Improved fast correlation attacks on stream ciphers via convolutional codes

Abstract

This paper describes new methods for fast correlation attacks, based on the theory of convolutional codes. They can be applied to arbitrary LFSR feedback polynomials, in opposite to the previous methods, which mainly focus on feedback polynomials of low weight. The results improve signifficantly the few previous results for this general case, and are in many cases comparable with corresponding results for low weight feedback polynomials.