Research On The Choice And Combination Of Web Services Based On QoS

With the continuous improvement of computers and communication technologies,cloud computing has developed rapidly.The development of cloud computing not only provides convenience for people's work and life,but also saves public resources and improves resource utilization.Cloud computing not only provides users with unlimited storage services,but also combines single-function services to solve complex problems for users.Therefore,the research on the choice and combination of web services based on QoS(Quality of Service)in cloud computing has been a hot topic.At present,the main problem in the selection and combination of web services is a large number of candidate services for a single node.In addition,the selection algorithm is easy to fall into local optimum and runs for a long time.There are many non-functional properties of web services.There are no uniform specifications and standards to measure web services.The system cannot select a service combination based on the user's preference.The results are not satisfactory to the users.In view of the above problems,this thesis studies the basic service combination model and service selection algorithm.This thesis proposes a QoS combination model based on user preference to measure user preferences.The improved model converts the user's preference for attributes into specific weights through the judgment matrix.A specific fitness formula is obtained by combining the corresponding attribute values in the actual workflow.Particle swarm optimization has the advantages of fast convergence and few parameters,so this thesis chooses particle swarm optimization as a service selection algorithm.Improves for the problems of particle swarm optimization,such as premature convergence,easy to fall into local optimum and low accuracy.Dynamic inertia weights are added to improve the accuracy of particle swarm optimization.Gaussian perturbation and Levi's flight are added to improve the problem of premature convergence and local optimality of particle swarm optimization.For the improved algorithm,this thesis first uses the control variable method to optimize the parameters of the improved algorithm.The improved algorithm was simulated in Matlab,and the experimental results were compared with the other four algorithms.From the experimental results,the service fitness value of the improved algorithm is better than the other four algorithms under the same running time.Under the same number of iterations,the improved algorithm is longer in running time,but it has also achieved better fitness than the other four algorithms.With the increase in the number of candidate services,the improved algorithm proposed in this thesis is more effective.In summary,the improved algorithm is practical and effective.