AlcaLu Can Do 400G Too

LOS ANGELES -- OFC/NFOEC 2012 -- Alcatel-Lucent (NYSE: ALU) is announcing a new chip for 100Gbit/s and 400Gbit/s optical transmission, and it's eerily similar to what Ciena Corp. (NYSE: CIEN) announced five days ago. (See Ciena Pushes Ahead to 400G.)

AlcaLu has announced the photonic service engine (PSE), an inhouse-designed chip for the vendor's 1830 PSS WDM optical platform. It lowers the power of 100Gbit/s links and also puts AlcaLu in a position to offer a 400Gbit/s wavelength (built up from two 200Gbit/s channels) by the end of next year.

AlcaLu and Ciena's stories match almost point for point. Both companies are using a transmit-side digital signal processor (DSP), a new feature that lets them pack wavelengths more closely together. This way, both companies say, they could cram more wavelengths onto a fiber. While the specifications violate the DWDM grid set by the International Telecommunication Union (ITU) , the capabilities could be useful with the advent of flexible-grid reconfigurable optical add/drop multiplexers (ROADMs).

Both companies say this wavelength-packing ability lets them sandwich two 200Gbit/s carriers close enough together to create commercially viable 400Gbit/s transmission. The 400Gbit/s wavelength would fit one 100GHz spot on the ITU grid.

One difference: AlcaLu says the PSE's 100Gbit/s reach is 3,000 km. Ciena talked about WaveLogic 3 reaching 2,500 km. AlcaLu is also talking about using dynamic demodulation (an internally developed technology) to get that kind of reach.

What AlcaLu is announcing Tuesday is the PSE chip itself. The 1830 PSS will need a software upgrade to use line cards that include the PSE, and that upgrade is due in the second half of 2012.

Why this matters
Vendors will spend the next few years trying to squeeze cost and power out of 100Gbit/s, and the AlcaLu and Ciena announcements suggest that everyone will be trying similar methods. (Or maybe it just means someone needs to stop copying someone else's test papers ... we're not suggesting who.)

Both companies say the market isn't ready for 400Gbit/s yet, but the announcements do show that 400Gbit/s is attainable using current technology.

At least one merchant chip vendor is aiming for 400Gbit/s in a similar way. In customer meetings at OFC/NFOEC, ClariPhy Communications Inc. -- which sells chips for 40Gbit/s modules -- will talk about producing a 200Gbit/s coherent chip sometime in 2013, Chief Strategy Officer Paul Voois tells Light Reading. That would be the basis for other companies to produce 400Gbit/s products similar to those now being pitched by AlcaLu and Ciena.

For more
Keep up with our coverage from the Los Angeles Convention Center at the Light Reading OFC/NFOEC Show Site. And here's some more about 400Gbit/s:

— Craig Matsumoto, Managing Editor, Light Reading

digits 12/5/2012 | 5:40:30 PM
re: AlcaLu Can Do 400G Too

As the industry eases itself into 100Gbit/s is there any sense when any network operator might want to look at deploymernt of 400Gbit/s wavelengths in a production network?

viper99 12/5/2012 | 5:40:25 PM
re: AlcaLu Can Do 400G Too


I think its easy to not look past the headline of "400G" -- what both Ciena and ALU really have announced is denser Nx100G coherent solutions, in otherwords allowing for more 100G waves over the same fiber to maximize fiber utilization and deliver the most efficiency.  There is technically no 1x400G waves here.  

400G is just a catchy marketing headline -- more sexy than 4x100G.  

To scale to 400G (really its 4x100G in a single channel spacing is what they both are talking about), they are trading off spectral efficiency and ultimately reach in order to get 4x100G -- which might be appropriate in Metro applications.  

IJD 12/5/2012 | 5:40:23 PM
re: AlcaLu Can Do 400G Too

From the announcements, both Ciena and ALU plan to use QAM16 with one laser to get 200Gb/s into a 50GHz channel for shorter-reach applications, and bond two of these together to get 400Gb/s in 100GHz bandwidth -- which is double the bits/sec/Hz of 100G coherent using DP-QPSK, which will still be used for maximum reach long-haul.

So in this case they're 2x200G on the fiber (2 channels) not 4x100G (4 channels).

Sign In