Robust localization based on non-parametric kernel technique

Abstract

Parametric approaches are primarily used in the context of robust localization. However, the localization performance is degraded when there is a mismatch between the assumed model and the actual situation. To circumvent this problem, in this letter, a robust weighted least squares (WLS) method based on the non-parametric kernel density estimator (KDE) and kernel regressor (Nadaraya-Watson estimator) is proposed. First, the line-of-sight (LOS)/non-LOS mixture distribution is obtained using the KDE and the support corresponding to the first peak is determined as a distance estimate. Subsequently, kernel regression is performed to calculate the conditional mean and variance of the conditional mean is then estimated. Moreover, the transformed range and its variance are obtained. Subsequently, the two-step WLS method is applied with this information. The simulation results demonstrate that the proposed algorithms outperform the conventional methods in terms of localization.