Design of a Peer-to-Peer 3-D Pedestrian Positioning System Based on UWB and Sensor Fusion

In ultra-wideband (UWB)-based indoor positioning, the peer-to-peer (P2P) mode to estimate the relative position between two devices has great advantages over infrastructure mode using multiple anchors. However, the implementation of 3-D P2P positioning systems having comparable accuracy to infrastructure mode can be a challenge, because it is more difficult for P2P systems to cope with non-line-of-sight (NLOS) conditions. In this study, we design a P2P 3-D positioning system with high accuracy in LOS/NLOS mixture environments. In our design, four major innovations are imbedded and combined organically. First, we suggest two measurement-calibration methods for an existing commercial UWB device. Second, to effectively cope with NLOS environments, we design a sensor-based 3-D pedestrian dead reckoning (PDR) scheme, which is a fusion of conventional 2-D PDR and altitude estimation using vertical velocity and barometric pressure. Third, to identify whether the UWB link state is LOS or NLOS, we design a two-stage channel classifier which combines machine learning (ML)-based classification with classification using the distance difference between the PDR-based position and UWB-based position. Finally, we provide a hybrid Bayesian nonlinear filter which fuses UWB-based 3-D positioning with 3-D PDR. We implement a prototype of the proposed system and evaluate its performance in typical indoor environments. Experimental results show that the proposed system achieves a root mean square error (RMSE) of about 15 cm and a 95-percentile error of about 28 cm, which implies that the system is a practical and cost-effective solution for real-world indoor IoT applications. A demonstration video is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://youtu.be/94-gZVSdaK8</uri>.