Adaptive Load Balancing Scheme for Software-Defined Networks Using Fuzzy Logic Based Dynamic Clustering

This paper presents an intelligent load balancing strategy among the controllers at the control plane of software-defined networking (SDN) paradigm. Due to the separation of user plane and control plane in SDN framework, the assignment of switches and the respective controller is a tremendously complex task. The use of a central allocation scheme for dynamic controller assignment in SDN results into a huge amount of data at the central super controller. The processing of this large amount of data presents a trade-off situation between the quality of service (QoS) parameters. This issue has been resolved in this work by proposing a cluster-based approach for a dynamic controller allocation scheme. Anon-uniform intelligent clustering scheme is proposed in this work to provide a robust load distribution technique in the presence of dynamic and uncertain network and traffic environments. The formation of clusters and the cluster head selection is done using fuzzy logic control to incorporate a greater number of parameters for decision making and to enhance the performance in terms of QoS requirements. The theoretical analysis of the optimal load balancing technique for SDN is supported by the experimental analysis. The performance of fuzzy logic-based clustering has shown a significant improvement in the average latency and packet loss as compared to the cluster-less load balancing and conventional cluster-based method for the Open NetworkOperating System (ONOS) controllers.