Identifying Line-of-Sight (LOS) and Non-LOS (NLOS) channel conditions is a big concern in many works. In this paper, we aim to achieve channel classification for millimeter wave (mmWave) systems. The results of mmWave measurement conducted at 28 GHz in typical shopping mall scenario are presented. Specifically, we analyzed the identification performance of single- valued feature extracted from channel impulse response (CIR), and confirmed that rise-time is an attractive features for identification, particularly precise for wideband mmWave signal. Furthermore, a classification algorithm Gradient Boosting Decision Tree (GBDT) is used for LOS/NLOS identification. Experimental results demonstrate an overall detection accuracy of 97.9\%, which shows that our method has a better classification performance than the single feature classifier. Moreover, extensive experimental evaluation considering various propagation distances and obstruction diversity have validated the feasibility of this method.
NLOS Identification for Wideband mmWave Systems at 28 GHz
01.04.2019
7614696 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch