The Design And Implementation Of The Road-matching Algorithm Based On Hadoop And The Vehicle Trajectory Tracking System

In recent years, with the rapid development of urban road traffic, traffic congestion, traffic pollution and other traffic problems are increased seriously. And they have become the main impediment of urban development. To solve these problems, many countries use the Internet of Things to build the Intelligent Transportation System (ITS). The Road-matching for large-scale locating data is the core technology in ITS. The traditional way which using stand-alone to store or process those mass data, is not only high cost but lows efficiency as well. It’s difficult to meet needs of ITS. Instead, as an open-source distributed cloud computing framework, Hadoop has great advantages in both storage and large-scale processing. Based on the study of typical Road-matching algorithms and Hadoop common components, a Road-matching algorithm which is suitable for large-scale locating data processing is proposed in this paper, and furthermore, it is migrated to Hadoop platform. The details of work are as following:1) In order to lay the foundation for the following study, the principle, advantages and disadvantages of typical Road-matching algorithms, as well as the architecture and operating principle of Hadoop common components are studied.2) Pretreated the experimental data, that is the7days GPS locating data of Beijing taxis and Beijing road data. The road network model according to the characteristic of the road data is designed, and Beijing digital map database with map grid is established.3) Based on Beijing digital map database, a grid searching based Road-matching algorithm which can speedily achieve candidate roads by calculating grid number is proposed, and the improved direct projection method is used to find the result. Furthermore, it is migrated to Hadoop platform by using HDFS and HBase to store input and output data, and running Road-matching jobs by MapReduce. 4) In order to show the trajectory of a vehicle visually for users, based on Beijing digital map database and match result database, the Vehicle Trajectory Tracking System which can display the digital map and the road-matching results is designed and implemented.5) Hadoop cluster is built, and the above Road-matching jobs is running on it. The results prove the efficiency of the algorithm. The Vehicle Trajectory Tracking System is used to compare pre-matching position with post-matching position, the results prove the feasibility of the algorithm.