A train has derailed at the edge of a city, spreading toxic chemicals and fumes over a wide area. Before rescue and decontamination workers can enter the danger zone, they need more information: How widespread is the contamination? Where are the hotspots? Where and how are toxic gases moving?

A helicopter swoops over the area, releasing a flurry of tiny devices, each about the size of a dime. They contain sensors that sample the air for toxins and tiny radio transceivers that allow them to communicate with one another and report to a van at the fringe of the disaster area. Inside the van, a screen lights up showing where the contamination is and how it’s spreading.

Such a system is the goal of a new research project at Cornell University that brings together molecular biologists, device physicists, telecommunications engineers, information and game theorists, and civil engineers to develop “self-configuring”sensor networks for disaster recovery. The project, involving researchers from Cornell and the Wadsworth Center of the New York State Department of Health, is funded by a $2.5 million, five-year Information Technology Research (ITR) grant from the National Science Foundation (NSF).

Sensors distributed over a disaster area will map their own locations and report back to observers in a plane or ground vehicle, giving information about the distribution of toxic chemicals, disease, or in some cases, locations of survivors.

While initial research will focus on the detection of biohazards, the underlying principles can be applied to many other situations, including searches for earthquake victims (using audio and body-heat sensors) and monitoring of municipal water systems for leaks or contamination, according to Stephen Wicker, Cornell professor of electrical and computer engineering, who heads the research team.

In the aftermath of a disaster—whether it be an earthquake, fire, building collapse, or terrorist attack—the most pressing need is for information, Wicker explains. Because often it would be dangerous or even impossible to collect data manually, the plan is to create an automated self-configuring remote sensor network. The idea grew out of studies of how such networks could be used on the battlefield, Wicker notes, but the NSF project focuses on civilian applications. “If they can save the lives of soldiers, you can use them to save the lives of civilians,” he says.

—Blaine P. Friedlander Jr., Cornell News Service