Research On QoS-based Service Selection In Dynamic Web Service Composition

Web Service extracts the advantage of distributed computation, Grid computation and XML, and solves the problems of heterogeneous distributed computation and data reuse. Web Service has the characters of inter-operation, multi-platform and loose coupling, so it becomes the leading development trend of WWW. But single Web Service has limited function, service reuse and sevice increment can be obtained only through combining multiple Web Services. In the environment of distributed, heterogeneous and multi-autonomous-fields Wide Area Network which is represented by Internet, highly dynamic network environment and changeful user requirement desire that Web Service Composition can cast off the static componsition mode, rectify and optimize composite web services in good time so that provide efficient, cheap and dynamic service composition.Quality of Service is the life of composite web services. As the development of Web Service, service functionality continually subdivides and deepens, and users have higher and various demands of QoS.The capability of realizing QoS-based web service composition with is the necessary condition that Web Service Composition can has practicability.In the context of dynamic Web services application environment, the dissertation emploies variety of mathematical models and tools to study the QoS model of Web Service, and the problem of Qos-based service selection, The main points of the dissertation are described as follows:1) The dissertation proposes a multi-lateral, hierarchical, unified General QoS model for web service. Compared with existing web service QoS model,the structure and attribute definitions of our model make more clear and comprehensive for QoS attribute definition. The Unified definition of the structure supports the scalability for QoS attribute stochastically.2) With complete information, It presents a method for data normalization, which can process the hybrid QoS criterions judgement matrix. The dissertation uses an improved flexible combination of subjective and objective methods to determine the weights of QoS attribute, and emploies gray correlation degree of the ideal point method to sort web service. Compared with other similar researchs, the method has a certain flexibility and can process the weight with only part of information data. System parameters can be set up to regulate the size of the difference between the weight of QoS attributes. The use of gray correlation degree affiliation ways can reflect the different user needs more comprehensibly. Compared with other similar researchs, our method can support more types of QoS attributes. It also has a clear concept, relatively simple calculation steps, and easy to program implementation.3) With incomplete information, the dissertation proposes a method for QoS-driven web service selection. It can support recruiting default in data, data discretization, web service filtration. So, we can use rough set similarity and weight to select web service with incomplete QoS data and unknown QoS attribute weight.4) It presents an improved particle swarm optimization algorithm based on disturbance mutation update, and emploies the algorithm to optimiz the web service selection under global QoS constraints. It combines advantages of wide-ranged exploration with mutation operation,to maintain diversity of new found non-dominated solutions via proportional distribution in the external repository, and deep exploitations of PSO to enhance the solution searching abilities of particles. Introduction of all the refined method allow the improved PSO algorithm to maintain the capbility of fast convergence and prevent solutions from falling into the local optimal.