Research On Geographical Information Synchronizing In Grid-Gis
Jian Tan,Xiangtao Fan,Junjie Zhu,Xiaoping Du,Weibing Wang,Zhaoming Zhong,Chaoji Ma
DOI: https://doi.org/10.1109/EORSA.2008.4620325
2008-01-01
Abstract:Geographical GRID system is of great importance at fields as public security, military action, emergency response etc. The homogenizing distributed geographic environment system requires same geographical information for operations in each node. The bottle neck is how to reliably and accurately synchronize the great volume geographical data. This paper solves the problem in three ways. First, Message server queue is constructed for stable message delivery. In this way, the message server always has its alternative in preparation for breakdowns, and the whole GRID always has single working message server. Then the message server queue can be constructed and effectively woks. This mode has the advantages of the other two modes that the message delivery is more reliable and less time-costing. Second. Both push and pull modes are adopted to send messages in time. Push mode means the node which has altered its data is responsible for the delivery of the changed part, like "push" the data to the message server. While pull mode means the demand node or the message server is responsible to check the data status in other nodes and "pull" the new data from the source. In push mode, if the network between the sponsor node and the message server breakdown, the message could be missing or the sponsor could be halted, when the network resumed, the update action could not be invoked again. And in pull mode, the message server needs to check the data and collect update parts in the whole grid, it is a time-costing operation that could not be executed frequently. So the combination mode is adopted. In combination mode, not only each node has its own update trigger to invoke the delivery of the new data, but also the message server also can recurrently check the data status after an assigned interval according to the network situation and the computation ability, then the duly update can be guaranteed.Three, extended GML is developed to wrap the geographical data. GML defines a lot of types of elements and attributes to describe geographical entity in detail. But to synchronize geo-information in GRID-GIS, these definitions are not adequate. Because the spatial data must be wrapped into small flexible and linkable unit to cut down the time of delivering and receiving which are the most unstable periods in synchronizing course and to resend and assembly the units in unambiguous order. So our system developed the extended GML format, in which granularity level, including relation, inner string length are defined. By its help, the volume of data message is controllable and it is more reliable and accurate to resend and assembly the data fragments.These three methods are the key solutions to the geographical information synchronizing in GRID-GIS. Their validity has proved in practice.