Hide/Show Apps

On statistical analysis of synchronous stream ciphers

Download
2008
Sönmez Turan, Meltem
Synchronous stream ciphers constitute an important class of symmetric ciphers. After the call of the eSTREAM project in 2004, 34 stream ciphers with different design approaches were proposed. In this thesis, we aim to provide a general framework to analyze stream ciphers statistically. Firstly, we consider stream ciphers as pseudo random number generators and study the quality of their output. We propose three randomness tests based on one dimensional random walks. Moreover, we theoretically and experimentally analyze the relations of various randomness tests. We focus on the ideas of algebraic, time memory tradeoff (TMTO) and correlation attacks and propose a number of chosen IV distinguishers. We experimentally observe statistical weaknesses in some of the stream ciphers that are believed to be secure.