A Method For Web Service Composition With QoS Constraints

With the development of Internet and Web technologies, traditional distributed computing technologies have not been suitable for the Web application environment. Therefore, Web services technology comes forth and is widely accepted as a paradigm of service-oriented computing. Perspective of Web services is most accredited in the view of application of e-business. However, some key technologies of Web services are not solved well. Therefore, Web services technology becomes the focus in the area of industry and academia.Generally, Web services are designed for special purposes based on the functionalities of their providers. Moreover, to decentralize and simplify the application logic, and improve the reusability of Web services, a single service should not be complex. However, in the real world, a user request may be so complex that it can not be satisfied by a single service. It is therefore necessary to compose some pre-existing available Web services to provide new value-added services. This procedure is called service composition. The number of Web services is dramatically increasing with the popularity of Web services technology. There will be multiple services providing identical functionality. Non-functional properties of services, also known as Quality of Service (QoS), become more and more important in service composition.This thesis mainly discusses Web service composition with QoS constraints. The major work and contributions are as following: ·Having studied Web service composition, this thesis proposes an automatic service composition method based on composition tree. After obtaining the mapping relationship from output parameters to Web services by establishing the indexing for service output parameters, the method makes use of the divide-and-conquer strategy to compose services automatically. Compared to traditional graph-based methods, it reduces the search space significantly and avoids the repetition search, so it can obtain the best service satisfying user requirements according to service selection algorithm in a short time.·A method for QoS property value acquisition is presented. When a Web service has just been published, the QoS property values provided by service providers are used as current values of QoS properties. Then, the current values are constantly updated according to feedback values of QoS properties from service consumers. With the increasing number of feedback values, the QoS property values provided by service providers and the feedback values from a few service consumers have smaller and smaller influence on quality of service. So the method can reduce the negative influence of cheating behavior, and reflect the quality of service under the current circumstance objectively and fairly.·A service selection algorithm is designed. First, the algorithm standardizes the QoS property values of candidate services based on the user's QoS constraints, such that QoS property values are converted to [0,1] interval and have the same orientation. Then, it calculates the QoS synthetic score of each candidate service according to user preference, and selects the best one.