Real-valued robust DOA estimation method for uniform circular acoustic vector sensor arrays based on worst-case performance optimization

Abstract

The minimum variance distortionless response (MVDR) method suffers performance degradation when the considerable ambient noise and the steering vector mismatch exist. Therefore, this paper proposes a real-valued robust beamforming for the uniform circular acoustic vector sensor array (UCAVSA) based on worst-case performance optimization. The paper first extends the phase-mode transformation into the UCAVSA, and the relationship between the transformation matrices of the UCAVSA and the circular pressure sensor array is derived. The paper then utilizes the spatial correlation characteristic between the acoustic pressure and particle velocity to construct a cross-covariance matrix to eliminate the isotropic noise, and then transforms the cross-covariance matrix into a real-valued one by the unitary transformation. Finally, the paper derives the real-valued robust beamforming algorithm via optimizing the worst-case performance. Theoretical analyses and simulation results indicate that the proposed method is superior over previous methods under scenarios such as low signal-to-noise (SNR), steering vector mismatch and coherent sources. Moreover, the proposed method has a lower computational cost than the complex-valued operation in calculating the weight vector. The experiment results verify the effectiveness of the proposed method.