Research And Implementation Of Intelligent Traffic System Based On Crowd Sensing

With the improvement of the people's quality of life and the increasing of the number of vehicles,the problem of traffic jams is becoming more and more serious.Traffic jams not only result in the increasing of travel time costs,but also cause the waste of resources and even environmental pollution.However,in China,the research on the Intelligent Traffic System is still in its initial stage.Further,it is of great significance for the study of the traffic system.Based on the functional requirements of the Intelligent Traffic System,this dissertation studies and implies the Intelligent Traffic System based on Crowd Sensing.The main contents are as follows:(1)This dissertation designs and implies the Intelligent Traffic System based on Crowd Sensing were completed.The functional modules of the system are designed in detail,and the system has realized functions such as real-time road condition delivery,carpool friend recommendation,and carpool communication.Finally,the system is tested in terms of both function and performance.The test result shows that the system can conform the users' requirements well.(2)In the key technology research of the system,aiming at the shortcomings of the old methods that the high cost of getting the road condition and the lack of flexible,this dissertation proposes a scheme based on Crowd Sensing which uses smart phones to collect road information and determine the road conditions.At the same time,the new quantitative analysis index TAP is proposed.Through this index,the applicability of different models to this proposal can be judged.After qualitative and quantitative analysis,XGBoost in the three models used in the experiment is the most suitable for this program.Moreover,the classification accuracy of method based on XGBoost can reach more than 90% that meets the needs of road classification well.(3)For storage and processing of path information,this dissertation considers the server needs to process and store a large amount of users' information.A combination of MapReduce and HBase solutions is designed to solve the big data processing and storage problems.(4)At the same time,in order to against the shortcomings of the path matching algorithm,this dissertation studies the Circle-based algorithm and improves it.It eliminates the blind spot area of the original algorithm and improves the accuracy of the algorithm.