Radar Waveform Recognition based on Deep Residual Network

This article presents our initial results in deep learning for the complex multiple radar waveforms recognition. The method is composed of time-frequency analysis and deep residual network (ResNet). Firstly, we transform one-dimensional radar signals into two-dimensional time-frequency images (TFIs), which can reveal more characteristics of the signals. And then, we preprocess these images by grayscale, image opening operation and image resizing. Meanwhile, we design a ResNet and sent these preprocessed images into the network for off-line training. Finally, we use the trained model to recognize different modulation radar signals on-line and test the performance of the method. From our simulation results, the approach can achieve considerable performances that the overall recognition rate of 14 types radar signal is close to 96% when the SNR is −2dB.

[1]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[2]  Ming Zhang,et al.  Convolutional Neural Networks for Automatic Cognitive Radio Waveform Recognition , 2017, IEEE Access.

[3]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[4]  Chao Wang,et al.  Automatic radar waveform recognition based on time-frequency analysis and convolutional neural network , 2017, 2017 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[5]  Gaoming Huang,et al.  Automatic Radar Waveform Recognition Based on Deep Convolutional Denoising Auto-encoders , 2018, Circuits Syst. Signal Process..