A comprehensive review on autonomous and plan-based QoS-aware service composition approaches in service community
Résumé
In the evolving landscape of the Internet of Things (IoT), many smart devices provide basic functionalities as services that often fail to meet users’ requirements. To address this gap, fine-grained services are combined to create coarse-grained services, giving rise to the QoS-aware service composition challenge. The growing number of IoT devices has resulted in numerous fine-grained services with similar functional properties (inputs and outputs) but different non-functional properties, typically related to quality of service (QoS). The QoS-aware service composition approaches can be categorized into two main classes: plan-based approaches and autonomous ones. Existing surveys in this domain primarily focus on either a single class or technical aspects, such as composition representation and utility functions. In contrast, this study offers a comprehensive and systematic review of both classes, emphasizing resolution and validation methodologies. The contributions of this paper are fourfold: 1) introduce a novel taxonomy that classifies QoS-aware service composition approaches from the past decade based on their reliance on preexisting abstract composition plans, 2) provide a detailed comparative analysis of these approaches using the most relevant criteria in the service composition field, 3) investigate their resolution and validation methodologies along with the quality and appropriateness of the datasets used in the reviewed approaches, and 4) discuss future directions for advancing the field. This review not only bridges the knowledge gap in understanding these approaches but also serves as a foundation for researchers to develop more robust and efficient QoS-aware service composition methods.