We were investigating methods to use DTNs, energy management, and programming languages to improve the state of the art in tracking small, mobile wildlife.

Faculty:  Mark D. Corner and Emery D. Berger.


Students: Jacob Sorber, Alexander Kostadinov, Matthew Garber, Matthew Brennan.

The motivation for our research comes, in part, from the efforts of conservation biologists to protect threatened turtles. The Wood Turtle (Clemmys insculpta) is found throughout the Northeast and Great Lakes regions and into Canada. They live primarily in and along streams; however, they are also terrestrial for about 4 months of the year. Wood Turtles are of particular interest since their numbers are rapidly declining due to habitat destruction and highway mortality. Unfortunately, conservation efforts have been hindered by a general lack of data due to current tracking methods. Researchers currently track turtles manually using radio telemetry and are limited to taking a single location fix every 2-3 days for each animal being studied. The turtles often travel up to 1 kilometer between fixes and practical concerns preclude the collection of location information at night. In order to more accurately understand how these turtles behave and use their habitat, new tracking methods are required to collect data at finer granularity.

Recent advances in sensing platforms make it possible to use more sophisticated sensors, such as a GPS receiver, to collect more frequent movement information. A sensor platform such as a Crossbow Mica2Dot, a GPS receiver, a flexible solar panel, and a small 250mAhr lithium polymer battery are within the acceptable weight and size requirements for studying these animals; however, energy management remains a significant concern. A typical GPS receiver will completely drain a 250mAhr battery in less than 2 hours, which necessitates carefully selecting an appropriate duty cycle based on the energy stored in the battery as well as the expectation of energy from its solar panel. Due to the high variance of solar radiation and the unknown mobility patterns of the individual turtles, it is necessary for us to dynamically adjust device behavior during operation. Our goal is to provide a simple networking, energy managment, and programming that meets these needs.

The data-flow oriented programming style used in the Flux language (another UMass project) has many similarities with the programming style used in many sensor systems. A Flux program is a directed acyclic graph that describes how data and control flow from event sources through computational tasks. While this program style was originally designed to support the programming needs of high-performance servers, network embedded sensor applications also typically consist of a sequence of tasks that are performed in response to an event, such as an expired timer or the arrival of a network message.

Along these lines, we are developing Eon, which extends Flux to include an application adaptation policy which describes an ordering of behavior adjustments, which include alternate control paths and adjustable timers. We are developing an Eon compiler and runtime system that implements this policy, by profiling the energy cost and frequency of individual execution paths and determining an appropriate operating state.

In August 2008, we deployed GPS-enabled tracking devices on 13 gopher tortoises in southern Mississippi, in collaboration with herpetologists Carl Qualls and Josh Ennen, at the University of Southern Mississippi.  Initial status/results from this effort can be seen here.



Hamed Soroush, Nilanjan Banerjee, Aruna Balasubramanian, Mark D. Corner, Brian Neil Levine, and Brian Lynn. In Proc. ACM Intl. Workshop on Hot Topics of Planet-Scale Mobility Measurements (HotPlanet), June 2009. PDF
Aruna Balasubramanian, Brian Neil Levine, and Arun Venkataramani. IEEE/ACM Transactions on Networking, 18(2):596--609, April 2010. PDF.
Nilanjan Banerjee, Mark D. Corner, and Brian Neil Levine. IEEE/ACM Transactions on Networking, 18(2):554--567, April 2010. PDF
Architecting Protocols to Enable Mobile Applications in Diverse Wireless Networks. Aruna Balasubramanian. PhD thesis, University of Massachusetts Amherst, Amherst, MA, February 2011.
System support for perpetual mobile tracking Ph.D. Thesis. Univ. of Massachusetts Amherst
Improved Network Consistency and Connection in Mobile and Sensor Systems Ph.D. Dissertation, University of Massachusetts, Amherst, September 2009 Winner of the 2009 UMass/Yahoo! Outstanding Dissertation Award!