Design and Analysis of Optimized Scheduling in Cloud Computing using Improved Metaheuristic Algorithm

Abstract

In the dynamic field of cloud computing, efficient task scheduling is paramount for optimizing newlineresource utilization, reducing costs, and enhancing system performance. This research newlineintroduces the novel Grey Wolf-Cuckoo Search Algorithm (GWO-CSA), a hybrid optimization newlinemethod combining the predatory behavior of grey wolves with the parasitic breeding strategies newlineof cuckoos. Designed to improve scheduling processes, this algorithm was assessed using newlineCloudSim a framework that simulates cloud environments, allowing for detailed comparison newlineagainst established methods like Particle Swarm Optimization (PSO) and Ant Colony newlineOptimization (ACO). The innovation of GWO-CSA lies in its dual-strategy approach, which newlineenhances global and local search capabilities, ensuring comprehensive exploration and newlineexploitation of the solution space. This approach is particularly effective in cloud environments newlinewhere task loads and resource availability are subject to frequent changes. newlineThe effectiveness of GWO-CSA was rigorously tested across various performance metrics such newlineas cost-efficiency by considering makespan, resource utilization, throughput, and scalability. newlineEmpirical results indicate that GWO-CSA significantly outperforms traditional algorithms. For newlineinstance, in tests involving different numbers of tasks and virtual machines, GWO-CSA newlineconsistently showed superior performance in makespan time and operational costs. It achieved newlinea makespan time of 19.6 seconds for 10 tasks, which scaled up linearly to 167.5 seconds for newline100 tasks, demonstrating both high efficiency and scalability. In contrast, PSO and ACO newlineexhibited steeper increases in makespan with increasing tasks, highlighting potential scalability newlineissues and less efficiency under load. In terms of cost, GWO-CSA maintained lower newlineoperational costs across all scenarios, starting at 203 for 10 tasks and peaking at only 520 for newline100 tasks, indicating its exceptional cost-effectiveness. newline