Error Mitigation for TDoA UWB Indoor Localization Using Unsupervised Machine Learning

Abstract

Indoor positioning systems based on ultrawideband (UWB) technology are gaining recognition for their ability to provide cm-level localization accuracy. However, these systems often encounter challenges caused by dense multipath fading, leading to positioning errors. To address this issue, in this article, we propose a novel methodology for unsupervised anchor node selection using deep embedded clustering (DEC). Our method uses an autoencoder (AE) before clustering, thereby better separating UWB features into separable clusters of UWB input signals. Afterward, we rank these clusters based on their cluster quality, allowing us to remove untrustworthy signals. Our method is novel, as it is the first error mitigation approach for time difference of arrival (TDoA)-based UWB localization that uses unsupervised machine learning (ML), thereby avoiding costly labeling efforts and significantly reducing the localization error. Our experiments show that our method can reduce the mean absolute error (MAE) by a significant 23.1% overall, and in dense multipath areas by 26.6%, and the 95th percentile error by 49.3% when compared with without anchor selection.