Dual-Domain Denoising Transformer for Precise Anti-Jamming Signal Reconstruction

Abstract

In modern wireless communication, radar, and intelligent sensing, recovering signals exposed to complex and adaptive interference is an issue that needs to be addressed. This paper proposes a novel dual-domain denoising transformer structure (D-DDT) that can effectively integrate time-frequency domain information, thereby enhancing anti-jamming capabilities. Considering the urgent need for accurate signal recovery in dynamic electromagnetic fields, a dual-branch encoder path and adaptive cross-domain attention mechanism are introduced. Conducted on multiple datasets, including real-world and simulated interference environments, with rigorous testing using filter-based and learning-based algorithms. DDT enhances noise suppression effects, improving the signal-to-noise ratio (SNR) compared to other methods in the comparison. Bilateral domain representation and adaptive attention are two necessary conditions for achieving good stability. In practical application tests, it can effectively handle various resource levels under time constraints. As a new participant in the complex anti-jamming signal recovery benchmark test, it contributes to subsequent communication and detection capabilities. Future research will enhance the adaptability of interference models and the optimization of edge deployments.