100G Ethernet

ADVA Opens Another 100-Gig Front

SAN DIEGO -- OFC/NFOEC 2009 -- Just when you thought you knew all the modulation schemes out there, ADVA Optical Networking comes up with DPSK-3ASK.

It's a proprietary scheme, being announced today, for 100-Gbit/s optical transport in metro spans.

Why go through the trouble, and why use so many letters to do it?

Well, the second part has to do with whoever named "three-amplitude shift keying" (3ASK). As for the first part, ADVA says it's found a way to create a metro link using 40-Gbit/s components -- meaning it can be done now.

The scheme is proprietary, but that's OK for line-side connections (the links that traverse long distances between nodes in a network). Because a carrier usually owns both ends of such connections, a line-side link can be proprietary without causing any interoperability fuss.

"If you look at 40 Gbit/s, it's never really embaced standards on the transport side," says Jim Theodoras, ADVA director of technical marketing.

ADVA has talked about this idea for a while and presented some simulation results last fall. This week, the company plans to announce it's gotten the concept working in the lab, producing results that match the simulation.

DPSK-3ASK is a spin on the dual phase-shift keying (DPSK) that's more common in optical modulation. It sends a 40-Gbit/s optical signal but transmits five bits of data for every two symbol transmissions -- which multiplies out to a 100-Gbit/s data rate.

By using this format, a carrier could send the data using off-the-shelf 40-Gbit/s equipment and components, rather than having to wait for every missing piece needed to make pure 100-Gbit/s transport work.

"We're not having to wait for coherent detectors. We're not having to wait for new DSPs [digital signal processors] or a polarization tracker," Theodoras says.

Other examples of new parts needed at 100 Gbit/s include a Serializer/Deserializer (SerDes), something Sierra Monolithics Inc. and CoreOptics Inc. are working on separately.

A new type of analog-to-digital coverter (ADC) chip, capable of running 56 billion samples per second, could also be a requirement; Fujitsu Microelectronics Europe (FME) is the only company so far claiming such a device. (See Sierra Strikes Forth for 100G and Fujitsu Micro Tackles 100G.)

ADVA is targeting the metro because it sees a no man's land there between standards.

Short-reach connections -- 10 km and under -- are being covered by the Institute of Electrical and Electronics Engineers Inc. (IEEE) 802.3ba standard for 40 and 100 Gbit/s Ethernet, which is expected to be ratified in mid-2010. The Optical Internetworking Forum (OIF) , meanwhile, is looking at standards for long links, the kind reaching up to 2,000 km.

"The need is real. I think what's been holding it up is the massive investment required to go 2,000 km," which has pushed much of the standards attention to those longer spans, Theodoras says. Because that represents a "limited number of links," ADVA feels some more immediate work should be done to bring 100-Gbit/s transport to the metro network.

— Craig Matsumoto, West Coast Editor, Light Reading

tailpin 12/5/2012 | 4:08:30 PM
re: ADVA Opens Another 100-Gig Front

Great comment -- "In particular there is zero chance of interoperability for any modulation acronym with more than 3 letters - it's Newton's 4th Law of Optics :-) "

Especially when no one can agree on what interop means.

hyperunner 12/5/2012 | 4:08:30 PM
re: ADVA Opens Another 100-Gig Front

Hi Craig,

Not sure why you mention this so often in the article.- Can you name any optical modulation scheme that's "standard"?- By that I mean that there's a genuine optical mid span meet.

Even different optical boxes from the same vendor oftentimes do not interoperate on the line side.

In particular there is zero chance of interoperability for any modulation acronym with more than 3 letters - it's Newton's 4th Law of Optics :-)


Pete Baldwin 12/5/2012 | 4:08:29 PM
re: ADVA Opens Another 100-Gig Front hR -- OK, fair point. I was trying to emphasize that unlike, say, the varieties of QPSK/DPSK (which at least go under the same names, even if they aren't 100% compatible), nobody else that i know of is doing the 3ASK thing -- but that that's not so much a problem on the line side. Belaboring an obvious point, maybe.
Mostly Harmless 12/5/2012 | 4:08:28 PM
re: ADVA Opens Another 100-Gig Front

>>"We're not having to wait for coherent detectors. "

Just wondering how they'll detect the interference in the receiver then.- All phase-based detectors are coherent.

