re: Nortel Rolls On With 40-GigBoozon, I would be very interested to know why PM-QPSK has no none linear effect issues at 100G while you claim that there are issues at 40G? This does not make any sense to me.
In fact there might be issues with high launch power into DCSMs with any phase modulated technology, regardless whether its 40G or 100G. You say that PM-QPSK will be the industry choice for 100G and I totally agree. But that's exactly why "100G ready" optical line systems either avoid DSCMs at all or you make sure that the launch power into a DSCM does not exceed about 0dBm.
re: Nortel Rolls On With 40-GigHave you read the OFC OTuM5 paper by ALU that shows the high nonlinear cross talk that 40G PM-QPSK (or DPQPSK) suffers from neighboring 10G NRZ channels in a system with in-line DCMs? Unless the carriers get rid of all their in-line DCMs it's difficult to see how this modulation format will become the de facto choice at 40G. As a matter of fact, a few thousands of 40G line cards deployed to date on existing infrastructure use PSBT (aka ODB) and DPSK modulation schemes, whereas you can find only a few tens (at most) of PM-QPSK cards in the field. Maybe there's the answer to your question...
Whereas the 40G best modulation scheme is still a very debated topic it seems that everybody converges to PM-QPSK for 100G, where the non linear effects with neighboring channels are not as bad as at 40G. This is good news as large economies of scale will be achievable if all systems use the same components.
re: Nortel Rolls On With 40-GigThe DCM issue is significant... If the PM-QPSK terminal equipment can only tolerate adjacent 10G lambda in systems without DCMs, then this could significantly limit where this new gear could be deployed.
Perhaps this is a known limitation of the first Nortel PM-QPSK gear, but a truly flexible solution would be for upcoming designs their PM-QPSK systems to accommodate 10G neighbors with DCMs.
re: Nortel Rolls On With 40-GigNoCopper, who said that there are no nonlinear effects at PM-QPSK 100G? There are but they're not as bad as at 40G. Why? Because the 40G symbol duration matches the 10G bit period, whereas this is not the case at 100G. Therefore there is some averaging of the non linear effects at 100G whereas at 40G the NL effects experienced in every span add up. This is the Mickey Mouse explanation, and I don't pretend that it's very rigorous. You can find a more scientific one the several technical papers on this topic (starting from the OFC one I mentioned).
re: Nortel Rolls On With 40-GigIs it because that 100G will be running at 4 WDM lanes instead of a single WDM lane used by the current 40G DPQPSK implementation?
re: Nortel Rolls On With 40-GigIs this the same Lucent research team that spent the last five years publishing papers showing that electronic dispersion pre-compensation could not possibly work, when Nortel we selling working systems?
Yes there are problems with XPM when you launch at high powers into DCMs. The answer? Dial down the power, and use coherent detection to improve the SNR. Same trick that submarine systems have been using for years.
If the Nortel 40G did not work, how are they selling it to so many people?
In fact there might be issues with high launch power into DCSMs with any phase modulated technology, regardless whether its 40G or 100G. You say that PM-QPSK will be the industry choice for 100G and I totally agree. But that's exactly why "100G ready" optical line systems either avoid DSCMs at all or you make sure that the launch power into a DSCM does not exceed about 0dBm.
Unless the carriers get rid of all their in-line DCMs it's difficult to see how this modulation format will become the de facto choice at 40G.
As a matter of fact, a few thousands of 40G line cards deployed to date on existing infrastructure use PSBT (aka ODB) and DPSK modulation schemes, whereas you can find only a few tens (at most) of PM-QPSK cards in the field.
Maybe there's the answer to your question...
Whereas the 40G best modulation scheme is still a very debated topic it seems that everybody converges to PM-QPSK for 100G, where the non linear effects with neighboring channels are not as bad as at 40G. This is good news as large economies of scale will be achievable if all systems use the same components.
Perhaps this is a known limitation of the first Nortel PM-QPSK gear, but a truly flexible solution would be for upcoming designs their PM-QPSK systems to accommodate 10G neighbors with DCMs.
-AD
There are but they're not as bad as at 40G.
Why?
Because the 40G symbol duration matches the 10G bit period, whereas this is not the case at 100G. Therefore there is some averaging of the non linear effects at 100G whereas at 40G the NL effects experienced in every span add up.
This is the Mickey Mouse explanation, and I don't pretend that it's very rigorous.
You can find a more scientific one the several technical papers on this topic (starting from the OFC one I mentioned).
Yes there are problems with XPM when you launch at high powers into DCMs. The answer? Dial down the power, and use coherent detection to improve the SNR. Same trick that submarine systems have been using for years.
If the Nortel 40G did not work, how are they selling it to so many people?
P
That is on the line side.
on the client side, it will likely be 4x25g using 4 optical wavelengths up to 10km...
sailboat