STANDING ON THE roof of the Facebook lab in Woodland Hills, California, I can’t see the airplane I’m looking for. It’s too small and too far away. So I duck under a white tent beside me where a bunch of engineers are watching a live video feed from a nearby camera on a massive flat-panel display. Even at heavy magnification, the aircraft is a tiny spec against the blue sky.
Beside the tent, a ten-foot-tall dish antenna is trying to establish a connection with the plane. On the flat-panel, a small red circle shows the precise point in the sky where the rooftop antenna is aiming. The red circle flits around the tiny spec, without quite reaching it. At times, one will zoom away from the other. But after several minutes, they come together. “We got lock!” yells Abhishek Tiwari, a lead engineer at Facebook, as the two connect.
And then, just as quickly, the connection breaks.
This is one of the first airborne tests of a new wireless technology Facebook is building for use with Aquila, the company’s Boeing-737-sized, solar-powered, long-range drone that’s designed to deliver internet access. Facebook engineers hope this wireless system will one day transport enormous amounts of data from ground stations to dozens of drones in the stratosphere, miles away. The drones will then beam the signal down to ground stations that can serve a city or rural area using more traditional communications—or even send the signal straight to smartphones down below, like a flying cellular antenna.
Either way, Facebook’s plan aims to bring the internet (and, of course, a certain social network) to new areas without the need to dig holes, install towers, and stretch expensive wire lines across the planet.
But the company has a long way to go before reaching that goal. That plane circling the San Fernando Valley in the distance? It’s not the actual drone but a stand-in, a Cessna that Tiwari and his team have outfitted with an extra bulge of equipment on its belly.
The key to the whole idea is an old but suddenly resurgent signaling paradigm called millimeter wave technology. Millimeter waves are smaller than the radio waves that transmit cell phone and Wi-Fi signals, and since this portion of the airwaves is not as widely used as others, Facebook can use it to send much larger about of data. Others have used millimeter wave systems to send data between two distant points, like ground stations and satellites. But it’s always been a bulky, power-hungry setup. Facebook is pushing toward the kind of lightweight, energy efficient applications that have only been contemplated in scientific journals and maybe some secret government labs. “There are a lot of theoretical papers on this,” says Robert Heath, an electrical engineer at the University of Austin. “But there aren’t a lot of people who are actually making it work.” Heath says that Facebook’s millimeter wave team rivals any in the world.
The team has already proven that its millimeter wave system can trade data between two fixed points eight miles away from each other at a rate of nearly 20 Gigabits per second. That’s roughly 400 times the speed of your home Internet connection. Facebook believes this is a world record for equipment that’s so light and consumes so little power.
But achieving that rate in connection with a moving target—and pushing that data rate even faster—are enormously difficult. First the team has to calibrate the precise orientation of the ground antenna using the position of the Sun. Then the flying antenna must lock in on that ground antenna—while hurtling through the sky at more than a hundred miles an hour, eight miles away, connecting with nothing a millimeter wave beam. It’s is like trying to thread a moving needle from the other side of a room.
During the test flight, hours go by: red circle zooming past tiny spec, locking briefly, falling away, the team anxiously waiting for a consistent lock.
Made for Sharing
Tiwari and his team built their first prototype of their wireless system during a Facebook hackathon in January. At the time, engineers at the Woodland Hills lab were already working on laser communication technologies for use with Aquila, but they knew lasers wouldn’t work for getting data from the ground to the sky. For that, they needed a signal that could pass through clouds. That’s where they decided millimeter wave technology would come in. (The lasers will still be handy for sending signals between Aquilas in the stratosphere.)
Nine months after the hackathon, they were ready for the test flight.
Three hours before he heads up to the roof in Woodland Hills, Tiwari—who worked on millimeter wave tech for Darpa before he came to Facebook—is on his knees inside a hangar at Whiteman Airport, about 20 miles away, helping install a millimeter wave antenna on the underside of the Cessna, a plane typically used for aerial photography. The engineers can’t get one of the screws in, and until they do, the plane can’t take off.
The antenna fits into a hole in the belly of the plane. It’s a dish little bigger than your hand, and it sits on a two-axis gimbal that controls its direction. The bulk of the apparatus is made of carbon-fiber to reduce weight. It’s about the size of a basketball, covered with a plastic globe to protect it from the wind, and wires connect the antenna to a Dell laptop and some other equipment in the cockpit.
The antenna uses the E band, radio frequencies from about 60-GHz to 90-GHz on the electromagnetic spectrum. Because this band includes so much available spectrum, it’s ideal for what’s called “backhaul”—moving enormous data between two different wireless hubs. Plus, says Hamid Hemmati, who oversees the Woodland Hills lab, the E band isn’t as heavily licensed as other frequencies. That means it’s easier to actually deploy this technology—and that, too, is important for the company’s grand plan.
Rather than keep all this technology proprietary, Facebook is open sourcing all its Aquila designs, including its millimeter wave hardware. Meanwhile, other Facebook teams are building terrestrial antennas that can help blanket cities with cellular and Wi-Fi signals or beam a signal from a city to distant rural areas. The company will share all these designs freely through an organization the company calls the Telecom Infrastructure Project.
Ultimately, Facebook’s bosses want telecoms, governments, and emergency response organizations to build and deploy this equipment themselves—and improve on it. And for that prospect to become real, Facebook’s Internet drone must be easy to build and operate at low cost. That means sending as much data as possible from an ultra-lightweight payload driven only by power Aquila gets from the sun. “It must be cheaper than any other means of getting communications out there,” Tiwari says.
And, of course, it also has to work. After few hours of trying, the Woodland Hills team is finally able to make a consistent connection between the antenna on the lab roof and the flying Cessna. The connection is nowhere near 20 Gigabits a second, but it’s an important step nonetheless. “We want to get this into people’s hands as soon as possible,” Hemmati says. Next step: Replacing that Cessna with one of Facebook’s giant, v-shaped Aquilas.