Identifiability Results for Nonuniform Linear and Rectangular Sensor Arrays

Sensor arrays with simple geometries play an important role in solving direction of arrival estimation and source separation problems. To reduce the number of sensors used, nonuniform linear and rectangular array geometries have been proposed. Unlike the case of uniform linear and rectangular arrays...

Full description

Saved in:
Bibliographic Details
Published in2022 30th European Signal Processing Conference (EUSIPCO) pp. 1736 - 1740
Main Authors Sorensen, Mikael, Sidiropoulos, Nicholas D.
Format Conference Proceeding
LanguageEnglish
Published EUSIPCO 29.08.2022
Subjects
Array processing
Array signal processing
canonical polyadic decomposition
Direction-of-arrival estimation
Europe
Geometry
harmonic retrieval
identifiability
Matrix decomposition
missing sensors
nonuniform linear array
nonuniform rectangular array
Parallel processing
Source separation
tensor
Online AccessGet full text
ISSN2076-1465
DOI10.23919/EUSIPCO55093.2022.9909588

Cover

More Information
Summary:Sensor arrays with simple geometries play an important role in solving direction of arrival estimation and source separation problems. To reduce the number of sensors used, nonuniform linear and rectangular array geometries have been proposed. Unlike the case of uniform linear and rectangular arrays, identifiability conditions for direction of arrival estimation and source separation using nonuniform linear and rectangular arrays are not well-studied. Based on rank properties of Fourier matrices and tools from algebraic geometry, we present generic identifiability conditions for direction of arrival estimation and source separation problems when nonuniform linear and rectangular array geometries are used. Furthermore, based on properties of bilinear factorizations subject to polynomial/monomial equality constraints, we also briefly discuss how to obtain deterministic identifiability conditions.
ISSN:2076-1465
DOI:10.23919/EUSIPCO55093.2022.9909588