Fast Adaptive Beamforming Using Deep Learning for Digital Phased Array Radars

All-digital polarimetric phased array radar has the potential to achieve the high temporal and spatial resolution requirements needed for polarimetric meteorological observations. It offers maximum flexibility in scanning strategies that support the use of the traditional “pencil” beams, spoiled beams, and/or multiple beams. Two digital beamforming (DBF) methods are of interest for polarimetric meteorological measurements, the Fourier and Capon methods. While beamforming weights can be pre-computed with the conventional Fourier DBF method, they are computed adaptively with the Capon method. The Capon method can provide improved data quality over the Fourier method by lowering antenna sidelobe levels in directions with high return power and improving the angular resolution of the scanning beam. However, it is challenging to implement the Capon method in real-time due to the high computational cost. This paper presents a computationally efficient Deep Learning Adaptive digital beamForming (DLAF) method using a neural network trained with the Capon method. The proposed DLAF model is evaluated using a simulated point target and simulated weather signals for a planar digital array architecture. The performance of DLAF is measured by its computational efficiency compared to the Capon method and its ability to mitigate contamination coming from antenna sidelobes. Preliminary results show that DLAF can be used to significantly reduce DBF computation time while improving the data quality of observations.