Research On UWB Localization Algorithm In NLOS Environment

With the development of wireless communication,people’s research on indoor positioning is in full swing,but how to maintain high-precision positioning in complex indoor scenes is the focus of people’s research.Compared with traditional indoor positioning methods,Ultra Wideband(UWB)has the characteristics of anti-multipath,strong penetrability,and low energy consumption.It is more suitable for complex indoor environments and achieves long-term,high-precision positioning effects.Therefore,this paper focuses on the algorithm to improve the UWB positioning accuracy in the Non-Line-of-Sight(NLOS)environment.Firstly,the definition and characteristics of UWB are introduced,and it s adaptability in the NLOS environment is explained.Four indoor scenarios of IEEE802.15.4a are simulated.The algebraic solution method based on geometric factors such as distance and angle and the fingerprint localization method based on scene prior information are studied,which provide a theoretical basis for improving the localization accuracy in NLOS environment.Then a localization algorithm based on the improved residual weighting method is proposed.First,it is determined that NLOS error is a var iable model related to distance,and the residual weighting method is selected to suppress NLOS error.The traditional residual weighting method is improved in terms of computational complexity and weight function selection.By simulating the static positi oning and dynamic positioning under the algorithm and comparing the positioning results with other traditional algorithms,it can be seen that the algorithm can guarantee the high-precision positioning effect under NLOS conditions.Finally,in order to solve the influence of multipath,NLOS in UWB positioning,this paper improves the traditional fingerprint positioning method.In the stage of establishing the fingerprint database,by adding K-means classification,the effective division of the fingerprint database is realized,and the BP neural network is used to train the fingerprint database after classification,and the accurate corresponding model of feature data and position coordinates is obtained.In the online matching stage,the obtained feature data is first compared with the sub-category center,and then substituted into the neural network model of the corresponding category,so as to achieve accurate and efficient positioning of the points to be measured.Compared with the traditional fingerprint positioning algorithm,the algorithm has higher positioning accuracy and positioning accuracy under the same situation.