Radio Modulation Recognition Based on Attention-based Convolutional Neural Network and Time-frequency Analysis

Radio modulation recognition is the key link of modern electronic warfare. This paper applies the idea of deep learning to radio modulation recognition. Since the modulation type is the most important information about the subtle features of the radio signal, we use short time Fourier transform (STFT) to convert the radio signal into a time-frequency distribution map to form a joint distribution of frequency and time. The attention-based convolutional neural network is used to classify and recognize the modulation types of time-frequency distribution maps. The simulation results show that time-frequency analysis can improve the accuracy of recognition, and adding attention can effectively improve the performance of the network.