Research On Grid Service Evaluation And Composition

The ultimate goal of grid is to elimate the isolated information island and to realize the real sharing of resources. Surpported by Henan Research Program of Application Foundation and Advanced Technology–Dmain in Adaptive Grid Serive Composition Algorithm, aiming at the grid service evaluation and composition, the study aspects are as follows.(1) After introducing the concept of the grid, the thesis analyzes the characterisitics of the grid-distributivity, sharing, dynamic, diversity, collaboration, and so on. The grid architecture and especial OGSA (Open Grid Services Architecture) are introduced and discussed.(2) Based on the detailed analysis on the service description model and the appraisal concept tree, the evaluation procedure based on BP neural network is presented. The reasonable evaluation for the grid service can be gotten through the establishment of BP neural network model of service quality, which lays the foundation for the high quality service. Simulation results show that the grid service quality evaluation method based on BP neural network through a highly nonlinear mapping, reveals the essential relationship between the grid service evaluation result and the related factors, had the high rationality and applicability.(3) Grid service composition algorithms based on the domain template are discussed. The fundamental concept of the domain template, template generation method, template matching method and automatic updating method are analyzed. The main structure of the service composition algorithm is proposed. The procedures of the domain template matching algorithm, the matching similarity algorithm for the input parameters and the auto updating algorithm based on fuzzy clustering are discussed and analyzed. Grid service composition algorithm based on the domain template can provide the transparent composition services in the business layer for users, which can satisfy the dynamic requirement.