Dynamic Resource Provisioning in Cloud Computing Using Optimized Wasserstein Deep Convolutional Generative Adversarial Networks

Abstract

Cloud computing (CC) has revolutionized the way resources are managed and delivered by providing scalable, on-demand services. However, dynamic resource provisioning remains a complex challenge due to unpredictable workloads, varying user demands, and the need to maintain cost efficiency. Traditional resource allocation techniques lack the adaptability required to optimize resource usage under dynamic conditions. This manuscript presents a novel approach for dynamic resource provisioning using an Optimized Wasserstein Deep Convolutional Generative Adversarial Network (DRP-WDCGAN-AHBA). Initially, the input data are collected from the Grid Workloads Dataset, which provides a comprehensive representation of workload patterns in cloud environments. The input data undergo rigorous preprocessing using Adaptive Self-Guided Filtering (ASGF) to ensure data quality. Then, Wasserstein Deep Convolutional Generative Adversarial Network (WDCGAN) is used to forecast CPU utilization over specified time intervals of 5, 15, 30, and 60 min. The Adaptive Hybrid Bat Algorithm (AHBA) is employed to optimize resource allocation dynamically and ensure efficient utilization. The proposed DRP-WDCGAN-AHBA model attains 20.36%, 18.63%, and 21.24% lower energy consumption and 16.78%, 23.64%, and 26.32% lower response time when compared with existing models, such as Multi-agent QoS-aware autonomic resource provisioning method BPM in containerized multi-cloud environs for elastic (DRP-QoS-EDSAE), Multi-objective dependent Scheduling Method for Effective Resource Utilization in Cloud Computing (DRP-LS-CSO-ARNN), and Energy-aware fully adaptive resource provisioning in collaborative CPU-FPGA cloud environs: Journal of Parallel and Distributed Computing (EFARP-CPU-FPGA).