A Comparative Analysis of Bioinspired Algorithms for Solving the Problem of Optimization of Circulant and Hypercirculant Networks

Abstract

The solution of the optimization problem of constructing regular networks (graphs) that are optimal over the average diameter is investigated. Two classes of parametrically described regular networks are investigated - circulant and hypercirculant networks. An approach using bioinspired algorithms for the automatic synthesis of parametric descriptions of optimal circulant and hypercirculant networks has been developed. A comparative analysis of five different bioinspired algorithms (genetic algorithm, differential evolution, particle swarm optimization, algorithm of artificial bee colony and firefly algorithm) was carried out using them to solve this optimization problem. For the found optimal networks structural characteristics such as diameter, average diameter, bandwidth, reliability were obtained, and these networks were compared according to these characteristics.