Convolutional Neural Networks for Classification of Drones Using Radars

The ability to classify drones using radar signals is a problem of great interest. In this paper, we apply convolutional neural networks (CNNs) the Short-Time Fourier Transform (STFT) spectrograms simulated reflected from drones. vary in many ways that impact STFT spectrograms, including blade length and rotation rates. Some these physical parameters are captured Martin Mulgrew model which was used produce datasets. We examine data under X-band W-band simulation scenarios show CNN approach leads an F1 score 0.816±0.011 when trained on with signal-to-noise ratio (SNR) 10 dB. network 2 kHz pulse repetition frequency shown perform better than aforementioned radar. It remained robust drone pitch its performance varied directly linear fashion SNR.