Towards controller placement problem for software-defined network using affinity propagation

This Letter investigates multi-controller placement problem for software-defined network aiming at minimising the propagation latency between switches and associated controllers. Distinct from the standard clustering-based partition approaches, a modified exemplar-based clustering method based on affinity propagation is presented to address the problem. Instead of specifying beforehand the number and locations of controllers, this method can learn the optimal number and locations of controllers adaptively according to network topology. Extensive simulation results demonstrate that the proposed method can provide more stable and accurate controller placement results in terms of latency minimisation compared with k-median and k-centre methods.