Radio Frequency Fingerprinting Identification for Drones Based on the Generalized S-Transform

The rapid proliferation of Unmanned Aerial Vehicles (UAVs) has introduced critical safety concerns, while traditional UAVs identification approaches have proven inadequate in addressing these challenges. As a result, integrating radio frequency (RF) signal recognition with deep learning has become a critical research focus. This study presents a novel deep learning-based RF signal recognition framework that adeptly addresses the non-stationary nature of UAVs RF signals by transforming raw in-phase/quadrature (I/Q) samples into spectrograms using the Generalized S-transform. Furthermore, the framework introduces a novel data augmentation technique called Phase-Shift Data Augmentation (PSDA), improving the model's robustness. The proposed approach is anticipated to deliver an effective solution for UAVs management and control, with the PSDA method demonstrating superior performance across specific signal-to-noise ratio (SNR) ranges compared to existing Radio Frequency Fingerprinting identification(RFFI) techniques.