TopBot: Automated Network Topology Detection
WithaMobileRobot
Paul Blaer and Peter K. Allen
Department of Computer Science, Columbia University, New York, NY 10027
f
pblaer,allen
g
@cs.columbia.edu
Abstract— We have demonstrated that a properly-
equipped mobile robot can easily construct a detailed map
of the wireless coverage of an urban environment. The Au-
tonomous Vehicle for Exploration and Navigation of Urban
Environments (AVENUE) mobile robot was successfully used
to generate such maps in both manual and autonomous
modes of operation. The resulting database contained a
wealth of information for many different positions in the
region, with a list of all access points viewable from each
location together with a quality measure (the signal-to-noise
ratio) of every detected signal. At a later time, the AVENUE
system effectively used the data in this map to determine the
approximate position of the robot as it traveled through the
urban area.
I. INTRODUCTION
In the process of exploring urban environments, a mo-
bile robot passes through areas rich in wireless networks.
Signals from the networks’ access points can be collected
and analyzed by the robot, with the acquired information
incorporated into the robot’s navigational system. During
an initial exploratory phase, the robot can construct a
detailed signal map of the geographical region. At a later
time, the robot can use this map to assist in localization
and navigation. The map can also identify deficiencies in
the signal coverage of the area and help in the design of
a better network infrastructure. The Autonomous Vehicle
for Exploration and Navigation of Urban Environments,
AVENUE [6], has been adapted to perform these wireless
mapping tasks.
AVENUE is a mobile robot (see Fig. 1) whose overall
goal is to automate the site modeling process which
includes building geometrically accurate and photomet-
rically correct models of complex outdoor urban envi-
ronments. In particular, the AVENUE system can au-
tonomously model a targeted building. The system plans
a path to a desired viewpoint, navigates the mobile robot
to that viewpoint, acquires images and three-dimensional
range scans of the building, and then plans for the next
viewpoint. The system then fuses all of the collected data
into an accurate, texture-mapped, three-dimensional model
of the targeted building.
In this paper, we describe how the AVENUE system has
been extended so that it can autonomouslymap an outdoor
This work was supported in part by NSF grant ANI-00-99184.
region by utilizing signals from a wireless network. With
the addition of new software modules to monitor and an-
alyze wireless transmissions used in conjunction with the
existing path planning and navigation software, AVENUE
has been able to map the coverage of the numerous access
points installed on the Columbia University campus. It has
then been able to use this map for subsequent localiza-
tion and navigation purposes, which is especially useful
at times when GPS and odometry are not available or
reliable.
Our paper is organized as follows. In the next section,
we indicate previous work that is related to our project.
We then describe in section III our equipment and the
environment in which we worked. In section IV, we
detail the process of building the wireless-signal maps of
the northern part of Columbia’s campus and we present
a representative map of the locations at which wireless
signals were measured. Some statistics associated with the
viewable access points are also given. We then present
and discuss in section V the implementation of one of the
uses of the signal map, localization of the mobile robot. In
the concluding remarks of section VI, we summarize our
results and discuss additional possible uses of the signal
map for the AVENUE and other projects.
II. RELATED WORK
The work presented in this paper is part of the AVENUE
project for automated modeling of an outdoor urban
environment by a mobile robot ([1], [6]).
The use of existing 802.11b wireless network signals
as a means of locating a user was originally presented in
Microsoft Research’s RADAR project [2]. The Microsoft
group collected the signal data manually in an indoor
environment and then used this information for estimating
the position of a user at a later time. Other groups have
also made use of manually-obtained 802.11b signals for
indoor localization [8]. We have extended the work of
these groups by having our mobile robot autonomously
construct the database, while covering a much larger
outdoor urban environment.
There have also been a number of systems [9] based
on the characteristics of cellular signals and designed for
geolocating cellular telephone users in outdoor environ-
ments. In addition, there have been attempts to use RF
