Lecture Notes in Computer Science Edited by G Goos, J Hartmanis and J van Leeuwen 1880 Berlin Heidelberg New York Barcelona Hong Kong London Milan Paris Singapore Tokyo Mihir Bellare (Ed.) Advances in Cryptology – CRYPTO 2000 20th Annual International Cryptology Conference Santa Barbara, California, USA, August 20-24, 2000 Proceedings 13 Series Editors Gerhard Goos, Karlsruhe University, Germany Juris Hartmanis, Cornell University, NY, USA Jan van Leeuwen, Utrecht University, The Netherlands Volume Editor Mihir Bellare University of California, Department of Computer Science and Engineering, 0114 9500 Gilman Drive, La Jolla, CA 92093, USA E-mail: mihir@cs.ucsd.edu Cataloging-in-Publication Data applied for Die Deutsche Bibliothek - CIP-Einheitsaufnahme Advances in cryptology : proceedings / CRYPTO 2000, 20th Annual International Cryptology Conference, Santa Barbara, California, USA, August 20 - 24, 2000 Mihir Bellare (ed.) [IACR] - Berlin ; Heidelberg ; New York ; Barcelona ; Hong Kong ; London ; Milan ; Paris ; Singapore ; Tokyo : Springer, 2000 (Lecture notes in computer science ; Vol 1880) ISBN 3-540-67907-3 CR Subject Classification (1998): E.3, G.2.1, D.4.6, K.6.5, F.2.1-2, C.2, J.1 ISSN 0302-9743 ISBN 3-540-67907-3 Springer-Verlag Berlin Heidelberg New York This work is subject to copyright All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broadcasting, reproduction on microfilms or in any other way, and storage in data banks Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer-Verlag Violations are liable for prosecution under the German Copyright Law Springer-Verlag is a company in the BertelsmannSpringer publishing group © Springer-Verlag Berlin Heidelberg 2000 Printed in Germany Typesetting: Camera-ready by author, data conversion by Steingrăaber Satztechnik GmbH, Heidelberg Printed on acid-free paper SPIN: 10722418 06/3142 543210 Preface Crypto 2000 was the 20th Annual Crypto conference It was sponsored by the International Association for Cryptologic Research (IACR) in cooperation with the IEEE Computer Society Technical Committee on Security and Privacy and the Computer Science Department of the University of California at Santa Barbara The conference received 120 submissions, and the program committee selected 32 of these for presentation Extended abstracts of revised versions of these papers are in these proceedings The authors bear full responsibility for the contents of their papers The conference program included two invited lectures Don Coppersmith’s presentation “The development of DES” recorded his involvement with one of the most important cryptographic developments ever, namely the Data Encryption Standard, and was particularly apt given the imminent selection of the Advanced Encryption Standard Mart´ın Abadi’s presentation “Taming the Adversary” was about bridging the gap between useful but perhaps simplistic threat abstractions and rigorous adversarial models, or perhaps, even more generally, between viewpoints of the security and cryptography communities An abstract corresponding to Mart´ın’s talk is included in these proceedings The conference program also included its traditional “rump session” of short, informal or impromptu presentations, chaired this time by Stuart Haber These presentations are not reflected in these proceedings An electronic submission process was available and recommended, but for the first time used a web interface rather than email (Perhaps as a result, there were no hardcopy submissions.) The submission review process had three phases In the first phase, program committee members compiled reports (assisted at their discretion by sub-referees of their choice, but without interaction with other program committee members) and entered them, via web forms, into web-review software running at UCSD In the second phase, committee members used the software to browse each other’s reports, discuss, and update their own reports Lastly there was a program committee meeting to discuss the difficult cases I am extremely grateful to the program committee members for their enormous investment of time, effort, and adrenaline in the difficult and delicate process of review and selection (A list of program committee members and subreferees they invoked can be found on succeeding pages of this volume.) I also thank the authors of submitted papers —in equal measure regardless of whether their papers were accepted or not— for their submissions It is the work of this body of researchers that makes this conference possible I thank Rebecca Wright for hosting the program committee meeting at the AT&T building in New York City and managing the local arrangements, and Ran Canetti for organizing the post-PC-meeting dinner with his characteristic gastronomic and oenophilic flair VI Preface The web-review software we used was written for Eurocrypt 2000 by Wim Moreau and Joris Claessens under the direction of Eurocrypt 2000 program chair Bart Preneel, and I thank them for allowing us to deploy their useful and colorful tool I am most grateful to Chanathip Namprempre (aka Meaw) who provided systems, logistical, and moral support for the entire Crypto 2000 process She wrote the software for the web-based submissions, adapted and ran the webreview software at UCSD, and compiled the final abstracts into the proceedings you see here She types faster than I speak I am grateful to Hugo Krawczyk for his insight and advice, provided over a long period of time with his usual combination of honesty and charm, and to him and other past program committee chairs, most notably Michael Wiener and Bart Preneel, for replies to the host of questions I posed during the process In addition I received useful