WLS Localization Using Skipped Filter, Hampel Filter, Bootstrapping and Gaussian Mixture em in LOS/NLOS Conditions

Abstract

We present robust range-based localization algorithms for which range measurements are used to estimate the location parameter. Non-line-of-sight (NLOS) propagation of signal can deteriorate the estimation performance severely in the indoor and crowded urban areas. A study for localization has been intensively performed in the line-of-sight (LOS) conditions, but the work for the positioning in the mixed LOS/NLOS environments is comparatively rare. Thus, we aim at the robust localization in the LOS/NLOS mixture environments. The Hampel and skipped filters-based weighted least squares (WLS) methods are proposed for situations where the variance for inliers is unknown in LOS/NLOS mixture environments. For the unsupervised clustering algorithm, Gaussian mixture expectation maximization-based WLS algorithm is utilized. It is demonstrated that the positioning accuracy of the proposed methods is higher than that of conventional methods through extensive simulation.