Research On The Services Composition Based On Granular Computing And QoS Restriction

With development of CPS, Web of Things, Cloud, Grid and Web, the service isplaying an increasingly more important role in the field of IT. Considering the fact thatthe number of services is growing and the kinds of services are expanding, this paperfocuses on the services organization and composition based on the Service–OrientedAchitecture (SOA), by means of which integrates previous available services to form anew and value-added composite service. In order to make sure the efficiency,requirement and quality within the services composition, the research on servicescomposition based on granular computing and QoS restriction is made. The main worksof this paper are as follows:(1) The Services Composition Workflow Model–PRR is proposed.Based on workflow modeling techniques, the Services Composition WorkflowModel–PRR (Process, Resource and Requirement) is proposed. With the relationshipsof process, resource and requirement, the framework of services composition isprovided, where the Process Model is the core, the Resource Model is the base and theRequirement Model is the aim. And then, the definitions and descriptions of the threemodels of PRR are put forward, among which the Resource Model shows the basicdescription, function description and QoS description for the service, the RequirementModel shows the function requirement and QoS requirement for the user, and theProcess Model provides the activity network graph for the job.(2) The services organizational mechanism and adjustment strategy are provided.Based on granular computing theory, the organizing and management for theservice space are made in this paper. By means of the Job-driven Generating Method,the service granule is generated with the function and QoS requirement as its externaltoken and the service set as its internal element. According to the rules of granulation,compostion and hierarchy, the hierarchical service granules architecture is built, and theservice granular computing based on quotient space theory is given. Furthermore, thedynamical adjustment strategy for the service space is designed in case of the generatedservice granule, the addition of new service and the removal of existent service, in thisway the timeliness and orderliness of the service space are guaranteed.(3) The service selection and optimization with QoS restriction is designed.On the basis of services compostion workflow model and hierarchical servicegranules architecture, the execution of service composition is made in this paper. Withthe method of iterative computing, the Composite Service QoS Model (CSQM) isproposed, explaining the measurement and computing of the QoS forrequirement-related composite service. And then the Services Composition ProgressiveComputing Algorithm (SCPCA) is designed to do the execution of service composition with two phases: one is the service selection and another is service optimization. As forthe service selection, four service granule match modes are defined, and the servicegranule-oriented selection algorithm is designed. When it comes to the seviceoptimization, the algorithm based on improved PSO is put forth, which guarantees theQoS of the composite service.In the end, a series of experiments are carried out to validate the effectiveness ofthe proposed method in terms of efficiency and effect.