advice from many members of our community including Silvio Micali, Tal Rabin, Ron Rivest, Phil Rogaway, and Adi Shamir Finally thanks to Matt Franklin who as general chair was in charge of the local organization and finances, and, on the IACR side, to Christian Cachin, Kevin McCurley, and Paul Van Oorschot Chairing a Crypto program committee is a learning process I have come to appreciate even more than before the quality and variety of work in our field, and I hope the papers in this volume contribute further to its development June 2000 Mihir Bellare Program Chair, Crypto 2000 CRYPTO 2000 August 20–24, 2000, Santa Barbara, California, USA Sponsored by the International Association for Cryptologic Research (IACR) in cooperation with IEEE Computer Society Technical Committee on Security and Privacy, Computer Science Department, University of California, Santa Barbara General Chair Matthew Franklin, Xerox Palo Alto Research Center, USA Program Chair Mihir Bellare, University of California, San Diego, USA Program Committee Alex Biryukov Weizmann Institute of Science, Israel Dan Boneh Stanford University, USA Christian Cachin IBM Research, Switzerland Ran Canetti IBM Research, USA Ronald Cramer ETH Zurich, Switzerland Yair Frankel CertCo, USA Shai Halevi IBM Research, USA Arjen Lenstra Citibank, USA Mitsuru Matsui Mitsubishi Electric Corporation, Japan Paul Van Oorschot Entrust Technologies, Canada Bart Preneel Katholieke Universiteit Leuven, Belgium Phillip Rogaway University of California, Davis, USA Victor Shoup IBM Zurich, Switzerland Jessica Staddon Bell Labs Research, Palo Alto, USA Jacques Stern Ecole Normale Sup´erieure, France Doug Stinson University of Waterloo, Canada Salil Vadhan Massachusetts Institute of Technology, USA David Wagner University of California, Berkeley, USA Rebecca Wright AT&T Laboratories Research, USA Advisory members Michael Wiener (Crypto 1999 program chair) Entrust Technologies, Canada Joe Kilian (Crypto 2001 program chair) Intermemory, USA VIII Organization Sub-Referees Bill Aiello, Jeehea An, Olivier Baudron, Don Beaver, Josh Benaloh, John Black, Simon Blackburn, Alexandra Boldyreva, Nikita Borisov, Victor Boyko, Jan Camenisch, Suresh Chari, Scott Contini, Don Coppersmith, Claude Cr´epeau, Ivan Damg˚ ard, Anand Desai , Giovanni Di Crescenzo, Yevgeniy Dodis, Matthias Fitzi, Matt Franklin, Rosario Gennaro, Guang Gong, Luis Granboulan, Nick Howgrave-Graham, Russell Impagliazzo, Yuval Ishai, Markus Jakobsson, Stas Jarecki, Thomas Johansson, Charanjit Jutla, Joe Kilian, Eyal Kushilevitz, Moses Liskov, Stefan Lucks, Anna Lysyanskaya, Philip MacKenzie, Subhamoy Maitra, Tal Malkin, Barbara Masucci, Alfred Menezes, Daniele Micciancio, Sara Miner, Ilia Mironov, Moni Naor , Phong Nguyen, Rafail Ostrovsky, Erez Petrank, Birgit Pfitzmann, Benny Pinkas, David Pointcheval, Guillaume Poupard, Tal Rabin, Charlie Rackoff, Zulfikar Ramzan, Omer Reingold, Leo Reyzin, Pankaj Rohatgi, Amit Sahai, Louis Salvail, Claus Schnorr, Mike Semanko, Bob Silverman, Joe Silverman, Dan Simon, Nigel Smart, Ben Smeets, Adam Smith, Martin Strauss, Ganesh Sundaram, Serge Vaudenay, Frederik Vercauteren, Bernhard von Stengel, Ruizhong Wei, Susanne Gudrun Wetzel, Colin Williams, Stefan Wolf, Felix Wu, Yiqun Lisa Yin, Amir Youssef, Robert Zuccherato Table of Contents XTR and NTRU The XTR Public Key System Arjen K Lenstra, Eric R Verheul A Chosen-Ciphertext Attack against NTRU 20 ´ Eliane Jaulmes, Antoine Joux Privacy for Databases Privacy Preserving Data Mining 36 Yehuda Lindell, Benny Pinkas Reducing the Servers Computation in Private Information Retrieval: PIR with Preprocessing 55 Amos Beimel, Yuval Ishai, Tal Malkin Secure Distributed Computation and Applications Parallel Reducibility for Information-Theoretically Secure Computation 74 Yevgeniy Dodis, Silvio Micali Optimistic Fair Secure Computation 93 Christian Cachin, Jan Camenisch A Cryptographic Solution to a Game Theoretic Problem 112 Yevgeniy Dodis, Shai Halevi, Tal Rabin Algebraic Cryptosystems Differential Fault Attacks on Elliptic Curve Cryptosystems 131 Ingrid Biehl, Bernd Meyer, Volker Mă uller Quantum Public-Key Cryptosystems 147 Tatsuaki Okamoto, Keisuke Tanaka, Shigenori Uchiyama New Public-Key Cryptosystem Using Braid Groups 166 Ki Hyoung Ko, Sang Jin Lee, Jung Hee Cheon, Jae Woo Han, Ju-sung Kang, Choonsik Park Message Authentication Key Recovery and Forgery Attacks on the MacDES MAC Algorithm 184 Don Coppersmith, Lars R Knudsen, Chris J Mitchell 530 Palash Sarkar and Subhamoy Maitra It can be checked that there are variables between X5 , , X1 which occur odd number of times overall in the above sequence Hence the degree of f is Also the nonlinearity of f can be shown to be 480 Note that the constructed function is not a saturated function and its Walsh spectrum is five-valued (0, ±32, ±64) In Table 2, we list some of the best known functions Also Table provides some open problems Table Some best known functions The (8, 3, 4, 112) and (9, 4, 4, 224) functions are from Theorem and the (10, 3, 6, 480) function is from Theorem All the other constructions were known previously [21,19] n (7,1,5,56), (7,3,3,48), (7,4,2,32) (8,1,6,112), (8,2,5,112), (8,3,4,112), (8,4,3,96), (8,5,2,64) (9,1,7,240), (9,2,5,240), (9,3,5,224), (9,4,4,224), (9,5,3,192), (9,6,2,128) 10 (10,1,8,484), (10,2,7,480), (10,3,6,480), (10,4,5,448), (10,5,4,448), (10,6,3,384), (10,7,2,256) Table Existence of functions with these parameters is not known n 10 (7,2,−,56) (8,1,−,116) (9,1,−,244), (9,2,6,240) (10,1,−,492), (10,1,−,488), (10,2,−,488), (10,4,−,480) Notes : In a recent work [24], Tarannikov showed that the maximum possible nonlinearity of an n-variable, m-resilient function is 2n−1 − 