Continuous Distance Queries And Updating Over Moving Objects

In recent decades,moving objects databases have been widely studied in various fields such as traffic management,target tracking,and battlefield situation analysis.The main task is to manage spatial objects continuously changing their locations over time in the database.The continuous query of moving objects is one of the key queries in moving objects database.The difficulty is that the continuous query needs to return the moving objects that fulfill the query condition at each time point in a given time window.The paper proposes a new query called the continuous distance query of moving object trajectories and the corresponding updating technique.The performance evaluation over millions of moving objects are carried out,and the performance superiority and data updating efficiency of the proposed method are verified.The query can be used for the movement pattern analysis and search of urban vehicles,aircraft and animal movement and other moving objects.Specific research work is as follows:(1)Trajectory data preprocessing.The position of the moving object is acquired by the positioning device,for example,getting the longitude and latitude of the moving object by the GPS device.However,the position of moving object which got by device is not perfect due to the noise of positioning device,and there is erroneous data and positional deviation data.For the format error in the moving object position data,the dirty data is filtered by regular expression matching.In order to simplify the correlation calculation of the moving object position,the longitude and latitude of the GPS data are projected in the plane rectangular coordinate system.The noise data in the moving object trajectory is filtered by the filtering algorithm,and save trajectory data into the database,and the indexing mechanism of the trajectory data is established.(2)We propose continuous distance queries over trajectory data and develop efficient query algorithms.This paper presents continuous distance queries over large trajectory data,each query returns a subset of trajectories that away from the query trajectory between d1 and d2 in the time interval [t1,t2].The algorithm of continuous distance queries based on 3D R-tree was proposed,and the filter of trajectory data is also optimized.For the long trajectory,it is split into several shorter trajectories so as to improve the efficiency of continuous distance queries.(3)The bulk load method of updating both trajectory data and the index is proposed.The method considers how to effectively update the trajectory data and the 3D R-tree index when new trajectory data is added to the database,and efficient continuous distance queries are still supported while the trajectory data is being continuously added.The 3D R-tree bulk load method based on grid partitioning is proposed to improve the efficiency of bulk load update of 3D R-tree.The experimental results show that the 3D R-tree bulk load method based on grid partitioning has better efficiency than 3D R-tree bulk load based on Z-curve ordering.