Correlation Attack on Stream Cipher Based on Genetic Algorithms

Abstract

Pseudonoise sequences generated by linear feedback shift register (LFSRs) with some nonlinear combining functions have been proposed as running key generators in stream ciphers.Siegnthaler has shown that the number of trials to break these ciphers can be significantly reduced by using correlation methods.This paper considers a new approach to cryptanalysis based on the application of a directed random search algorithm mcalled a Genetic algorithm. It shown that such an algorithm can be used to reduce the number of trials which are needed to determine the initial state and the primitive feedback polynomials of the attacked generator using cipher text only attack.Well known system are taken for the case of study: Pless system [14], Geffe system [6], Bruer system [3],J-k, OR, Multiplying, and police systems [2].