liu.seSearch for publications in DiVA
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
On Calibration of Ground Sensor Networks
Linköping University, Department of Electrical Engineering, Automatic Control. Linköping University, The Institute of Technology.
2012 (English)Licentiate thesis, comprehensive summary (Other academic)
Abstract [en]

Sensor networks are everywhere around us. Developments in sensor technology and advances in hardware miniaturization open up brand-new application areas. In the future networks of cheap and small sensor nodes will be deployed for a variety of purposes. Military needs have been a major motivation for the development in the past, but today it has changed. Other applications such as traffic monitoring, security threat detection, ecology and environmental protection are the new driving forces behind further development.

The thesis considers the problem of calibration of ground sensor networks. In order to perform its operational tasks – detection, classification and tracking ofobjects of interest, the network has to be correctly calibrated. Improper calibration might result in a degraded performance, problems with data association and appearance of multiple track instances representing one object.

In order to find the unknown calibration parameters (biases), in most cases we need to use reference targets with known positions. If such targets are not available, one has to use opportunistic targets and simultaneously estimate both target positions and bias parameters. In this thesis, the expectation maximization algorithm is applied to that problem, where the unknown states are treated as latent (unknown) variables in the process of bias estimation.

Next, the problem of estimating a large number of calibration parameters is tackled. In the case when the measurement data is not informative enough – due to a limited range of sensors or a small number of samples – standard approaches such as the least squares algorithm might provide unreliable results. One solution to the problem is to apply a regularization (or prior in a Bayesian case). In this thesis, the problem of selecting the parameters (the so called hyper-parameters) for the regularization process, based on the set of measurements, is considered. The solution is provided through the evidence approximation method, where both the bias parameters and the hyper-parameters are estimated simultaneously. As a result, one obtains a robust algorithm that, thanks to the application of Occam’s razor, allows to find the good trade-off between model complexity and its fit to the data.

Finally, both methods are combined together, in order to provide a robust and accurate algorithm for the calibration of sensor networks using targets of opportunity.

The applicability of algorithms was also verified during field trials with good final outcome, confirming the expected performance.

Abstract [sv]

Sensornätverk finns  överallt omkring oss. Klassiska  exempel är meteorologiska stationer för väderprognoser samt seismometrar för att lokalisera jordbävningar och explosioner. Utvecklingen av små  och billiga  sensorer möjliggör  sensornät- verk i en helt ny omfattning jämfört med tidigare. En vanlig  vision är att sprida ut  tusentals små  enheter som  decentraliserat samverkar för  att lösa  komplexa problem. Ett specifikt exempel på en sådan  vision är att alla världens smarta tele- foner utbyter sensordata med varandra. I princip kan accelerometrarna användas för att upptäcka jordbävningar, mikrofonerna kan lokalisera bullerkällor och ex- plosioner, gps-mottagarna kan modellera atmosfäriska fenomen, barometrar och termometrar kan  matas  in i väderprognosmodeller, dopplereffekter i mottagna radiosignaler kan  användas för att följa alla jordens  fordon, etc. Algoritmer för detta finns redan idag.

En flaskhals i exemplet ovan  och rent allmänt i utrullning av sensornätverk är att sensorerna måste  vara  väl kalibrerade samt att deras  positioner måste  vara kända för att meningsfulla slutsatser ska kunna dras. Denna  avhandling behand- lar detta problem i detalj.  Idag krävs  tidsödande manuellt arbete  för att mäta  ut exakt  var sensorerna placeras samt att jämföra  deras  så kallade biasparametrar. Båda problemen kan lösas halvautomatiskt, t. ex. för ett mikrofonnätverk genom att flytta  runt en ljudkälla till kända positioner. Denna  avhandling beskriver en metod  som gör detta helautomatiskt genom  att utnyttja tillfälliga ljudkällor som råkar  passera förbi,  t. ex. en bil eller  fågel.  Metoden bygger  på att alla  kalibre- ringsparametrar skattas simultant med ljudkällans position.

De sensorer som  använts i detta arbete  har  mellan tre  och  tio  kalibreringspa- rametrar, så för nätverk med  hundratals eller  kanske tusentals sensorer så kan skattningarna lätt bli dåliga.  Avhandlingen beskriver en utvidgning av en grund- metod  som bygger på att de flesta parametrar oftast  är tillräckligt välkalibrerade, och detekterar automatiskt vilka parametrar som behöver kalibreras.

Metoderna är utvärderade på data  från  fältförsök med  akustiska sensornätverk med utmärkta resultat.