If this is DPSK I'm guessing they're usign a delayed copy of the signal to create the interference.- But that's still cohrent detection - just no local oscillator.

Happy to be corrected - physics class was a while ago ;-)



opticaljunkie 12/5/2012 | 4:08:25 PM
re: ADVA Opens Another 100-Gig Front Technically the technique they use is called differential detection, using a delay-add interferometer, phase information is detected. However, if there's PMD in the link, the delay length is cannot be optimized and the you usually get the wrong bits. On top of that, delay and add is achieved using only the signal (both t baud and t+1 baud) contains noise from link. Coherent uses a local oscillator, much cleaner.
hyperunner 12/5/2012 | 4:08:24 PM
re: ADVA Opens Another 100-Gig Front

Well to be picky, Mostly Harmless is correct.- Any detector that measures the interference caused by phase differences is coherent.- We may have vendors choosing to ignore it and use their own definitions, but Physics is Physics.

To quote Wikipedia..."Two waves are said to be coherent if they have a constant relative phase, which also implies that they have the same frequency."- Bottom line - this is a phase-based, coherent-detection method.

Differential techniques use a copy of the signal itself, but techniques currently being promoted by the long haul 40G and 100G vendors use local oscillators.- Both approaches have pros and cons.- If ADVA is proposing this for the metro only then they should be able to keep the implementation less complex and so be cheaper.- But to carry that 100G DPSK-3ASK signal over a long haul network you'll need to re-modulate the signal.

Anyway-ADVA are definitely using a coherent detector - even if their marketing guy thinks they don't :-)



paolo.franzoi 12/5/2012 | 4:08:23 PM
re: ADVA Opens Another 100-Gig Front


I read the quote differently.- My thought here is that they are modulating ASK on top of DPSK.- The DPSK symbols are running at 40 Gb/s and therefore use standard 40G optical components in that area.- So, they "don't have to wait" for pure 100G components to be available.



hyperunner 12/5/2012 | 4:08:22 PM
re: ADVA Opens Another 100-Gig Front

Hi seven,

I don't think that's what Jim Theodoras meant, and here's my reasoning.

All of these fancy modulation schemes (that I'm aware of) operate with lower speed electronics.- For example, Nortel's 40G PM-QPSK takes 4 x 10Gbit/s electronic signals and sends them using multiple phases and polarizations.- The 100G technique goes further still and splits 8 x 12.5G signals over phases, polarizations and two separate wavelengths.

So what I think Mr.Theodoras was saying is that ADVA isn't waiting for progress on "coherent detectors with local oscillators and DSP chips" (this is what guys like Nortel, Ciena and Huawei mean when they speak of "coherent detection").

Nortel has proven that this technology works, ans works very well.- But if you've seen the board - especially the 100G board - it's not pretty.

OIF is promoting a single modulation technique for 100G to allow guys like JDS and Avanex to build PIC-based transmitter and receiver modules to try and control the number of components and fiber couplings.- But I wouldn't hold my breath for those PICs if I were you.

Note to Craig:- OIF is not working on a "standard" modulation technique.- There is no plan at this stage to have different vendor line sides interoperating using these PICs. It's purely to give the component guys a chance to get some volume to try and help offset the huge expense of getting a PIC development to work.

Disclaimer:- I'm not an expert in these matters, but we had a very good presentation on this topic last week.


teleliar 12/5/2012 | 4:08:21 PM
re: ADVA Opens Another 100-Gig Front With this modulation scheme the baud rate will be 1/log2(6) of the bit rate (not 2.5 as the article claims).- Using existing 40G components, there will not be much space left for FEC overhead that is required for metro application.
Sign In