22 Sekunden ermitteln. Dem Angreifer kann theoretisch bis zu 1/3 der Nachricht unbekannt sein, allerdings ist dann der benötigte Zeitaufwand deutlich größer. Wir präsentieren die Testergebnisse der Methoden im Detail, zeigen theoretische und praktische Grenzen auf und treffen eine Aussage über ihre Relevanz im Bezug auf die Sicherheit von RSA. Literatur [1] Don Coppersmith: Finding a Small Root of a Univariate Modular Equation. In Advances in Cryptology EUROCRYPT 96, Bd d. Reihe Lecture Notes in Computer Science, S , Springer-Verlag, [2] Nicholas Howgrave-Graham: Finding Small Roots of Univariate Modular Equations Revisited. In Crytography and Coding IMA 1997, Bd d. Reihe Lecture Notes in Computer Science, S , Springer-Verlag, [3] Alexander May: Using LLL-Reduction for Solving RSA and Factorization Problems: A Survey. LLL+25 Conference in honour of the 25th birthday of the LLL algorithm, [4] Mathias Herrmann, Alexander May: On Factoring Arbitrary Integers with Known Bits. In INFORMATIK Informatik trifft Logistik, Band 2, Bd. P-110 d. Reihe Lecture Notes in Informatics, S ,

127 Arithmetic operators for pairing-based cryptography Jérémie Detrey Computer Security group, B-IT, Bonn, Germany Since their introduction in constructive cryptographic applications, pairings over (hyper)elliptic curves are at the heart of an ever increasing number of protocols. Software implementations being rather slow, the study of hardware architectures became an active research area. In this talk, I will first describe an accelerator for the η T pairing over F 3 [x]/(x 97 + x ). Our architecture is based on a unified arithmetic operator which performs addition, multiplication, and cubing over F This design methodology allows us to design a compact coprocessor (1888 slices on a Virtex-II Pro 4 FPGA) which compares favorably with other solutions described in the open literature. References [1] Jean-Luc Beuchat, Nicolas Brisebarre, Jérémie Detrey, and Eiji Okamoto. Arithmetic operators for pairing-based cryptography. In P. Paillier and I. Verbauwhede, editors, 9th International Workshop on Cryptographic Hardware and Embedded Systems (CHES 07), pages , Vienna, Austria, September Springer-Verlag. 1 6

128 Kryptographische Missbrauchssicherungen für die Vorratsdatenspeicherung Heiko Stamer Universität Kassel, Fachbereich Elektrotechnik/Informatik Wilhelmshöher Allee 71-73, D Kassel Auf Grundlage der EU-Richtlinie 2006/24/EG hat die Bundesregierung im April 2007 einen umstrittenen Gesetzentwurf [BT07a] vorgelegt, welcher alle Telekommunikations- und Internetanbieter verpflichtet, die Kommunikationsverbindungsdaten ihrer Kunden für mindestens 6 Monate auf Vorrat zu speichern. Bei der Sachverständigenanhörung im Rechtsausschuss des Deutschen Bundestages [BT07b] ist kürzlich sehr deutlich geworden, dass nicht nur verfassungs- und datenschutzrechtlich große Bedenken bestehen, sondern das auch viele technische Fragen noch ungeklärt sind. Zwar gab es auf europäischer Ebene bereits verschiedene Expertentreffen [AKV1] und das Europäische Institut für Telekommunikationsnormen (ETSI) hat mittlerweile ebenfalls in Form von Standardisierungsbestrebungen [ETSI1, AKV2, ETSI2] reagiert, jedoch sind bisher kaum datenschutzfreundliche Lösungen für die Speicherung der sensiblen Verbindungsdaten öffentlich bekannt geworden. Das ist im Zusammenhang mit der Brisanz von Telekommunikationsverbindungsdaten mehr als erstaunlich, da diese Daten ein enormes Missbrauchspotential in sich bergen. Im Vortrag werden verschiedene Speicher- und Zugriffsszenarien sowie entsprechende Implementierungsansätze diskutiert, die mit Hilfe von Kryptographie (Verschlüsselung, PIR u. ä.) eine datenschutzfreundlichere Vorratsdatenspeicherung ermöglichen könnten. Diese konstruktive Herangehensweise soll jedoch nicht darüber hinweg täuschen, dass ich persönlich jegliche Vorratsspei- cherung prinzipiell für unverhältnismäßig, ineffizient und datenschutzrechtlich gefährlich halte. Literatur [BT07a] Gesetzentwurf der Bundesregierung. Gesetz zur Neuregelung der Telekommunikationsüberwachung und anderer verdeckter Ermittlungsmaßnahmen sowie zur Umsetzung der Richtlinie 2006/24/EG. BT-Drucksache 16/5846. Online verfügbar unter anhoerungen/24 tkue vorratsdatenspeicherung/01 Bundestagsdrucksachen/ pdf [BT07b] Zusammenfassung der Stellungnahmen der Sachverständigen im Rechtsausschuss. Telefonüberwachung und Datenspeicherung umstritten. Online verfügbar unter aktuell/archiv/2007/telekom kw38/index.html [AKV1] Arbeitskreis Vorratsdatenspeicherung. Protokoll des Expertentreffens zu Implementierungsfragen der Vorratsdatenspeicherung vom 16. Mai Online verfügbar unter: des Expertentreffens zu Implementierungsfragen der Vorratsdatenspeicherung vom 16. Mai 2007 [AKV2] Arbeitskreis Vorratsdatenspeicherung. ETSI-Dokumente. Online verfügbar unter: [ETSI1] European Telecommunications Standards Institute. Retained data requirements law enforcement. [ETSI2] European Telecommunications Standards Institute. Security framework in Lawful Interception and Retained Data environment. Ref. DTR/LI-00044, Start of work on

129 Tracking Dog - A Privacy Tool against Google Hacking Martin Keßler, Stefan Lucks and Emin Islam Tatlı Bauhaus Universität Weimar, Medien Fakultät Bauhausstraße 11, Weimar {martin.kessler, stefan.lucks, emin-islam.tatli}@medien.uni-weimar.de Protection of personal data is a privacy right from both ethical and legislation perspectives. Internet users require safeguarding their privacy against misuses and exploits. On the other hand, internet search engines and especially the most popular Google threaten user privacy. Google Hacking is a general term describing how Google can be used to find out vulnerable servers, files and web applications, unauthenticated programs, various online devices, etc.[ghdb]. As threats to privacy, Google indexes and reveals sensitive and confidential user data that include names, addresses, CVs, files containing usernames-passwords, confidential s and forum postings, private directories, chat log files, secret and private keys, etc. to unauthorized persons [Tatli06, Tatli07]. As a countermeasure, the users should be equipped with privacy enhancing tools to protect their privacy. In this talk, we present Tracking Dog [Kessler07], a penetration testing tool for searching cryptographic secrets and personal private data for a given host and/or given person name. The tool helps the individuals to detect if any of their confidential data have become public over the internet via Google. Tracking Dog supports both English and German language-specific queries and enables the users to edit raw search queries. References [Tatli06] Emin Islam Tatlı. Google Reveals Cryptographic Secrets. 1. Kryptowochenende, Kloster Bronbach, July [Tatli07] Emin Islam Tatlı. Google Hacking Against Privacy. 3. International IFIP Privacy Summer School, Karlstad-Sweden, August [Kessler07] Martin Keßler. Bachelorarbeit: Tracking Dog - Implementation of a penetration testing tool for searching cryptographic secrets and personal secrets with Google. Bauhaus Universität Weimar, Fakultät Medien, Oktober [Ghdb] Google Hacking Database

130 Visual Mutual Authentication - an approach to secure online banking Denise Doberitz and Sebastian Gajek Horst Görtz Institut für IT-Sicherheit Ruhr-Universität Bochum Today, most applications are only as secure as their underlying system. Since the design and technology of malware has improved steadily, their dectection is a difficult problem. As a result it is nearly impossible to be sure whether a computer, that is connected to the internet, can be considered trustworthy and secure or not. The question is, how to handle applications, that require a high level of security, such as online banking. In consequence we are interested in a solution, that allows us to establish a secure and trusted communication, although the underlying system is untrusted. In this paper we present an approach, that is based on Visual Cryptography. In [NaorS94] Shamir and Naor presented the concept of Visual Cryptography, that handles the plaintext to be encrypted as a graphic, that is processed pixel by pixel and thus implies interesting characteristics concerning security and fault tolerance. In [NaorP97] Naor and Pinkas demonstrated how to use Visual Cryptography for the authentication of one party without trusting the underlying system. Based on this, we develope a scheme for mutual visual authentication of a user and a server, that allows us to establish a trusted channel between the two parties, and can be used for secure communication. We describe the Visual Mutual Authentication - Scheme and demonstrate its application as a protocol on the example of online-banking. The protocol provides a practical and user-friendly approach to secure online banking, that is not only resilient to malware, but also resistant to phishing. Using Visual Cryptography, the user s key is a transparency that has to be applied on the computer monitor. With the tansparency, the decrypted message can simply be read from the monitor by the user. We transfer this application to a TAN-scheme and integrate a one-time pad structure. As a result, we achieve a user friendly scheme with a high level of secrecy. References [NaorS94] Moni Naor and Adi Shamir. Visual Cryptography, EUROCRYPT 94, Volume 950 of Lecture Notes in Computer Science, pp Springer-Verlag, [NaorP97] Moni Naor and Benny Pinkas. Visual Authentication and Identification. CRYPTO 97, Volume 1294 of Lecture Notesin Computer Science, pp Springer-Verlag,

131 Entropieberechnung des inneren Zustandes eines FCSR s Andrea Röck Projet CODES, INRIA Paris-Rocquencourt, France Lineare rückgekoppelte Schieberegister (LFSR) sind weitverbreitete Primitive in der Kryptographie. Um deren Linearität zu vermeiden, kann auf die nichtlineare Variante des rückgekoppelten Schieberegisters mit Übertrag (engl. Feedback with Carry Shift Register, kurz FCSR) ausgewichen werden, wie z.b. im Fall des estream Kandidaten F-FCSR [AB05] für eine Stromchiffre. Ein FCSR ist ein endlicher Automat, welcher unabhängig von Klapper und Goresky [KG97], Marsaglia und Zamand [MZ91] und Couture und L Ecuyer [CL94] eingeführt wurde. Er ist ähnlich aufgebaut wie ein LFSR, jedoch wird für die Rückkopplung nicht ein XOR, sondern eine Addition mit Übertrag angewandt. Daher benötigen wir an jeder Rückkopplungsposition einen zusätzlichen Speicherplatz für den Übertrag. Im Folgenden betrachten wir binäre FCSR s. Wir interessieren uns für die Entropie des inneren Zustandes des FCSR s. Eine verringerte Entropie erleichtert es einem Angreifer, den Wert des Zustandes und somit die nachfolgende Ausgangssequenz des FCSR s zu erraten. Nehmen wir an, der innere Zustand des FCSR s bestehe aus einem Hauptregister der Länge n und l Übertragsbits an den Rückkopplungspositionen. Unter der Vorraussetzung, dass der Anfangszustand jeden der 2 n+l möglichen Werte mit der gleichen Wahrscheinlichkeit annimmt, hat dieser FCSR eine Anfangsentropie von n + l Bits. Da die Iteration des FCSR s keine bijektive Funktion ist, wird die Entropie mit jeder Iteration solange verringert, bis der FCSR einen periodischen Kreislauf erreicht. Ab diesem Zeitpunkt verhält sich eine Iteration wie die Anwendung einer Permutation, und die Entropie bleibt konstant. Im Allgemeinen sinkt die Entropie des inneren Zustandes relativ schnell. Arnault, Berger und Minier zeigen in [ABM07], dass ein FCSR mit maximaler Periode seinen periodischen Kreislauf und somit seine endgültige Entropie nach maximal n + 3 Iterationen erreicht. Indem wir betrachten, von wie vielen Zuständen ein bestimmter Zustand erzeugt werden kann, können wir zeigen, dass schon nach einer Iteration die Entropie auf n + l 2 Bits absinkt. Zusäztlich können wir einen Algorithmus angeben, welcher die endgültige Entropie für einen beliebigen FCSR berechnet. Dieser Algorithmus kann auch verwendet werden, um zu zeigen, dass die Entropie immer größer als n Bits ist, und dass die Verschiebung einer Rückkopplungsposition um eine Stelle in Richtung des höchstwertigen Bits die Entropie erhöht. Literatur [AB05] F. Arnault and T. P. Berger. F-FCSR: Design of a new class of stream ciphers. In FSE, Lecture Notes in Computer Science, pages Springer, [ABM07] F. Arnault, T. P. Berger, and M. Minier. On the security of FCSR-based pseudorandom generators. In SASC ECRYPT Workshop on stream ciphers, January [CL94] R. Couture and P. L Ecuyer. On the lattice structure of certain linear congruential sequences related to AWC/SWB generators. Math. Comput., 62(206): , [KG97] A. Klapper and M. Goresky. Feedback shift registers, 2-adic span, and combiners with memory. J. Cryptology, 10(2): , [MZ91] G Marsaglia and A Zaman. A new class of random number generators. Annals of Appl. Prob., 1(3): ,

132 Ein Angriff auf Bivium mittels SAT Solver Tobias Eibach and Enrico Pilz Universität Ulm Universität Ulm Trivium (und damit auch die reduzierte Variante Bivium) ist ein aussichtsreicher Stufe 2 Kandidat in dem europäischen estream Projekt. Das Design der Stromchiffre ist schlicht und elegant, vorgestellt wurde die Chiffre 2005 von Cannière und Preneel [CB05]. Bislang wurde noch kein wirkungsvoller Angriff auf die Chiffre gefunden. Zu den bekanntesten Angriffen auf Strom Chiffren zählen die Algebraischen Angriffe, die darauf abzielen ein Gleichungssystem aufzustellen und zu lösen. Bekannte Verfahren hierbei sind der F4- und F5-Algorithmus die auf Gröbner Basen basieren, die sich in Softwarepaketen wie Magma und Singular finden. In [BCJ07] wird vorgeschlagen SAT Solver zum Lösen solcher (dünner) Gleichungssysteme zu verwenden und dieser Gedanke wird in [MCP07] speziell auf Bivium angewendet und durchgeführt. Hierin wurden auch erste interessante experimentelle Ergebnisse veröffentlicht. SAT Solver sind seit langer Zeit Forschungsgegenstand der Theoretischen Informatik, es geht darum, die Erfüllbarkeit von Booleschen Formeln (in CNF) zu überprüfen und ggf. eine Lösung auszugeben. Das Erfüllbarkeitsproblem ist das kanonische NP-vollständige Problem, man geht davon aus, dass es keinen Algorithmus gibt, der das Problem in polynomieller Laufzeit lösen kann. Trotz dieser theoretisch nicht-effizienten Laufzeit ging viel Forschungsarbeit in die Entwicklung von möglichst effizienten SAT Algorithmen, die hoch optimiert sind und auf Teilklassen (hier: industrial instances ) besonders schnell arbeiten. Für den Angriff auf Stromchiffren bedeutet das: Zuerst muss das Problem in ein Gleichungssystem umgeformt werden (dessen Lösung der gesuchte interne Zustand der Chiffre ist). Das algebraische Gleichungssystem wird dann in eine äquivalente CNF transformiert, die dann als Eingabe für einen SAT Solver dient. Den oben beschriebenen Angriff haben wir implementiert und damit die Ergebnisse aus [MCP07] überprüft und weitere Experimente durchgeführt. Teilweise konnten wir Ergebnisse bestätigen, teilweise nicht. Neue Ergebnisse betreffen Fragestellungen auf dem Weg zu einem möglichst effizienten Angriff und sollen in dem Vortrag vorgestellt werden. Zum Bsp. geben wir eine empirische Antwort darauf, welche Bits des internen Zustandes am wertvollsten sind und somit am besten erraten werden sollten. Auerdem haben wir untersucht welcher SAT solver am Besten für diese Problemklasse geeignet ist und ob mehr Redundanz im Gleichungssystem das Problem einfacher macht. Abschlieend wird noch vorgestellt welche Fragestellungen wir in den nchsten Wochen noch untersuchen werden. Literatur [CB05] Cannière and Preneel. TRIVIUM - a stream cipher construction inspired by block cipher design principles, [BCJ07] Bard and Courtois and Jefferson. Efficient Methods for Conversion and Solution of Sparse Systems of Low-Degree Multivariat Polynomials over GF(2) via SAT-Solvers,2007. [MCP07] Cameron McDonals and Chris Charnes and Josef Pieprzyk. Attacking Bivium with MiniSat,

133 Auf der Suche nach schwachen Schlüsseln bei Trivium Deike Priemuth-Schmid und Alex Biryukov Laboratory of Algorithmics, Cryptology and Security University of Luxembourg Die Stromchiffre Trivium wurde als Kandidat für das estream-projekt [ECR05] im Jahr 2005 von C. De Cannière und B. Preneel [CP05] entwickelt. Seit der Veröffentlichung von Trivium wurde das Design der Chiffre nicht geändert, denn auch nach Jahren gibt es keine wirklichen Angriffe auf Trivium. Dies und seine sehr guten Eigenschaften im Bereich Performanz und das geringe Zeit-Flächen-Produkt machen Trivium zu einer interessanten Stromchiffre. Trivium verwendet einen 80 Bit Schlüssel, einen 80 Bit Initialwert und eine nichtlineare Update- Funktion. Von den Designern wurde die Chiffre mit einem 288 Bit Register vorgestellt, welches jedoch auf natürliche Weise in drei Register der Länge 93, 84 und 111 Bit zerfällt. Mit jeder Taktung der Chiffre wird diese aktualisiert und ein Schlüsselstrombit ausgegeben. Die maximale Anzahl an Schlüsselstrombits pro Schlüssel-Initialwert-Paar wurde von den Designern mit 2 64 Bits festgelegt. Im Key-Setup wird der Schlüssel in das erste und der Initialwert in das zweite Register geladen, jeweils beginnend mit den ersten 80 Bit. Alle restlichen Bits werden mit Null initialisiert, mit Ausnahme der letzten 3 Bit des dritten Registers, welche auf Eins gesetzt werden. Dieser Zustand mit Schlüssel, Initialwert und den fest gesetzten Bits ist der Startzustand. Nachdem alle Bits initialisiert sind, wird das Register mal unter Verwendung der Update-Funktion getaktet, ohne dass Schlüsselstrombits produziert werden. Danach beginnt die Erzeugung des Schlüsselstroms. Wir untersuchen die Fragestellung, ob es Paare von Schlüsseln und Initialwerten gibt, welche wieder einen Startzustand reproduzieren, wie erstmals von Jin Hong [JH05] erwähnt. Im Startzustand sind 128 Bits festgelegt. Nutzen wir diese festen Positionen, um den zweiten Startzustand als Gleichungssystem darzustellen, so erhalten wir 128 Gleichungen in 160 Variablen, davon jeweils 80 Schlüssel- und Initialwertbits. Der maximale Grad dieser Gleichungen beträgt entweder 3 oder 4. Da die Anzahl der Variablen größer ist, als die der Gleichungen, erhalten wir 2 32 mögliche Lösungen für das Gleichungssystem. Unser Ziel ist die effiziente Beschreibung dieser Lösungen. Desweiteren wollen wir auch andere Chiffren auf solche schwachen Schlüssel untersuchen. Literatur [CP05] C. De Cannière and B. Preneel. TRIVIUM - a stream cipher construction inspired by block cipher design principles. estream, ECRYPT Stream Cipher Projekt, Report 2005/030, [ECR05] estream: The ECRYPT Stream Cipher Projekt, [JH05] Jin Hong. Discussion Forum. certain pairs of key-iv pairs for Trivium, created September 13, :11PM

134 13

135 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie. Er ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 8. Kryptotag am 11. April 2008: Wilhelm-Schickard-Institut für Informatik der Universität Tübingen. Kontakt: Michael Beiter, Claudia Kaesser und Anja Korsten. Bisherige Kryptotage 7. Kryptotag am 9. November 2007: Bonn-Aachen International Center for Information Technology. Kontakt: Michael Nüsken und Daniel Loebenberger. 9 Einreichungen und 36 Anmeldungen. 6. Kryptotag am 19. Februar 2007: Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. 8 Einreichungen. 5. Kryptotag am 11. September 2006: Universität Kassel, Fachbereich Mathematik/Informatik, Fachgebiet Theoretische Informatik. Kontakt: Heiko Stamer. 8 Einreichungen. 1. Kryptowochenende am Juli 2006: Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 14 Einreichungen und 21 angemeldete Teilnehmer. 4. Kryptotag am 11. Mai 2006: Ruhr Universität Bochum, Horst-Görtz Institut. Kontakt: Ulrich Greveler. 10 Einreichungen und 32 angemeldete Teilnehmer. 3. Kryptotag am 15. September 2005: Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldeten Teilnehmer. 2. Kryptotag am 31. März 2005: Universität Ulm, Abteilung fâ ur Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldeten Teilnehmer. 1. Kryptotag am 1. Dezember 2004: Universität Mannheim, Lehrstuhl für Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldeten Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (K.U.Leuven, Belgien) verantwortlich für die Organisation der Kryptotage. Für eventuelle Rückfragen bitte an sie wenden. 14

136 Tagungsband zum 8. Kryptotag Workshop der Fachgruppe Kryptographie in der Gesellschaft für Informatik Universität Tübingen 11. April 2008 c 2008 Michael Beiter, Claudia Schmidt, Anja Korsten

137 Universität Tübingen, Wilhelm-Schickard-Institut für Informatik, Arbeitsbereich Diskrete Mathematik Technischer Report Nr. WSI ISSN

138 Programm 9:30 Kaffee und Gebäck 10:00 Begrüßung 10:15 Session 1: 10:15 Vladimir Kolesnikov, Thomas Schneider, Volker Strehl Practical Secure Function Evaluation 10:35 Gerald Madlmayr, Josef Scharinger Sicheres Over-the-air Management von SmartCard Applikationen in einem Near Field Communication Ökosystem 10:55 Kaffeepause 11:20 Session 2: 11:20 Ralph Holz, Heiko Niedermayer A Protocol for Inter-Domain Authentication With a Trust-Rating Mechanism 11:40 Simon Fischer, Willi Meier, Dirk Stegemann Equivalent Representations of the F-FCSR Keystream Generator Family 12:00 Mittagspause 1 14:00 Invited Talk: Sebastian Schreiber (SySS GmbH) 15:00 Kaffeepause 15:20 Rump Session 15:30 Session 3: Live Hacking - Angriffe live demonstriert 15:30 Michael Gorski Related-Key Boomerang Attacks on 7, 8 and 9 Round Reduced Versions of AES :50 Bernd Borchert Trojaner-sicheres Online-Banking mit der Permutations-TAN 16:10 Franz Scheerer Eine kryptographische Methode, welche informationstheoretisch sicher ist und nachweislich nicht gebrochen werden kann 16:30 Verabschiedung 16:45 Fachgruppentreffen (Raum A302) 18:00 Ausklang des Kryptotages bei schwäbischen Spezialitäten im Gasthaus s Urige 2 2 Mensa Wilhelmstrasse, Hinfahrt: Linie 2 (Richtung Omnibusbahnhof), 12:09 oder 12:24 (bis Haltestelle Hölderlinstr.). Treffpunkt für die Rückfahrt: 13:25 im Foyer der Mensa (Linie 2 fährt ab Haltestelle Geschwister-Scholl-Platz um 13:43.) 2 Treffpunkt: Bushaltestelle Nonnenhaus, 17:55. Hinfahrt: Linie 2 (Richtung Omnibusbahnhof), 17:39. 1

139 2

140 Inhaltsverzeichnis Practical Secure Function Evaluation Vladimir Kolesnikov, Thomas Schneider, Volker Strehl Sicheres Over-the-air Management von SmartCard Applikationen in einem Near Field Communication Ökosystem. Gerald Madlmayr, Josef Scharinger A Protocol for Inter-Domain Authentication With a Trust-Rating Mechanism Ralph Holz, Heiko Niedermayer Equivalent Representations of the F-FCSR Keystream Generator Family Simon Fischer, Willi Meier, Dirk Stegemann Related-Key Boomerang Attacks on 7, 8 and 9 Round Reduced Versions of AES-192 Michael Gorski Trojaner-sicheres Online-Banking mit der Permutations-TAN Bernd Borchert Eine kryptographische Methode, welche informationstheoretisch sicher ist und nachweislich nicht gebrochen werden kann Franz Scheerer

141 Practical Secure Function Evaluation Vladimir Kolesnikov, Thomas Schneider, Volker Strehl Friedrich-Alexander Universität Bell Laboratories Erlangen-Nürnberg 600 Mountain Ave. Murray Hill, NJ Germany USA Since the first publication of Yao [Yao86], Secure Function Evaluation (SFE) is a well-researched problem. Continuing advances in available computational power and communication have made secure computation of many useful functions affordable. Recent work like Fairplay [MNPS04] demonstrate practicability of general SFE. This thesis focuses on several practical aspects of SFE. Our new improved SFE protocol allows free evaluation of XOR gates and is provably secure against semi-honest adversaries in the random oracle model - the same assumptions that Fairplay relies on. The protocol merges elements of the information-theoretic SFE protocol GESS [Kol05] with Fairplay. This results in substantial performance improvements of 50% for many important circuit structures like addition or number comparison. SFE is extended to allow the evaluated function to be secret and only known by one party, called SFE of private functions (PF-SFE). These settings occur naturally in applications like no-fly-list-, credit report-, or medical history checking. It is known that PF-SFE can easily be reduced to SFE of universal circuits (UC). We give a practical UC construction [KS08] that is up to 50% smaller than the best UC of Valiant [Val76] when used in today s PF-SFE. FairplayPF was implemented as extension of Fairplay to demonstrate practicability of PF-SFE based on the new UC construction. Using the improved SFE protocol, UC-based PF-SFE can be improved by another factor of 4. Besides these circuit-based approaches for SFE and PF-SFE new protocols for SFE and PF-SFE of functions represented as Ordered Binary Decision Diagrams (OBDDs) are given that are based on [KJGB06]. This SFE protocol for OBDDs is extended to the malicious model and shown how to obtain a PF-SFE protocol for OBDDs at the cost of a small overhead only. The results of this thesis substantially improve general SFE for many practical functions and demonstrate practicability of general PF-SFE for small functions. References [Kol05] Vladimir Kolesnikov. Gate evaluation secret sharing and secure one-round two-party computation. In Advances in Cryptology ASIACRYPT05, volume 3788 of LNCS, pages Springer, [KS08] Vladimir Kolesnikov and Thomas Schneider. A practical universal circuit construction and secure evaluation of private functions. In Financial Cryptography and Data Security, FC08, LNCS. Springer, [KJGB06] Louis Kruger, Somesh Jha, Eu-Jin Goh, and Dan Boneh. Secure function evaluation with ordered binary decision diagrams. In CCS, pages ACM Press, [MNPS04] Dahlia Malkhi, Noam Nisan, Benny Pinkas, and Yaron Sella. computation system. In USENIX, [Val76] Fairplay - a secure two-party Leslie G. Valiant. Universal circuits (preliminary report). In Proc. 8th ACM Symp. on Theory of Computing, pages , New York, NY, USA, ACM Press. [Yao86] Andrew C. Yao. How to generate and exchange secrets. In Proc. 27th IEEE Symp. on Foundations of Comp. Science, pages , Toronto, IEEE. 4

