Acoustic Vector Sensor Based Source Localization

The conventional techniques to estimate the Direction-of-Arrival (DOA) of acoustic sources includes employing multiple Acoustic Pressure Sensors (APSs) in an array configuration. Recently with the advent of MEMS technology, the sensors are capable of capturing acoustic particle velocity along with acoustic pressure and these sensors are referred to as Acoustic Vector Sensors (AVSs). In order to capture the velocity field information, AVSs contain two (three) vector sensors which are aligned along corresponding coordinate axes in the two (three) dimensional acoustic space. Based on narrow-band processing techniques and for a single far-field source scenario, we discuss the advantages obtained by an AVS array in comparison to a conventional APS array based on their respective beam patterns. Later the discussion is extended to a configuration where each AVS in the array contains only one vector sensor (which measures one component of the vector field), such that it can have a variable orientation, and a pressure sensor, which is referred to as a 1D AVS. We discuss the similarities and differences between 2D AVS arrays and 1D AVS arrays through beam patterns. Also we discuss the bias in the source location estimate obtained by using a 1D AVS array. Further we derive Cramer-Rao bounds for 2D AVS, 1D AVS and APS array configurations. Finally we compare the performance of 1D AVS arrays and 2D AVS arrays in the covariance domain in terms of the localization of quasi-stationary sources.

To reference this document use:

http://resolver.tudelft.nl/uuid:cfd81ca7-e44c-4d0d-8680-a9150f9d1162

Embargo date

2016-06-23

Part of collection

Student theses

Document type

master thesis

Rights

Files

PDF

thesis_KPNR_latest.pdf

11.93 MB

Close viewer