Successive Multi-RFI Suppression in Raw SAR Data via Space-Frequency Domain Processing

Abstract

The growing prevalence of Radio Frequency Interference (RFI) in Synthetic Aperture Radar (SAR) systems poses a critical challenge to data quality, particularly at the raw signal level, severely degrading image quality and compromising the accuracy of Earth observation applications. This paper introduces a successive multi-RFI mitigation framework that leverages space-frequency domain processing to suppress interference directly at the raw data level. The proposed approach integrates successive detection, localization, and suppression of multiple narrowband interference sources using a combination of Constant False Alarm Rate (CFAR)- based spectral analysis and adaptive spatial filtering. Key RFI characteristics, including center frequency, bandwidth, Signal-to-Interference Ratio (SIR), and azimuthal position, are estimated to guide dynamic suppression masks that preserve underlying SAR signal structure. The framework is validated through extensive Monte Carlo simulations on raw synthetic datasets generated using an open-source SAR emulator (SEMUS), as well as on real raw level-0 TerraSAR-X data acquired over Los Angeles. Results demonstrate accurate frequency and spatial localization of RFI, with a mean SIR improvement of 5.9 dB, azimuth localization recall of 95.05%, and high suppression effectiveness across diverse RFI scenarios. Additional validation confirms that the proposed approach outperforms the conventional notch filter while preserving the underlying signal structure. The approach provides a computationally efficient and scalable solution for RFI mitigation in raw SAR data, advancing the feasibility of real-time onboard interference suppression in next generation SAR missions.