SAR Interference Mitigation using a Filter Bank Time-Frequency Distribution and L-Statistics

Interference originating from ground-based active sources, such as radar systems, can represent a substantial obstacle to achieving high-quality and dependable Synthetic Aperture Radar (SAR) imagery. This interference can lead to partial image degradation, errors in image interpretation, and deviations in parameter extraction. Notably, the Sentinel-1, operating in C-band, is among the most widely used SAR remote sensing satellites. Given the prevalence of ground-based radar systems operating in the C-band, interference significantly impacts Sentinel-1 data, thereby necessitating effective mitigation strategies. In this article, we propose an innovative approach for mitigating interference in SAR images by employing a combination of a filter bank time-frequency distribution (dual of the classical Short-Time Fourier Transform) computed on the spectrum of the range compressed signal and a linear combination of order statistics (L-statisics). The methodology involves sorting each constant-time line of the time-frequency distribution matrix in ascending order, which places the interference at the right-hand side of the time-frequency plane. Subsequently, the interference-free range profile is determined by coherently summing along the frequency axis of the bins identified as unaffected by interference (the first q% of frequency bins from the sorted distribution). The method is evaluated on a dataset acquired by Sentinel-1A on 26.04.2021 over the city of Doha, Qatar, which is affected by interference from a radar of the Patriot Missile System.