Web Service Composition Research Based On Imperialist Competitive Algorithm

Large-scale development of cloud computing and Web services has not only brought new experience and to both enterprises and users,but also caused great challenges.A single Web service in the traditional resource pool is difficult to meet the complex and changeable needs of users,while Web service composition in cloud computing environment can realize added value based on the existing services by combining different Web services as a more powerful composite Web service.Presently,Quality of Service(QoS)is widely used as the indicator in evaluating Web services,which can only evaluate services in performance without consideration of users' satisfaction.This thesis mainly studies the use of imperialist competitive algorithm in solving Web service composition problem based on QoE.Firstly,a Web service composition model based on QoE is studied and then abstracted as a mathematical optimal problem,which is according to the relationship between QoS and QoE and fuzzy expert system.Secondly,a parametric adaptive ICA based on splitting mechanism is proposed,which uses chaos initialization,parameter adaptation mechanism and adds empire splitting mechanism to keep the population diversity both at the begginning and in the process of iteration,avoid falling into local optimum by dynamically adjusted assimilation migration angle and improve the global search ability.The Basic test functions are used to verify the performance of AGICA.Finally,the performance simulation of AGICA,GICA,PSO algorithm and DE algorithm are carried out.The experimental results show that AGICA has better effectiveness and stability,faster convergence rate but longer execution time compared with other algorithms in solving the problem of Web service composition.