ADMM-Net for Communication Interference Removal in Stepped-Frequency Radar

Complex ADMM-Net, a complex-valued neural network architecture inspired by the alternating direction method of multipliers (ADMM), is designed for interference removal in super-resolution stepped frequency radar angle-range-doppler imaging. Tailored to an uncooperative scenario wherein MIMO shares spectrum with communications, ADMM-Net recovers image---which assumed be sparse---and simultaneously removes communication interference, which modeled as sparse domain owing underutilization. The motivates $\ell_1$-minimization problem whose ADMM iteration, turn, undergirds design, yielding set generalized iterations that have learnable hyperparameters and operations. To train we use random data generated according signal models. In numerical experiments exhibits markedly lower error computational cost than CVX.