The fifth dimension

Commercial applications for space-division multiplexing could arrive sooner than we think.


Commercial applications for space-division multiplexing could arrive sooner than we think.

Optical communications technology has made phenomenal progress in recent decades, with new ways to boost the transmission capacity of a single optical fiber. Almost every available degree of freedom has been exploited - modern coherent optical systems multiplex signals using wavelength, polarization, amplitude, and phase. Each new degree of freedom has provided an exponential increase in transmission capacity.

Future gains will not be so readily achieved, however. There is widespread recognition that the base of installed optical fiber is close to reaching the theoretical transmission limits described by Shannon's theorem.

"Interface cards have made this huge jump from 10G to 100G, and they've been able to make this jump in capacity without sacrificing reach or performance. But if you look forward, you will never see this kind of step again," says Maxim Kuschnerov, DWDM product manager at Coriant. "We have reached a plateau [in transmission capacity]."

Yet, one potentially massive source of additional capacity has been left virtually untapped. Space-division multiplexing (SDM) uses parallel spatial channels to transmit multiple independent optical signals simultaneously.

A variety of different modes can be supported via few-mode fiber.

"Optical fibers can easily support hundreds of spatial modes, but today's commercial systems - singlemode or multimode - make no attempt to use these as parallel channels," notes Prof. David Richardson, deputy director of the Optoelectronics Research Centre at the University of Southampton in the U.K.

The notion of increasing fiber capacity with SDM is not new, he says; in fact it's almost as old as fiber-optic communications itself, with the fabrication of fibers containing multiple cores reported as far back as 1979. But interest in the technology has really picked up over the last few years, and now post-deadline paper sessions at the major optical conferences are dominated by new results obtained using SDM.

New fibers

Unfortunately, while conceptually simple, SDM turns out to be extremely challenging in practice. SDM will require optical vendors to design completely new transmission fibers as well as the associated optical components to multiplex, amplify, and switch the spatial modes and electronic circuitry to untangle the information they contain. Placing optical signals in close proximity also generates crosstalk that must be compensated.

Packaged photonic lanterns, which couple power from N input fibers to N fiber modes, are commercially available.

It almost goes without saying that carriers are reluctant to put new fiber in the ground. But they recognize that they need to be prepared for the day when the capacity of their existing plant is exhausted, even if it seems far away right now. "There is probably enough fiber in the ground to deal with capacity upgrades until 2025, but when you look at what's required [in terms of technology development], that's a very short time," says Ian Giles, CEO of Phoenix Photonics in the U.K. and project manager for MODE-GAP, a collaborative European project investigating the potential of SDM.

Several different approaches have been proposed and investigated. Multi-core fiber (MCF) contains multiple independent optical cores inside a single cladding - up to 19 have been demonstrated. Few-mode fibers (FMFs) are designed to guide a restricted number of spatial modes, typically 6-12 distinct modes, inside a single core slightly larger than that of a typical singlemode fiber. Photonic crystal fibers (PCFs) guide light through a hollow core, which has additional properties of interest, including ultra-low optical nonlinearity, lower latency, and the potential for lower losses than solid-core fibers.

Researchers have already achieved results that establish the potential for high capacity transmission via SDM. In Japan, the National Institute for Communications Technologies (NICT) has sponsored collaborative research into SDM based on MCF. In 2012, NTT and its partners demonstrated petabit optical transmission over a single optical fiber at a distance of 52.4 km. In the experiment, the researchers generated polarization-multiplexed 32-QAM signals at 380 Gbps on each of 222 wavelengths to yield a transmission capacity of 84.5 Tbps for each core, for an aggregate capacity of 1.01 Pbps. The fiber contained 12 cores arranged in a nearly concentric pattern to reduce crosstalk by reducing the number of nearest neighbors.

Another result worthy of note came out of the U.S. later that same year. Researchers from NEC Labs in Princeton, NJ, and Corning's Sullivan Park Research Center in Corning, NY, claimed a new record, achieving 1.05-Pbps transmission over an MCF that contained 12 singlemode and two few-mode cores. It's interesting to see that different types of spatial guidance can be combined in the same fiber, but it does raise the question of whether a common approach across the industry would enable more rapid progress.

In Europe, SDM research is being pushed forward by the MODE-GAP project, which favors an approach based on multiple modes and PCF (also known as photonic bandgap fiber). "The problem with [spatially overlapping modes in] solid-core fiber is the nonlinear effects, whereas with photonic bandgap fiber the limits are much higher. You can increase the capacity of the fiber just by pumping more power down it," explains Giles.

