Localizing multiple targets with a single particle velocity sensor

Abstract

The acoustic velocity sensor (AVS) measures the direction of a propagating plane wave by measuring instantaneously the particle velocity over a small volume in space. When multiple sound sources are present, the AVS no longer reports the correct directions as the sound intensity vectors (the product of pressure and velocity vector) get mixed up nonlinearly. A few methods have been proposed to extract the individual components from just one sensor by assuming, for example, spectral separability or nongaussianity. This paper shows it is in fact possible to extract the components without making such assumptions. The approach here is to constrain the signal with the acoustic impedance equation for multiple superposing plane waves and, applying various techniques, such as sparsity decomposition, to solve the derived least squares problem.