Comparative Study of Real-Time Beamforming Techniques for Railway Communication Systems in LoS and NLoS Scenarios

This paper presents a comparative study of real-time beamforming techniques for railway communication systems in Line-of-Sight (LoS) and Non-Line-of-Sight (NLoS) scenarios. Using the 3GPP TR 38.901 channel model, initial machine learning (ML)-based beam prediction solutions are evaluated, and advanced models are proposed to address their limitations. The results demonstrate significant improvements in beam prediction accuracy and robustness, emphasizing the importance of diverse data collection to capture channel variability.