Finding firefighters through heavy smoke
An RF-based locator system may someday help fire departments pinpoint firefighters in trouble.
By Martin Rowe, Senior Technical Editor -- Test & Measurement World, 5/1/2009 2:00:00 AM
 LEARN MORE Information on the December 1999 fire in Worcester, MA, that took the lives of six firefighters is available from the Worcester Telegram: www.telegram.com/static/fire Complete information about the history of the PPL project is available on the WPI Web site: www.ece.wpi.edu/Research/PPL See a diagram of the PPL test site as well as photos of the equipment and the project team. See other articles from our May 2009 issue. |
WORCESTER, MA—In December 1999, six firefighters in this city lost their lives because they couldn’t find their way out of a burning warehouse building through heavy smoke. Had firefighters outside the building been able to locate their lost comrades, some, if not all, might have been saved.
This tragedy led professors at WPI (Worcester Polytechnic Institute) to initiate the PPL (Precision Personnel Locator) system project in 2000. When completed, the RF-based PPL system should enable rescue teams to find injured firefighters in burning buildings. It may also have military applications.
Today, finding a firefighter in heavy smoke requires a rescue team to search each room in a building, usually by feeling around because smoke completely obscures their vision. With the PPL, an incident commander outside the building could direct rescuers to the room containing the firefighter in distress. That could drastically cut the time needed to find the firefighter, possibly saving a life.
ECE (electrical and computer engineering) professors R. James Duckworth and David Cyganski lead a team of graduate students who develop and test the PPL. Graduate student Vincent Amendolare works on the project as part of his PhD program. Five other graduate students and technician Bob Boisse round out the team. Duckworth and Cyganski also call on other ECE faculty members when they need additional expertise.
Boisse supports the efforts of the graduate students, who design and test hardware, firmware, and software. Through bench tests, open-field tests, and indoor tests, the students collect data that they use to make design changes and add features that improve the system’s performance.
PPL at a glance
The PPL consists of battery-powered RF transmitters that are worn by firefighters (Figure 1), four receiver stations that can be located in fire trucks around the site, and a base station that resides in one of the trucks. The base-station computer calculates the locations of the transmitters by measuring the phase shift in transmitted signals received from up to four “bowtie” antennas per receiver. The computer then displays the locations of the transmitters.
 Figure 1. A firefighter tests the Precision Personnel Locator system, whose transmitter unit is mounted to the oxygen tank. |
Figure 2 (click on figure name to view figures) illustrates the relationship of the system components. Each transmitter sends a signal consisting of approximately 100 unmodulated carriers over a range of 550 MHz to 700 MHz to the receiver. Figure 3 shows a spectrum analyzer display of the signal. Each receiver unit downconverts the signals to a bandwidth from 30 MHz to 180 MHz. The receiver then digitizes the signals with four 400-Msamples/s ADCs (analog-to-digital converters) for each channel. After performing an FFT (fast Fourier transform) on the signals, the receiver encapsulates the frequency-domain data into Ethernet packets and sends the data to the base station (a Linux-based PC) over Ethernet. Early designs used wired Ethernet, which is impractical at a fire scene. The current design uses wireless Ethernet (IEEE 802.11).
The base station runs Matlab scripts that process the data in the frequency domain. From the phase shifts, the station can calculate the locations of the transmitters and display those locations on its screen in real time (Figure 4), which provides enough information for firefighters. As part of the system development, researchers can analyze data in the lab for greater insight.
The transmitters are also equipped with sensors that monitor the firefighter’s heart rate and temperature, sending that data to the receivers over a 915-MHz data link. Because firefighters can only find each other through touch in heavy smoke, the PPL system needs location accuracy to within 30 cm. To reach that goal, students perform bench tests, open-area tests, and building tests. From the test results, they make improvements to hardware and firmware in the transmitters and receivers, modify the software in the base station, and then repeat the tests. Thus far, tests inside buildings have shown accuracy to within 10 cm.
