A Research And Implementation Of WebGIS Based Multi Agents

Under the drive of network globalization and information digitalization, Geographic Information System (GIS) based on Internet as a promising research field brings new opportunities to the ways of geographic information' s accessing, sharing and disseminating. However, the inherent limitation of current Internet GIS architectures based on connection makes them incompetent to fulfill various requirements of the coming GIS applications under dynamic, complicated and distributed computing environment such as the Mobile Internet. Seeking for new distributed computing modes of networked GIS is an important research area for sharing effectively and real timely the global distributed geographic information. The technology of Multi Intelligent Agents is important way for developing distributed intelligent system. Mobile agent as a recently developed paradigm in distributed computing can especially benefit WebGIS on providing asynchronous task execution, reducing communication bandwidth, enabling off-line processing and disconnected operation. Using XML/Java can overcome the problem of transmission and representation on WebGIS. Considering the advantages of WebGIS, Multi Agents technology and XML/JAVA standard, we propose a new paradigm in constructing WebGIS with the combination of both technologies, and further explore the related issues of its principles and techniques. Major contributions of the thesis include:1. A new paradigm of distributed WebGIS based on multi agents is proposed. The architecture of our MAGIS is presented, which is based on the techniques of SVG/XML Java and Multi Agents. The architecture is adaptable to Internet especially for Mobile Internet application environment with limited bandwidth and vulnerable connection, in which Multi agents are the kernel technology and makes it the cornerstone of constructing WebGIS. Agents are used to overcome the drawbacks of distributed systems on Internet2. Based on Bayesian Multi Intelligent Agent is built in MAGIS. HCI Agent, JavaBayes agent and JESS expert system agent are built. The relation of them is set, too. The decision system designed on Multi Intelligent Agent and data mining is built four level agent.3. Key techniques related to mobile agent (or GISMAgent, which is the mobile agent in MAGIS) building in MADGIS are explored comprehensively. Docking system is advanced, it can overcome traveling defeat effectively.4. The application strategy on MAGIS of SVG/XML is advanced. The MAGIS model Based SVG/XML is built. The implementation strategy for each user is designed.5. The security model and some security services of MAGIS are offered for hosts and GISMAgents protection.6.A prototype of MAGIS is developed based on Multi Agents. The successful running of the prototype in China grain depot information and intelligent decision system shows that MAGIS is feasible and effective.