2m+1 for 2n−7 ≤ m ≤ n − and functions achieving this nonlinearity must have maximum possible algebraic degree n − m − Also a construction method for such n-variable functions with the additional restriction that each variable occurs in a maximum degree term is provided for m in the range 2n−7 ≤ m ≤ n − log2 n−2 3 − Acknowledgement The authors are grateful to the anonymous referees for many comments which helped to improve the presentation of the paper Nonlinearity Bounds and Constructions of Resilient Boolean Functions 531 References P Camion, C Carlet, P Charpin, and N Sendrier On correlation immune functions In Advances in Cryptology - CRYPTO’91, pages 86–100 Springer-Verlag, 1992 C Carlet More correlation immune and resilient functions over Galois fields and Galois rings In Advances in Cryptology - EUROCRYPT’97, pages 422–433 Springer-Verlag, May 1997 S Chee, S Lee, D Lee, and S H Sung On the correlation immune functions and their nonlinearity In Advances in Cryptology, Asiacrypt 96, number 1163 in Lecture Notes in Computer Science, pages 232–243 Springer-Verlag, 1996 C Ding, G Xiao, and W Shan The Stability Theory of Stream Ciphers Number 561 in Lecture Notes in Computer Science Springer-Verlag, 1991 E Filiol and C Fontaine Highly nonlinear balanced Boolean functions with a good correlation-immunity In Advances in Cryptology - EUROCRYPT’98 SpringerVerlag, 1998 X Guo-Zhen and J Massey A spectral characterization of correlation immune combining functions IEEE Transactions on Information Theory, 34(3):569–571, May 1988 X Hou Covering radius of the Reed-Muller code R(1, 7) - a simpler proof Journal of Combinatorial Theory, Series A, 74(3):337–341, 1996 X Hou On the covering radius of R(1, m) in R(3, m) IEEE Transactions on Information Theory, 42(3):1035–1037, 1996 X Hou On the norm and covering radius of the first order Reed-Muller codes IEEE Transactions on Information Theory, 43(3):1025–1027, 1997 10 T Johansson and F Jonsson Fast correlation attacks based on turbo code techniques In Advances in Cryptology - CRYPTO’99, number 1666 in Lecture Notes in Computer Science, pages 181–197 Springer-Verlag, August 1999 11 T Johansson and F Jonsson Improved fast correlation attacks on stream ciphers via convolutional codes In Advances in Cryptology - EUROCRYPT’99, number 1592 in Lecture Notes in Computer Science, pages 347–362 Springer-Verlag, May 1999 12 F J MacWillams and N J A Sloane The Theory of Error Correcting Codes North Holland, 1977 13 S Maitra and P Sarkar Highly nonlinear resilient functions optimizing Siegenthaler’s inequality In Advances in Cryptology - CRYPTO’99, number 1666 in Lecture Notes in Computer Science, pages 198–215 Springer Verlag, August 1999 14 W Meier and O Staffelbach Fast correlation attack on stream ciphers In Advances in Cryptology - EUROCRYPT’88, volume 330, pages 301–314 SpringerVerlag, May 1988 15 S Palit and B K Roy Cryptanalysis of LFSR-encrypted codes with unknown combining functions In Advances in Cryptology - ASIACRYPT’99, number 1716 in Lecture Notes in Computer Science, pages 306–320 Springer Verlag, November 1999 16 E Pasalic and T Johansson Further results on the relation between nonlinearity and resiliency of Boolean functions In IMA Conference on Cryptography and Coding, number 1746 in Lecture Notes in Computer Science, pages 35–45 SpringerVerlag, 1999 17 R A Rueppel Analysis and Design of Stream Ciphers Springer Verlag, 1986 532 Palash Sarkar and Subhamoy Maitra 18 R A Rueppel and O J Staffelbach Products of linear recurring sequences with maximum complexity IEEE Transactions on Information Theory, IT-33:124–131, January 1987 19 P Sarkar and S Maitra Construction of nonlinear Boolean functions with important cryptographic properties In Advances in Cryptology - EUROCRYPT 2000, number 1807 in Lecture Notes in Computer Science, pages 491–512 Springer Verlag, 2000 20 J Seberry, X M Zhang, and Y Zheng Nonlinearly balanced Boolean functions and their propagation characteristics In Advances in Cryptology - CRYPTO’93, pages 49–60 Springer-Verlag, 1994 21 J Seberry, X M Zhang, and Y Zheng On constructions and nonlinearity of correlation immune Boolean functions In Advances in Cryptology - EUROCRYPT’93, pages 181–199 Springer-Verlag, 1994 22 T Siegenthaler Correlation-immunity of nonlinear combining functions for cryptographic applications IEEE Transactions on Information Theory, IT-30(5):776– 780, September 1984 23 T Siegenthaler Decrypting a class of stream ciphers using ciphertext only IEEE Transactions on Computers, C-34(1):81–85, January 1985 24 Y V Tarannikov On resilient Boolean functions with maximum possible nonlinearity Cryptology ePrint Archive, eprint.iacr.org, No 2000/005, 2000 Almost Independent and Weakly Biased Arrays: Efficient Constructions and Cryptologic Applications Jă urgen Bierbrauer1 and Holger Schellwat2 Department of Mathematical Sciences, Michigan Technological University, Houghton, Michigan 49931, USA jbierbra@mtu.edu ă ă Department of Natural Sciences, Orebro University, SE-70182 Orebro, Sweden holger.schellwat@nat.oru.se Abstract The best known constructions for arrays with low bias are those from [1] and the exponential sum method based on the WeilCarlitz-Uchiyama bound They all yield essentially the same parameters We present new efficient coding-theoretic constructions, which allow farreaching generalizations and improvements The classical constructions can be described as making use of Reed-Solomon codes Our recursive construction yields greatly improved parameters even when applied to Reed-Solomon codes Use of algebraic-geometric codes leads to even better results, which are optimal in an asymptotic sense The applications comprise universal