Classification and Characterization of RFI Events for Passive Earth Observation Bands

Passive Microwave remote sensing is a very energy-efficient and time-proven technology for remote sensing. Microwave Radiometry and Global Navigation Satellite System (GNSS)-Reflectometry are two widely-used examples. The increasing effects of Radio-Frequency Interferences (RFI) on the performance of these receivers have sparked serious concerns due to their proliferation. Detection and mitigation of RFI heavily rely on the identification and location of the interference sources. Due to the vast amount of degrees of freedom found in these RFIs, classification using classical signal processing techniques becomes problematic, especially if combined RFIs are found within the same bandwidth. In this work, continuous recordings of real RFI events from a real-time system to detect and locate RFI sources are used as training data for an automated classification system, the description and performance of which are presented and analyzed.