Pseudorandom numbers generators quality assessment using graphic tests

摘要

In this paper we address testing methodologies of pseudorandom numbers generators (PRNGs) and propose improved approaches which increase testing efficiency. At present time, PRNG quality is usually assessed using two main groups of tests: evaluation tests and graphic tests. When executing an evaluation test on a pseudorandom sequence, the assessment is based on computing specific statistics and its comparison to predefined critical values. Graphical tests are less common, however, in some cases, executing graphic tests allows detecting flaws in the PRNG's design which could be overlooked by common evaluation tests. In this paper, authors review some graphic tests implementations for PRNG statistical properties evaluation.