NCPSO: A Solution of the Controller Placement Problem in Software Defined Networks

摘要

Controller placement is an important problem in software defined networks (SDN). Most of the placement strategies focused on propagation latency, but ignored the load of controllers, which is a critical factor in real networks. In this paper, we propose Network Clustering Particle Swarm Optimization Algorithm (NCPSO), which taking into consideration of the load of controllers, propagation latency and load balancing. By generating diverse individuals with high clustering efficiency and overcoming the disadvantages of PSO used in discrete problem, the NCPSO can solve the controller placement problem better than other algorithms. The evaluation shows that the new algorithm can significantly reduce the number of required controllers, reduce the load of the maximum-load controller, and have a good performance of load balance.