A Comparative Study of PSO, PSO Variants, and Random Scheduling in Solving Workflow Scheduling Problem in Cloud Computing Environment

Abstract

To obtain an optimal solution for an optimization problem, the most important and, hence, the crucial step is to make the right choice of the optimization algorithm from a diverse range of algorithms available. Cloud computing is a methodology that dynamically as well as simultaneously provides services and allocates resources to remotely residing users through Internet-based tools using the pay-for-use model. The unlimited storage, ease of use, backup and amp, recovery, and security are some of the features of a cloud computing environment which make its high demand even higher. This demand is the reason that calls for better cloud optimization which minimizes the processing cost using scheduling policies and algorithms. The objective here is to minimize the total cloudlet processing cost using an optimal scheduling algorithm on virtual machines. In this paper, the optimization heuristics named particle swarm optimization, its variants, and random scheduling are compared. The comparison of the results shows that different variants of the PSO heuristic perform better in comparison with random scheduling. Among different variants of PSO used for task scheduling purposes, PSO using constriction factor and PSO variant using both linear decreasing inertia weight (LDIW) and amp; constriction factor is found to perform better than others.