FPGA-based adaptive digital beamforming for spaceborne systems: A neural network-based approach compared with MVDR

Neural networks are transforming spaceborne communication systems by offering energy-efficient alternatives to traditional beamforming algorithms like MVDR. This paper proposes a neural network-based digital beamforming (NN-DBF) system and presents a comparative analysis of FPGA implementations of both approaches in adaptive digital beamforming for phased array systems. Experimental results reveal that neural networks not only match MVDR’s beamforming performance but also significantly reduce power consumption, computational complexity, and execution time. These findings demonstrate the potential of neural networks as a robust solution for next-generation satellite communications.