142 Sicheres Over-the-air Management von SmartCard Applikationen in einem Near Field Communication Ökosystem. Gerald Madlmayr, Josef Scharinger NFC Research Lab Hagenberg, Johannes Kepler Universität Linz Near Field Communication (NFC) ist eine kontaktlose Übertragungstechnologie, um Daten über eine Distanz von bis zu 10 cm austauschen zu können. NFC ist kompatibel zum Smartcard- Standard ISO und wird bevorzugt in Mobiltelefonen eingesetzt, damit diese die Aufgaben von kontaktlosen Smartcards übernehmen können. Durch die Integration diese Technologie in ein Mobiltelefon besteht die Möglichkeit, den Smartcard-Chip, auch als sicheres Element bezeichnet, nicht nur über die kontaktlose Schnittstelle, sondern auch über ein Weitverkehrsnetz wie GSM oder UTMS anzusprechen [GSMA07]. So können zu jedem beliebigen Zeitpunkt entsprechende Smartcard-Applikationen over the air (OTA) in den sicheren Smartcard-Chip des Telefons geladen werden. Mit diesem System wird es in Zukunft möglich sein, auf Smartcards zu verzichten und Applikationen in einer sicheren Art und Weise auf einem NFC-Mobiltelefon zu jedem beliebigen Zeitpunkt zu installieren und zu verwalten. Diese Methode spart Zeit und Prozessaufwand im Zuge der Ausstellung und erlaubt zudem auch ein rasches Handeln im Falle von Verlust oder Diebstahl. Zudem besteht für den Benutzer die Möglichkeit, Daten, die im sicheren Element des Telefons ablegt sind, am Bildschirm des mobilen Endgerätes darzustellen. Das Gerät entspricht sozusagen einer Smartcard mit einem Display. Aus diesem Grund ist eine entsprechende Infrastruktur notwendig, welche die Fernverwaltung eines oder mehrerer Smartcard-Chips und den darauf befindlichen Applikationen in einem NFC- Mobiltelefon in einer sicheren Art und Weise regelt. Schwerpunktmäig wird die Zertifizierung der Instanzen und Komponenten des Systems, deren Funktionen sowie die Absicherung der Kommunikation und Protokolle diskutiert. Hierfür wird auf eine Public-Key-Infrastruktur zurückgegriffen. Zudem werden die Rollen der Service Anbieter (zb Bank), Mobilfunknetzbetreiber sowie Stellen, die sichere Elemente (zb SIM Karten, sichere SD-Karten) ausgeben bzw. verwalten [GN07], in einem solchen Ökosystem analysiert sowie deren Anforderungen definiert. Des Weiteren wird erläutert, welche Kriterien von einer Applikation am Mobiltelefon erfüllt sein müssen, damit diese auf das sichere Element zugreifen kann, um dessen Inhalt anzuzeigen oder Daten darin verändern zu können. Ein besonderes Schutzbedürfnis für ein derartiges System leitet sich insofern ab, als Smartcard- Applikationen häufig im Zahlungsverkehr und zur Personenidentifikation verwendet werden. Um die Privatsphäre der Anwender zu schützen, ist neben einer sensiblen Behandlung der Daten auch die Absicherung der Datenübertragung und der Verwaltung erforderlich. Literatur [GSMA07] Mobile NFC Technical Guidelines, GSM Association, NFC working group, February [GN07] SIM as Secure Key Storage in Communication Networks, Gyorgy Kalman and Josef Noll, in Prodceedings of ICWMC,

143 A Protocol for Inter-Domain Authentication With a Trust-Rating Mechanism Ralph Holz, Heiko Niedermayer University of Tübingen We present an authentication scheme and new protocol for domain-based scenarios with interdomain authentication. It is primarily intended for domain-structured Peer-to-Peer systems, but is applicable in any domain scenario. We make use of Trusted Third Parties in the form of Domain Authentication Servers (DAS) in each domain. These act on behalf of their clients, resulting in a four-party protocol. Our scheme differs from traditional protocols in two regards. First, its design is such that communication between domains is strictly limited to two channels: communication between client principals, and communication between the DAS of two domains. This decouples domains as much as possible and is of advantage for scalability. Second, it also enables us to introduce a mechanism that addresses the (frequent) use case of domains that have no a priori knowledge of each other, where the DAS have not been able to securely exchange their keys. Our protocol is designed to fulfill the goals Authentication as Fresh Injective Agreement [2], Key Establishment, Fresh Session Key, Mutual Belief in Key, and (optionally) Perfect Forward Secrecy. We sketch the protocol flow: an initiator B obtains a Credential from its DAS S B and presents it to the responder A. A cannot verify the Credential and passes it to her DAS S A. S A is in possession of S B s public key and verifies the Credential. The freshness is confirmed with a query to S B, and S A also obtains an Authentication Token for A to authenticate to B. The security of our protocol lies solely with the properties of the channel between the DAS. If the channel is secure, our protocol can provide secure authentication. If not, it follows from the work of Boyd [1] that the authentication cannot be made secure. We address this special case and exploit domain structures in order to provide what we call trust-rated authentication. The DAS also communicate their knowledge about the inter-domain channel and about the other domain to their clients in the form of a Trust Token. If S B is the DAS for B and S A the DAS for A, the information that S B passes to B is: 1) whether there exists a secure channel to S A and 2) knowledge about S A, the other domain, and A. The first is a simple yes/no statement. Knowledge is delivered in the form of a set of properties, e.g. details on channel, a-priori knowledge, observations, etc. The clients evaluate this information and decide whether to proceed. Such knowledge can provide basic trust or distrust into the other party. We have verified that our protocol achieves the set security goals and evaluated protocol security with the AVISPA model checker [3]. Due to the size of the state space, this was limited to systems with three concurrent sessions. The evaluation showed our protocol to be secure for these scenarios. Our scheme allows key revocation without the need to distribute Key Revocation Lists. It can be seen as a PKI that is distributed over several domains, yet operates without a single global authority. References [1] Boyd, C.: Security architecture using formal methods. IEEE Journal on Selected Topics in Communications 11 (1993) [2] Lowe, G.: A hierarchy of authentication specifications. In: Proceedings of the 10th IEEE Computer Security Foundations Workshop, Rockport, MA, USA (CSFW 97). (1997) [3] The AVISPA Project: Automated Validation of Internet Security Protocols and Applications (homepage). (January 2008) 6

144 Equivalent Representations of the F-FCSR Keystream Generator Family Simon Fischer, Willi Meier, Dirk Stegemann FHNW Windisch Switzerland University of Mannheim Mannheim Germany Linear feedback shift registers (LFSRs) are widely used devices for producing pseudorandom sequences with good statistical properties. When used in stream ciphers, the inherent linear structure of their output has to be overcome with additional methods such as nonlinear filter or combination functions. Another approach is to replace LFSRs by nonlinear feedback shift registers (NFSRs), i.e., feedback shift registers with nonlinear feedback functions. The F-FCSR stream cipher family [1,2], which has made its way into the final phase of the EUfunded stream cipher project estream [4], is based on a special type of NFSRs called feedback with carry shift registers (FCSRs) [3,6,7]. In contrast to the linear mod 2-addition used in LFSRs, the feedback function of FCSRs is based on a (nonlinear) addition with carry. Similarly to LFSRs, feedback with carry shift registers can be equivalently represented in Galois architecture and in Fibonacci architecture [5]. We show how this correspondence can be used to derive three additional equivalent representations of the F-FCSR ciphers. How these representations may be exploited for cryptanalytic attacks is subject to further research. References [1] F. Arnault and T. P. Berger, Design and Properties of a New Pseudorandom Generator Based on a Filtered FCSR Automaton. In IEEE Transactions on Information Theory, 54(11): , [2] F. Arnault, T. P. Berger, and C. Lauradoux. Update on F-FCSR Stream Cipher. In estream, ECRYPT Stream Cipher Project, Report 2006/025. See also [4]. [3] R. Couture and P. L Ecuyer. On the lattice structure of certain linear congruential sequences related to AWC/SWB generators. In Math. Comput., 62(206): , [4] estream - The ECRYPT Stream Cipher Project - Phase 3. See [5] M. Goresky and A. Klapper. Fibonacci and Galois Representations of Feedback-With-Carry Shift Registers. In IEEE Transactions on Information Theory, 48(11): , [6] A. Klapper and M. Goresky. Feedback Shift Registers, 2-Adic Span, and Combiners with Memory. In Journal of Cryptology, 10: , [7] G. Marsaglia and A. Zaman. A new class of random number generators. Annals of Appl. Prob., 1(3): ,

145 Related-Key Boomerang Attacks on 7, 8 and 9 Round Reduced Versions of AES-192 Michael Gorski Bauhaus-University of Weimar In this paper we present the first implementable attack on 7-round reduced AES-192. We use the related-key boomerang attack to obtain an attack which has a times lower time complexity than the best known attack so far. It also has a very low data complexity which makes it possible for implementation. We also extend this attack to an 8 and 9-round attack on AES-192 which has the lowest time complexity than previously known attacks. Keywords: Differential Cryptanalysis, Block Ciphers, AES, Related-Key Boomerang Attack. The related-key boomerang attack was published first in [1], but not used to break reduced versions of AES. It utilizes the fact, that the boomerang attack uses short differentials and thus the fewer diffusion of differences in the subkeys can better be used than in the original related-key differential attack.we present the first related-key boomerang attack on AES-192. Another application of the ordinary boomerang attack on AES-128 can be found in [3], which can break up to 5 and 6 out of 10 rounds of AES-128. Our related-key boomerang attack uses 4 related keys to break 7 and 8 out of 12 rounds of AES-192. Up to now our attack on 7-round AES-192 is the best known attack on AES in terms of data and time complexity. We decrease the complexity of the best known attack [2] on 7-round AES-192 by a factor of for the data and a factor of for the time complexity. Our 9-round attack has a 2 36 times lower time complexity than previous attacks. References [1] Eli Biham, Orr Dunkelman, and Nathan Keller. Related-Key Boomerang and Rectangle Attacks. In Ronald Cramer, editor, EUROCRYPT, volume 3494 of Lecture Notes in Computer Science, pages Springer, [2] Eli Biham, Orr Dunkelman, and Nathan Keller. Related-Key Impossible Differential Attacks on 8- Round AES-192. In David Pointcheval, editor, CT-RSA, volume 3860 of Lecture Notes in Computer Science, pages Springer, [3] Alex Biryukov. The Boomerang Attack on 5 and 6-Round Reduced AES. In Hans Dobbertin, Vincent Rijmen, and Aleksandra Sowa, editors, AES Conference, volume 3373 of Lecture Notes in Computer Science, pages Springer,

146 Trojaner-sicheres Online-Banking mit der Permutations-TAN Bernd Borchert Universität Tübingen Online Banking mit dem PIN/iTAN Verfahren ist nicht trojaner-sicher: ein auf den Rechner des Benutzers eingeschleppter Virus kann die PIN abhören ( keylogger ) und Überweisungen fälschen ( Man-in-the-Middle ). Es wird eine einfache Abänderung des Verfahrens vorgestellt, die trojaner-sicher ist: anstatt einer itan Liste bekommt der Bankkunde eine Liste mit Permutationen der 10 Ziffern und Positionsangaben. 9

147 Eine kryptographische Methode, welche informationstheoretisch sicher ist und nachweislich nicht gebrochen werden kann Franz Scheerer Scheerer Software Wiesbaden Deutschland Seit dem berühmten Artikel von Claude Shannon [Sh49] wissen wir, dass es eine kryptographische Methode gibt, welche informationstheoretisch sicher ist und nachweislich nicht gebrochen werden kann. Diese Arbeit untersucht praktische Schwierigkeiten, die möglicherweise bei der Implementierung dieser Methode auftreten könnten. Es zeigt sich dabei, dass die vermeintlichen Probleme, ein Schlüssel, der so lang ist wie die Nachricht selbst, nur einmal verwendet werden darf und geheim gehalten werden muss, mit heute verfügbaren Speichermedien kein ernstes Problem darstellt. Kern des Verfahrens ist eine kryptographisch sicherer Algorithmus zur Erzeugung einer zufälligen, also nicht vorhersehbaren statistisch gleichverteilten Folge von Bits oder Bytes. Derartige Verfahren können mit geringem technischen Aufwand auf der Grundlage von der sicheren Hashfunktion SHA-1 [SHA02] als Webapplikation auf Basis der weit verbreiteten Programmiersprache PHP leicht realisiert werden. Zugleich können entsprechende Verfahren ebenfalls problemlos in den Programmiersprachen C, C++ in einer PC-Anwendung implementiert werden. Die Implementierung ist dabei mit minimalem Programmieraufwand verknüpft, so dass die Programme leicht überschaubar sind und auf korrekte Funktion überprüft werden können. Es kann dabei mit hoher Sicherheit ausgeschlossen werden, dass die Programme unerwünsche Zusatzfunktionalitäten, vergleichbar Tojanischen Pferden, enthalten. Als kryptographischer Zufallsgenerator hat der Algorithmus RC4, der sich durch extrem einfache Implementierung auszeichnet, weite Verbreitung. Doch allein durch die Initialisierung der Zustandsvariablen bietet dieser einfache Algorithmus bei Schlüsselwerten mit festen und variablen Anteilen reale Angriffsmöglichkeiten [SF02]. Durch Verknüpfung mehrerer Zufallsgeneratoren durch verschiedene Formen der Addition lassen sich weitere Generatoren mit verbesserten Eigenschaften entwickeln. Die statistische Gleichverteilung wird an konkreten Beispielen systematisch untersucht. Literatur [Sh49] Claude Shannon. Communication Theory of Secrecy Systems. Bell System Technical Journal, Vol 28, Oktober 1949 [SHA02] National Institute of Standards and Technology (NIST). SECURE HASH STANDARD. Federal Information Processing Standards Publication 180-2, August [SF02] Scott Fluhrer, Itsik Mantin, and Adi Shamir. Weaknesses in the Key Scheduling Algorithm of RC4. SAC,

148 11

149 Teilnehmer Mathias Bartl (Uni Michael Beiter (Uni Tübingen) Jonas Bochtler (Uni Rainer Boie (Uni Tübingen) Bernd Borchert (Uni Finja Büchel (Uni Tübingen) Ingo Czerwinski (Uni Tübingen) ingo Tobias Eibach (Uni Ali Fessi (Uni Tübingen) Marc Fouquet (Uni Tübingen) Markus Geyer (Uni Tübingen) Michael Gorski (Uni Weimar) Jens Gramm (Robert-Bosch GmbH) Peter Hauck (Uni Tübingen) Florian Heimgärtner (Uni Tübingen) Mathias Herrmann (Uni Marius Hofmeister (Uni Tübingen) Ralph Holz (Uni Tübingen) Michael Hubig (AMK Bonn) Anja Korsten (Uni Berthold Laschinger (HS Furtwangen) Moritz Laschinger Daniel Loebenberger (Uni Bonn) Gerald Madlmayr (FH Hagenberg) Christian Morlok (Uni Tübingen) Thomas Nestmeyer (Uni Tübingen) Heiko Niedermayer (Uni Tübingen) Emilios Papadopoulos (Uni Bochum) Sebastian Pape (Uni Kassel) Enrico Pilz (Uni Ulm) Deike Priemuth-Schmid (Uni Luxemburg) Klaus Reinhardt (Uni Tübingen) Franz Scheerer (Scheerer Software, Wiesbaden) Claudia Schmidt (Uni Tübingen) Thomas Schneider (Uni Bochum) Jamshid Shokrollahi (Robert-Bosch GmbH) Mandeep Singh (Uni Tübingen) Heiko Stamer (Uni Dirk Stegemann (Uni Tobias Thilde Gunnar Völkel (Uni Jörg Willmann Christopher Wolf (Uni Bochum)

150 13

151 Der Kryptotag ist eine zentrale Aktivität der GI-Fachgruppe Angewandte Kryptologie. Er ist eine wissenschaftliche Veranstaltung im Bereich der Kryptologie und von der organisatorischen Arbeit der Fachgruppe getrennt. Grundgedanke des Kryptotages ist, dass er inklusive Anreise wirklich nur einen Tag dauert und Nachwuchswissenschaftlern, etablierten Forschern und Praktikern auf dem Gebiet der Kryptologie die Möglichkeit bieten, Kontakte über die eigene Universität hinaus zu knüpfen. Die Vorträge können ein breites Spektrum abdecken, von noch laufenden Projekten, die ggf. erstmals einem breiteren Publikum vorgestellt werden werden, bis zu abgeschlossenen Forschungsarbeiten, die zeitnah auch auf Konferenzen präsentiert wurden bzw. werden sollen oder einen Schwerpunkt der eigenen Diplomarbeit oder Dissertation bilden. Die eingereichten Abstracts werden gesammelt und als technischer Bericht veröffentlicht. Es handelt sich damit um eine zitierfähige Arbeit. Sie können von den Seiten der Fachgruppe herunter geladen werden. Geplante Kryptotage 9. Kryptotag 10. November 2008, Fachhochschule Gelsenkirchen Bisherige Kryptotage 8. Kryptotag am 11. April 2008 Universität Tübingen, WSI für Informatik, Diskrete Mathematik. Kontakt: Michael Beiter, Claudia Schmidt, Anja Korsten. 7 Einreichungen und 43 angemeldete Teilnehmer. 7. Kryptotag am 9. November 2007 Bonn-Aachen International Center for Information Technology. Kontakt: Michael Nüsken und Daniel Loebenberger. 9 Einreichungen und 36 angemeldete Teilnehmer. 6. Kryptotag am 19. Februar Universität des Saarlandes, Information Security and Cryptography Group und Sirrix AG. Kontakt: Michael Backes und Ammar Alkassar. 8 Einreichungen und 30 angemeldete Teilnehmer. 5. Kryptotag am 11. September Universität Kassel, Fachbereich Mathematik/Informatik, Theoretische Informatik. Kontakt: Heiko Stamer. 8 Einreichungen und 22 angemeldete Teilnehmer. 1. Kryptowochenende am Juli Tagungszentrum Kloster Bronnbach der Universität Mannheim. Kontakt: Frederik Armknecht und Dirk Stegemann. 14 Einreichungen und 21 angemeldete Teilnehmer. 4. Kryptotag am 11. Mai Ruhr Universität Bochum, Horst-Görtz Institut. Kontakt: Ulrich Greveler. 10 Einreichungen und 32 angemeldete Teilnehmer. 3. Kryptotag am 15. September Technische Universität Darmstadt, Theoretische Informatik. Kontakt: Ralf-Philipp Weinmann. 13 Einreichungen und 35 angemeldete Teilnehmer. 2. Kryptotag am 31. März Universität Ulm, Abteilung für Theoretische Informatik. Kontakt: Wolfgang Lindner und Christopher Wolf. 10 Einreichungen und 26 angemeldete Teilnehmer. 1. Kryptotag am 1. Dezember Universität Mannheim, Theoretische Informatik. Kontakt: Stefan Lucks und Christopher Wolf. 15 Einreichungen und 37 angemeldete Teilnehmer. Innerhalb der Fachgruppe für Angewandte Kryptologie sind Stefan Lucks (Universität Mannheim) und Christopher Wolf (Ruhr-Universität Bochum) verantwortlich für die Organisation der Kryptotage. Für eventuelle Rückfragen bitte an sie wenden.

152

153 Fault Attacken und Gegenmaßnahmen Marcel Medwed TU Graz So far many software countermeasures against fault attacks have been proposed. However, most of them are tailored to a specific cryptographic algorithm or focus on securing the processed data only. In this work we present a generic and elegant approach by using a highly fault secure algebraic structure. This structure is compatible to finite fields and rings and preserves its error detection property throughout addition and multiplication. Additionally, we introduce a method to generate a fingerprint of the instruction sequence. Thus, it is possible to check the result for data corruption as well as for modifications in the program flow. This is even possible if the order of the instructions is randomized. Furthermore, the properties of the countermeasure allow the deployment of error detection as well as error diffusion. We point out that the overhead for the calculations and for the error checking within this structure is reasonable and that the transformations are efficient. In addition we discuss how our approach increases the security in various kinds of fault scenarios. Dazu kommt noch eine kleine Einführung in Side channel und Fault Attacks. 1 of 55

154 1 Introduction Parallel Generation of l Sequences 1 Cedric Lauradoux and Andrea Röck Princeton University Department of electrical engineering Princeton, NJ 08544, USA claurado@princeton.edu INRIA Paris-Rocquencourt Team SECRET Le Chesnay Cedex, France andrea.roeck@inria.fr The generation of pseudo-random sequences at a high rate is an important issue in modern communication schemes. The representation of a sequence can be scaled by decimation to obtain parallelism and more precisely a sub-sequences generator. Sub-sequences generators and therefore decimation have been extensively used in the past for linear feedback shift registers (LFSRs). However, the case of automata with a non linear feedback is still in suspend. We have studied how to transform a feedback with carry shit register (FCSR) into a sub-sequences generator. We examine two solutions for this transformation, one based on the decimation properties of l-sequences, i.e. FCSR sequences with maximal period, and the other one based on multiple steps implementation. We show that the solution based on the decimation properties leads to much more costly results than in the case of LFSRs. For the multiple steps implementation, we show how the propagation of carries affects the design. The synthesis of shift registers consists in finding the smallest automaton able to generate a given sequence. This problem has many applications in cryptography, sequences and electronics. The synthesis of a single sequence with the smallest linear feedback shift register is achieved by the Berlekamp-Massey [Mas69] algorithm. In the case of FCSRs, we can use algorithms based on lattice approximation [KG97] or on Euclid s algorithm [ABN04]. We are interested in the following issue in the synthesis of shift registers: given an automaton generating a sequence S, how to find an automaton which generates in parallel the sub-sequences associated to S. We will refer to this problem as the sub-sequences generator problem. We aim to find the best solution to transform a 1- bit output pseudo-random generator into a multiple outputs generator. In particular, we investigate this problem when S is generated by a feedback with carry shit register (FCSR) with a maximal period, i.e. S is an l-sequence. This class of pseudo-random generators was introduced by Klapper and Goresky in [KG93]. FCSRs and LFSRs are very similar in terms of properties [GK02, GKnt]. However, FCSRs have a non-linear feedback which is a significant property to thwart algebraic attacks [CM03] in cryptographic applications [ABM07]. The design of sub-sequences generators has been investigated in the case of LFSRs [LE71] and two solutions have been proposed. The first solution [Rue86] is based on the classical synthesis of shift registers, i.e. the Berlekamp-Massey algorithm, to define each sub-sequence. The second solution [LE71] is based on a multiple steps design of the LFSR. We have applied those two solutions to FCSRs. Our contributions are as follows: We explore the decimation properties of l-sequences for the design of a sub-sequences generator by using an FCSR synthesis algorithm. We show how to implement a multiple steps FCSR in Fibonacci and Galois configuration. 1 This extended abstract is a short version of an article accepted at SETA 08, Lexington, Kentucky, USA. 1 2 of 55

155 2 Motivation Decimation is the main tool to transform a 1-bit output generator into a sub-sequences generator. This allows us to increase the throughput of a pseudo-random sequence generator (PRSG). Let S = (s 0, s 1, s 2, ) be an infinite binary sequence of period T, thus s j {0, 1} and s j+t = s j for all j 0. For a given integer d, a d decimation of S is the set of sub-sequences defined by: S i d = (s i, s i+d, s i+2d,, s i+jd, ) where i [0, d 1] and j = 0, 1, 2,. Hence, a sequence S is completely described by the sub-sequences: Sd 0 = (s 0, s d, ) Sd 1 = (s 1, s 1+d, )... S d 1 d = (s d 1, s 2d 1, ). A single automaton is often used to generate the pseudo-random sequence S. In this case, it is difficult to achieve parallelism. The decomposition into sub-sequences overcomes this issue as shown by Lempel and Eastman in [LE71]. Each sub-sequence is associated to an automaton. Then, the generation of the d sub-sequences of S uses d automata which operate in parallel. Parallelism has two benefits, it can increase the throughput or reduce the power consumption of the automaton generating a sequence. Throughput The throughput T of a PRSG is defined by: T = n f, with n is the number of bits produced in every cycle and f is the clock frequency of the PRSG. Usually, we have n = 1, which is often the case with LFSRs. The decimation achieves a very interesting tradeoff for the throughput: T d = d γf with 0 < γ 1 the degradation factor of the original automaton frequency. The decimation provides an improvement of the throughput if and only if γd > 1. It is then highly critical to find good automata for the generation of the sub-sequences. In an ideal case, we would have γ = 1 and then a d-decimation would imply a multiplication of the throughput by d. Power consumption The power consumption of a CMOS device can be estimated by the following equation: P = C Vdd 2 f, with C the capacity of the device and V dd the supply voltage. The sequence decimation can be used to reduce the frequency of the device by interleaving the sub-sequences. The sub-sequences generator will be clocked at frequency γf d and the outputs will be combined with a d-input multiplexer clocked at frequency γf. The original power consumption can then be reduced by the factor γ d, where γ must be close to 1 to guarantee that the final representation of S is generated at frequency f. The study of the γ parameter is out of the scope of this work since it is highly related to the physical characteristics of the technology used for the implementation. In the following, we consider m-sequences and l-sequences which are produced respectively by LFSRs and FCSRs. We detail different representations of several automata. We denote by x i a memory cell and by (x i ) t the content of the cell x i at time t. The internal state of an automaton at time t is denoted by X t. 2 3 of 55

156 3 Previous Results An LFSR is an automaton which generates linear recursive sequences. A detailed description of this topic can be found in the monographs of Golomb and McEliece [Gol81, McE87]. The decimation of LFSR sequences has been used in cryptography in the design of new stream ciphers [MR84]. There exists two approaches to use decimation theory to define the automata associated to the sub-sequences. Construction using LFSR synthesis. This first solution associates an LFSR to each subsequence. It is based on well-known results on the decimation of LFSR sequences. It can be applied to both Fibonacci and Galois representation without any distinction. Theorem 3.1 ([Zie59, Rue86]). Let S be a sequence produced by an LFSR whose characteristic polynomial Q(x) is irreducible in F 2 of degree m. Let α be a root of Q(x) and let T be the period of Q(x). Let Sd i be a sub-sequence resulting from the d-decimation of S. Then, Si d can be generated by an LFSR with the following properties: The minimum polynomial of α d in F 2 m is the characteristic polynomial Q (x) of the resulting LFSR. The period T of Q (x) is equal to T gcd(d,t ). The degree m of Q (x) is equal to the multiplicative order of Q(x) in Z T. In practice, the characteristic polynomial Q (x) can be determined using the Berlekamp-Massey algorithm [Mas69]. The sub-sequences are generated using d LFSRs defined by the characteristic polynomial Q (x) but initialized with different values. In the case of LFSRs, the degree m must always be smaller or equal to m. Construction using a multiple steps LFSR. This method was first proposed by Lempel and Eastman [LE71]. It consists in clocking the LFSR d times in one clock cycle by changing the connections between the memory cells and by some duplications of the feedback function. We obtain a network of linearly interconnected shift registers. All the cells x i of the original LFSR, such that i mod d = k, are gathered to form a sub-shift register, where 0 k d 1. This is the basic operation to transform a LFSR into a sub-sequences generator with a multiple steps solution. In the case of the Fibonacci setup, we apply the update function f at the states X t, X t+1,..., X t+d 1 to obtain the new feedback functions. In the case of the Galois setup, we have to define the new feedback function at each feedback position. Comparison. Let wt(q(x)) denote the Hamming weight of the polynomial Q(x), i.e. the number of non-zero monomials. The method based on LFSR synthesis proves that there exists a solution for the synthesis of the sub-sequences generator. With this solution, both memory cost ( d m memory cells ) and gate number ( d wt(q ) logic gates) depends on the decimation factor d. The method proposed by Lempel and Eastman [LE71] uses a constant number of memory cells ( m ) for the synthesis of the sub-sequences generator, the number of logic gates ( d wt(q) ) still depends on d. 3 4 of 55

157 4 Sub-Sequences Generators and m-sequences FCSRs were introduced by Klapper and Goresky in [KG93]. Instead of addition modulo 2, FCSRs use additions with carry, which means that they need additional memory to store the carry. Their non linear update function makes them particularly interesting for areas where linearity is an issue, like for instance stream ciphers. As for the LFSRs, there exists a Fibonacci and a Galois setup [GK02]. The contribution of our work is to apply the two methods used in the previous section on the case of l-sequences, i.e.sequences with maximal period. Construction using FCSR synthesis. There exists algorithms based on Euclid s algorithm [ABN04] or on lattice approximation [KG97], which can determine the smallest FCSR to produce Sd i. These algorithms use the first k bits of Si d to find h and q such that h /q is the 2-adic representation of the sub-sequence, q < h 0 and gcd(q, h ) = 1. Subsequently, we can find the feedback positions and the initial state of the FCSR in Galois or Fibonacci architecture. The value k is in the range of twice the linear 2-adic complexity of the sequence. For our new sequence Sd i, let h and q define the values found by one of the algorithms mentioned above. By T and T, we mean the periods of respectively Sd i and S. For the period of the decimated sequences, we can make the following statement, which is true for all periodic sequences. Lemma 4.1. Let S = (s 0, s 1, s 2,...) be a periodic sequence with period T. For a given d > 1 and 0 i d 1, let S i d be the decimated sequence with period T. Then, it must hold: If gcd(t, d) = 1 then T = T. T T gcd(t, d). (1) In the case of gcd(t, d) > 1, the real value of T might depend on i, e.g. for S being the 2-adic representation of 1/19 and d = 3 we have T/gcd(T, d) = 6, however, for S 0 3 the period T = 2 and for S 1 3 the period T = 6. A critical point in this approach is that the size of the new FCSR can be exponentially bigger than the original one. In general, we only know that for the new q it must hold that q 2 T 1, where T can be as big as T/gcd(T, d). In the case of gcd(t, d) = 1, we know from [GKMS04] that 2 T/ Based on a conjecture in [GK97] we can even assume that q is always bigger than q, if gcd(t, d) = 1. This means that the space complexity of this method is much worse than for the original FCSR which is also an interesting aspect for decimation attacks on FCSR sequences. Construction using a multiple steps FCSR. We will apply the same technique than for the LFSR, however this time we have to take care of the carry path, i.e. we need the value of the carry at time t to compute the carry at time t + 1. The computation of these subsequent carry bits seems to reduce the effectiveness of the decimation. However, it can be done efficiently by using n-bit ripple carry adders, which are well-known arithmetic circuits. An example of a multiple steps Galois FCSR can be found in Figure of 55

