Vehicle Trajectory Data Management And Analysis Based On HBase

The efficient management of spatiotemporal data is of vital importance in the field of spatial data mining, particularly in the modeling and analysis of spatio-temporal data. As a common type of spatio-temporal data, vehicle trajectory data can not only represent the complete vehicle path but also reflect the traffic condition directly. Moreover, it indirectly illustrates the geometric characteristics of the road networks. With the development of various positioning technology and the Internet technology, positioning devices such as GPS and BDS(BeiDou Navigation Satellite System), are widely equipped on vehicles, which makes it greatly easier to obtain the vehicle trajectory data and harder to manage the explosively growing vehicle trajectory data. Thus, new methods of vehicle trajectory data management and analyzation are urgently needed to be proposed.Traditional stand-alone databases are trapped in troubles like the high cost of the maintenance and upgrading. More significantly, huge amounts of data cannot be well processed in this mode. This thesis firstly introduces Hadoop, which is known as an open source distributed ecosystem, and the Hadoop distributed file system (HDFS) as well. HBase is then introduced based on HDFS as a distributed column database to manage the vehicle trajectory data. The demand of application analysis must be taken into consideration when organizing data. Travel time prediction is one important research content in ITS(Intelligence Transport System). As an effective transport information, travel time prediction results may assist traffic management department in adjusting traffic flow through time-dependent rules,and assist travelers in optimizing travel plans. And this thesis presents an encoding approach of travel origin and destination for travel time prediction that adapt to distributed database which based on Geohash, and verified the validity of this encoding method by experiment.Then, the HBase distributed storage scheme for vehicle trajectory data with such encoding as rowkey index is studied, and the database import method of mass data is optimized. Finally,the data query efficiency of the HBase distributed storage scheme and MySQL is tested. The HBase distributed vehicle trajectory storage scheme is more efficient in the concurrent network environment.In order to verify the validity of encoding method of the vehicle trajectory data OD and vehicle trajectory data distributed storage scheme, this thesis takes the taxi trajectory data of Chengdu as an example, using the encoding method above to encode the data. And then, the travel time is predicted. The experimental results show that the encoding method proposed in this paper is easy to use and feasible, and the distributed storage scheme is efficient and has good prediction results.