The Research Of High-performance Geo Computing Engine Based On CSP And Workflow

With the rapid growth of spatial data, the serial processing paradigm of traditional Geographic Information System failed to meet the requirements of geocomputation. Using high-performance geocomputation to solve complex geographical problem became a trend. In order to learn how the geocomputation job process operate in high-performance environment correctly and efficiently, the following aspects are discussed in the paper.First of all, based on directed acyclic graph, the paper describes the geocomputing processes, and analyzes the dependencies of jobs in the processes. Based on high-performance computing,the paper designs a geocomputing services framework with REST style.Secondly, the paper makes a modeling for high-performance geocomputing service based on communicating sequential processes. The model describes the relationship of jobs because of the dependence in the processes, and adds user's controllable operations. The paper validates the model of geocomputing service with CSP tool.Thirdly, this paper designed a method for processing concurrent processes based on multithread. As the core technology of the executing engine for the geocomputation job process, this technique provides support for parsing process model correctly, parallel execution geographical job and full use of system resources. In addition, some experiments are designed to validate the geocomputing engine.Finally, a prototype system of parallel executing the geocomputation job process is designed and implemented. Through showing an application demo of calculating suitable areas for housing, the geocomputing service has good practical performance.