NLOS Identification Based UWB and PDR Hybrid Positioning System

Abstract

Some of the researches on indoor positioning have been conducted, but there are still many constraints on indoor positioning approaches. Among these approaches, ultra-wideband (UWB) provides a fast and precise positioning performance but requires a sufficient infrastructure and a clear line-of-sight (LOS) channel. However, inertial sensor-based pedestrian dead reckoning (PDR) operates without infrastructure, but it requires position initialization and has error drift problems. In this study, we propose a hybrid positioning system that fully combines UWB and PDR to overcome such constraints and improve the positioning performance. This hybrid positioning system uses a Kalman filter (KF) based fusion method that identifies non-line-of-sight (NLOS) environments and mitigates UWB errors through PDR. We also evaluated the proposed system implemented using practical testbed devices at indoor environments classified as LOS, weak NLOS, and hard NLOS. The evaluation results showed that the proposed system significantly improves the positioning performance and alleviates the positioning constraints, as compared to the single positioning system. Our system has been designed to be lightweight compared to the existing extended KF-based convergence system, but is more robust to both weak and hard NLOS environments. Eventually, it improved positioning performance by 35.5% than existing hybrid systems in the hard NLOS environments.