UWB Indoor Localization Based on XGBoost NLOS Identification and DS-TWR Ranging

Indoor environments present significant challenges for ultra-wideband (UWB) localization due to ranging errors and non-line-of-sight (NLOS) propagation. This paper proposes a robust UWB indoor localization framework that integrates double-sided two-way ranging (DS-TWR), XGBoost-based NLOS identification, residual-weighted localization, and Kalman filter (KF). The main contribution of this work is the unified use of NLOS identification in both ranging correction and localization fusion, significantly improving localization accuracy in complex environments. Experimental results demonstrate improvements in ranging accuracy of up to 53.7% and 47.22% under human-body and wooden-board occlusions. In dynamic experiments, the proposed method outperforms conventional UWB localization, KF, and weighted least squares methods with positioning accuracy improvements of 38.64%, 28.95%, and 12.9%, respectively. These results confirm the framework’s effectiveness in mitigating NLOS impact and enhancing UWB localization robustness. Keywords: Ultra-Wide Band (UWB), indoor localization, Non-Line-of-Sight (NLOS) identification, Double-Sided Two-Way Ranging (DS-TWR), XGBoost algorithm