Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714



Hi Ian,

For PAM-4 and DMT (as proposed in 802.3bs), the modulation penalties are 5 dB and 9 dB, respectively. 

So the relative modulation penalty of DMT compared to PAM-4 is 4 dB of which 3.5 dB is due to PAPR. 

Chris 


> On Aug 26, 2014, at 1:13 PM, "Chris Cole" <chris.cole@xxxxxxxxxxx> wrote:
> 
> Hi Ian,
> 
> Optical specifications and optical link budgets are an established methodology for describing optics. They are a simple booking formalism for summarizing results of comprehensive analysis and measurements. Even the most complex impairments like jitter can be captured by a penalty, as long as it is done carefully. In all IEEE 802.3 projects, we have always been able to summarize the effects of jitter and other complex impairments with an optical link budget.
> 
> Sampled data systems are not so difficult and not so complex that they are beyond description in understandable and established terms. It does require more work, but that should not discourage us from doing it.
> 
> It also appears that your colleagues view optical link budgets as an important tool, as shown in the July 2014 "400G Discrete Multi-Tone PMD for 10 km SMF" presentation:
> 
> http://www.ieee802.org/3/bs/public/14_07/tanaka_3bs_01a_0714.pdf#page=6  
> 
> The penalties in this optical link budget are minimal and many more have to be added, but the presentation shows clear agreement that this is the proper approach to summarizing optical link performance. 
> 
> Chris
> 
> -----Original Message-----
> From: Dedic, Ian [mailto:Ian.Dedic@xxxxxxxxxxxxxx] 
> Sent: Tuesday, August 26, 2014 12:18 PM
> To: Chris Cole
> Cc: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: RE: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> Hi Chris
> 
> I wasn't disputing that DMT has a 9dB penalty due to PAPR, but your phrasing implied that it had a 9dB penalty compared to PAM4 which is not correct when pre-emphasis and frequency response are taken into account, since the actual PAPR for PAM4 is no longer the ideal value at either TX or RX.
> 
> Of course this -- and other non-idealities like jitter -- can be accounted for with an increased implementation penalty for PAM4. But they do need to be included somehow, and this can be difficult to do realistically with a simple loss budget calculation like yours, a system simulation with degradations of real components is more accurate -- or of course, measurements using actual hardware.
> 
> Maybe this is why our comparison of DMT and PAM doesn't show the advantage with PAM that simple calculations predict ☺
> 
> Cheers
> 
> Ian
> 
> 
> 
> Sent from my Android phone using TouchDown (www.nitrodesk.com)
> 
> 
> -----Original Message-----
> From: Chris Cole [chris.cole@xxxxxxxxxxx]
> Received: Tuesday, 26 Aug 2014, 19:06
> To: Dedic, Ian [Ian.Dedic@xxxxxxxxxxxxxx]
> CC: IEEE_listserver [STDS-802-3-400G@xxxxxxxxxxxxxxxxx]
> Subject: RE: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> Hi Ian,
> 
> 
> 
> Thank you for the nice additional detail on quantization noise. 
> 
> 
> 
> We assumed no pre-emphasis in the optical signal out of the 50GBaud PAM-4 transmitter. There could be pre-distortion driving the DAC, but that is intended to cancel the transmitter optics response. We also assumed no DFE in the 50GBaud PAM-4 RX DSP, only a 13-tap FFE. We have also not modelled RX clock jitter effects on sampling. 
> 
> 
> 
> Generally, I agree with you that the analysis of 50GBaud PAM-4 is idealized and optimistic, and that there are many implementations penalties which are not accounted for. I invite you to present a more detailed analysis of 50GBaud PAM-4 that accounts for more penalties. 
> 
> 
> 
> The 9dB Modulation penalty calculation for DMT is unrelated to the PAM-4 calculation. It is an absolute calculation of the reduction in eye size because of higher modulation order and reduction in signal power to support tails of the DMT distribution. Additional 50GBaud PAM-4 implementation penalties will not reduce the DMT 9 dB Modulation penalty. 
> 
> 
> 
> Chris
> 
> 
> 
> 
> 
> From: Dedic, Ian [mailto:Ian.Dedic@xxxxxxxxxxxxxx] 
> Sent: Tuesday, August 26, 2014 10:51 AM
> To: Chris Cole
> Cc: IEEE_listserver
> Subject: RE: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Hi Chris
> 
> 
> 
> The classical definition of “quantisation noise” is the additive noise caused by slicing the signal into 2^N levels, so it is directly related to the resolution of the converter. For an 8b resolution ADC or DAC the SNR for a full-scale sinewave (PAR=3dB) is 50dB down from full scale (6.02N+1.78 dB). An ADC or DAC with this SNR would have 8 effective bits (ENOB=8.0), but this can never be achieved in practice due to other sources of noise and distortion not related to the resolution of the converter.
> 
> 
> 
> When all sources of noise and distortion (including quantisation noise) are added together we get a lower ENOB figure for the converter, for example ENOB=6.0 (your assumption, which is somewhat pessimistic) would correspond to 38dB SNR for a full-scale sinewave – but this is not quantisation noise as this is normally defined. If you mean “the total noise and distortion contribution of the A/D conversion process) this is the definition of ENOB.
> 
> 
> 
> Your calculation of 9dB PAPR penalty for DMT vs. PAM4 is based on two assumptions – basically, that the PAM4 PAR is close to the ideal value for the constellation:
> 
> 
> 
> 1.       Assumes no pre-emphasis is used for PAM4 to compensate for bandwidth limitations in the TX chain (DAC/package/PCB/driver/modulator/laser)
> 
> a.       depending on bandwidths, typically 3dB of DAC dynamic range (0.5 ENOB) is used up with PAM4 pre-emphasis
> 
