Research On GNSS Non-line-of-sight Detection And Multipath Mitigation Algorithm

Global Navigation Satellite System(GNSS)is the most widely used positioning technology at present.When studying the positioning problem in urban canyons,the interference of Non-Line-of-Sight(NLOS)signals and multipath(MP)signals exists at the same time,which seriously affects the positioning accuracy of the receiver.In this dissertation,NLOS signal detection and MP parameter estimation are used to reduce the NLOS and multipath effect,so as to improve the positioning accuracy.The main research contents are as follows:(1)Aiming at the degradation of positioning performance caused by NLOS signal propagation.In order to solve this problem,this dissertation combines the satellite signal classification results of unsupervised learning with the particle filter algorithm,and designs the k-means NLOS signal detection and optimization particle filter positioning algorithm based on unsupervised learning,and reduces the influence of NLOS effect from the two processes of NLOS signal detection and location result optimization.Firstly,GNSS signal classification results were used as NLOS signal detection results to participate in the subsequent calculation of pseudorange difference.Then,the particle filter is used to optimize the positioning results of the receiver,and the kernel k-means updates the particle weight to evaluate the position candidate points.Finally,the pseudorange observations of the optimized GPS/BDS dual-mode satellite navigation system was used to participate in the final positioning results.Experimental results show that in the urban scene,the average positioning accuracy of the proposed particle filter based on unsupervised learning improves from 15 m to about 5m,and the convergence time is shortened from 500 s to about 200 s.(2)For the positioning performance degradation caused by multipath signal propagation.In this dissertation,the IF suppression problem of GNSS receiver is reduced to the joint state of a direct signal and the time-varying model estimation of multipath signal parameters.Therefore,a Gaussian Mixture model-expectation Maximization(GMM-EM)algorithm is proposed to estimate the joint state and time-varying parameters of MP interference suppression in GNSS receiver.The MP mitigation method based on GMM-EM proposed in this dissertation is decomposed into two steps.In the first step,GMM model is used to approximate the posterior Probability Density Function(PDF)and the expected logarithmic likelihood Function of LOS signal parameters required in step E of the EM algorithm.In the second step,the maximum likelihood solution of MP signal parameters is obtained in step M.The convergence of the EM algorithm is analyzed by using the convergence theorem of the EM algorithm.Finally,a comprehensive simulation study is carried out,and the positioning performance of this method in real scene is compared.The code tracking error is controlled within 0.01 slice.For most PRNs,the estimation obtained by using GMM-EM method is 10% to 45% better than the traditional parameter estimation method.