Leveraging Deep Learning-Based Approach for IoT Service Composition Through Local Service Selection
Résumé
The advent of the Internet of Things (IoT) has created opportunities for developing innovative smart systems, but it has also introduced complex requirements often subject to Quality of Service (QoS) constraints. Efficient service composition, essential for addressing these needs, involves selecting and integrating suitable services. Given the rapid expansion of IoT and the exponential increase in available services, the selection of IoT services poses significant challenges. In this context, several approaches have been proposed in the literature broadly categorized into two main types: global selection and local selection. Global selection approaches suffer from scalability issues, while local selection methods rely on local QoS constraints derived from a predefined set of candidate services, which is a limiting assumption in large and dynamic environments. In this paper, we present a novel deep learning-based QoS decomposition approach for IoT service local selection. Our method efficiently transforms global QoS constraints of complex needs, formalized as abstract processes, into distinct local constraints. This decomposition facilitates local selection without relying on a predefined set of services, enabling a flexible and dynamic selection process that accommodates the inherent variability and expansiveness of IoT environments. We have implemented and evaluated the proposed approach. The results show significant advancements in addressing the IoT service selection problem compared to the existing methods in the literature.
Domaines
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |