Support Vector Machine based proactive fault-tolerant scheduling for Grid Computing Environment

Abstract

To classify the reliable resources accurately and perform a proactive fault tolerant scheduling in grid computing environment, a combination of support vector machine (SVM) with the quantum-behaved particle swarm optimization using Gaussian distributed local attractor point (GAQPSO) is proposed in this paper. When tuned with appropriate kernel parameters, the SVM classifier provides high accuracy in reliable resource prediction. The higher diversity of GAQPSO compared to other variants of QPSO, reduces the makespan of the schedule significantly. The performance of the SVM-GAQPSO scheduler is analysed in terms of the makespan, reliability, and accuracy. The empirical result shows that the reliability of the SVM-GAQPSO scheduler is 14% higher than the average reliability of the compared algorithms. Also, the accuracy of prediction using the SVM classifier is 92.55% and it is 37.2% high compared to classification and regression trees (CART), linear discriminant analysis (LDA), K-nearest neighbourhood (K-NN), and random forest (RF) algorithm.