随机序列的扑克检测优化研究

PDF下载HTML

Title:

Study on optimization of poker test random sequences

作者:

杨先伟¹, 康红娟², 廖祖华^3,4

1. 无锡职业技术学院 基础部, 江苏 无锡 214121;
2. 四川长虹电器股份有限公司, 四川 成都 610041;
3. 江南大学, 江苏 无锡 214122;
4. 江南大学 智能系统与网络计算研究所, 江苏 无锡214122

Author(s):

YANG Xianwei¹, KANG Hongjuan², LIAO Zuhua^3,4

1. Department of Fundamental Courses, Wuxi Institute of Technology, Wuxi 214121, China;
2. Sichuan Changhong Electric Co., Ltd., Chengdu 610041, China;
3. School of Science, Jangnan University, Wuxi 214122, China;
4. Institute of Intelligence System & Network Computing, Jiangnan University, Wuxi 214122, China

关键词:

二元序列; 随机序列; 随机数发生器; 随机性检测; 扑克检测; 密码算法; 效率分析; 余不完全伽玛函数

Keywords:

binary sequence; random sequence; pseudorandom bit generator; randomness test; poker test; encryption algorithms; efficiency analysis; incomplete gamma functions.

分类号:

TP18

DOI:

10.11992/tis.201606002

摘要:

现代计算机系统的安全性依赖于二元随机序列，随机性检测利用概率统计方法对二元序列的随机性进行分析测试。我国国家密码管理局发布了随机性检测规范，扑克检测为其中一个检测项。本文通过充分分析扑克检测效率不高的原因有针对性地提出一种新的快速实现算法，优化算法充分利用CPU字长一次处理多个比特，将m为4和8的情况整合在一起，减少不必要的处理流程。同时精简并优化统计量的计算和判断过程，避免余不完全伽马函数的计算。分析和实验的结果表明该优化算法可以使得扑克检测的速度提升9.5倍左右。

Abstract:

The security of modern computer systems depends on binary random sequences, such as cipher algorithms keys, RSA algorithm prime numbers, the digital signature system, the identity authentication system, etc. Randomness tests analyze and test the randomness of sequences, using probability and statistics. The Chinese National Cryptography Administration has released national randomness test specifications and the Poker test is one of these. This paper analyzed the reasons for the low efficiency of the Poker test, then proposes a fast implementation algorithm. This new algorithm deals with bytes by making full use of CPU word length, integrates the detection process, and reduces some unnecessary operations under the conditions when m equals 4 and 8. At the same time, the method reduces and optimizes the computation and assessment of statistical quantity, avoiding computation of incomplete gamma functions. The results show that the efficiency of the new algorithm increases 9.5 fold.