Data fusion applied to beamforming measurement

Abstract

The aim of this work is use data fusion in a set of data obtained from measurements done with a microphone array in different times to improve beamforming results. Beamforming is a technique that basically samples the sound field with an array of sensors. The correct summation of these signals will render a reinforcement of the recorded sound for a chosen direction in space. In addition, processing a set of possible incoming directions enables the creation of sound maps. The spatial resolution in beamforming is directly related to array’s constructive factors and frequency of analysis. One way to improve resolution is increasing array’s size and number of sensors. Considering the measured source statistically stationary, it is possible to use signals obtained in different times to evaluate it. In this way, the array can be placed in different positions, and the data acquired can be processed and fused in order to create a single set of data corresponding to a virtual array composed by all aforementioned positions.