A Study of Swarm Topologies and Their Influence on the Performance of Multi-Objective Particle Swarm Optimizers

It has been shown that swarm topologies influence the behavior of Particle Swarm Optimization (PSO). A large number of connections stimulates exploitation, while a low number of connections stimulates exploration. Furthermore, a topology with four links per particle is known to improve PSO’s performance. In spite of this, there are few studies about the influence of swarm topologies in Multi-Objective Particle Swarm Optimizers (MOPSOs). We analyze the influence of star, tree, lattice, ring and wheel topologies in the performance of the Speed-constrained Multi-objective Particle Swarm Optimizer (SMPSO) when adopting a variety of multi-objective problems, including the well-known ZDT, DTLZ and WFG test suites. Our results indicate that the selection of the proper topology does indeed improve the performance in SMPSO.