A team at the Optical Networks Group at University College London has sent 178 terabits per second through a commercial singlemode optical fiber that has been on the market since 2007. It's a record for the standard singlemode fiber widely used in today's networks, and twice the data rate of any system now in use. The key to their success was transmitting it across a spectral range of 16.8 terahertz, more than double the the broadest range in commercial use.
The goal is to expand the capacity of today's installed fiber network to serve the relentless demand for more bandwidth for Zoom meetings, streaming video, and cloud computing. Digging holes in the ground to lay new fiber-optic cables can run over $500,000 a kilometer in metropolitan areas, so upgrading transmission of fibers already in the ground by installing new optical transmitters, amplifiers, and receivers could save serious money. But it will require a new generation of optoelectronic technology.
A new generation of fibers have been in development for the past few years that promise higher capacity by carrying signals on multiple paths through single fibers. Called spatial division multiplexing, the idea has been demonstrated in fibers with multiple cores, multiple modes through individual cores, or combining multiple modes in multiple fibers. It's demonstrated record capacity for single fibers, but the technology is immature and would require the expensive laying of new fibers. Boosting capacity of fibers already in the ground would be faster and cheaper. Moreover, many installed fibers remain dark, carrying no traffic, or transmitting on only a few of the roughly 100 available wavelengths, making them a hot commodity for data networks.
"The fundamental issue is how much bandwidth we can get" through installed fibers, says Lidia Galdino, a University College lecturer who leads a team including engineers from equipment maker Xtera and Japanese telecomm firm KDDI. For a baseline they tested Corning Inc.'s SMF-28 ULL (ultra-low-loss) fiber, which has been on the market since 2007. With a pure silica core, its attenuation is specified at no more than 0.17 dB/km at the 1550-nanometer minimum-loss wavelength, close to the theoretical limit. It can carry 100-gigabit/second signals more than a thousand kilometers through a series of amplifiers spaced every 125 km.
Generally, such long-haul fiber systems operate in the C band of wavelengths from 1530 to 1565 nm. A few also operate in the L band from 1568 to 1605 nm, most notably the world's highest-capacity submarine cable, the 13,000-km Pacific Light Cable, with nominal capacity at 24,000 gigabits per second on each of six fiber pairs. Both bands use well-developed erbium-doped fiber amplifiers, but that's about the limit of their spectral range.
To cover a broader spectral range, UCL added the largely unused wavelengths of 1484 to 1520 nm in the shorter-wavelength S band. That required new amplifiers that used thulium to amplify those wavelengths. Because only two thulium amplifiers were available, they also added Raman-effect fiber amplifiers to balance gain across that band. They also used inexpensive semiconductor optical amplifiers to boost signals reaching the receiver after passing through 40 km of fiber.
Another key to success is format. "We encoded the light in the best possible way" of geometric coding quadrature amplitude modulation (QAM) format to take advantage of differences in signal quality between bands. "Usually commercial systems use 64 points, but we went to 1024 [QAM levels]...an amazing achievement," for the best quality signals, Gandino said.
This experiment, reported in IEEE Photonics Technology Letters, is only the first in a planned series. Their results are close to the Shannon limit on communication rates imposed by noise in the channel. The next step, she says, will be buying more optical amplifiers so they can extend transmission beyond 40 km.
"This is fundamental research on the maximum capacity per channel," Galdino says. The goal is to find limits, rather than to design new equipment. Their complex system used more than US$2.6 million of equipment, including multiple types of amplifiers and modulation schemes. It's a testbed, not optimized for cost, performance, or reliability, but for experimental flexibility. Industry will face the challenge of developing detectors, receivers, amplifiers and high-quality lasers on new wavelengths, which they have already started. If they succeed, a single fiber pair will be able to carry enough video for all 50 million school-age children in the US to be on two Zoom video channels at once.
This article appears in the October 2020 print issue as “Same Old Fibers, Record-Breaking Speeds.”
Jeff Hecht writes about lasers, optics, fiber optics, electronics, and communications. Trained in engineering and a life senior member of IEEE, he enjoys figuring out how laser, optical, and electronic systems work and explaining their applications and challenges. At the moment, he’s exploring the challenges of integrating lidars, cameras, and other sensing systems with artificial intelligence in self-driving cars. He has chronicled the histories of laser weapons and fiber-optic communications and written tutorial books on lasers and fiber optics.