Research On Real-time And Accurate Positioning Methods In Vehicular Ad-hoc Networks

Location information of vehicles is a very important factor to realize the security application,improve the network performance and ensure the quality of service in vehicular ad-hoc networks(VANETs).It is of great significance for solving traffic congestion,ensuring traffic safety,improving driving experience and promoting the development of intelligent transportation systems.This dissertation focuses on the accurate and real-time positioning problem in VANET based on roadside unit(RSU),from two aspects of vehicle positioning algorithm and RSU deployment.The main contents of the study include the followings.The single-station positoning algorithm based on matrix pencil and virtual RSU in non line of sight(NLOS)environment,the low complexity single-station positioning algorithm in line of sight(LOS)environment,the optimal RSU deployment strategy in full coverage scenario and non-full coverage scenario.Through the above research,the VANET-based accurate and real-time positioning of vehicles can be effectively realized,and the cost efficiency of RSU deployment and the positioning performance of the system can be improved.The main contributions of this dissertation are as follows.Pointing at the vehicle positioning problem for NLOS environment,a method based on matrix pencil and virtual RSU is proposed to realize single station VANET positioning.This method includes an enhanced matrix pencil(EMP)algorithm,the virtual RSU technique and a two-step weighted least squares(WLS)estimator.Firstly,for improving the poor performance of parameter estimation at low signal-to-noise ratio(SNR),EMP algorithm is proposed to implement the high-resolution estimation of time-of-arrival(TOA)and angle-of-arrival(AOA).EMP algorithm combines the traditional 2D-MP and matrix enhancement techniques to extract channel poles from Hankel block matrixca of the received channel frequency response(CFR)data and obtain the multi-path TOA and AOA estimates by matching operation.At the same time,the adaptive threshold is adopted to improve the estimation of subspace dimensions,which enhances the robustness and accuracy of the estimation process.Then,according to the known diagram of reflector plane,the possible position candidate set of the vehicle and virtual RSU coordinates are generated iteratively through the analysis of geometric relations.Finally,combined with the multi-path TOA/AOA estimation obtained by EMP algorithm and the generated virtual RSU coordinates,a two-step WLS estimator is proposed to calculate the real-time position of the vehicle,while fusing the information of motion models.The simulation results show that the proposed single station positioning method has good positioning performance even in the case of low SNR.Pointing at the vehicle positioning problem for LOS environment,a low complexity single station VANET positioning method is proposed.This method includes a power-based matrix pencil(PMP)algorithm and the nonlinear fitting technique.In this method,PMP algorithm is proposed firstly,in which the signal subspace of received covariance matrix is approximated by matrix power operation,avoiding the eigenvalue decomposition or singular value decomposition with a large amount of computations.Based on this approximation,the AOA is estimated and the time cost of the algorithm is reduced.Then,by using the frequency diversity characteristics of the subcarriers of orthogonal frequency division multiplexing(OFDM)signals,the high-resolution estimation of TOA is modeled as a nonlinear fitting problem.The ill-conditioned state of the fitting function is also dealt with,which further reduces the computational complexity of the algorithm and improves the reliability of the results.After obtaining the estimation of AOA and TOA,the position of the target vehicle can be located by geometric relation and a WLS estimator.The simulation results show that the time overhead of this method is only about 50% of that of the traditional 2D-MP algorithm,and the better accuracy of positioning can be achieved under the condition of high bandwidth and a greater number of elements.To solve the deployment problem of RSU full coverage scene in VANET positioning,an optimal RSU deployment strategy for the full coverage scene is proposed.The deployment strategy is divided into two steps.Firstly,a cost-effective optimal placement pattern(OPP)is determined to realize the optimal RSU layout in one coverage scenario.Secondly,in K coverage scenario,the geometric dilution of precision(GDOP)metric is used to measure the quality of RSU layout,which is suitable for both RSS-based and joint TOA/AOA estimation-based localization method.Then the RSU deployment problem can be transformed into hierarchical OPP deployment optimization problem,and the K coverage optimal RSU layout can be obtained by solving the optimization problem using asynchronous particle swarm optimization algorithm.The deployment strategy uses the minimum number of RSU for each layer,so it can be considered to be achieve the optimal cost-effective.The simulation results show that the performance of the proposed deployment strategy is better than the traditional uniform strategy under the condition of different coverage degrees and communication radii.To solve the deployment problem of RSU incomplete coverage scenario in VANET positioning,an optimal RSU deployment strategy for the incomplete coverage scene is proposed.Firstly,based on the lower bound of the location estimation error of nonlinear filtering in dynamic VANET positioning,the mathematical expression of average GDOP in the positioning region is derived according to nonlinear recurrent model and Fischer information matrix.Then,considering the deployment cost of RSU,the RSU deployment problem for incomplete coverage scenario is modeled as an optimization problem related to average GDOP and deployment spacing.The optimal deployment strategy can be obtained by using center particle swarm optimization algorithm to solve this problem.The performance of the proposed strategy is verified by simulations,and the impact of vehicle speed and RSU communication radius on RSU deployment efficiency in vehicle positioning is analyzed.The simulation results show that the strategy can optimize the cost-effectiveness of RSU deployment,and can provide some references for VANET network planning in incomplete coverage scenarios.