An RFID-Based Object Localization Framework and System

Locating objects is a key requirement in several of the emerging
computing paradigms. The problem of locating objects has been
extensively studied from a variety of technological and technique
-oriented perspectives. Recently, Radio Frequency Identification
(RFID), a wireless automated identification technology, has come
forth as a viable platform for locating objects, particularly in indoor
environments.

While rapid advances in RFID-based object localization are evident,
current approaches lack adaptability, reliability, and scalability.
This thesis addresses these issues and presents an RFID-based
object localization framework and system to help locate stationary
and mobile objects with high accuracy. Our RFID-based object
localization framework and system is resilient in select environmental
conditions, accommodates numerous use-case scenarios, and is tag
orientation and vendor hardware –agnostic. We demonstrate that
radio signal strength, a technique used in our location system and
traditionally considered unreliable, can be used as a reliable metric
for locating objects in selective cases. Additionally, we show that
tag sensitivity caused by manufacturing variation influences object
localization performance and we present tag selection and binning
techniques. This ensure range and cost -optimized uniformly
sensitive tags, leading to a reliable and high-performance object
localization.

We further improve the object localization characteristics of our
system by matching tags to readers and demonstrating that reference
tags could be made optional without significant loss in performance.
Rigorous experimental evidence suggests that our RFID-based object
location system can simultaneously locate several stationary and mobile
objects in realistic noisy indoor environments with localization accuracy
in the range of 0.15-0.84 meters. We have also developed several
visualization applications focusing on a variety of computing platforms
to help visualize the targeted object’s location.