158 1-decimation x 3 x 2 x 1 x 0 2-decimation x 3 x 1 c0 a 1 b 1 carry out r1 c0 x 2 x 0 a 0 b 0 r0 carry in (a) Galois FCSR for q = 19. Figure 1: Example for a Galois FCSR with q = 19. (b) 2-bit ripple carry adder. References [ABM07] Francois Arnault, Thierry P. Berger, and Marine Minier. On the security of FCSR-based pseudorandom generators. In State of the Art of Stream Ciphers - SASC 2007, February available at [ABN04] Francois Arnault, Thierry P. Berger, and A. Necer. Feedback with Carry Shift Registers synthesis with the Euclidean Algorithm. IEEE Transactions on Information Theory, 50(5), [CM03] Nicolas Courtois and Willi Meier. Algebraic Attacks on Stream Ciphers with Linear Feedback. In Advances in Cryptology - EUROCRYPT 2003, Lecture Notes in Computer Science 2656, pages Springer Verlag, [GK97] Mark Goresky and Andrew Klapper. Arithmetic crosscorrelations of feedback with carry shift register sequences. IEEE Transactions on Information Theory, 43(4): , [GK02] Mark Goresky and Andrew Klapper. Fibonacci and Galois representations of feedback-with-carry shift registers. IEEE Transactions on Information Theory, 48(11), [GKMS04] Mark Goresky, Andrew Klapper, Ram Murty, and Igor Shparlinski. On Decimations of l- Sequences. SIAM Journal of Discrete Mathematics, 18(1): , [GKnt] Mark Goresky and Andrew Klapper. Algebraic Shift Register Sequences. preprint. [Gol81] Solomon W. Golomb. Shift Register Sequences. Aegean Park Press, [KG93] Andrew Klapper and Mark Goresky. 2-adic shift registers. In Fast Software Encryption - FSE 93, Lecture Notes in Computer Science 809, pages Springer Verlag, [KG97] Andrew Klapper and Mark Goresky. Feedback shift registers, 2-adic span, and combiners with memory. Journal of Cryptology, 10: , [LE71] Abraham Lempel and Willard L. Eastman. High Speed Generation of Maximal Length Sequences. IEEE Transactions on Computer, 2: , [Mas69] James L. Massey. Shift-register synthesis and BCH decoding. IEEE Transactions on Information Theory, 15: , [McE87] Robert J. McEliece. Finite field for scientists and engineers. Kluwer Academic Publishers, [MR84] James L. Massey and Rainer A. Rueppel. Linear ciphers and random sequence generators with multiple clocks. In Advances in Cryptology - EUROCRYPT 84, Lecture Notes in Computer Science 209, pages Springer Verlag, [Rue86] Rainer A. Rueppel. Analysis and Design of Stream Ciphers. Springer Verlag, [Zie59] Neal Zierler. Linear recurring Sequences. Journal of the Society for Industrial and Applied Mathematics, 2:31 48, of 55

159 Slide Attacks on Hash Functions Michael Gorski 1, Stefan Lucks 1, and Thomas Peyrin 2 1 Bauhaus-University of Weimar, Germany {Michael.Gorski, Stefan.Lucks}@medien.uni-weimar.de 2 Orange Labs and University of Versailles Thomas.Peyrin@gmail.com Abstract. This paper studies the application of slide attacks to hash functions. Slide attacks have mostly been used for block cipher cryptanalysis. But, as shown in the current paper, they also form a potential threat for hash functions, namely for sponge-function like structures. As it turns out, certain constructions for hash-function-based MACs can be vulnerable to forgery and even to key recovery attacks. In other cases, we can at least distinguish a given hash function from a random oracle. To illustrate our results, we describe attacks against the Grindahl-256 and Grindahl-512 hash functions. To the best of our knowledge, this is the first cryptanalytic result on Grindahl-512. Furthermore, we point out a slide-based distinguisher attack on a slightly modified version of RadioGatún. We finally discuss simple countermeasures as a defense against slide attacks. Key words: slide attacks, hash function, Grindahl, RadioGatún, MAC, sponge function. 1 Introduction A hash function H : {0, 1} {0, 1} n is used to compute an n-bit fingerprint from an arbitrarilysized input. Established security requirements for cryptographic hash functions are collision resistance, preimage and 2nd preimage resistance but ideally, cryptographers expect a good hash function to somehow behave like a random oracle. Current practical hash functions, such as SHA-1 or SHA-2 [13, 14], are iterated hash functions, using a compression function with a fixed-length input, say h : {0, 1} n+l {0, 1} n, and the Merkle- Damgård (MD) transformation [6, 12] for the full hash function H with arbitrary input sizes. The core idea is to split the message M into l-bit blocks M 1,..., M m {0, 1} l (with some padding, to ensure all the blocks are of size l-bit), to define an initial value X 0, and to apply the recurrence X i = h(x i 1, M i ). The final chaining variable X i is used as the hash output. The main benefit of the MD transformation is that it preserves collision resistance: if the compression function is collision resistant, then so is the hash function. Recent results, however, highlight some intrinsic limitations of the MD approach. This includes being vulnerable to multicollision attacks [7], long second-preimages attacks [9], and herding [8]. Even though the practical relevance of these attacks is unclear, they highlight some security issues, which designers of new hash functions should avoid. In general, and due to certain structural weaknesses, MD-based hash functions do not behave like a random oracles. Consider, e.g., a secret key K, a message M and define a Message Authentication Code MAC(K, M) = H(K M). If we model H as a random oracle, this is obviously secure. But for an MD-based hash function H, one can easily forge authentication codes: given MAC(K, M) = H(K M), compute a valid MAC(K, M Y ) = H(K M Y ) without knowing the secret key K. Coron et al. [5] recently discussed a formal model to prove hash functions being free from such structural weaknesses (but still weak against multicollision attacks). 7 of 55

160 Our contribution. Newly proposed hash function designs should not suffer from length extension. So for a new and well-designed hash function, the MAC(K, M) = H(K M) should be a secure MAC. We will show that this is not the case for some recently proposed hash functions. In contrast to the case of MD-based hash functions, where one can forge messages but cannot recover K, our attacks allow, in general, the adversary to find K (much faster than by exhaustively searching for it). Our attacks are an application of slide attacks. These are a classical tool for block ciphers cryptanalysis, but have so far not been used for hash function cryptanalysis. The Targets for Our Attacks. A natural idea for thwarting the MD limitations is to increase the size of the internal chaining variables in the iterated process, see, e.g., [11]. Using a similar patch, sponge functions [2] followed the idea to employ a huge internal state (to hold a huge chaining variable) and to claim a capacity c, typically c n. This defends against attackers even if these can perform 2 n/2 operations (but are still restricted to 2 c/2 units of work). Here n is considered a typical hash function output size (sponge functions may also provide for arbitrary output sizes, rather than for a fixed n). Several recent hash functions follow this approach, including Grindahl [10] and Radio- Gatún [1]. As far as we know, there are no cryptanalytic attack on either RadioGatún or the 512-bit version of Grindahl while some collision attacks for the 256-bit version of Grindahl have already been published [15]. In the current paper, we study the applicability of slide attacks for sponge functions. Our results indicate that slide attacks can be a serious threat for hash functions fitting into the sponge framework. On the other hand, if the hash function designer is aware of slide attacks, we believe it is easy to defend against such attacks. We give concrete examples by providing attacks against Grindahl [10] and two slightly tweaked versions of RadioGatún [1]. Our attack applies for both published flavours of Grindahl, the 256-bit version and the 512-bit version. As far as we know, this is the first cryptanalytic result for the 512-bit version. References 1. Guido Bertoni, Joan Daemen, Michael Peeters, and Gilles Van Assche. Radiogatun, a belt-and-mill hash function. Presented at Second Cryptographic Hash Workshop, Santa Barbara (August 24-25, 2006). See 2. Guido Bertoni, Joan Daemen, Michael Peeters, and Gilles Van Assche. On the Indifferentiability of the Sponge Construction. In Nigel P. Smart, editor, EUROCRYPT, volume 4965 of Lecture Notes in Computer Science, pages Springer, Alex Biryukov, editor. Fast Software Encryption, 14th International Workshop, FSE 2007, Luxembourg, Luxembourg, March 26-28, 2007, Revised Selected Papers, volume 4593 of Lecture Notes in Computer Science. Springer, Gilles Brassard, editor. Advances in Cryptology - CRYPTO 89, 9th Annual International Cryptology Conference, Santa Barbara, California, USA, August 20-24, 1989, Proceedings, volume 435 of Lecture Notes in Computer Science. Springer, Jean-Sébastien Coron, Yevgeniy Dodis, Cécile Malinaud, and Prashant Puniya. Merkle-Damgård Revisited: How to Construct a Hash Function. In Victor Shoup, editor, CRYPTO, volume 3621 of Lecture Notes in Computer Science, pages Springer, Ivan Damgård. A Design Principle for Hash Functions. In Brassard [4], pages Antoine Joux. Multicollisions in Iterated Hash Functions. Application to Cascaded Constructions. In Matthew K. Franklin, editor, CRYPTO, volume 3152 of Lecture Notes in Computer Science, pages Springer, of 55

161 8. John Kelsey and Tadayoshi Kohno. Herding Hash Functions and the Nostradamus Attack. In Serge Vaudenay, editor, EUROCRYPT, volume 4004 of Lecture Notes in Computer Science, pages Springer, John Kelsey and Bruce Schneier. Second Preimages on n-bit Hash Functions for Much Less than 2 n Work. In Ronald Cramer, editor, EUROCRYPT, volume 3494 of Lecture Notes in Computer Science, pages Springer, Lars R. Knudsen, Christian Rechberger, and Søren S. Thomsen. The Grindahl Hash Functions. In Biryukov [3], pages Stefan Lucks. A Failure-Friendly Design Principle for Hash Functions. In Bimal K. Roy, editor, ASIACRYPT, volume 3788 of Lecture Notes in Computer Science, pages Springer, Ralph C. Merkle. One Way Hash Functions and DES. In Brassard [4], pages National Institute of Standards and Technology. FIPS 180-1: Secure Hash Standard. April See National Institute of Standards and Technology. FIPS 180-2: Secure Hash Standard. August See Thomas Peyrin. Cryptanalysis of Grindahl. In ASIACRYPT, pages , of 55

162 Kosten von Malware und Spam Oliver Schmid TU Darmstadt Am 29. Mai 2002 verkündete FBI-Director Robert Mueller die Prioritätenliste des FBI, an deren dritter Stelle Protect the United States against cyber-based attacks and high-technology crimes steht. Dabei musste der Kampf gegen Cybercrime lediglich den Aufgaben Terrorbekämpfung und Spionageabwehr weichen, die im Gegensatz zur Internet-Kriminalität eindeutig den Charakter von Geheimdiensttätigkeiten besitzen. Insofern ist die Aussage zulässig, dass die Bekämpfung der Kriminalität im Cyberspace die höchste Priorität unter den nicht-geheimdienstlichen, bundespolizeilichen Aufgaben in den Vereinigten Staaten von Amerika erlangt hat. Ausdrücklichen Vorrang hat dabei nach Angaben der Cyber Investigations -Website des FBI to stop those behind the most serious computer intrusions and the spread of malicious code. Strafrechtlich fällt die Verbreitung von Malware meistens in den Bereich der Datenveränderung oder der Computersabotage und wird in der vom BKA jährlich veröffentlichten Polizeilichen Kriminalstatistik gesondert ausgewiesen. Diese Zahlen müssen jedoch unter dem Vorbehalt gesehen werden, dass von einem extrem großen Dunkelfeld ausgegangen werden muss, wie der Erste Periodische Sicherheitsbericht (PSB, 2001) betont. Gestützt wird diese Annahme durch die Angaben, dass es in Deutschland im Jahr 2000 durch den I love you -Wurm 1 bei 21 % der mit Computern ausgestatteten Arbeitsplätze zu Behinderungen und Arbeitsausfällen kam, dem BKA bei der Veröffentlichung des PSB aber lediglich vier Anzeigen wegen Computersabotage vorlagen. Die Kosten, die durch einen solchen Malware-Vorfall verursacht werden, sind vielfältig und nur teilweise messbar. Neben dem reinen Produktivitätsverlust durch Ausfallzeiten von einzelnen Rechnern oder ganzen Computersystemen sowie den Investitions- und Personalkosten für IT-Sicherheit spielen vor Allem Imageverluste betroffener Unternehmen eine große Rolle. In diesem Risiko, wegen (möglicher) Mängel in der IT-Sicherheit Kunden zu verlieren, sehen auch die Autoren des Periodischen Sicherheitsberichts einen wichtigen Grund für die geringe Anzeigebereitschaft und somit implizit für das Dunkelfeld. Die Folgen von Problemen mit Malware gehen aber bis hin zu Vertrauensverlusten in die Computertechnik und spielen so beispielsweise auch bei der Nutzung von und der damit verbundenen Bekämpfung von Spam durch Filtermaßnahmen eine arbeitspsychologische Rolle. Die Angaben zu Schäden, welche durch einen Virus oder einen Wurm verursacht wurden, müssen häufig kritisch betrachtet werden und halten dann oft dieser Inaugenscheinnahme nicht stand. Dabei ist meistens nicht die Glaubwürdigkeit der Quelle das Problem, sondern eine mangelnde Aufschlüsselung der betrachteten Kostentreiber oder teilweise einfach nur unpräzise Angaben. So bezifferte das Bundesamt für Sicherheit in der Informationstechnik (BSI) in der Viren-Chronik auf seiner Website den durch Code Red im Jahr 2001 verursachten Schaden auf zwei Milliarden US- Dollar. Andere Quellen wie beispielsweise Computerwoche.de berichteten hingegen unter Berufung auf das US-amerikanische Marktforschungsunternehmen Computer Economics von 2,6 Milliarden US-Dollar, je nach Quelle variieren die veröffentlichten Werte zwischen 1,2 und 8,7 Milliarden US- Dollar. Mit Ausnahme von Computer Economics schlüsselte jedoch keine Quelle genauer auf, aus welchen Kosten sich die Summe zusammensetzt, teilweise wurde nicht einmal angegeben, ob sich der Wert auf die weltweiten Kosten bezieht oder ob nur eine Region wie Nordamerika oder Europa betrachtet wurde. 1 Der Wurm I love you wird im PSB allgemein als Virus bezeichnet of 55

163 Die Kosten durch Produktivitätsausfälle beliefen sich laut Computerwoche.de auf ungefähr 1,5 Milliarden, während die Bekämpfung des Wurms Kosten von 1,1 Milliarden verursachte, darunter das Testen, Säubern und Patchen der Systeme. Die Software zum Schließen der Sicherheitslücke stand jedoch schon einen Monat vor dem ersten Auftreten des Schadprogramms auf der Website von Microsoft zur Verfügung, so dass die Infektion durch das präventive Patchen hätte verhindert werden können. Insofern ist fraglich, inwiefern die korrektive Installation wirklich zu den Kosten durch den Wurm gerechnet werden darf, und ob nicht vielmehr ein Großteil der Kosten durch Versäumnisse in der Systempflege begründet sind. Nach Angaben von Microsoft lagen in der Vergangenheit zwischen der Veröffentlichung eines Patches und dem Erscheinen von Schadprogrammen, welche die zu schließende Lücke nutzten, teilweise Monate, wenn auch mit stark sinkender Tendenz 2. Demnach sollte es oft im Vorfeld einer Viren-Epidemie möglich sein, sich durch präventive Maßnahmen gegen die Schadprogramme zu schützen. Außer Frage steht indes, dass die Kosten für das Schließen von Sicherheitslücken zumindest teilweise jenen Kosten zugerechnet werden können, die durch Schadprogramme, also Viren, Würmer und Trojaner, verursacht werden. Werden Unternehmen befragt, wie hoch sie die Schäden schätzen, die ihnen durch Schadprogramme entstanden sind, so wäre es durchaus möglich, dass die Angaben auf Basis jener Kosten gemacht werden, die beispielsweise durch die Beseitigung eines Virus aus dem System entstanden sind. Doch spätestens bei der nicht näher präzisierten Frage nach den Kosten werden die Angaben divergieren, da manche Unternehmen beispielsweise die Personalkosten für Administratoren berücksichtigen, andere hingegen die Meinung vertreten, diese Fachkräfte würden ohnehin beschäftigt, auch ohne konkrete Vorfälle. Auch bleibt in diesem Zusammenhang offen, ob bei der direkten Frage nach den entstandenen Kosten die Angaben zu eventuellen Vorfällen mit Schadprogrammen beschönigt werden oder Vorkommnisse völlig verschwiegen werden. Diesen Eindruck bestätigen die <kes>-studien aus den Jahren 2004 und 2006, in der jüngeren Studie wurde sogar die Aussage getroffen Auch mit Schätz- oder Erfahrungswerten zu Ausfallzeiten und Kosten taten sich die meisten Befragten schwer [... ]. So hatte die Frage zu durch Malware- Infektionen verursachten Ausfallzeiten und Kosten nur gut jedes dritte teilnehmende Unternehmen beantwortet, bei den weiteren Sicherheitsvorfällen sank dieser Anteil sogar auf ein Sechstel, auch unterlagen die angegebenen Werte zum Teil stärkeren Schwankungen. In der Wissenschaft wurden die Phänomene Malware und Spam in den vergangenen Jahren hauptsächlich aus technischer und juristischer Sicht behandelt, die ökonomische Betrachtung dieser Gefahren aus dem Internet hat bislang noch wenig Eingang in die Forschung gefunden. Zur Spam- Problematik gibt es bereits erste Studien, M. Clement et al. (2008) sprechen in ihrer aktuellen Veröffentlichung dennoch von einer Forschungslücke. In der Studie wurden die Kosten von unerwünschten Werb s einer systematischen Analyse unterzogen, bei welcher unter den Mitarbeitern der Universität Kiel eine umfangreiche Erhebung in zwei Befragungswellen durchgeführt wurde. Bei einer Stichprobe von Teilnehmern wurde durch Regressionen das Mittel der jährlichen Kosten je Mitarbeiter auf 531 Euro (2004) bzw. 447 Euro (2005) geschätzt. Diese Werte wurden auf Basis der Arbeitszeitverluste in Verbindung mit den Bruttoarbeitslöhnen errechnet, vorrangigen Einfluss hatte das teilweise tägliche manuelle Aussortieren bzw. die Überprüfung des automatisierten Vorgangs auf False Positives. Berücksichtigt wurde aber auch die Suche nach geeigneten Filtermaßnahmen sowie die Installation und Konfiguration der entsprechenden Software, die zumeist dezentral von den einzelnen Mitarbeitern durchgeführt wurde. 2 Nicht immer erscheinen die Patches vor den dazugehörenden Schadprogrammen, so dass ein regelmäßiges Patchen der Systeme keine Sicherheitsgarantie darstellt of 55

164 Im Rahmen der vorliegenden Arbeit wurde in Zusammenarbeit mit dem Zentrum für Europäische Wirtschaftsforschung (ZEW) in Mannheim eine Unternehmensbefragung zur Feststellung der Kosten von Schadprogrammen bzw. Spam durchgeführt. Dabei wurde die Studie auf die sogenannte Contingent Valuation Method (CVM) ausgerichtet, bei welcher nicht nach durch Schadprogramme verursachten Kosten gefragt wird, sondern die Zahlungsbereitschaft zur Bekämpfung von Viren und Würmern erfragt wird. Weiterhin wurden die Unternehmen gebeten, eine Zahlungsbereitschaft zur Eindämmung der Spam-Problematik zu äußern, da dieses Phänomen im - Alltag zwar nicht direkt als Bedrohung zu sehen ist, aber als eine Belästigung, die in gewissem Maße ebenfalls die Gewährleistung der IT-Sicherheit einschränkt. Bei der Contingent Valuation Methode handelt es sich um ein umfragebasiertes Konzept zur ökonomischen Bewertung von Gütern, für die es keine Marktdaten gibt, oder die den Charakter eines öffentlichen Gutes besitzen. Gemäß Paul R. Portney (1994) werden bei diesem Verfahren im Rahmen von Befragungen individuelle Zahlungsbereitschaften für üblicherweise hypothetische (aber auch tatsächlich erwogene) Projekte oder Programme erfragt. Ferner wies Portney darauf hin, dass die von den Probanden geäußerten Zahlungsbereitschaften von dem im Fragebogen dargestellten Marktszenario abhängig seien. Nach Portney gibt es keine standardisierte Vorgehensweise für das Design einer CV-Studie, dennoch bestehen beinahe alle Anwendungen aus drei Teilen: Zuerst muss ein klar verständliches und möglichst detailliertes Szenario geschaffen werden, ein (hypothetisches oder tatsächlich geplantes) Programm oder politisches Konzept, zu welchem der Proband Stellung beziehen oder welches er bewerten soll. Dabei soll dem Probanden ein klares Bild dessen gegeben werden, was er im Rahmen der Studie bewerten soll. Weiterhin muss die Erhebung einen Mechanismus zur Verfügung stellen, durch welchen der Proband einen gewählten Wert oder eine getroffene Entscheidung kundtun muss. Diese Mechanismen sind vielfältig, einschließlich offener Fragen, Gebote oder Referenda. Zu guter Letzt erheben CV-Studien üblicherweise Informationen zu sozioökonomischen Merkmalen der Teilnehmer (u. a. Alter, Geschlecht, Einkommen), ebenso wie für das spezifische Szenario relevante Eigenschaften. Diese Variablen ermöglichen bei der Auswertung die Schätzung einer Zahlungsbereitschaftsfunktion, welche diese Charakteristika als mögliche erklärenden Variablen beinhaltet. Des Weiteren kann mit Hilfe von Kontrollfragen festgestellt werden, ob der Antwortende die Aufgabe verstanden hat und die Fragen ernsthaft beantwortet wurden. Die theoretische Grundlage für die CVM schuf S. V. Ciriacy-Wantrup mit einem Aufsatz, in dem er die Methode zur Bestimmung der Nachfrage nicht handelbarer Güter durch die Befragung der Bevölkerung über ihre willingness to pay vorschlug. Die praktische Umsetzung dieser Idee erfolgte erstmals durch R. Davis, der mittels einer Umfrage den Wert eines Erholungsgebiets zu schätzen versuchte und dabei eine hohe Korrelation zwischen den Resultaten seiner CVM-basierten Umfrage und den Ergebnissen der etablierten Reisekostenmethode feststellte. Während sich die CVM neben ihrem Anwendungsschwerpunkt in der Ressourcen- und Umweltökonomik auch schon in einigen anderen Bereichen der Nationalökonomie etabliert hat, kommt sie in der empirischen Kriminalitätsforschung bisher kaum zum Einsatz, so fokussierten die Artikel von J. Ludwig und P. Cook (2001) sowie M. Cohen et al. (2004) auf die Kosten von Gewaltkriminalität. In Zusammenarbeit mit der IKT-Abteilung des ZEW wurde im ersten Quartal 2006 der Fragebogen der vierteljährlich stattfindenten ZEW Konjunkturumfrage auf den Themenschwerpunkt IT-Sicherheit ausgerichtet. Um zwei im Zusammenhang mit der CVM auftretende Effekte kontrollieren zu können, wurden die teilnehmenden Unternehmen zufällig in vier Gruppen unterteilt, deren Fragen zur Zahlungsbereitschaft entsprechend variiert wurden of 55

165 Die Fragen rund um die Erfassung der Zahlungsbereitschaften waren dabei auf die Bedürfnisse der CVM zugeschnitten, die Kernfragen für die Ermittlung der Zahlungsbereitschaft für zwei Aspekte der IT-Sicherheit wurden wie folgt formuliert: Angenommen, Sie hätten keine Möglichkeit, einen eigenen Spamfilter einzurichten. Wie viel wären Sie bereit pro Jahr an eine europäische Institution zur Bekämpfung von Spam zu zahlen, um den Anteil der unerwünschten s (Spam) zu senken? Je nach Befragungsgruppe wurden die Teilnehmer der Studie nach ihrer Zahlungsbereitschaft (in Euro) für eine Senkung um zwei unterschiedliche prozentuale Anteile gefragt, die Fragestellung zu Schadprogrammen mit Viren und Trojanern als explizit genanntem Beispiel wurde analog formuliert. Des Weiteren wurde die Anordnung der Fragen vertauscht, um einen möglichen Einfluss der Reihenfolge auf das Antwortverhalten entdecken zu können, dieser sogenannte Sequencing-Effekt konnte aber in der vorliegenden Studie nicht nachgewiesen werden. An der Umfrage im Frühjahr 2006 nahmen 844 Unternehmen aus der Branche der IKT-Dienstleister aktiv teil, damit liegt die Zahl der zur Verfügung stehenden Beobachtungen nahe der Anzahl an Unternehmen, welche sich üblicherweise an der Umfrage beteiligen. Ein im Vorfeld der Erhebung als problematisch erachtetes Thema war die Frage nach durch Malware verursachten Vorfällen in den Unternehmen, insofern überraschte, dass sich 807 der teilnehmenden Unternehmen zu diesem heiklen Thema äußerten. Fast die Hälfte dieser Teilnehmer gab an, noch nie Vorfälle durch Schadprogramme gehabt zu haben, von den verbleibenden 411 Unternehmen hatte ungefähr die Hälfte Malware-Probleme im Jahr 2005, ebenso wie im Jahr 2004, auch vor dem Jahr 2004 hatte jedes zweite der Unternehmen Vorfälle zu beklagen. Mit 127 Unternehmen hatte immerhin mehr als jeder vierte Teilnehmer mit Malware-Befall Vorfälle in den Jahren 2004 und 2005 zu vermelden, besonders hart traf es 83 Unternehmen, die nach eigenen Angaben sowohl in den Jahren 2005 und 2004, als auch davor Probleme mit Schadprogrammen gehabt haben. Zur Frage nach dem Anteil der Spam-Mails am gesamten -Aufkommen machten 727 Unternehmen Angaben, dabei fällt auf, dass die Belastung der IKT-intensiven Dienstleister erheblich geringer ausfällt als bei branchenübergreifenden Studien des gleichen Zeitraums. So lag bei der Hälfte der Unternehmen der Anteil unerwünschter s am täglichen Mail-Aufkommen bei maximal 15 %, ein Viertel der Teilnehmer gab sogar nur bis zu 5 % an, lediglich jedes fünften Unternehmen empfängt mehr Spam als erwünschte s. Zur Schätzung der Zahlungsbereitsschaft der Unternehmen für eine (fiktive) Reduzierung des Malware- bzw. des Spam-Aufkommens um 100 % wurden nun die Ergebnisse für die Angaben zu den erhobenen Prozentangaben mittels schrittweiser Regressionen geschätzt. Bei diesen Schätzungen überraschte, dass die geäußerten Zahlungsbereitschaften hauptsächlich von der Anzahl der im Unternehmen beschäftigten Mitarbeiter abhingen, wenige andere Variablen jedoch nur in einzelnen Beobachtungsgruppen signifikanten Einfluss hatten. Eine Bereinigung der Zahlungsbereitschaften um die Mitarbeiterzahl führte hingegen zu keinen durchgängig signifikanten Ergebnissen. Die Resultate für die verschiedenen Prozentangaben wurden nun mittels einer Potenzfunktion interpoliert, um den Funktionswert für einen theoretischen, da nicht realisierbaren Schutz von 100 % zu berechnen. Demnach läge die jährliche Zahlungsbereitschaft bei der Bekämpfung von Malware bei 103 Euro je Mitarbeiter, bei der Vermeidung von Spam bei 47 Euro. Gemäß der zugrundegelegten Theorie der CVM entstehen der Branche der IKT-Dienstleister somit jährlich Kosten von 220 Mio. Euro durch Schadprogramme sowie 100 Mio. Euro durch unerwünschte s of 55

