Articles via Databases
Articles via Journals
Online Catalog
Research & Information Literacy
Interlibrary loan
Theses & Dissertations
About / Contact Us
Littman Architecture Library
This site will be removed in January 2019, please change your bookmarks.
This page will redirect to in 5 seconds

The New Jersey Institute of Technology's
Electronic Theses & Dissertations Project

Title: A bayesian approach to wireless location problems
Author: Zorych, Ivan
View Online: njit-etd2005-143
(xiii, 94 pages ~ 7.4 MB pdf)
Department: Department of Mathematical Sciences
Degree: Doctor of Philosophy
Program: Mathematical Sciences
Document Type: Dissertation
Advisory Committee: Bhattacharjee, Manish Chandra (Committee co-chair)
Madigan, David (Committee co-chair)
Dhar, Sunil Kumar (Committee member)
Michalopoulou, Eliza Zoi-Heleni (Committee member)
Spencer, Thomas (Committee member)
Date: 2005-08
Keywords: Bayesian
Graphical model
Semiparametric model
Wireless networks
Availability: Unrestricted

Several approaches for indoor location estimation in wireless networks are proposed. We explore non-hierarchical and hierarchical Bayesian graphical models that use prior knowledge about physics of signal propagation, as well as different modifications of Bayesian bivariate spline models. The hierarchical Bayesian model that incorporates information about locations of access points achieves accuracy that is similar to other published models and algorithms, but by using prior knowledge, this model drastically reduces the requirement for training data when compared to existing approaches. Proposed Bayesian bivariate spline models for location surpass predictive accuracy of existing methods. It has been shown that different versions of this model, in combination with sampling/importance resampling and particle filter algorithms, are suitable for the real-time estimation and tracking of moving objects. It has been demonstrated that "plug-in" versions of the bivariate Bayesian spline model perform as good as the full Bayesian version. A combination of two Bayesian models to reduce the maximum predictive error is proposed. Models presented in this work utilize MCMC simulations in directed acyclic graphs (DAGs) to solve ill-posed problem of location estimation in wireless networks using only received signal strengths. Similar approaches may be applied to other ill-posed problems.

If you have any questions please contact the ETD Team,

ETD Information
Digital Commons @ NJIT
Theses and DIssertations
ETD Policies & Procedures
ETD home

Request a Scan

NJIT's ETD project was given an ACRL/NJ Technology Innovation Honorable Mention Award in spring 2003