On the placement of controllers for designing a wide area software defined networks

The newly emerged Software Defined Networks (SDN) can apply to Wide Area Network (WAN) for network control and management. To achieve this, a logically centralized but a physically distributed control architecture is normally required. This type of control architecture consists of multiple controllers. For maintaining the global consistency of the network, controller placement is a key issue in a Wide Area SDN. The controller placement problem (CPP) refers to selecting the number and proper positions of the controllers to improve the performance of the SDN control plane. Although various approaches are made to solve CPP for the small and medium-size network, still it requires an alternative solution for a large scale network like WAN. It needs enough time to select the number of controllers and their location, within a limited resource constraint. In this work, we propose two stochastic meta-heuristic techniques for finding optimal locations of the controllers that optimize the latency between the switch to a designated controller. We develop the algorithm for Particle Swarm Optimization and Firefly that solve CPP, with the fitness function as the controller to switch latency, which needs to be minimized. The extensive simulation shows that Firefly provides a better result in terms of computation time and cost that are close to optimal.