Hardware Reference
In-Depth Information
Network Location and Physical Location
Locating things is one of the most common tasks people want to achieve with sensor
systems. Once you understand the wide range of things that sensors can detect, it's
natural to get excited about the freedom this affords. All of a sudden, you don't have to
be confined to a chair to interact with computers. You're free to dance, run, jump—and
it's still possible for a computer to read your action and respond in some way.
The downside of this freedom is the perception that in a
networked world, you can be located anywhere. Ubiqui-
tous surveillance cameras and systems like Wireless E911
(which locates mobile phones on a network), make it seem
as though anyone or anything can be located anywhere
and at any time—whether you want to be located or not.
The reality of location technologies lies somewhere in
between these extremes.
Getting a good location starts with cultural and behavioral
cues. If you want to know where you are, ask another
person near you. In an instant, she's going to sum up all
kinds of things—your appearance, your behavior, the
setting you're both in, the things you're carrying, and
more—in order to give you a reasonably accurate and
contextually relevant answer. No amount of technology
can do that, because the connection between where
you are and why you want to know is seldom explicit
in the question. As a result, the best thing you can do
when you're designing a locating system is to harness
the connection-making talents of the person who will be
using that system. Providing him with cues as to where
to position himself when he should take action, and what
actions he can take, helps eliminate the need for a lot of
technology. Asking him to tell your system where things
are, or to position them so that the system can easily find
them, makes for a more effective system.
Locating things on a network is different than locating
things in physical space. As soon as a device is connected
to a network, you can get a general idea of its network
location using a variety of means—from address lookup to
measuring its signal strength—but that doesn't mean that
you know its physical location. You just know its relation-
ship to other nodes of the network. You might know that
a cell phone is closest to a given cell transmitter tower, or
that a computer is connected to a particular WiFi access
point. You can use that information along with other data
to form a picture of the person using the device. If you
know that the cell transmitter tower is less than a mile
from you, you'd know that the person with the cell phone
is going to reach you soon, and you can act appropriately
in response. For many network applications, you don't
need to know physical location as much as you need to
know relationship to other nodes in the network.
For example, imagine you're making an interactive space
that responds to the movements of its viewers. This is
popular among interactive artists, who often begin by
imagining a “body-as-cursor” project, in which the viewer
is imagined as a body moving around in the space of
the gallery. Some sort of tracking system is needed to
determine his position and report it back in two dimen-
sions, like the position of a cursor on a computer screen.
•
Step 1: Ask a Person
People are really good at locating things. At the physical
level, we have a variety of senses to throw at the problem
as well as a brain that's wonderful at matching patterns of
shapes and determining distances from different sensory
clues. At the behavioral level, we've got thousands of
patterns that make it easier to determine why you might
be looking for something. Computer systems don't have
these same advantages, so when you're designing an
interactive system to locate things or people, the best
tool you have to work with—and the first one you should
consider—is the person for whom you're making your
system.
What's missing here is the reason why the viewer might be
moving in the first place. If you start by defining what the
viewer's doing, and give him cues as to what you expect
him to do at each step, you can narrow down the space
in which you need to track him. Perhaps you only need to
know when he's approaching one of several sculptures in
the space so that you can trigger the sculpture to move
in response. If you think of the sculptures as nodes in
a network, the task gets easier. Instead of tracking the
viewer in an undefined two-dimensional space, now all
you have to do is determine his proximity to one of several
