Reasearch On Nlos Identification And Mitigation In Common Frequency Band Position System

With the development of the national economy and the improvement of people’s living standards,building a smart society based on location services has become an important part of the national development plan.In outdoor environments,Global Navigation Satellite System(GNSS)can provide high-precision location services.In indoor environments,satellite positioning signals cannot reach due to occlusion,and GNSS cannot provide location services.The Common Frequency Band Positioning System has the advantages of wide coverage,low promotion cost,and positioning accuracy within the meter level,which can provide an important guarantee for indoor location services.However,the indoor environment is complex and changeable.When the line-of-sight(LOS)path of the transceiver equipment is blocked by obstacles,the positioning signal may propagate through non-line-of-sight(NLOS)such as reflection and diffraction,thereby introducing non-line-of-sight error.The non-lineof-sight error leads to a serious drop in positioning accuracy,which cannot meet people’s needs for high-precision location services.In order to solve the problem of positioning accuracy degradation caused by non line-of-sight errors in a common frequency band positioning system,this paper studies the formation mechanism of non-line-of-sight errors,the impact of non-line-of-sight errors on positioning accuracy,and proposes new non-line-of-sight recognition and mitigation techniques.The main contributions of this paper are as follows:(1)Aiming at the low recognition rate of non-line-of-sight paths,a neural network model based on pseudo-code autocorrelation waveform signal features is proposed.The correlation analysis of the selected 12 features was carried out by the Spearman correlation coefficient method.According to the analysis results,a 4-branch input neural network model is built.The constructed convolutional neural network is trained by using the NLOS and LOS autocorrelation waveform datasets measured in indoor scenes.The optimal model test results show that the proposed neural network model has an accuracy of 94%in identifying NLOS paths,which is superior to existing machine learning models.(2)Aiming at the problem that the positioning accuracy drops sharply when the line-of-sight measurement values do not exist and the number of line-of-sight paths is less than 3,a semi-definite planning TOA positioning based on common NLOS errors and a second-order cone planning TDOA positioning based on the upper bound of NLOS errors are proposed.Through semi-definite relaxation and second-order cone relaxation techniques,the nonlinear and non-convex maximum likelihood positioning model in the non-line-of-sight environment is converted into a convex optimization model.In the semi-definite programming,the non-line-ofsight error in the ranging value of each base station is replaced by the common error,which solves the problem that the optimization variables are more than the constraint equations and the model does not converge.In the second-order planning,the upper bound of the non-line-of-sight errors are replaced by the initial positioning residuals to solve the problem that the upper bound of the non-line-of-sight errors cannot be obtained.Simulation results show that the two NLOS errors mitigation algorithms proposed in this paper can not only mitigate NLOS errors well,but also solve them faster than similar convex optimization algorithms.(3)Build a NLOS identification and mitigation experimental system.In this paper,the neural network NLOS error identification method based on pseudo-code autocorrelation waveform,the NLOS error mitigation method based on semi-definite programming,and the NLOS error mitigation method based on second-order cone programming are integrated into the designed positioning framework.And related positioning performance tests were carried out on the actual common frequency band positioning system.The experimental test results show that the whole set of NLOS identification and suppression scheme proposed in this paper can effectively improve the positioning performance.