Distributed Processing for 3-D Localization Using Acoustic
Vector Sensors on the Seabed or Battlefield

M. Hawkes and A. Nehorai
Department of Electrical Engineering and Computer Science
The University of Illinois at Chicago
Email: hawkes@sensor.eecs.uic.edu

 

Abstract We derive fast wideband algorithms for locating the bearing and 3-D location of acoustic targets using a distributed array of acoustic vector sensors located at a reflecting boundary.  Our technique is relevant to the localization of submarine and surface sources using seabed sensors, and of airborne targets over a battlefield using ground sensors.  An acoustic vector sensor (AVS) measures the acoustic pressure and all three components of particle velocity at a single point and such sensors have been constructed for underwater and atmospheric applications.  The sensors, whose locations may be arbitrary because an AVS array suffers from no ambiguities due to spatial undersampling, each make a local estimate of the bearing from their position to the target. We derive a fast bearing estimate based on the acoustic intensity vector and another based on an adaptation of minimum-variance beamforming, for both the underwater and in-air scenarios.  These local bearing estimates are then transmitted to a central processor where they are combined to determine the 3-D location.  We propose a closed-form weighted least-squares (WLS) and a reweighted LS algorithm to achieve this.  Since the AVS components are co-located and correlations between measurements at different locations are not used, these algorithms are applicable to wideband signals.  Numerical simulations for both the underwater and battlefield scenario show the effectiveness of our solutions.  We also present a bound on the mean-square angular error of the local bearing estimate and use it along with the data to adaptively determine the weights for the WLS routine, as well as asses the quality of the local bearing estimators.

Presentation in PDF format