Robust Mobile Location Estimation in NLOS Environment Using GMM, IMM, and EKF

Indoor mobile localization in real-life scenarios often suffers from frequent transitions of sensor measurements between line-of-sight (LOS), non-line-of-sight (NLOS), and/or mixed LOS and NLOS conditions (LOS-NLOS). To address this, we propose GIMM-EKF by integrating Gaussian mixture model (GMM), interacting multiple model (IMM), and extended Kalman filter (EKF). In GIMM-EKF, GMM aims at modeling the distribution of a set of mixed LOS-NLOS range estimates. Then, a Kalman-based IMM framework is introduced with the estimated state probabilities from the GMM. Finally, an EKF is employed to estimate the target's location based on the resulting range estimates. The proposed GIMM-EKF works in a synergistic manner and outperforms several challenging baselines significantly. Experimental results demonstrate the feasibility of GIMM-EKF in mitigating the adverse impacts of severe NLOS errors and accurately estimating the mobile location in the LOS/NLOS/LOS-NLOS transition conditions.