A Load Balancing Mechanism Based on Non-Cooperative Game in Traffic Environment

Multi-controller resource scheduling problem is crucial in the edge computing. Nonetheless, since the network has to be constantly scaled up to match the increasing application’s diversity and complexity, the original SDN with single controller suffers from the problem of lower processing capacity of the control plane in the edge computing environments. The currently popular solution in the literature tends to deploy multiple controllers to form a distributed control plane, while this method may lead to the load imbalance among controllers, bringing about the large network latency and low resource utilization. In this paper, to balance the load among controllers, we propose ATS-NCGT, which is an Adaptive Traffic Scheduling algorithm based on the Non-Cooperative Game Theory. We consider the load balancing problem of multi-controller as a non-cooperative game according to the interaction characteristics between the switches and the controllers. We also design a hybrid genetic algorithm based on simulated annealing to improve the optimization accuracy and convergence speed in the solving process. The simulation results demonstrate that our scheme is effective in terms of improving the load-balancing rate and lowering the controller response delay, even in the case of high request rate.