Multi-Objective Load Balancing in Heterogeneous DistributedComputing Systems Using Particle Swarm Optimization

Load balancing distributes incoming traffic across a pool of resources to improve system performance and prevent overutilization or underutilization of resources. In heterogeneous distributed computing systems, dynamic load balancing involves continuous monitoring and redistribution of tasks among available nodes, considering their diverse processing capabilities. As a result, efficient task distribution remains a major concern in these systems. The study proposes an approach that combines probability-based load migration and weighted-based task reallocation using Particle Swarm Optimization (PSO). The primary goal is to minimize task execution time while maximizing overall resource utilization in a heterogeneous environment, with a focus on optimizing memory and CPU usage. The proposed dynamic algorithm effectively adapts to handle numerous tasks and fluctuating nodes. Simulations are conducted using ‎CloudSim. The performance is compared with existing PSO variants: Time-Conscious PSO, PPTS-PSO, EASA-MORU, and ‎HEPGA. Results show that ‎the proposed method significantly outperforms other existing approaches. Specifically, the algorithm achieves an average resource utilization of 98%, reduces execution time, improves throughput, and ensures fair workload distribution across all nodes. The proposed method is highly suitable for distributed computing environments characterized by dynamic task demands and varying computational resource requirements.