A high-quality pseudorandom numbers generator based on twi-layer couple cellular automata

This paper proposes a new class of cellular automata, twi-layer couple cellular automata (TLCCA), with specific application to pseudorandom number generation. TLCCA consists of two layer each of which is a one dimensional CA. Two different rules are selected in the lower-layer CA on account of hybrid CA had more complex behavior. The upper-layer CA is divided into two parts. These two parts have a novel neighbourhood, which called couple-structure neighbourhood. By this neighbourhood, two parts in upper layer interplay with each other. ENT test suites are adopted to test the randomness of PRNG. In order to find a stable PRNG, entropy, chi-square and serial correlation coefficient and their variability need to be considered. So a multi-objectives optimization algorithm is proposed. The results of experiment indicate that the TLCCA PRNG can obtain credible random number using no less than 48 cells. The merits of TLCCA PRNG are simpler structure, higher efficiency and better robusticity.