A Robust Filter for TOA Based Indoor Localization in Mixed LOS/NLOS Environment

One of the major problems in TOA (Time of Arrival) based localization is the presence of non-line-of-sight (NLOS), which is caused by intermittent blockage of the signal propagation path between the transmitter and the receiver. The NLOS measurement can be isolated with varies distribution tests or mitigated with time domain smoothing. This paper presents a robust extended Kalman filter (RKF) that adapting NLOS with the variance inflation method and proposes a new robustness performance indicator named bias mitigation ratio (BMR). The RKF performs a normalized residual test to identify the NLOS measurement and adapting the identified NLOS measurement with the variance inflation method. This approach is computationally efficient and does not require any prior information about NLOS distribution. The simulation results indicate that the RKF can mitigate the effect of NLOS efficiently and achieves 0.5m positioning accuracy by integrating Wi-Fi signal and pedestrian dead reckoning (PDR). The robustness of RKF is examined by BMR. BMR enables to analyze the robustness of RKF quantitatively, which reveals the performance improvement of RKF subject to the extended Kalman filter (EKF).