The LOS/NLOS Classification Method Based on Deep Learning for the UWB Localization System in Coal Mines

A localization system is one of the basic requirements for coal mines. Ultra-wideband (UWB), as a technology with broad application prospects, considered to have great potential in absence satellite signals, especially underground mine environment, it can provide rescue assistance. However, state-of-the-art UWB position systems mines cannot efficiently differ line-of-sight from all communication links, which results deterioration accuracy. In this paper, we propose LOS/NLOS classification method based on deep learning algorithm. Specifically, utilize Generative Adversarial Networks (GAN) generate diagnostic data frame transmission under non-line-of-sight (NLOS) condition. Then, Convolutional Neural Network (CNN) adopted identify NLOS communication. Finally, trilateral centroid positioning algorithm (TCPA) ranging used verify effect our Field experiments show that accurately accuracy 91.19%. The TCPA obtain 3.11% improvement compared scenario without using method.