Detecting and Classifying Radar Signals in SAR Raw Data Obtained by Satellites

The growing number of satellites-especially in low Earth orbit (LEO). Many of these satellites are radar satellites providing an essential civilian and military service. Increasingly, radar-satellite constellations have emerged yielding a much higher re-visit rate, with other tactical advantages, than is possible with a single radar-satellite. For example, radar satellite constellations like Sentinel-1 which carry a C -band synthetic aperture radar (SAR) providing an essential Earth observation and remote sensing service. SAR systems provide reliable, allweather imaging, but data quality can be impacted by Radio Frequency Interference (RFI) from external systems that are in-band to the radar-satellite. This is exactly the focus of this work - to identify the sources of interference in-band to the SAR imaging satellite. Of particular interest is the presence of other, interfering, radar systems that are in-band of the SAR system. This work analyses satellite SAR data and implements a signal detection method to detect interfering radar signals and to estimate key parameters of the interfering-radar, such as bandwidth, center frequency, duration, and chirp rate. The method is evaluated using simulated and real satellite SAR data (Sentinel1), demonstrating effective detection accuracy and robustness to parameter variations, with promising results for wider RFI detection and feature extraction. This approach integrates CFAR filtering, gradient-based shape analysis, and DBSCAN clustering into a unified pipeline, enabling robust pulse characterization directly from Sentinel-1 raw I/Q data. This method extracts RFI signal features-including bandwidth, chirp rate, and center frequency-proving to be an important component of spacebased intelligence, surveillance, and reconnaissance (SBISR)