Non-line-of-sight error mitigation for range estimation in dynamic environments

Localization is an important component in many applications such as the promising Ultra-Wideband (UWB) wireless sensor network for medical treatment. For the majority of localization technologies, it is essential to measure the ranges between a target and several reference nodes before the target can be localized. Existing range estimation techniques rely on the measurements of time-of-arrival (TOA) and received-signal-strength (RSS) which suffer from environmental change. Dynamic environment such as human mobility can cause non-line-of-sight (NLOS) measurements which will significantly degrade the accuracy of range estimation. Therefore, range estimation methods are desired to be robust to NLOS measurements. In this paper, it is proposed to use hypothesis tests to identify whether there are NLOS measurements mixed in with the measurement dataset. For those NLOS corrupted measurement datasets, a new range estimation method based on a Log-normal model is found to be capable of reducing the range estimation error. Another advantage of this new range estimation method is that NLOS measurements are not required to be excluded from its analysis. However, simulation results show that the range estimation accuracy can be further improved if NLOS measurements are excluded.