基于特征分布函数的分组密码算法差异性分析

Analysis of Differences on Block Cipher Algorithms Based on Characteristic Distribution Function

基于密文的密码体制识别是分组密码分析领域的重要研究方向之一,也是实际背景下展开密码分析的前提保证。建立高效准确的密码体制识别方案,能够为破译密文及恢复密钥提供正确的指导等。以机器学习方法中的特征工程和统计学中的特征分布函数相似度指标为手段和研究方法,证明了在随机情况下一种SPN结构与一种Feistel结构之间存在密文特征分布上的差异性,并加以应用。提出在一般随机条件下区分两种不同结构密码算法的依据与方法,即提取密文相关特征并拟合其分布函数及计算相似度指标的方法,解决了在随机密钥的条件下如何寻找两种不同结构类型密码算法的差异之处。将统计学方法与密码学问题结合起来,为解决随机密钥下唯密文加密算法识别问题提供新的思路。

Cryptosystem identification based on ciphertext is an important research focus in block cryptanalysis,which is also a prerequisite for cryptographic analysis in authentic situation.An efficient cryptographic system identification scheme provides guidance for recovering keys.In this field,the identification schemes of cryptographic algorithms based on machine learning have made very remarkable progress.This paper uses the feature engineering in machine learning and the feature distribution function similarity index in statistics as the methods to prove that there are some differences in the distribution of cipher text features between the SPN structure and Feistel structure under random conditions.Further,the basis for distinguishing two kinds of cryptographic algorithms with different structures under general random conditions is proposed,that is,the method of extracting the relevant features of the ciphertext,fitting its distribution function and calculating the similarity index.This method solves the problem under the condition of random key to find the differences between two different structures of cryptographic algorithms.By combining statistical methods with cryptographic problems,this paper provides new ideas for solving the problem of ciphertext-only encryption algorithm identification under random keys.