166 Literatur [EPS01] Bundesministerium des Innern, Bundesministerium der Justiz (o. V.) (2001). Erster Periodischer Sicherheitsbericht (Tech. Rep.). Berlin. [CPB08] Clement, M., D. Papies und H. J. Boie (2008). Kosten und Kostentreiber von unerwünschten Werb s (Spam) - Eine empirische Analyse auf Provider- und Anwenderseite. Erscheint in: Zeitschrift für Betriebswirtschaft [CR+04] Cohen, M. A., R. T. Rust, S. Steen und S. T. Tidd (2004). Willingness-to-Pay for Crime Control Programs. Criminology 42 (1), S [DH94] [H94] Diamond, P. A. und J. A. Hausman (1994). Contingent Valuation: Is Some Number Better than No Number? Journal of Economic Perspectives 8 (4), S Hanemann, W. M. (1994). Valuing the Environment Through Contingent Valuation. Journal of Economic Perspectives 8 (4), S [KES06a] <kes> Die Zeitschrift für Informations-Sicherheit (2006a). Lagebericht zur Informations- Sicherheit. <kes> - Zeitschrift für Kommunikations- und EDV-Sicherheit Nr. 4, August / September, S SecuMedia Verlag, Ingelheim. [KES06b] <kes> Die Zeitschrift für Informations-Sicherheit (2006b). Lagebericht zur Informations- Sicherheit (3). <kes> - Zeitschrift für Kommunikations- und EDV-Sicherheit Nr. 6, Dezember, S SecuMedia Verlag, Ingelheim. [LC01] [P94] Ludwig, J. und P. J. Cook (2001). The Benefits of Reducing Gun Violence: Evidence from Contingent-Valuation Survey Data. International Review of Law and Economics 22 (3), S Portney, P. R. (1994). The Contingent Valuation Debate: Why Economists Should Care. Journal of Economic Perspectives 8 (4), S of 55

167 1 Introduction SPAM Some Cryptographic Thoughts Christopher Wolf Horst Görtz Institute for IT-Security, Ruhr-University Bochum, Germany or When introducing electronic mail or , the designers of the Simple Mail Transfer Protocol (SMTP) imagined a cooperative Internet. In its first incarnation as a research network, this was certainly true. However, since the first newsgroup crosspostings and the successive rise of spam, this paradigm has certainly proven wrong. Due to the lack of even the most basic security features (e.g. sender authentication) in SMTP or its current version ESMPT, spam is on the rise. Actually, using some basic physical laws, it is even predicted in that spam will cause a resonance-cathastrophy in the entire -system which will first render totally useless as a mean of communication and second will lead to failures in the entire Internet. Hence, drastic measures seem to be necessary to elude spam. As spam has quite a long history, at least by Internet standards, it is not surprising, that such a drastic solution has already been proposed, namely Internet Mail Before outlining Internet Mail 2000, we will first look at the basic difference between traditional mail and . Although these differences are rather simple, this will help us to understand why spam exists in but was and still is virtually unknown in traditional mail. In traditional or surface mail, all a potential receiver has to do is to put up a post-box, attach her name to it and to distribute her address to potential senders, e.g. her friends, her bank, or favourite online-shops. After that, mail will arrive and will be paid for by the sender, e.g. through a stamp. Hence, the whole infrastructure for delivering surface mail, i.e. post-boxes and offices, store houses, lorries and staff, is paid by through the senders. The only exception are companies who may charge for shipping and handling or otherwise include this fee into their price calculation. In electronic mail, the situation is fundamentally different. Here, the receiver pays by far and large the mayority of the costs through operating a server 24h per day and in line with latest security findings, keeping it connected to the rest of the Internet and so forth. And instead of improving, the situation gets worse as all commonly used anti-spam techniques such as content filtering, fingerprinting, black- and white-listing have to be paid by the sender, not the receiver. In contrast, gray-listing is neutral in this respect as it requires both the sender and the receiver to invest in a connection: on the sender side, an has to be placed in a queue and resent, while on the receivers side, some memory has to be spent to remember the Internet address (IP address) of the prospective sender. Internet Mail 2000 (IM2000) tried to change this fundamental asymmetry between sender and receiver in the SMTP protocol and also this fundamental difference between traditional and electronic mail by requiring the sender to operate co-called out-box servers, which had to be kept online 24h a day. As we saw above, this is in fact the mayority of costs in the SMTP protocol. To notify a user about new messages, IM2000 suggested small notification messages. Upon receiving such a notification, the receiver would connect to the corresponding out-box server and download 1 15 of 55

168 his message or decide not to do this in case of unwanted . Hence, it would have been on the costs of the receiver to store bulks of unwanted messages. Although technically brilliant, IM2000 was obviously too drastic a change to supersede SMTP. Unfortunately, network effects have had a big impact here. However, the basic observation of IM2000 is still valid, i.e. in a fair system, the sender should pay a large or at least larger than today cost to send out s then the receiver. This observation is also strengthened by the statement Users rarely learn about the they do not receive, and even if they do, are seldom competent enough to assign the blame to their [Internet Service Provider] blocking it rather than the sender having made an error.. Although a bit cynical, the statement is certainly true in the sense that receivers will usually not learn about they do not receive while senders will learn about this, e.g. by answers which do not arrive or through bounce messages. Hence, the incentive for a sender to invest into her outbound infrastructure is larger than the incentive for a receiver to invest into his inbound infrastructure. Equipped with this observation, we therefore advocate a paradigm shift in SMTP: rather than having the receiver bearing the mayority of the costs, we show how to shift them to the sender, using different, partly surprisingly simple strategies (Sec. 2. However, the example of IM2000 has to make us deliberate if we want to see this paradigm shift to happen in practice (and hence, to find far less SPAM in our own inbox). Hence, we have to make sure that our suggested changes are compatible with existing techniques and the spirit of the SMTP protocol. We will then go through several use-cases to see how SF-MTP is expected to work in practice (Sec. 3). 2 Let Sarah Pay As already outlined above, we will outline in this section a few simple strategies which can be used to combat spam by shifting as much workload as possible from the receiver Robert to the sender Sarah. For the purpose of our article, it is not important if Sarah or Robert are actual human beings or computer programmes. Although this view is a bit of a simplification in terms of the current SMTP protocol, we found it a valid simplification to ease explanation. We will hint in later sections, at which stage we suggest to implement it. In most cases, you should assume Sarah and Robert being computer programmes or servers. In a nutshell, we suggest External White-and-Gray-Lists Sender-Call-Back Proof-of-Work User Interaction Standardized Abuse & Bounce Messages Distributed Repudiation System Before going into the details of our suggestions, we start with a brief sketch of a typical SMTP session to give the reader not familiar with the Simple Mail Transfer Protocol an idea of possible problems in terms of authentication and hence the ancor points for spammers of 55

169 2.1 SMTP session Hier kommt ein Bild 1. Sarah 220: Greeting (ESMTP) 2. Robert HELO/EHLO + domain 3. Sarah 250: Hello back 4. Robert MAIL FROM: Mail-Adress-Robert 5. Sarah 250: Sender OK 6. Robert RCPT TO: Mail-Adress-Sarah 7. Sarah 250: Recipient OK 8. Robert DATA (komplette , inkl. From, To,...) 9. Sarah 250 accepted for delivery 10. Robert QUIT 11. Sarah 221: Bye Not shown here is the fact that Robert and Sarah need to know each others Internet or IP address to send each other messages. However, we see that Robert can send any message he wants to sent to Sarah. In particular, Sarah cannot be sure about anything but the validity of her own address (given by Robert in Step 6), and Robert s IP addresss. Therefore, it has been common among spammers to use fake from addresses. In addition, the from-address given in Step 4 does not necessarily match the from-address given in the text in Step 8. On this other hand, this property allows for such features as -forwarding or -lists. 3 Use Senarios We assume that the following use scenarios need to be addressed by any solution which can superseed SMTP: Normal User Automated Forwarding Roaming Users Mailing List 3 17 of 55

170 The Indifferentiability Framework - Revisited Ewan Fleischmann, Stefan Lucks Bauhaus-University of Weimar, Germany Abstract. At Crypto 2005, Coron et al. introduced a formalism to study the presence or absence of structural flaws in iterated hash functions: If one cannot differentiate a hash function using ideal primitives from a random oracle, it is considered structurally sound, while the ability to differentiate it from a random oracle indicates a structural weakness. This model was devised as a tool to see subtle real world weaknesses while in the random oracle world. In this paper we take in a practical point of view. We show, using well known examples like NMAC and MCM, how we can prove a hash construction secure and insecure at the same time in the indifferentiability setting. These constructions do not differ in their implementation but only on an abstract level. Naturalls, this gives rise to the question what to conclude for our hash function. Our results cast doubts about the notion of indifferentiability from a random oracle as a practically relevant criterion (as e.g. proposed by Knudsen [KNU08] for the SHA-3 competition) to separate good hash function structures from bad ones. Keywords. hash function, random oracle, uninstantiable, indifferentiable 1 Introduction 1.1 Hash Functions Cryptographic hash functions map an input of unlimited size to a small fixed-size output. Due to their complexity and for theoretical purposes, cryptographic hash functions are often modeled as random oracles. In practice, cryptographic hash functions are usually implemented based on some compression function(s) with fixed-size input and output. If the compression functions are insecure, we can hardly expect much security from the iterated hash function. On the other hand, even if the compression functions are secure, there could be flaws in the hash structure which an adversary might possibly exploit. Current hash functions, for example, almost universally follow the Merkle- Damgård structure [Me89,Da89], which suffers from length-extendibility: Given the hash H(M) of a message M, one can easily compute some string X and some hash H(M X) without even knowing the message M itself. A good practical hash function H should prevent this. Motivated by the practical need to say anything about structural flaws in the design of H itself, Coron et al. [C05] presented a new notion of security for cryptographic hash functions called indifferentiability. In short, if one models the compression functions as random oracles with fixed-size inputs, then the iterated hash function composed from these compression functions should be indifferentiable from a random oracle with variably-sized inputs. This could be a criterion, e.g., to separate practical hash functions with a good structure from those which might suffer from structural flaws, e.g., in the context of the search for new standard hash functions [NIST]. The current paper discusses this issue of 55

171 Preliminaries. In the current paper, we use notions such as efficient, significant and negligible as usual in theoretical cryptography, e.g., an algorithm is efficient, if its running time is bounded from above by a polynomial in the security parameters. 2 Random Oracles and Indifferentiability Random oracles are a mathematical abstraction used in cryptographic proofs, abstracting away nearly all (instantiation-)specific details. Namely, a random oracle responds to every fresh query with a random response chosen uniformly from its output domain. If a query isn t fresh, i.e., if the same query has been made before, the random oracle repeats its previous answer, thus ensuring that random oracles always behave like well-defined functions. Random oracles are typically used when no known implementable function provides the mathematical properties required for the proof or when it gets too tedious to precisely formalise these properties. We consider cryptographic schemes with access to an oracle. The adversary is allowed to access the same oracle as the legitimate user. The idea is to prove the scheme secure assuming the random oracle, but, when used in practice, instantiate the oracle by some efficient function. From a theoretical point of view, it is clear that a security proof in the random oracle model is only a heuristic indication of the security of a system when instantiated with a particular function. Nevertheless, a formal proof in the random oracle model seems to indicate that there might be no structural flaws in the design (i.e. structure) of the system. 2.1 Being Indifferentiable from a Random Oracle a Security Notion for Hash Functions The notion of indifferentiability for hash functions is defined as follows [MRH04]: Definition 1. A Turing machine H Alg with oracle access to a set of ideal primitives G is said to be computationally indifferentiable from an ideal primitive H Rnd if there exists a simulator S, such that for any distinguisher D it holds that the difference P r[d H Alg,G = 1] P r[d H Rnd,S = 1] is negligible. Here, the simulator S is efficient, has oracle access to H, and simulates the primitive. The distinguisher D also is efficient (i.e., runs in polynomial time and thus makes at most a polynomial number of queries). See figure 1 for a visualisation of indifferentiability. As shown in [MRH04, Proposition 2], if H Alg is computationally indifferentiable from H Rnd, we can take any cryptosystem using H Rnd (i.e. a random oracle) and replace H Rnd by H Alg, without destroying the cryptosystem s security. This makes indifferentiability in general a useful tool for cryptography. Typically, an ideal n-bit hash function is thought of as a random function H Rnd : {0, 1} {0, 1} n. In practice, hash functions usually iterate some compression function(s) F : {0, 1} n {0, 1} m {0, 1} n. We denote such a compression function as an ideal primitive, if it is chosen randomly, uniformly distributed from the set of all such functions. (Clearly, we can model random functions as random oracles!) 2 19 of 55

172 Case I: Hash Algorithm using ideal primitives Case II: Random Hash Function Simulator for primitives Hash Algorithm Halg F G G? Distinguisher D Random Hrnd Simulator S F G Ideal Function Algorithm A B: A has access to B Figure 1. Defining H being indifferentiable from a random oracle RO Coron et al. [C05] used the following formalism to characterise the structural soundness of hash functions. Assume as set G of ideal primitives and a hash function H Alg utilising these primitives. We denote, H Alg as computationally indifferentiable from a random oracle, if no efficient adversary A can distinguish between the following two cases with any significant advantage: Algorithm: oracle access to H Alg and oracle access to the ideal primitives contained in G and Random: oracle access to a random oracle H Rnd and an efficient simulator S answering the oracle queries to the primitives in G. (The simulator S has no access to the H Rnd -oracle queries of the attacker but access to the random oracle H Rnd itself.) Here, the advantage of A is defined as the difference of two probabilities, the probability P r[a H Alg,G = 1] that A outputs 1 when given access to H Alg and the ideal primitives in G, and P r[a H Rnd,S = 1] of A outputting 1 when given access to a random oracle H Rnd and a simulator S for the primitives: Adv(A) = P r[a H Alg,G = 1] P r[a H Rnd,S = 1]. In the following, we will sometimes briefly write secure instead of computationally indifferentiable from a random oracle. Note that by assuming ideal primitives even in the Algorithm case, this definition is inherently based on the random oracle model. In the standard model we cannot assume ideal primitives (at least not without allowing an exponentially-sized memory to store a description of the function), so this notion of security only makes sense in the random oracle model. Nevertheless, as we understand [C05], a part of their motivation was to introduce a formalism for aiding the design of practical hash functions. Showing the above kind of security in the random oracle model ought to indicate the absence of structural flaws in the hash function. On the other hand, if one can efficiently differentiate a hash function (using ideal primitives) from a random oracle, this appears to indicate a weakness in the hash function structure. With this reasoning, we again follow the example of Coron et al., who debunk certain hash function structures as insecure by pointing out efficient differentiation attacks [C05, sections 3.1 and 3.2] of 55

173 2.2 Indifferentiability vs. Indistinguishability 3 (In)Security in the Indifferentiability World In the following we will examine a bunch of constructions secure in the indifferentiability framework involving one or more random oracles and show how slight modifications to them or partial instatiations drive them insecure (at least in this framework). Section Secure Insecure Insecure (partial instantiation or modification) (extension) ( 3 ) RO 1 X 1 RO X 2 X RO ( 3 ) RO RO RO X RO RO X X RO X RO RO ( 3 ) RO MD RO RO MD Z RO MD RO X (NMAC) X MD RO X RO MD RO ( 3 ) RO i X RO i RO x X Y RO i X RO i Y (MCM) Y X RO x Y RO i X RO i X RO x Y ( 3 ) Secure Structure Instantiate one of Add a finalization or with at least 2 RO s the RO s using a CROWF preprocessing CROWF Table 1. RO denotes a random oracle (with fixed or variable length input), RO i an injective random oracle, RO x a (fixed or variable length, injective or not) random oracle, X,Y and Z collision resistant one-way functions (CROWF). Our results are stronger than indicated by Table 1. Motivational, informal Example. Say we want to design a secure hash function and come up with the idea to design our hash function as a concatenation of a preprocessing function modelled as a random oracle RO and a collision resistant one way function (CROWF) X. Consequently, our hash function H for a message M is H := (RO X)(M). So we try to proof its security in the indifferentiability framework and come to the conclusion that this hash function is in fact insecure (see Theorem 1.1 (iii) 3 ). In the indifferentiability framework we have at least three straightforward approaches to get H secure: 1. Remove the CROWF X: H 1 = (RO)(M). 2. Strengthen X and make it a random oracle: H 2 = (RO RO)(M). 3. Weaken X and make it an easily invertible function: H 3 = (RO X)(M). The hash functions H 1, H 2 and H 3 can be proven secure in the indifferentiability framework (see Thm. 1.1(i) and (ii) 3 ). 1 Proofs are trivial and are not given here. A random oracle alone can of course not be instantiated partially. 2 same as right below in this table 3 see full version of the paper, available soon on our homepage medsec.medien.uni-weimar.de 4 21 of 55

174 Indifferentiability was devised as a tool to see subtle real world weaknesses while in the random oracle world. But we can prove H insecure and H 3 secure. In the real world (i.e. comparing the instantiated hash functions) H 3 is sure to be substantially weaker than H. Additionally, the hash functions H and H 2 could be implemented using the same algorithms but one is proved to be insecure, the other one is secure. What shall we conclude for the security of our instantiated hash function in the real world? We don t see how we can conclude that H has some real world weaknesses that H 2 has not. References [BBP04] M. Bellare, A. Boldyreva, A. Palacio. An Uninstantiable random oracle-model Scheme for a Hybrid- Encryption Problem. Eurocrypt [BR06] M. Bellare, T. Ristenpart. Multi-Property-Preserving Hash Domain Extension and the EMD Transform. Asiacrypt [BR93] M. Bellare, P. Rogaway. Random Oracles are Practical: A Paradigm for Designing Efficient Protocols. ACM Conference on Computer and Communications Security, [CCH04] Canetti and Goldreich and Halevi, The Random Oracle Methodology, Revisited, JACM: Journal of the ACM, volume 51, 2004 [CCH04a] Canetti and Goldreich and Halevi, On the random oracle Methodology as Applied to Length-Restricted Signature Schemes. Theory of Cryptography Conference (TCC) [C05] J. Coron, Y. Dodis, C. Malinaud, P. Puniya. Merkle Damgård Revisited: How to Construct a Hash Function, Crypto [Da89] I. Damgård. A Design Principle for Hash Funcitons. Crypto 89. [Go06] O Goldreich, On Post-Modern Cryptography, Cryptology eprint Archive, Report 2006/461, [GT03] S. Goldwasser, Y. Taumann, On the (In)security of the Fiat-Shamir Paradigm. Symposium on Foundations of Computer Science (FoCS) [Ko07] N. Koblitz, The Uneasy Relationship Between Mathematics and Cryptography, NOTICES: Notices of the American Mathematical Society, volume 54, 2007 [KNU08] Knudsen, Lars R. Hash Functions and SHA-3, Invited Talk at FSE 2008, slides/day_1_sess_2/knudsen-fse2008.pdf [Li06] Constructing Secure Hash Functions from Weak Compression Functions: The Case for Non-Streamable Hash. Selected Areas in Cryptography (SAC) [MRH04] U. Maurer, R. Renner, C. Holenstein. Indifferentiability, Impossibility Results on Reductions, and Applications to the Random Oracle Methodology. Theory of Cryptography Conference (TCC), [Me89] R. Merkle. One-Way Hash Functions and DES. Crypto [Ni02] Nielsen, Separating Random Oracle Proofs from Complexity Theoretic Proofs: The Non-committing Encryption Case. Crypto [NIST] National Institute of Standards and Technology (NIST). Tentative Timeline of the Development of New Hash Functions. [RS08] Ristenpart Thomas, Shrimpton Thomas. How to Build a Hash Function from Any Collision-Resistant Function, 2007, ASIACRYPT, pages , of 55

175 Code based signature schemes Pierre-Louis Cayrel Université de Limoges, XLIM-DMI, 123, Av. Albert Thomas Limoges, France Abstract. In this paper, I propose a survey of signature schemes using error correcting codes. I describe coding theory, its application to cryptography and the different identification and signature schemes using error correcting codes. Next I briefly describe a secure implementation of the Stern Scheme against DPA [3], an identity based Signature Scheme [2] and the Kabatianskii Krouk Smeets Signature Scheme [4]. 1 Introduction Coding theory deals with the error-prone process of transmitting data across noisy channels, via clever means, so that a large number of errors that occur can be corrected. It also deals with the properties of codes, and thus with their fitness for a specific application. We can use the coding theory in cryptography. It is partly on the difficulty of the Syndrome Decoding problem and the Goppa Parameterized Bounded Decoding problem that the safety of several cryptosystems (e.g. McEliece [10] or Niederreiter [11]) lies. This makes it an issue for cryptographic applications of error-correcting codes is not only being NP-complete, but also the fact that, in general, a random instance of this problem is always difficult. It is difficult to prove specific results on this matter, but it seems that in terms of complexity in average, the problem is almost always difficult. Code-based cryptography was introduced by McEliece [10], two years after the introduction of public key cryptography by Diffie and Hellman [6] in In his paper he proposed a public key encryption scheme. In 1986 Niederreiter proposed [11] an equivalent code-based cryptosystem [9]. McEliece first had the idea in 1978, of using the theory of error-correcting codes for cryptographic purposes, and more specifically for asymmetric encryption algorithm. The principle of the protocol described is for Alice to send a message containing a large number of mistakes, mistakes that only Bob knows how to detect and correct. In this paper, I present recent improvements in code based cryptography. After I detailed the identification and signature schemes using error-correcting codes, I briefly describe three new results on code based signature schemes : a secure implementation of the Stern Scheme Section 3, an identity-based Signature Scheme Section 4, a stud of the Kabatianskii Krouk Smeets Signature Scheme Sectiin 5. 2 Identification and signature with error correcting codes An open question was the existence of a signature scheme based on coding theory, if one puts aside the signature scheme induced by the Fiat-Shamir paradigm from the Stern identification scheme of 1993, the first signature scheme was proposed in 2001 by Courtois, Finiasz and Sendrier in [5]. It adapts the Full Domain Hash approach of Bellare and Rogaway [1] to Niederreiter s encryption scheme. Another approch to this problem was investigated in 1997 by Kabastianskii, Krouk and Smeets who proposed a (few times) digital signature scheme based on random errorcorrecting codes. 2.1 Stern identification and signature scheme At Crypto 93, Stern proposed a new identification and signature scheme based on coding theory [12]. Stern s Scheme is an interactive zero-knowledge protocol which aims at enabling any user (usually called the prover and denoted here by P ) to identify himself to another user (usually called the verifier and denoted here by V ). 23 of 55

176 Let n and k be two integers such that n k. Stern s Scheme assumes the existence of a public (n k) n matrix H defined over F 2. It also assumes that an integer t n has been chosen. For security reasons (discussed in [12]) it is recommended that t is chosen slightly below the value given by the Gilbert-Varshamov bound. The matrix H and the weight t are protocol parameters and may be used by several (even numerous) different provers. Each prover P receives a n-bit secret key s P (also denoted by s if there is no ambiguity about the prover) of Hamming weight t and computes a public identifier i V such that i V = Hs T P. This identifier is calculated once in the lifetime of H and can thus be used for several identifications. When a user P needs to prove to V that he is indeed the person associated with the public identifier i V, then the two protagonists perform the following protocol where h denotes a standard hash function: 1. [Commitment Step] P randomly chooses y F n and a permutation σ defined over F n 2. Then P sends to V the commitments c 1, c 2 and c 3 such that : c 1 = h(σ Hy T ); c 2 = h(σ(y)); c 3 = h(σ(y s)), where h(a b) denotes the hash of the concatenation of the sequences a and b. 2. [Challenge Step] V sends b {0, 1, 2} to P. 3. [Answer Step] Three possibilities: if b = 0 : P reveals y and σ. if b = 1 : P reveals (y s) and σ. if b = 2 : P reveals σ(y) and σ(s). 4. [Verification Step] Three possibilities: if b = 0 : V verifies that c 1, c 2 have been honestly calculated. if b = 1 : V verifies that c 1, c 3 have been honestly calculated. if b = 2 : V verifies that c 2, c 3 have been honestly calculated, and that the weight of σ(s) is t. 5. Iterate the steps 1,2,3,4 until the expected security level is reached. Fig. 1. Stern s Protocol During the fourth Step, when b equals 1, it can be noticed that Hy T derives directly from H(y s) T since we have: Hy T = H(y s) T i V = H(y s) T Hs T. As proved in [12], the protocol is zero-knowledge and for a round iteration, the probability that a dishonest person succeeds in cheating is (2/3). Therefore, to get a confidence level of β, the protocol must be iterated a number of times k such that (2/3) k β holds. When the number of iterations satisfies the last condition, then the security of the scheme relies on the NPcomplete problem SD. By using the so-called Fiat-Shamir Paradigm [7], it is theoretically possible to convert Stern s Protocol into a signature scheme, but then the signature is very long (about 140-kbit long for a security of 2 80 binary operations). Notice that it is large in comparison with classical signature schemes, but close to, or less than, the size of many files used in everyday life. One also can notice that the memory of common devices has considerably increased when comparing to the time (1978) when the first code-based system was proposed. 2.2 CFS signature s scheme As we already mentioned, contrarely to the RSA scheme which is naturally invertible, the McEliece or the Niederreiter schemes are not invertible, i.e. if one starts from a random element y of F n 2 and a code C[n, k, d] that we are able to decode up to d/2, it is almost sure that we won t be able to decode y into a codeword of C. This comes from the fact that the density of the whole space which is decodable is very small. Courtois, Finiasz and Sendrier proposed in [5] a practical signature scheme based on coding theory (CFS). The idea of the CFS scheme is to fix parameters [n, k, d] such that the density of decodable codewords is reasonable and pick up random elements until one is able to decode it. More precisely, given M a message to sign and h a hash function of {0, 1} n k. We try to find a way to build s F n 2 of given weight t such that h(m) = Hs T. For D() a decoding algorithm, the algorithm works as follows: 2 24 of 55

177 1. i 0 2. while h(m i) is not decodable do i i compute s = D(h(M i)) where arg 1 arg 2 notes the concatenation of arg 1 and arg 2. Fig. 2. The CFS signature scheme We get at the end an {s, j} couple, such that h(m j) = Hs T. Let us notice that we can suppose that s has weight t = [d/2]. Using a Goppa code, we start from a word of length k = n mt that we transform into a codeword of length n = 2m, and then we have an error of weight t. The decoding algorithm (the trapdoor) permits us to decode any word built that way, meaning, the set of all words to a distance t or less of a codeword. These words represent a set of spheres of balls of radius t centered on the codewords. 2.3 Kabatianskii Krouk Smeets KKS signature s scheme Kabatianskii et al. [8] proposed a digital signature scheme based on random linear codes. They exploited the fact that for every linear code the set of its correctable syndrome contains a linear subspace of relatively large dimension. Firstly, we suppose that C is defined by a random parity check matrix H. We also assume that we have a very good estimate d of its minimum distance. Next, we consider a linear code U of length n n and dimension k defined by a generator matrix G = [g i,j ]. We suppose that there exist two integers t 1 and t 2 such that t 1 u t 2 for any non-zero codeword u U. Let J be a subset of {1,..., n} of cardinality n, H(J) be the sub matrix of H consisting of the columns h i where i J and define an r n matrix F def = H(J)G T. The application f : GF (q) k M n,t is then defined by f(m) = mg for any m GF (q) k where G = [gi,j ] is the k n matrix with g i,j = g i,j if j J and gi,j = 0 otherwise. The public application χ is then χ(m) = F m T because HG T = H(J)G T. The main difference with the CFS s signature scheme (Section 2.2) resides in the verification step where the receiver checks that: t 1 wt(z) t 2 and F m T = H z T. 1. Setup. The signer chooses a random matrix H = [I r D] that represents the parity check matrix of a code C[n, n r, d]. He also chooses a random generator matrix G that defines a code U[n, k, t 1] such that wt(u) t 2 for any u U. He chooses a random set of size n J {1,..., n} and he forms F = H(J)G T. 2. Parameters. Private key. The set J {1,..., n} and the k n matrix G Public key. The r k matrix F and the r n matrix H 3. Signature. Given m GF (q) k, the signer sends (m, m G ) 4. Verification. Given (m, z), the receiver verifies that: t 1 wt(z) t 2 and F m T = H z T. Fig. 3. KKS signature scheme. Compared with other public-key cryptosystems which involve modular exponentiation, these schemes have the advantage of high-speed encryption and decryption. However, they suffer from the fact that the public key is very large. 2.4 Pros and cons Stern s Scheme : The operations used are really simple, the identification is possible, the signature is done quickly and in a zero-knowledge context, using quasi-cyclic construction the public key is small but the signature produced is of large size of 55

