Abstract
Random numbers are highly used in applications like computer simulation, gambling, statistical sampling and cryptography. Only a high-entropy random source can produce a better quality of random data. In this paper, we work on the generation of high-quality random numbers based on cellular automata (CA). We propose three-dimensional (3D) CA to generate a random sequence. We also propose various enhancements to CA-based pseudorandom number generator (CA-PRNG) and discuss its five variants by using two-dimensional (2D) and 3D CA. We analyze the switching between 2D and 3D CA for the same sequence to achieve high-quality randomness. We provide the architectures, CA rules and the algorithms for all the proposed variants with respect to 2D and 3D cell structures. We have done a software implementation of all the proposed variants and tested their produced sequences by using the internationally accepted standard for random numbers known as DIEHARD. The analysis of our results is made as per the recommendations of ANSI standards. A very similar criteria are applied to compare all the methods. Based on our results we are able to find the best CA-PRNG for producing high-quality random sequence.