Research On LOS/NLOS Propagation Path Identification Technology Based On Statistical Information And Geometric Features

In view of the possibility of obstacles in the wireless signal communication transmission,the signal finally reaches the receiving end through reflection or refraction.The propagation method based on reflection and refraction will not only cause the time of arrival(TOA)of the received signal to be greater than the transmission time of the straight-line distance from the sender to the receiver,but also the angle of arrival(AOA)of the signal is different from the angle of arrival of the linear propagation.The non-linear distance propagation method of wireless signals is defined as non-line-of-sight(NLOS)propagation.At the same time,the distance and angle of arrival errors generated by NLOS propagation are collectively referred to as NLOS errors,and the linear distance propagation method is defined as line-of-sight propagation(LOS).If the parameters received during NLOS propagation are used to locate the sender,the obtained position information will be inaccurate.Therefore,in order to ensure the accuracy of position information estimation and eliminate the impact of NLOS errors,this thesis has conducted in-depth research on the identification methods of LOS/NLOS propagation paths for different characteristics of LOS/NLOS propagation.The main work statement are as follows:First of all,by referring to a large number of references,the current LOS/NLOS recognition methods of domestic and foreign are summarized,and the LOS/NLOS recognition algorithms are specifically divided into three categories: The first category is based on training data sets,using machine learning and other methods Identify LOS/NLOS signals;the second type is to use the received signal statistical characteristics,comprehensive signal detection theory and mathematical characteristics to achieve the identification of LOS/NLOS signals;the third type is based on the spatial geometric relationship,through geometric constraints on the LOS/NLOS signal Judgment.The first type of method is mainly applicable to the indoor environment that is basically unchanged,so this thesis mainly studies the second and third methods.Secondly,the signal detection method in the LOS/NLOS environment for the binary decision of the receiving parameters TOA and AOA is studied and analyzed.When the prior information is unknown,the N-P criterion is used to decide that the signal belongs to the specific category of LOS/NLOS.The research analyzes the NLOS recognition method(FINE algorithm)based on error vector subspace information,and compares it with the NLOS recognition method(SRNI algorithm)based on sparse representation.The recognition rate of FINE algorithm gradually decreased.In response to this problem,an improved FINE algorithm for normalized residuals is proposed.This algorithm can first identify a group of LOS base stations using normalized residuals when the FINE algorithm cannot recognize the results or the recognition results are incorrect.Use this group of LOS base stations to correct the position of the intermediate mobile station.Simulation results show that the proposed improved algorithm can improve the accuracy of NLOS recognition of FINE algorithm in complex environments.Then,by analyzing the characteristics of the positioning geometric relationship,the crosscircle area recognition algorithm and the positioning position residual recognition algorithm are studied respectively.The principles and performance of these two algorithms are analyzed and verified.Based on their principles,a method based on LOS recognition algorithm for common string intersection error.The core idea of the algorithm is to identify the LOS base station combination by using the difference in the influence of measurement noise and NLOS error on the intersection point of the observation circle common chord of the positioning base station combination.The results of simulation experiments show that this algorithm has a higher recognition rate when the number of LOS base stations is greater than 3,compared with the RT method,the location position residual recognition algorithm and the cross circle area recognition algorithm.Finally,the research content of the full text is summarized,and the shortcomings and future prospects are analyzed.