NLOS Identification And Error Suppression Algorithm Research In UWB Indoor Positioning

Ultra Wide Band(UWB)is widely used in indoor positioning.The indoor positioning wireless channel transmission environment is complex,and the UWB signal propagation process is obstructed by obstacles.The direct path(DP)propagation channel between the positioning tag and the node is blocked,and the signal reaches the positioning node through reflection and diffraction channels,resulting in non-line of sight(NLOS)propagation.In the NLOS propagation environment,it is extremely challenging to ensure the positioning accuracy and continuity of the UWB tag.Therefore,effective NLOS identification and error suppression are of great importance to improve indoor positioning accuracy.This paper addresses the two problems of NLOS identification and localization error suppression in UWB indoor positioning,and the main research contents and contributions are as follows:(1)For the first time,the XGBoost algorithm is applied to NLOS recognition in UWB indoor positioning,and two XGBoost NLOS recognition methods based on Channel Impulse Response(CIR)and key signal features are proposed respectively.For XGBoost NLOS recognition based on impulse channel response,the collected LOS(line of sight,LOS)and NLOS propagation signal impulse channel response are input into the XGBoost algorithm,and the NLOS recognition model is obtained after training.For the XGBoost NLOS recognition based on key signal features,the four key features of the UWB localization signal are extracted and then input to the XGBoost algorithm,and the NLOS recognition model is obtained after training.The experiments show that the two XGBoost NLOS recognition methods proposed in the paper are more accurate than several other commonly used machine learning algorithms.(2)To address the problem of large NLOS errors or even divergence,a camel particle filter UWB localization algorithm based on Levy flight improvement is proposed.Firstly,a camel marching strategy is adopted to improve the particle degradation and particle depletion caused by resampling in UWB NLOS localization by traditional particle filtering;then a Lévy flight strategy is used to perturb the position of the camel particle filtering algorithm to assist the algorithm to find the best position and improve the position estimation accuracy and stability of the algorithm.(3)The experimental environment is built and the experimental procedure is designed to verify the algorithm.The experimental results show that the two XGBoost NLOS recognition algorithms proposed in the paper outperform traditional machine learning algorithms in terms of accuracy;the improved camel particle filter UWB localization algorithm based on Lévy flight improves the localization accuracy in mixed LOS/NLOS scenarios while solving the particle degradation problem.