A general construction method for Pseudo-Random Number Generators based on the Residue Number System

In this paper a general construction method for Pseudo-Random Number Generators (PRNG) is described. The main idea is to give form to a topology similar to a Hopfield neural network in which each neuron is replaced by a small PRNG. The Residue Number System (RNS) is used to perform the involved calculations in each feedback line in order to both introduce non-linearity to the system, and to use binary representations of little number of bits. Different tests are applied to measure the statistical properties of the generated sequence of numbers. These tests are basic statistical tests, the Diehard test, statistical complexity measurements, zipping complexity and the Hurst exponent.