Reliability-Driven Dynamic Web Services Selection Technology

Web service is an application program, which is based on internet, distributed, self-included, self-described and modular. It possesses the attributes of interoperability, cross-platform, loose coupling and so on. With the development of Web services, composite Web services which can provide value-added functions have received a lot of attentions. Composite Web services are a kind of Web services which combine the existing, relatively simple services based on certain business logic in order to make the composed one achieve more powerful and more complete business functions. However, because of the dynamic, autonomous and other features, the reliability of composite Web services become more prominent, it is the key point for Web services to work correctly or not. Therefore, how to provide the assured reliability of composite Web services which can meet the functional requirements and non-functional constraints from users attracted much attention. It is one of the hottest issues in recent researches.To this problem, this thesis discusses the problem of the service selection in the dynamic service composition, and study the reliability-driven approach for dynamic Web service selection in semi-automatic service composition. Firstly, in order to effectively assess the reliability, this thesis puts forwards a reserve-based reliability evaluation model. Secondly, the thesis presents a reliability-driven dynamic Web service selection approach. The approach targets optimal reliability of composite Web services in the service selection stage. Each task does not correspond to only one service instance, but that it selects a number of alternative services for each task of the composite process on the basis of meeting the constraints of users, that is, bases on the redundancy to ensure the reliability of composite Web services. For service selection problems of execution time constraints, the thesis presents a reliability-driven service selection algorithm based on particle swarm optimization. As the selection of composite Web services is an NP problem, it can make good use of intelligent algorithms to optimize composite Web services selection in order to improve the quality of results and execution time. In the premise of meeting the constraints of users, the thesis uses particle swarm intelligent optimization algorithm to find appropriate alternative services for each task based on reliability evaluation model to get the maximum reliability of composite Web services. Eventually, the thesis designs three types of experimental schemes to analyze and evaluate the approach on the algorithm efficiency, the parameter settings and the adaptable conditions respectively. These experiments verify the effectiveness and practicality of the reliability-driven service selection approach.