Ph.D
Group :
Clock synchronization and localization for wireless sensor network
Starts on 01/10/2015
Advisor : LAMBERT, Alain
Funding : Bourse pour étudiant étranger
Affiliation : Université Paris-Saclay
Laboratory : LRI
Defended on 12/11/2018, committee :
- M. Alain LAMBERT Laboratoire de Recherche en Informatique Université Paris-Sud
- M. Hichem SNOUSSI University of Technology of Troyes
- M. Marcus SHAWKY Université de Technologie de Compiègne
- M. Thomas NOWAK Laboratoire de Recherche en Informatique Université Paris-Sud
- M. Jaulin LUC Ecole d'ingénieurs et centre de recherche
- M. Michel KIEFFER Laboratoire des signaux et systèmes, Centrale Supelec
- Mme Farah CHEHADE Université de Technologie de Troyes Département Recherche Opérationnelle, Statistiques Appliquées, Simulation
Research activities :
Abstract :
Wireless sensor networks (WSNs) play an important role in applications such as environmental monitoring, source tracking, health care, etc. In WSN, sensors have the ability to perform data acquisition, distributed computing and information fusion. To perform such tasks, information about their localization in some reference frame, and the availability of well-synchronized clocks is very important.
WSNs have been widely studied in the past years, and the scientific literature reports many outcomes that make them applicable to various fields such as environmental monitoring, health care, and the internet of things. For some others, research still needs to find solutions to some of the challenges posed by the limitations of sensors, such as battery limitation, dynamicity, and low computing clock rate. With the aim of contributing to the research on WSN, this thesis proposes new algorithms for both clock synchronization and localization. The proposed synchronization algorithm is able to tolerate a low connectivity dynamic WSN and results in a much smaller clock skew than its previous work.
Localization of sensors has been widely studied. Nevertheless, the characterization of estimation uncertainty is usually overlooked. Asymptotic techniques are usually employed, assuming that the noise is Gaussian and that many measurements are available. These hypotheses are seldom satisfied. This thesis applies the Leave-out Sign-dominant Correlated Regions (LSCR) algorithm to localization problem from received signal strength measurements. With LSCR, one evaluates the accurate estimates of confidence regions with prescribed confidence levels. From this confidence region, an estimate of the sensor location may also be obtained. In this thesis, several localization approaches are implemented and compared. The results show under mild assumptions, LSCR obtains competitive localization performance compared to other methods.