IJPSO: An Improved Hybrid PSO Algorithm For Multi-Type And Large Scale Data Scheduling In Cloud Computing

Abstract

The services of cloud computing are expanding quickly. The objective of cloud computing is to offer online services as it is a pay-per-use model through which users can manage data as per their needs. The number of cloud users is growing, so task scheduling is considered an important issue for allocating tasks to resources. In recent years, numerous nature-inspired meta-heuristic algorithms have been introduced and can be applied to solve task scheduling issues. However, these techniques do not consider simultaneously multi-type workload scheduling such as workflows and independent tasks. In this research, various task scheduling algorithms have been examined and a new algorithm has been designed named IJPSO using JAYA and particle swarm algorithms to enhance the cloud system performance. PSO suffers from local stagnation issue; therefore, JAYA acts as a local refinement operator. Adaptive inertia weight is adopted to balance the local and global searching capabilities. Simulations were performed using scientific workflows and independent Google traces 2019 dataset. The newly designed algorithm has performed better than Particle Swarm Optimization (PSO), Grey Wolf Optimization (GWO) and IPSO in terms of makespan and monetary cost.