Improving algebraic attacks on stream ciphers based on linear feedback shift register over $\mathbb {F}_{2^k}$

Abstract

In this paper we investigate univariate algebraic attacks on filter generators over extension fields $\mathbb {F}_q=\mathbb {F}_{2^n}$ with focus on the Welch–Gong (WG) family of stream ciphers. Our main contribution is to reduce the general algebraic attack complexity on such cipher by proving new and lower bounds for the spectral immunity of such ciphers. The spectral immunity is the univariate analog of algebraic immunity and instead of measuring degree of multiples of a multivariate polynomial, it measures the minimum number of nonzero coefficients of a multiple of a univariate polynomial. In particular, there is an algebraic degeneracy in these constructions, which, when combined with attacks based on low-weight multiples over $\mathbb {F}_q$, provides much more efficient attacks than over $\mathbb {F}_2$. With negligible computational complexity, our best attack breaks the primitive WG-5 if given access to 4 kilobytes of keystream, break WG-7 if given access to 16 kilobytes of keystream and break WG-8 if given access to half a megabyte of keystream. Our best attack on WG-16 targeted at 4G-LTE is less practical, and requires $2^{103}$ computational complexity and $2^{61}$ bits of keystream. In all instances, we significantly lower both keystream and computational complexity in comparison to previous estimates. On a side note, we resolve an open problem regarding the rank of a type of equation systems used in algebraic attacks.