Place, publisher, year, edition, pages
Linköping: Linköping University Electronic Press, 2012. , 65 p.
Series
Linköping Studies in Science and Technology. Thesis, ISSN 0280-7971 ; 1611
National Category
Signal Processing
Identifiers
URN: urn:nbn:se:liu:diva-96809ISBN: 978-91-7519-540-7 (print)OAI: oai:DiVA.org:liu-96809DiVA: diva2:643476
Presentation
2013-09-18, Visionen, B-Huset, Campus Valla, Linköping University, Linköping, 15:15 (English)
Opponent
Supervisors
Funder
EU, FP7, Seventh Framework Programme, 238710
Available from: 2013-09-11 Created: 2013-08-27 Last updated: 2013-09-11Bibliographically approved
List of papers
1. Expectation Maximization Algorithm for Calibration of Ground Sensor Networks using a Road Constrained Particle Filter
Open this publication in new window or tab >>Expectation Maximization Algorithm for Calibration of Ground Sensor Networks using a Road Constrained Particle Filter
2012 (English)In: 15th International Conference on Information Fusion (FUSION), 2012, IEEE , 2012, 771-778 p.Conference paper, Published paper (Refereed)
Abstract [en]

Target tracking in ground sensor networks requires an accurate calibration of sensor positions and orientations, as well as sensor offsets and scale errors. We present a calibration algorithm based on the EM (expectation maximization) algorithm, where the particle filter is used for target tracking and a non-linear least squares estimator is used for estimation of the calibration parameters. The proposed algorithm is very simple to use in practice, since no ground truth of the target position and time synchronization are needed. In that way, opportunistic targets can also be used for calibration. For road-bound targets, a road-constrained particle filter is used to increase the performance. Tests on real data shows that a sensor position accuracy of a couple of meters is obtained from only one passing target.

Place, publisher, year, edition, pages
IEEE, 2012
National Category
Signal Processing
Identifiers
urn:nbn:se:liu:diva-96808 (URN)978-1-4673-0417-7 (ISBN)978-0-9824438-4-2 (ISBN)
Conference
15th International Conference on Information Fusion (FUSION 2012), 9-12 July 2012, Singapore
Funder
EU, FP7, Seventh Framework Programme, 238710
Available from: 2013-08-27 Created: 2013-08-27 Last updated: 2013-09-17Bibliographically approved
2. Simultaneous Tracking and Sparse Calibration in Ground Sensor Networks using Evidence Approximation
Open this publication in new window or tab >>Simultaneous Tracking and Sparse Calibration in Ground Sensor Networks using Evidence Approximation
2013 (English)Conference paper, Published paper (Refereed)
Abstract [en]

Calibration of ground sensor networks is a complex task in practice. To tackle the problem, we propose an approach based on simultaneous tracking of targets of opportunity and sparse estimation of the bias parameters. The evidence approximation method is used to get a sparse estimate of the bias parameters, and the method is here extended with a novel marginalization step where a state smoother is invoked. A simulation study shows that the non-zero bias parameters are detected and well estimated using only one target of opportunity passing by the network.

Place, publisher, year, edition, pages
IEEE conference proceedings, 2013
National Category
Engineering and Technology
Identifiers
urn:nbn:se:liu:diva-97376 (URN)10.1109/ICASSP.2013.6638230 (DOI)000329611503054 ()
Conference
The 38th International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Vancouver, Canada, May 26-31, 2013
Available from: 2013-09-11 Created: 2013-09-11 Last updated: 2014-02-20Bibliographically approved
3. Expectation Maximization Algorithm for Simultaneous Tracking and Sparse Calibration of Sensor Networks
Open this publication in new window or tab >>Expectation Maximization Algorithm for Simultaneous Tracking and Sparse Calibration of Sensor Networks
(English)Manuscript (preprint) (Other academic)
Abstract [en]

The performance of ground sensor networks (gsn) relies on accurate knowledge of sensor positions and sensor linearity parameters. Calibration of these should be done automatically after the deployment and later on regular intervals. Preferably, only targets of opportunity should be used to avoid manual intervention. We present a Bayesian solution that scales well with network size and efficiently utilizes a limited number of data to get a sparse estimate of a potentially huge number of calibration parameters. The approach is based on iteratively solving two different estimation problems using the expectation maximization (em) algorithm. First, the target trajectory and sensor parameters are estimated for a given prior on the parameters. Second, the prior is estimated using the evidence approximation (ea) principleto get a sparse solution. The algorithm successfully jointly estimates the sparsity structure, calibration parameters and target trajectory as demonstrated on both simulated and real data from a microphone network with a passing vehicle.

National Category
Engineering and Technology
Identifiers
urn:nbn:se:liu:diva-97377 (URN)
Available from: 2013-09-11 Created: 2013-09-11 Last updated: 2013-09-11Bibliographically approved

Open Access in DiVA

omslag(35 kB)19 downloads
File information
File name COVER01.pdfFile size 35 kBChecksum SHA-512
6821dd3cc56416e6c3bbb750489976066f0dac8bac3325b50ad31133733c6473ebfaea02cc6f32c3d66e9de86274145b3c98228aa765ca7cf4fc292a3cd3afb0
Type coverMimetype application/pdf

Authority records BETA

Syldatk, Marek

Search in DiVA

By author/editor
Syldatk, Marek
By organisation
Automatic ControlThe Institute of Technology
Signal Processing

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 345 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf