Robust localization under NLOS environment in the presence of isolated outliers by full-Set TDOA measurements

Abstract

Different from the traditional approach of using the reduced set of time-difference-of-arrival (TDOA) measurements, localization by the full TDOA set is discussed here. It avoids choosing a specific reference sensor and exploits more measurements, conferring robustness against outliers. Nevertheless, the none-line-of-sight (NLOS) propagation still presents challenges for the existing full-set positioning algorithms. In this work, we develop an iterative scheme that utilizes the full TDOA measurements for localization capable of simultaneously mitigating the influence of outliers and NLOS-path measurements. The proposed scheme consists of the localization and identification modules, and they are designed with low complexity to meet real-time positioning demands. The theoretical optimal TDOA methods using full-set measurements are proposed in the localization module. Besides, to identify the NLOS measurements, the identification module adopts a greedy search-based random sample consensus (GS-RSC) algorithm. It can work cooperatively with the proposed lightweight localization module, returning a set free of outliers, NLOS measurements, and an accurate source location estimate. The theoretical analysis and experiments illustrate the superiority of the proposed scheme over the state-of-the-art TDOA full-set algorithms for handling outliers and NLOS measurements in localization.