 Figure 2. The PPL system consists of transmitters, receivers, and a base station. Firefighters carry transmitters, and the receivers are located on fire trucks. |
One recent change to the design involved changing the transmitter signal bandwidth from 60 MHz to 150 MHz. Before running an open-field test, the students measured the transmitted spectrum in the lab using an Agilent Technologies spectrum analyzer. Bench testing lets the students verify the engineering characteristics of new hardware, firmware, and software. The wider bandwidth resulted in reduced errors when the students tested the system in a campus building (Ref. 2).
Open-field tests
Rather than go from performing tests on the bench to performing a test in a building, the team first conducts trials in an open area such as a football field, something they didn’t do in the early stages of the project. “Several years ago, we learned not to go from the lab to a test in a building,” said Duckworth. “We found some errors that needed correcting, resulting in invalid data and wasted time.”
 Technician Bob Boisse (left) and graduate student Vincent Amendolare work on a receive station that contains two RF receiving (bowtie) antennas and a wireless Ethernet antenna that receives data from another receive station. The receive unit is inside the bin. Watch a report of the Worcester firefighters testing the PPL system on NECN. |
Performing tests in an open field eliminates a significant variable. In a building, multipath signals caused by reflections can produce location errors. To eliminate that variable, the team will first test the system in a field where they have line-of-sight conditions and minimal signal reflections. This intermediate step provides the team with a “sanity check” that the system is functional before testing it in a building. “Multipath is by far the greatest challenge to RF precision positioning,” said Cyganski. “No matter what RF specifics are involved, multipath is a fact often overlooked by newcomers to the problem of indoor location.”
Recently, the students added a wireless Ethernet link between the receiver stations and the base station, a change that created a new technical requirement: the synchronization of ADCs among the receiver units. Synchronization is crucial because of the nanosecond timing required to accurately measure signal timing delays (signals travel at about 30 cm/ns). The students needed an open-field test to verify that the system could synchronize receiver ADCs.
A clock in a receiver unit can synchronize ADCs within that unit, but it cannot synchronize the ADCs in other receivers when communicating over wireless Ethernet. Graduate student Amendolare explained how the PPL achieves synchronization. The clocks in the receivers tend to drift relative to each other over time. A reference transmitter in the system transmits a signal that the receiver units lock onto. The system calculates the timing errors from the received data and corrects them.
At an indoor line-of-sight test on December 22, 2008, on the WPI campus, the students ran a test to verify synchronization of the ADC clocks. The test, conducted in a 15-m² area, proved that software synchronization could bring the clocks to within 0.5 ns of each other.
A 15-m² test area is good enough for a small building. But the project’s current sponsor hopes to use the system at a much greater distance, perhaps up to 1 km² as a vehicle locator. So, the team set out to try the system on a 200-m² grid at a former military airstrip on March 17, 2009. I attended that open-field test, the first one conducted off campus.
Test prep
Prior to the field test, the students simulated the increased transmission distance at the 15-m² test. Amendolare explained that they simulated the greater distance by reducing the transmitter’s power using RF attenuators. Those tests gave them confidence that the system could work at the larger test site.
 Figure 3. Transmitters worn by firefighters broadcast about 100 unmodulated carriers over a 150-MHz bandwidth. |
The team must conduct a survey of the test site before bringing in the PPL equipment. On March 16, several students conducted a spectral survey of the airstrip test site with an Agilent spectrum analyzer. The PPL operates in the 550 MHz to 700 MHz band, which is currently used for broadcast TV. (The team needs a temporary FCC license to conduct an open-field test or a test in a building.) Depending on the results of the spectral survey, the team may need to avoid using carriers at certain frequencies. Students can change the waveform with a command over the wireless Ethernet link, but the mobile locators must be reprogrammed over a cable.
Students also performed a site survey where they formed a grid by placing a mark every 20 m on two adjacent sides of the runway. The PPL team had several plans for this test:
-
Check that the wireless Ethernet links between the receivers and the base station functioned properly, which they did.
-
Verify the integrity of the tripods that hold the receiving antennas to see if they needed guy wires to prevent them from tipping in the wind. (There was virtually no wind on this day.)
