UWBLoc: NLoS Mitigation for Practical UWB Localization

Ultra-WideBand (UWB) technology has been employed for high-precision localization of targets due to its high resolution, low power consumption, and strong anti-interference capability. Though promising, the None-Line-of-Sight (NLoS) issue caused by walls and human occlusions in real-world scenarios significantly degrades ranging accuracy and localization performance. Existing methods have been developed to tackle this issue, however, they typically incur a high computational cost and are not easily adaptive to different environments. In this paper, we propose UWBLoc, a UWB-based localization system that achieves accurate localization under NLoS scenarios. Our insight is that Channel Impulse Response (CIR) from UWB device exhibits significant differences in Line-of-Sight (LoS) and NLoS environments, which can be utilized for link status identification. Meanwhile, we dynamically update the ranging weights of multiple anchors based on the link status identification and ranging residuals to characterize the influence of the environment on ranging. Then, we propose a hybrid distance and angle measurements algorithm based on the Taylor Series-based Least-Square (TS-LS) method, which effectively mitigates the impact of NLoS links on practical localization. We implement UWBLoc on commodity UWB devices and evaluate it in real office scenarios with obstructions such as walls and pillars. The evaluation results demonstrate that UWBLoc can achieve an average localization accuracy of 25.4 cm for single-point localization and 27.5 cm for real-time trajectory tracking.