Proximal Subgradient Descent Method for Cancelling Cross-Interference in FMCW Radars

In this work, we address the problem of mitigating cross-interference which is commonly encountered in a dense deployment of Frequency-Modulated Continuous Wave (FMCW) radars. We present a novel proximal subgradient descent-based method to jointly infer and cancel any cross-interference potentially present in the received baseband signal of an FMCW radar. The proposed method is designed to exploit the temporal sparsity of the cross-interference and the sparsity of the interference-free received baseband signal in the range-Doppler frequency domain for improved estimation and cancellation of the cross-interference. Experiments using synthetic and real-world FMCW radar measurements demonstrate the real-time implementation potential and the superior performance of our proposed method compared to the existing cross-interference mitigation methods.