Improved indifferentiability security bound for the JH mode

Indifferentiability security of a hash mode of operation guarantees the mode’s resistance against all generic attacks. It is also useful to establish the security of protocols that use hash functions as random functions. The JH hash function was one of the five finalists in the National Institute of...

Full description

Saved in:
Bibliographic Details
Published inDesigns, codes, and cryptography Vol. 79; no. 2; pp. 237 - 259
Main Authors Moody, Dustin, Paul, Souradyuti, Smith-Tone, Daniel
Format Journal Article
LanguageEnglish
Published New York Springer US 01.05.2016
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:Indifferentiability security of a hash mode of operation guarantees the mode’s resistance against all generic attacks. It is also useful to establish the security of protocols that use hash functions as random functions. The JH hash function was one of the five finalists in the National Institute of Standards and Technology SHA-3 hash function competition. Despite several years of analysis, the indifferentiability security of the JH mode has remained remarkably low, only at n / 3 bits, while the two finalist modes Keccak and Grøstl offer a security guarantee of n / 2 bits. Note all these three modes operate with n -bit digest and 2 n -bit permutations. In this paper, we improve the indifferentiability security bound for the JH mode to n / 2 bits (e.g. from approximately 171 to 256 bits when n = 512 ). To put this into perspective, our result guarantees the absence of (non-trivial) attacks on both the JH-256 and JH-512 hash functions with time less than approximately 2 256 computations of the underlying 1024-bit permutation, under the assumption that the underlying permutations can be modeled as an ideal permutation. Our bounds are optimal for JH-256, and the best known for JH-512. We obtain this improved bound by establishing an isomorphism of certain query-response graphs through a careful design of the simulators and bad events. Our experimental data strongly supports the theoretically obtained results.
ISSN:0925-1022
1573-7586
DOI:10.1007/s10623-015-0047-9