Abstract Efficient composition of cloud services under Quality of Service (QoS) constraints is a critical challenge in service-oriented computing. Traditional metaheuristic approaches often struggle to balance global exploration and SLA compliance in high-dimensional, dynamic environments. To address this, we propose an Enhanced Prairie Dog Optimization (EPDO) algorithm that integrates Lévy flight dynamics with behaviorally adaptive search phases. The EPDO is validated through benchmark functions and real-world service composition situations and exhibits a 94.1% SLA conformity rate, which is better than baseline algorithms GA, PSO, MFO, and original PDO. It also demonstrates a 15–25% speedup in convergence and maintains similar runtime efficiency. As demonstrated by the findings, EPDO’s future prospects are in its capable, constraint-aware optimization in extremely massive-scale cloud service systems.
Daxin Tian (Mon,) studied this question.