Direction finding method based on differential covariance matrix diagonalization via acoustic vector sensor array under colored noise

To mitigate the challenge of degraded direction of arrival (DOA) estimation accuracy for acoustic vector sensor array (AVSA) under colored noise conditions, a novel differential covariance matrix diagonalization (DCMD) method is proposed in this paper. To effectively suppress the colored noise, a temporal differencing operation is first applied to the received data, which isolates the noise component and rigorously derives its variance. Subsequently, a truncated noise analysis is performed, theoretically demonstrating that the differencing transforms the original long-memory autoregressive noise into a finite-memory moving-average process, thereby achieving tridiagonalization of the temporal noise covariance matrix. Finally, the denoised covariance matrix is reconstructed by subtracting the modeled noise, enabling high-precision DOA estimation via the multiple signal classification algorithm. Simulation results demonstrate that the proposed DCMD method exhibits effectiveness and robustness in DOA estimation under colored noise environments compared to the existing state-of-the-art techniques.