Acoustic Source Localization using Microphone Arrays by TDOA Method

Abstract

Acoustic source localization using an array of microphones and time difference of arrival (TDOA) method is demonstrated. Adaptation of such a scheme into combustors or reactors is expected to help assist in identification of local acoustic sources that may be present for improved operation of the system. Acoustic source localization algorithm made in MATLAB TM is implemented using microphone array arranged in defined configurations. For any pair of microphones, the surface on which the TDOA is constant is a hyperboloid of two sheets. Acoustic source localization algorithm presented here groups all microphones into pairs for estimating the TDOA between them and then estimating the point where all associated hyperboloids most nearly intersect. Acoustic source positioned in two-dimensional plane is estimated using defined configurations of microphone array in space. Acoustic source localization by hyperbolic position fix has been examined using different microphone array geometries. The effect of selected configurations to identify the noise source was carried out by changing the coordinates of the noise source and estimating the position of the noise source by using hyperbolic position location algorithm. Different geometries were investigated by changing number of sensors and source locations to localize the noise source. Simulations and experiments were carried out for localizing the noise source in a two-dimensional plane. Amongst the geometries investigated, curvilinear array with multiple microphones showed favorable results in a two-dimensional plane. Increasing the number of sensors in a given array configuration increased the performance of source location estimation based on the results obtained from array of different configurations. Effect of increasing number of sensors in a given arrangement on the performance of hyperbolic position location showed improved location of the noise source with large local acoustic powers.