RSSI-based methods for LOS/NLOS channel identification in indoor scenarios

In this paper, we investigate classification methods aiming at identifying the Line-Of-Sight (LOS) or Non-LOS (NLOS) condition of a wireless channel. Our approach is based on the computation of statistical features over N consecutive channel measurements at the receiver (namely, N Received Signal Strength Indicator, RSSI, values). First, threshold classification criteria, on the considered features, are derived in order to perform LOS/NLOS identification. The thresholds' values are tuned according to the "behaviour" of the statistical features in the considered environment. This method is compared to a sample-based (whose aim is to detect the data distribution) and a machine learning-based approaches. Although our approach is general, we present experimental results for IEEE 802.11 indoor channels. Our results show that simple threshold-based classification criteria on the considered statistical features may yield approximately 85÷90% LOS/NLOS classification accuracy, making them an attractive strategy for future 5G systems.