Multi-Controller Load Balancing Mechanism Based on Improved Genetic Algorithm
Aixin Xu, Shimin Sun, Ze Wang, Xiao-Fan Wang, Li Han · 2022
To solve the load imbalance problem of controllers in Software Defined Networks (SDN), we present a controller reselection mechanism based on non-cooperative game theory. Load balancing among controller clusters is achieved by dynamic migration of SDN switches. The load balance of controller cluster, the average latency and the switch migration cost were adopted as utility functions. To meet Nash equilibrium strategy, we propose a Genetic Algorithm for Improved Multi-objective Optimization (GAIMO). To prevent global optimal solution from slipping into a local optimum, a similarity operator is designed to speed up the convergence and to improve the accuracy of the proposed algorithm. Experimental results show that the mechanism effectively optimizes overall resource utilization of controllers, and reduces average network latency, communication overhead, as well as the cost of switch migration, and thus optimizes entire network performance.