Bell Labs Boasts Optical Eureka Moment
With communications service providers worldwide wondering how to build networks that can cope with the data and video traffic loads of the future, researchers at Bell Labs believe they have solved the major challenge facing optical network planners with space-division multiplexed optical multiple-input-multiple-output (MIMO-SDM). (See Bell Labs Claims Optical Capacity Breakthrough.)
(Try saying that after four pints of Murray Hill Moosehead Beer...)
According to Bell Labs , the research arm of Alcatel-Lucent (NYSE: ALU) that is soon to be folded into the Nokia Corp. (NYSE: NOK) empire, this R&D breakthrough could boost the potential capacity of a single fiber from today's 20 Tbit/s to 1 Petabit-per-second (1,000 Tbit/s). (See AlcaLu's Weldon Lands Nokia CTO Role.)
Industry projections suggest such capacity expansion will be needed sooner rather than later: Studies from the likes of Cisco Systems Inc. (Nasdaq: CSCO), Ericsson AB (Nasdaq: ERIC) and Bell Labs offer a sobering view of the demands likely to be placed on access and transport networks within just a few years. (See Bell Labs Sees 560% Boost in Metro Data Traffic, Cisco's Visual Networking Index, Cisco VNI: Make Way for HD & 4K Video and this story that includes Ericsson Mobility Report forecasts.)
That's one of the reasons why the smart folks at Bell Labs have been looking for the next big thing in optical transport -- and now they think they've found it with MIMO-SDM, with which the researchers aim to overcome the Shannon Limit that sets a threshold for the maximum throughput on a single optical fiber.
"There's a big problem with the upcoming capacity crunch in optical. Current commercial transport networks are going to peak at 20 Tbit/s per fiber and the industry needs a way to add capacity without also increasing the cost and power consumption per bit," notes Peter Winzer, head of the Optical Transmission Systems and Networks Research Department at Bell Labs.
"So we need integration [at the optical component level]… multiprocessor integration is the way forward. But close integration also creates crosstalk and we have found a way to counter that noise using real-time MIMO processing."
Winzer believes this is the first time that multi-path MIMO, beyond the 2x2 MIMO technology used in coherent optical technology, has been shown. "This is a lightbulb moment for the industry… this is the next step on from coherent," he states. "It's hard to say exactly what impact this will have on network capacity but it will be orders of magnitude greater than now. This is the only way the industry can achieve multiple gains and maintain the kind of gains in price and energy use erosion per bit that we have seen in optical in recent years. It will probably start in the metro, for example in data center interconnection scenarios but this will also have an impact in long-haul networks," he told Light Reading, echoing the view of parent vendor Alcatel-Lucent that data center interconnect needs plenty of attention. (See Data Center Interconnects in Desperate Need of SDN – AlcaLu .)
Currently, though, the Bell Labs team is just at the prototype stage, but it can still inform the components vendors that can turn this from lab tests into commercial reality, says Winzer. "Once the components companies," such as Lumentum (formerly JDSU), Finisar Corp. (Nasdaq: FNSR) and Oclaro Inc. (Nasdaq: OCLR), realize the potential, "they will say 'hallelujah'."
The big question for the industry, says the Bell Labs man, is: "How much can the components companies take cost out through integration? What we need to end up with is arrays of everything -- modulators, receivers, amplifiers... If a company such as Oclaro built an array of modulators, integrating tens of them into each package, packing as many as they can on one wafer... the closer the modulators are to each other, the greater the crosstalk. Currently crosstalk would make the product unusable. But with MIMO processing it can work. This development is opening up a whole new design space for components manufacturers," says Winzer. "In ten years, every optical system will have a MIMO processor that has more than 2x2 inside," he believes.
— Ray Le Maistre, , Editor-in-Chief, Light Reading