Start Date: January 1, 2009
Segou Olga
The availability of a Location Estimation Service is often regarded as a significant component in a multitude of Pervasive Computing systems implementations. The collection and transmission of the location information of a moving device or person, typically requires the existence of dedicated, integrated infrastructure, which is often pre-deployed in the area wherein location-based services are desired. The present dissertation aims to provide an in-depth analysis of the capabilities of Location Estimation systems and their application in the context of indoor spaces. The use of satellite technology (such as the Global Positioning System – GPS) in an indoor environment is usually not feasible, typically leading to the design and implementation of specialized equipment. This document focuses on location estimation systems that utilize the existing wireless networking infrastructure to provide the location information, while other cases such as optical or magnetic systems, are considered to be out of the scope of this work.The definition of the concepts of space, location, distance and the coordinate system is of fundamental importance towards the definition of a location estimation system. The design and implementation of such a system proves to be a challenge due to the large number of factors affecting their operation. The unique characteristics of the wireless channel in indoor spaces (where there is a multitude of obstructions, moving persons etc.) may affect the quality of the received signal and degrade the location information accuracy. Other important factors that need to be taken into account include the technical characteristics of the equipment (such as the wireless standard and specification, the available metrics etc.), the network and software architecture (i.e. the localization algorithm, centralized versus mobile-based processing etc.), the system’s deployment topology etc. The multitude of factors that may affect a localization system’s architecture and accuracy, often makes comparison between systems difficult or irrelevant.
A software system focusing on the simulation of location estimation processes was implemented within this work. The Locus software platform is based on the accurate simulation of the indoor channel and a variety of localization algorithms. It integrates a large number of algorithms across many fields and disciplines such as Wireless Propagation, Computational Geometry, Signal Processing, Location Estimation, Pseudorandom Number Generation, Exploratory Data Analysis and Probabilistic Modeling, etc. thus comprising a unique analytical framework. This software permits the accurate simulation of various localization methods in different yet controlled propagation conditions, leading to the creation of a results database of significant size. The large stochastic size of this database allows the comparison between different methods and drawing of generalized conclusions, especially since the time-consuming nature of the process of setting up a system and collecting empirical measurements would be prohibitive towards this goal. Furthermore, the comparison between different real deployments is often irrelevant, as the propagation conditions cannot be controlled. The use of empirical measurements within the simulation environment is also possible, thus taking advantage of the existing analytical framework.
The proposed software platform allows fast experimentation with localization methods, the study of underlying error distributions etc. while it promoted the development of novel statistical and analytical simulation methods. Relevant results are presented in the form of Use Cases, also showcasing significant conclusions pertaining to the expected boundaries of localization system performance, the modeling of localization error, the efficient placement of wireless nodes to form a viable localization topology etc. This work also enables the extensive simulation of the operational boundaries of a Received Signal Strength localization system and the extraction of the boundary models. This work allows the creation of a scale on which systems may be ranked according to their performance, enabling, to an extent, comparisons among different systems that would otherwise be prohibitive.
Empirical results from the deployment of three real-time localization systems based on the ZigBee, RFID and Chirp Spread Spectrum specifications are also herein presented. The implementation of the protocol stack of the ZigBee based system is thoroughly discussed, as it details the basic software components and sheds light on the common pitfalls that should be avoided in order to successfully reach the end-result of a fully functioning system. A large number of empirical measurements is collected in a variety of deployments for all three systems, in order to estimate their performance capabilities. Results are shown to align with previous conclusions drawn by extensive simulation and include the implementation of a novel topology-based smoothing method. This method is also shown to significantly improve the performance of all trilateration-based systems.
Keywords: Location Estimation Services, Wireless Propagation, Localization Algorithm, Probabilistic Modeling, Computational Geometry, Received Signal Strength, Time of Arrival, Pseudorandom Number Generation, Locus software.