100G Ethernet

AlcaLu Climbs to 16.4 Tbit/s

SAN DIEGO -- OFC/NFOEC -- It's not quite like winning the Super Bowl, but Alcatel-Lucent (NYSE: ALU) appears to have claimed the championship for most 100 Gbit/s signals jammed onto a fiber.

In a post-deadline paper presented here yesterday, AlcaLu described how it sent a 16.4-Tbit/s transmission over 2,550 kilometers of fiber in the lab. It was one of four post-deadline papers the company published. (See AlcaLu Claims Optical Records.)

An OFC/NFOEC post-deadline paper is a big deal. The technical sessions here are filled with new experimental results, but companies sometimes save their niftiest work for the post-deadline papers, which traditionally get published and presented on the conference's penultimate day.

Last year's post-deadline crop included an experiment involving 10 wavelengths of 111 Gbit/s apiece, sent down 2,400 km of fiber. That one was presented by CoreOptics Inc. , the Eindhoven University of Technology , and Siemens AG (NYSE: SI; Frankfurt: SIE). (See CoreOptics Demos 100G.)

AlcaLu's 16.4-Tbit/s result is a kind of sequel to that paper. Both experiments used a modulation scheme called (deep breath) polarization division multiplexed quadrature phase-shift keying (PDM-QPSK). But AlcaLu took the idea further by packing many more wavelengths, using up the entire C and L bands.

PDM-QPSK is designed to be robust to polarization mode dispersion (PMD), one of the optical effects that causes high-speed signals to degrade as they traverse long distances. Companies often cite PMD as one of the bugaboos that arises with serial 100-Gbit/s transmission.

Post-deadline papers don't represent products that are coming out any time soon. But AlcaLu's results -- and CoreOptics's a year ago -- show that it's feasible to send 100 Gbit/s signals down the infrastructure that was built for 10 Gbit/s. That's considered a critical factor in developing 100-Gbit/s Ethernet.

One of the biggest questions around serial 100-Gbit/s transmission is the modulation scheme. Most companies prefer some kind of "quad" format, where data gets sent four bits at a time. That would turn the 100 Gbit/s problem into an easier 25 Gbit/s one.

Differential quadrature phase shift keying (DQPSK) is the modulation scheme most talked about as a 100 Gbit/s candidate.

— Craig Matsumoto, West Coast Editor, Light Reading

Pete Baldwin 12/5/2012 | 3:46:30 PM
re: AlcaLu Climbs to 16.4 Tbit/s Um..... Not to crash the mood, but this came out of Alcatel France, with help from Kylia.

But the U.S. half, which I'm presuming means Bell Labs, did have three other postdeadlines:

hyperunner 12/5/2012 | 3:46:30 PM
re: AlcaLu Climbs to 16.4 Tbit/s This is a great achievement and it shows there's still innovation coming out of Bell Labs.

great job!

trzwuip 12/5/2012 | 3:46:29 PM
re: AlcaLu Climbs to 16.4 Tbit/s I didn't know there were any employees left at what is called "bell labs." Bell Labs is a great part of our history, but it is just that for now. When it comes to putting the last of the Lucent people in charge, you "can't fix stupid" as documented by Alcatel's stock since the buy and Lucent leadership. This merger was just plain old dumb!
desiEngineer 12/5/2012 | 3:46:29 PM
re: AlcaLu Climbs to 16.4 Tbit/s I like this headline on the Kylia website, following the News link:

Nov 2007 : Kylia toots Ultra DWDM Mux/Demux

Looks like the French are tooting more than just their horn :-)

hyperunner 12/5/2012 | 3:46:27 PM
re: AlcaLu Climbs to 16.4 Tbit/s Craig is right. This is a French team. My buddy read the paper and it seems this is very much a science experiment. They only use one signal source and "copy" it to all the other channels. I guess that makes sense if you think about it - it'd be impractical to have hundreds of separate transponder cards feeding one fiber. The space and power that would take up would be crazy.

I wouldn't be so hard on Bell Labs. They've got some smart people there and a great track record of innovation. It's just that the innovation never gets turned into anything commercially useful!

boozon 12/5/2012 | 3:46:21 PM
re: AlcaLu Climbs to 16.4 Tbit/s you're right: this is a french team and it is a science experiment.
Still, it is a bloody good hero experiment!
The real question is whether this will materialise in a product (and hence dollars, hum...euros I should say) in a not too far future.
Unfortunately for them, Alcatel seems to have a big track record of hero experiments that never come out of the science books (see the optical regens for example)...
jaclermont 12/5/2012 | 3:40:41 PM
re: AlcaLu Climbs to 16.4 Tbit/s I updated the Kylia news page.
Sign In