The New Advanced Society
by Sandeep Kumar Panda, Ramesh Kumar Mohapatra, Subhrakanta Panda, S. Balamurugan
17PSO-Based Hybrid Weighted k-Nearest Neighbor Algorithm for Workload Prediction in Cloud Infrastructures
N. Yamuna, J. Antony Vijay and B. Gomathi*
Hindusthan, College of Engineering and Technology, Coimbatore, India
Abstract
Elasticity in Cloud Infrastructure is used to take huge computation and repository demands in an effective way. Since the load in cloud environments varies from time to time that will be the obstacle to providing guaranteed Quality of Service (QoS) to end users. The workload prediction in cloud environments improves proper utilization of resources and service level agreement at a stable level. Hence, Particle Swarm Optimization (PSO) based Hybrid Wavelet Weighted k-Nearest Neighbors (PHWkNN) algorithm is proposed to predict workload in Cloud Data Centre. This algorithm combines Wavelet Transform with a Weighted kNN algorithm and seeing weights to get better accuracy of workload prediction. PSO algorithm is utilized as a Parameter Optimization algorithm to find best possible values for the parameters in HWkNN model. Leave One Out Cross-Validation (LOOCV) method is adopted to validate the accuracy of the proposed model. In addition, it helps to adjust the weight value of the proposed algorithm and avoid the premature convergence concurrently. However, Google CPU and Memory workload dataset make use to assess accomplishment of the proposed algorithm. The evaluation outcomes demonstrate that PHWkNN algorithm is better than ANN and KSVR algorithms in accuracy ...