摘要
自相关检测是一种用以检测一个长度为n的二元序列与其左移d位后序列的关联程度的随机性检测算法。d的选择范围很大,对所有参数逐一进行检测不现实,需要研究检测参数之间的关系。定义了检测参数之间可能存在的3种关系,以分组长度为m的分组密码随机性检测为对象,综合考虑分组密码和自相关检测的特点,利用统计实验研究了自相关检测参数子集D={1,2,m/4,m/2,3m/4,m,2m}中参数的关系。研究结果表明,对分组密码进行自相关检测时,检测参数应该首选d=m。该方法和结果为研究其他类型密码算法的随机性检测参数选择提供了新思路。
Autocorrelation test was a statistical test to evaluate the correlation between one sequence and the corresponding non-cyclic left-shifted d bits sequence. It was impractical to adopt all the values of d since its range was often very wide. Three relations between parameters of randomness test were defined firstly. Then the relationships among the subclass D={ 1,2,m/4,m/2,3m/4,m,2m}of autocorrelation test for block cipher were studied by statistical experiments, where m was block length. The experiments show that the prefer choice of parameter d is m when doing autocorrelation test for block cipher. The method is also available for parameter selection of other randomness test for other types of cryptosystem.
出处
《通信学报》
EI
CSCD
北大核心
2009年第7期86-90,共5页
Journal on Communications
基金
国家自然科学基金资助项目(60503014
60603013)
国家高技术研究发展计划("863"计划)基金资助项目(2007AA01Z470
2008AA01Z417)
北京市自然科学基金资助项目(4072026)~~
关键词
信息安全
分组密码
统计检测
自相关检测
参数选择
information security
block cipher
statistical test
autocorrelation test
parameter selection
