Research On The QoS-related Problems In Web Service Composition

With the fast development of Internet and the gradual maturity of web services technology, the applications based on web services thrive, and a large number of web services are published on the Internet. Due to the limited utility of a single web service, it is natural and effective to compose single ones into a more powerful composite service for the purpose of satisfying users’changing requirements. Thus more and more attention is paid to web service composition. The prediction of a single Web service’s QoS, the QoS modeling of the composite web service and QoS-based web services selection are the three research hotspot.The prediction of a single web service’s QoS is the basis of the QoS modeling of composite web service and QoS-based web services selection. The recent methods mostly predict the QoS by mining similarity between services or users, and they deal with different QoS in one way. Based on the in-depth analysis of QoS, the prediction method based on SVM for objective direct QoS and the prediction method based on environment matching for objective statistical QoS are proposed.In recent research, only member services’QoS and the structure of composite service are considered in the composite service QoS model. Actually, the execution of composite service must rely on the engine, and the QoS would be affected by the engine. The QoS model of composite service is improved in this thesis, considering the affection of engine.Local optimal selection and global optimal selection are included in the QoS-based web services selection. The key problem of local optimal selection is to get the weight of QoS properties, thus a new combination weighting approach is proposed. Global optimal selection has been shown to be an NP-hard combinatorial optimization problem, and lots of methods have been used to deal with this problem, such as linear programming, constrained optimization methods, heuristic methods, Genetic Algorithm and so on. Estimation of Distribution Algorithm is introduced to solve the problem, and the good results are received in this thesis.