Hybrid Metaheuristics Using Rough Sets for QoS-Aware Service Composition

Abstract

Quality of Service (QoS)-aware service composition plays an increasingly important role in various computational paradigms and delivery models, predominantly cloud computing. The proliferation of services with expanding quality attributes navigates this problem towards big service compositions, which fall under the umbrella of NP-hard. Within the realm of big services, performing composition also became a computationally expensive and challenging task. Since service composition is an NP-hard problem, numerous research aimed to determine optimal or near-optimal solutions within a reasonable budget. A large body of evidence suggests that metaheuristics could realize this goal to some extent. However, the proliferation of services with expanding quality attributes (search dimensions) may fail the most efficient techniques. In order to deal with the problem of big service composition, one trending approach has been devising hybrid metaheuristics methods by incorporating clustering techniques to minimize search space. This paper proposes a hybrid metaheuristic incorporated with a maximal discernibility heuristic based on rough set theory to perform composition in the subset of search space. Moreover, it introduces a parallel architecture and monitoring mechanism to provide immunity against premature convergence when the search space is minimized. The experiment was conducted for 25 datasets generated incrementally from a real-world QWS dataset, where the proposed hybrid solution effectively improves solution quality and reduces execution time with a statistical significance of 99 % confidence interval across diverse metaheuristics and datasets.