178 CFS s Scheme : On one hand, we have seen that both keysize and signing cost will remain high. On the other hand the signature length and verification cost will always remain extremely small. KKS s Scheme : The signature is quite small and the system is fast, using almost quasi-cyclic construction the public key is small but the fact that we can just make few times signature is a real drawback. 3 Secure implementation of the Stern Scheme In [3], the authors presente the first implementation on smartcard of the Stern authentication and signature scheme. This gives a securization of the scheme against side channel attacks. On the whole, this provides a secure implementation of a very practical authentication (and possibly signature) scheme which is mostly attractive for light-weight cryptography. A quick analysis of Stern s protocol shows that the different steps are merely composed of the four following main operators: Matrix-vector product: the multiplication of a vector by a random double circulant matrix; Hash function: the action of a hash function; Permutation: the generation and the action of a random permutation on words; PRNG: a pseudo-random generator used to generate random permutations and random vectors. The securisation is based on the fact that the linear operations (scalar products or bit-permutations) are easy to implement in hardware and are very efficient to secure against DPA attacks. We obtain an authentication in 5, 5 seconds and and a signature in 22 seconds for a security of The communication cost is around 40-kbits in the authentication scheme and around 140-kbits for the signature. It must be noticed that the timing performances would be highly improved by using a hardware SHA-256 instead of a software implementation. This implementation doesn t include countermeasure but the value of the overhead is really small ( 3). 4 Identity-based Signature Scheme In [2], the authors presente an IBI and an IBS scheme based on error-correcting code. Here is the scheme : Given C a q-ary linear code of length n and of dimension k. Let H be a parity check matrix of C. Given H = V HP with V invertible and P a matrix of permutation. Let h a hash function with values in {0, 1} n k. Let id A Alice s identity, id A can be compute by everyone. Similarly, H is public. The decomposition of H is, on the contrary, a secret of the authority and not of Alice. We shall describe an identity-based authentication method : Alice the prover is identifying herself to Bob the verifier. Alice has to authentify herself in a classic way, to get the private key which will then allow her to authentify herself to a third person as Bob. For that purpose, we use variation on identity. Let us admit that we know Alice s identity id A. Given h a hash function with values in {0, 1} n k. We search a way to find s E q,n,t such that h(id A ) = Hs T. The main point is to decode h(id A ). The main problem is that h(id A ) is not in principle in the arrival space of x Hx T. That is to say that h(id A ) is not in principle in the space of decodable elements of F n 2. That problem can be solved thanks to the CFS signature scheme. We get at the end a couple {s, j}, such that h(id A j) = Hs T. We can note that we have s of weight t or less. We use a slight derivation of Stern s protocol. We suppose that A obtained a couple {s, j} verifying h(id A j) = Hs T. h(id A j) is A s public key. The new protocol is based on Stern s protocol but with two changes (steps 1 and 4). The security of this system is the same as the security of Stern s one. It is possible to derive a signature scheme from the zero-knowledge authentication scheme of Stern by classical constructions. Hence it permits to derive an IBS scheme. This scheme is the first proposed non number theory based identity based scheme. The scheme combines two well known schemes and inherits from bad properties of these schemes: the public data is large, the communication cost for the IBI scheme is large and the signature length for the IBS scheme is also very large but besides these weaknesses our scheme presents the first alternative to number theory for ID-based cryptography and may open new area of research of 55

179 5 Kabatianskii Krouk Smeets Signature Scheme In [4], the authors investigate the security and the efficiency of their proposal. They show that a passive attacker who may intercept just a few signatures can recover the private key. Thay give precisely the number of signatures required to achieve this goal. This enables them to prove that all the schemes given in the original paper can be broken with at most 20 signatures. They improve the efficiency of these schemes by firstly providing parameters that enable to sign about 40 messages, and secondly, by describing a way to extend these few-times signatures into classical multi-time signatures. They finally study their key sizes and a mean to reduce them by means of more compact matrices (quasi-cyclics matrices). Such matrices are completely determined if the first row is known. An r n matrix M with r 2 and n 2 is a almost quasi-cyclic matrix if each row vector is rotated one element to the right relative to the preceding row vector: c 1 c c n c n c 1 c 2 c n 1 M = c n 1 c n c 1 c 2 c n c n r+2... c n 1 c n c 1 c 2... c n r+1 Their new scheme relies on the use of random almost quasi-cyclic codes rather than pure random linear codes. 6 Conclusion Recently a few improvements have been done in code based signature schemes. The reduction of the size of the public key is a real breakthrought for an use in low ressource context [3]. We have shown in this paper that the code based signature scheme could be interesting in different areas (embedded devices Section 3, identity based signature Section 4 and few times signatures schemes Section 5). Besides this, the code based cryptography has also the 3 following advantages : it can be an alternative to the number-theory based protocols in case a putative quantum computer may exist; code based cryptography is just based on linear operations (scalar products or bit-permutations); the secret key is smaller than the one of the other protocols (a few hundred bits) for the same security level. References 1. M. Bellare and P. Rogaway. Random oracles are practical : A paradigm for designing efficient protocols. In ACM Conference on Computer and Communications Security, pages 62 73, P.-L. Cayrel, P. Gaborit, and M. Girault. Identity-based identification and signature schemes using correcting codes. In D. Augot, N. Sendrier, and J.-P. Tillich, editors, WCC INRIA, P.-L. Cayrel, P. Gaborit, and E. Prouff. Secure implementation of the stern authentication and signature schemes for lowresource devices. CARDIS, P.-L. Cayrel, A. Otmani, and D. Vergnaud. On kabatianskii-krouk-smeets signatures. WAIFI 2007, Springer Carlet C. and Sunar B. LNCS: , N. Courtois, M. Finiasz, and N. Sendrier. How to achieve a mceliece based digital signature scheme. Springer-Verlag, W. Diffie and M. E. Hellman. New directions in cryptography. IEEE Transactions on Information Theory, IT-22(6): , A. Fiat and A. Shamir. How to prove yourself: Practical solutions to identification and signature problems. In C. Pomeronce, editor, Advances in Cryptology CRYPTO 87, volume 263 of LNCS, pages Springer-Verlag, G. Kabatianskii, E.Krouk, and B. J. M. Smeets. A digital signature scheme based on random error-correcting codes. IMA Int. Conf., Springer LNCS 1355: , Y. Li, R. Deng, and X. Wang. On the equivalence of mceliece s and niederreiter s cryptosystems. IEEE Trans. on Information Theory IT-40, 40 no. 1: , R. J. McEliece. A public-key cryptosystem based on algebraic coding theory. JPL DSN Progress Report, pages , H. Niederreiter. Knapsack-type cryptosystems and algebraic coding theory. Problems Control Inform. Theory, 15(2): , J. Stern. A new identification scheme based on syndrome decoding. In D. Stinson, editor, Advances in Cryptology CRYPTO 93, volume 773 of LNCS, pages Springer-Verlag, of 55

180 Building Stream Ciphers from FCSRs Dirk Stegemann Theoretical Computer Science University of Mannheim Mannheim, Germany 1 Introduction Stream ciphers are commonly used for fast encryption of arbitrarily long data streams. A common design principle is to use a keystream generator to produce a pseudorandom stream of running key bits z = (z t ) t 0, which is added bitwise to the plaintext stream (p t ) t 0 in order to obtain the ciphertext (c t ) t 0 with c t = p t z t. The receiver computes the same keystream z and recovers the plaintext via p t = c t z t. Many keystream generators operate as finite state machines (FSMs). Their initial state is derived from a secret key K and a public initialization vector IV by a procedure named key/iv setup, and in each clock t the FSM outputs a piece of keystream and changes its state according to the state transition function. For the construction of the FSM, linear feedback shift registers (LFSRs) have been widely used in practical applications due to their pseudorandomness properties and their efficiency. However, in the light of recent cryptanalysis results on LFSR-based stream ciphers (especially correlation attacks and algebraic attacks) it seems appropriate to look for other constructions that provide similar pseudorandomness properties but are more resistant to known attacks. Feedback with carry shift registers (cf. for example [4]), which we address in this paper, have been identified as possible alternatives. 2 Preliminaries We call a state of a finite state machine (e.g., an FCSR) periodic if, left to run, the machine will return to that same state after a finite number of steps. We call a sequence u = (u i ) i 0 strictly periodic (or simply periodic) with period T if u i+t = u i for all i 0. We call a sequence u eventually periodic if there exists a t 0 such that u = (u i ) i t is periodic. By wt(d), we denote the Hamming weight of a binary vector d. We will identify a vector (u 0,..., u k 1 ) {0, 1} k with the integer u = k 1 i=0 u i2 i. A 2-adic integer is a formal power series α = i=0 u i2 i with u i {0, 1}. The collection of all such formal power series forms the ring of 2-adic numbers. This ring especially contains rational numbers p/q, where p and q are integers and q is odd. 2-adic numbers and eventually periodic sequences are linked by the following Theorem [1]. Theorem 1 There is a one-to-one correspondence between rational numbers α = p/q (with odd q) and eventually periodic binary sequences u which associates to each such rational number α the bit sequence u = (u 0, u 1,...) of its 2-adic expansion. The sequence u is strictly periodic if and only if α 0 and α of 55

181 a n 2 a 1 a 0 x n 1 x 1 x 0 b div 2 mod 2 y n 1 y n 2 y 0 d 0 d 1 d n 1 d n 1 d n 2 d 1 d 0 Figure 1: FCSR in Galois and Fibonacci architecture 3 Feedback With Carry Shift Registers (FCSRs) 3.1 Galois and Fibonacci Architectures An FCSR of length n in Fibonacci architecture contains n binary register cells (y 0,..., y n 1 ) with fixed binary feedback taps (d 0,..., d n 1 ) and an additional l-bit memory b, see Fig. 1. From an initial configuration (y, b), the FCSR outputs in each clock t the value y 0, computes the sum σ = b + n 1 i=0 y id n i 1 over the integers and updates the register and memory according to b = σ div 2 and y = (σ mod 2, y n 1,..., y 1 ). If the Fibonacci FCSR is in a periodic state, we have 0 b < wt(d), i.e., we will need at most log 2 (wt(d) 1) + 1 memory bits to store b [4]. An FCSR of length n in Galois architecture contains n binary register cells x i with fixed binary feedback taps (d 0,..., d n 1 ) and n 1 memory cells a 0,..., a n 2, see Fig. 1. Starting from an initial configuration (x, a), the Galois FCSR outputs in each clock the value x 0, computes the sums σ i = x i+1 + a i d i + x 0 d i for 0 i < n (with x n = 0 and a n 1 = 0) and updates x i to σ i mod 2 and a i to σ i div 2 for all 0 i < n 1. We will assume that memory cells are only present at those positions with feedback taps, i.e., a i = 0 if d i = 0 for all 0 i < n Properties of FCSR-Sequences Identify a Galois state (x, a) with the integer p = x + 2a, a Fibonacci state (y, b) with the integer n 1 p = b2 n k=0 i=0 k q i y k i 2 k, and define for both architectures the connection integer q as q = 1 2d. Then the output of the FCSR is the 2-adic expansion of α = p q (cf. [1], Proposition 1, and [4], Theorem 2.4). Theorem 1 implies that the output will be strictly periodic if and only if 0 p q. Proposition 2 For an FCSR with connection integer q and a periodic initial state corresponding to p {0,..., q 1}, the sequence of integer representations of the states (p t ) t 0 is given by p t = 2 t p mod q and the t-th output bit can be computed as z t = p t mod 2 = (2 t p mod q) mod 2. If 0 < p < q, q odd, and p and q are coprime, Prop. 2 implies that the period of the sequence (p t ) t 0 is the order of 2 modulo q (cf. Theorem 2 in [1]). The period reaches its maximum q 1 if q is a prime for which 2 is a primitive root. We call such FCSRs maximum-length FCSRs and the 2 29 of 55

182 sequences they produce l-sequences. The properties of 2-adic numbers imply that an l-sequence consists of two half-periods, where the second half is the binary complement of the first [1]. For a binary sequence u, we denote by its 2-adic complexity the length of the shortest FCSR that produces the sequence. The 2-adic complexity along with an actual FCSR that produces the given sequence can be efficiently computed using FCSR-synthesis algorithms (cf. [6] for an overview). Clearly, the 2-adic complexity is upper-bounded by the length of the sequence, and we can expect the 2-adic complexity of a random sequence to be close to this bound, whereas its value will be small for an FCSR-sequence. This bias can be used to efficiently distinguish FCSR-sequences from truely random sequences. However, no other statistical tests except for computing the 2-adic complexity are known to reveal biases from the behaviour that is expected from truely random sequences. Particularly, the expected autocorrelation of maximum-length FCSR-sequences is zero [10], and the linear complexity is high [8]. Experiments indicate that FCSR-sequences pass the NIST statistical test suite [7]. 3.3 Particular Properties of the Galois Architecture Implementors will often prefer the Galois architecture to the Fibonacci architecture since the size of the memory is intrinsically limited and the memory bits can be updated in parallel. For a Galois FCSR, we have 0 p q. The output corresponds to the all-zero sequence if p = 0 and to the all-one sequence if p = q [1]. We call Galois states (x, a) with p {0, q } invariant states. The sequence (x i,t ) t 0 of values taken by the main register cell i is again an FCSR-sequence, more precisely the 2-adic expansion of p i /q with p i = F i (x, a) q + M i p, F i (x, a) = n 1 j=i (x j + 2a j )2 j i, and with constants M i = 2 n 1 j=i d j2 j i [2]. This expression can be further simplified for periodic initial states [3]. Proposition 3 For a maximum-length Galois FCSR with connection integer q, a periodic initial state (x(0), a(0)), and p 0 = x + 2a, the sequence (x i,t ) t 0 of values taken by a fixed main register cell i corresponds to (p t+si mod 2) t 0 with s i = log 2 (M i ) mod q and M i = 2 n 1 j=i d j2 j i. Proposition 3 implies that the sequence (x i,t ) t 0 corresponds to the sequence produced by the whole FCSR (which is in turn equal to the sequence (x 0,t ) t 0 ) shifted by s i positions. Note that the phase shifts s i are independent of the initial state p and depend on i and q only. Note that although equivalent states produce the same output, the sequence of states (x(t), a(t)) t 0 obtained by running the FCSR from equivalent starting states may be different. 3.4 Mappings between Galois and Fibonacci States There is an onto function E : {(x, a)} {(1,..., 1; a 0,..., a n 2 )} Z q that assigns to a Galois state (x, a) the number E(x, a) = x + 2a mod q [4]. Moreover, there exists a one to one mapping S from Z q onto the set L of strictly periodic states of the Fibonacci FCSR with connection integer q except for the state (1,..., 1; wt(q + 1) 1). More precisely, the bijective mapping S assigns to a p Z q the initial Fibonacci main register state y i = ((2 i p mod q) mod 2) for 0 i n 1 and the initial memory state ( ) b = 1 n 1 k 2 n p + q i y k i 2 k. k=0 i= of 55

183 Conversely, for a given periodic Fibonacci state (y, b) the corresponding integer p will satisfy 0 p < q. Obviously, the mapping E from the Galois states to Z q is not one to one, i.e., generally more than one state is mapped to the same p Z q. However, we can compute for a given p Z q the uniquely determined corresponding periodic state (x, a) [3]. Proposition 4 For all p Z q, the only strictly periodic state (x, a) with x+2a = p of a maximumlength Galois FCSR with connection integer q is given by x i = M i p mod q mod 2 and a = (p x)/2 with M i defined as in Prop. 3. For an initial state of a Galois FCSR with connection integer q, we can thus compute a (periodic) initial state of a Fibonacci FCSR with connection integer q (and vice versa) such that the two registers will produce the same output (cf. [4], Corollary 3.7). Moreover, we can map between periodic Fibonacci states and periodic Galois states using Prop. 4. Corollary 5 Let R g and R f denote Galois resp. Fibonacci FCSRs with the same connection integer q. Then there is a bijective mapping T between the periodic states of R g and R f. 4 From FCSRs to Stream Ciphers We have observed in Sect. 3.2 that FCSR-sequences and especially l-sequences have many desirable pseudorandomness properties but low 2-adic complexity, which prevents them from being used directly as keystream. In the context of LFSRs, which suffer from a similar weakness (their low, efficiently computable linear complexity), popular approaches are combination and filter generators. A combination generator consists of a small number of feedback shift registers and a Boolean function f that combines the output sequences of the internal registers in order to produce the output keystream. A filter generator contains only one feedback shift register and a Boolean filter function g that produces the output keystream from the current content of certain register cells. It seems natural to apply the ideas of combination and filter generators to the FCSR-context. This has been done in the case of the F-FCSR stream cipher family [1], which is to the best of our knowledge the first FCSR-based stream cipher. F-FCSR is a filter generator based on a single Galois FCSR and a filter function g that simply computes the binary XOR of its inputs. The initialization procedure of F-FCSR ensures that the initial state of the generator is periodic. Hence, by Prop. 3, the keystream generation procedure of F-FCSR is equivalent to taking the bitwise XOR-sum of different parts of the same FCSR-sequence. This design is motivated by the conjecture that linear and 2-adic operations are unrelated and that the correlation between two distant parts of the same FCSR-sequence is low. At present, we are not aware of any cryptanalytical results that contradict this assumption. 1 Proposition 3 further implies that a Galois-based filter generator like F-FCSR can be equivalently represented as a combination generator based on Galois FCSRs that contains as many FCSRs as the filter function g has inputs, where the FCSR producing the sequence (x i,t ) t 0 is initialized with p si. Furthermore, one or more Galois registers in the combination generator may be replaced by Fibonacci registers producing the same output according to Corollary 5. We can also build an equivalent filter generator based on a Fibonacci FCSR with the following result [3]. 1 Generic attack strategies have been reported in [1] and [9] but are less efficient than exhaustive search for wellchosen cipher parameters of 55

184 Proposition 6 The value x i of the i-th cell in the main register of the Galois FCSR can be computed from the (strictly) periodic state (y, b) of the corresponding Fibonacci FCSR by n 1 k x i = M i b2 n d j 1 y k j 2 k mod q mod 2. k=0 j=0 Note that these representations are all equivalent in the sense that they produce the same keystream sequence from a given (possibly transformed) starting state, but one representation may be more suitable for further cryptanalysis than the other. Finally, we want to point out that the security of the keystream generation (based on a secret initial state) does not imply the security of the whole cipher, since the initial state of the FSM is not entirely secret but partly depends on the publicly known initialization vector. In fact, many stream ciphers (particularly an early version of F-FCSR [5]) have been broken by exploiting weaknesses in the key/iv setup. References [1] F. Arnault and T. P. Berger. Design and properties of a new pseudorandom generator based on a filtered FCSR automaton. IEEE Trans.Comp., 54(11): , [2] F. Arnault, T. P. Berger, and M. Minier. Some results on FCSR automata with applications to the security of FCSR-based pseudorandom generators. IEEE Trans. Inform. Theory, 54(2): , [3] S. Fischer, W. Meier, and D. Stegemann. Equivalent representations of the F-FCSR keystream generator. In Christophe de Cannière and Orr Dunkelman, editors, The State of the Art of Stream Ciphers (SASC 2008) Workshop Record, pages 87 96, [4] M. Goresky and A. Klapper. Fibonacci and Galois representations of feedback-with-carry shift registers. IEEE Trans. Inform. Theory, 48(11): , [5] È. Jaulmes and F. Muller. Cryptanalysis of the F-FCSR Stream Cipher Family In B. Preneel and S. Tavares, editors, Proc. of Selected Areas in Cryptography (SAC 2005), volume 3897 of LNCS, pages Springer, [6] A. Klapper. A survey of feedback with carry shift registers. In T. Helleseth et al., editors, Proc. of SETA 2004, volume 3486 of LNCS, pages Springer, [7] NIST. A statistical test suite for the validation of random number generators and pseudo randoom number generators for cryptographic applications [8] C. Seo, S. Lee, Y. Sung, K. Han, and S. Kim. A lower bound on the linear span of an FCSR. IEEE Trans. Inform. Theory, 46(2): , March [9] D. Stegemann. Extended BDD-based cryptanalysis of keystream generators. In C. Adams, A. Miri, and M. Wiener, editors, Proc. of Selected Areas in Cryptography (SAC 2007), volume 4876 of LNCS, pages Springer, [10] H. Xu and W.-F. Qi. Autocorrelations of maximum-length FCSR sequences. SIAM J. Discrete Math., 20(3): , of 55

185 Werner Rozek FH Schmalkalden Verfahren zur Authentisierung und Authentifizierung von Personen und Einheiten Das Verfahren basiert auf lokale Security Einheiten und auf personzugeordnete Einheiten, den so genannten Sicherheitsidentifikationseinheiten (SID Einheiten). Die Grundlagen für die Authentisierung und Authentifizierung sind: biometrische Merkmale ein, allen Einheiten, bekanntes, nach außen geheimes, für ein zufallsbestimmten Zeitabschnitt gültiges, globales Zufallsbezugsdatum 80 Bit lange öffentliche und beglaubigte Person und/oder Einheit kennzeichnende Daten lokal, am Ort des Absenders, generierte Zufallsdaten (128Bit,16 Bit Zufallsdaten u. a. m.) dynamisch zufällige, global oder separat geltende Räume, Zufallsbezugspunkte, Translations / Rotationsvektoren relative Daten Authentisierung Unter Authentisierung versteht man die Echtheitsprüfung der eigenen Identität durch sich selbst. Sie erfolgt beim Verfahren in der Regel durch die personbezogene und personzugeordnete SID Einheit anhand von biometrischen Merkmalen wie z. B. Fingerabdruckdaten. Das besondere Merkmal des Verfahrens ist, dass die Fingerabdruckdaten nur in der SID Einheit gespeichert werden. Die Fingerabdruckdaten verbleiben somit immer bei der Person. Sie werden nicht mit anderen Personen ausgetauscht. Authentifizierung Der Begriff Authentifizierung bezeichnet die Echtheitsprüfung des Gegenübers vom Gegenüber. Zur Authentifizierung sieht das Verfahren öffentliche und beglaubigte Anschriftdaten (Adresse, Identität) vor. Öffentlich sind Person kennzeichnende Daten wie Name, Geburtsdaten, Wohnort u. a. m. Verfahrenstechnisches, lokal zufallsbestimmtes Einweben von 16Bit Zufallsmerkmalen in den Anschriftdaten ermöglicht die unmanipulierbare Authentifizierung Die 80Bit langen beglaubigten Anschriftdaten [= beglaubigte öffentliche Anschriftdaten in der SID Einheit] des Absenders und die 80Bit langen öffentlichen Anschriftdaten des Adressaten werden am Ort des Absenders mit je einem 16Bit langen Zufallsdatum zu 96Bit langen geheimen Anschriftdaten des Absenders und des Adressaten verflochten. Die 16Bit langen Zufallsdaten und deren zugeordneten Flechtsteuerdaten werden am Ort des Absenders zufallsbestimmt aus dem globalen Zufallsbezugsdatum abgelesen. Der Ort des Ablesens wird in Form eines zwei Byte langen Bitortdatums innerhalb eines relativen 128Bit langen Steuerdatums zum Ort des Adressaten übertragen. 33 of 55

186 Dynamisch zufällig sich ändernde globale Räume, Zufallsbezugsdaten und Translations / Rotationsvektoren sorgen für Geheimnisschutz und Unmanipulierbarkeit Mit Hilfe von Transformationen in ein dynamisch zufällig ändernden Anschriftenraum und Bezugsraum werden die 96Bit langen geheimen Anschriftendaten in 128Bit lange geheime Anschriftendaten umgerechnet, aus denen in Bezug auf lokal bestimmte Zufallsbezugsdaten und zentral vorgegebenen Translations / Rotationsvektoren ihre relativen Daten bestimmt werden. Geschickte Bezüge der Anschriftdaten ermöglichen die Echtheitsprüfung des Gegenübers Um die Authentifizierung des Absenders durch den Adressaten auch bei vollständiger Kenntnis aller Person kennzeichnenden Daten unmanipulierbar zu gestalten, sind bei der Berechnung ihrer relativen Daten am Ort des Absenders die geheimen öffentlichen Anschriftdaten des Adressaten auf die geheimen, beglaubigten Anschriftdaten des Absenders indirekt bezogen. Durch diese Bezüge kann der Adressat am Ort des Adressaten die Echtheit seiner öffentlichen Anschriftdaten anhand seiner beglaubigten Anschriftdaten überprüfen, wobei gleichzeitig die Echtheit der beglaubigten Anschriftdaten des Absenders kontrolliert ist. Dynamisch zufällig sich ändernde separate Räume, Zufallsbezugsdaten und Translations / Rotationsvektoren erhöhen die Unmanipulierbarkeit Die relativen Daten der 128Bit langen geheimen Anschriftdaten des Absenders und des Adressaten werden mit weiteren relativen Daten, die auf separate Räume und Zufallsbezugspunkte und Translations / Rotationsvektoren bezogen sind, in vordefinierter Reihenfolge aneinander gereiht und mit ihrem Hashwert versehen. Der Hashwert wird in Teile der aneinander gereihten Zahlenfolge geflochten. Die geflochtene Zahlenfolge wird in mehrere 1024Bit lange Teildaten zerlegt. Das erste 1024Bit lange Teildatum wird in Bezug auf ein globalen Transferraum, Zufallstransferbezugsdatum und Translations /Rotationsvektor in sein relatives Datum umgeformt. Aus allen anderen 1024Bit langen Teildaten werden in Bezug auf ein separaten Transferraum, auf separate Zufallstransferbezugsdaten und Translations / Rotationsvektoren ihre 1024Bit langen relativen Daten ermittelt. Byteweise zufallsbestimmtes Verweben relativer Datenbestandteile verhindert die Zuordnung zum relativen Datum Das erste 1024Bit lange, relative Datum wird zufallsbestimmt und byteweise in die folgenden 1024Bit langen relativen Daten eingeflochten, so dass die Zuordnungen zum jeweiligen relativen Datum verloren gehen. Der so entstandene Datenstrom wird zum Adressaten gesendet. Die lokale Security Einheit vom Adressaten entflechtet die geflochtenen relativen Daten und bestimmt aus den relativen Daten die geheimen Anschriftdaten. Am Ort des Adressaten prüft die SID Einheit des Adressaten im Zusammenspiel mit der lokalen Security Einheit oder nach Vollmachterteilung die lokale Security Einheit allein, ob er berechtigter Empfänger der Nachricht ist. Unter Verwendung des Bitortdatums sind die Entflechtsteuerinformationen für die 16Bit Zufallsmerkmale bekannt. Nach dem Entflechten stehen die vom Absender übermittelten beglaubigten Anschriftdaten des Absenders und öffentlichen An Prof. Dr. Ing. W. Rozek Tel: FH Schmalkalden Fax: FB Elektrotechnik w.rozek@fh sm.de 34 of 55

187 schriftdaten des Adressaten sowie die jeweiligen 16Bit Zufallsmerkmale für die Echtheitsprüfung zur Verfügung. Der Vergleich der vom Absender übermittelten Anschriftdaten mit den beglaubigten Anschriftdaten des Adressaten zeigt die Echtheit der Anschriftdaten. Da die übermittelten relativen geheimen Anschriftdaten auf die beglaubigten Anschriftdaten des Absenders bezogen wurden, ist gleichzeitig die Echtheit der beglaubigten Anschriftdaten des Absenders überprüft. Vergleiche der entflochtenen Zufallsmerkmale mit den intern bekannten Merkmalen unterstützen zusätzlich die Echtheitsprüfung. Anwendungsmöglichkeiten 1. Allen P2P Kommunikationen 2. EC Karte, Gesundheitskarte, Kundenkarte, Ausweis 3. Authentisierte und authentifizierte Geldbörse 4. Datei und Ordnerzugriff 5. Ressourcenzugang u. v. a. m. Literatur: Prof. Dr. Ing. W. Rozek, Dipl. Ing. Th. Rozek, Fachinformatiker J. Rozek: Verfahren zur Authentisierung und Authentifizierung von Personen und Einheiten. DE Prof. Dr. Ing. W. Rozek Tel: FH Schmalkalden Fax: FB Elektrotechnik sm.de 35 of 55

188 Slid Pairs in Salsa20 Deike Priemuth-Schmid and Alex Biryukov FSTC, University of Luxembourg 6, rue Richard Coudenhove-Kalergi, L-1359 Luxembourg (deike.priemuth-schmid, The stream cipher Salsa20 is one of the finalists of the estream project [4] which is in the final portfolio of new promising stream ciphers. In 2005 Bernstein [2] submitted the original Salsa20 with 20 rounds, later 8 and 12 rounds versions Salsa20/8 and Salsa20/12 were also proposed. We show that initialization and key-stream generation of this cipher is slidable, i.e. one can find distinct (Key, IV) pairs that produce closely related key-streams. Previous attacks on Salsa used differential cryptanalysis exploiting a truncated differential over three or four rounds. The first attack was presented by Crowley [3] which could break the 5 round version of Salsa20 within claimed trials. Later a four round differential was exploited by Fischer et al. [5] to break 6 rounds in trials and by Tsnunoo et al. [6] to break 7 rounds in about trials. The currently best attack by Aumasson et al. [1] covers 8 round version of Salsa20 with estimated complexity of We show that the following observation holds: suppose that you are given two black boxes, one with Salsa20 and one with a random mapping. The attacker is allowed to chose a relation F for a pair of inputs, after which a secret initial input x is chosen and a pair (x, F(x)) is encrypted either by Salsa20 or by a random mapping. We stress that only the relation F is known to the attacker. The goal of the attacker is given a pair of ciphertexts to tell whether they were encrypted by Salsa20 or by a random mapping. To make the life of the attacker more difficult the pair may be hidden in a large collection of other ciphertexts. It is clear that for a truly random mapping no useful relation F would exist and moreover there is no way of checking a large list except for checking all the pairs or doing a birthday attack. On the other hand Salsa20 can be easily distinguished from random in both scenarios if F is a carefully selected function related to the round-structure of Salsa20. Moreover since in Salsa20 the initial state is initialized with the secret key, known nonce, counter and constant it is not only a distinguishing but also a complete key-recovery attack. Our attacks are independent of the number of rounds in Salsa and thus work for all the 3 versions of Salsa. We also show a general birthday attack on 256-bit key Salsa20 with complexity which can be further sped up twice using sliding observations. Brief Description of Salsa20 The Salsa20 encryption function uses the Salsa20 hash function in a counter mode. The internal state of Salsa20 is a matrix of 32-bit words. A vector (y 0, y 1, y 2, y 3 ) of four words is transformed into (z 0, z 1, z 2, z 3 ) by calculating 1 z 1 = y 1 ((y 0 + y 3 ) 7), z 2 = y 2 ((z 1 + y 0 ) 9), z 3 = y 3 ((z 2 + z 1 ) 13), z 0 = y 0 ((z 3 + z 2 ) 18). 1 Here the symbol + denotes the addition modulo 2 32, the other two symbols work at the level of the bits with as XOR-addition and as a shift of bits of 55

