REEF: Putting Sensors to the Test

By Daniel Luke, Stephen Theophanis, William Dowling, and Dave Allen

Although MITRE researchers have proved that even parking lots can serve as successful testing spots, many aspects of the Netted Sensors Initiative require a richer, more versatile testing environment.

MITRE's Netted Sensors Research Evaluation and Experimentation Fabric, or REEF, provides an infrastructure with which we can develop technologies and algorithms, quantify capabilities and limitations, evaluate and integrate components, and model systems as we help our sponsors explore the capabilities of netted sensors.

REEF consists of two major components: a simulation testbed for modeling netted sensor components and systems and a hardware testbed for supporting experiments that require laboratory hardware and field deployments.

The simulation testbed can replicate a multitude of sensor types (radar, acoustic, seismic, and video) and target types (vehicles, aircraft, people, and animals). It can also host live or recorded sensor data, as well as high-fidelity 3D virtual environments. The simulation testbed has proved to be a rapid tradeoff and performance analysis tool with which to evaluate the effectiveness of various sensor networks. To enable easy integration and scalability with Department of Defense (DoD) projects, the testbed employs the Distributed Interactive Simulation protocol standard employed by the DoD and is designed to migrate, as necessary, to the new High Level Architecture standard the DoD has adopted for all new distributed simulation developments.

Similar to the simulation testbed, the hardware testbed is an adaptable infrastructure designed to facilitate netted sensors experimentation. Its hardware and software components are modular in design, which allows for flexibility and rapid reuse. The hardware testbed consists of a variety of sensors (acoustic, seismic, electro-optic, magnetic, radar, imaging, etc.) that can be controlled by individual node computers. The node computers can be connected over a radio network in multiple configurations that allow the sensor network to adapt to a variety of applications.

Modeling Mexico

During the Netted Sensors Initiative's first year, we employed REEF as we tackled the border monitoring and perimeter security challenge problem. We wanted to investigate the ability of netted sensors to monitor, detect, and classify illegal vehicle and human traffic across borders. In this case, we chose to model the U.S./Mexico border.

REEF offers high-fidelity dynamic target modeling capabilities. We used these capabilities to add realistic humans and vehicles into our sensor simulation scenarios. In the case of the human models, various modes of locomotion, including crawling, walking, and running, were characterized using a 3D-motion model composed of 53 skeletal joints. We are also designing animal models in order to simulate the false alarms and erroneous classifications that can be triggered by animals crossing a sensor's coverage field. (No need to keep tabs on every coyote loping along the U.S./Mexican border or every bear pacing back and forth between the U.S. and Canada!)

To evaluate a sensor's real world performance, we have to accurately characterize the environment where the sensor network will reside. REEF's ability to incorporate a high-fidelity 3D virtual world allows us to fully evaluate limiting factors on sensor performance such as terrain obscuration, background clutter, and signal propagation media. We re-created the Arizona/Mexico border region stretching from New Mexico on the east to Nogales, Arizona, on the west in order to simulate various sensor laydown and border crossing scenarios.

Our REEF simulations confirmed what we've learned from our previous research in netted sensors. Algorithms are required for node localization, timing synchronization, data acquisition, resource management, information management, blue force tracking, and sensor exploitation. Clearly these algorithmic components require a broad range of computational power. The REEF simulations once again supported MITRE's notion that a tiered system of nodes is the most efficient way to supply a netted sensor system with the computational power necessary to fulfill its purpose.

The REEF Lab is a focal point for development of high-fidelity 3D virtual environments in support of distributed sensor simulations.

What's Next

The next challenge problem slated for the Netted Sensors Initiative is "Situational Awareness in Support of Combat Identification." We hope to prove the benefit of netting stand-alone unattended ground sensor systems to perform the tasks of target detection, classification, and tracking of multiple targets over a wide area.

Success in this and other netted sensor challenges rests on the development of several key technologies and components. With the help of REEF, we are confident that the development of those key technologies and components will soon be at hand.

For more information, please contact Daniel Luke, Stephen Theophanis, William Dowling or Dave Allen 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