Controller Placement in SDN with Low Latency Using Meta-heuristic Algorithms

Abstract

Software-Defined networks (SDNs) are a new generation of computer networks that have eliminated many of the problems of traditional networks. These networks use a three-tier architecture in which the physical layers, controller, and management are located at different levels. This new architecture has made the network very dynamic, and many of the previous problems in the network have been solved. As the size of the network increases, using a controller across the network will cause issues such as increasing the average latency between the switches and the controller, as well as forming a bottleneck in the controller. For this reason, it is recommended to use multiple physical controllers on the control plane. Due to the cost of purchasing and maintaining the controller, it is necessary to solve the mentioned problem with the least controllers. The question is, to achieve a goal such as reducing latency to an acceptable threshold, at least how many controllers are needed, where the controllers should be located, and which switches should be monitored by which controller? Since this is an NP-Hard problem, methods based on meta-heuristic algorithms can be effective in solving it. In this article, we have solved the problem of controller placement in software-based networks to reduce latency using the cuckoo meta-heuristic algorithm. The simulation results show that the efficiency of our proposed method is between 16 to 70 percent better than the method proposed by the PSO algorithm.