189 This nonlinear operation is called quarterround and it is the basic part of the columnround where it is applied to columns (y 0, y 4, y 8, y 12 ), (y 5, y 9, y 13, y 1 ), (y 10, y 14, y 2, y 6 ) and (y 15, y 3, y 7, y 11 ) as well as of the rowround which transforms rows (y 0, y 1, y 2, y 3 ), (y 4, y 5, y 6, y 7 ), (y 8, y 9, y 10, y 11 ) and (y 12, y 13, y 14, y 15 ). A so called doubleround consists of a columnround followed by a rowround. The doubleround function of Salsa20 is repeated 10 times. If Y denotes the matrix a key-stream block is defined by Z = Y + doubleround 10 (Y ). In this feedforward the symbol + denotes the addition modulo Salsa20 has 976 word operations in total for one encryption. The cipher takes as input a 256-bit key (k 0,..., k 7 ), a 64-bit nonce (n 0, n 1 ) and a 64-bit counter (c 0, c 1 ). A 128-bit key version of Salsa20 copies the 128-bit key twice. We mainly concentrate on the 256-bit key version. The remaining four words are set to fixed publicly known constants, denoted with σ 0, σ 1, σ 2 and σ 3. Slid Pairs The structure of a doubleround can be rewritten as columnround then a matrix transposition another columnround followed by a second transposition. Now the 10 doublerounds can be transferred into 20 columnrounds each followed by a transposition. We define F to be a function which consists of a columnround followed by a transposition. If we have 2 triples (key1, nonce1, counter1) and (key2, nonce2, counter2) so that starting state (key2, nonce2, counter2) = F [ starting state (key1, nonce1, counter1)] then this property holds for each point during the round computation of Salsa20 especially for the end of the round computation. Pay attention that the feedforward at the end of Salsa20 destroys this property. We call such a pair of a 1 st and 2 nd starting state a slid pair. The relation of a slid pair is shown in Fig. 1. In a starting state four words are constants and 12 words can be chosen freely which leads to a total amount of possible starting states. If we want that a starting state after applying function F results in a 2 nd starting state we obtain four wordwise equations. This means we can choose eight words of the 1 st starting state freely whereas the other four words are determined by the equations as well as the words for the 2 nd starting state. This leads to a total amount of possible slid pairs. For the 128-bit key version no such slid pair exists due to the additional constrains of four fewer words freedom in the 1 st starting state and four more equations in the 2 nd starting state. S F 19 F X Z S 19 F F X Z Figure 1: relation of a slid pair 2 37 of 55

190 With function F we get two equations S = F(S) and X = F(X). The words for the matrices of the two starting states we denote with S = σ 0 k 0 k 1 k 2 k 3 σ 1 n 0 n 1 c 0 c 1 σ 2 k 4 k 5 k 6 k 7 σ 3 S = σ 0 k 0 k 1 k 2 k 3 σ 1 n 0 n 1 c 0 c 1 σ 2 k 4 k 5 k 6 k 7 σ 3 The set up of the system of equations for a whole Salsa20 computation is too complicated but the equations for the computation of F are very clear. Due to the eight words freedom we have in a 1 st or 2 nd starting state there are some relations in the 12 non-fixed words. For the 2 nd starting state these relations are very clear as they deal only with words 0 = k 2 + k 1, 0 = k 3 + n 1, 0 = c 1 + c 0 and 0 = k 7 + k 6, (1) whereas for the 1 st starting state these relations depend on the bits and thus are more complicated. Sliding by the function F is applicable to any version of Salsa20/r where r is even. For r odd there would be no transposition at the end of the round computation, equations are a bit different, though still solvable. Related Key Key-Recovery Attack Let us assume that we know two ciphertexts and the corresponding nonces and counters. We do not know both keys but we know that both starting states differ by function F which gives the relation shown in Fig. 2 and that the 2 nd starting state conforms to the initial state format of Salsa20 (i.e. has proper 128-bit constant on the diagonal). We can show that this information is sufficient to completely recover the two related 256-bit secret keys of Salsa20 with O(1) complexity.. σ 0 k 0 k 1 k 2 σ 0 k 3 c 0 k 5 σ 0 k 0 k 1 k 2 k 3 σ 1 n 0 n 1 column k 0 σ 1 c 1 k 6 trans- k 3 σ 1 n 0 n 1 c c σ k round k 7 position k σ 2 c 1 n 0 k 1 0 c 0 1 σ 2 k k σ 3 k 7 n 1 4 σ 3 k k 7 5 k 6 k 2 k 5 6 σ 3 Figure 2: relation of the 1 st and 2 nd starting state A Generalized Related Key Attack on Salsa20 Suppose we are given a (possibly large) list of ciphertexts with the corresponding nonces and counters and we are told that in this list the slid pair is hidden. The question is, can we find slid pairs in a large list of ciphertexts efficiently? As shown for the key-recovery attack, given such slid ciphertext pair it is easy to compute both keys. The task is made more difficult by the feedforward of Salsa20, which destroys the sliding relationship. Nevertheless we can show that given a list of ciphertexts of size O(2 l ) it is possible to detect a slid pair with memory and time complexity of just O(2 l ). The naive approach which would require to check for each possible pair the equations from function F will have complexity O(2 2l ) 3 38 of 55

191 Table 1: Complexities for different list seizes list size memory time which is too expensive. Our idea is to reduce the amount of potential pairs by sorting them by eight precomputed words, so that only elements where these eight words match have the possibility to yield a slid pair. After decreasing the number of possible pairs in that way we can check the remaining pairs using additional constraints coming from the sliding equations. In total we have at least filtering power of bits. Thus we expect that only the correct slid pairs survive this check. The remaining pairs are the correct slid pairs for which we completely know both keys. A summary for the complexity of different lists is given in Table 1. The memory is given in words and the time in Salsa encryptions. Time-Memory Tradeoff Attacks on Salsa Salsa20 has possible starting states. We notice that this is less than the square root of the keyspace size for keys longer than 192-bits. Thus a trivial birthday attack on 256-bit key Salsa20 would proceed as follows: During the preprocessing stage generate a list of randomly chosen starting states and run for each of them Salsa20 to get a sample of ciphertexts. Afterwards we sort this list by the ciphertexts. During the on-line stage we capture ciphertexts for which we want to find the keys. We do not have to store these ciphertexts and can check each of them immediately for a match with the sorted array of precomputed ciphertexts. If we have a match we retrieve the corresponding key from our table. Of course due to very high memory complexity this attack can be only viewed as a certificational weakness. If we can choose the nonce or the counter there exist only different starting states reducing the attack to precomputation and memory complexity of and if we can choose both the state space drops to and the attack complexity drops to Similar reasoning for 128-bit key Salsa20 would yield attack with 2 64 complexity. Thus it is crucial for the security of Salsa20 that nonces are chosen at random for each new key and the counter is not stuck at some fixed value (like 0, for example). The complexities are summarized in Table 2 where R stands for a complete run of Salsa20 and M for a matrix of Salsa (16 words). Improved Birthday Using the Sliding Property. We can use the sliding property to increase the efficiency of the birthday attack twice (which can be translated into reduction of memory, time or increase of success probability of the birthday attack). During the preprocessing stage we generate a sample of 2 nd starting states by using (1) and choose the rest eight words at random. We compute the corresponding ciphertexts for these states as well as the eight specified words for the corresponding 1 st starting states. We use two kinds of pointers to sort this generated list by the ciphertexts for the 2 nd starting and by the eight words for the corresponding 1 st starting state. We capture ciphertext from the key-stream where we also know the nonce and the counter 4 39 of 55

192 Table 2: Complexities for the Birthday attack attack precomputation memory time captured ciphert. chosen nonce and counter R M chosen nonce or counter R M general R M using sliding property R M and check if it is matching a 2 nd starting state from our list (direct birthday) or is a correct 1 st starting state for one of the states from our collection (indirect birthday). In both cases we learn the key for this ciphertext. Conclusion We have described sliding property of Salsa20 which leads to distinguishing, key recovery and related-key attacks on this cipher. We also show that Salsa20 does not offer 256-bit security due to a simple birthday attack on its 384-bit state. Since the likelihood of falling in our related key classes by chance is relatively low (2 256 out of ) this attack does not threaten most of the real-life usage of this cipher. However designer of protocols which would use this primitive should be definitely aware of these non-randomness property, which can be exploited in certain scenarios. References [1] J.-P. Aumasson, S. Fischer, S. Khazaei, W. Meier and C. Rechberger. New Features of Latin Dances: Analysis of Salsa, ChaCha and Rumba To appear in Proceedings of Fast Software Encryption 2008 (FSE 2008), LNCS, Lausanne, Switzerland, February 10-13, Full version as IACR eprint, [2] D. J. Bernstein. Salsa20. estream, ECRYPT Stream Cipher Project, Report 2005/025, [3] P. Crowley. Truncated differential cryptanalysis of five rounds of Salsa20. SASC Stream Ciphers Revisited, [4] estream: The ECRYPT Stream Cipher Projekt, [5] S. Fischer, W. Meier, C. Berbain, J.-F. Biasse, M. J. B. Robshaw. Non-randomness in estream Candidates Salsa20 and TSC-4. INDOCRYPT, volume 4329 of LNCS, pages Springer, [6] Y. Tsunoo, T. Saito, H. Kubo, T. Suzaki and H. Nakashima. Differential Cryptanalysis of Salsa20/8. SASC The State of the Art of Stream Ciphers, of 55

193 Joint State Theorems for Public-Key Encryption and Digital Signature Functionalities with Local Computation Max Tuengerthal (joint work with Ralf Küsters) Universität Trier Composition theorems in simulation-based approaches allow to build complex protocols from subprotocols in a modular way. However, as first pointed out and studied by Canetti and Rabin [3], this modular approach often leads to impractical implementations. For example, when using a functionality for digital signatures within a more complex protocol, parties have to generate new verification and signing keys for every session of the protocol. This motivates to generalize composition theorems to so-called joint state theorems, where different copies of a functionality may share some state, e.g., the same verification and signing keys. In this paper [5], we present a joint state theorem which is more general than the original theorem of Canetti and Rabin [3], for which several problems and limitations are pointed out. We apply our theorem to obtain joint state realizations for three functionalities: public-key encryption, replayable public-key encryption, and digital signatures. Unlike most other formulations, our functionalities model that ciphertexts and signatures are computed locally, rather than being provided by the adversary. To obtain the joint state realizations, the functionalities have to be designed carefully. Other formulations are shown to be unsuitable. Our work is based on a recently proposed, rigorous model for simulation-based security by Küsters [4], called the IITM model. Our definitions and results demonstrate the expressivity and simplicity of this model. For example, unlike Canetti s UC model [1, 2], in the IITM model no explicit joint state operator needs to be defined and the joint state theorem follows immediately from the composition theorem in the IITM model. References [1] R. Canetti. Universally Composable Security: A New Paradigm for Cryptographic Protocols. In Proceedings of the 42nd IEEE Symposium on Foundations of Computer Science (FOCS 2001), pages IEEE Computer Society, [2] R. Canetti. Universally Composable Security: A New Paradigm for Cryptographic Protocols. Technical Report 2000/067, Cryptology eprint Archive, December Available at eprint.iacr.org/2000/067. [3] R. Canetti and T. Rabin. Universal Composition with Joint State. In D. Boneh, editor, Advances in Cryptology, 23rd Annual International Cryptology Conference (CRYPTO 2003), Proceedings, volume 2729 of Lecture Notes in Computer Science, pages Springer, [4] R. Küsters. Simulation-Based Security with Inexhaustible Interactive Turing Machines. In Proceedings of the 19th IEEE Computer Security Foundations Workshop (CSFW ), pages IEEE Computer Society, [5] R. Küsters and M. Tuengerthal. Joint State Theorems for Public-Key Encryption and Digital Signature Functionalities with Local Computation. In 21st IEEE Computer Security Foundations Symposium (CSF 2008), pages IEEE Computer Society, of 55

194 Anargumentforexacttravellingsalesmanproblem VadymFedyukovych Abstract We introduce an argument of knowledge protocol for a variant of travelling salesman problem. We consider a problemtodecidewhetherahamiltonpathofagivencostexistsinagivengraph. OurprotocolallowsaProver to show his knowledge of a solution to this problem without giving any useful information about his solution, and withoutachanceforaverifiertoshowexistenceofasolutiontoanythirdparty.ourprotocolusepolynomialgraph representation and a novel proof system to show properties of polynomials. Our protocol is of knowledge and is special honest verifier zero knowledge. 1 Introduction Travelling salesman is a classical combinatorial problem. This problem is an alternative to factoring and discrete logarithm useful for cryptographic applications. A proof of knowledge protocol for Hamilton cycle with binary challenges was introduced by Blum[1](citing by[2]). ArelatedproblemtodecidewhetheraHamiltonpathofagivencost(accordingtoacostmatrix)existsinagraphis Exact Travelling Salesman Problem(XTSP) and is known to be NP-hard[8]. An authentication protocol with ternary challenges was introduced in[9]. Weconsidera modular variantofxtspwithcostsfromafinitefield F q forsomeprimeq. Wealsoconsider thecaseofaprimenumberofverticesinthegraph.ourresultalsoapplyforahamiltonpathwithdifferentstart/end nodes. WeintroduceanefficientprotocoltoletaProvershowhisknowledgeofasolutiontoXTSP.Ourprotocoluses Verifier challenges chosen from a large set, and achieve soundness without repeating. This protocol uses a novel polynomial graph representation such that existence and cost of a Hamilton cycle are show as properties of polynomials. Soundness of this protocol is based on a probability estimate to choose a root of a polynomial at random(schwartz- Zippel lemma[13]). Protocol is an argument of knowledge on assumption of taking logarithms is hard for Prover in the group used. Protocol is of knowledge, and is special honest verifier information-theoretic zero knowledge. A polynomial set representation with a characteristic function was introduced for set reconciliation protocol[10]. A proof protocol for set similarity was introduced in[3]. Protocols for graph isomorphism and colourability were introduced in[5]. Three new protocols to show properties of polynomials were introduced in[6]. A protocol to show knowledgeofacodewordofgoppacode,codeparameters,andaanerrorofaboundedweightwasintroducedin[4]. A protocol to show substring matching was introduced in[7]. 2 Preliminaries Let F q beafinitefieldforsomeprimeq.consideradirectedgraph Γ.Lettherebeasubgraph P(Γ)thatisaHamilton cycle.let (a ij )becostsmatrixwitha ij F q.cyclecostbisthesumofcostsassociatedwithedgesinthecycle. Exacttravellingsalesmanproblem(XSTP)istodecidewhetheraHamiltoncycleofagivencostbexistsinthe givengraph Γforacostmatrix (a ij ). InthisreportweconsideraproblemtodecidewhetheraHamiltoncycleofagivencostexistsinthegivengraph suchthatl= V(Γ) isaprime.weusecommitmentstothecostofthecycleandtocostsmatrixasproverinput. WeekendofCryptography2008,Tabarz,Germany of 55

195 WeusePedersencommitments[11] producedwithtwogroupelementsg G,h GoforderqsuchthatProver doesnotknowlog h (g). WeusevalidresponsesofSchnorrprotocol[12]toproducepolynomialsinchallengeof Verifier.WeconsiderL,qchosensuchthatL q 1andL 2 q 1. 3 Protocoldesign WemarkedgesofthegraphthatarethesolutiontoXTSPwith1flags,andmarkallotheredgeswith0flags.Prover commit to all flags and run Schnorr protocol to show knowledge of values committed. Weuseamultiplicative(moduloq)cyclicgroup G L oforderl.verifierchoosessomegeneratorcofthisgroupas hischallenge.proverassignsvaluesc k G L toverticesofthegraphthatarehissolutiontoxtspproblem.prover commit to the values assigned to all vertices and run Schnorr protocol to show knowledge of values committed. Let {U e }bepedersencommitmentstoflagsedgesofthegrapharemarkedwithandlet {W v }bepedersencommitments to values assigned to vertices of the graph: Let {Φ e },{Θ v }beresponsesofschnorrprotocolwithchallenges (e,d): Consider a graph verification polynomial: U e =g ue h βe u e {0,1} e E(Γ) (1) W v =g wv h αv v V(Γ) (2) Φ e =eu e + δ e e E(Γ) (3) Θ v =dw v + γ v v V(Γ) (4) F(x,y,c,d,e;Γ) = (e + Φ HjT (d +xθ Hj +yθ Tj )) (5) j H jt j E(Γ) Inthefollowing,weonlyconsidervalidresponses {Θ v },{Φ e }producedaccordingto(3,4).verificationpolynomial isofpowern = E(Γ) inesuchthatit stopcoefficientisaproductofpolynomialsrelatedtoedgesofthecycle: NowF n (x,y)isapolynomialindofpowerl: n F(x,y,c,d,e;Γ) = e i F i (x,y,c,d) (6) i=0 F n (x,y,c,d) = (d +xθ Hj +yθ Tj ) (7) H jt j E(P) L F n (x,y,c,d) = d l f l (x,y,c) (8) l=0 L 1 f L (x,y,c) = (1 +xw Hj +yw Tj ) = (1 +xc k +yc k+1 ) (9) H jt j E(P) k=0 Itcanbeshownf L (x,y,c)isoftheform(9)ifandonlyifedgesselectedwith 1 flagsareacycle.verifiercanevaluate f L (x,y,c)onlyusinghischallenges,andcanevaluatef(x,y,c,d,e;γ)usinghischallengesandresponsesofprover. ProvercanassistVerifierwhiletestingthatF(x,y,c,d,e;Γ)matchesf L (x,y,c). Leta ij becostassociatedwithpathfromitoj,bbecostofthesolutiontoxtsp,andletb,{a ij }bepedersen commitments to costs: Let,{Λ ij }beresponsesofschnorrprotocol: A ij =g aij h rij B =g b h t (10) Λ ij =ea ij + η ij (11) =eb + σ (12) 2 43 of 55

196 ConsideracostverificationpolynomialforthegivencostBofthewholepathoftheform: M(e;Γ) = e + Λ HjT j Φ HjT =m j 2 e 2 +m 1 e +m 0 (13) H jt j E(Γ) NowVerifierassessXTSPsolutionbytestingthatM(e)isatmostlinearine:m 2 =0. 4 Informal description of the protocol Weassumecommitmentstocostsmatrix,tocostofsolution,andtoflagsassignedtoedgesareinputtotheprotocol. Verifier chooses a proper challenge c and sends it to Prover. Prover responds with commitments to values assigned to vertices. Verifiers chooses some pair (x, y) as his second challenge and sends it to Prover. Prover sends commitments to expansioncoefficientsf l (x,y,c),l =0...L 1.Verifierproducesf L (x,y,c)withoutanyhelpofprover. Verifierschoosesandsendsanotherchallenged.Proverproducesresponses Θ v.proveralsoproducescommitments tocoefficientsf i (x,y,c,d),i =0...n 1.Atthisstep,VerifierhasinformationaboutF n (x,y,c,d)alreadythatheneeds totestthesolution.proverproducescommitmentstom 0 andm 1. Verifierchoosesandsendshislastchallengee.Proverproducesresponses Φ e,λ ij,. NowVerifierteststhatM(e) =m 1 e +m 0 andthatf L (x,y,c)isthetopcoefficientoff(x,y,c,d,e;γ)ine,d. 5 Full description of the protocol 1. Verifierchoosesanonzerochallengec F q suchthatc L =1.VerifiersendsctoProver. 2. Proverassignsvalues {w k } = {1,c,c 2,...,c L 1,1}toverticesinthecycle(thatis,firstandlastnodesinthepath arethesame).proverproducescommitmentstovaluesassigned:w k =g wk h ρk.proversends {W k }toverifier. 3. Verifierchoosesachallenge (x c,y c ) F 2 qandsendsittoprover. 4. Proverchoosessome {γ v }atrandomfrom F q.proverproducesexpansioncoefficients {f l }ofthepolynomial overedgesinthecycle P.Proverchoosessome {µ v },{τ l }atrandomfrom F q,producescommitments {D l }and initialcommitments {X v }: Proversends {D l },{X v }toverifier. L z l f l = (z +x c (zw Hj + γ Hj ) +y c (zw Tj + γ Tj )) (14) l=0 H jt j E(P) 5. Verifierchoosesanonzerochallenged F q andsendsittoprover. 6. Prover produces responses D l =g fl h τl l =0...L 1 (15) X v =g γv h µv (16) Θ v =dw v + γ v (17) Ω v =dα v + µ v (18) Proverchoosessome δ e from F q atrandomandproducesexpansioncoefficientsf i. Proverchoosessome λ i,ν e,η e,π e,σ, χfrom F q atrandom,producescommitmentsp i andinitialcommitmentsz e,v e,k: n z i F i = (z + (zu e + δ e )(d +x c Θ Hj +y c Θ Tj )) (19) i=0 e= H jt j E(Γ) P i =g Fi h λi i =0...n 1 (20) Z e =g δe h νe (21) V e =g ηe h πe (22) K =g σ h χ (23) 3 44 of 55

197 Proverproducesexpansioncoefficients {n e },{o e }.Proverchoosessome ξ e,ζ e atrandomfrom F q andproduces commitments {N e },{O e }: n e z +o e = (zu e + δ e )(z(u e 1) + δ e ) (24) N e =g ne h ξe (25) O e =g oe h ζe (26) Proverproducesexpansioncoefficientsm 1,m 0. Proverchoosessome θ 0,θ 1 atrandomfrom F q andproduces commitmentsm 0,M 1 : Proversends Θ v,ω v,p i,z e,v e,k,n e,o e,m 0,M 1. m 1 z +m 0 = z(zb + σ) + (za e + η e )(zu e + δ e ) (27) e E(Γ) 7. Verifierchoosesanonzerochallengee F q andsendsittoprover. 8. Prover produces responses 9. Verifier produces Verifier accepts if M i =g mi h θi i =0,1 (28) Φ e =eu e + δ e e E(Γ) (29) Ψ v =eβ e + ν e (30) Λ e =ea e + η e (31) ϒ e =er e + π e (32) Σ e =eξ e + ζ e (33) =eb + σ (34) Ξ =et + χ (35) R =e E L 1 d l τ l +e E LL 1 e i λ i (36) l=0 i=1 Y =eθ 1 + θ 0 (37) L 1 C = (1 +x c c j +y c c j+1 ) (38) j=0 S = (e + Φ HjT (d +xθ Hj +yθ Tj )) (39) j H jt j E(Γ) Q = e + Λ e Φ e (40) e E(Γ) g Θv h Ωv Wv d =X v (41) g Φe h Ψe Ue e =Z e (42) g Φe(Φe e) h Σe N e eo e =1 (43) g Λe h ϒe A e e =V e (44) g h Ξ B e =K (45) ( ) en g S h Rn 1 (P i ) ei g CdLL 1 (D l ) dl =1 (46) i=0 l=0 g Q h Y M e 1 M 0 =1 (47) 4 45 of 55

198 6 Protocolproperties ItcanbeverifiedhonestVerifieralwaysacceptsforahonestProverandavalidXTSPsolution. Itcanbeshown soundness holds on assumption of hardness of DL. Knowledge extractor algorithm is the same as that for Schnorr protocol. Considersimulatoralgorithmcandidateasfollows.Givenchallenges (c,x c,y c,d,e),verifier: 1. Choosesresponses (Θ v,ω v,φ e,ψ e,σ e,λ e,ϒ e,,ξ,r,y)atrandomfrom F q ; 2. Choosessomegroupelementsfor ({D l },{P n...p 1 },{N e },M 1 )atrandom; 3. Produces C, S, Q according to the protocol, produces X v =g Θv h Ωv W d v (48) Z e =g Φe h Ψe U e e (49) O e =g Φe(Φe e) h Σe N e e (50) V e =g Λe h ϒe A e e (51) K =g h Ξ B e (52) ( ) en P 0 =g S h Rn 1 (P i ) ei g CdLL 1 (D l ) dl (53) i=1 l=0 M 0 =g Q h Y M e 1 (54) Simulated transcript has uniform distribution and is indistinguishable from transcript of a protocol with a Prover. References [1] ManuelBlum.Howtoproveatheoremsonooneelsecanclaimit.InInternationalCongressofMathematicians, pages , [2] Ran Canetti and Marc Fischlin. Universally composable commitments. In CRYPTO, pages 19 40, [3] Vadym Fedyukovych. A signature scheme with approximate key matching(a presentation in russian). In Information Security conference(kiev), [4] Vadym Fedyukovych. Argument of knowledge of a bounded error. In Coding Theory Days(accepted), [5] Vadym Fedyukovych. Protocols for graph languages. In(submitted), [6] Vadym Fedyukovych. Proving polynomial identities(in russian). In Information Security conference(kiev), [7] Vadym Fedyukovych and Vitaliy Sharapov. No disclosure pattern matching.(submitted), [8] Stefan Lucks. How to exploit the intractability of exact tsp for cryptography. In FSE, pages , [9] Stefan Lucks. How traveling salespersons prove their identity. In IMA Conf., pages , [10] Yaron Minsky, Ari Trachtenberg, and Richard Zippel. Set reconciliation with nearly optimal communication complexity. IEEE Transactions on Information Theory, 49(9): , [11] Torben P. Pedersen. Non-interactive and information-theoretic secure verifiable secret sharing. In CRYPTO, pages , [12] Claus-Peter Schnorr. Efficient identification and signatures for smart cards. In CRYPTO, pages , [13] J. T. Schwartz. Fast probabilistic algorithms for verification of polynomial identities. J. ACM, 27(4): , of 55

199 CypherMatrix as cryptographic Data Generator Ernst Erich Schnoor, Munich Abstract The original intention was to search for a new encryption method possibly in excess of today usually used bit manipulations no bits, but with whole characters and numbers (bytes) just as cryptographers have done since centuries especially when using modern technology. The procedure found as a result named by the Author: CypherMatrix method (filed patent DPMA as of ) has little common actions with those methods mostly used today. The CypherMatrix procedure uses simple arithmetik, MODULO calculations and higher number systems. Divergent from actual methods CypherMatrix works with two different sectors: Data generator and Coding area. Both sectors are combined with each other but may be used even separately. The Data generator controls the whole procedure and results in several control parameters necessary for processing combined applications. The CypherMatrix especially is used for: 1. Creating unlimited character series (random number generator), 2. calculating collision free dynamic hash values, 47 of 55

200 3. generating extended digital signatures and 4. all modes of encryptions. The input (pass phrase) of optimal 42 bytes here denoted as Start sequence - initializes and controls the total course of the procedure, but is no initial vector (IV) in a conventional sense. Examples: 7 kangaroos jumping along the Times Square Im Thüringer Wald gibt es 7132 Fliegenpilze Pass phrases should be somewhat funny, but unusual. They have to be easy to remember and it's not necessary to write them down. Because of their length they will not be object of lexical attacks or iterative search techniques. Because of their still somewhat short length the input has to be widened and assigned to a univalent and collision free control value CypherMatrix (hash value) by the following four steps: 1. Calculating of a position weightened and collision free interim result H(k), 2. expanding to a hashfunktion series (HFR) and another partial hash value H(p): Expansion, 3. contracting hashfunktion series to a (BASIC VARIATION) (array of 16x16 elements) Contraction and 4. threefold permutations to perform the CypherMatrix (16x16 characters) as final result and a definite mapping of the input. All control parameters necessary to perform cryptographic applications are extracted out of the CypherMatrix, especially the Matrix key, a series of 42 bytes, which are to be led back (loop) to the beginning of the cycle in order to initialize the next round. In addition for encryptions purposes block keys and a cipher alphabet of 128 characters are extracted from the CypherMatrix. The four steps outlined above generate a new CypherMatrix repeated in each cycle as long as the cryptographic task is performed or the procedure is terminated. Start sequence as an input of length (n) comprises a series of definite signs a(i): a(1) a(2) a(3)... a(i)... a(n) To assign the start sequence univalent with a value H(k) each byte a(i) is denoted with an index value (ASCII character set) and the bytes have to be linked together in a qualified manner. Linking together by addition results as follows: n H(k) = ( a(i) + 1 ) i = 1 (Note: Single values for a(i) are increased by (+1) because otherwise ASCII-zero (0) would not be considered)) 48 of 55

