Particle Swarm Optimization for Multiple Multicast Routing Problem

The optimization problem of multiple multicast routing with both bandwidth and delay constraints is more complicated than the multicast routing problem. To get the optimal solution of the multiple multicast routing problem quickly, this paper proposes a particle swarm optimization algorithm based on evolution of tree structure. In the proposed algorithm a particle, as a feasible solution of the problem, is represented as a vector, and the components of the particle are represented by tree structure coding. The flight of particles in the search space is implemented through the evolution of trees. Visual radius of a particle is introduced in the orbicular social structure of particle population to enhance the ability of particle neighborhood learning. The tree structure mutation method is designed to increase the possibility of which the algorithm jumps out of local optima. To the infeasible solutions unsatisfied with constraints in the population, penalty strategy is adopted to penalize the particle and its components according to the situation unsatisfied with constraints. Simulation experiments have been carried out on different network topologies produced by random for networks consisting of 26, 50 and 100 nodes. The results of solving the routings of multiple multicast requests show that the proposed algorithm performs better in searching optimal solution and converging speed.