A Novel Jamming Signal Recognition Method Based on Data Augmentation Using 1D-GAN under Small Sample Condition

Various jamming signals in the complex electromagnetic environment pose a serious threat to radar detection. Effective recognition of jamming type is of great significance for anti-jamming. In recent years, jamming recognition algorithms based on deep learning have been proposed. With a sufficient number of samples, these algorithms can obtain high recognition accuracy. However, in the actual battlefield environment, it is difficult to accurately obtain large amounts of measured samples with clear labels, and the accuracy of jamming recognition cannot be guaranteed. This paper proposes a jamming recognition method based on a one-dimensional adversarial generative network (1D-GAN) for small sample conditions. Extra samples are generated by using this 1D-GAN and these generated samples are combined with the original data to form an augmented data set. Then a one-dimensional convolutional neural network (1D-CNN) is used for jamming recognition. The fidelity of the generated samples is verified in three dimensions data domain, feature domain, and jamming recognition accuracy. The experiment results show that the generated data has high similarity with the real data, and data augmentation by GAN can effectively improve the jamming recognition performance in the case of a small number of samples.