Unit:
Κατεύθυνση Ηλεκτρονική και Ραδιοηλεκτρολογία (Ρ/Η, με πρόσθετη εξειδίκευση στις Τηλεπικοινωνίες και στην επεξεργασία και διοίκηση της Πληροφορίας)Library of the School of Science
Author:
Αλιμπέρτης Γεώργιος
Supervisors info:
Ανδρέας Πολύδωρος Καθηγητής ,Άρης Μουστάκας Επίκ. Καθηγητής, Γεώργιος Τόμπρας Καθηγητής
Original Title:
Μελέτη και επεξεργασία αλγορίθμων για τον προσδιορισμό της θέσης μιας πηγής σε ένα ασύρματο δίκτυο αισθητήρων
Translated title:
Single source localization in wireless sensor networks
Summary:
This thesis studies certain approaches to the localization problem in wireless
sensor networks using received-signal-strength measurements.In order to
circumvent the nonconvexity of the conventional maximum likelihood (ML)
estimator,the problem is firstly reformulated under the equivalent exponential
transformation of the conventional path loss measurement model and the
unscented transformation (UT).It is approximately approached by the maximum
likelihood (ML) parameter estimation. This estimation is referred to as the
weighted least squares (WLS) approach.Secondly,the use of convex estimator is
being examined for the same RSS-based localization problem. In this case, a
nonconvex estimator approximates the ML estimator but has no logarithm in the
residual.The semidefinite relaxation technique is then applied to the nonconvex
estimator and the convex estimator is developed.Finally the use of second order
cone programming is being explored as an approach to relax the estimator by
applying efficient convex relaxations.All approaches have been critically
compared in terms of geometric layout, standard deviation and number of sensors
in order to determine their effectiveness.
Keywords:
Semidefinite Programming, Second order cone programming, Localization, Weighted Least Squares, Maximum Likelihood
