Estimation Of Direction-Of-Arrival Based On Acoustic Vector Sensor Array

Acoustic vector sensor is consist of pressure sensor and particle velocity sensor, which can synchronously pick up the pressure information(scalar) and velocity information(vector) in the sound field. Therefore, as compared with the conventional pressure sensor, the acoustic vector sensor may receive more complete information of sound field. But most of existing acoustic vector sensor array signal processing techniques are just using the velocity vector sensor information as an additional independent array elements to deal with, rather than make full use of related characteristics of the far-field sound pressure and vibration velocity. In fact, in the isotropic noise field, the sound pressure and velocity are uncorrelated; applying this feature, we can reduce the impact of interference generated by Gaussian white noise.This paper reviews the basic principles of acoustic vector sensor, the underlying physical basis of sound pressure velocity joint information processing and acoustic vector sensor array signal processing technology based on subspace algorithms. In addition, this paper focuses on the sound pressure velocity combined information processing technology to make a better use in acoustic vector sensor array signal processing, which can efficiently locate the target orientation in low SNR environment.This paper also introduces the joint information based on sound pressure and particle velocity of P-V cross-covariance matrix to process the received data signal. The new covariance matrix projects velocity information onto a viewing direction, avoiding the dimension of the received data matrix increasing. Thus the computational complexity of the acoustic vector sensor array signal processing reduces. In the subsequent processing of the received signal matrix, this paper combined the existing Unitary-MUSIC algorithm and Root-MUSIC algorithm to proposed a novel Unitary-Root MUSIC improved algorithm based on acoustic vector sensor array cross-covariance matrix. This new algorithm not only enhance the detection performance compared to the original algorithm, but also reduces the computational complexity. There are many simulations in the paper, such as RMSE from different SNR conditions, the probability of detection, spatial power spectrum and time-consuming. By comparing with the existing algorithms, simulations verify the excellent detection performance of the proposed algorithm.