Performance analysis and improved DOA estimation for a single acoustic vector sensor under non-orthogonal deviation

Acoustic vector sensor (AVS) has gained widespread application in estimating the direction of arrival (DOA). However, the orthogonality of the velocity axes of the AVS cannot usually be guaranteed, which may significantly impact the performance of DOA estimation. To illustrate the influence of non-orthogonal deviation (ND) on DOA estimation, the Cramer-Rao Lower Bound (CRLB) and Root Mean Squared Error (RMSE) of the non-orthogonal AVS (N-AVS) models are deduced. The theoretical analysis and simulation results show that the ND brings a considerable impact on the accuracy of DOA estimation. Furthermore, to suppress the impact of the ND on the performance of DOA estimation, a ND matrix modification (NDMM) algorithm is proposed. Simulation and experimental data results have verified that the proposed method improves the DOA estimation accuracy using a single N-AVS.