A Research On Web Service Composition Based On Nonlinear Programming Optimization

With the rapid development of Web services, Web-based service compositionQos Web optimization has become a hot topic. Web service composition is NP-hardproblem. A typical Qos attributes are associated with the Web service response time,throughput, arrival rate and so on.Research on the optimization of Web servicescomposition based Qos issues can be more fully meet user requirements for quality ofservice, which is of great significance for the development of Web services.This paper describes the current Web services for solving combinatorialoptimization problems several commonly used algorithms. There are geneticalgorithms, particle swarm optimization, and so on. Genetic algorithms are widelyused in solving optimization problems in Web services, but most rely on theirexperience to select the parameters, can not guarantee the validity of the solution,and the optimal solution in the search process is easy to fall into local convergence.Ant colony algorithm global search ability, but it can not be directly applied to theactual model and optimization of Web services composition.This article will be mapped to BPEL service composition process queuingnetwork using queuing network theory for response time, throughput and arrival rateQos attributes are analyzed, and the data is calculated using Mathematica. The resultsshow that the method has a strong practical.Contents of this paper is in a given Web service using nonlinear programmingalgorithm preferred a set of Web services, making this group of services under thepremise of satisfying the constraints of the total cost to a minimum. Based on queuingnetwork theory, the sequential structure BPEL, BPEL BPEL loop structures andconditions for the queuing network structure mapping, the use of the characteristicsof queuing network throughput under three structures of Web services is calculated,and then solve the response time on the Web server node the expression service rates.Through the response time and service rate of each node is set constraints, anonlinear programming model for service rate in Mathematica nodes to solve it. Ratewill be solved through the service with a given service rate compared to Web services,to solve the service selection and the smallest difference in the rate of a set of Webservices, is seeking herein optimal Web service composition.