Research On Cooperative Vehicle Localization Technology In Non-Line-of-Sight Environment

With the development of intelligent transportation system(ITS),global navigation satellite system(GNSS)is gradually unable to meet the needs of high-precision vehicle position in complex environment.Cooperative positioning technology can effectively improve the position accuracy through the communication between nodes and the integration of a variety of measurement information.However,non-line-of-sight(NLOS)vehicle nodes and malicious attacks in vehicular networks cause serious distance measurement errors,which lead to cooperative positioning errors.Therefore,how to apply the cooperative location technology in vehicular network and eliminate the influence of NLOS on location has become a hot topic of current research.This thesis focuses on the corresponding detection methods,which detect and discard NLOS vehicle and malicious vehicle,and study a Bayesian cooperative vehicle location method based on the normal LOS vehicle from the detection methods.The research content and innovations of the thesis are as follows:(1)By analyzing the characteristics of autocorrelation matrix of inter-vehicle distance error,it is concluded that the NLOS distance error vector is the eigenvector corresponding to the maximum eigenvalue of the autocorrelation matrix.Based on this conclusion,this thesis proposes a NLOS detection method.In this method,firstly,the difference between GNSS Euclidean distance and measurement distance is used to approximate the intervehicle distance error vector,and the statistical average is used to replace the mathematical expectation to obtain the autocorrelation matrix;secondly,the decision threshold under the specific probability is obtained by the fitting exponential distribution of the eigenvalue;finally,NLOS detection is carried out by the decision on the eigenvalue.Simulation results show that the proposed method can effectively detect NLOS,the detection rate is more than 90%,and the false alarm rate is approximate 10%.(2)The probability model of LOS,NLOS and malicious inter-vehicle distance measurement is established by using the statistical characteristics of vehicle GNSS position.It is concluded that the three measurements can be regarded as following the same distribution.Based on this conclusion,this thesis proposes an abnormal vehicle detection method to realize the unified detection of NLOS vehicles and malicious vehicles.In the method,firstly,based on the established probability model,the conservative preliminary decision is made,and some normal LOS distance measurements are obtained as reference sets;then,the statistical differences between other distance measurements and reference sets are analyzed by Mahalanobis distance(MD),and the credibility of the neighboring vehicle is obtained by chi-square test;finally,the abnormal vehicle is detected by the second decision on credibility.Simulation results show that the proposed method can achieve good detection results for both NLOS and abnormal vehicles.The average NLOS detection rate and malicious detection rate are above 90%,and the average false alarm rates are below 5%.(3)Based on the normal LOS neighboring vehicles obtained by the methods proposed in(1)and(2),this thesis proposes a Bayesian collaborative vehicle location method.Using the GNSS position information filtered by Kalman algorithm and the normal LOS inter-vehicle distance to approximate the prior probability of vehicle position and intervehicle distance respectively.Based on the Bayesian theory,the posterior probability formula of vehicle position in the multi adjacent vehicle scene is derived,and the maximum posterior probability principle and the minimum mean square error principle are used for position estimation.The simulation results based on Monte Carlo show that the proposed method can effectively improve the GNSS position accuracy in complex environment.