MITRE's Contributions to the DARPA NEST Research Program

By Kenneth W. Parker

Several years ago, the Defense Advanced Research Projects Agency (DARPA) expanded the idea of using large scale deployments of proximal sensors. MITRE has been working hand in hand with them ever since to perfect the idea.

From the beginning, the agency has prized MITRE's ability to identify technology gaps that undermine the practicality of DARPA's research and to find a way to fill these gaps. When DARPA's Network Embedded Systems Technology (NEST) program hit a snag with one of its projects, the agency invited MITRE to help unknot it.

NEST's Extreme Scaling Demonstration was a perimeter protection application designed to cover multiple square kilometers. At the onset of the demonstration, questions arose over the appropriate sensor density. MITRE's analysis and influence convinced DARPA that the application required a lower node density than DARPA had initially conceived. However, because the existing sensor ranges were short, simply deploying devices at a lower density would have created huge gaps in coverage. We estimated that ranges up to eight times greater than what had been demonstrated would be required to support reasonable military applications.

MITRE had independently developed an algorithm for vehicle classification based on acoustic signatures using the same commercial-off-the-shelf hardware that was in use by the NEST program. As a result of this work, we believed that acoustic sensing could provide the necessary solution to the range problem. However, developing new acoustic detection algorithms required significant expertise in fixed-point signal processing techniques. MITRE stepped in and formed a development team that delivered an initial capability within six weeks of authorization by the program manager. We then spent six months refining the acoustic sensing capability.

As part of this work, we discovered that a significant source of false alarms in most settings came from non-targeted natural phenomena such as birds and thunderstorms. It was clear that some kind of classification capability was necessary and that the simplest approach was to exploit the signatures in the frequency domain. MITRE developed a fixed-point Fast Fourier Transform (FFT) algorithm in TinyOS, the prevalent mote-based operating system, suitable for use on a large family of motes. Prevailing wisdom at the time was that this was unachievable within reasonable energy budgets. But with only modest effort, we were able to implement an FFT on a mote that exhibited acceptable round-off error and was more than 900 times more energy efficient than the previously published algorithm.

For more information, please contact Kenneth W. Parker using the employee directory.

Solutions That Make a Difference.®
Copyright © 1997-2013, The MITRE Corporation. All rights reserved.
MITRE is a registered trademark of The MITRE Corporation.
Material on this site may be copied and distributed with permission only.

Privacy Policy | Contact Us