Robust Lightweight RF Fingerprinting Identification for High-Mobility Scenarios

RF fingerprinting has gained widespread application in wireless communication, particularly in the Internet of Things (IoT), due to its low cost and high security based on physical-layer characteristics. However, rapidly changing channels introduce multipath fading and Doppler effects in high-mobility scenarios, making RF fingerprint extraction challenging. Introducing additional training parameters to mitigate channel variations is impractical due to the resource limitations of IoT devices. To address these challenges, this paper proposes a lightweight and efficient RF fingerprint extraction and classification framework. The method involves generating spectrograms through Multi-angle Averaging Linear Canonical Transform (MALCT), and classifying the features with a lightweight deep neural network. Experimental results show that our method achieves superior performance with minimal computational and storage costs, while exhibiting outstanding robustness in high-mobility scenarios. This makes it highly suitable for deployment on resource-constrained IoT devices, effectively addressing the challenges posed by rapid channel variations.