|
Numerical methods and programming, 2015, Volume 16, Issue 1, Pages 61–77
(Mi vmp519)
|
|
|
|
This article is cited in 2 scientific papers (total in 2 papers)
Problems of search for collisions of cryptographic hash functions of the MD family as variants of Boolean satisfiability problem
I. A. Bogachkovaa, O. S. Zaikinb, S. E. Kochemazovb, I. V. Otpushchennikovb, A. A. Semenovb, O. O. Khamisova a Irkutsk State University
b Matrosov Institute for System Dynamics and Control Theory of Siberian Branch of Russian Academy of Sciences, Irkutsk
Abstract:
An implementation of the differential attacks on cryptographic hash functions MD4 (Message Digest 4) and MD5 (Message Digest 5) by reducing the problems of search for collisions of these hash functions to the Boolean satisfiability problem (SAT) is considered. The novelty of the results obtained consists in a more compact (compared to already known) SAT encodings for the algorithms considered and in the use of modern parallel and distributed SAT solvers in applications to the formulated SAT problems. Searching for single block collisions of MD4 in this approach turned out to be very simple. In addition, several dozens of double block collisions of MD5 are found. In the process of the corresponding numerical experiments, a certain class of messages that produce the collisions is found: in particular, a set of pairs of such messages with first 10 zero bytes is constructed.
Keywords:
CDCL (Conflict-Driven Clause Learning), cryptographic hash functions, collisions, differential attacks, Boolean satisfiability problem, SAT, CDCL (Conflict-Driven Clause Learning), parallel SAT solvers.
Received: 18.01.2015
Citation:
I. A. Bogachkova, O. S. Zaikin, S. E. Kochemazov, I. V. Otpushchennikov, A. A. Semenov, O. O. Khamisov, “Problems of search for collisions of cryptographic hash functions of the MD family as variants of Boolean satisfiability problem”, Num. Meth. Prog., 16:1 (2015), 61–77
Linking options:
https://www.mathnet.ru/eng/vmp519 https://www.mathnet.ru/eng/vmp/v16/i1/p61
|
Statistics & downloads: |
Abstract page: | 197 | Full-text PDF : | 154 |
|