Deep Learning-Based Individual Drone Identification Using Raw IQ Data

In this letter we propose a novel deep learning-based method for drone identification using raw in-phase and quadrature (IQ) data, addressing the increasing need for monitoring and identifying drones. The proposed approach directly inputs the collected raw IQ signals into convolutional neural networks (CNNs) without requiring traditional time-frequency transformations, thus simplifying the preprocessing steps. Experimental results demonstrate that the method achieves nearly 100% identification accuracy on 10 different individual drones, significantly outperforming existing methods in classification performance and computational complexity. This study validates the feasibility and superiority of using raw IQ data with CNNs for individual drone recognition, offering a new solution to enhance the real-time reliability of drone identification systems.