Using Service Clustering and Self-Adaptive MOPSO-CD for QoS-Aware Cloud Service Selection

Abstract

A promising way to effectively manage the composition of services in a heterogeneous and dynamic environment is to make workflow management able to self-adapt at runtime to react to changes in its environment by autonomously reconfiguring itself. Most of the proposed methodologies address this issue as a QoS-aware service selection problem in which a web service broker can dynamically select the “right” service that takes part in the composition, and adaptively change the bound service when the delivered QoS has changed.In this paper, we propose a self-organizing framework that uses a service clustering based discovery approach to effectively and efficiently support the selection of services in which runtime changes in the QoS of the services are taken into account. Two bio-inspired algorithms are designed to support a QoS-aware dynamic service selection mechanism. An ant-based clustering algorithm enhanced with a template mechanism that guides the artificial ants to move data items to construct and maintain a specific topology is adopted as a method for efficient service discovery. As a consequence, services can dynamically be discovered in a shorter time and with lower network traffic. To select the actual concrete services that best meet the user QoS requirements a Self-Adaptive Multi-Objective Particle Swarm Optimization Algorithm using crowding distance technique (MOPSO-CD) is executed using the topological map generated by the ant algorithm. In the end, simulation results show the effectiveness of the method proposed.