Electrical engineer Margaret Martonosi (center) and her students Hidekazu Oki (fore-ground) and Philo Juang (background) have begun field tests of wireless devices that will be hung from the necks of wild zebras in Kenya and used to monitor the animals' behavior. The project, called ZebraNet, is a collaboration between faculty members in electrical engineering and Dan Rubenstein of ecology and evolutionary biology.Internet, ethernet ... ZebraNet?
That is one of the many queries that have been flying back and forth between the E-Quad and Eno Hall as part of an interdisciplinary project called ZebraNet. Martonosi and engineering colleagues Vince Poor and Li-Shiuan Peh are collaborating with Rubenstein, chair of ecology and evolutionary biology, to develop a network of wireless devices that will hang from the necks of zebras and monitor their every move.
Funded by a three-year grant from the National Science Foundation, the researchers are building a prototype of the network and hope to test it on zebras in Kenya next summer.
The purpose is to study the animals -- ultimately not just zebras, but lions, hyenas, entire systems of predators and prey -- and analyze their activities with a level of detail that was previously impossible.
"We think this technology is going to open up the study of not only animal movements, but also information processing among animals -- the way the animal mind works," said Rubenstein, who specializes in the study of zebras and other equine animals.
The fruits of such knowledge go beyond the basic desire to unlock the secrets of the animal kingdom. As human development in Africa -- and around the world -- encroaches on animal habitats, ecologists want to know why animals do what they do and how flexible they are in the choices they make. Understanding such questions could mean the difference between development that coexists with animals and development that destroys them, said Rubenstein.
The payoffs are expected to be large for the engineers as well. Academic, commercial and military engineers are increasingly interested in "remote sensing networks" -- autonomous, but communicating devices that could perform unmanned military surveillance, monitor underground water pollution or track the weather, said Martonosi, a specialist in low-power-consumption computing and networks.
"There are many potential applications," she said, "and they raise interesting technical challenges regarding how to deploy sensing computers that use little energy, that run autonomously in harsh environments and that communicate with each other to interpret data or forward it to base stations."
ZebraNet was born a year ago when Rubenstein heard of a research project Martonosi had advised in which undergraduates created an electronically guided campus tour -- a digital assistant, like a Palm, equipped with a global positioning system, would read out information about the campus depending on where the person walked. Some of the same ideas, Rubenstein realized, could be employed with zebras instead of tour-takers.
Martonosi and her students, including graduate students Philo Juang and Hidekazu Oki, are now developing prototype devices that will be attached to collars and put on zebras. The devices will monitor the animals around the clock, using the global positioning system to record their location and other gadgetry to note whether the animals are eating, moving or resting.
The collars will periodically broadcast a signal to search for other collars in the area. Once two collars establish a connection -- zebra to zebra -- they will trade data. With each data swap, a collar will collect information about not only the other animal, but all the others that have already been in contact with that animal. In this way, the collars will accumulate redundant databases, with each animal carrying information about many others.
Every few weeks, biologists could find some of the collared zebras and download detailed data about the entire herd in a few quick transactions. Once a collar reports its data to a researcher, it would earmark it for deletion and propagate that information back through the herd.
Because collars will transmit data only when zebras are close to each other, they will require far less energy than if they had to broadcast to a distant base station or to a satellite. The redundancy also improves the reliability of the system, said Martonosi.
If it works, the system will cause a sea change in the way field ecologists work, said Rubenstein. Currently biologists can study only animals that someone has personally tracked down and observed. "In the past, you put a radio collar on an animal and tried to go find it -- it can take you an hour or two to get a fix on it -- then you drive to that location, and then, if you don't disturb the animal, you can watch it," said Rubenstein. With ZebraNet, he said, "our collars will allow us to go to the animals when we want and also to get data when we're not there."
|Dan Rubenstein, chair of ecology and evolutionary biology, hopes to use the ZebraNet technology to monitor animals with a far greater level of detail than previously possible.
Such data is crucial to many unresolved questions about ecology and animal behavior. One of the main subjects Rubenstein hopes to address is the relation between local movements and long-distance migrations. "We still don't know what triggers animals to leave one area and go to another," he said. "We don't really understand why they choose to go to particular places. We don't understand why they stop and then what triggers them to return to their starting point. We don't know how flexible they are in making these decisions."
