关于伪随机数串统计性检测的研究

Many fields need random and pseudorandom numbers, in particular in cryptographic applications. d-bit segment binary pseudorandom numbers can be easier used for block encryption. This paper studies on the statistical test for binary d string pseudorandom number generator (PRNG). Three postulates on the randomness for ideal pseudorandom d-bit segment sequences have been proposed. Based on the FIPS140-2 tests, a statistical test suite for d-bit segment sequences generated by PRNG has been proposed. Using the test suite tests the 100 key streams generated by RC4 PRNG and Matlab PRNG, respectively. The test results show that in the two 100 key streams generated via the two PRNGs with different seeds，about 97% 8-bit strings have passed the test suite, respectively.