Optimal controller placement in large scale software defined networks based on modified NSGA-II

摘要

Software Defined Network (SDN) is an emerging approach to overcome challenges of traditional networks. One particularly important issue in SDN architectures is that of controller placement problem (CPP), i.e., deploying a desired number of controllers within a network while some possibly conflicting requirements have to be fulfilled. A single optimal placement may not be possible and decision makers need to seek for an appropriate trade-off among the metrics. Although an exhaustive evaluation of all possible placements can be practically performed well for small and medium-sized networks, regarding realistic time and resource restrictions, heuristic approaches are required for large-scale networks. Hence, a heuristic called Multi-Start Hybrid NSGA-II (MHNSGA-II) is introduced which yields faster computation times and needs reasonable memory to perform. The obtained results on several topologies extracted from Internet Topology Zoo showed the efficiency of the proposed approach.