Non-Line of Sight Detection Based on 5G Signal Strength- and Quality Measurements and its impact on mobile positioning accuracy

5G radio signal strength-and quality measurements can be used to distinguish between Line of Sight (LOS) and non- Line of Sight (NLOS) channels with good accuracy. Classification results, obtained with the use of Machine Learning (ML) methods, can be used to significantly improve positioning accuracy in complex radio environments. Two measurement datasets were collected, first with 5 G modem and second one with dedicated 5 G measurement receiver. At first, we showed that by implementing ML methods the LOS/NLOS classification accuracy of 96% was achieved. Influence of obtained classification performance on positioning accuracy was simulated next. Those results showed that positioning errors can be decreased up to five times by discarding the NLOS measurements, given that sufficient LOS measurements are available.