Cheap sensors plus mesh networking could yield an effective alarm system for biological and chemical attacks.
We haven’t heard much about biological warfare since envelopes containing anthrax spores were mailed to Congress and some news media in the fall of 2001. But it’s still a concern for the military and the Department of Homeland Security. So researchers at Lincoln Laboratory are pursuing the use of commercially available mesh networking technology to make populations safer from airborne pathogens.
Sensor networks have existed for several years. Companies like Crossbow, Ember, and Dust Networks manufacture radio platforms that autonomously communicate with their neighbors, forming mesh networks that relay such data as temperature, moisture, and other environmental characteristics back to a central point. These systems haven’t been used for biological or chemical defense applications, though, because sensors that can detect specific biological or chemical threats are not designed for low-power, low-cost deployments, says Adam Norige, a biomedical engineer in the Laboratory’s Biodefense Systems Group. State-of-the-art detectors incorporate sophisticated technologies that are usually geared toward improved performance rather than reduced cost, and scattering dozens throughout a city could be prohibitively expensive.
Norige’s aim is to develop a sensing architecture focused on low-cost distributed detection, and to test these networks with various sensor prototypes. “Inexpensive biochemical sensors, which may even approach disposable in terms of cost, don’t get a lot of attention from the R&D community,” he says.
Although the sensitivities of inexpensive sensors may be a couple of orders of magnitude lower than those of the most advanced devices, linking them in a network can increase their overall detection performance. For one thing, no matter how sensitive a detector is, it works only if it’s in the right spot to detect a threat—and given how unpredictably clouds of toxins can blow around, choosing the right location can be a difficult task. Scattering large numbers of sensors increases the odds that one of them will intercept the plume of threat particles that might be drifting toward a commercial district. “You don’t have to be as sensitive, because the odds are increased that at least one or two sensors will be closer to the release point,” Norige explains. And while a single sensor gives information at only one point, an array could provide spatial data about the plume’s structure and propagation. “Once you have a network of sensors out there, you’re gathering much more information than you would with a single, typical sensor.”
Sensor at sunrise: a grid of high-fidelity smoke detectors and collocated anemometers was tested at Fort Devens, Mass., in August 2007. Such an array of sensor nodes can monitor the smoke plume’s shape and movement—information that would be critical in a biological or chemical release.
For instance, with a network of sensors, officials could look at which sensor registered a signal at a given time and draw up a map of the plume’s spread. That would help track the plume in real time—yielding clues as to where it came from and where it was headed.
“That would be a very helpful adjunct,” says Robert Weiss, founder of Physical Sciences Inc., a company that develops sensors and other new technologies for government and industry. “The systems that are out there now are very large, very expensive, and few in number.” He thinks the best arrangement may be to deploy one or two expensive sensors and then add Lincoln Laboratory’s cheaper network, for a better mix of wide range and high sensitivity.
Weiss has discussed the work with Norige and thinks it shows promise for dealing with the threat of bioweapons, which he doesn’t feel has been sufficiently addressed. “I don’t think enough people are working on it,” Weiss says.
The Lincoln Laboratory group has conducted proof-of-concept experiments using smoke as a threat cloud and a particle detector at each network node. In addition, each node included an anemometer to measure wind speed and direction. With such information as part of a real alert system, Norige says, officials could say, “We’re seeing the threat here. Look at the way the wind’s blowing. The people over here have so many minutes to get out of there.” Norige and his colleagues are also working on algorithms to help distinguish between real detections and false alarms and track the plume’s propagation. They are performing outdoor field experiments to test with plumes generated in an uncontrolled environment and supplementing their results with computational fluid dynamic analysis.
These networks could help detect chemical plumes as well as biological ones. Toward this end, Lincoln Laboratory is working with Timothy Swager, head of MIT’s chemistry department, to develop an inexpensive chemical-agent sensor. Swager’s device is based on carbon nanotubes and polymers with attached molecules that are designed to bind to specific chemical agents. When the agent binds to the polymer, it exerts pressure on the nanotubes, increasing the system’s electrical resistance and thus signaling a detection. The Lincoln Laboratory team has built Swager’s sensors into prototypes that incorporate commercially available mesh networking technology and is currently characterizing the prototypes in the laboratory.
Norige hopes to be able to build the sensor network for about $1000 per node; if it becomes commercialized, the price could drop to $100 to $200 per node, complete with sensor and networking equipment. At such a price, local officials could spread the sensors around to provide warning of biological threats. As part of its standard procedure in securing an area, the military could distribute sensors while setting up an operations base. The Environmental Protection Agency could place them near industrial plants to assure compliance with clean air rules or to warn if there’s a chemical leak. It’s just a matter of showing people that such a setup could work effectively and affordably, Norige explains.
“Low-cost, low-power, distributed detection is a hot field,” Norige says. “We’re trying to find it a home in biological and chemical defense.”