A Robust Statistical Approach to Non-Line-of-Sightmitigation

Mitigation of non-line-of-sight (NLoS) errors in the geolocation problem has received much attention. It is well-known that these errors degrade the robustness and accuracy of localization systems, whereby localization is performed using time-of-arrival (ToA) measurements. In this paper, we propose a robust approach to mitigate the NLoS effect in location estimation. The approach is based on modeling NLoS errors as contaminated Gaussian noise, which masks the unknown ToA. We then suggest a non-parametric estimate of the noise density, obtained from observations. It provides robustness against the outliers caused by the NLoS errors common in urban areas. Simulation results show an improvement over conventional approaches.