EFFICIENT CLOUD TASK SCHEDULING THROUGH ADAPTIVE INERTIA WEIGHT PARTICLE SWARM OPTIMIZATION

Abstract

Cloud computing has revolutionized the delivery of computing resources, enabling on-demand access to servers, storage, and applications over the internet. An improved Particle Swarm Optimization(PSO) is presented for task scheduling, utilizing a linearly decreasing adaptive inertia weight to enhance performance. The key enhancement is the use of a linearly decreasing adaptive inertia weight, which shifts from 0.9 to 0.4 over 500 iterations. This dynamic adjustment better balances global exploration and local exploitation compared to traditional PSO, which uses a constant weight. Evaluated in CloudSim, the improved PSO was compared against traditional PSO and conventional algorithms (FCFS, SJF, RR). The results show significant performance gains: a 21.18% reduction in makespan, 9.45% reduction in completion time, and an 11.22% reduction in waiting time, with only a 0.68% cost increase. This highly effective and practical solution requires minimal code modification for real-world deployment.