Multi-Agent Genetic Algorithm for One Criteria Network Routing Optimization

Network design is widely used in practice in an ever wider range of applications. The shortest path models is one of the core models of network design and as we know well in one of NP-complete Problem in network design. This paper presents a multi-agent genetic algorithm to the shortest path routing problem, this algorithm is named MAGA-Rout. Priority-based encoding and decoding have been used for encoding the chromosome (strings). In this algorithm, we design a variant of weight mapping crossover operator based on neighborhood so as to obtain useful information from its neighbors and avoid random recombination. Results after experimentation for a sample test network have been presented to demonstrate the capabilities of the proposed approach to generate a much better quality of solution (route optimality) and much higher rate of convergence than other algorithms.