Low-Altitude Target Angle Estimation Based on Soft Threshold and Attention Mechanism

The traditional neural network-based angle-direction estimation methods are mainly for non-coherent signals, and it is difficult to effectively achieve the accurate estimation of direction of arrival in the low altitude multipath environment with low SNR. In this paper, we propose an innovative neural network algorithm that combines the soft threshold module and attention mechanism. The introduction of the soft threshold module aims to reduce signal noise and the attention mechanism is employed to further explore the connection between different channel features in order to enhance angle measurement performance. Simulation results demonstrate that the proposed algorithm significantly improves the accuracy of angle measurement and exhibits robustness under low SNR.