Probabilistic Analysis of Cellular Automata Rules and its Application in Pseudo Random Pattern Generation

The present work is an extension of the work that appeared in the article titled “Pseudorandom Pattern Generation by a 4-Neighborhood Cellular Automata (4NCA) using a Probabilistic Analysis”. In this paper we propose a probabilistic analysis based technique for selecting good CA rules that can be used in pseudo random pattern generation. The proposed technique is applied on a CA of neighborhood four to construct one dimensional, non-uniform 4NCA random number generators. Another set of 4NCA random number generators were evolved using Cellular Programming (CP), a technique proposed by Tomassini and Sipper. A comparison is made between the pseudo random patterns generated by the proposed method and those obtained using CP. The results show that our approach outperforms CP both in terms of average time taken to evolve CA rules and in terms of quality of pseudo random patterns generated. The proposed approach is also shown to be better than the common generators such as Shift Register, Congruential Generator and Lagged Fibonacci Generator.