Statistical Characterization and Mitigation of NLOS Errors in UWB Localization Systems

TL;DR

This paper presents experimental insights into NLOS bias in UWB localization, evaluates ML algorithms for bias mitigation, and highlights environment-dependent accuracy variations.

Contribution

It offers new statistical characterization of NLOS bias and assesses the performance of ML-based mitigation algorithms in different propagation conditions.

Findings

NLOS bias significantly affects localization accuracy.

ML algorithms' performance varies with LOS/NLOS conditions.

Experimental results validate the proposed statistical models.

Abstract

In this paper some new experimental results about the statistical characterization of the non-line-of-sight (NLOS) bias affecting time-of-arrival (TOA) estimation in ultrawideband (UWB) wireless localization systems are illustrated. Then, these results are exploited to assess the performance of various maximum-likelihood (ML) based algorithms for joint TOA localization and NLOS bias mitigation. Our numerical results evidence that the accuracy of all the considered algorithms is appreciably influenced by the LOS/NLOS conditions of the propagation environment.