MCF-Net: A Mamba-based Efficient Network for Radar Jamming Recognition

Deception jamming exhibits high coherence with target echoes, resulting in severe time-frequency aliasing that obscures discriminative features and poses significant challenges for accurate jamming recognition in complex and dynamically varying electromagnetic environments. Existing methods often transform signals into image representations and apply computer vision models for classification. However, such transformations introduce additional computational overhead during dataset generation and risk information loss. To tackle with these problems, we propose the Mamba-based Complex Signal Fusion Network (MCF-Net), an efficient deep learning architecture and the first to apply the Mamba framework to radar jamming recognition. MCF-Net directly processes complex-valued radar jamming, enabling dynamic perception of signal variations and efficient modeling of cross-channel dependencies. A complementary dynamic low-rank projection attention module efficiently captures contextual dependencies. Additionally, an orthogonal projection regularization term is incorporated to enhance feature discriminability and inter-class separability. Experimental results demonstrate that MCF-Net achieves state-of-the-art recognition accuracy with the lowest inference latency among competitive models, making it highly promising for real-time deployment.