Hey there, human — the robots need you! Vote for IEEE’s Robots Guide in the Webby Awards.

Close bar

When Disaster Strikes, Flying Cell Towers Could Aid Search and Rescue

Base stations carried by drones would form an ad hoc network and connect first responders

11 min read
A resident calls out to a rescue crew in a flooded Houston neighborhood after Hurricane Harvey struck the region.
Photo: Joe Raedle/Getty Images

A resident calls out to a rescue crew in a flooded Houston neighborhood after Hurricane Harvey struck the region.  A Call for Help: A resident waves to a rescue crew in a flooded Houston neighborhood after Hurricane Harvey inundated the region.Photo: Joe Raedle/Getty Images

When Hurricane Harvey struck land in southeast Texas on Friday, ferocious winds and rain nearly obliterated the telecommunications network in Aransas County, a coastal community that lies directly in the storm’s path. Only one of the county’s 19 cell towers was still working on Saturday, according to a report [PDF] by the Federal Communications Commission.

A day later, as the storm continued to batter the region with rain that the National Weather Service has called “unprecedented,” 85 percent of towers [PDF] in nearby Refugio and Calhoun Counties also went offline.

Before the storm, wireless carriers took careful steps to prepare, filling backup generators with fuel and stationing response crews in the area. And throughout the past four days, they have managed to keep most cell towers running—the FCC estimates about 5 percent of cell towers in the official disaster area are not functioning.

The three hardest-hit counties—Aransas, Refugio, and Calhoun—were under mandatory evacuation orders when the hurricane hit land. Still, anyone who remained in the area, and the first responders who are trying to assess damage, can most likely not count on reliable cell service at this critical time.

I see a better way. Emergency responders could use a fleet of drones to send up temporary base stations that operate in midair. With this system in place, responders from any agency could share information, as well as photos and videos of the scene, to coordinate their response.

The U.S. government even described the merits of such an approach in 2011, but the idea never took off. Part of the reason is that there is still no commercial drone-based system for disaster communications.

So in 2013, I started to build a model for one with my colleagues at the University of North Texas and our partners at MIT Lincoln Laboratory, Pennsylvania State University, and several other institutions. Our system is designed to be stored in the cargo hold of a fire truck or ambulance. Crews could launch the first drones, fused together with a miniature base station, within minutes of arriving at a disaster scene—long before any telecom company shows up. As the situation evolves, the drones would automatically reposition themselves as aerial base stations to maximize coverage.

We have tested several versions of our system over the past three years. Now we intend to partner with private companies and public safety agencies to help emergency responders around the world stay connected at times when communicating is, quite literally, a matter of life and death.

Photo: Allison V. Smith

Maurice Griffin, a drone pilot, screws a GreenCell base station to the bottom of an AR200 drone.

First responders struggle to maintain basic communications during a disaster in part because of the disjointed nature of the telecom system: In the United States alone, public safety agencies use more than 10,000 separate radio networks.

The consequences can be harrowing. On 11 September 2001, officials in New York City transmitted urgent instructions for emergency responders to evacuate the World Trade Center approximately 21 minutes before the second of the two towers collapsed. Many police officers in the tower heard the dispatch and escaped, but firefighters in the area never received the warning: Their radios operated on different channels.

A decade later, a magnitude-9.0 earthquake struck the eastern shore of Japan, followed closely by a massive tsunami. The NTT Group, one of Japan’s largest telecommunications companies, sustained damage to 6,700 pieces [PDF] of base-station equipment, and floods ruined 65,000 of the telephone poles in its network. Mobile phones, laptops, and even landlines performed poorly during the critical four days after the earthquake.

Whenever a telecommunications company is faced with extensive network damage, technicians begin hauling temporary base stations to damaged tower sites. They also dispatch mobile units, which are large trucks with base stations attached to antenna masts that stick up from the back. However, these trucks are expensive to maintain and store, and the mobile units can reach only the areas that remain accessible by road. Once the mobile units are in place, their antennas can’t reach as high as a typical cell tower can, which means that the units are more susceptible to interference. In a crisis, this can be deadly: During disasters, call volumes have surged to 40 times [PDF] their normal levels.

In the United States alone, public safety agencies use more than 10,000 separate radio networks

An aerial communications system supported by drones could be deployed much faster and operate with minimal interference. In 2013, we started to think about what such a drone-based communications system for public safety agencies might look like. We knew it would need a shared radio-frequency channel for first responders, drone-portable base stations, a power supply, and a digital database for exchanging information. We would also need controllers that would be easy enough for a licensed drone pilot to operate in a crisis.

Our first major challenge was to find a base station small enough for a drone to support. Drones under 25 kilograms—the limit now imposed by U.S. air-safety regulators—can carry a maximum payload of about 2 kg, so we would need a base station that weighed less, even with its battery.

First, I approached several major telecommunications and aviation companies, but not a single company thought this business would be profitable. Next, I reached out to the National Public Safety Telecommunications Council, which was just beginning to investigate drone-based communications systems in 2015. It has since started to develop operational requirements to support the concept of drone-deployed telecom networks. However, the council, as an industry consortium, is not in a position to manufacture anything.

