UAV Complex Interference Signal Recognition Based on Lightweight Neural Network

In order to improve the complex interference signals recognition capability of UAVs, this paper proposes an efficient and accurate approach for classifying complex interference signals based on lightweight neural network. Firstly, thirteen types of different interference signals are generated. Then, these signals are transformed into time-frequency spectrum by short-time fourier transform (STFT) to construct the complex communication interference signal dataset. Finally, the MobileNetV2 lightweight neural network model is used to extract interference features and classify the interference signals. Simulation results show that compared with other interference signal recognition algorithms based on lightweight neural networks, including PPLCNetV2, ShuffleNetV2 and GhostNet, the proposed algorithm has better performance than others in terms of recognition accuracy and inference time.