“Combing” Through Light May Give Us Faster, More Powerful Internet

A lab experiment used a device called a frequency comb to send fiber optic data a record-breaking distance with no signal loss

This online article is brought to you by fiber optic cables. (Andrew Brookes/Corbis)

Fiber optic cables make up the backbone of modern communications, carrying data and phone calls across countries and under oceans. But an ever-expanding demand for data—from streaming movies to Internet searches—is putting pressure on that network, because there are limits to how much data can be pushed through the cables before the signal degrades, and new cables are expensive to build.

Now a team at the University of California, San Diego, might have a solution by borrowing a technique used in other fields as a measurement tool: the frequency comb. These laser-based devices allowed the team to remove distortions that would usually appear before the signal got to the end of a cable. The researchers sent data further than ever before—7,456 miles—without the need to boost the signal along the way.

If their experimental technique holds up in the real world, fiber optic cables would need fewer expensive repeaters to keep signals strong. In addition, greater signal stability within a data stream would mean more channels could be stuffed into a single transmission. Right now, a fundamental trade-off in fiber optics is the more data you want to transmit, the shorter the distance you can send it.

Fiber optic signals are simply encoded light, either generated by a laser or an LED. This light travels down thin glass cables, reflecting off their inside surfaces until it comes out the other end. Just like radio broadcasts, a laser beam will have a certain bandwidth, or range of frequencies, it covers, and a typical strand of fiber optic cable can carry more than one bandwidth channel.

But the signals can't travel forever and still be decoded due to so-called non-linear effects, specifically the Kerr effect. For fiber optics to work, the light inside the fiber has to refract, or bend, a certain amount as it travels. But electric fields will alter how much glass bends light, and light itself generates a small electric field. The change in refraction means that there are small changes in the wavelength of the transmitted signal. In addition, there are small irregularities in the glass of the fiber, which isn't an absolutely perfect reflector.

The small wavelength changes, called jitter, add up and cause cross-talk between the channels. The jitter appears random because a fiber optic transmission carries dozens of channels, and the effect on each channel is a bit different. Since the Kerr effect is non-linear, mathematically speaking, if there's more than one channel you can't just subtract it—the calculation is much more complex and nearly impossible for today's signal processing equipment. That makes the jitters hard to predict and correct.

"We realized that the fuzziness, ever so slight, causes the whole thing to appear as though it is not deterministic," says Nikola Alic, a research scientist from the Qualcomm Institute at UCSD and one of the leaders of the experimental work.

In the current fiber optics setup, channel frequencies have to be far enough apart that jitter and other noise effects don’t make them overlap. Also, because the jitter increases with distance, adding more power to the signal only amplifies the noise. The only way to deal with it is to put costly devices called repeaters on the cable to regenerate the signal and clean up the noise—a typical transatlantic cable has repeaters installed every 600 miles or so, Alic said, and you need one for each channel.

The UCSD researchers wondered whether they could find a way to make jitter look less random. If they knew exactly how much the wavelength of light in every channel would change, then they could compensate for it when the signal got to a receiver. That's where the frequency comb came in. Alic says the idea came to him after years of working in related fields with light. “It was sort of a moment of clarity,” he says. A frequency comb is a device that generates laser light at lots of very specific wavelengths. The output looks like a comb, with each "tooth" at a given frequency and each frequency an exact multiple of the adjacent ones. The combs are used in building atomic clocks, in astronomy and even in medical research.

Alic and his colleagues decided to find out what would happen if they used a frequency comb to calibrate the outgoing fiber optic signals. He likens it to a conductor tuning an orchestra. “Think of the conductor using a tuning fork to tell everyone what the middle A is,” he says. The team built simplified fiber optic systems with three and five channels. When they used the comb to calibrate the outgoing signal wavelengths, they still found jitter, but this time, all the channels were jittering in the same way. That regularity allowed the signal to be decoded and sent at a record distance with no repeaters. “It makes the process deterministic,” says Alic, whose team reports the results this week in Science.  

Sethumadhavan Chandrasekhar, distinguished member of the technical staff at the global telecom company Alcatel-Lucent, is one of many scientists who have been working on the fiber optic jitter problem for a number of years. His published work involves transmitting phase-conjugated signals—two signals that are exactly 180 degrees out of phase with each other. This setup means that any of the nonlinear effects that cause noise would be canceled out.

The UCSD work is important, but it isn't a complete solution yet, Chandrasekhar says. "What is missing is that most systems now have dual polarization," he says, meaning that the systems boost capacity by sending light signals that are polarized differently. "Most systems today transmit information in the two polarization states of light, and the UCSD team needs to demonstrate that their technique works as well under such a transmission scenario," he says.

Alic says that the team's next set of experiments will address that very issue. So far, they think this technique can be adapted for real-world use, though it will require building and deploying new hardware, which will take time. Either way, increasing the reach of signals will allow for a much more aggressive build-out, yielding more data and more distance without worries over signal loss. "There's no reason to be afraid anymore," he says.

About Jesse Emspak

Jesse Emspak is a freelance science writer based in New York City. His work has appeared in Scientific American, The Economist, New Scientist, Livescience.com, The Christian Science Monitor and Astronomy Magazine.

Read more from this author |

Comment on this Story

comments powered by Disqus