hashing, authentication, resilient functions and pseudorandomness Key Words: Low bias, almost independent arrays, Reed-Solomon codes, Hermitian codes, Suzuki codes, Fourier transform, Weil-Carlitz-Uchiyama bound, exponential sum method, Zyablov bound, hashing, authentication, resiliency Introduction The concepts of limited dependence and low bias have manifold applications in cryptography and complexity theory We mention universal hashing, authentication, resiliency against correlation attacks, pseudorandomness, block ciphers, derandomization, two-point based sampling, zero-knowledge, span programs, testing of combinatorial circuits, intersecting codes, oblivious transfer, interactive proof systems, resiliency (see [19,16,18,17,1,11,25,10,6,9,7,13,16]) A basic notion underlying these concepts are families of −biased random variables The WeilCarlitz-Uchiyama bound and several constructions from the influential papers by Naor and Naor [18] and by Alon, Goldreich, H˚ astad and Peralta [1] provide families of −biased random variables All these classical constructions yield very similar parameters In this paper we describe methods, which generalize these constructions and yield far-reaching improvements Essential ingredients are linear codes and the Fourier transform M Bellare (Ed.): CRYPTO 2000, LNCS 1880, pp 533–543, 2000 c Springer-Verlag Berlin Heidelberg 2000 534 Jă urgen Bierbrauer and Holger Schellwat Bias and Dependency We use neutral notation which is suited to describe all the applications (hashing, authentication, derandomization, pseudorandomness, ) Definition Let p be a prime An (n, k)p −array A is an array with n rows and k columns, where the entries are taken from a set with p elements Definition Let p be a prime, v = (v1 , v2 , , ) ∈ IFnp For every i ∈ IFp = ZZ/pZZ let νi (v) be the frequency of i as an entry of v Let ζ be a primitive complex p−th root of unity The bias of v is defined as bias(v) = n νi (v)ζ i i∈IFp We have ≤ bias(v) ≤ As i∈IFp ζ i = the bias is low if all elements of IFp occur with approximately the same frequency as entries in v Definition Let ≤ < An (n, k)p −array is −biased if every nontrivial linear combination of its columns has bias ≤ The bias of an array is a property of the IFp −linear code generated by the columns The bias of the array is low if and only if every nonzero word of the code has low bias While the bias of a vector depends on the choice of the root of unity, the bias of an array is independent of this choice Definition Let ≤ < An (n, k)p −array is t-wise −biased if every nontrivial linear combination of at most t of its columns has bias ≤ Definition Let ≤ < An (n, k)p −array A is t-wise -dependent if for every set U of s ≤ t columns and every a ∈ IFsp the frequency νU (a) of rows of A, whose projection onto U equals a satisfies νU (a) − 1/ps ≤ n The notion of a t-wise -dependent array generalizes the combinatorial notion of an orthogonal array of strength t (equivalently: t−universal family of hash functions in the sense of Carter/Wegman [11]) An array is t-wise independent (=0-dependent) if and only if it is an orthogonal array of strength t The most important of these concepts from the point of view of applications is t-wise -dependency It captures the familiar theme of representing a family of random variables (the columns of the array) on a small sample space (the rows of the array, with uniform distribution) such that any t of the random variables are almost statistically independent Almost Independent and Weakly Biased Arrays 535 We want to point out in the sequel that the construction problem of t-wise -dependent arrays can be efficiently reduced to the construction of −biased arrays This is the basic idea behind [18] The following construction of t-wise −biased arrays is essentially from [18] Theorem Let the following be given: – An (n, k)p −array B, which is −biased – A linear code [N, N − k, t + 1]p Then we can construct an (n, N )p −array, which is t-wise −biased Theorem An array, which is t-wise −biased, is also t-wise for some < −dependent The fundamental Theorem is proved in a nontrivial but standard way by using the Fourier transform, see [5] The following construction from the journal version of [16] is obvious and useful: Theorem If there is an array (n, k )p , which is t −wise −dependent, and t ≤ t /l, k ≤ k /l, then there is an array (n, k)pl , which is t−wise −dependent We see that indeed the central problem is to efficiently construct −biased arrays Linear codes are then used to construct t −wise −biased arrays via Theorem The standard method is to use BCH codes The resulting t −wise −biased arrays are also t −wise −dependent by Theorem Because of Theorem it is possible to concentrate entirely on binary arrays We turn to the basic problem of constructing weakly biased arrays Definition Denote by fp (b, e) the minimum a such that there is an array (pa , pb )p , which is p−e −biased Clearly fp (b, e) is weakly monotonely increasing in both arguments The construction from [1] shows the following: Theorem There is an efficient construction showing fp (b, e) ≤ 2(b + e) The Weil-Carlitz-Uchiyama Construction The celebrated Weil-Carlitz-Uchiyama bound [8] may be understood as a limit on the bias of dual BCH-codes More precisely, let (aj ) be a basis of IFpf |IFp and T r : IFpf −→ IFp the trace Consider the array A whose rows are indexed by the elements α ∈ IFpf and whose columns are indexed by aj X i , where i ≤ n and i is not a multiple of p The corresponding entry is T r(aj αi ) The WCU bound asserts that this (pf , f (n − n/p ))p −array has bias ≤ (n − 1)p−f /2 Comparison reveals that the WCU construction (exponential sum method) yields parameters which are very