AESA Adaptive Beamforming Using Deep Learning

Alberto Raimondi, Simone Bianco, Paolo Napoletano, Pietro Vinetti, Maurizio Feo, Giovanni Petraglia

Abstract: In this work we propose a method for the adaptive beam-forming of an antenna array using Deep Learning. The proposed method is based on a deep Convolutional Neural Network that takes as input an image-like radiation pattern encoding the desired behavior and computes the optimal currents needed to adapt the antenna to the new beam specification. The proposed approach drastically reduces the computation time (up to 1700×) introducing a smart mapping of a classic iterative algorithm to an antenna to reproduce it. After training the model is able to compute optimal currents successfully in a single forward pass, avoiding the need of expensive iterative optimizations to find the needed currents.