LOS/NLOS Channel Identification Technology Based on CNN

作者：Fei Wang, Zhan Xu, Ruxin Zhi, Jinhui Chen, Peiyue Zhang · 发表于：2019 6th NAFOSTED Conference on Information and Computer Science (NICS) · 年份：2019 · DOI：10.1109/nics48868.2019.9023805 · 被引用次数：31 · 研究领域：Indoor and Outdoor Localization Technologies、Ultra-Wideband Communications Technology、Target Tracking and Data Fusion in Sensor Networks

In the field of indoor positioning, ultra-wideband (UWB) communication technology is the key technology. But the errors caused by non-line of sight (NLOS) communication channels could greatly affect the positioning accuracy. Therefore, identifying the channel type can effectively avoid the above-mentioned errors, thereby helping to reduce the loss of positioning accuracy. This paper proposes a LOS/NLOS channel identification method by using a convolutional neural network (CNN) to identifying the impulse response figures of the channels. This method is capable of identifying four different types of channel impulse response figures and providing excellent identification accuracy. The results show that the accuracy of identifying LOS (0-4m), NLOS (0-4m), NLOS (4-10m) and extreme NLOS channel impulse response figures can reach 98.24%.