Research On QoS-aware Web Services Selection Based On Multi-objective Artificial Fish Swarm Algorithm

With cloud computing technology being prevalent, the research of Web service,which is a mainly cloud service technology, will be re-valued, while individual Webservice is not capable of meeting the needs of users in the cloud environment, Webservice composition, which Web service selection is one of the key point for, becomesincreasingly significant. At present, most of the Web service selection algorithms havethe drawback of single target and local optimum, and the research of multi-objectiveQoS global optimal Web service selection algorithms, which are likely to in line withthe actual users’ demands, is not common. Multi-objective fish swarm algorithm is anew type of intelligent optimization algorithm, which has characteristics of strongrobustness, global convergence and is less sensitive to the initial value, has beenwidely used in many fields.Firstly, QoS global optimal Web service selection problem was transformedinto a constrained condition multi-objective services combinatorial optimizationproblem and the model of service selection was established, and then QoS modelsof atom service and services composition were established. Secondly,MOAFSA4WSS algorithm was proposed to solve the model of service selection,and then artificial fish coding, construction of Pareto optimal solution set andadjustment of four basic behavioral were handled. At same time, finite Markovchain theory is used to analysis the algorithm in order to prove its convergence tothe Pareto front with probability1. Finally, MOGA4WSS and MOPSO4WSSalgorithms were designed in order to analysis the algorithm objectively andcomprehensively.Multi-objective fish swarm algorithm is used to solve the problem of Webservice selection and a Pareto optimal services set meeting needs of customer canbe generated in a limited iterations times, hence a new method to solve the problemof Web service selection is provided. Convergence proof of MOAFSA4WSSalgorithm not only provides a theoretical basis for application of multi-objectivefish swarm algorithm in other areas but also provides a reference for convergenceproof of other multi-objective evolutionary algorithms.