Research On Key Technologies For Web Service Selection Based On QoS Measure

Service oriented computing can seamlessly integrate the existing services or applications into new value-added services for satisfying users'requirements. However, due to the varied, open network and fast increasing services, how to find the most suitable services for service users quickly and reliably has become an important issue of service computing.Although service oriented architecture can effectively support the registration, discovery and composition of Web services, it remains difficult to how to select the most suitable services according to users' QoS (quality of service) reliably and quickly. The difficulty has attracted attention from industry and academia, particularly, the difficulty of service selection technology based on QoS measure during service composition. Hence, this paper mainly researches on the key technologies for Web service selection based on QoS measure and reaches the following results:1. In open Web service environment, it is difficult to guarantee the authenticity of the feedback ratings reported by users after they consumed some services, which leads to large deviation between evaluation results and actual values of service reputation, and therefore much more failures in service selection. To address the above problem, the paper proposes a reputation evaluation approach for Web service selection. The main idea of the proposed approach is evaluating the reputation by three reputation evaluation model including feedback checking, feedback adjustment and feedback detection to reduce the affect of chaos feedbacks, preference feedbacks and malicious feedbacks and improve the accuracy of reputation evaluation.2. Traditional QoS-aware Web service selection approaches often assume that the QoS published by service providers be trusted. However, the assumption is wrong because QoS of Web service is often influenced by several factors. To overcome the above problem, the paper proposes a synthetic QoS measure method for Web service. Its principle is to take the three factors, service providers, the context of customers and historical statistics into QoS measure. The aim of the approach is to improve the reliability and success ratio of service selection3. In Web service environment, the inherent uncertainty of QoS makes users selection deviate from their actual requirements, which often leads to service composition failures. To end up the above problem, the paper proposes a QoS uncertainty computing approach based on Cloud Model. Based on the thorough analysis on the reason of QoS inherent uncertainty, the approach adopts the backward cloud algorithm of Cloud Model to transform the quantitative service transaction records to qualitative service transaction records. Then by setting the corresponding threshold parameters, the QoS uncertainty degree of services can be determined.4. In order to satisfy users'real time requirements during service selections, that is, finding the services that satisfies users' end-to-end QoS requirements within a short computation time, the paper proposes a fast service selection approach. The core of the approach is using adaptive adjustment method based on fuzzy logic and adaptive particle swarm optimization algorithm to adaptively decompose global QoS Constrains into local constraints, and then the composition services which satisfy users end-to-end QoS requirements can be obtained by using local selection.