Dissertação
Utilizando a infraestrutura wi-fi disponível para prover a localização de smartphones em ambientes fechados
Autor
Silva, Bolívar Menezes da
Institución
Resumen
The popularization of mobile devices with increasingly sensors and embedded resources
has boosted several types of research in the area of context-awareness computing. Among the
most relevant contextual information is the location. In outdoor environments, GPS technology
is already widespread. However, in general, people tend to spend most of their time indoors,
such as university buildings, hospitals, shopping malls, supermarkets, airports or even in their
homes. Due to interference caused by various obstacles, the accuracy of GPS location can often
be compromised. In order to overcome the problem of location indoors, several approaches,
mainly using radiofrequency technologies, have been proposed. So far, there is no widely accepted
solution that solves the problem of location indoors. In this sense, the present work uses
an opportunistic approach, which takes advantage of the Wi-Fi infrastructure available in the
environment, to provide the location of mobile stations. Based on this objective, the WALDO
architecture was developed, linking some characteristics of different approaches developed in
recent years, taking into account the techniques that present the best results at each stage, together
with a zone-based approach and rankings. This approach seeks to make RSS reads that
have noises ignored during the localization phase. After performing tests in different scenarios,
using the WALDO architecture, the presented results were satisfactory. Although this is an
opportunistic approach, the tests present most of the estimates with an average error between 2
and 4 meters, depending on the dataset used. In the first set of data, corresponding to an area of
66 m2 (computer lab), 80.24% of the tests presented location estimates between 0 and 4 meters
from the true location (zero being the correct position of smartphone at the time of the test).
On the other hand, in the tests performed with the second set of data, in an area of 560 m2
(composed of some rooms), the results between 0 and 5 meters corresponded to 75.5% of the
tests.