Vehicle Trajectory Analysis System Based On Cloud Computing

Nowadays,with the improvement of transportation infrastructure,various monitoring devices collect information all the time,forming a large and diverse traffic data set.In addition,with the rapid development of Internet of things technology,traffic monitoring equipment is no longer an isolated node.However,traditional traffic data analysis is often based on the data obtained by a single monitoring device,which does not take into account the correlation between the data of each node,and lacks the linkage analysis of the overall data.Therefore,this thesis proposes a vehicle trajectory analysis system based on cloud computing.It introduces Storm real-time flow processing and Hadoop distributed processing framework to analyze vehicle trajectory data in real-time and off-line phases,and digs the internal relationship between trajectory data to identify illegal vehicles and assist traffic supervision.The research work of this thesis mainly includes the following aspects:1)Design a real-time vehicle violation detection scheme based on Storm.Combining Kafka message buffering queue with Storm to build a real-time processing platform.According to the characteristics of trajectory flow data and real-time calculation,a real-time vehicle violation detection algorithm is designed to quickly judge the overspeed and fake plate vehicles.In order to enhance the real-time performance of the system,a heuristic shortest path search algorithm is proposed based on the spatial characteristics of road network and Dijkstra algorithm.2)A lightweight cryptographic scheme with dynamic key is designed to enhance the communication security between traffic monitoring devices and servers.Under the conditions of ensuring real-time communication and without too much increase of the computing burden of monitoring equipment,the functions of bidirectional identity authentication,communication encryption and session key synchronization are realized.The security of the scheme is demonstrated by BAN logical analysis.3)Design a parallel vehicle history trajectory analysis scheme based on Hadoop.The vehicle trajectory splitting algorithm,the fake plate vehicles detecting algorithm and the lost point trajectory restoring algorithm are proposed.Combined with the above algorithms and the MapReduce computing model,a parallel analysis algorithm is designed for offline analysis of massive historical trajectory data.4)A vehicle trajectory analysis system based on cloud computing is designed,which integrates functions of real-time overspeed detection,real-time fake plate vehicles detection,vehicle trajectory restoration,trajectory analysis of fake plate vehicles,repair of lost trajectory and trajectory data visualization.The system is implemented and analyzed with real urban road data.Data analysis shows that the system can effectively realize the required functions and is suitable for processing mass vehicle trajectory data.