 |
Sensing nature’s ways
Tiny sensors keep a
watchful eye on remote habitats
By David Pescovitz
| 
In late May, about 5,000 Leach’s
storm petrels descend in the dark of night on Great Duck
Island, one of two popular Northeast breeding grounds for
the elusive bird. Researchers hope to learn about the island’s
micro-habitat to determine whether it is unique and in need
of protection.
JOE POLASTRE PHOTO
|
Protecting the delicate ecosystems on our planet is impossible
if we don’t understand them. From the microclimates in forest
canopies to the mysterious breeding practices of seabirds, field
ecologists yearn for a way to observe without interfering. To
that end, a team of engineers and scientists from Maine and Berkeley
are collaborating to develop a powerful and unobtrusive sensing
system that biologists can leave behind in remote environments
as a window onto nature’s wonders.
Late last spring, Berkeley computer science graduate students
Joe Polastre and Rob Szewczyk brought their research on the system
out of the laboratory and into the real world. As summer interns
at the Intel Research Berkeley Laboratory, the students are part
of a dedicated team of researchers that was dispatched to Great
Duck Island, the 237-acre island off the coast of Maine, where
5,000 Leach’s storm petrels come to nest. During their two-week
field trip, the researchers deployed a wireless network of small
sensors that will monitor the shy seabird’s habitat long
after the air grows cold with the fall and the last human has
returned to the mainland.
The trip begins in earnest on a dock at the College of the Atlantic
in Bar Harbor. There, the group loads a 35-foot motorboat with
several weeks of supplies and begins the journey to the island.
One hundred yards from the island, the gear is transferred to
small rowboats. All the electronic equipment is packed tightly
in custom-made Tupperware-like containers to protect it from the
waves that bash the small boat and its passengers. From shore,
it’s less than half a mile to the old lightkeeper’s
cottage — mission control and living quarters for this field
study.
The petrels, the locale’s natural summer denizens, have
a much easier time getting to Great Duck Island. Indeed, it’s
these ground-nesting seabirds that Polastre, Szewczyk, and their
colleagues from UC Berkeley, Intel Research, and the College of
the Atlantic have traveled all this way to see. The goal of the
trip, the first of two last summer, was the deployment of a new
habitat monitoring system of 100 sensors to collect raw data about
what may be one of the largest petrel breeding colonies in the
eastern United States.
"The data can stream continuously from the island onto the
Internet so it can be analyzed by biologists anywhere in real
time," says David Culler, Berkeley computer sciences professor
and director of the Intel Research Berkeley Laboratory, who oversees
this project.
The tiny black seabird — so small it fits comfortably in
the palm of a hand — is a challenge to study because it
lives most of its life on the water, save for its breeding period
from the end of May through October. And even then the petrels,
an easy mark for predatory sea gulls, emerge from their burrows
only in pitch dark. In the summer months in Maine, that’s
less than two hours a night. John Anderson, College of the Atlantic
conservation biologist and Berkeley alumnus who spearheaded the
collaborative Great Duck Island effort, hopes the data the researchers
gather will help scientists identify the petrel’s attraction
to Great Duck Island — a site it prefers over thousands
of other local islands.
|
|
Unobtrusive
sensor motes are placed into the wall of a petrel burrow,
where they monitor the site and transmit data to a central
base station for distribution via the Internet.
PEG SKORPINSKI PHOTO
|
"There’s nothing else like this sensor network available
for conservation biologists, nothing that can provide good data
in such dense numbers," says Anderson. "What’s
really exciting here is that we can get a feel for what happens
on the island when humans aren’t there. This sensor network
will have a profound effect on how we do field ecology."
The petrel project began last year when Polastre, Szewczyk, and
Intel Research scientist and Berkeley alumnus Alan Mainwaring
collaborated with Anderson’s team to deploy a network of
32 sensor motes on the island. The system, Polastre explains,
had to be built from scratch — from identifying the desired
sensing capabilities for the motes, to designing the network architecture,
to installing the photovoltaic system needed to power the laptop
computer in the lighthouse. It was an iterative process that spanned
several scientific and engineering disciplines, Culler explains.
