| Due to the economic boom and rapid urbanization in 21st century, the number of vehicles is growing very fast. The growing traffic demand and the limit number of road infrastructure become a big problem. It becomes very demanding to build a dynamic road traffic information system. A dynamic navigation based real time traffic information system includes two main parts:the real time traffic information service and dynamic navigation. These two parts are also very important research area of LBS, ITS and GIS-T. The above-mentioned system can provide real time traffic information, traffic predict and route planning service for users. The users can make use of these updated real time information to plan their trips in advance, and make changes according to real time traffic condition. The information system can make the trip safer, more reliable and more convenient.Based on the related research program and related work experience, this thesis is focus on the theoretic research of real time spatial data modeling, especially the dynamic traffic information modeling and data organization. Moreover, researches have been done on the key techniques for application, including the storage of high volume real time traffic data, the distribution of traffic information and the design of dynamic navigation system. The theoretical contributions of this thesis are based on the working experience of the author and form a good foundation for future research and application building. The main research contribution is as follows:1) The navigation data modeling is reviewed:Navigation data modeling has been evolved from planar model to non-planar model and from road lane based model to traffic direction based model. The LRS model, MDLRS model and the unified traffic network model are all linear referencing based GIS-T data models. These three models are designed for traffic information management and they are not capable to support the key functions of dynamic navigation service. Moreover, the key techniques for dynamic traffic information storage, coding and distribution for real time traffic information service are reviewed.2) The Multi-Level, Multi-Dimensional Reference Navigation Data Model (MLMD-NDM) is proposed. The proposed MLMD-NDM model is divided into three levels:the real time traffic information level, the spatial road network level and geographic information infrastructure level. The traffic features are the abstract of the various traffic data in the traffic information system. These features are composed by spatial object, time object, theme object and event object. The model is built in a multi-dimensional referencing space, which is built based on two-dimensional linear referencing system by the relative importance of the features. In addition, the model gives a detailed description of the data objects, the geometric and topologic road network data of each level.3) The modeling and organization of traffic information, traffic event and moving objects are proposed. The proposed model is able to integrate road network information with traffic information at the geometric level, the semantic level and the data structure level. The model provides the basis for dynamic traffic information data management and application. The traffic data modeling is based on the spatial and time features of the road networks. A data structure is designed for the storage of real time traffic data status and estimated traffic time, which makes the information fusion between the real time traffic information and road networks. Moreover, an adaptive and self-organization table is designed to store traffic event location information, which is based on linear referencing and dynamic segmentation techniques. An information update model that is suitable for the storage of moving object is built based on the road network and the linear referencing methods.4) The Sum Square Error (SSE) and Cross Validation (CV) are proposed for sampling and storage traffic information data. The SSE and CV are proposed to manage large volume of dynamic traffic information. The trajectories of moving objects are compressed based on the characteristics of the road networks and traffic speed.5) A road segment/node based traffic information location referencing system is proposed by integrating the node location referencing, segment referencing and linear referencing methods. Based on the current condition of domestic traffic information capturing and distribution characteristics, an applicable real time traffic information distribution protocol is design. The protocol is supported by several data distribution methods and has been taken into practice. The proposed protocol can be served as the basis and beginning for domestic traffic information standardization.In the last chapter, an experimental system for the real time dynamic traffic information service is designed. Detailed information about the system framework, experimental data flow, database design and traffic information distribution is given. Real time traffic information based dynamic route planning and Multi-Level Multi-Dimension traffic data visualization is realized on cellphones and navigation terminals. The key functions of the experimental system are tested. The results prove the capability and usage of the model. |
