| Trajectory data of moving vehicles becomes an important supplement to conventional traffic data. Using trajectory data to solve traffic problems is the background of this research. This thesis specifically deals with the management of trajectory data and seeks answers to the research question, "how to efficiently represent and access trajectory data." Correspondingly, the objectives of this research are to develop a trajectory data model and to develop a trajectory data access method. To fulfill these objectives, a LRS-based trajectory data model (LTDM) and a topology-based mixed index structure (TMIS) are developed. Given the importance of road networks, a trajectory-oriented, carriageway-based road network data model (CRNM) is also developed to provide the foundation for LTDM and TMIS. In order to integrate the developed approaches into the solution of real traffic problems, a tentative framework for trajectory data applications, namely, cooperative intelligent transportation system (CITS), is set up. Major contributions of this thesis consist of the following four points.; First, the CRNM provides network-based spatial references for location points of trajectory data. Based on the CRNM, the LTDM adopts a novel approach to select key points from location points. An experimental test shows that these models have a better performance than existing ones. These models, as extensions of geographic representation in the spatio-temporal domain, also compensate for the lack of capability of Geographical Information System for Transportation (GIS-T) to handle dynamic traffic features, e.g., moving vehicles.; Second, since the TMIS is based on the LTDM and CRNM, the number of spatial dimensions of trajectory data is decreased, which effectively reduces the complicacy of index structures. The TMIS consists of a number of small and classical index structures (e.g. R-tree and B-tree) instead of a huge and complicated index structure. These small index structures are linked by network topology. It is easy to implement and maintain the TMIS, and with different combinations of these small index structures, the TMIS can support more spatio-temporal query types of trajectory data than existing access methods. These ideas employed by the TMIS hopefully open a new horizon in building index structures of network-based trajectory data.; Third, the CITS, though still a conceptual framework at the current stage, can develop in a benign circle if being realized and can facilitate traffic management to a great extent. Especially, the proposed models and methods can be integrated into the framework and can provide the foundation for advanced applications, such as travel behavior analysis based on trajectory data mining.; Fourth, in order to avoid confusion, some concepts, including spatial attribute, aspatial attribute, spatio-temporal object (STO), point STO, region STO, spatio-temporal queries, etc., are redefined or extended into the spatio-temporal domain. An event-state analysis method is also developed to illustrate how the value of an attribute changes over time. These concepts and method may provide the foundation for relevant research in the future. |
