Time-Varying RFI Mitigation for SAR Systems via Graph Laplacian Clustering Techniques

As a wideband radar system, the synthetic aperture radar (SAR) usually conflicts with several electromagnetic systems, such as frequency modulation (FM), TV, and other communication systems. These signals, which are radio frequency interference (RFI) for radar systems, severely interfere with SAR systems to generate a high-resolution image. Some previous parametric methods focused on the time-varying RFI model; however, they cannot realize the comparable effectiveness and efficiency against semi-parametric methods. However, previous semiparametric methods did not focus on the time-varying RFI case. Hence, in this letter, a graph Laplacian clustering (GLC) semiparametric algorithm is proposed to suppress RFIs by constructing the Laplacian embedding connections between different pulses of signals. As a result, locally time-varying interferences are clustered in a nonlinear low-dimensional manifold and can be effectively mitigated. The real SAR data with measured RFIs are provided to demonstrate the effectiveness and efficiency of the proposed algorithm.