The Testing of Pseudorandom Generators Using Correlation Properties

Abstract

The statistical properties of a pseudorandom sequence being characterized by the distribution of the values near to the uniform distribution may differ from the expected ones. On doing a quantitative evaluation of the differences, it is advantageous to start with the properties of an ideally random sequence. The results of the study of the absolute value of autocovariance sequence point show that the estimate is characterized by the unilateral normal distribution with the parameters dependent on the sequence length. Thanks to this it is possible to determine the criteria that enable the relatively simple testing of the behaviour of pseudorandom sequence generators. These criteria have been verified on the four generators.