Research On QoS-Aware Web Service Discovery And Composition

The emergence of Service-Oriented Computing (SOC) provides a flexible solution for building distribution applications and enterprise information integrations. While Service-Oriented Architecture (SOA) is an infrastructure of SOC, and Web service became the element of SOA. The functional attributes and non-functional attributes are under heaviy studing by both industry and academic. Further more, the describtion of Web service, the discovery and matching, and the composition of Web services became the kernel problems of current SOC research. The issues of Quality of Service (QoS), the main part of a Web service's non-functional attributs, are addressed in this thesis.This thesis takes a view of service register center. The key problems and technologies including: the formal describtion of Web service's QoS, QoS-aware Web service discovery, Web service composition and the uncertainty of Web service. Some specific contributions of our work are listed as follows:1) A novel Web service QoS model is proposed. The diversity and complexity of QoS criterias are considered by the new QoS model. It proveds a multiple view of QoS, covers most of current QoS criterias and more it is easy to extand. The experiments show that it is very suit for a service register center.2) A QoS-aware Web service discovery and matching technology is proposed, which is based on both user's Context and user's QoS preferences, so it is user centric. This technology is build on the proposed QoS model, user Context is used for filting QoS information, and user's QoS preferences and QoS criterias are synthesized by Analytic Hierarchy Process.3) A QoS-aware Web service composition framework is proposed. Each basic composition pattern is studied, and algorithms are provided. This framework contains basic planning algorithms, run time replanning trigger algorithm and replannning policies. The uncertainty is considered in this framework and this is an important future. Thus the services are not bingding optimally at first time; rather the service bingding is oftern changes in the run time according to the real time information of the composition service. The experiments show that these algorithms is suit for a low uncertainty.4) A novel Web service composition method using Hierarchy Markov Decision Process is proposed for the case of a heavy uncertainty environment. The general Web service compostion is modeled by a Hierarchy Markov Decision Process, and the optimal policy serviced as a service execution plan, thus there is no need to replan during the run time and the composite service would adjust it self adaptively.This thesis addresses key technologies of Web service QoS management in a service register center. A novel QoS description model is proposed and an effective service discovery mechanism is based on it which takes user's context and preferences into consideration. Algorithms for QoS-aware service composition are propsed, and the uncertainty nature of Web services is studied. Service register center is the kernel componet of Service Oriented Computing solution. The technologies provided by this thesis make the service register center QoS-aware and enhanced the SOC solutions.