MD-4原象攻击的分析与改进

Analysis and improvement of preimage attack on MD-4

首先分析了Leurent提出的MD-4原象攻击方法,该方法利用MD-4布尔函数的吸收性质,迭代函数的可逆性以及消息扩展方式的特殊性,首先形成伪原象攻击,之后利用基于树的方法将伪原象转变为原象攻击。采用随机图的方法,对其后一部分进行了改进,提高了攻击效率,将复杂度从2102降低到298。

The preimage attack of MD-4 is discussed.Leurent’s method explores chiefly three properties of MD-4：The absorption properties of boolean function,reversibility of the iterative process and specificity of message expansion.After their pseudo preimage attack a tree-based method is followed to transform pseudo preimage to true preimage.Tree-based method is replaced with random graph method.The complexity of the attack reduces to 2 ^98 from 2^ 102 .