Use of MaSE methodology for designing a swarm-based multi-agent system

Abstract

Swarm intelligence is a rapidly growing branch of artificial intelligence with numerous heuristics being developed. These offer faster and more accurate ways to achieve good solutions. Ant colony optimization (ACO) and particle swarm optimization (PSO) are the most common metaheuristics. Based on the behaviour of collectives of ants, birds, fish and others, they offer algorithms which can be used to solve different NP-hard problems. Multi-agent systems (MAS) offer new perspectives on organisation of code into specific components as autonomic, decentralized or distributed agents. The Multi-agent Systems Engineering (MaSE) methodology allows the combination of both aforementioned approaches. This paper presents research based on the AgentSwarm application which evaluates the efficiency of applying MaSE methodology to solve the traveling salesman problem (TSP) using ACO and PSO metaheuristics. Initial results have shown that MaSE is a comprehensive methodology, solid and reliable in building and developing swarm-based multi-agent systems.