Modified multi-objective TLBO for location of controllers in software defined networks

Software Defined Network is a new idea that enables administrator/operator to build a highly automated and manageable network. However, this architecture encounters various challenges such as scalability and fault tolerant. Multiple controllers are often required to alleviate these challenges. Nonetheless, the deployment of a desired number of controllers influence various metrics that may be conflicting together. Therefore, based on the fact that various types of objectives should be taken into consideration, this matter regarded as a multiobjective combinatorial optimization problem (MOCO). A particular efficient method to solve a typical MOCO, which is used in the relevant literature, is to find the actual Pareto frontier first and give it to the decision maker to select the most appropriate solution(s). However, this problem when applied for large sized or dynamic networks, behaves as a NP-hard problem, therefore, use of heuristic approaches are required. In this study, a heuristic algorithm called Modified Multi-Objective Teaching Learning Based Optimization (MMOTLBO) is introduced to solve the problem. Efficiency of the algorithm is tested using real network topologies from Internet Topology Zoo. Obtained results prove that the algorithm has superior performance from efficiency and computation time point of views comparing to the previous studies.