Hybrid Resource Allocation Technique Based on Double Q Learning and Metaheuristic Algorithms in Optimizing the Allocation of Tasks in Cloud Computing

Abstract

Cloud computing (CC) has grown as a core component of modern computing facilities, providing users with adaptable, on-demand services. In dynamic environments, the effective utilization of resources is significant for performance, cost-effectiveness, and user satisfaction. The optimization of resource allocation is important in CC due to the dynamic and complex nature of cloud infrastructure. This study suggests a hybrid resource allocation procedure that combines Double Q-learning(DQ), a reinforcement learning (RL) approach, with meta-heuristic algorithms, including the Whale Optimization Algorithm (WOA), and it is compared with Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), and Genetic Algorithms (GA) to optimize task allocation in CC settings. This combined system overcomes task allocation challenges such as usage of resources, energy consumption, and SLA compliance by integrating Q-learning's adaptive decision-making skills with meta-heuristic methods' global search efficiency. The proposed approach dynamically adapts optimum resource allocation guidelines while considering a variety of options for near-optimal task scheduling and resource assignment. The results of the experiment reveal that the hybrid strategy outperforms the PSO, ACO, and GA techniques in terms of makespan, reaction time, throughput, resource utilization of resources, cost, and rate of resource violation. This study provides an effective approach to handling complex resource allocation difficulties in a variety of dynamic cloud environments.