CA based cost optimized PRNG for Monte-Carlo simulation of distributed computation

Abstract

Monte-Carlo (M-C) simulations are widely used today to evaluate the reliability of any distributed system. M-C simulations are generated using high quality pseudo-random numbers. Different pseudo-random number generators (PRNGs) are being used as a source of random patterns for M-C simulation. In this research, we have emphasized a cost effective design methodology for the pseudo-random numbers, using Cellular Automata (CA). We have introduced a behavioural discussion and examined the benefits of our proposed Equal Length Cellular Automata (ELCA) based PRNG. This research is expected to contribute to reducing the overhead use for fault coverage in generation of random integers using Cellular Automata. In the proposed design methodology, a high degree of randomness is maintained in generated patterns, while focusing on the reduction of various associated complexities, like design complexity, time complexity and searching complexity. The result achieved in the experiment reflects the high quality of randomness in the generated patterns for ELCA PRNG over the Maximum Length Cellular Automata (Max CA) PRNG and M-C PRNG.