Dynamic Selection for Service Composition Based on Temporal and QoS Constraints

Abstract

To implement abstract business processes, elementary services are selected for each abstract task. Because of uncertainties of Quality of Service (QoS) values during execution, services may become faulty and cause the violation of end-to-end awaited constraints. Additionally, due to the dynamic nature of service systems, several environment changes may occur at run-time. In fact, services can join or leave the system or change their offerings. To deal with possible changes and maintain the feasibility of the selected solution, enabling dynamic service selection during execution is essential. This is not a trivial task especially in the presence of several constraints and dependencies between services namely QoS and temporal constraints. Existing approaches do not consider the specificities of temporal properties and usually handle violations after they have occurred. In this paper, a novel proactive dynamic service selection approach is proposed to deal with changes during execution while considering both QoS and temporal constraints. Experiments show that, by using our approach, faults can be successfully handled in a reasonable time while guaranteeing overall constraints.