similar to (a little better than) Theorem All constructions based on one of these classical methods will produce about the same parameters 536 Jă urgen Bierbrauer and Holger Schellwat The Zyablov Construction As remarked earlier Theorem makes it possible to base the construction on binary −biased arrays This has the advantage that a direct link to coding theory can be used An array (n, k)2 is −biased if and only if the code generated by the columns has dimension k and the relative weights of all nonzero codewords are in the interval of length centered at 1/2 This elementary observation yields an immediate reduction of the construction problem of binary weakly biased arrays to the construction of linear codes containing the all-1-vector Theorem Let ≤ < The following are equivalent: – An (n, k)2 -array, which is −biased – A binary linear code of length n and dimension k + 1, which contains and whose minimal distance d satisfies 1− d ≥ n Constructing families of −biased (n, k)2 −arrays which are asymptotically nontrivial (meaning that is fixed and k/n ≥ R > 0) is equivalent to constructing asymptotically nontrivial families of binary linear codes containing the all-1vector The question of determining the asymptotics of binary codes is one of the most famous and most well-studied problems in coding theory The question is how incisive the additional condition is A famous simple result is the GilbertVarshamov bound: for every prime-power q and δ < (q − 1)/q the rate R = − Hq (δ) can be asymptotically reached by families of q−ary linear codes It can be managed that the all-1-word is contained in all these codes Unfortunately this bound is not constructive The construction given in [15] does not yield linear codes The Justesenmethod [14,21] is constructive, but the all-1-word is not contained in the resulting codes The Justesen method when applied to families of algebraic-geometric codes yields precisely the Zyablov bound However, for the same reason as above this does not yield families of binary −biased arrays More interesting for our problem is the original semi-constructive proof of the Zyablov bound [27] In fact, apply concatenation to a Reed-Solomon code [q m , rq m , (1 − r)q m ]qm as outer code and a code [n, m, d]q as inner code, where it is assumed that the inner code asymptotically meets the Gilbert-Varshamov bound (d/n = µ, m/n = − Hq (µ)) The concatenated code has parameters [q m n, rq m m, (1 − r)q m d]q , with relative distance δ = (1 − r)µ and rate R = r(1 − Hq (µ)) This construction shows that for every µ < (q − 1)/q and δ < µ we can construct families of q−ary linear codes with relative distance δ and rate R ≥ (1 − Hq (µ)(1 − δ/µ) The only drawback is that this is not really constructive However, for short inner codes this may be feasible Let us explore the situation in more detail We aim at a lower bound for f2 (b, e) Choose r = 2−(e+1) , µ = 12 − 2−(e+2) As the relative distance of the concatenated code is (1 − r)µ we obtain as bias Almost Independent and Weakly Biased Arrays 537 = 1−2(1−r)µ = 1−(1−2−(e+1) )2 = 2−e −2−(2e+2) It follows that < 2−e We have b = m + log(m) − e − and a = m + log(n) What is the order of magnitude of the rate S = − H2 (µ) of the inner code guaranteed by Gilbert-Varshamov? We have µ = 12 − 2−(e+2) = (2e+1 − 1)/2e+2 , − µ = (2e+1 + 1)/2e+2 and S = 1−2−(e+2) ((2e+1 −1)(e+2−log(2e+1 −1))+(2e+1 +1)(e+2−log(2e+1 +1)) Collecting the terms without log yields S = 2−(e+2) ((2e+1 − 1) log(2e+1 − 1) + (2e+1 + 1) log(2e+1 + 1)) − (e + 1) Divide the arguments of the log −terms by 2e+1 The term obtained from compensating for that is e + and cancels against the last summand We obtain S = 2−(e+2) ((2e+1 − 1) log(1 − 2−(e+1) ) + (2e+1 + 1) log(1 + 2−(e+1) )) Using the series for ln(1 ± x) we obtain S= 2−(e+2) (((2e+1 −1)(−2−(e+1) −2−2e−3 − .)+(2e+1 +1)(2−(e+1) −2−2e−3 +) ln(2) (−1 + 2−(e+1) − 2−e−2 + + 2−(e+1) − 2−e−2 ), ln(2) where terms involving −2e in the exponent and higher have been omitted This yields S ∼ 2−(2e+3) / ln(2) = 2−(e+2) Theorem The Zyablov method needs the construction of binary [n, m, d]2 codes, where n = ln(2)22e+3 m and d/n = 12 − 2−(e+2) The output is a weakly biased array showing f2 (m + log(m) − e − 1, e) ≤ m + log(m) + 2e + Theorem states f2 (b, e) ≤ b + 3e + It improves on the bound from Theorem when b > e A Coding-Theoretic Construction of Weakly Biased Arrays In Section we used an equivalent coding-theoretic interpretation of binary weakly biased arrays to obtain constructions Observe however that this does not seem to lead to explicit asymptotic constructions The Zyablov method presupposes exhaustive search for codes of moderate length attaining the GilbertVarshamov bound When p > an equivalent reduction to coding theory is not available Our next theorem provides a general link, which allows the use of linear codes in the construction of p−ary weakly biased arrays As this leads to efficient constructions, it is interesting even in the binary case Theorem Let C be a code [n, k, d]q , where q = pm and B an (n0 , m)p -array of bias We can construct an (nn0 , km)p −array with bias = − δ + δ < − δ + , where δ = d/n is the relative distance of code C A proof of Theorem is in [5] Application of Theorem to Reed-Solomon codes [pm , Rpm , (1 − R)pm ]pm and inner unbiased arrays (pm , m)p (consisting of all m-tuples) yields the following: 538 Jă urgen Bierbrauer and Holger Schellwat Theorem For every natural number m and every rational number < < with denominator pm we can construct an array (p2m , m pm )p with bias ≤ In particular Theorem yields yet another proof for the parameters from Theorem and from the WCU construction Our Theorem is much more general In order to obtain