Algorithms for Localization and Tracking of Acoustic Sources with Widely Separated Sensors

摘要

Multiple sensor arrays distributed over a region provide the means for accurate localization of the (x, y) position of a source. When microphone arrays are used to measure aeroacoustic signals from ground vehicles, random fluctuations in the air lead to frequency-selective coherence of the signals that arrive at widely-separated arrays. We have shown previously that even in cases of imperfect spatial coherence, improvements in source localization accuracy are possible when the data from widely-separated arrays are processed jointly by a fusion center. Further, we have shown that a distributed processing scheme involving bearing estimation at individual arrays and time-delay estimation between pairs of widely-separated sensors performs nearly as well as the optimum scheme, with significantly lower communication bandwidth. These results were obtained by studying the Cramer-Rao bound (CR13) on source localization accuracy based on a statistical model for the data measured at the sensors. The contributions of this paper include the presentation of more accurate performance bounds (Ziv-Zakai), and the development of a narrowband subspace algorithm for source localization with distributed arrays and partially coherent signals. We demonstrate through analysis, simulation, and processing of measured data that the performance of both algorithms is limited by ambiguities that arise from the narrowband signals and the large spacing between arrays.