基于Feistel网络分组密码的实际安全性研究

Practical Security Research for Block Ciphers with Feistel Networks

密码最大差分特征概率的上界和最大线性特征概率的上界是衡量分组密码抵抗差分密码分析和线性密码分析能力的重要指标。文章根据线性密码分析的串联规则提出了一种新的求取密码最大线性特征概率上界的算法,适用于密钥异或作用下采用Feistel网络的分组密码,指出了原有的评估一类基于混沌函数的广义Feistel密码实际安全性的结论有误,得到了其线性活动轮函数的最小个数和差分活动轮函数的最小个数不总是相等的结论。

The upper bounds of maximum differential characteristic and linear approximation probabilities are an important measure to evaluate the security of block ciphers against differential cryptanalysis and linear cryptanalysis. In this paper,a new method based on the concatenation rules of linear cryptanalysis is proposed for seeking the upper bounds of maximum linear approximation probability for block ciphers, which is especially applicable to block ciphers with Feistel networks that key is XORed with data. we also draw a conclusion that the least of differential active round functions and the least of linear active round functions are not always the same for a class of Generalized Feistel Ciphers based on chaotic maps which was evaluated wrongly.