Radar Signal Recognition Based on Deep Residual Network with Attention Mechanism

To address the problem of poor recognition success rate (RSR) of low probability of intercept (LPI) radar signals such as polyphase codes at low signal-to-noise ratio (SNR), In this paper, we propose a method of deep neural network with attention mechanism for modulation recognition of radar signals. The method first transforms the received radar signal into a two-dimensional time-frequency image(TFI), and then we obtain a binary time-frequency image by threshold segmentation, and resize TFIs through bicubic interpolation. While a deep residual network with Convolutional Block Attention Module (CBAM) is designed to recognize the radar signals. We take advantage of the attention mechanism to find the effective feature information in the noise and increase the effective information weight to further improve the recognition success rate of radar signals under low SNR. Finally, the average RSR of the method is nearly 96% for 7 signals at -6dB.