"We were deploying a new technology into a space that we'd
never dealt with before," he says. Previously, keeping an
eye on the petrels' private matters involved either reaching inside
the burrows or installing portable video systems for remote feeds.
These methods are risky, as too much human disturbance can cause
the birds to abandon their nests, chicks, and eggs.
That’s where the sensors can improve matters. The motes
were developed in a partnership between the UC Berkeley-based
Center for Information Technology Research in the Interest of
Society (CITRIS) and the Intel Research Berkeley Laboratory. Inexpensive
to manufacture and easy to deploy, the tiny devices boast myriad
applications, from energy monitoring in offices to diagnosing
a building’s structural stability after an earthquake.
Each of the environmental monitoring motes designed especially
for the Great Duck Island research is just under an inch wide
and just as tall, substantially smaller than the batteries that
power it. A hard plastic cylinder protects the batteries and delicate
microprocessor from the elements. Every five minutes, onboard
humidity and temperature sensors take a data snapshot of the surroundings.
Meanwhile, an infrared sensor scans for body heat to determine
if a petrel is inside the burrow. Above ground, strategically
placed weather station motes keep tabs on atmospheric pressure,
sunlight, and the other general environmental conditions on the
island.
|
Digitizing redwoods in the UC Berkeley Botanical Gardens
requires an agile student tree climber who can slip through
the redwood branches to wrap as many as 50 sensors along
the tree’s lanky trunk.
PEG SKORPINSKI PHOTO
|
Thanks to Culler’s TinyOS operating system, the motes self-organize
into an ad hoc wireless network and pass their data from one to
another, bucket-brigade style, until the information reaches a
gateway sensor above ground. Eventually, all the data makes its
way to a laptop computer tucked inside the lighthouse where it’s
relayed to a Web site via satellite.
"It’s ironic to be somewhere that has amazing Internet
access, but no running water," Szewczyk says. According to
the researchers, the data gathered during the preliminary deployment
was not crisp enough for biological study. Still, mining the massive
database containing more than one million readings from the initial
deployment revealed a gold mine of information about the network’s
performance in its first real-world test.
"It’s not just a matter of the motes behaving when
they’re supposed to," Culler says. "You also have
to think about things like what happens when the birds decide
to chew on them for a while."
Even as the Great Duck Island data pours in, Culler, Szewczyk,
and Polastre are embarking on another deployment of their environmental
monitoring system. Last summer in collaboration with Berkeley
Professor Todd Dawson of integrative biology, the researchers
mounted their motes high in the redwood trees in the UC Botanical
Garden’s Mather Redwood Grove.
Currently, Polastre explains, biologists use a winch to raise
large trolleys of sensors up and down a tree to acquire a sampling
of environmental data at different heights in the canopy. The
aim is to better understand the micro-climates that the redwoods
create as a consequence of their enormity. The beauty of the motes
— outfitted with sensors to detect weather conditions and
photosynthetic activity — is that hundreds of them can be
deployed simultaneously in a cross-section of the canopy to provide
a three-dimensional picture of the ecosystem across space and
time.
"The idea is to get a sense of what’s really going
on in the grove in order to better understand the plants' physiology,"
Polastre says. Once the technology is proven, the team will install
some 200 sensors in Big Basin Redwoods State Park and the Russian
River area where Dawson conducts most of his research.
"This kind of environmental monitoring technology is really
a new microscope," Culler says. "It gives scientists
the ability to perceive what they’ve never perceived before."
David Pescovitz is a contributing writer to
Wired and the creator of Lab
Notes, the College's online research digest. His work
has appeared in Scientific American, New Scientist, the
New York Times, and Salon.
|
