Semantic Feature-Enhanced Graph ATtention Network for Radar Target Recognition in Heterogeneous Radar Network

Radar target recognition (RTR), as a key technique of intelligent radar systems, has been widely investigated. Accurate RTR at low signal-to-noise ratios (SNRs) still remains an open challenge. Considering that most existing methods are based on single or the homogeneous network, we extend to heterogeneous network improve robustness RTR, which uses cross Section (RCS) signals SNRs by further exploiting frequency-domain information. In this article, Semantic Feature-Enhanced Graph ATtention Network (SFE-GAT) is proposed, extracts semantic features from both source and transform domains via long short-term memory (LSTM) GAT layers, then fuses them in space using attention mechanism, distills higher-level layer before classification. Extensive experiments carried out validate proposed SFE-GAT model can greatly accuracy SNR region.