201 But the thus calculated value is far away to serve as a hash value. In order to obtain a unmistakable result some more features have to be added. With respect to Renè Descartes every fact can be exactly determined by coordinates for subject, location and time (cartesian coordinate system). Subject is the single character a(i) and as location we state position p(i) inside the sequence, but time is not relevant, here (t(i)=1): Each character a(i) is position weighted by multiplying with its location p(i): n H(k) = ( a(i) + 1 ) * p(i) * t(i) i = 1 t(i) = 1 Occurrences of collisions are excluded by solving with the hash constant C(k) which depends on length of start sequence (n) and an individual fixed user code (1 up to 99). C(k) = n * (n 2) + code n = sqrt (C(k) + 1 code) + 1 Including hash constant C(k) a position weighted partial hash value H(k) is calculated as follows: n H(k) = ( a(i) + 1 ) * (C(k) + p(i)) i = 1 The calculated value H(k) excludes collisions but is still to narrow to establish a secure unchallengeable hash result, To obtain more security an expansion function is introduced which widens the determining factors to a voluminous scale without loosing their quality of being collision free. The decimal values a(i) are converted into digits of higher number systems here to base 77 or base 85 - the results are added to another partial hash value H(p) and accumulated to a hashfunction series (HFR) of about 160 to 2400 digits. n H(p) = ( ( a(i) + 1 ) * p(i) * H(k) ) + ( p(i) + code ) i = 1 This expansion constitutes a first one-way function of the procedure. There is no way back to the initial start sequence. In order to reduce the digits of hashfunction series to decimals again the figures are reconverted by MODULO 256 to decimal numbers (0 to 255) and stored in an array of 16x16 elements: BASIC VARIATION. To achieve the reverse convertion the hashfunktion series is assumed to be digits in number system on expansion base + 1 (base 78 or base 86). This contraction is the second one-way-function 49 of 55

202 of the procedure. The function is not reversible and retrograde destination of the hashfunction series is not achievable. From BASIC VARIATION the procedure generates in each round and three loops (permutations) the CypherMatrix with 16x16 elements. CypherMatrix is a definite mapping of the start sequence. A new CypherMatrix is created in each round. Due to principles of probability a repetition of an identical distributuion of matrix elements will occur once in 256! (faculty) = 8E+506 cases. Working together of generator and coding area in order to achieve encryption and decryption are demonstrated by the following scheme: In symmetric mode sender and recipient insert an identical start sequence which controls the total procedure on both sites. In each round identical matrixes and control parameters are generated. An attacker would be able to enter the procedure only if he finds the start sequence (42 bytes and not written down anywhere). Key words Data-generator, start sequence, one-way-hashfuncion, expansion, contraction, position-weighted, hash-function-series, hash-constant, BASIC VARIATION, CypherMatrix, matrix key Munich, July 1, of 55

203 Templateless Biometric-Enforced Non-Transferability of Anonymous Credentials 1 Introduction Sebastian Pape Databases and Interactive Systems Research Group University of Kassel Wilhelmshöher Allee 73, Kassel pape@db.informatik.uni-kassel.de Most cryptographic primitives for authentification schemes in the digital world are based on the knowledge of a private key or secret. For example digital signatures or zero-knowledge proofs. In many cases there is (at least) an implicit binding of the secret to a person. If you receive a signed mail, you assume it is signed by the regular owner of the private key; if you authenticate yourself with a zero-knowledge proof you are expected not to give the secret to other persons. One may not put too many trust into this assumption since this secrects are eventually digital data which can be copied without evidence. Two obvious situations cross one s mind: On one hand cryptographic secrets are not very memorable for human beings in general, so they are usually stored somewhere they could be stolen. On the other hand the user may want to share his secret while the authorizing organisation does not want him to do so. While the first situation could be released by storing the key at a safe place (e.g. a tamper-proof smartcard or encrypted with a human memorable passphrase it is much harder to achieve the same in the latter situation. If you want to ensure the non-transferability of knowledge you must keep it secret from the user or make him want to keep it secret. In the following work we especially focus on the non-transferability of anonymous credentials since they offer less contact points than e.g. digital signatures, where the user is known and could be sued for abuse. Nevertheless the following approach can also be used to ensure non-transferability of non-anonymous authentification. Anonymous credentials [CE87] introduced by Chaum [Cha85] usually consist of cryptographic tokens which allow the user (herein after referred to as prover) to prove a statement or relationship with an organisation to another person or organisation (herein after referred to as verifier) without being identified. While some anonymous credential systems are related to the concept of untraceable or anonymous payments [Cha83] and hence it should be possible to transfer them easily to another person there are some situations where credentials should not be transferable. E.g. if the prover wants to show the possession of a valid driving licence the verifier probably does not want to see a transfered driving licence which would rather prove the statement I know someone who has a valid driving licence. Other examples of anonymous credentials are age verification, the proof of a country s citizenship and weekly, monthly or yearly tickets for train rides or baths. 2 Different Approaches of Ensuring Non-Transferability As abovementioned there are two general principals to ensure the non-transferability of tickets. One approach tries to make the secret more valuable for the prover making it unpleasant for him to share the credential. The other approach is of technical nature and tries to prevent the prover from sharing with biometrics of 55

204 2.1 Embedding Valuables There are already approaches to prevent the transfer of credentials where sharing a credential implies also sharing a valuable secret outside a system [DLN97, GPR98, LRSW00] or even all of the prover s secrets inside the system [CL01]. Nevertheless this protection naturally will not prevent all users from sharing credentials. Be it they share their credentials incautious, be it they really trust someone else. In addition those valuable secrets raise the system s value. So users have to take care of thiefs and have to immanently trust the system s architecture. As a first conclusion we notice that the system s effectiveness fundamentally depends on the embedded value. 2.2 Biometrically Enforcing Non-Transferability Another possibility to make sure the credentials are only used by the person the credential was created for, is to make use of the person s biometric information. Using biometrics however usually causes privacy concerns, especially since in contrast to passwords or tokens you cannot change biometric attributes. Therefore extraordinary care has to be taken to protect the user s data [BBB + 08]. It can be easily seen that allowing the verifier to check the prover s biometric attributes conflicts with the prover s wish of anonymity. In 1998 Bleumer [Ble98] combined anonymous credentials with biometric authentification making use of a variant of the wallet-with-observerarchitecture introduced by Chaum and Pedersen [CP93]. In the wallet-with-observer-architecture there exists a user trusted device (wallet) which runs a local process (observer). The credential issuing organization (herein after referred to as authority) trusts that the observer only performs legitimate operations. Impagliazzo and Miner More [IM03] transfered that design to a personal digital assistant (PDA) with a tamper-resistant smartcard. The smartcard (observer) is issued and trusted by the authority and it s tamper-resistance makes sure that the user cannot read and tamper with its content. In contrast the PDA (warden) protects the prover s interests and makes sure the smartcard does not diverge from the specified protocol. Both approaches have in common that biometric authentification is not part of the underlying credential system, but instead prerequisite for the credential protocol to start. 3 Templateless Biometric-Enforced Non-Transferability Analogous to [IM03] we prefer a setup where each user has a PDA and a smartcard handed out by the credential issuing authority. Following [BBB + 08] renders it obvious that the biometric device has to be connected straightly to the smartcard. The only devices which fulfill our needs today are fingerprintreaders. Either they are embedded into the smartcard [Bio, fid08] or they work the way cash-card-terminals act when asking for the users personal identification number (PIN). In this case the reader s input is directly sent to the card, the smartcard decides about acception or rejection (called match-on-card [Nor04, NH04]) and no attached computer is able to eavesdrop on it. Note that relying on fingerprints is only a compromise solution. On the one hand some people do not have suitable fingerprints, forging fingerprints has been done with acceptable effort and there are several attacks on biometric systems based on fingerprints [UJ04]. On the other hand today fingerprint readers are the only devices which could be embedded into smartcards. Following the parameters and comparison of [Jai04] the most desirable attributes for us would be to have a low Circumvention (the ease of use of substitutes is hard) and high Universality (as many people as 2 52 of 55

205 possible have those attributes), Uniqueness (people can be well seperated) and Permanence (the attributes resist aging well). That would lead us to the use of DNA-recognition which can t be done on a smartcard. Since we do not rely on special attributes of fingerprints it seems suitable to us to base our contribution on fingerprints and should there exists an on-the-fly DNA-recognition (or any other suitable biometric identification method) some day which could be embedded into smartcards, the system could easily be switched. Due to the template s compact nature it was commonly assumed that it is infeasible to extract the complete biometric information from them. Since there are some attacks [JRU05] we do not want to store templates on the card. Instead of that we rely on fuzzy extractors proposed by Dodis et al. [DORS08]. Fuzzy extractors provide the same output, even if the input changes, but remains reasonably close to the original. Since Dodis et al. also claim that their fuzzy extractor s output is nearly distributed uniformly and information-theoretically secure, we translate each user s fingerprint to a unique identifier with it. Let us assume the underlying anonymous credential protocol is based on the Fiat-Shamir identification protocol [FFS87] or simular to the protocol proposed in [IM03]. Both are Zero-Knowledge- Protocols (ZKP) and have in common, that the prover possesses secret information which he proves to a verifier without revealing to him what that secret information is. If the prover should not be able to share this information, it has to be kept in a tamper-prove device, e.g. a smartcard. Since we want to restrict the use of such a smartcard to a specific person, the secret information is not stored straightly on the smartcard but instead connected to the output of the fuzzy extractor introduced in the previous paragraph. This can be done by a simple modulo addition. 3.1 Creating and Showing Credentials Depending on the underlying ZKP, public parameters and the secret information are determined. The authority stores the secret information s on the card an sets it to initialisation state. After reading the user s fingerprint fp u, the card computes the value of the fuzzy extractor fe, stores s := s fe(fp u ) and deletes any occurence of s. Now the card is personalized to a specific user, as the secret information s can only be restored with the value of the fuzzy extrator. To prevent a later change of s the card has to be switched to proving state at this point. Proving the possession of a credential is also only slightly changed. The user s fingerprint is read and the following calculations/proofs are done with s + fe(fp u ). It can be easily seen, that this operation restores s. Therefore the underlying ZKP does not need to be changed. We illustrate that by means of an example in the next subsection. 3.2 Example: Modified Feige-Fiat-Shamir Identification Protocol Setup: The authority chooses two large prime integers p and q and calculates their product n = pq. n is then stored on the smartcard and given to the verifier and the prover, p and q are kept secret. Next the authority generates secret numbers s 1,..., s k with gcd(s i, n) = 1 and computes v i s 2 i (mod n). The verifier and prover receive the numbers v i while the numbers s i are stored on the card. When the card is initialised to the prover his fingerprint fp u is read from the card and the stored s i are overwritten by s i s i fe(fp u ) (mod n). The card is set to proving state then of 55

206 Proving: The smartcard chooses a random integer r, a random sign σ { 1, 1}, computes σx r 2 (mod n) and sends this number to the verifier. The verifier chooses numbers a i {0, 1} and sends them to the card. The prover now has to give his fingerprint to the smartcard and the card computes y r(s 1 + fe(fp u)) a 1 (s 2 + fe(fp u)) a2 (s k + fe(fp u)) a k (mod n). y is sent to the verifier and he checks if y 2 ± xv a 1 1 va 2 2 va k k (mod n) to decide if the prover has passed authorisation. Notice that the user is able to follow the procedure since he is provided with n and v i and listens to the communication of the card and the verifier to make sure the card follows the protocol. 4 Conclusion Our contribution proposes a general procedure to ensure the non-transferability of credentials without storing a template anywhere. It achieves the same goal as [IM03] to establish an anonymous credential system that makes use of biometrics to obtain non-transferability. Since our approach is very similar, there are almost the same problems and restrictions that apply to [IM03]. To call them by name: The security of the system depends very much on the tamper-resistance of the smartcard. If a prover begins to interact with a verifier he needs to be kept isolated from the outside world. This is necessary, because otherwise the verifier cannot be sure which card he is communicating with. E.g. if the prover has radio contact to another card, non-transferability suffers and it is possible to share cards. However there are some improvements compared to [IM03]. Our approach abandons the storage of the user s fingerprint. Although it is not easy to tamper a smartcard, there may be sidechannelattacks like timing attacks [Koc96] or differential power analysis [KJJ99] attacking the concrete implementation of a card. This especially affects match-on-card systems where the fingerprint is read by an external reader and submitted to the card. This way it is a lot easier to create test data and it may be possible to extract information about the stored template on the card if the card is lost. Also if the stored secret is revealed it is useless without the information about the fuzzy extractor s value of the user s fingerprint (and vice versa), which might be usefull if the card is lost. Another advantage is that the biometric information is stronger embedded into the anonymous credential system and not only a prerequisite. A consequence of this is that it is easier to create a biometric authentification failed -awareness for the verifier since he will notice rather the fail of a ZKP than a failed biometric authentification on the card itself. References [BBB + 08] H. Biermann, M. Bromba, C. Busch, G. Hornung, M. Meints, and G. Quiring-Kock. White Paper zum Datenschutz in der Biometrie. Technical report, TELETRUST Deutschland e.v., Arbeitsgruppe Biometrie, [Bio] [Ble98] [CE87] Biometric Associates, Inc. The BAI Authenticator Smart Card Datasheet. Technical report, from Gerrit Bleumer. Biometric yet privacy protecting person authentication. Lecture Notes in Computer Science, 1525:99 110, David Chaum and Jan-Hendrik Evertse. A secure and privacy-protecting protocol for transmitting personal information between organizations. In Proceedings on Advances in cryptology CRYPTO 86, pages , London, UK, Springer Verlag of 55

207 [Cha83] David Chaum. Blind signatures for untraceable payments. In D. Chaum, R.L. Rivest, and A.T. Sherman, editors, Advances in Cryptology Proceedings of Crypto 82, pages , [Cha85] David Chaum. Security without identification: transaction systems to make big brother obsolete. Communications of the ACM, 28(10): , [CL01] [CP93] Jan Camenisch and Anna Lysyanskaya. An efficient system for non-transferable anonymous credentials with optional anonymity revocation. In EUROCRYPT, volume 2045 of Lecture Notes in Computer Science, pages , David Chaum and Torben P. Pedersen. Wallet databases with observers. In CRYPTO 92: Proceedings of the 12th Annual International Cryptology Conference on Advances in Cryptology, volume 740, pages , London, UK, Springer Verlag. [DLN97] Cynthia Dwork, Jeff Lotspiech, and Moni Naor. Digital signets: Self-enforcing protection of digital information. In Proc. 28th Ann. ACM Symp. on Theory of Computing, [DORS08] Yevgeniy Dodis, Rafail Ostrovsky, Leonid Reyzin, and Adam Smith. Fuzzy extractors: How to generate strong keys from biometrcis and other noisy data. Technical report, [FFS87] U. Feige, A. Fiat, and A. Shamir. Zero knowledge proofs of identity. In STOC 87: Proceedings of the nineteenth annual ACM conference on Theory of computing, pages , New York, NY, USA, ACM. [fid08] Homepage of Fidelica Microsystems, Inc., June [GPR98] Oded Goldreich, Birgit Pfitzmann, and Ronald L. Rivest. Self-delegation with controlled propagation or what if you lose your laptop. In Advances in Cryptology - CRYPTO 98, volume 1462 of Lecture Notes in Computer Science, pages , [IM03] Russell Impagliazzo and Sara Miner More. Anonymous credentials with biometrically-enforced non-transferability. In Proceedings of the 2003 ACM Workshop on Privacy in the Electronic Society (WPES 03), pages ACM, [Jai04] A. K. Jain. Biometric recognition: how do i know who you are? In Proceedings of the IEEE 12th Signal Processing and Communications Applications Conference, pages 3 5, April [JRU05] A. K. Jain, A. Ross, and U. Uludag. Biometric template security: Challenges and solutions. In Proc. of 13th European Signal Processing Conference (EUSIPCO), Antalya, Turkey, September [KJJ99] Paul Kocher, Joshua Jaffe, and Benjamin Jun. Differential power analysis. Lecture Notes in Computer Science, 1666: , [Koc96] Paul C. Kocher. Timing attacks on implementations of Diffie-Hellman, RSA, DSS, and other systems. Lecture Notes in Computer Science, 1109: , [LRSW00] Anna Lysyanskaya, Ronald L. Rivest, Amit Sahai, and Stefan Wolf. Pseudonym systems. In Selected Areas in Cryptography, 6th Annual International Workshop, SAC 99, volume 1758 of Lecture Notes in Computer Science. Springer, [NH04] Jonas Nilsson and Michael Harris. Match-on-card for java cards white paper. Technical report, Precise Biometrics, April [Nor04] Björn Nordin. Match-on-card technology white paper. Technical report, Precise Biometrics, [UJ04] U. Uludag and A.K. Jain. Attacks on biometric systems: a case study in fingerprints. In Proc. SPIE-EI 2004, Security, Seganography and Watermarking of Multimedia Contents VI, pages , of 55

208 Tagungsband zum 9. Kryptotag Workshop der Fachgruppe Kryptographie in der Gesellschaft für Informatik Institut für Internet-Sicherheit Fachhochschule Gelsenkirchen Fachbereich Informatik 10. November 2008

2. Krypto-Tag Workshop über Kryptographie Universität Ulm

2. Krypto-Tag Workshop über Kryptographie Universität Ulm CURANDO 2. Krypto-Tag Workshop über Kryptographie Universität Ulm Wolfgang Lindner (Hg.) Universität Ulm Christopher Wolf (Hg.) K.U.Leuven 31. März 2005 UNIVERSITÄT ULM DOCENDO SCIENDO Ulmer Informatik-Berichte

Mehr

Titelbild1 ANSYS. Customer Portal LogIn

Titelbild1 ANSYS. Customer Portal LogIn Titelbild1 ANSYS Customer Portal LogIn 1 Neuanmeldung Neuanmeldung: Bitte Not yet a member anklicken Adressen-Check Adressdaten eintragen Customer No. ist hier bereits erforderlich HERE - Button Hier nochmal

Mehr

Efficient Design Space Exploration for Embedded Systems

Efficient Design Space Exploration for Embedded Systems Diss. ETH No. 16589 Efficient Design Space Exploration for Embedded Systems A dissertation submitted to the SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH for the degree of Doctor of Sciences presented by

Mehr

Englisch-Grundwortschatz

Englisch-Grundwortschatz Englisch-Grundwortschatz Die 100 am häufigsten verwendeten Wörter also auch so so in in even sogar on an / bei / in like wie / mögen their with but first only and time find you get more its those because

Mehr

BISON Instantiating the Whitened Swap-Or-Not Construction September 6th, 2018

BISON Instantiating the Whitened Swap-Or-Not Construction September 6th, 2018 BION Instantiating the Whitened wap-or-not Construction eptember 6th, 2018 Horst Görtz Institut für IT icherheit Ruhr-Universität Bochum Virginie Lallemand, Gregor Leander, Patrick Neumann, and Friedrich

Mehr

prorm Budget Planning promx GmbH Nordring Nuremberg

prorm Budget Planning promx GmbH Nordring Nuremberg prorm Budget Planning Budget Planning Business promx GmbH Nordring 100 909 Nuremberg E-Mail: support@promx.net Content WHAT IS THE prorm BUDGET PLANNING? prorm Budget Planning Overview THE ADVANTAGES OF

Mehr

ELBA2 ILIAS TOOLS AS SINGLE APPLICATIONS

ELBA2 ILIAS TOOLS AS SINGLE APPLICATIONS ELBA2 ILIAS TOOLS AS SINGLE APPLICATIONS An AAA/Switch cooperative project run by LET, ETH Zurich, and ilub, University of Bern Martin Studer, ilub, University of Bern Julia Kehl, LET, ETH Zurich 1 Contents

Mehr

Ein Stern in dunkler Nacht Die schoensten Weihnachtsgeschichten. Click here if your download doesn"t start automatically

Ein Stern in dunkler Nacht Die schoensten Weihnachtsgeschichten. Click here if your download doesnt start automatically Ein Stern in dunkler Nacht Die schoensten Weihnachtsgeschichten Click here if your download doesn"t start automatically Ein Stern in dunkler Nacht Die schoensten Weihnachtsgeschichten Ein Stern in dunkler

Mehr

FACHKUNDE FüR KAUFLEUTE IM GESUNDHEITSWESEN FROM THIEME GEORG VERLAG

FACHKUNDE FüR KAUFLEUTE IM GESUNDHEITSWESEN FROM THIEME GEORG VERLAG FACHKUNDE FüR KAUFLEUTE IM GESUNDHEITSWESEN FROM THIEME GEORG VERLAG DOWNLOAD EBOOK : FACHKUNDE FüR KAUFLEUTE IM GESUNDHEITSWESEN Click link bellow and free register to download ebook: FACHKUNDE FüR KAUFLEUTE

Mehr

Darstellung und Anwendung der Assessmentergebnisse

Darstellung und Anwendung der Assessmentergebnisse Process flow Remarks Role Documents, data, tool input, output Important: Involve as many PZU as possible PZO Start Use appropriate templates for the process documentation Define purpose and scope Define

Mehr

EEX Kundeninformation 2007-09-05

EEX Kundeninformation 2007-09-05 EEX Eurex Release 10.0: Dokumentation Windows Server 2003 auf Workstations; Windows Server 2003 Service Pack 2: Information bezüglich Support Sehr geehrte Handelsteilnehmer, Im Rahmen von Eurex Release

Mehr

CNC ZUR STEUERUNG VON WERKZEUGMASCHINEN (GERMAN EDITION) BY TIM ROHR

CNC ZUR STEUERUNG VON WERKZEUGMASCHINEN (GERMAN EDITION) BY TIM ROHR (GERMAN EDITION) BY TIM ROHR READ ONLINE AND DOWNLOAD EBOOK : CNC ZUR STEUERUNG VON WERKZEUGMASCHINEN (GERMAN EDITION) BY TIM ROHR PDF Click button to download this ebook READ ONLINE AND DOWNLOAD CNC ZUR

Mehr

Algorithms & Datastructures Midterm Test 1

Algorithms & Datastructures Midterm Test 1 Algorithms & Datastructures Midterm Test 1 Wolfgang Pausch Heiko Studt René Thiemann Tomas Vitvar

Mehr

Funktion der Mindestreserve im Bezug auf die Schlüsselzinssätze der EZB (German Edition)

Funktion der Mindestreserve im Bezug auf die Schlüsselzinssätze der EZB (German Edition) Funktion der Mindestreserve im Bezug auf die Schlüsselzinssätze der EZB (German Edition) Philipp Heckele Click here if your download doesn"t start automatically Download and Read Free Online Funktion

Mehr

Corporate Digital Learning, How to Get It Right. Learning Café

Corporate Digital Learning, How to Get It Right. Learning Café 0 Corporate Digital Learning, How to Get It Right Learning Café Online Educa Berlin, 3 December 2015 Key Questions 1 1. 1. What is the unique proposition of digital learning? 2. 2. What is the right digital

Mehr

PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB

PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB Read Online and Download Ebook PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB DOWNLOAD EBOOK : PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: Click link bellow

Mehr

WP2. Communication and Dissemination. Wirtschafts- und Wissenschaftsförderung im Freistaat Thüringen

WP2. Communication and Dissemination. Wirtschafts- und Wissenschaftsförderung im Freistaat Thüringen WP2 Communication and Dissemination Europa Programm Center Im Freistaat Thüringen In Trägerschaft des TIAW e. V. 1 GOALS for WP2: Knowledge information about CHAMPIONS and its content Direct communication

Mehr

DAS ERSTE MAL UND IMMER WIEDER. ERWEITERTE SONDERAUSGABE BY LISA MOOS

DAS ERSTE MAL UND IMMER WIEDER. ERWEITERTE SONDERAUSGABE BY LISA MOOS Read Online and Download Ebook DAS ERSTE MAL UND IMMER WIEDER. ERWEITERTE SONDERAUSGABE BY LISA MOOS DOWNLOAD EBOOK : DAS ERSTE MAL UND IMMER WIEDER. ERWEITERTE Click link bellow and free register to download

Mehr

Group signature schemes and payment systems based on the discrete logarithm problem

Group signature schemes and payment systems based on the discrete logarithm problem Research Collection Doctoral Thesis Group signature schemes and payment systems based on the discrete logarithm problem Author(s): Camenisch, Jan Publication Date: 1998 Permanent Link: https://doi.org/10.3929/ethz-a-001923735

Mehr

HIR Method & Tools for Fit Gap analysis

HIR Method & Tools for Fit Gap analysis HIR Method & Tools for Fit Gap analysis Based on a Powermax APML example 1 Base for all: The Processes HIR-Method for Template Checks, Fit Gap-Analysis, Change-, Quality- & Risk- Management etc. Main processes

Mehr

Contents. Interaction Flow / Process Flow. Structure Maps. Reference Zone. Wireframes / Mock-Up

Contents. Interaction Flow / Process Flow. Structure Maps. Reference Zone. Wireframes / Mock-Up Contents 5d 5e 5f 5g Interaction Flow / Process Flow Structure Maps Reference Zone Wireframes / Mock-Up 5d Interaction Flow (Frontend, sichtbar) / Process Flow (Backend, nicht sichtbar) Flow Chart: A Flowchart

Mehr

Security Patterns. Benny Clauss. Sicherheit in der Softwareentwicklung WS 07/08

Security Patterns. Benny Clauss. Sicherheit in der Softwareentwicklung WS 07/08 Security Patterns Benny Clauss Sicherheit in der Softwareentwicklung WS 07/08 Gliederung Pattern Was ist das? Warum Security Pattern? Security Pattern Aufbau Security Pattern Alternative Beispiel Patternsysteme

Mehr

Lehrstuhl für Allgemeine BWL Strategisches und Internationales Management Prof. Dr. Mike Geppert Carl-Zeiß-Str. 3 07743 Jena

Lehrstuhl für Allgemeine BWL Strategisches und Internationales Management Prof. Dr. Mike Geppert Carl-Zeiß-Str. 3 07743 Jena Lehrstuhl für Allgemeine BWL Strategisches und Internationales Management Prof. Dr. Mike Geppert Carl-Zeiß-Str. 3 07743 Jena http://www.im.uni-jena.de Contents I. Learning Objectives II. III. IV. Recap

Mehr

Aufbau eines IT-Servicekataloges am Fallbeispiel einer Schweizer Bank

Aufbau eines IT-Servicekataloges am Fallbeispiel einer Schweizer Bank SwissICT 2011 am Fallbeispiel einer Schweizer Bank Fritz Kleiner, fritz.kleiner@futureways.ch future ways Agenda Begriffsklärung Funktionen und Aspekte eines IT-Servicekataloges Fallbeispiel eines IT-Servicekataloges

Mehr

Newest Generation of the BS2 Corrosion/Warning and Measurement System

Newest Generation of the BS2 Corrosion/Warning and Measurement System Newest Generation of the BS2 Corrosion/Warning and Measurement System BS2 System Description: BS2 CorroDec 2G is a cable and energyless system module range for detecting corrosion, humidity and prevailing

Mehr

miditech 4merge 4-fach MIDI Merger mit :

miditech 4merge 4-fach MIDI Merger mit : miditech 4merge 4-fach MIDI Merger mit : 4 x MIDI Input Port, 4 LEDs für MIDI In Signale 1 x MIDI Output Port MIDI USB Port, auch für USB Power Adapter Power LED und LOGO LEDs Hochwertiges Aluminium Gehäuse

Mehr

arlanis Software AG SOA Architektonische und technische Grundlagen Andreas Holubek

arlanis Software AG SOA Architektonische und technische Grundlagen Andreas Holubek arlanis Software AG SOA Architektonische und technische Grundlagen Andreas Holubek Speaker Andreas Holubek VP Engineering andreas.holubek@arlanis.com arlanis Software AG, D-14467 Potsdam 2009, arlanis

Mehr

Magic Figures. We note that in the example magic square the numbers 1 9 are used. All three rows (columns) have equal sum, called the magic number.

