Research On Real-Time Performance On Distributed Marine Environment Monitoring And Information Publication System

China is a developing country with a large area of ocean. So it concerns deeply on the development of economics, politics, national defense, and culture that marine resources are exploited and utilized properly and legally. Along with massive and excessive development of marine resources, however, marine calamities such as red tides emerge more frequently as a result of continuing marine pollution activities, which caused serious impact on our marine ecosystem and huge loss on economics directly or indirectly. Thus it has pressing needs to devise marine monitoring and calamities publication system and research on its real-time ability and reliability, which will help maintain a proper and health marine development environment.Originates from research funds on marine monitoring of Shandong Province, this thesis mainly concerns on two subsystem of the whole project: monitoring and information publication subsystem. The research focus on techniques and strategies used to enhance and guarantee system real-time ability, and supply effective marine calamities monitoring.The thesis first analyzed the background and research achievements on marine monitoring and publication system, then introduced its main research topics and content construction. And architecture design and key techniques analysis of the two subsystems was presented subsequently. The whole system employs a three-tiered architecture, which includes monitoring sites, station sites and data center. Monitoring sites which distributes over target ocean area are responsible for data collection, and they belong to one unique station site at one time, where data from them are inter-graded. The two types of sites connected with wireless communication network and mined collected data using distributed clustering algorithm under the supervision of data center. Meanwhile these marine data are sent to data center for further mining and publication.As to monitoring subsystem, the thesis mainly considered how to promote real-time performance on data collection and encapsulation, which include techniques like wireless network communication, GPS location service, data encapsulation and access using XML, etc. Information publication subsystem introduced is based on station sites, which lies in the central role in the whole monitoring system. In order to implement multi-target and multi-form publication, a distributed subsystem architecture was constructed, and so as to the detail module division. The proposed publication platform provides multi-form publication services to multi-type targets such as monitoring subsystem, Internet explorer and cellphone. Among the modules of the platform, the performance of embedded GIS publication platform plays the key role on the performance of the whole information publication subsystem. Thus the thesis continued with a survey on GIS system on linux platform, and described its real-time architecture and class diagrams after analysis of platform design requirements, then researched on the techniques and strategies platform employed to promote real-time performance and reduce system storage.The thesis finally gives two real-time clustering algorithms under different application environments. One is cater for the need of distributed data clustering in station-based publication subsystem, which improved the traditional algorithm architecture based on corset technique, thus effectively reduced the data communication amount among sites and accelerated information clustering. Another one aims to fulfill real-time data mining and history database update requirements on data center. The proposed new incremental algorithm based on entropy theory has a good performance in its application of the clustering mining task on data center, which being proved by simulation experiment results.