essential improvements on Theorem let us consider a recursive application Apply Theorem with a Reed-Solomon code [pm , Rpm , (1 − R)pm ]pm , where R = /2 and an /2−biased (4m2 / , m)p −array We obtain the following: Theorem We can construct arrays (4m2 pm / , m pm /2)p , which are −biased The choice m = pj , = p−e yields fp (pj +j −e−1, e) ≤ pj +2j +2e+2 Theorem states in particular fp (b, e) ≤ b + 3e + j + 3, where j ∼ log(b + e) In the binary case this is very close to Theorem and it yields an essential improvement over Theorem when b > e Example Apply Theorem to a p4 −ary Reed-Solomon code of dimension p3 (relative minimum distance > − (1/p)) and an inner array (p2 , 4)p , which is (1/p)−biased Such an array follows from the WCU construction We can describe it as follows: Its rows are (x, y, xy, x2 + cy ), where x, y ∈ IFp and c is a non-square The result is an − biased (p6 , 4p3 )p − array, p which is better than what results from the WCU construction Example In the same style apply Theorem to a pm −ary Reed-Solomon code of dimension pm−1 and an (1/p)−biased array (m2 p2 , m)p , whose existence is guaranteed by Theorem We obtain an − biased (m2 pm+2 , mpm−1 )p − array p This is much better than a corresponding WCU-array Theorem with the same bias and the same number of columns would use m2 p2m /4 rows So far the only ingredients used in our constructions have been Reed-Solomon codes Next we want to show that algebraic-geometric codes can be used to great advantage Let us start by pointing out that many important classes of algebraicgeometric codes can be just as efficiently implemented as Reed-Solomon codes In the next section this is exemplified in the case of the Hermitian codes Hermitian Codes for the User We describe how to obtain generator matrices for the Hermitian codes Consider the field extension IFq2 | IFq and the corresponding trace tr and norm N, where tr(x) = x + xq , N (x) = xq+1 Our codes are defined over IFq2 and have length q (see [24]) Almost Independent and Weakly Biased Arrays 539 The coordinates are parametrized by the pairs (α, β), where N (α) = tr(β) So we need to calculate traces and norms of all elements in the field and to list all these pairs in some order There are q such pairs The general build-up: We construct a (q − g, q )− matrix G with entries from IFq2 Here g = 2q The first k rows of G generate the k−dimensional Hermitian code It has parameters [q , k, q − k + − g]q2 The pole-order test: For n = 0, 1, 2, we have to decide if n is a pole-order or not If n is a pole-order we determine its coordinate vector (i, j) This is done as follows: Let r be the remainder of n mod q, where ≤ r ≤ q − and −s the (negative) remainder of n mod q + 1, where ≤ s ≤ q Then n is a pole-order if and only if x= n−r n+s ≥ = y q q+1 If n ≥ 2g, then the pole-order test does not need to be performed Every such number is a pole-order If n is a pole-order, then n = (q + 1)i + qj, where i = (x − y)q + r, j = s The coordinate vector of n is (i, j) Constructing the rows of G : Let u1 = 0, u2 = q, u3 = q + be the first pole-orders If uk has coordinate-vector (i, j), then the entry of row k of G in coordinate (α, β) is β i αj We conclude that the use of Hermitian codes requires the usual field arithmetic, just as Reed-Solomon codes Using Hermitian and Suzuki Codes Use Theorem with the Hermitian codes as ingredients, q = pm The codes have parameters [p3m , k, p3m − (k + p2m /2)]p2m Use as inner arrays the unbiased arrays (p2m , 2m)p Choose e ≤ m and k ∼ p3m−e − p2m /2 With this choice the resulting array has bias ≤ p−e As we have an array (p5m , 2km)p and logp (2km) ∼ 3m − e + logp (m) it follows fp (3m − e + logp (m), e) ≤ 5i, where m ≥ e Let now e and b be given, where b ≥ 2e Determine m ≥ e such that b + e = 3m (provided b + e is a multiple of 3) We have seen that fp (b, e) ≤ 5m = 53 (b + e), which clearly represents an improvent on Theorem and on the W CU −construction If b < 2e, then fp (b, e) ≤ fp (2e, e) ≤ 5e, still an improvement upon Theorem when b ≥ 32 e The Suzuki codes in characteric (see [12]) have parameters [24f +2 , 2j , 24f +2 − (2j + 23f +1 )]22f +1 Use Theorem with an unbiased array as inner array If f ≥ e and j = 4f − e + we obtain ≤ 2−e , and hence f2 (4f − e + 1, e) ≤ 6f + This presupposes b + e = 4f + > 4e, hence b > 3e 540 Jă urgen Bierbrauer and Holger Schellwat Theorem 10 The Hermitian codes show fp (b, e) ≤ (b + e) if b ≥ 2e The Suzuki codes show f2 (b, e) ≤ (b + e) + if b > 3e The results of Theorem 10 are superior to all the constructions discussed earlier, for the parameter range when Theorem 10 applies The strength of Theorem is its universality and simplicity For b < e it seems to be hard to obtain improvements upon the WCU-construction Another construction principle for weakly biased arrays, first introduced in [18], uses expander graphs and asymptotically nontrivial families of codes as ingredients However, this construction seems to work best when k is large with respect to 1/ (b large with respect to e) and it cannot improve upon the results presented above in that parameter range We conclude this section with an application of Theorem to Hermitian codes The p2 −ary Hermitian code of dimension k ∼ p2 /2 has relative minimum weight δ = − 1/p The unbiased (p2 , 2)p −array yields an (1/p)−biased (p5 , p2 )p −array Example For every odd prime p we can produce an (1/p)−biased (p5 , p2 )p −array by applying Theorem to a Hermitian code and an unbiased array Observe that the WCU construction when applied in the case of p5 rows and = 1/p produces a number of columns of the order of magnitude p3/2 Construction of Authentication Schemes Unconditional authentication was originally introduced