After several Worcester firefighters died in a burning warehouse, unable to find their way out, four WPI professors are teaming technology and expertise to prevent future tragedies.
Lost in the mazelike layout of a massive warehouse filled with thick black smoke, two firefighters gasped for breath. Their air tanks were nearly empty; the men were running out of time. It was Dec. 3, 1999, and Worcester firefighters Paul Brotherton and Jerry Lucey were trapped inside the Worcester Cold Storage warehouse. Separately, two pairs of their brethren had answered their radio distress call, but they, too, became disoriented in the dense smoke and roar of the flames. Before the night was over, all six lost their lives, each within 100 feet of exits they simply couldn’t locate. The Worcester Cold Storage tragedy made international headlines and shone a spotlight not just on the dangers of the firefighting profession, but on the enormous challenge of tracking the whereabouts of emergency personnel when they enter buildings.
From heroic sacrifice, a better idea
Firefighters who lose their bearings typically rely on ropes to find their way out. This system works—if the rope doesn’t go up in flames or get lost in the murk. An alarm that sounds when a firefighter stops moving has also proven unreliable; the alarms on Brotherton and Lucey were drowned out by the fire’s roar.
Professors Cyganski, Duckworth, Orr, and Michalson
In the days after the Worcester fire, John Orr, professor of electrical and computer engineering and then head of the department, began to think that there had to be a better way. On Dec. 9, 1999, he joined tens of thousands of mourners who lined the streets of Worcester to watch a three-mile-long procession of 30,000 firefighters from around the world wind its way to the Centrum (now the DCU Center) for a memorial service for the six firefighters: Brotherton, 41; Lucey, 38; Timothy P. Jackson, 51; James F. Lyons III, 34; Joseph T. McGuirk, 38; and Lt. Thomas E. Spencer, 42.
It was there, in that solemn setting, that Orr decided that he and his colleagues could harness WPI’s expertise and come up with a better system for locating firefighters trapped or lost in a building fire. But the idea needed funding. The issue caught the attention of Senators John Kerry (D-MA) and Edward Kennedy (D-MA), and Representative James McGovern (D-MA). By February 2003, the legislators had secured $1 million from the National Institute of Justice’s Office of Science and Technology to fund the development of a locator system. Orr mapped out a three-year project, which will culminate with a functioning prototype, and assembled a team of ECE faculty to make the system—intended for firefighters and police—a reality. (Orr, professor David Cyganski, the project co-leader, and associate professors William Michalson and James Duckworth are assisted by graduate and undergraduate students.)
Homing in on the signal
The team first had to determine which communication technologies would help them deliver on the complex criteria required by the First Responder Locator System.
They began by analyzing the Global Positioning System (GPS). “Most people assume that any GPS worth its salt would be able to locate people inside buildings,” says Cyganski. But GPS has proven to be inaccurate indoors. Its satellite-broadcast signals are weak, and when those signals bounce off walls and other surfaces, accuracy suffers. It is also incapable of pinpointing location—30 feet is the best it can do, far from the one foot needed for the First Responder system.
The team also reviewed impulse UWB (ultra-wideband), which relies on sharp pulses for tracking; again, there were drawbacks. “The sharper the pulse, the more radio spectrum it takes up,” explains Cyganski. “You’d have to disrupt all other radio-related services in the area to use a system based on impulse-UWB, which is, of course, wholly impractical.”
The engineers continued their survey and found two communication tools suitable for the job. Super-resolution radar, also called synthetic aperture radar (SAR), extracts great detail from radar signals by applying sophisticated computational methods that were not practical, especially for mobile systems, before recent innovations in computer technology. And orthogonal frequency division multiplexing (OFDM), another recent innovation, transmits high-speed data via wireless devices and integrates well in the radio spectrum.
Once the team had settled on SAR and OFDM as the technological backbone for the system, they needed to find a way to channel the power of these technologies.
A true team effort
In the labs run by Cyganski, Duckworth, and Michalson, a prototype assembly line has been set up that utilizes the expertise of each team member.
Cyganski is the math guy who’s designing from scratch a system that will be able to transmit, receive, and process the signals. In addition, he calculates the best types of signals and the best way to generate them. The customized OFDM signals are emitted continuously by the transmitters worn by each first responder. The receivers are able to decipher the signals and determine their distance from the transmitters.
