A High Resolution Method for Acoustic Localization in Far-field Based on Sliding DFT Algorithm

Abstract

In order to accurate the time delay estimation and orientation localization, in this paper, we are aimed at finding a high resolution algorithm to improve the deviation in the far-field source localization. Compared with the traditional method based on the cross-correlation algorithm which often corresponds with the interpolation algorithm, the method of Sliding DFT is a novel time delay estimation approach based on the Sliding DFT analysis window, which is more comprehensive and accurate. The signal is specified into a singularity frequency signal and this algorithm of Sliding DFT uses the sample number at the maximum magnitude of the spectrum as the first estimation and utilize its corresponding phase offset for compensation with the aim to estimate the time delay of the signal, which is a rather new method applied to the acoustic localization field in this paper. The simulation includes a simple acoustic source with singularity frequency using the traditional cross-correlation algorithm and the improved Sliding DFT algorithm, accompanied with the matrix transformation method to compute the position of the assumed signal. After calculating and comparing the statistic of both algorithms, the improvement of the accuracy can be easily seen.