Taylor CFRO-Based Deep Learning Model for Service-Level Agreement-Aware VM Migration and Workload Prediction-Enabled Power Model in Cloud Computing

Abstract

In this research, Taylor Chaotic Fruitfly Rider Optimization (TaylorCFRO)-based Deep Belief Network (DBN) approach is designed for workload prediction and Service level agreement (SLA)-aware Virtual Machine (VM) migration in the cloud. In this model, the round robin technique is applied for the task scheduling process. The Chaotic Fruitfly Rider Optimization driven Neural Network (CFRideNN) is also introduced in order to perform workload prediction. The DBN classifier is employed to detect SLA violations, and the DBN is trained using devised optimization model, named the TaylorCFRO technique. Accordingly, the introduced TaylorCFRO approach is newly designed by incorporating the Taylor series, Chaotic Fruitfly Optimization Algorithm (CFOA), and Rider Optimization Algorithm (ROA). The developed TaylorCFRO-based DBN scheme outperformed other workload and SLA Violation (SLAV) detection methods with violation detection rate of 0.8048, power consumption of 0.0132, SLAV of 0.0215, and load of 0.0033.