The process of identifying the transmitter’s exact location is complicated by something known as multipath—the tendency of signals to radiate out from the transmitters and bounce off walls, floors, and ceilings, arriving in a jumbled fashion at the receivers. But while each receiver picks up a multitude of signals, appearing to have arrived via many different paths, when all of the paths are compared some will converge on a single reference point. That point is the true location of the transmitter and the first responder wearing it.
After Cyganski generates the mathematical representations of the signals, he hands them over to Michalson and Duckworth. “We then translate these representations into a flow of electrons,” explains Michalson, the team’s wireless navigation expert. Recent prototypes, including an analog-to-digital converter, lie on tables in Michalson’s lab, along with the tools needed to construct and repair these complex devices. “The models are big now, because fingers are fat,” he says, holding aloft a board about two feet square. “Their size makes it easy to change components and saves us a lot of time.”
He gestures to a board, to which black boxes and a spaghetti of wire are affixed. “This interface between analog and digital data samples the signal two hundred million times per second. Then the FPGA [field programmable gate array] dumps this data into memory and transfers the data to the laptop for processing.”
The FPGA—a small chip critical to the project—is an integrated circuit that the engineers can program to make the transmitter and receiver handle the system’s complex signals with almost no additional components. When the design is complete, the FPGA functions can be mass produced in an even smaller chip, enabling the final devices to be both compact and cheap.
“For the system’s next generation,” Michalson says, “instead of these prototypes scattered across several feet of boards, we’ll have stacks of three boards that you can hold in your hand.”
Enter Duckworth, the embedded system designer; his circuit board designs are created on a computer. Using software, he draws thousands of spiderweb-fine lines, color-coded in brilliant red, blue, yellow, and orange, representing the signals to be used by the system. Squares and circles designate components. “This will be our digital controller board,” he explains, pointing to a design on his screen, “which goes into the receiver and transmitter systems.”
From an envelope, Duckworth removes a small fiberglass square densely packed with lines and shapes matching his computer rendition: it’s a newly minted circuit board, manufactured off-site, using his specifications.
Working toward a deadline
Currently the transmitter and receiver are each made up of three circuit boards. Later, only one board will be required for the transmitter and another for the receiver.
In the interim, the team plans a demonstration of the system with four retrofitted laptops—three to act as receivers, one as transmitter—by summer 2005.
Within the next couple of years, issues such as monitoring the physiological status of emergency responders and making the transmitters impervious to the crematory-like fire environments in which firefighters are called upon to work will be addressed. “But those items are for a later pass,” says Michalson. “If you try to do everything at once, you end up doing nothing well.”
Although Cyganski cautions that “we don’t know how many roadblocks we’ll encounter along the way,” the team believes the final product will be ready within the initially conceived three-year timeline, possibly hitting the marketplace within five years.
“This is a project we had to do,” says Orr. “Technology can solve the problem that killed the firefighters in the Worcester Cold Storage warehouse. We think we now understand better than anyone else why precision indoor position location is such a difficult problem. And we remain confident that we will solve it.”
How the First Responder Locator Will Perform
Fire department personnel arrive at a blazing building. Each firefighter in full protective gear wears a badge-sized transmitter. Three fire trucks equipped with transmitter-receivers are positioned at roughly even intervals around the building to permit three-dimensional maps to be generated on the fly.
Several firefighters walk around the building’s exterior, tapping their badges when they reach an entrance/exit. The taps, relayed from the trucks’ transmitter-receivers, show up as glowing dots on a map displayed on the site commander’s computer.
Firefighters enter the building. Their badges send continuous signals to the receivers, which display lines tracing their every movement. The system senses changes in elevation; at second- and subsequent-floor walk-throughs, the lines on the display change colors. As the lines build up, they create a “picture” of hallways, stairwells, room layouts—a clear, three-dimensional map of the building.
If exits become impassable, the site commander’s display corrects for such changes, showing alternate routes out of the building. The site commander knows where each responder is, to within 12 inches, and can help anyone get out while there’s still time, whether or not firefighters can see through the email@example.com
Maintained by: firstname.lastname@example.org
Last modified: Jan 03, 2005, 13:51 EST