Research On Parallel Processing Technique Of Forest Distribution Information With Web Service

It's a paradigm for current web application to process data with spatial computing function in Internet. It can provide such characters as low-cost and maintainability. But for the complexity of spatial computing, performance of traditional serial processing in single machine is dissatisfying. The method of strengthen hardware is improvident and inefficient. This problem is serious in application of forest distribution information processing. Parallel processing with Web Service is proposed as a new solution in the paper.The solution includes such parts as parallel processing of forest distribution information, parallel technology in open network, application architecture. The first part is design and validate for algorithm of parallel processing for forest distribution information. The second part is parallel technology for web service in open network, which includes parallel computing system model, load balance, analysis method for application's Parallelizability. This part provides theoretical support for parallel processing for web service-based parallel processing of forest distribution information. Application architecture for software project is studied in third part to implement the application which includes net structure, layer model, principle for application implement, and function module is designed for application.The followings are achievement in the paper:The first, based on analyzing and summarizing on mechanism of mechanical drawing for forest distributing map, method is proposed for abstracting forest distribution information from data of sub-compartment, which comminutes the results coming from processing for fine-grained administrative regionalization.The second, characters such as heterogeneity, difficulty to synchronize, delay and communication cost is disclosed in the paper, and based on improvement of existing models, a new parallel computing model is proposed which includes central scheduling for load, distributed storage, asynchronous collaboration with message, and independently processing communication cost and delay by processor. Load balancing method is proposed which can schedule job adaptively in load status of node, and features are abstracted for parallel spatial computing service. A new method for load information renew in initiative measure is designed to resolve the conflict between time effect for load information renew and communication cost.The third, method to judge feasibility of parallel for application is advanced based on analyzing the relation of time cost for parallel processing and un-parallel processing.The fourth, architecture and system design are proposed for application of forest distribution information parallel processing. Architecture includes net structure with in-p2p, four layer model, implement principle with SOA.The fifth, according to demand of result on parallel theory and architecture, technology which build upon agent and multi-thread programming is advanced to implement parallelizable spatial computing service. It accords with the requirements of self-determination and administrabilityThe sixth, the primary technique such as service appoint in run-time, service requirement dynamically, notification, agent of Web Service, service chain have been studied in the end of paper. Prototype system are implement under system architecture.It can be concluded from the test that arithmetic of parallel processing for extracting forest distribution information is feasible. The architecture of application is viable. And parallel processing can cut down timing cost distinctly and improves performance greatly when data is deployed in each server.