A Novel Bi-Velocity Particle Swarm Optimization Scheme for Multicast Routing Problem

Abstract

A nondeterministic polynomial (NP) with complete Multicast routing problem is defined using a bi-velocity particle swarm optimization (BVDPSO) is proposed in this paper. The shift of particle swarm optimization to the discrete or binary domain, stepping away from the continuous domain is the major impact of the work. Initially a bi-velocity strategy is built such that it characterizes each dimension in terms of 0 and 1. The basic function of this strategy is to describe the MRP’s binary characteristics such that 0 stands for the node not being selected while 1 stands for selection. Based on the location and velocity of the original PSO in the continuous domain, the BVDPSO is updated. This will preserve the global search ability and fast convergence speed of the original PSO. 58 instances of large, medium and small scales are used for experimentation in the OR-Library. Based on the results, it is identified that it is possible to get near-optimal or optimal solutions for BVDPSO as it requires generation of limited multicast trees. This approach is found to be optimal over its peers and outperforms recent heuristic algorithms and many advanced techniques used for the MRP problem. They also outperform several PSO, ant colony optimization and genetic algorithms.