A QoS enhanced framework and trust model for effective web service selection

Abstract

Service Oriented Architecture (SOA) has become a promising paradigm for software development due to its benefits in cost-efficiency, agility, adaptability, and leverage of legacy investment. One of the most important research topics in SOA is to identify appropriate Web services for a service composition so as to fulfill business requirements. Web service selection focuses on selecting best services among a bunch of services with same or similar functions but having different QoS (Quality of Service). Many previous approaches, such as QoS models with quality criteria and selection algorithm, have been proposed to optimize web service selection. However, in current research, quality values normally come from service providers, who have high possibility to exaggerate these values for advertisement. It is also argued that reputation based on an average user rating is not enough to indicate the trust degree of Web services and service provider. In addition, handling dynamic nature of web services is still a challenging problem for dynamical web service selection. In this thesis, these problems are focused to solve. First QoS enhanced framework for effective Web service selection is proposed. Then a Trust model is built, which is composed of TQoS model, Decision model and Trust correction. It is claimed that a Web service can be regarded as trustful if QoS values received by consumers and tested by registry are no less than QoS values promised by providers. A prototype of the proposed framework is implemented, including SC agent, SR agent and QoS Enhanced SR. In addition, a scenario about a Tour agency’s Web service selection according to its business process is implemented. To validate effectiveness of proposed framework and Trust model, several comparisons were done. First, the effectiveness of multi-source QoS data is verified by comparing with single source QoS data on obtained quality values. Then proposed approach is compared with other QoS models for Web serv...