Magic Figures. We note that in the example magic square the numbers 1 9 are used. All three rows (columns) have equal sum, called the magic number. Magic Figures Introduction: This lesson builds on ideas from Magic Squares. Students are introduced to a wider collection of Magic Figures and consider constraints on the Magic Number associated with such

Mehr

Customer-specific software for autonomous driving and driver assistance (ADAS)

Customer-specific software for autonomous driving and driver assistance (ADAS) This press release is approved for publication. Press Release Chemnitz, February 6 th, 2014 Customer-specific software for autonomous driving and driver assistance (ADAS) With the new product line Baselabs

Mehr

Kryptographie II. Introduction to Modern Cryptography. Jonathan Katz & Yehuda Lindell

Kryptographie II. Introduction to Modern Cryptography. Jonathan Katz & Yehuda Lindell Kryptographie II Introduction to Modern Cryptography Jonathan Katz & Yehuda Lindell Universität zu Köln, WS 13/14 Medienkulturwissenschaft / Medieninformatik AM2: Humanities Computer Science Aktuelle Probleme

Mehr

ISO 15504 Reference Model

ISO 15504 Reference Model Process flow Remarks Role Documents, data, tools input, output Start Define purpose and scope Define process overview Define process details Define roles no Define metrics Pre-review Review yes Release

Mehr

EVANGELISCHES GESANGBUCH: AUSGABE FUR DIE EVANGELISCH-LUTHERISCHE LANDESKIRCHE SACHSEN. BLAU (GERMAN EDITION) FROM EVANGELISCHE VERLAGSAN

EVANGELISCHES GESANGBUCH: AUSGABE FUR DIE EVANGELISCH-LUTHERISCHE LANDESKIRCHE SACHSEN. BLAU (GERMAN EDITION) FROM EVANGELISCHE VERLAGSAN EVANGELISCHES GESANGBUCH: AUSGABE FUR DIE EVANGELISCH-LUTHERISCHE LANDESKIRCHE SACHSEN. BLAU (GERMAN EDITION) FROM EVANGELISCHE VERLAGSAN DOWNLOAD EBOOK : EVANGELISCHES GESANGBUCH: AUSGABE FUR DIE EVANGELISCH-LUTHERISCHE

Mehr

Franke & Bornberg award AachenMünchener private annuity insurance schemes top grades

Franke & Bornberg award AachenMünchener private annuity insurance schemes top grades Franke & Bornberg award private annuity insurance schemes top grades Press Release, December 22, 2009 WUNSCHPOLICE STRATEGIE No. 1 gets best possible grade FFF ( Excellent ) WUNSCHPOLICE conventional annuity

Mehr

Die Bedeutung neurowissenschaftlicher Erkenntnisse für die Werbung (German Edition)

Die Bedeutung neurowissenschaftlicher Erkenntnisse für die Werbung (German Edition) Die Bedeutung neurowissenschaftlicher Erkenntnisse für die Werbung (German Edition) Lisa Johann Click here if your download doesn"t start automatically Download and Read Free Online Die Bedeutung neurowissenschaftlicher

Mehr

Was heißt Kryptographie I? Understanding Cryptography Christof Paar und Jan Pelzl

Was heißt Kryptographie I? Understanding Cryptography Christof Paar und Jan Pelzl Was heißt Kryptographie I? Understanding Cryptography Christof Paar und Jan Pelzl Die Autoren Dr.-Ing. Jan Pelzl Prof. Dr.-Ing. Christof Paar Gliederung Historischer Überblick Begrifflichkeiten Symmetrische

Mehr

Mitglied der Leibniz-Gemeinschaft

Mitglied der Leibniz-Gemeinschaft Methods of research into dictionary use: online questionnaires Annette Klosa (Institut für Deutsche Sprache, Mannheim) 5. Arbeitstreffen Netzwerk Internetlexikografie, Leiden, 25./26. März 2013 Content

Mehr

Wie man heute die Liebe fürs Leben findet

Wie man heute die Liebe fürs Leben findet Wie man heute die Liebe fürs Leben findet Sherrie Schneider Ellen Fein Click here if your download doesn"t start automatically Wie man heute die Liebe fürs Leben findet Sherrie Schneider Ellen Fein Wie

Mehr

Die Kunst des Programmierens...

Die Kunst des Programmierens... Die Kunst des Programmierens... Wo die Kosten anfallen Der Mythos Wiederverwendung: Design für Wartung als eigentliches Ziel, Objekt Spektrum 4/2009 software maintainers sped 45 percent of their time seeking

Mehr

LEBEN OHNE REUE: 52 IMPULSE, DIE UNS DARAN ERINNERN, WAS WIRKLICH WICHTIG IST (GERMAN EDITION) BY BRONNIE WARE

LEBEN OHNE REUE: 52 IMPULSE, DIE UNS DARAN ERINNERN, WAS WIRKLICH WICHTIG IST (GERMAN EDITION) BY BRONNIE WARE LEBEN OHNE REUE: 52 IMPULSE, DIE UNS DARAN ERINNERN, WAS WIRKLICH WICHTIG IST (GERMAN EDITION) BY BRONNIE WARE DOWNLOAD EBOOK : LEBEN OHNE REUE: 52 IMPULSE, DIE UNS DARAN EDITION) BY BRONNIE WARE PDF Click

Mehr

Ressourcenmanagement in Netzwerken SS06 Vorl. 12,

Ressourcenmanagement in Netzwerken SS06 Vorl. 12, Ressourcenmanagement in Netzwerken SS06 Vorl. 12, 30.6.06 Friedhelm Meyer auf der Heide Name hinzufügen 1 Prüfungstermine Dienstag, 18.7. Montag, 21. 8. und Freitag, 22.9. Bitte melden sie sich bis zum

Mehr

VGM. VGM information. HAMBURG SÜD VGM WEB PORTAL USER GUIDE June 2016

VGM. VGM information. HAMBURG SÜD VGM WEB PORTAL USER GUIDE June 2016 Overview The Hamburg Süd VGM Web portal is an application that enables you to submit VGM information directly to Hamburg Süd via our e-portal Web page. You can choose to enter VGM information directly,

Mehr

WAS IST DER KOMPARATIV: = The comparative

WAS IST DER KOMPARATIV: = The comparative DER KOMPATATIV VON ADJEKTIVEN UND ADVERBEN WAS IST DER KOMPARATIV: = The comparative Der Komparativ vergleicht zwei Sachen (durch ein Adjektiv oder ein Adverb) The comparative is exactly what it sounds

Mehr

Bayerisches Landesamt für Statistik und Datenverarbeitung Rechenzentrum Süd. z/os Requirements 95. z/os Guide in Lahnstein 13.

Bayerisches Landesamt für Statistik und Datenverarbeitung Rechenzentrum Süd. z/os Requirements 95. z/os Guide in Lahnstein 13. z/os Requirements 95. z/os Guide in Lahnstein 13. März 2009 0 1) LOGROTATE in z/os USS 2) KERBEROS (KRB5) in DFS/SMB 3) GSE Requirements System 1 Requirement Details Description Benefit Time Limit Impact

Mehr

Big Data Analytics. Fifth Munich Data Protection Day, March 23, Dr. Stefan Krätschmer, Data Privacy Officer, Europe, IBM

Big Data Analytics. Fifth Munich Data Protection Day, March 23, Dr. Stefan Krätschmer, Data Privacy Officer, Europe, IBM Big Data Analytics Fifth Munich Data Protection Day, March 23, 2017 C Dr. Stefan Krätschmer, Data Privacy Officer, Europe, IBM Big Data Use Cases Customer focused - Targeted advertising / banners - Analysis

Mehr

Support Technologies based on Bi-Modal Network Analysis. H. Ulrich Hoppe. Virtuelles Arbeiten und Lernen in projektartigen Netzwerken

Support Technologies based on Bi-Modal Network Analysis. H. Ulrich Hoppe. Virtuelles Arbeiten und Lernen in projektartigen Netzwerken Support Technologies based on Bi-Modal Network Analysis H. Agenda 1. Network analysis short introduction 2. Supporting the development of virtual organizations 3. Supporting the development of compentences

Mehr

Symbio system requirements. Version 5.1

Symbio system requirements. Version 5.1 Symbio system requirements Version 5.1 From: January 2016 2016 Ploetz + Zeller GmbH Symbio system requirements 2 Content 1 Symbio Web... 3 1.1 Overview... 3 1.1.1 Single server installation... 3 1.1.2

Mehr

Risiko Datensicherheit End-to-End-Verschlüsselung von Anwendungsdaten. Peter Kirchner Microsoft Deutschland GmbH

Risiko Datensicherheit End-to-End-Verschlüsselung von Anwendungsdaten. Peter Kirchner Microsoft Deutschland GmbH Risiko Datensicherheit End-to-End-Verschlüsselung von Anwendungsdaten Peter Kirchner Microsoft Deutschland GmbH RISIKO Datensicherheit NSBNKPDA kennt alle ihre Geheimnisse! Unterschleißheim Jüngste Studien

Mehr

SARA 1. Project Meeting

SARA 1. Project Meeting SARA 1. Project Meeting Energy Concepts, BMS and Monitoring Integration of Simulation Assisted Control Systems for Innovative Energy Devices Prof. Dr. Ursula Eicker Dr. Jürgen Schumacher Dirk Pietruschka,

Mehr

Hardwarekonfiguration an einer Siemens S7-300er Steuerung vornehmen (Unterweisung Elektriker / - in) (German Edition)

Hardwarekonfiguration an einer Siemens S7-300er Steuerung vornehmen (Unterweisung Elektriker / - in) (German Edition) Hardwarekonfiguration an einer Siemens S7-300er Steuerung vornehmen (Unterweisung Elektriker / - in) (German Edition) Thomas Schäfer Click here if your download doesn"t start automatically Hardwarekonfiguration

Mehr

Netzwerke und Sicherheit auf mobilen Geräten

Netzwerke und Sicherheit auf mobilen Geräten Netzwerke und Sicherheit auf mobilen Geräten Univ.-Prof. Priv.-Doz. DI Dr. René Mayrhofer Antrittsvorlesung Johannes Kepler Universität Linz Repräsentationsräume 1. Stock (Uni-Center) 19.1.2015, 16:00

Mehr

Preisliste für The Unscrambler X

Preisliste für The Unscrambler X Preisliste für The Unscrambler X english version Alle Preise verstehen sich netto zuzüglich gesetzlicher Mehrwertsteuer (19%). Irrtümer, Änderungen und Fehler sind vorbehalten. The Unscrambler wird mit

Mehr

Materialien zu unseren Lehrwerken

Materialien zu unseren Lehrwerken Word order Word order is important in English. The word order for subjects, verbs and objects is normally fixed. The word order for adverbial and prepositional phrases is more flexible, but their position

Mehr

Listening Comprehension: Talking about language learning

Listening Comprehension: Talking about language learning Talking about language learning Two Swiss teenagers, Ralf and Bettina, are both studying English at a language school in Bristo and are talking about language learning. Remember that Swiss German is quite

Mehr

PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB

PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB Read Online and Download Ebook PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: ENGLISCH LERNEN MIT JUSTUS, PETER UND BOB DOWNLOAD EBOOK : PONS DIE DREI??? FRAGEZEICHEN, ARCTIC ADVENTURE: Click link bellow

Mehr

Group and Session Management for Collaborative Applications

Group and Session Management for Collaborative Applications Diss. ETH No. 12075 Group and Session Management for Collaborative Applications A dissertation submitted to the SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZÜRICH for the degree of Doctor of Technical Seiences

Mehr

Notice: All mentioned inventors have to sign the Report of Invention (see page 3)!!!

Notice: All mentioned inventors have to sign the Report of Invention (see page 3)!!! REPORT OF INVENTION Please send a copy to An die Abteilung Technologietransfer der Universität/Hochschule An die Technologie-Lizenz-Büro (TLB) der Baden-Württembergischen Hochschulen GmbH Ettlinger Straße

Mehr

https://portal.microsoftonline.com

https://portal.microsoftonline.com Sie haben nun Office über Office365 bezogen. Ihr Account wird in Kürze in dem Office365 Portal angelegt. Anschließend können Sie, wie unten beschrieben, die Software beziehen. Congratulations, you have

Mehr

ZWISCHEN TRADITION UND REBELLION - FRAUENBILDER IM AKTUELLEN BOLLYWOODFILM (GERMAN EDITION) BY CHRISTINE STöCKEL

ZWISCHEN TRADITION UND REBELLION - FRAUENBILDER IM AKTUELLEN BOLLYWOODFILM (GERMAN EDITION) BY CHRISTINE STöCKEL Read Online and Download Ebook ZWISCHEN TRADITION UND REBELLION - FRAUENBILDER IM AKTUELLEN BOLLYWOODFILM (GERMAN EDITION) BY CHRISTINE STöCKEL DOWNLOAD EBOOK : ZWISCHEN TRADITION UND REBELLION - FRAUENBILDER

Mehr

Klassische Themen der Computerwissenschaft

Klassische Themen der Computerwissenschaft Klassische Themen der Computerwissenschaft 3 VO, 1 UE, 716.113 / 716.114 Bernhard Aichernig Oswin Aichholzer Alexander Felfernig Gerald Steinbauer Franz Wotawa Institut für Softwaretechnologie, TU Graz

Mehr

Social Innovation and Transition

Social Innovation and Transition Social Innovation and Transition Dmitri Domanski & Jürgen Howaldt TU Dortmund University Sozialforschungsstelle Dortmund Co-innovation theory and practice to facilitate change Wageningen, October 28-29,

Mehr

ISO 15504 Reference Model

ISO 15504 Reference Model Prozess Dimension von SPICE/ISO 15504 Process flow Remarks Role Documents, data, tools input, output Start Define purpose and scope Define process overview Define process details Define roles no Define

Mehr

TMF projects on IT infrastructure for clinical research

TMF projects on IT infrastructure for clinical research Welcome! TMF projects on IT infrastructure for clinical research R. Speer Telematikplattform für Medizinische Forschungsnetze (TMF) e.v. Berlin Telematikplattform für Medizinische Forschungsnetze (TMF)

Mehr

Cloud Architektur Workshop

Cloud Architektur Workshop Cloud Architektur Workshop Ein Angebot von IBM Software Services for Cloud & Smarter Infrastructure Agenda 1. Überblick Cloud Architektur Workshop 2. In 12 Schritten bis zur Cloud 3. Workshop Vorgehensmodell

Mehr

SmartGrid Security. Jörn Müller-Quade, KASTEL, FZI, KIT. kastel.kit.edu KOMPETENZZENTRUM FÜR ANGEWANDTE SICHERHEITSTECHNOLOGIE

SmartGrid Security. Jörn Müller-Quade, KASTEL, FZI, KIT. kastel.kit.edu KOMPETENZZENTRUM FÜR ANGEWANDTE SICHERHEITSTECHNOLOGIE SmartGrid Security Jörn Müller-Quade, KASTEL, FZI, KIT KOMPETENZZENTRUM FÜR ANGEWANDTE SICHERHEITSTECHNOLOGIE KIT Universität des Landes Baden-Württemberg und nationales Forschungszentrum in der Helmholtz-Gemeinschaft

Mehr

Killy Literaturlexikon: Autoren Und Werke Des Deutschsprachigen Kulturraumes 2., Vollstandig Uberarbeitete Auflage (German Edition)

Killy Literaturlexikon: Autoren Und Werke Des Deutschsprachigen Kulturraumes 2., Vollstandig Uberarbeitete Auflage (German Edition) Killy Literaturlexikon: Autoren Und Werke Des Deutschsprachigen Kulturraumes 2., Vollstandig Uberarbeitete Auflage (German Edition) Walther Killy Click here if your download doesn"t start automatically

Mehr

Mercedes OM 636: Handbuch und Ersatzteilkatalog (German Edition)

Mercedes OM 636: Handbuch und Ersatzteilkatalog (German Edition) Mercedes OM 636: Handbuch und Ersatzteilkatalog (German Edition) Mercedes-Benz Click here if your download doesn"t start automatically Mercedes OM 636: Handbuch und Ersatzteilkatalog (German Edition) Mercedes-Benz

Mehr

Unit 1. Motivation and Basics of Classical Logic. Fuzzy Logic I 6

Unit 1. Motivation and Basics of Classical Logic. Fuzzy Logic I 6 Unit 1 Motivation and Basics of Classical Logic Fuzzy Logic I 6 Motivation In our everyday life, we use vague, qualitative, imprecise linguistic terms like small, hot, around two o clock Even very complex

Mehr

Exercise (Part II) Anastasia Mochalova, Lehrstuhl für ABWL und Wirtschaftsinformatik, Kath. Universität Eichstätt-Ingolstadt 1

Exercise (Part II) Anastasia Mochalova, Lehrstuhl für ABWL und Wirtschaftsinformatik, Kath. Universität Eichstätt-Ingolstadt 1 Exercise (Part II) Notes: The exercise is based on Microsoft Dynamics CRM Online. For all screenshots: Copyright Microsoft Corporation. The sign ## is you personal number to be used in all exercises. All

Mehr

The Future Internet in Germany and Europe

The Future Internet in Germany and Europe The Future Internet in Germany and Europe David Kennedy Direktor Eurescom GmbH Heidelberg, Deutschland Starting point.. The innovative society and sustainable economy of Europe 2020 will be based on ICT

Mehr

Algorithms for graph visualization

Algorithms for graph visualization Algorithms for graph visualization Project - Orthogonal Grid Layout with Small Area W INTER SEMESTER 2013/2014 Martin No llenburg KIT Universita t des Landes Baden-Wu rttemberg und nationales Forschungszentrum

Mehr

Kryptographie im 21. Jahrhundert

Kryptographie im 21. Jahrhundert Kryptographie im 21. Jahrhundert Willi Meier Simon Fischer Jean-Philippe Aumasson Institut IAST 1 / 25 Einsatz von Kryptographie Absicherung von Computernetzwerken Sicherer elektronischer Geldtransfer

Mehr

VGM. VGM information. HAMBURG SÜD VGM WEB PORTAL - USER GUIDE June 2016

VGM. VGM information. HAMBURG SÜD VGM WEB PORTAL - USER GUIDE June 2016 Overview The Hamburg Süd VGM-Portal is an application which enables to submit VGM information directly to Hamburg Süd via our e-portal web page. You can choose to insert VGM information directly, or download

Mehr

11. Das RSA Verfahren und andere Verfahren

11. Das RSA Verfahren und andere Verfahren Chr.Nelius: Kryptographie (SS 2011) 31 11. Das RSA Verfahren und andere Verfahren Eine konkrete Realisierung eines Public Key Kryptosystems ist das sog. RSA Verfahren, das im Jahre 1978 von den drei Wissenschaftlern

Mehr

Number of Maximal Partial Clones

Number of Maximal Partial Clones Number of Maximal Partial Clones KARSTEN SCHÖLZEL Universität Rostoc, Institut für Mathemati 26th May 2010 c 2010 UNIVERSITÄT ROSTOCK MATHEMATISCH-NATURWISSENSCHAFTLICHE FAKULTÄT, INSTITUT FÜR MATHEMATIK

Mehr

Patentrelevante Aspekte der GPLv2/LGPLv2

Patentrelevante Aspekte der GPLv2/LGPLv2 Patentrelevante Aspekte der GPLv2/LGPLv2 von RA Dr. Till Jaeger OSADL Seminar on Software Patents and Open Source Licensing, Berlin, 6./7. November 2008 Agenda 1. Regelungen der GPLv2 zu Patenten 2. Implizite

Mehr

Die besten Chuck Norris Witze: Alle Fakten über den härtesten Mann der Welt (German Edition)

Die besten Chuck Norris Witze: Alle Fakten über den härtesten Mann der Welt (German Edition) Die besten Chuck Norris Witze: Alle Fakten über den härtesten Mann der Welt (German Edition) Click here if your download doesn"t start automatically Die besten Chuck Norris Witze: Alle Fakten über den

Mehr

KURZANLEITUNG. Firmware-Upgrade: Wie geht das eigentlich?

KURZANLEITUNG. Firmware-Upgrade: Wie geht das eigentlich? KURZANLEITUNG Firmware-Upgrade: Wie geht das eigentlich? Die Firmware ist eine Software, die auf der IP-Kamera installiert ist und alle Funktionen des Gerätes steuert. Nach dem Firmware-Update stehen Ihnen

Mehr

MobiDM-App Handbuch für Windows Mobile

MobiDM-App Handbuch für Windows Mobile MobiDM-App Handbuch für Windows Mobile Dieses Handbuch beschreibt die Installation und Nutzung der MobiDM-App für Windows Mobile Version: x.x MobiDM-App Handbuch für Windows Mobile Seite 1 Inhalt 1. WILLKOMMEN

Mehr

9 Schlüsseleinigung, Schlüsselaustausch

9 Schlüsseleinigung, Schlüsselaustausch 9 Schlüsseleinigung, Schlüsselaustausch Ziel: Sicherer Austausch von Schlüsseln über einen unsicheren Kanal initiale Schlüsseleinigung für erste sichere Kommunikation Schlüsselerneuerung für weitere Kommunikation

Mehr

p^db=`oj===pìééçêíáåñçêã~íáçå=

p^db=`oj===pìééçêíáåñçêã~íáçå= p^db=`oj===pìééçêíáåñçêã~íáçå= Error: "Could not connect to the SQL Server Instance" or "Failed to open a connection to the database." When you attempt to launch ACT! by Sage or ACT by Sage Premium for

Mehr

RECHNUNGSWESEN. KOSTENBEWUßTE UND ERGEBNISORIENTIERTE BETRIEBSFüHRUNG. BY MARTIN GERMROTH

RECHNUNGSWESEN. KOSTENBEWUßTE UND ERGEBNISORIENTIERTE BETRIEBSFüHRUNG. BY MARTIN GERMROTH RECHNUNGSWESEN. KOSTENBEWUßTE UND ERGEBNISORIENTIERTE BETRIEBSFüHRUNG. BY MARTIN GERMROTH DOWNLOAD EBOOK : RECHNUNGSWESEN. KOSTENBEWUßTE UND Click link bellow and free register to download ebook: RECHNUNGSWESEN.

Mehr

Presentation of a diagnostic tool for hybrid and module testing

Presentation of a diagnostic tool for hybrid and module testing Presentation of a diagnostic tool for hybrid and module testing RWTH Aachen III. Physikalisches Institut B M.Axer, F.Beißel, C.Camps, V.Commichau, G.Flügge, K.Hangarter, J.Mnich, P.Schorn, R.Schulte, W.

Mehr

Routing in WSN Exercise

Routing in WSN Exercise Routing in WSN Exercise Thomas Basmer telefon: 0335 5625 334 fax: 0335 5625 671 e-mail: basmer [ at ] ihp-microelectronics.com web: Outline Routing in general Distance Vector Routing Link State Routing

Mehr

Finite Difference Method (FDM)

Finite Difference Method (FDM) Finite Difference Method (FDM) home/lehre/vl-mhs-1-e/folien/vorlesung/2a_fdm/cover_sheet.tex page 1 of 15. p.1/15 Table of contents 1. Problem 2. Governing Equation 3. Finite Difference-Approximation 4.

Mehr

Workflow, Business Process Management, 4.Teil

Workflow, Business Process Management, 4.Teil Workflow, Business Process Management, 4.Teil 24. Januar 2004 Der vorliegende Text darf für Zwecke der Vorlesung Workflow, Business Process Management des Autors vervielfältigt werden. Eine weitere Nutzung

Mehr

Addressing the Location in Spontaneous Networks

Addressing the Location in Spontaneous Networks Addressing the Location in Spontaneous Networks Enabling BOTH: Privacy and E-Commerce Design by Moritz Strasser 1 Disappearing computers Trends Mobility and Spontaneous Networks (MANET = Mobile Ad hoc

Mehr

Ways and methods to secure customer satisfaction at the example of a building subcontractor

Ways and methods to secure customer satisfaction at the example of a building subcontractor Abstract The thesis on hand deals with customer satisfaction at the example of a building subcontractor. Due to the problems in the building branch, it is nowadays necessary to act customer oriented. Customer

Mehr

Benjamin Whorf, Die Sumerer Und Der Einfluss Der Sprache Auf Das Denken (Philippika) (German Edition)

Benjamin Whorf, Die Sumerer Und Der Einfluss Der Sprache Auf Das Denken (Philippika) (German Edition) Benjamin Whorf, Die Sumerer Und Der Einfluss Der Sprache Auf Das Denken (Philippika) (German Edition) Sebastian Fink Click here if your download doesn"t start automatically Benjamin Whorf, Die Sumerer

Mehr

Level 1 German, 2014

Level 1 German, 2014 90886 908860 1SUPERVISOR S Level 1 German, 2014 90886 Demonstrate understanding of a variety of German texts on areas of most immediate relevance 9.30 am Wednesday 26 November 2014 Credits: Five Achievement

Mehr

Exercise (Part XI) Anastasia Mochalova, Lehrstuhl für ABWL und Wirtschaftsinformatik, Kath. Universität Eichstätt-Ingolstadt 1

Exercise (Part XI) Anastasia Mochalova, Lehrstuhl für ABWL und Wirtschaftsinformatik, Kath. Universität Eichstätt-Ingolstadt 1 Exercise (Part XI) Notes: The exercise is based on Microsoft Dynamics CRM Online. For all screenshots: Copyright Microsoft Corporation. The sign ## is you personal number to be used in all exercises. All

Mehr

Mail encryption Gateway

Mail encryption Gateway Mail encryption Gateway Anwenderdokumentation Copyright 06/2015 by arvato IT Support All rights reserved. No part of this document may be reproduced or transmitted in any form or by any means, electronic

Mehr

There are 10 weeks this summer vacation the weeks beginning: June 23, June 30, July 7, July 14, July 21, Jul 28, Aug 4, Aug 11, Aug 18, Aug 25

There are 10 weeks this summer vacation the weeks beginning: June 23, June 30, July 7, July 14, July 21, Jul 28, Aug 4, Aug 11, Aug 18, Aug 25 Name: AP Deutsch Sommerpaket 2014 The AP German exam is designed to test your language proficiency your ability to use the German language to speak, listen, read and write. All the grammar concepts and

Mehr

Einkommensaufbau mit FFI:

Einkommensaufbau mit FFI: For English Explanation, go to page 4. Einkommensaufbau mit FFI: 1) Binäre Cycle: Eine Position ist wie ein Business-Center. Ihr Business-Center hat zwei Teams. Jedes mal, wenn eines der Teams 300 Punkte

Mehr

IT-Forum Darmstadt 16. Mai 2008

IT-Forum Darmstadt 16. Mai 2008 Headline IT-Sicherheit einzeilig made in Germany?! oder Themen-Panel zweizeilig IT-Sicherheit IT-Forum Darmstadt 16. Mai 2008 Claudia Eckert 16.05.2008, Prof. Dr. Claudia Eckert, FhI-SIT 1 Bedeutung der

Mehr

IT-Sicherheit Kapitel 3 Public Key Kryptographie

IT-Sicherheit Kapitel 3 Public Key Kryptographie IT-Sicherheit Kapitel 3 Public Key Kryptographie Dr. Christian Rathgeb Sommersemester 2013 1 Einführung In der symmetrischen Kryptographie verwenden Sender und Empfänger den selben Schlüssel die Teilnehmer

Mehr

Unternehmensweite IT Architekturen

Unternehmensweite IT Architekturen Unternehmensweite IT Architekturen Part 1: IT Systems Architecture, Roles and Responsibilities of IT Architects Part 2: Solution Architecture, based on a practical Case Study Part 3: SOA (Service Oriented

Mehr

Technische Universität Kaiserslautern Lehrstuhl für Virtuelle Produktentwicklung

Technische Universität Kaiserslautern Lehrstuhl für Virtuelle Produktentwicklung functions in SysML 2.0 La Jolla, 22.05.2014 12/10/2015 Technische Universität Kaiserslautern Lehrstuhl für Virtuelle Produktentwicklung Dipl. Wirtsch.-Ing. Christian Muggeo Dipl. Wirtsch.-Ing. Michael

Mehr

Vorstellung RWTH Gründerzentrum

Vorstellung RWTH Gründerzentrum Vorstellung RWTH Gründerzentrum Once an idea has been formed, the center for entrepreneurship supports in all areas of the start-up process Overview of team and services Development of a business plan

Mehr

Multicriterial Design Decision Making regarding interdependent Objectives in DfX

Multicriterial Design Decision Making regarding interdependent Objectives in DfX Overview Multicriterial Design Decision Making regarding interdependent Objectives in DfX S. Bauer The Design Process Support of the Design Process with Design for X Visualization of Decision Problems

Mehr

EU nimmt neues Programm Mehr Sicherheit im Internet in Höhe von 55 Millionen für mehr Sicherheit für Kinder im Internet an

EU nimmt neues Programm Mehr Sicherheit im Internet in Höhe von 55 Millionen für mehr Sicherheit für Kinder im Internet an IP/8/899!"#$$%&')*+%,%-.%"/ EU nimmt neues Programm Mehr Sicherheit im Internet in Höhe von Millionen für mehr Sicherheit für Kinder im Internet an Ab. Januar 9 hat die EU ein neues Programm für eine sicherere

Mehr