Reliability-Aware Beam Prediction for UAVs

This paper addresses the critical challenge of reliable communication for UAVs in the rapidly growing low-altitude economy. It moves beyond static weighting in multi-modal beam prediction, which is a significant advancement. The proposed SaM2B framework's dynamic weighting scheme, informed by reliability, and the use of cross-modal contrastive learning to improve robustness are key contributions. The focus on real-world datasets strengthens the paper's practical relevance.