Interference Mitigation in FMCW Automotive Radars using the Short-Time Fourier Transform and L-Statistics

In this paper we propose a novel method for the mitigation of uncorrelated frequency-modulated continuous-wave (FMCW) automotive radar interferences based on the short-time Fourier transform (STFT) and L-statistics. The idea of the method is to compute the STFT of the beat signal and sort each constant frequency line of the spectrogram in ascending order, which places the bins affected by interference towards the right hand side of the time-frequency plane. Onwards, the interference-free range profile is computed by a coherent summation along the time axis of the bins considered to be unaffected by interference (the first q% of time bins from the sorted spectrogram). The impact of the STFT parameters (sliding window length and step size) and the summation percentage q on the method's performances and computation time is investigated using a synthetic dataset that contains 24000 samples of realistic automotive radar signals (with multiple targets and interference sources). It is shown that the developed method can mitigate multiple interference sources. Moreover, by increasing the sliding window step size to half the window's length we obtain similar performances to the case with unitary step size (the classical STFT definition) and the computation time decreases significantly, which makes the method more suitable for real time implementation on a digital signal processor. The method is compared with state-of-the-art interference mitigation approaches on the synthetic dataset and real data acquired with a W-band automotive radar.