The project team has successfully transmitted data of 57.6 Tbps (73.7 Tbps including error-correction overheads) over hollow-core fiber supporting several spatial modes, which has now been certified as a Guinness World Record. The consortium detailed the feat in a post-deadline paper at OFC 2014 in San Francisco last March.

Cross-sectional view of a hollow- core fiber.

In addition to Phoenix Photonics, project members include Southampton University (where PCF is manufactured in research quantities), the COBRA Institute at the Technische Universiteit Eindhoven, Tyndall National Institute at Ireland's University College Cork, OFS Fitel Denmark, Eblana Photonics, ESPCI ParisTech, Aston University in Birmingham, U.K., as well as the German business of optical systems vendor Coriant. The systems house has provided a testbed for demonstrations and a perspective on systems development. To help define the customer requirements of an optical transport system based on SDM, Coriant talked to more than 20 customers from around the world, both traditional telecom operators and cloud content providers, Kuschnerov asserts.

Cost of SDM

Reaching higher capacities isn't the only motivation to develop SDM, of course; the metric that carriers really care about is cost per bit. To help inform this discussion before such systems have actually been built, Coriant also modeled the cost of building from scratch an optical transport system based on 10 parallel fibers and compared it to the cost of building a system of equivalent capacity running over a single fiber using 10 spatial modes.

Roughly half the system cost is in the optical interfaces, which is difficult to reduce even with SDM. But overall the SDM system is up to about 20% cheaper to build because the amplifiers and ROADMs are more integrated and more efficient, and therefore less expensive. "It doesn't strike you at first as revolutionary," says Kuschnerov. "You have to argue for benefits of SDM in a different way. It's not a pure capex discussion; it is also about operational simplicity, power savings in the amplifiers, and the lowest possible footprint. This conversation is not as obvious to the customer."

1411lw Feat02 Mech Lpg 2
The variable LP01-LP11 mode coupler pictured here creates a periodic mode coupling in dual-mode fibers at the beat length between the two modes.

Coriant found that telecom operators had a different view from the cloud content providers. Telecom operators were generally reluctant to think about new fiber types, according to Kuschnerov. "Carriers have had bad experience with these kinds of choices in the past," he explains. "They were told that lower dispersion in the fiber was good at 10G, but then new coherent technology came along, and it was actually good to have high dispersion because it reduces nonlinearities. People installed a lot of fiber that turned out to be very bad for coherent and they never saw this coming."

Telecom operators also said that once the full ecosystem of components and systems has been developed and thoroughly tested, they would need a smooth transition to the new technology. In the ideal scenario, components and systems designed for SDM will also be compatible with standard transmission systems based on singlemode fiber and could be installed as part of a gradual upgrade.

In response, Coriant set up a field trial with Austrian cable operator A1 to show that it was possible to upgrade individual links in the network. The field trial, carried out in 2013, used 100G and 200G optical carriers running over a 660-km route between Salzburg and Vienna that included a span containing FMF and a multimode amplifier. "We showed A1 that fitting in new technology is not disruptive. The customer can begin to extract the benefits of the new technology without disturbing the equipment in the field," says Kuschnerov.

Cloud content providers, on the other hand, are less tied to legacy infrastructure as they build out new data-center networks in response to spiraling levels of data traffic. With shorter network upgrade cycles and tremendous pressure to deliver connectivity at the lowest cost, cloud providers are also more receptive to new technologies, Kuschnerov reports.

Cloud providers perked up when they heard about hybrid-core fiber (HCF) because it delivers a 30% improvement in latency compared to solid-core fiber, simply because light travels faster in air. "The connection length inside a data center is typically between 500 m and 2 km. You could save up to 3 msec over this distance," reasons Kuschnerov. That would boost response times and overall performance inside the data center, an advantage to any data-center operator, but especially those with customers in the financial sector. This is a niche application that could use SDM technology "as soon as it is ready," he adds.

There's still plenty of work to do, but progress has been steady. Over the last 12 months, the MODE-GAP project has increased the transmission distance over HCF by a factor of 10. Researchers at OFS Laboratories and Coriant recently demonstrated a 100G linecard running over a 2.75-km length of HCF, which is long enough to cover every link inside most data centers. The work will be presented as an invited paper at OFC 2015. "This experiment has put HCF on our product roadmaps; it could become real as soon as three years from now," concludes Kuschnerov.

More in High-Speed Networks