| In the medium to late 1990's there came forth a new computing environment which is termed as GRID nowadays. The GRID, with the aim to collaborate the share of geographically distributed computing resources beyond different organizations, is compared with power grid. Now GRID computing environment is developing quickly in its own way. At about the same period, in the GIS field, a tide to turning up GIScience is prominently. And spatial analysis, said to be a key content of GIScience, takes the emphasis from many scientists. To study the implementation of spatial analysis in the GRID computing environment, this dissertation firstly tries to bring forward a framework for spatial analysis from the computing aspect. The dissertation also takes the research of 3 cases to explore the computing model suit for solving in the GRID computing environment the tense computation spatial analysis problem, multi-organizations involved spatial decision and multi-organizations involved spatial simulation problem.This dissertation is divided into 6 chapters. In the 1st chapter, the progress for GIS to GIScience is discussed briefly to show that spatial analysis is important to augment the understand of geographic information, and to illustrate the recognition of computation in geography circles. The GRID computing environment is also discussed in detailed to find out that it can be classified into computational GRID and cooperative GRID. The characters of these two type of GRID is also discussed.In chapter 2, the concept and content of spatial analysis is explored fully with review on many scientists' viewpoints. A computational system of spatial analysis is brought forward based on the difficulty to solve a specific spatial analysis problem. The difficulty, is divided to 2 aspects, one is computational complexity and another is the number of involved organizations, for the more organizations involved, the more difficult to deal with all the data and computing. We call the later multi-organizationality. Finally, spatial analysis is classified into 4 types: low tense computation few organizations (1st type), high tense computation few organizations (2nd type), low tense computation many organizations (3rd type) and high tense computation many organizations (4th type). It is argued that GRID computing environment can help in solving the non-1st type of spatial analysis problems.In chapter 3, a case of high tense computation spatial analysis problem is chosen to be parallelized on a computational GRID built up with CONDOR. The case is a P-Median problem, which is NP-hard, and approximate resolution approaches are often used. 2 parallelization strategies are studied to argue that the scatter search strategy is much suitable for the master-worker paradigm on the loose-couple distributed environment of computational GRID, in which the computation nodes are very dynamical and not always well connected. The development of a task management middleware and a task monitoring middleware is also discussed in this chapter.The 4th chapter, which contains a case study of a spatial decision with many organizations involved,argue that in generally spatial decision is a process that spatial data are transformed and collected form many sites to form the final decision. The many organizations are due to provide spatial data services and spatial analysis services. Hence, a computing model termed collective computing is advanced. In this model, the way to provided spatial data and spatial analysis functions is to set up Web services, and an extended GML is designed to descript the user's spatial request.Chapter 5 is based on the case study of a spatial simulation with many organizations involved. It is argue that spatial simulation is different from spatial decision, for the organizations don't serve the user, but serve together. Every organization provides data to the other and gets information vice versa. They collaborate. We call this collaborative computing. Each collaborator is a peer. In the chapter, the case is a MAS to simulate the develop...
|
PDF Full Text Request