Research On Localization And Tracking Strategy Based On Large Scale Antenna Array

In the era of mobile Internet,location-based services have become an important part of the wireless positioning technology has been widely used in people's clothing,food,housing,travel and other aspects,effectively improving people's living efficiency and quality of life.The location of the mobile terminal can cope with the challenges of high mobility,high reliability,number of connections,ultra-low latency,and traffic density of 5G new services.Large-scale antenna array and beamforming technology provide important technical support and guarantee for positioning technology,especially in ultra-dense networks,which can achieve high-density coverage,greatly enhance the probability of signal line-of-sight transmission,and make the positioning by angle of arrival(AOA)possible.Secondly,since the circular array antenna can control the beam to scan 360° in the horizontal plane,and its gain pattern and other characteristics do not change with the scanning angle,it has direction finding advantages.This paper concentrates on AOA positioning and tracking technology using uniform circular array.Firstly,this paper studies the AOA estimation part of the positioning method,and compares the angular resolution,positioning accuracy and time complexity of the existing AOA estimation method.The MVDR algorithm is an optimization algorithm for CBF,but there is still a problem of low resolution as CBF.In the super-resolution algorithm,the accuracy of the ESPRIT algorithm is relatively low,and the time complexity of the WSF algorithm is very high.Among them,the MUSIC algorithm has higher angular resolution and positioning accuracy,but the spectrum search requires a lot of time.To solve this problem,a hierarchical AOA estimation method for narrowing the spectrum search range is proposed.The method is based on a circular array antenna,and uses RSS information to determine the range of the arrival angle of the signal.Then,using the MUSIC algorithm based on the beamspace circular array,the spectrum search is performed within the specified range after the judgment to obtain an accurate angle of arrival,and the final target position is obtained according to the principle of triangular positioning.The improved hierarchical AOA estimation method based on MUSIC algorithm greatly shortens the AOA estimation time.The method is simulated to compares the positioning accuracy and time complexity of the method.It demonstrates that the method improves the positioning efficiency while maintaining the positioning accuracy.Regarding the part of suppressing the NLOS error in the positioning process,the residualbased NLOS suppression algorithm,such as residual weighting,iterative minimum residual,and selection residual weighting algorithm,is studied.NLOS recognition methods such as skewness detection and kurtosis detection,Grubbs test,shapiro-Wilk test,etc.through distribution are studied.Aiming at the difference of AOA detection error distribution between NLOS base station and LOS base station,based on N-P criterion and likelihood ratio test,a less complex method is proposed to replace the high complexity likelihood ratio calculation.The simulation results show that the traditional NLOS error suppression algorithm based on residuals will calculate the positioning result of NLOS base station into the final result,which will lead to an increase in error.By first identifying the NLOS base station,the positioning accuracy of the remaining LOS base station is greatly improved.When a reasonable false alarm probability is set,a high detection probability can be achieved.After identifying the NLOS base station,exclude them and use the remaining LOS base station for positioning,and then clustering is used to eliminate the abnormal point.This process can prevent some NLOS identification failures.The NLOS base station can be excluded from the calculation of the final positioning result.The methed obtains the mean value by multiple measurements,and counts the number of times the single measurement value deviates from the mean value beyond a certain threshold to judge the NLOS base station,and removes the abnormal point by clustering.Combination of the front and the back quickly and effectively suppresses the NLOS positioning error.The method is simulated to compare and analyzed with the traditional RWGH error suppression method,and the simulation demonstrates the effectiveness of the proposed method.Finally,for the tracking and filtering part,several common motion models are studied.In view of the fact that multiple motion states are interacted in the actual motion process,it is demonstrated that the interactive multi-model method can achieve better tracking effect.The basic principle and parameter selection of the interactive multi-model method are analyzed,as well as its advantages and disadvantages.When the Kalman filter is used for the filtering method in the interactive multi-model method,the parameter Q can change the Kalman filter gain,and then it can change the weighting weights of the predicted value and the measured value.According to the difference in the measurement error caused by the difference in the number of participating base stations,an adaptive Q interactive multi-model Kalman filter tracking method based on the number of located base stations is proposed.When there are many base stations participating in the positioning,the measurement error is small,thereby increasing Q and increasing the measured value weight.The simulation results show that the IMMKF method with dynamic change of Q has better tracking accuracy and higher stability during the tracking process.