DOA Estimation in FMCW Automotive Radars with Interference: ML and VMD Approaches

In this paper, we deal with the problem of estimating the Direction of Arrival (DOA) of vehicles under the interference assumption in frequency division multiplexing-multiple input-multiple output (FDM-MIMO) Automotive radar systems. To estimate the DOA, we propose two approaches. In the first approach, with the help of a designed machine learning network, waveforms that are associated with interference are identified and excluded from the MUSIC algorithm for the DOA estimation. In the second approach, instead of removing the waveforms contaminated with interference, we use the variational mode decomposition (VMD) technique to reduce the interference of the waveforms. Finally, the DOA is estimated with the help of a novel weighted-MUSIC algorithm whose weights are determined based on the signal to the interference ratio (SIR) output from the VMD algorithm.