Terminal location method with NLOS exclusion based on unsupervised learning in 5G-LEO satellite communication systems

We investigate the terminal location method in 5G-Low Earth Orbit (5G-LEO) satellite communication systems. To overcome the dependence on the external Global Navigation Satellite System (GNSS), we propose to use a single LEO satellite in 5G-LEO satellite communication systems for terminal location and utilize the downlink synchronization detection for pseudorange differential measurement. Then, a data clustering method of unsupervised machine learning is proposed to classify the measured data into line-of-sight (LOS) and non-line-of-sight (NLOS) signals. Furthermore, the NLOS data are excluded, and the Taylor series expansion iteration method is used to calculate the terminal coordinates. Simulation results show that the proposed method can effectively reduce the influence of NLOS error on measurement results and improve the accuracy of terminal location. In simulated urban scenario, the average location accuracy is improved from 10 km by traditional methods to 0.7 km and the convergence time is reduced from 400 to 250s.