Adaptive Not-direct-path Identification in UWB Localization

Abstract In specific ultra-wideband (UWB) localization scenarios, conventional non-line-of-sight (NLOS) identification algorithm cannot detect other not-direct-path conditions. We proposed an adaptive not-direct-path identification method based on machine learning and artificial neural networks. Compared to the artificial division of not-direct-path data sets, we verified the advantages of data set partitioning by the method. The performance of NLOS recognition using different methods in different scenes is also analyzed, and the average identification accuracy in different scenarios can reach 92% or more.