by Simmons [22,23] An (n, k)q −array is −almost strongly universal2 (ASU2 ) if each column has bias and for any two different columns c, c and any entries e, e the conditional probability P r(ci = e | ci = e ) is bounded by , where the probability refers to a choice of a row i according to the uniform distribution of rows In the application rows are keys, columns are source states and entries are authentication tags A composition construction based on codes is used in [4,2,3] In [13] a direct link is established between the WCU construction of weakly biased arrays and ASU2 arrays We generalize this construction as follows: Theorem 11 If there is an −biased (n, k)p −array then for every t ≤ k there is an −ASU2 array (pt n, pk )pt , where = p−t + Proof Let C be the linear [n, k]p -code generated by the columns of the -biased array The columns of the ASU2 −array A are indexed by f ∈ C, the rows are indexed by tuples (i, α1 , , αt ), where i is a coordinate of C and αr ∈ IFp It is easy Almost Independent and Weakly Biased Arrays 541 to see that we can find linear mappings Mr : C −→ C, r = 1, , t such that every nontrivial IFp −linear combination of the Mr is non-singular Define the entry of A in row (i, α1 , , αt ) and column f as (M1 (f )(i) + α1 , , Mt (f )(i) + αr ) It is obvious that each column of A is unbiased Let f, g be different columns and (βr ), (γr ) be two entries Let ν be the number of rows i of the original array such that Mr (f − g)(i) = βr − γr for all r We have to show that ν/n ≤ p−t + This follows from Theorem and the linear independence of the Mr (f − g) We see that via Theorem 11 essential improvements upon the parameters of weakly biased arrays yield improved authentication2 codes Example Continuing from Example we obtain (p6 , pp )p arrays, which are (2/p)−ASU2 Not surprisingly this is better than the constructions from [4,13] based on Reed-Solomon codes and it reproduces the parameters of the construction from [2] based on Hermitian codes Example An application of Theorem 11 to the arrays from Example prom−1 )p , which are (3/p)−ASU2 duces arrays (m2 pm+3 , pmp A refinement of the theory of unconditional authentication is introduced in [16] An (N, m)p −array is (δ, t)−almost strongly universal (short (δ, t)−ASU) if for every set U = U0 ∪ {u} of t columns and every a ∈ IFt−1 p , x ∈ IFp the frequencies νU0 (a ) and νU (a , x) satisfy | νU (a , x)/νU0 (a ) |≤ δ The idea is to use the same key for t subsequent messages while still bounding the opponent’s probability of success The link between almost independent arrays and (δ, t)−ASU codes has been established in [16] (and is almost obvious): Theorem 12 A t-wise − dependent array is (δ, t)−ASU, where δ = (p−t + )/(p−(t−1) − ) The following theorem generalizes the method used in [16] Theorem 13 Let f2 (b, lt) ≤ a Then there is an (2−(l−1) , t)−ASU with 2l entries, 2b /(lt) − log2 (l) source bits and a key bits Proof A BCH-code [2j , 2j − ljt, lt + 1]2 , where jlt = 2b , yields an lt−wise 2−e −biased array (2a , 2j ) By Theorem this yields an array (2a , 2j )2l , which l is t−wise 2−lt −biased Apply Theorem 12 We obtain δ < 2/(2l − 1) ∼ 2−(l−1) The number of rows is still 2a Resiliency A number of interesting applications of the WCU construction are in [16] They can all be generalized to admit the use of arbitrary weakly biased arrays We consider the case of almost resilient functions The construction from [16] is an application of Theorem to check matrices of binary BCH codes A straightforward generalization is as follows: 542 Jă urgen Bierbrauer and Holger Schellwat Theorem 14 Assume the following exist: – A systematic −biased (2t , s)2 −array, and – a linear code [m, m − s, k + 1]2 Then there exists a systematic k−wise −dependent (2t , m)2 −array The proof is similar to the proof for the special case used in [16] The end m−t product of Theorem 14 allows the construction of a function : IFm such −→ IF2 that whenever k of the input parameters are fixed the output is close to being unbiased (for details see [16]) Note that the study of almost resilient functions can be motivated from an analysis of the wire-tap channel of type II [20] A discussion of that aspect is in [26], where the close link to the coding-theoretic and geometric notion of generalized Hamming weights is pointed out 10 Conclusion The concepts of sample spaces which are statistically close to being unbiased or independent is fundamental for large areas of computer science and cryptology The best known constructions all yield very similar parameters The various constructions from [1] excel by their simplicity and universality, whereas the WeilCarlitz-Uchiyama construction yields slightly better parameters In this paper we used several new coding-theoretic construction procedures to obtain essential improvements for vast parameter ranges These improvements can already be obtained by restricting the ingredients to Reed-Solomon codes Algebraicgeometric codes produce further improvements in suitable parameter ranges We pointed out that Hermitian codes, a particularly useful class of AG codes, are just as efficiently computable as Reed-Solomon codes In the applications we concentrated on universal hashing, unconditional authentication and almost resilient functions A large number of applications are documented in the literature It is expected that more applications will be discovered References Alon, N., Goldreich, O., H˚ astad, J., Peralta, R.: Simple constructions of almost kwise independent random variables, Random Structures and Algorithms (1992), 289-304, preliminary version: Symposium 31st FOCS 1990, 544-553 Bierbrauer, J.: Universal hashing and geometric codes, Designs, Codes and