-
Test the system with eight RF receiving antennas instead of 16 to simplify setup and reduce costs.
-
Test the accuracy of the locator on the larger grid by placing a transmitter at each of the 20-m marks along the grid, transmitting a signal, and then moving the transmitter to the next marked location.
 Figure 4. With the base station’s display, an incident commander can view the location of firefighters and monitor their physiological conditions such as temperature and heart rate. |
The students arrived early on March 17 to set up a receive station at each corner of the grid. The tripod for each of the four receive stations held two bowtie RF antennas connected to a receiver unit. Each receiver unit’s Ethernet port connected to an Ethernet switch, which in turn connected to a wireless Ethernet transmitter on the tripod. A gas-powered generator at each corner of the grid provided power to the receivers and base station. (See a diagram of the test site as well as photos of the site’s equipment and project team.)
Three of the four receivers communicated to the base station over wireless Ethernet. The receiver at the corner closest to the base used a wired connection. Because three of the stations needed a wireless link to the base, the students needed three data-link receive antennas at the base station.
They placed two of the three wireless data-link receive antennas on a fifth tripod containing one bowtie antenna, which is used for the reference transmitter. Boisse mounted the third wireless data-link receive antenna on the tripod for the receive station that was wired to the base station. Each wireless data-link receive antenna was pointed to a different corner of the grid.
At the March 17 field test, the project team encountered a problem. At the longer (200-m²) distance, the system couldn’t find the reference transmitter’s signal. Students experimented by moving the reference transmitter across the grid, but they were still unable to get a strong enough signal.
 Professor R. James Duckworth with a receiver unit in the lab. |
Before leaving for the day, student Jorge Alejandro used an Agilent spectrum analyzer to measure the signal strength from the reference transmitter. Amendolare wanted the measurement to see if the reference signal was in fact strong enough to use on the large grid, as his bench tests had indicated. “We need to verify if the problem is with signal strength or if our software is not properly detecting the signal,” he said. The students left for the day, intending to return the following week to test the system again, but my schedule prevented me from returning with them.
During a typical open-field test, students check how accurately the system locates a transmitter at known locations. If the displayed locations are within tolerance, a student will move the transmitter around the grid to unknown locations by walking around. In the case of the 200-m² test site, the transmitter will be on a truck. The base station’s display should be able to keep up with the transmitter as it moves. Early versions of the PPL system lacked that feature. “Adding the real-time display saves a great deal of time,” said Professor Duckworth. “At first, we had to go back to the lab to process and analyze the data.”
Back at the lab
Even with the real-time display, students still process and analyze the data in the lab. Over the course of a week following an open-field test or building test, Amendolare will adjust parameters and compare results. For example, he can adjust the spatial density of the scan. That provides greater resolution on the display, which lets him see how finely he can measure displacement. Increasing spatial density slows the processing speed.
 Professor David Cyganski (foreground) advises graduate students such as Vincent Amendolare (background) on RF technology. |
Amendolare compares the results to get an idea of how the system performs under different conditions. For example, he can remove half the bandwidth from the transmitters and check location accuracy or remove the data from half of the RF receive antennas to see how that affects location accuracy. He performed that analysis on the data from the December 22 open-field test, which led him to believe that the system could perform satisfactorily with eight antennas.
The students have also performed PPL system tests in buildings, which is the real test. They started in a WPI lab and have since moved to a house on campus. For the past few years, they’ve conducted these tests as part of a workshop on the system. They’ve even conducted building tests with Worcester firefighters, who are eagerly awaiting a system they can use.
| References |
|
3G Technologies Complicate CDMA Testing
02/01/2000Precision Personnel Locator test site
04/30/2009Protocol stack testing for LTE
06/01/2008Base station analyzer
03/31/2005
-
FLIR offers IR camera for under $3000
-
Don't let the economy compromise quality (Guest commentary)
-
Danaher speeds up restructuring, acquires life-sciences businesses
-
Agilent’s Cover-Extend technology eliminates need for physical test points for in-circuit test
-
So many combinations: Testing a switch-matrix board
View All Webcasts