Finally, my search led me to a startup named Virtual Network Communications. This company, based in Chantilly, Va., sells a product called a GreenCell that seemed suitable. It’s a scalable LTE base station, known as a picocell, which is typically used to extend the reach of an existing network but can also generate its own network. The base station contains an E-UTRAN Node B radio with two antennas and a credit-card-size component called a Micro Evolved Packet Core, which uses LTE technology to form an ad hoc network with nearby radios. Then, that local network connects to a nationwide cellular network.

With these components, our GreenCell can support communications for up to 128 users at a time from a distance of up to about 2 kilometers on any LTE frequency. Better yet, it measures just 12.5 by 12.5 centimeters and weighs only 2 kg with its battery, just light enough to be lifted by a drone.

Once we had found a suitable base station, we still needed to find a suitable drone. Ideally, it would be affordable and be capable of flying for 10 to 12 hours before needing a recharge. Unfortunately, no such drone exists today. Most commercial drones can stay aloft for fewer than 45 minutes.

After some research, I found a company named CyPhy Works, which has developed a drone powered through a 150-meter cord that extends up from a grid or generator. Technically, this drone could stay in the air for as long as it had access to a power supply on the ground. But in a disaster scenario, it would have to be tethered to a van loaded with a generator and fuel. That would limit it to serving the same road-accessible places to which mobile units already travel. Another drawback: The drone’s tether restricts its mobility once it’s in the air. We wanted to be able to reconfigure our network in an instant.

We briefly considered using balloons instead of drones, but we discovered through trial and error that balloons are difficult to reposition and hold in place, especially during high winds.

We decided instead to use the AR200 drone from AirRobot, a company based in Arnsberg, Germany. The AR200 has six rotors that allow it to hover more steadily than the usual four. And because the AirRobot drone is battery powered, it can zoom off to any location.

imgPreparing for the Worst: Namuduri places a GPS tracker (built from a Raspberry Pi) alongside a patient to track the trip to the hospital during an emergency response drill.Photo: Allison V. Smith

After we selected our drone, we needed to figure out how to power the base station without adding too much extra weight to the entire unit. One option we considered was to rewire the drone’s battery so that it would power both the drone and the base station it carried. But siphoning off power from the battery to support the base station would reduce the drone’s flight time to about 30 minutes.

We ended up using a separate battery for the GreenCell. We attached a container with the base station to the bottom of the drone, with the GreenCell’s two antennas pointing toward the ground. From the standpoint of radio performance, this upside-down arrangement provided the clearest path for a signal between the base station and its users.

To operate our base station, we needed to use a frequency that first responders could count on without the risk of being interrupted or the channel becoming too crowded. We chose Band 14, which includes a block of spectrum within the range of 700 to 800 megahertz that is reserved in the United States for public safety communications. This band is managed by FirstNet, an independent authority within the U.S. Department of Commerce, which is planning to use it as the basis for a national public safety broadband network.

By testing our system on Band 14 from the start, we could ensure that it would be compatible with that future public safety network. And Band 14 is an excellent frequency to work with—the relatively long wavelengths travel farther and move more easily around hills and walls than waves at higher frequencies (such as the LTE band at 1920 MHz). We obtained approval from FirstNet to experiment within the band well in advance of our trials, and a company called Sonim Technologies, located in San Mateo, Calif., supplied us with mobile phones programmed to Band 14 for our tests.

With many of the hardware components in place, we turned our attention to software. Back in 2009 at MIT Lincoln Laboratory, a team of researchers had developed the Next-Generation Incident Command System (NICS) for wildland firefighters. This tool improves first responders’ situational awareness during a disaster and makes it easier for decision makers to allocate resources. The system has since been used by 450 public safety groups to coordinate their efforts during emergencies, including those that responded to a devastating mudslide that hit the town of Oso, Wash., in 2014.

We worked with the MIT Lincoln Laboratory team to integrate NICS into our trials, so that emergency responders could use our system not only to restore cellular service but also to share information by texting photos and videos to the system. To use NICS, agencies preregister all users and their devices in the system. Those users are assigned privileges based on their role and can access the system on their laptops or smartphones.

Once it’s live, NICS collects users’ GPS locations from their devices over a Wi-⁠Fi or cellular connection to give commanders a bird’s-eye view of their team in action. The system can also show the location of each drone, using its onboard GPS, to allow decision makers to view and adjust the coverage area. In future versions, we plan to deploy tiny sensors on the drones to measure air temperature in disaster zones.

Our drone-supported base station, which we call the Aerial Deployable Communications System, operates hand in hand with NICS to provide reliable coverage. We began to test early versions of this system in 2015, when we attached a Wi-Fi router to a large helium balloon (as a stand-in for a drone) that measured 1.5 meters in diameter. We flew the balloon at about 12 meters and later up to 21 meters on the University of North Texas (UNT) campus, located in Denton, showing that Wi-Fi coverage improved as the base station flew higher.

