RF-Based on Feature Fusion and Convolutional Neural Network Classification of UAVs

RF fingerprint identification is an identification technology that relies on the hardware characteristics of wireless devices. This paper proposes a UAV RF recognition scheme based on convolutional neural network and feature fusion of time-frequency spectrogram and bispectrogram. Firstly, we obtain the frequency spectrogram and bispectrogram characterizing the UAV RF signal, and then classify them by designing a two-way convolutional neural network after feature fusion in the fully connected layer, and the classification accuracy reaches 99.47%. Under the condition of low SNR, the classification accuracy of the model reaches 88.63% with good classification performance through the denoising mechanism.