Calibration of acoustic vector sensor based on MEMS microphones for DOA estimation

Abstract

A procedure of calibration of a custom 3D acoustic vector sensor (AVS) for the purpose of direction of arrival (DoA) estimation, is presented and validated in the paper. AVS devices working on a p-p principle may be constructed from standard pressure sensors and a signal processing system. However, in order to ensure accurate DoA estimation, each sensor needs to be calibrated. The proposed algorithm divides the calibration process into two stages. First, amplitude calibration is performed in order to compensate for amplitude differences between pairs of microphones situated on each axis. After the pressure and velocity signals are computed from the corrected microphone signals, the second stage is performed in order to correct phase differences between the pressure and velocity signals, which then allows for computing the intensity signals for each axis. In order to validate the calibration method, a reference AVS was constructed from low cost components, namely MEMS microphones and a DSP board. The method was engineered with an assumption that it will be applicable to any AVS working on the same principle. A set of experiments was performed in order to validate the calibration method and to compare the accuracy of the calibrated sensor with a commercial AVS. It was found in the experiments that DoA accuracy of the proposed 3D AVS calibrated with the proposed procedure matches that of a commercial, high cost, factory-calibrated sensor. Therefore, the proposed calibration method fulfills the requirements of accurate DoA estimation, and it is applicable to calibration of custom built, low cost AVS devices, that may be implemented in practical applications for determining the direction of sound sources, such as environmental monitoring, traffic monitoring and public security systems.