Dynamic broker-based service selection with QoS-driven recurrent counter classes

For Web services providing similar functionality, quality of service (QoS) is the main factor to differentiate them. In this paper, we propose a new approach which enables a more efficient service selection in dynamic environment of QoS characteristics. First, we present a new broker-based architecture for Web services in which the role of service broker is separated from service quality analyzer. Then we propose a three-phase technique that is performed by a third party broker. In phase one, we use the QoS Class concept to categorize service providers based on QoS issues. In phase two, we suppress the uncertainties of QoS provided by service provider leveraging recurrent counter classes. Finally, we consider a third phase which is referenced to an AI planning algorithm the result of which produces a chain of services while considering service requesters restrictions in service selection problems. In this paper, not only do we consider the past QoS acquisition of services, but we consider also the online variation of QoS characteristics in serving services. Hence, the QoS assessment is dynamic and performed online.