A New Metaheuristic-Based Load Balancing Framework Using Grey Wolf Optimization And Centroid Opposition-Based Learning

Abstract

One of the key challenges in cloud computing is the efficient allocation of load among the resources, which are distributed dynamically and may contain resources of different capacities in the form of virtual machines (VMs). This paper presents an improved metaheuristic, which is the Grey Wolf Optimiztion (GWO) algorithm combined with Centroid Opposition-Based Learning (COBL) algorithm for a novel load balancing scheme. COBL facilitates the convergence process to help improve the exploration/exploitation ratio of GWO by producing competitive opposite solutions around the centroid of the population. The COBL-GWO algorithm is proposed for task scheduling in cloud computing to enhance the main performance metrics such as make span, resource utilization, execution time and memory usage. The experimental tests with different task sizes and VM configurations (5 and 10) show that COBL-GWO is clearly superior to the standard GWO, IGWO, CPSO, PSO and GA approaches. Specifically, COBL-GWO shows good performance in terms of computing cost, resource utilization, and make span, and is thus a strong and scalable solution to efficiently schedule and balance tasks in cloud computing.