Research On QoS-aware Service Discovery And Service Composition

Service-objected computing is a usual computing model that is applying in the loose-coupled, open and distributed heterogeneous environment, and Web Service is a new service-objected architecture in the recent years. With the rapid development of the Internet information, the resources of Web Service become more and more enriching, meanwhile the applications across different fields of the Web Service become wide increasingly. Lots of UDDI registries exist in autonomous domain, so that realize Web Services' search, discovery, integration, and classification, which will become as challenged as Web pages search. For the sake of explosive increase of the number of the Web Services, a large number of Web Services with similar function exist in the Internet service virtual community, so how to utilize QoS (Quality of Service) attributions to select advertising services is one of key technologies of Web Service composition. In all, in the distributed, isomerous and autonomous-coupled environment, the quality guarantee of Web Service composition has affected the methods of the Web Service composition primarily.Two main problems have been tackled under the circumstance of Web. One is how to search and discover Web Services across multiple UDDI, and the other is how to provide the applications with quality guarantee according to the providers' requirements. Aiming at the above problems, service management framework and resource management component—WebJet Repository Builder (across multiple UDDI) has been proposed. At the same time the strategy of how to search and discover services has been also put forward. On the basis of the framework, a QoS-based architecture of Web Service discovery and composition—WebJet has been brought forward. And WebJet is designed for publication, management, discovery of the services, optimization prior to the composition and fault-tolerance at the running time.Pointing at the uncertainty or not of the users' preferences, based self-adjusting judgment matrix and the interval judgment matrix, the method computing the QoS attribution weights has been proposed. To realize the global optimization of Web Service composition, a 0-1 heuristic algorithm has been implemented. Additionally, compared with the branch bounding method of integer programming and GAs (Gene Algorithms) under the different running environments, the advantages of 0-1 heuristic algorithm have been verified. Currently, little optimization models of Web Service composition allow for dynamic commerce alliance of the services in the management of the supply chains. To improve the disadvantages, non-linear optimization model of Web Service composition supporting supply chain management has been put forward, as well, the optimization strategy for GAs has been brought forward. At last, practicability of the service alliance models has been proved through some experiments.