Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

6 Citationer (Scopus)
135 Downloads (Pure)

Resumé

In many spectral estimation and array processing problems, the process
of finding estimates of model parameters often involves the optimisation
of a cost function containing multiple peaks and dips. Such
non-convex problems are hard to solve using traditional optimisation
algorithms developed for convex problems, and computationally intensive
grid searches are therefore often used instead. In this paper,
we establish an analytical connection between the grid size and the
parametrisation of the cost function so that the grid size can be selected
as coarsely as possible to lower the computation time. Additionally,
we show via three common examples how the grid size depends
on parameters such as the number of data points or the number
of sensors in DOA estimation. We also demonstrate that the computation
time can potentially be lowered by several orders of magnitude
by combining a coarse grid search with a local refinement step.
OriginalsprogEngelsk
TitelSignal Processing Conference (EUSIPCO), 2016 24th European
ForlagIEEE
Publikationsdatoaug. 2016
Sider1653-1657
ISBN (Elektronisk)978-0-9928-6265-7
DOI
StatusUdgivet - aug. 2016
Begivenhed European Signal Processing Conference - Hotel Hilton Budapest, Budapest, Ungarn
Varighed: 29 aug. 20162 sep. 2016
http://www.eusipco2016.org/

Konference

Konference European Signal Processing Conference
LokationHotel Hilton Budapest
LandUngarn
ByBudapest
Periode29/08/201602/09/2016
Internetadresse
NavnProceedings of the European Signal Processing Conference (EUSIPCO)
ISSN2076-1465

Fingerprint

Array processing
Cost functions
Sensors

Citer dette

Nielsen, J. K., Jensen, T. L., Jensen, J. R., Christensen, M. G., & Jensen, S. H. (2016). Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing. I Signal Processing Conference (EUSIPCO), 2016 24th European (s. 1653-1657). IEEE. Proceedings of the European Signal Processing Conference (EUSIPCO) https://doi.org/10.1109/EUSIPCO.2016.7760529
Nielsen, Jesper Kjær ; Jensen, Tobias Lindstrøm ; Jensen, Jesper Rindom ; Christensen, Mads Græsbøll ; Jensen, Søren Holdt. / Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing. Signal Processing Conference (EUSIPCO), 2016 24th European. IEEE, 2016. s. 1653-1657 (Proceedings of the European Signal Processing Conference (EUSIPCO)).
@inproceedings{43a2bda48a7148bd896e381fc6d571f9,
title = "Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing",
abstract = "In many spectral estimation and array processing problems, the processof finding estimates of model parameters often involves the optimisationof a cost function containing multiple peaks and dips. Suchnon-convex problems are hard to solve using traditional optimisationalgorithms developed for convex problems, and computationally intensivegrid searches are therefore often used instead. In this paper,we establish an analytical connection between the grid size and theparametrisation of the cost function so that the grid size can be selectedas coarsely as possible to lower the computation time. Additionally,we show via three common examples how the grid size dependson parameters such as the number of data points or the numberof sensors in DOA estimation. We also demonstrate that the computationtime can potentially be lowered by several orders of magnitudeby combining a coarse grid search with a local refinement step.",
keywords = "optimisation, DOA estimation, fundamental frequency estimation, periodogram",
author = "Nielsen, {Jesper Kj{\ae}r} and Jensen, {Tobias Lindstr{\o}m} and Jensen, {Jesper Rindom} and Christensen, {Mads Gr{\ae}sb{\o}ll} and Jensen, {S{\o}ren Holdt}",
year = "2016",
month = "8",
doi = "10.1109/EUSIPCO.2016.7760529",
language = "English",
pages = "1653--1657",
booktitle = "Signal Processing Conference (EUSIPCO), 2016 24th European",
publisher = "IEEE",
address = "United States",

}

Nielsen, JK, Jensen, TL, Jensen, JR, Christensen, MG & Jensen, SH 2016, Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing. i Signal Processing Conference (EUSIPCO), 2016 24th European. IEEE, Proceedings of the European Signal Processing Conference (EUSIPCO), s. 1653-1657, Budapest, Ungarn, 29/08/2016. https://doi.org/10.1109/EUSIPCO.2016.7760529

Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing. / Nielsen, Jesper Kjær; Jensen, Tobias Lindstrøm; Jensen, Jesper Rindom; Christensen, Mads Græsbøll; Jensen, Søren Holdt.

Signal Processing Conference (EUSIPCO), 2016 24th European. IEEE, 2016. s. 1653-1657 (Proceedings of the European Signal Processing Conference (EUSIPCO)).

Publikation: Bidrag til bog/antologi/rapport/konference proceedingKonferenceartikel i proceedingForskningpeer review

TY - GEN

T1 - Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing

AU - Nielsen, Jesper Kjær

AU - Jensen, Tobias Lindstrøm

AU - Jensen, Jesper Rindom

AU - Christensen, Mads Græsbøll

AU - Jensen, Søren Holdt

PY - 2016/8

Y1 - 2016/8

N2 - In many spectral estimation and array processing problems, the processof finding estimates of model parameters often involves the optimisationof a cost function containing multiple peaks and dips. Suchnon-convex problems are hard to solve using traditional optimisationalgorithms developed for convex problems, and computationally intensivegrid searches are therefore often used instead. In this paper,we establish an analytical connection between the grid size and theparametrisation of the cost function so that the grid size can be selectedas coarsely as possible to lower the computation time. Additionally,we show via three common examples how the grid size dependson parameters such as the number of data points or the numberof sensors in DOA estimation. We also demonstrate that the computationtime can potentially be lowered by several orders of magnitudeby combining a coarse grid search with a local refinement step.

AB - In many spectral estimation and array processing problems, the processof finding estimates of model parameters often involves the optimisationof a cost function containing multiple peaks and dips. Suchnon-convex problems are hard to solve using traditional optimisationalgorithms developed for convex problems, and computationally intensivegrid searches are therefore often used instead. In this paper,we establish an analytical connection between the grid size and theparametrisation of the cost function so that the grid size can be selectedas coarsely as possible to lower the computation time. Additionally,we show via three common examples how the grid size dependson parameters such as the number of data points or the numberof sensors in DOA estimation. We also demonstrate that the computationtime can potentially be lowered by several orders of magnitudeby combining a coarse grid search with a local refinement step.

KW - optimisation

KW - DOA estimation

KW - fundamental frequency estimation

KW - periodogram

U2 - 10.1109/EUSIPCO.2016.7760529

DO - 10.1109/EUSIPCO.2016.7760529

M3 - Article in proceeding

SP - 1653

EP - 1657

BT - Signal Processing Conference (EUSIPCO), 2016 24th European

PB - IEEE

ER -

Nielsen JK, Jensen TL, Jensen JR, Christensen MG, Jensen SH. Grid Size Selection for Nonlinear Least-Squares Optimization in Spectral Estimation and Array Processing. I Signal Processing Conference (EUSIPCO), 2016 24th European. IEEE. 2016. s. 1653-1657. (Proceedings of the European Signal Processing Conference (EUSIPCO)). https://doi.org/10.1109/EUSIPCO.2016.7760529