ADMM-Net for Communication Interference Removal in Stepped-Frequency Radar
Jeremy Johnston, Yinchuan Li, Marco Lops, Xiaodong Wang
Abstract
Complex ADMM-Net, a complex-valued neural network architecture inspired by the alternating direction method of multipliers (ADMM), is designed for interference removal in stepped-frequency radar super-resolution angle-range-doppler imaging. We consider an uncooperative spectrum sharing scenario where the radar is tasked with imaging a sparse scene amidst communication interference that is frequency-sparse due to spectrum underutilization, motivating an <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\ell _1$</tex-math></inline-formula> -minimization problem to recover the radar image and suppress the interference. The problem's ADMM iteration undergirds the neural network design, yielding a set of generalized ADMM updates with learnable hyperparameters and operations. The network is trained with random data generated according to the radar and communication signal models. In numerical experiments ADMM-Net exhibits markedly lower error and computational cost than ADMM and CVX.