Cryptography 11 (1997), 207-221 Bierbrauer, J.: Authentication via algebraic-geometric codes, in: Recent Progress in Geometry, Supplemento Rendiconti del Circolo Matematico di Palermo 51 (1998), 139-152 Bierbrauer, J., Johansson, T., Kabatiansky, G., Smeets, B.: On families of hash functions via geometric codes and concatenation, Proceedings CRYPTO 93, Lecture Notes in Computer Science 773 (1994), 331-342 Bierbrauer, J., Schellwat, H.: Weakly biased arrays, almost independent arrays and error-correcting codes, submitted for publication in the Proceedings of AMSDIMACS Almost Independent and Weakly Biased Arrays 543 Boyar, J., Brassard, G., Peralta, R.: Subquadratic zero-knowledge, JACM 42 (1995), 1169-1193 Brassard, G., Cr´epeau, C., Santha,M.: Oblivious transfers and intersecting codes, IEEE Transactions on Information Theory 42 (1996), 1769-1780 Carlitz, L., Uchiyama, S.: Bounds for exponential sums, Duke Mathematical Journal 24 (1957), 37-41 Cohen, G D., Z´emor, G.: Intersecting codes and independent families, IEEE Transactions on Information Theory 40 (1994), 1872-1881 10 Gal, A.: A characterization of span program size and improved lower bounds for monotone span programs, Proceedings 13th Symposium of the Theory of Computing (1998), 429-437 11 Carter, J L., Wegman, M N.: Universal Classes of Hash Functions, J.Computer and System Sci 18 (1979), 143-154 12 Hansen, J P., Stichtenoth, H.: Group codes on certain algebraic curves with many rational points, AAECC (1990), 67-77 13 Helleseth, T., Johansson, T.: Universal hash functions from exponential sums over finite fields and Galois rings, Lecture Notes in Computer Science 1109 (1996), 31-44 (CRYPTO 96) 14 Justesen, J.: A class of asymptotically good algebraic codes, IEEE Transactions on Information Theory 18 (1972), 652-656 15 Katsman, G L., Tsfasman, M A., Vladut, S G.: Modular curves and codes with a polynomial construction, IEEE Transaction on Information Theory 30 (1984), 353-355 16 Kurosawa, K., Johansson, T., Stinson, D.: Almost k-wise independent sample spaces and their cryptologic applications, Lecture Notes in Computer Science 1233 (1997), 409-421 (Advances in Cryptology, Eurocrypt 97) 17 Lu, C J.: Improved pseudorandom generators for combinatorial rectangles, Proceedings of the 25th International Colloquium on Automata, Languages and Programming (1998), 223-234 18 Naor, J., Naor, M.: Small-bias probability spaces: efficient constructions and applications, SIAM Journal on Computing 22 (1993), 838-856, preliminary version: Proceedings STOC 1990, 213-223 19 Naor, M., Reingold, O.: On the construction of pseudo-random permutations: Luby-Rackoff revisited, Proceedings STOC 29 (1997), 189-199 20 Ozarow, L H., Wyner, A D.: Wire-Tap Channel II, AT&T Bell Laboratories Technical Journal 63 (1984), 2135-2157 21 Shen, B Z.: A Justesen construction of binary concatenated codes that asymptotically meet the Zyablov bound for low rate, IEEE Transactions on Information Theory 39 (1993), 239-242 22 Simmons, G J.: A game theory model of digital message authentication, Congressus Numerantium 34 (1992), 413-424 23 Simmons, G J.: Authentication theory/coding theory, in: Advances in Cryptology, Proceedings of Crypto 84, Lecture Notes in Computer Science 196 (1985), 411-431 24 Stichtenoth, H.: Algebraic function fields and codes, Springer 1993 25 Wegman, M N., Carter, J L.: New Hash Functions and Their Use in Authentication and Set Equality, J.Computer and System Sci 22 (1981), 265-279 26 Wei, V K.: Generalized Hamming weights for linear codes, IEEE Transactions on Information Theory 37 (1991), 1412-1418 27 Zyablov, V V.: An estimate of the complexity of constructing binary linear cascade codes, Problems in Information transmission (1971), 3-10 Author Index Abadi, Mart´ın 353 Abe, Masayuki 271 Ateniese, Giuseppe 255 Lee, Sang Jin 166 Lenstra, Arjen K Lindell, Yehuda 36 Beimel, Amos 55 Beth, Thomas 287 Biehl, Ingrid 131 Bierbrauer, Jă urgen 533 Black, John 197 Boneh, Dan 236 Maitra, Subhamoy 515 Malkin, Tal 55 Meyer, Bernd 131 Micali, Silvio 74 Mitchell, Chris J 184 Mă uller, Volker 131 Cachin, Christian 93 Camenisch, Jan 93, 255 Chan, Agnes 501 Cheon, Jung Hee 166 Coppersmith, Don 184 Coron, Jean-S´ebastien 229 Naor, Moni 236 Nielsen, Jesper Buus Damg˚ ard, Ivan 432 Desai, Anand 359, 394 Dodis, Yevgeniy 74, 112 Fischlin, Marc 413 Fischlin, Roger 413 Garay, Juan A 333 Geiselmann, Willi 287 Gennaro, Rosario 469 Gisin, Nicolas 482 Grassl, Markus 287 Halevi, Shai 112 Han, Jae Woo 166 Ishai, Yuval 55 Jaulmes, Eliane 20 Johansson, Thomas 300 Jă onsson, Fredrik 300 Joux, Antoine 20 Joye, Marc 255 Kang, Ju-sung 166 Knudsen, Lars R 184 Ko, Ki Hyoung 166 Okamoto, Tatsuaki 432 147, 271 Park, Choonsik 166 Pinkas, Benny 36 Rabin, Tal 112 Ramzan, Zulfikar 376 Reyzin, Leonid 376 Rogaway, Phillip 197 Rosen, Alon 451 Safavi-Naini, Reihaneh 316 Sarkar, Palash 515 Schellwat, Holger 533 Semanko, Michael 216 Staddon, Jessica 333 Steinwandt, Rainer 287 Tanaka, Keisuke 147 Tsudik, Gene 255 Uchiyama, Shigenori Verheul, Eric R Wang, Yejing 316 Wolf, Stefan 482 Wool, Avishai 333 Zhang, Muxiang 501 147 ... Bibliothek - CIP-Einheitsaufnahme Advances in cryptology : proceedings / CRYPTO 200 0, 20th Annual International Cryptology Conference, Santa Barbara, California, USA, August 20 - 24, 200 0 Mihir Bellare... Bellare (Ed.) Advances in Cryptology – CRYPTO 200 0 20th Annual International Cryptology Conference Santa Barbara, California, USA, August 20- 24, 200 0 Proceedings 13 Series Editors Gerhard Goos,... Bellare Program Chair, Crypto 200 0 CRYPTO 200 0 August 20? ??24, 200 0, Santa Barbara, California, USA Sponsored by the International Association for Cryptologic Research (IACR) in cooperation with IEEE