Optimization Research On Web Service Composition Based On Qos

In recent years, with the development of Web Service technology and servicecomposition, the number of Web service with the same or the similar function increasesgradually, so Web service selection based on QoS has become key technology of Webservice composition filed. Each attribute of QoS use fixed weights in the existing QoScomprehensive calculation model, which completely ignore the relationship between eachattribute weights and actual value and can’t reasonable to accurate calculation Web servicequality. So the selected candidate Web services according to the comprehensive value ofQoS may not be appropriate service, which reduces the total quality of Web servicecomposition. At the same time the service selection algorithm in services compositionmostly adopt classical mathematical optimization algorithm, which may serious impact onthe efficiency of Web service composition. To resolve above problems, this paper has thefollowing research.Firstly, this paper improves QoS comprehensive calculation method. Aiming at theusers’ individual preferences of Web services, the improved model establishes statevariable weight vector to adjust the weight of QoS attributes in a single service, improvesthe objectivity and accuracy of the comprehensive evaluation of QoS in Web service.Secondly, particle swarm optimization suffers some shortcomings and defects. Themodified algorithm in this paper, adjusts the convergence speed by dynamic learningfactor strategy and variable inertia weight. Then according to test function, the parametersof improved algorithm by continuously simulation experiment so that it will be able tobetter adapt to needs of the function and update swarms position speed, and improve theexecution efficiency of service composition.Finally, according to the requirements and restrictions of users for Web servicecomposition, this paper design fitness function of QoS including target function andconstraints. Compare with the results of particle swarm optimization algorithm and theimproved algorithm. Experimental results show that improved QoS comprehensivecalculation method gets better result comparing with the whole quality of the existing QoScomprehensive calculation method.