A Generalized XGBoost-Based Approach for Camera-Driven Beam Selection

Incorporating non-RF data in beam selection can speed up decision-making in situations that require a thorough search of all candidate options. Therefore, using camera sensory data at base stations to enhance millimeter-wave (mmWave) beam selection is becoming increasingly popular. This paper introduces a machine learning framework using a robust model along with XGBoost in beam prediction scenarios. Our approach enhances beam selection in both single and multi-candidate real-world vehicle-to-infrastructure (V2I) settings.