Research On Index And Query Techniques Of Moving Objects In Spatio-Temporal Databases

Advances in global position system, sensor network and wireless communication technologies have made it possible to track and record the positions of moving objects which continuous move with time. Many applications, such as, traffic control, meteorology monitoring, mobile computing, location-based services and digital battle, propote the generation and development of spatial-temporal database and moving object database. In recent years, spatial-temporal database and moving object database have emerged as an active topic in database community, and received considerable attentions from both the research area and industry area. The technique of indexing and querying moving objects is the key challenge that the spatial-temporal database and moving object database face. However, the spatial-temporal database and moving object database have not achieved industrialization and been used in real life, it requries deeper research. In addition, there exists a gap between the oversea research level and that of domestic in this area. Therefore, research on the techniques of indexing and querying moving objects has important academic value and extensive applications.Motivated by this and according to the analysis of disadvantages of the exsiting work, this dissertation studies the techniques of indexing and querying moving objects, aiming at enhancing the performance and intelligentize of moving object indexes, providing new queries of moving objects, and satisfying the requirement in real circumstances of moving objects. This dissertation studies the management of spatial attributes and other kinds of attributes over moving objects, and does some innovative research work as follows:(1) Research on the timeliness of moving object index:based on the analysis of the timeliness problems and influences in the real circumstances of moving objects, and in order to deal with these timeliness problems, this dissertation inctroduces the concept of "prediction life period" to the module of moving objects, and handles it in the process of indexing and querying. In addition, this dissertation allows the "prediction life periods" to be set uncertainly, and introduces corresponding probabilistic queries. In this way, this dissertation solves the timeliness problems and influences brought by moving objects with highly variable update frequencies.(2) Moving object indexes with high query performance:this dissertation proposes new indexes of moving objects, By-tree and aBy-tree, which considerablely improve the query performance, at the same time of ensruing the efficient update performance. Based on the proposed indexes, this dissertation designs and implements efficient query algorithms of range query, KNN query and continuous query. Extensive experiments show that the proposed indexes and query algorithms outperform the exsiting work significantly.(3) Self-tuning index framework of moving objects:motivated by the dynamic circumstances in which the requirement of update operation and query operation varies with time, this dissertation proposes a self-tuning index and a online tuning framework SIFUQMO, which can strike a balance between the performance of queries and updates, thereby achieving good and smooth overall performance for different proportion of updates and queries, without breaking the index service. Extensive experiments show that it is able to work well in these dynamic applications, and its overall performance is better than that of the exsiting work in different circumstances.(4) Predictive skyline queries for moving objects:based on the analysis of the spatial attributes and non-spatial attributes, dynamic attributes and static attributes of moving objects, this dissertation defines and handles the problem of predictive skyline queries for moving objects, including point qury, range query and subspace query. This dissertation proposes a new uniform index structure which indexes all kinds of attributes, and several query algorithms to handle the problem. Extensive experiments show that these algorithms are efficient, and compare their performance.