Improving RSS-based ranging in LOS-NLOS scenario using GMMs

Range estimation based on Radio Signal Strength (RSS) has been widely adopted by the indoor localization systems. Many existing works have been devoted to tackle the imprecise and unreliable RSS measurements caused by multipath fading. But there exist only a few works dealing with errors caused by non-line-of-sight (NLOS) radio propagation. In some circumstances, it is common for obstacles (e.g. human movements) to cause NLOS measurements, which could undermine the whole ranging process by introducing significant NLOS errors. In this letter, we propose to use a Gaussian Mixture Model (GMM) to model the distribution of a set of NLOS corrupted range estimations. In the GMM method, the distribution of LOS estimations and the distribution of NLOS estimations are represented by different Gaussian components. Consequently, the ranging quality is improved by employing soft exclusion of those Gaussian components associated with NLOS. An indoor field experiment has been performed to verify the proposed method.