A Self-Attention Residual Network for SAR Jamming Classification with Multi-Domain Feature Fusion

With the increasing widespread use of synthetic aperture radar(SAR) systems in various environments, jamming have become a serious problem that they face. In many situations, these jamming affect SAR systems' ability to gather information. Many anti-jamming methods are based on the classification of jamming types. In order to provide information about jamming types, it is necessary to design a classification method which can classify multiple types of jamming. Considering the complexity of jamming features, in this paper, multi-domain jamming feature analysis and a residual network with convolutional block attention module (CBAM) are proposed to classify SAR jamming. To train this jamming classification network and validate its effectiveness, a database containing different jamming simulations is generated. The simulation results show that this method has reliable classification performance for various types of jamming.