Decoder-Free Transformer-Based Track Prediction for Compact High-Frequency Surface Wave Radar

Compact high-frequency surface wave radar (HFSWR) systems play a critical role in maritime surveillance and over-the-horizon target tracking. However, due to their limited transmit power and low signal-to-noise ratio, compact HFSWR systems suffer from data discontinuity and interference from clutter and noise. As a result, they may fail to achieve accurate track prediction. This paper proposes a track prediction method based on a simplified, decoder-free Transformer model. It learns the motion patterns and long-term dependencies of target states from historical track data to achieve accurate and efficient state prediction under diverse conditions. Experimental results demonstrate the effectiveness of the proposed method in reducing prediction errors, resisting clutter interference, and improving the tracking continuity of compact HFSWR systems.