Research On Wireless Localization Technology Based On NLOS Propagation Path Identification And Error Mitigation

With the growing demand for various Location Based Service(LBS),the requirement for Location accuracy is getting higher and higher.In the process of wireless signal transmission,it is inevitable that various obstacles will be encountered,resulting in the signal can not be transmitted by LOS(Line-of-Sight)directly to the positioning receiver.If the signal reaches the positioning receiver through non-line-of-sight(NLOS)transmission,such as reflection or refraction,it will cause inaccurate measurement of positioning parameters such as TOA(Time of Arrival)and AOA(Angle of Arrival),which will further affect the positioning accuracy.The adverse effect of NLOS error on positioning accuracy has not been well solved so far.How to effectively identify the line-of-sight propagation path and the non-line-of-sight propagation path is very important for improving positioning accuracy.Therefore,this thesis focuses on the identification of LOS/NLOS propagation path and the elimination of NLOS error.The main work is as follows:First of all,through the analysis of the literature reading,the thesis summarizes the NLOS error elimination method of domestic and foreign,it can be divided into two categories: the first is direct to suppress and weaken NLOS error,the second type is based on the different signal propagation characteristics in LOS/NLOS environment,identify and eliminate NLOS base station,only LOS base station is used for positioning.Secondly,the thesis analyzes the NLOS propagation is how to produce adverse effects on the positioning accuracy,focuses on the three restrain NLOS error location algorithm based on the residual error,three kinds of algorithm performance is studied through simulation analysis as well as the advantages and disadvantages,the result shows that in the condition of less NLOS number of base stations and the NLOS error,has good effect on restraining NLOS error;With the increase of the number of NLOS base stations and the increase of NLOS error,the inhibition effect of the algorithm on NLOS error decreases sharply.Then,the statistical characteristics of TOA and AOA positioning parameters in LOS/NLOS environment were studied.For the cases where the prior probability was unknown,NLOS was identified by binary decision based on Neyman-Pearson criterion.When the prior probability is known,the likelihood ratio test is used for NLOS recognition.The simulation results show that the two NLOS recognition algorithms have good performance.In addition,the recognition algorithm based on the distance residuals(RRT)and the recognition algorithm based on the location residuals were analyzed and studied,and the distance residuals were improved by introducing the sum of the squares of the residuals to reduce the computational complexity of the RRT algorithm.When the number of LOS base stations is 3,the group of base stations with the smallest sum of measurement standard deviations is regarded as LOS base stations.Simulation results show that the computational complexity of the improved RRT algorithm is significantly reduced,and the recognition rate is greatly improved when the number of LOS base stations is 3.Then,a LOS/NLOS base station identification and location algorithm based on machine learning is proposed.The positioning base stations were grouped,and every 3 base stations were divided into a group.The TOA measurement distance was used as the feature,and the random forest classification algorithm was used to classify and identify the NLOS base stations.The simulation results show that the correct recognition rate of NLOS base station based on random forest classification algorithm is higher than that of K-nearest neighbor and support vector machine classification algorithm,and LOS base station can achieve better positioning performance after recognition.Finally,the research work of this thesis is summarized,and the future work is prospected.