A Hybrid Cuckoo Search and Manta Ray Foraging Optimization Algorithm for Multi-Objective Task Scheduling in Cloud Computing
Abstract
Task scheduling in cloud computing is an NP-hard multi-objective problem that requires the simultaneous optimization of makespan, execution cost, and energy consumption. Although individual metaheuristic algorithms can be strong optimizers, they often fail to achieve a reasonable exploration/exploitation balance across varying problem scales. This paper presents a hybrid algorithm, H-CS-MRFO, that combines the global exploration of Cuckoo Search (CS) through an adaptive Lévy flight mechanism with the local exploitation of Manta Ray Foraging Optimization (MRFO), which employs chain, cyclone, and somersault foraging strategies. A weighted-sum Pareto method simultaneously minimizes three conflicting objectives. Synthetic workload experiments (50–1,000 tasks, 13 heterogeneous VMs, 10 independent runs) indicate that H-CS-MRFO statistically outperforms three competitive baselines (MMWOA, MMRFO, MCS): makespan is reduced by up to 5.6%, execution cost is decreased by 1.4–2.1%, and power consumption is reduced by 4.5–6.8%, while a well-distributed Pareto front is obtained. Wilcoxon signed-rank tests confirm the statistical significance (p < 0.05) of each metric across each task scale.