Radio frequency interference suppression in ultra-wideband synthetic aperture radar using range-azimuth sparse and low-rank model

Ultra-wideband (UWB) Synthetic Aperture Radars (SAR) operate over a large bandwidth ranging from under 100 MHz to over a few Ghz. They often share spectrum with other systems such as radio, TV and cellular networks. The mitigation of radio frequency interference(RFI) from these sources is an important problem for UWB SAR systems. Traditional RFI suppression techniques such as notch filtering introduce side effects such as large sidelobes or poor peak-to-sidelobe ratio. More recently, methods based on sparsity and compressive sensing that do not have these side effects have been proposed. In particular, a sparse and low-rank method that models SAR data as a linear combination shifted SAR pulses and RFI to be of low-rank has been found to be effective. This model however uses the structure of SAR data in down-range direction only and ignores the structure in azimuth direction. In this paper, we propose to replace the data model with a new sparse model that incorporates structure in azimuth direction as well. We demonstrate that the new model has significantly better performance than the previously proposed model. It performs robustly even in the presence of high level of noise(-20 dB SNR) and does not suppress small targets like the previously proposed model did.