Research On Recovering Trajectories Of Urban Private Vehicles Based On Crowd-Sensing

In the urban road network,the travel path of urban private vehicles contains rich travel information.Through the analysis of travel path of urban private vehicles,we can get abundant information about the transportation on the urban road network,thus providing new ideas for the government and transportation department to collect valuable traffic information.However,due to the privacy of urban private vehicle users,only a small number of owners will share the exact location of their vehicles.The path information of most urban private vehicles can only be acquired from the sparse road surveillance camera.It is very hard to recover the accurate trajectories of urban private vehicles only by relying on the data of sparse road surveillance cameras.There has been no proper algorithm to solve the problem of recovering the trajectories of urban private vehicles.Some existing vehicle trajectory recovery algorithms are to mine the potential information and to recover the trajectories of the vehicle through vehicles' trajectory history.However,as the data of vehicles' trajectory history do not contain sufficient information,the trajectory recovery rate normally cannot meet people's requirement of the precision rate of trajectory recovery.Based on the above analysis,this thesis proposes a solution to recover the urban private vehicle trajectories based on crowd-sensing.First,through mining the data of sparse road surveillance cameras and the accurate trajectory data of a small number of vehicles with online telematics systems,this thesis analyzes the traffic patterns in the region and builds modeling of the travel time-cost of path with exponential distribution.Then,we use the Canopy-Kmeans clustering algorithm to cluster the sparse trajectories of urban private vehicles.The results of clustering optimized the road time model and laid a foundation for improving the trajectory recovery accuracy of urban private vehicles.In the end,this thesis puts forward an algorithm with the maximum utility value of data packet to transmit the data packet stored on vehicles.When the vehicle passes the intersection of the Wi-Fi packet upload point,the Wi-Fi packet upload point will collect and transmit the data to the comprehensive data analysis platform of the background.Through the analysis and mining of the collected data by comprehensive data analysis platform,the accurate trajectory information of all urban private vehicles can be obtained.This thesis uses historical data in real world to conduct simulation experiment and evaluate the effect of experiment.We adopted all road surveillance cameras data in Suzhou industrial park from April 1,2016 to May 30,2016,and the accurate trajectory data of a small number of vehicles with online telematics systems for simulation experiment and evaluate the effect of experiment.And then comparative analysis is conducted among data transmission algorithm with the maximum packet information utility that this thesis proposed,FIFO algorithm,packet residual life algorithm,and PhotoNet algorithm.The experimental results show that the data transmission algorithm based on packet information utility maximization that this thesis proposed achieves a higher precision of trajectory recovery than the other three commonly used data transmission algorithms.