AI-Based Metaheuristic Algorithm for Reducing Cost and VM Failure Rate in Task Scheduling within Cloud Computing Environments

Scheduling is an NP-hard problem, and heuristic algorithms are unable to find approximate solutions within a feasible time frame. In Cloud Computing (CC) environments, efficient Task Scheduling (TS) plays a critical role in minimizing operational expenses and enhancing system reliability. This paper presents a novel task scheduling approach that uses the Coati Optimization Algorithm (COA) to address two pivotal challenges: reducing the total cost (sum of computational cost and communication cost) and minimizing Virtual Machine (VM) failure rates. Inspired by the cooperative foraging and adaptive behavior of coatis in dynamic environments, the proposed algorithm leverages intelligent exploration and exploitation strategies to identify optimal task-to-VM mappings under fluctuating workloads. The COA incorporates cost-awareness and failure probability metrics into its fitness function to ensure robust scheduling decisions that align with budgetary constraints and fault tolerance requirements. To assess the performance of the proposed model, comprehensive simulations were conducted using the CEA-Curie real-world workload. The results were compared against three state-of-the-art approaches, MoHHOTS, RTATSA2C, and TS-GWO. Experimental evaluations demonstrate that COA significantly outperforms these existing methods by achieving a 19.8% reduction in overall cost and a 22.5% decrease in VM failure rate. These findings demonstrate that COA offer a promising pathway toward sustainable, cost-effective, and resilient task execution in large-scale cloud infrastructures, particularly under diverse and realistic workload scenarios.