A Research Of The Position Indexing Mechanism Of Moving Objects

As the mass application of mobile devices, wireless communication and GPS, a type of location based service was born, this location service provides the position information over time of a moving target to the enquirer, the database need to deal with constantly updating spatio-temporal data, while traditional database can only deal with data with relatively longer update cycle, and unable to respond to frequent updates, so the spatio-temporal database was born. In reality, most moving objects only moves in a limited two-dimensional way (i.e. buses travel on bus routes), rather than free movements(ships sails freely in the ocean), therefore it is more practical to study moving object database related technology, and to build up a high-efficiency index system is the key of moving target database. At present, the mature indexing structure can just only index the history track and real time position of the moving objects.On the basis of introduction of moving objects database and data model, this article analyzes on the traffic network model and moving object indexing technique。 Firstly considering traffic network topology, using roads as basic unit, discretely represents road information, learning from advantages from two commonly used indexing structure--FNR-Tree and MON-Tree, proposed a comprehensive indexing on road that has contains index to the all time position indexing of moving objects and supports nearest neighbor queiy.CRS is a two tier index structure, the upper tier builds up the index structure for the traffic network, the lower tier builds up the index structure for the moving objects on the road network. The upper road network uses roads as basic unit, builds a roads2DR树, to increase the quirying efficiency, introduce a roads_hash, we can locate the leaf node of R_tree through roads_hash directly, and won’t have to search the whole upper R树to find the corresponding roads, to increase the accuracy of prediction of future position and effectively support nearest neighbor query, introduce a turn_table at upper indexing structure。 The lower tier is composed of moving object R_tree, static object R_tree, mobj_hash and dynamic list, the node on every leaf of upper R_Tree points at a moving object R_Tree and a static object R_tree, moving object R_tree records all position information of this roads, and each static object points at a nearest neighbor list, to realize the nearest neighbor query from moving objects to static objects, to increase the efficiency of indexing the history track of moving objects, I introduced a dynamic list and a mobj_hash, as we can get the track information of the moving object directly through visiting hash table and dynamic list, without searching the whole forest of lower moving object R_Tree.In aspect of track querying of moving objects, use the number of times of access node as performance indicator, having compared the CRS indexing structure, MON-Tree and FNR-Tree, the result shows CRS indexing structure has the highest query efficiency. And In aspect of trajectory prediction and Nearest neighbor queries of moving objects, the results shows that CRS indexing structure has the highest accuracy.