In 2016, we completed another demo in Austin, Texas, during an exhibition for the Global City Teams Challenge, an event focused on improving public services through technology. There, we placed a GreenCell tuned to Band 14 on a tabletop and used smartphones from Sonim to show it was possible to transmit video in this band from some distance away.

During 2016, we spent four months working with local government and public safety officials in Denton to test NICS during a live simulation of the aftermath of an airplane crash. To set the stage, organizers scattered airplane parts all over the runway of the regional airport and recruited students from UNT to play the parts of victims. We attached a GPS sensor to a vehicle that was on scene for the drill, and programmed the sensor to upload its location data to NICS.


imgUrgent Care: An emergency responder wears a Pi-based GPS tracker built with extra sensors to monitor heart rate and air temperature (top). Four students from the University of North Texas watch live results from the tracker (bottom).Photo: Allison V. Smith

This year, we deployed our entire system as we originally envisioned it. On 3 May 2017, at a gun range in Waxahachie, Texas, our first AirRobot drone carried a GreenCell base station and provided live cellular coverage. Two days later we and Denton city officials tested a similar base station airlifted by a helium balloon to provide coverage to first responders during a drill held at Apogee Stadium on the UNT campus.

Our goal for both these demonstrations was to establish reliable cellular communications, and to enable sharing of critical information including voice, text, images, and video among first responders immediately after a disaster. In both cases, we used the Sonim smartphones programmed to operate on Band 14 to test two key measures of signal quality.

The measurements showed that with 250 milliwatts of transmit power and flying at an altitude of 120 meters, the system could provide cellular coverage to users within 2 km—many times farther than a traditional cell tower could reach with the same transmit power. By increasing the transmit power of a GreenCell to as much as 10 watts, cellular coverage could be extended to several kilometers. Several GreenCells in the air and on the ground, connected to a Micro Evolved Packet Core, could reach 10,000 users over about 230 square kilometers, about the size of Denton.

In the second drill, we also tested two wearable gadgets designed to transmit information to emergency coordinators. A firefighter wore one gadget that recorded his heart rate, GPS location, and surrounding air temperature and shared that data with a computer at the nearby operations center. We placed another gadget on an ambulance with a student playing the role of a patient and tracked the vehicle’s GPS location throughout the trip to the hospital. With further development, we think these passive sensors could play a critical role in augmenting the information that responders upload directly to NICS.

We’ve come a long way toward providing drone-based emergency communications, but neither the technology nor the market are ready for prime time just yet. On the technology side, we must still find a way to maintain a reliable connection even as a drone is flying around. Most base stations, including GreenCells, are designed to be stationary. When both a base station and its users are in motion, signals miss their targets more often. Our high-flying base stations may require additional signal processing or perhaps phased arrays, which can control signals more precisely than the fixed antennas used on GreenCells today, to deliver a steady data stream.

In addition, we need to figure out how fast and how high a drone should fly to provide the best coverage. We know that the average base station stands between 40 meters and 60 meters tall. In the United States, drones are prohibited by the Federal Aviation Administration from flying at an altitude over 122 meters. This year, we are completing trials that will help us find the sweet spot for cellular service in Band 14 within that range.

The final question is the cost. A commercial drone capable of carrying a sufficient payload may cost up to US $100,000. A GreenCell or similar portable base station costs tens of thousands of dollars. Each handheld mobile device also adds as much as a thousand dollars to the bill.

Over time, we are hopeful that we will see drone-based communications systems that are compact, affordable, and capable of staying aloft for several hours at a time. The good news is that we are no longer the only ones working on this idea: Facebook is developing a tethered drone that could provide Wi-⁠Fi to disaster zones, while a research team from Ghent University, in Belgium, has proposed a drone base station concept similar to our own. Drones are catching on with the public, too, which should lead to lower prices. Someday, the rewards of building such a system could be measured not just in costs avoided but in human lives saved.

A version of this article appears in the September 2017 print issue as “Flying Cell Towers to the Rescue.”

About the Author

Kamesh Namuduri is a professor of electrical engineering at the University of North Texas, in Denton. During a recent test of his drone-supported base station, the base station detached, plunged 120 meters, and shattered. He’s now looking for a better attachment method. He hopes the system will one day help search teams communicate after disasters. But he struggles to find funding for the project. “When it comes to saving people, I don’t think there is enough investment,” Namuduri says.

This article is for IEEE members only. Join IEEE to access our full archive.

Join the world’s largest professional organization devoted to engineering and applied sciences and get access to all of Spectrum’s articles, podcasts, and special reports. Learn more →

If you're already an IEEE member, please sign in to continue reading.

Membership includes:

  • Get unlimited access to IEEE Spectrum content
  • Follow your favorite topics to create a personalized feed of IEEE Spectrum content
  • Save Spectrum articles to read later
  • Network with other technology professionals
  • Establish a professional profile
  • Create a group to share and collaborate on projects
  • Discover IEEE events and activities
  • Join and participate in discussions