Research On Moving Region Oriented Moving Objects Database

Moving object database is a complex system related to manage the positions and extents of moving point and moving region. With increasing spatiotemporal applications based on moving region such as fire simulation and tsunami monitoring, it's very necessary to research on moving region because current research only focuses on the location management of moving objects. This dissertation makes a systematic and thorough study on key issues on moving region oriented moving objects database.The spatiotemporal semantics of moving objects are firstly studied, which are the basis to construct data models and even to the whole research on moving objects databases. A systematic classification on spatiotemporal changes is presented in the dissertation and a new complete approach to the description of spatiotemporal changes is propounded, based on which a spatiotemporal semantic model is further designed. The new approach to representing spatiotemporal changes, which is based on object identity and descriptors, describes spatiotemporal changes using object-level spatiotemporal changes and attribute-level spatiotemporal changes. The former is explicitly represented, while the latter is implicit.Inherited from the research on spatiotemporal semantics, a new data model named GPM (Grid Partition Model) is presented, which is based abstract data types and constraint database theory. GPM is founded on the object-relational data model. In GPM moving objects are represented as spatiotemporal relations and spatiotemporal tuples, and queries on spatiotemporal relations are implemented through an extended relational algebra. Spatiotemporal data types and development topology types represent the spatiotemporal data and changes of moving objects. Queries on spatiotemporal data and changes are supported by the operations defined on these data types. The data structures and algorithms for modeling the irregular extents of moving region are discussed in detail. Query operations and language are also involved. Several parameters are proposed to stand for the precision of modeling. Experiments results prove the effectiveness of modeling method.To support spatiotemporal queries based on GPM and estimate the disk cost of predictive queries, the dissertation proposes cost models for spatiotemporal selective and join queries based on TPR-tree structures. The cost computation of the models is only relative with the properties of datasets and TPR-tree, and need no knowledge of the underlying TPR-tree structures and data distribution, so the computation of cost will be reduced. Experiments results on simulated datasets show that the cost models are effective for estimating the cost of spatiotemporal queries. Based on the proposed cost models, TPR-tree and grid file, hybrid index structure is proposed to support spatiotemporal predictive querying of the current and predicted future extents. Experimental results performed on simulated datasets show that the proposed index structures are effective for querying the extents of moving objects.On the basis of GPM and spatiotemporal query, the approaches to the implementation of moving objects database management systems are discussed. As a result, the GPM-based STDBMS on object-relational database management systems prototype was developed. To support complicated spatial operations, we propose a new method that is founded on counterclockwise directed triangle, and the algorithms for complicated spatial operations are then defined too. Through the detailed analysis on the implementation processes and results of the GPM-based database management system, it is observable that the GPM model is easy to be implemented.Compared with previous researches, the new contributions of the thesis are:1. Systematic and complete analytic methods and implementation framework are proposed to research on moving regions oriented moving objects database. The Studies include spatiotemporal semantics, data model and modeling method, spatiotemporal query and index structure and the implementation prototype of moving objects database. The work completes the current research on moving objects database.2. Proposes a new approach to the description of spatiotemporal semantics. In the new approach the explicit object-level spatiotemporal changes and the implicit attribute-level spatiotemporal changes are synthetically used, so the approach can overcome the shortcomings of previous researches that lack in completeness and systematization such as thematic continuous change, and can guarantee the complete description of spatiotemporal changes in theory.3. Presents an innovative data model GPM, which is based on abstract data types and constraint databases theory. GPM can represent and query various spatiotemporal changes effectively. Meanwhile, because GPM is founded on type extension, so it can be transformed into and implemented on object-relational database management systems and can overcome the difficulties of implementation that exist in previous spatiotemporal data models. In GPM the discrete and continuous representation of moving region are modeled by unit grid.4. Proposes cost models for predictive spatiotemporal queries and hybrid index structure, which completes the research on moving objects databases.5. Proposes an implementation prototype to model the moving region and an improved approach to implementing spatial operations, which is based on counterclockwise-directed triangle. Different types of spatial operations can be implemented effectively founded on this approach. The approach avoids the influence of efficiency when to compute spatial operations through the list of vertexes directly.