Research On The Trajectory Models, Index Structures And Queries Of Moving Objects

With the rapid development of wireless communication technology, satellite global positioning systems and geographic information systems, it is possible to track and record the location of Moving Objects. Moving Objects Databases will accept a new challenge, which is how to effectively manage and query Moving Objects, and to provide precise location-based services.The research aim of this thesis is to establish the whole trajectory model of representing the Moving Objects position, and to solve the technology of Moving Objects on it, which is the trajectory update, the predicting policy, the index and the nearest neighbor query, and to propose the feasible scheme of solving.Because of the limitations of the previous modeling Moving Objects trajectory methods, which can't dispose the past, present, and future positions of Moving Objects, it is one of the questions be solved how to establish the whole trajectory model, which will highly efficiently support the past, present and future information processing. This thesis combines the ideas of the discrete modeling, and proposes the whole trajectory model, which is on the basis of the Moving Objects Spatio-Temporal (MOST) model and supports the past, present and future information processing. By adopting the linear interpolation of the based-point modeling method, the Moving Objects Spatio-Temporal model will dispose the historical information. It can dispose the present and near future information. To dispose the further future information queries, it will adopt the predicted-trajectories methods. The experimental results with theoretical analysis indicate that the whole trajectory model of Moving Objects is reasonable and feasible, which can support the queries of Moving Objects efficiently.In order to effectively query the Moving Objects data, it is necessary to adopt effective index technologies of Moving Objects. Most of the previous index methods can dispose the past and present information. However, the index methods are less and their efficiency is low, which can predict the future position and support the past, present and future information processing.This thesis comprehensively analyzes the advantages and disadvantages of the primary index technologies of Moving Objects from the perspective of the data disposal methods of Moving Objects, and further optimizes the index structures. It improves the query efficiency of Moving Objects Databases and reduces their frequency of update for index structures. By improving the TB-tree and combining the TPR*-tree, it also proposes the index structure of TB+_TPR*-tree, which supports the past, present and future information processing on the whole trajectory model. The effective analysis of theory ensures the correctness of the TB+_TPR*-tree. The experimental results indicate that it is feasible.Because the nearest neighbor query processes the huge amount of data, frequently querying will consume a large amount of time and space and impact the efficiency of querying seriously. Therefore, the algorithms which can dispose effectively a large number of querying nearest neighbors for Moving Objects are particularly important.By adopting a predicted-speed policy and a bottom-up update R-tree with an update first stored in memory, this thesis improves the querying method of k-nearest neighbors on the whole trajectory model of Moving Objects. When the speeds or routes of Moving Objects are changed, the immediately updated location information will be firstly stored in an update list of memory. Then, when the update list reaches the maximum value, the R-tree will be updated. By effective analysis of theory and experiment, the method reduces the disk accesses effectively, and also enhances the efficiency of queries.