Programmable metasurface DOA estimation via DnCNN and music

For programmable metasurface based direction of arrival (DOA) estimation, existing algorithms often fail to achieve the expected results under low signal-to-noise ratio (SNR) conditions. To tackle this problem, we propose a DnCNN-MUSIC algorithm that combines the De-Noising Convolutional Neural Network (DnCNN) with the multiple signal classification (MUSIC) algorithm. Specifically, the proposed algorithm first applies DnCNN to denoise the signal covariance matrix, and then uses the MUSIC algorithm for DOA estimation. Simulation results show that the proposed algorithm outperforms existing algorithms in DOA estimation accuracy under low SNR and small snapshot conditions. It also demonstrates advantages in resolving targets with small angle intervals.