Sweep-Free Beamforming via Optimal Angle of Arrival Estimation for 6G Ultra-Dense Networks

In this paper, we propose a sweep-free beamforming approach for 6G wireless systems by accurately estimating the angle of arrival (AoA) corresponding to the lowest path loss in multipath propagation environments. The proposed method directly estimates optimal azimuth and elevation angles using a lightweight ConvNeXt model trained on angular channel power matrices (ACPMs). The estimated AoA is reused as the angle of departure (AoD) for beamforming, enabling robust downlink transmission without exhaustive beam sweeping. Leveraging the sparse and dominant structure of ACPMs, the method achieves accurate AoA estimation with low model complexity and demonstrates improved beamforming gain compared to the quantized 5G DFT-codebook-based approach, even in non-line-of-sight scenarios. The results show that the proposed method enables high-precision beamforming suitable for 6G systems.