Wideband RFI Mitigation for Spaceborne SAR by Acquiring Prior Information via Time–Frequency Analysis
Abstract
Operating in shared frequency bands, synthetic aperture radar (SAR) faces a complex electromagnetic environment, making radio frequency interference (RFI), particularly wideband interference (WBI), an increasingly prominent issue. WBI typically exhibits time-varying characteristics across pulses, with its amplitude and temporal location varying significantly. Despite these variations, WBI originating from the same source generally maintains stable intrinsic properties, such as bandwidth and duration, which can serve as useful prior information for WBI mitigation. However, due to the difficulty of WBI modeling, existing mitigation methods often bypass WBI parameter extraction, neglecting the utilization of prior information. Furthermore, these methods typically apply fixed filtering strategies across the entire WBI-containing pulse, ignoring variations in scattering intensity across different terrains. This leads to degraded mitigation performance in complex scenes. To address these issues, a WBI mitigation method that acquires prior information is proposed in this article. Specifically, through time–frequency analysis, WBI prior information is extracted from representative pulses identified via a spectral flatness indicator. WBI localization can be achieved by independently clustering strong- and weak-scattering regions, further optimized by the prior information. Finally, the eigen-subspace projection (ESP) method is applied to the localized time–frequency regions to mitigate WBI. Comprehensive experimental results on SAR data from diverse scenes confirm the effectiveness and superiority of the proposed method.