Blue-Sky Thinking

Cyrus Behroozi wants to connect the whole world to the Internet. 

“Two-thirds of the world’s population still doesn’t have access,” says Behroozi, an engineer with Google X, the Internet giant’s experimental division. Although it’s easy to think of the Internet as a luxury, he says, it’s now inextricably tied to economic development. 

Considering that Google X is most widely known for engineering the driverless car, its solution to global connectivity might seem charmingly low-tech—balloons. But these aren’t everyday balloons. Behroozi leads the network engineering for Project Loon, an ambitious experiment by Google X that’s focused on creating a global wireless network of balloons floating around the world 12.5 miles (20 kilometers) aboveground, within the  stratosphere—twice the elevation flown by commercial airlines.

We spoke with Behroozi to hear about his path from Caltech to Google’s lofty experiment.

[VIDEO: Scroll down to see a short produced by Google of the balloons in action, along with an interview with Behroozi.]


Tell us about your time at Caltech.

My parents were physicists, and I had grown up with a bit of hero worship for Richard Feynman. So when I arrived at Caltech, I felt that I had found “my people.” I fell under the wing of physics professor Ken Libbrecht (BS ’90), who at the time was very interested in Bose-Einstein condensates [cooling subatomic particles to near absolute zero so that they coalesce into larger structures, in some cases observable on a macroscopic scale]. Being able to trap and observe atoms has a very obvious “cool factor,” so I continued to pursue it after graduation. A couple of years later, I was part of a research team at Harvard that was able to slow down light traveling through a Bose-Einstein condensate. That project received a lot of press and really seemed to capture the imagination of the public. Jay Leno even made a joke about it: “Researchers announced that they were able to slow light. You know how they did it? They took a laser and aimed it through the post office.”

What exactly are Google X and Project Loon?

Google X is a research lab of Google, but it’s not just focused on problems related to the company’s core business. We look for big problems that might require radical solutions and breakthrough technologies. The self-driving car is perhaps our best-known project and a great example. The problem: People are generally terrible drivers; we cause traffic jams and get into accidents. A radical solution might be to teach computers to drive. Technology exists that might make that possible, but it’s an enormous challenge to implement it. So—big problem, radical solution, breakthrough technology.

Project Loon is an attempt to solve the problem of Internet access around the world. Right now, two-thirds of the world’s population does not have the ability to use the Internet. You might consider it a luxury, but access is now so closely correlated with economic development. There is an emerging global digital divide. People in developing countries could be greatly helped by gaining access.

So that's the problem: global Internet access. The radical approach and the breakthrough technology are doing it by balloon—and not just stationary balloons hovering over one location and providing access to a small group willing to pay for it. Most likely, that approach actually turns out not to be technologically feasible. Instead, the idea is to let the balloons float freely, travel with the winds in the stratosphere, and have just enough of them that as some drift out of range, new balloons come from the other direction to take their places.

Why balloons? Why not use satellites?

Well, satellites are also a great solution, but they're extremely expensive to launch. There's a long time delay between when a satellite is proposed, when it's built, when you can get a launch window, and the duration that it has to last in orbit to sort of pay for itself. You might be developing technology for a satellite that launches five years from now. It has to last for the next 10 years to become economically viable. So you're dealing with technology that might be 10 or 15 years old by the time it really gets used. 

With balloons, we have more flexibility. We can make them cheaply. We can launch them cheaply. We can refresh and iterate the technology on a month-by-month basis. So it's sort of low-stakes aerospace. We get the altitude and coverage similar to a satellite and gain the ability to iterate quickly—all at a much lower cost. 

What are the challenges involved?

Behroozi on Google's campus in Mountain View, California, with a prototype of a balloon's communications array.

Behroozi on Google's campus in Mountain View, California, with a prototype of a balloon's communications array.

It’s a terrific challenge of engineering. There's a big jump between making a balloon that can last one day and one that can last 100. Going through the day-and-night cycle is treacherous, requiring what's called a super-pressure balloon. Then, there's the navigation. Winds within the stratosphere tend to travel in one direction, which we can take advantage of to navigate. The balloon needs to be able to change altitude to catch a current headed in the right direction. 

My responsibility is the communication, connecting the balloons to the end users on the ground. As balloons drift in and out of range, we need the connectivity to feel seamless. We also need to link the balloons to one another, back to ground stations, and the Internet. One of the first basic challenges we faced was the fact that most transmitters are polarized, meaning the signal transmits in only one direction. But balloons rotate. So we designed special antennae that provide a signal no matter which way the balloon is oriented.

So to make this all work, we need: balloons capable of staying aloft for 100 days, altitude-control systems, an “air-traffic control” system that can predict weather models, and a network to tie them all together.

What stage of the project are you in?

Well, we've already done pilot testing in New Zealand and Brazil. We've been working furiously and growing the team since then. We're launching and flying balloons continuously right now. We're definitely “all in” to make this work.

[Update: Since this interview was conducted, Astro Teller, the head of the Google X lab, announced that Project Loon had logged more than 1.2 million miles (2 million kilometers) in testing and was on track to have a “semi-permanent” ring of balloons in the Southern Hemisphere within the next year or so.

We really look for what we call ‘T-shaped people,’ — people with really diverse backgrounds, but who also have very deep expertise in one area. That describes most Techers.

How did your training at Caltech prepare you for this career?

I really think that Techers fit very well within Google X. We have a bunch here right now. We really look for what we call T-shaped people—people with really diverse backgrounds (the top of the T) who also have very deep expertise in one area (the stem). Because of the way that Caltech trains us, that describes most Techers. I also believe that the collaborative culture at Caltech is really helpful. If I have a hard problem, my first reaction is to go find somebody else who might be an expert in it rather than attack it myself.

As an engineer, I’m very pleased to be here. Just about every project in Google X is something that relates somehow to the physical world. We have generous resources to tackle them and an environment that encourages risk—even if we end up failing.

Professor Libbrecht always impressed me with his ability to jump between projects and follow his interests. He often joked, “If at first you don’t succeed, no need to make a fool of yourself.” I would encourage people to be adventurous and explore many paths in order to find the ones that suit them best.