Once other animals are included in the network, the researchers also will learn about predator-prey dynamics. "Are the prey pulling the predators around the landscape or are the predators pushing the prey?" Rubenstein asked.
With ZebraNet, the opportunities for answering such questions are so great that Rubenstein has organized a graduate seminar this fall to "intellectually get ahead of the curve" and define an agenda of research projects that take advantage of new technology. As a measure of the excitement it is generating, the seminar, called "Animal Movements and Animal Minds," has attracted participation from seven faculty members -- Rubenstein, Henry Horn, James Gould, Simon Levin, Stephen Pacala and Martin Wikelski from the ecology and evolutionary biology department, as well as Jonathan Cohen from the psychology department.
Among the subjects that once seemed off-limits is how animals process information, or, in a crude way, how they think. "We will be able to get into their cognitive processing: How does an animal gain information about a landscape? Does it have to know everything by itself? Can it use knowledge in others to get over gaps in its own knowledge?" said Rubenstein.
These questions, in turn, translate into a rich set of engineering problems. For Martonosi, one challenge is to take advantage of what biologists already know about zebra behavior so that the collars collect and distribute their data efficiently. In one scheme, the zebras would swap all their data each time they contact one another; in another, a collar would save energy by swapping only if it met a zebra -- or other animal -- with a track record of successfully returning the data to a base station.
Martonosi noted that it might turn out that lions are better carriers of data about zebras than the zebras themselves, because they move around a lot and zebras "tend to hang around in the same place and eat grass." (For engineers who worry about maintaining privacy in wireless networks, Martonosi acknowledged, there is a certain irony in making lions carry a database with the locations and habits of their prey.)
Efficient delivery of information is critical because radio transmissions consume a lot of battery power, and batteries contribute most of the weight and bulk of the device. The collars will be fitted with a solar panel to recharge the battery during the day.
In addition, the entire package has to work under extremely abusive conditions. "We can assume that, over the course of a year, just about everything will happen," said Martonosi. "Once you have the mindset that it's got to run for a year without babying it, then you really start to think about things differently. You can't just reboot if it quits working."
Such challenges epitomize the value of interdisciplinary research, said Martonosi. "It's a blast," she said. "You learn a lot about other fields that you would never have contact with." One example was the "Do zebras swim?" question. The answer: It doesn't matter if you call it swimming, "they go in the water enough that we need to worry about waterproofing," said Martonosi.
Interdisciplinary problems also can be more interesting to students than conventional engineering problems, said Martonosi, who has pioneered methods for reducing the energy consumption and heat buildup in computer chips.
"You can save 10 watts in a microprocessor and it doesn't have the same effect on students as putting 30 collars on zebras," she said.
"It's a lot more hands-on," agreed Juang, a third-year graduate student. "You can actually hold the product you are designing. That is really energizing."
The researchers have begun field-testing the devices and plan to try them on horses in the United States this fall before bringing the first collars to Africa in the summer.
The Bulletin is published weekly during the academic year, except during University breaks and exam weeks, by the Office of Communications. Second class postage paid at Princeton. Postmaster: Send address changes to Princeton Weekly Bulletin, Office of Communications, Princeton University, 22 Chambers St., Suite 201, Princeton, NJ 08542. Permission is given to adapt, reprint or excerpt material from the Bulletin for use in other media.
Subscriptions. The Bulletin is distributed free to faculty, staff and students. Others may subscribe to the Bulletin for $28 for the academic year (half price for current Princeton parents and people over 65). Send a check to Office of Communications, Princeton University, 22 Chambers St., Suite 201, Princeton, NJ 08542.
Deadline. In general, the copy deadline for each issue is the Friday 10 days in advance of the Monday cover date. The deadline for the Bulletin that covers Nov. 25-Dec. 8 is Friday, Nov. 15. A complete publication schedule is available at deadlines or by calling (609) 258-3601.
Editor: Ruth Stevens
Calendar editor: Carolyn Geller
Staff writers: Jennifer Greenstein Altmann, Steven Schultz
Contributing writers: Karin Dienst, Evelyn Tu
Photographer: Denise Applewhite
Design: Mahlon Lovett, Laurel Masten Cantor, Margaret Westergaard
Web edition: Mahlon Lovett