Researches On QoS-based Web Services Composition Methods

With the development of society and economy, the amount of information is in afast increase and the scale of the software system is also in a rapid expansion, whichgenerate new requirements for information technology. To satisfy the interconnectionof heterogeneous information across organizations, Service Oriented Architecture(SOA) and cloud-computing emerges. The advancement of Internet technology andWeb application promote the development of service-oriented technology and cloudcomputing. As an implementation of SOA, Web service has caused widespreadconcern in the industry and the research. With more and more web services applied inpublic information database, it becomes vital to provide an efficient web servicecomposition mechanism with respect to users’ requirement.The basic idea of SOA is to combine reusable web services with small granularityinto a compositional service with larger granularity to meet users’ requirements. Asthe widely application of web service, quality of service has been paid more attentionby users. Therefore, this paper researched the methods of web services combinationbased on the Quality of Service. Firstly, we present a complete evaluation system forWeb services composition and use hierarchical index system to measure the quality ofcost, time, reliability and availability. Also, we calculate QoS of compositionconsidering about the structure of the composition and the logic control relationsamong services. Secondly, quality risk is defined as the variance between actual valueand the expectation of quality. We propose a multi-Objective programming model,aiming at QoS and quality risk optimization. Finally, the ε-dominance basedmulti-Objective evolutionary algorithm is applied to solving the model with syntheticdata. Experiment results demonstrate that the presented model and algorithm canprovide effective and comprehensive information for decision makers to selectappropriate web services for composition.