Breaking the Swarm Paradigm: Single Candidate Optimization for Cloud Task Allocation

Cloud computing has become an indispensable platform for delivering scalable, on-demand services to a wide array of users. One of critical trials in cloud settings is efficient task scheduling across heterogeneous virtual machines (VMs), which requires balancing multiple conflicting objectives such as execution time, energy consumption, cost, and resource utilization. To address these tests, this paper proposes a novel metaheuristic, single candidate optimization algorithm (SCOA) is designed to overcome the limitations of traditional population-based procedures. Unlike swarm-based models, SCOA utilizes a single candidate solution, which evolves through a two-phase strategy for exploration and exploitation. The algorithm dynamically adjusts its search behavior using a decaying weight parameter and adopts an adaptive reinitialization mechanism to avoid stagnation in local optima. The performance of SCOA is evaluated through extensive simulations on the CloudSim platform besides compared against state-of-the-art procedures including Enhanced whale optimization algorithm (EWOA), WOA, ACO, HHO, and GA. The results demonstrate SCOA’s superiority across ten evaluation metrics, including resource utilization (91.2%), energy consumption (134.7 kWh), execution cost ($312.5), and SLA violation rate (1.2%). SCOA also exhibited the lowest makespan (408 sec) and highest task success rate (98.6%). These findings indicate that SCOA offers a highly efficient, scalable, and robust solution for multi-objective cloud task scheduling, paving the way for greener and more cost-effective cloud infrastructures.