Probability-Based Crossover Genetic Algorithm for Task Scheduling in Cloud Computing

Abstract

In order to meet the rising demand for cloud services and remain in compliance with Service Level Agreements (SLA), service providers require effective task scheduling solutions capable of adapting to cloud computing’s elastic and dynamic characteristics.In this paper, we propose a novel approach to optimize task scheduling in cloud computing called a ProbabilityBased Crossover Genetic Algorithm (PxGA) with a primary objective of minimizing the tasks execution makespan. PxGA is an improvement on the Genetic Algorithm (GA) achieved by introducing the concept of Virtual Machine (VM) fitness and applying it to implement an effective weighted probabilistic crossover technique. Using the CloudSim simulation toolkit, we conduct our analysis of PxGA and evaluate it against standard and more recent task scheduling algorithms. The results of the simulations show that our proposed task scheduling algorithm is superior to other task scheduling algorithms in terms of the makespan, the VMs energy consumption, and the degree of imbalance (DoI). Moreover, the computational time (CT) for the PxGA decreases when compared against the other evaluated algorithms, except for its base GA.