GOP-SDN: an enhanced load balancing method based on genetic and optimized particle swarm optimization algorithm in distributed SDNs

Abstract

One of the biggest challenges of distributed software defined networks (SDNs) is to create load balancing on controllers to reduce response time. Although recent studies have shown that switch migration is an efficient method for solving this problem, inappropriate decision making in selecting the target controller and the high number of switch migrations among controllers caused a decrease of throughput with an average increase in response time of the network. In the proposed method, named GOP-SDN, in first place, using a variable threshold based on controllers, the congestion or imbalance of the load is detected. Subsequently, regarding the capacity of controllers and switches connected to them and using the intelligent combination of genetic algorithm and OPSO, GOPS-SDN tried to choose the best controller with appropriate capacity to migrate. In other words, using genetic algorithm with the highest fitness and then the OPSO algorithm and using the speed of each particle to move to the best overall and best locations, the best solution is calculated from the particle imported into PSO. In parallel with the implementation of the PSO algorithm, GOSP-SDN used the same algorithm to compute the best weights for each particle in the algorithm (OPSO). Therefore, the best and optimal solution among the particles to migrate to the controller is found. The results of the implementation and evaluation of GOP-SDN in the Cbench simulator and Floodlight controller showed improvement of 24.72% in throughput and the number of migration has been reduced by 13.96%.