> 2.       Assumes the PAR seen by the ADC is the same as transmitted, which ignores the bandwidth of the RX chain (PD/TIA/PCB/package/ADC)
> 
> a.       the attenuation of high-frequency components (101010) means the PAR seen by the ADC is increased for PAM4
> 
> 
> 
> Especially for 100G/lambda PAM4 the effect of both these factors is significant because of the bandwidth requirements, and I’m not convinced that these factors have been allowed for in your calculations – if I’m wrong, please explain where these are included.
> 
> 
> 
> A third issue is clock jitter amplification at both TX (due to pre-emphasis) and RX (due to FFE) which is again significant for PAM4, especially at 100G/lambda where the baud rate is higher (less jitter can be tolerated) and more equalisation is needed (jitter amplification is bigger). A multi-tap DFE which does not amplify RX jitter has major implementation problems at these clock rates.
> 
> 
> 
> Another jitter issue is the required clock phase noise when running the RX at 1 sample per bit, because the clock recovery loop has to be narrow bandwidth to suppress noise from the low-gain phase estimator, which means the loop cannot then suppress phase noise in the sampling clocks (from reference clock noise and/or clock PLL). This either means that the required clock jitter is extremely (impractically?) low or that eye closure from jitter will be worse than expected.
> 
> 
> 
> All these are issues linked to real PAM4 implementation penalties which do not seem to be adequately included in your analysis – again, please correct me if I’m wrong.
> 
> 
> 
> (they are all included in our DMT simulation and measurement results – and comparisons with PAM -- because these are based on real hardware, with simulation models fitted to actual measurements)
> 
> 
> 
> It’s crucial that such real implementation penalties are included when trying to compare two very different modulation systems, and I’m concerned that the numbers being presented for PAM4 compared to DMT in your analysis are overly optimistic because they do not correctly include such penalties.
> 
> 
> 
> Cheers
> 
> 
> 
> Ian
> 
> 
> 
> Ian Dedic
> 
> Chief Engineer
> 
> Communications Business Unit
> 
> Fujitsu Semiconductor Europe GmbH
> 
> 3 Concorde Park, Concorde Road
> 
> Maidenhead SL6 4FJ, UK
> 
> Tel : +44 (0) 1628 504 711
> 
> Mob : +44 (0) 7795 534 253  *** NEW ***
> 
> 
> 
> From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx] 
> Sent: 26 August 2014 17:57
> To: Dedic, Ian
> Cc: IEEE_listserver
> Subject: RE: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Hi Ian,
> 
> 
> 
> The DMT Electrical BTB measurements on page 15 (page 14 footer numbering) show the combined effect of DAC and ADC quantization. This is an elegant measurement that has been used for the past 30 years to show the total electrical SNDR (Signal-to-Noise Distortion Ratio) in communications systems due to DAC quantization in TX and ADC quantization in RX. When the quantization contribution is uncorrelated with the signal, as is the case for the PMD proposals in 802.3bs, the contribution can be approximated as an additive noise term Nq’ in SNR at the decision slicer. The Wiki reference <http://en.wikipedia.org/wiki/Quantization_(signal_processing)>  discusses this nicely. 
> 
> 
> 
> SNR (w/o quantization) = S/N
> 
> SNR (w/ quantization) = ~S/(N + Nq’) 
> 
> 
> 
> The reason for the prime term is that this is a filtered version of the quantization noise that is measured by the electrical BTB measurement. Some DSP algorithms, for example for sigma delta converters, dramatically reduce the quantization term. 
> 
> 
> 
> For finite quantization:
> 
> Nq’ > 0
> 
> Therefore:
> 
>                S/N > S/(N + Nq)
> 
> 
> 
> Since Nq reduces or degrades the SNR, it is a penalty. As pointed out in an earlier email, the tricky part is that the is an additive term, whereas penalties are multiplicative (dB). To convert Nq effect on SNR to a penalty, it has to be calculated together with all other impairments, i.e. a full end to end link simulation. The penalty is the difference between SNR with finite quantization and SNR with no quantization. This is similar to how TDP is calculated. 
> 
> 
> 
> For the 2km LAN-WDM DMT proposal in 802.3bs, we calculated the SNR penalty with quantization (6 ENOB DAC and 6 ENOB ADC) as 3dB (excluding other RX design limitations) of which we allocated 2.5dB to TDP and 0.5dB as RX quantization penalty. The latter is in rough agreement with your 10% estimate, which corresponds to a 0.4dB SNR penalty. We also agree that this is not the dominant term. The dominant term is the modulation penalty including PAPR effects which is 9dB. 
> 
> 
> The SNR penalty for PAM-4 is less than DMT because there is no quantization clipping effect. We are tweaking this penalty and will have slightly revised numbers in our September presentation. 
> 
> 
> 
> Chris
> 
> 
> 
> 
> 
> From: Dedic, Ian [mailto:Ian.Dedic@xxxxxxxxxxxxxx] 
> Sent: Tuesday, August 26, 2014 2:49 AM
> To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: Re: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Just to make it clear, the page referred to does not really show the effects of quantisation noise as such – it shows the effect on bit rate of using better ADCs and DACs with higher ENOB (Effective Number Of Bits), which includes everything causing noise and distortion (including jitter).
> 
> 
> 
> At these sample rates ADC/DAC performance is not limited by quantisation noise in either case, other causes of noise and distortion dominate – for example if an 8 bit ADC or DAC has ENOB=6.3 (a good figure!) the ideal quantisation noise (ENOB=8.0) is 10dB below the overall noise, so only contributes 10% with 90% coming from other sources.
> 
> 
> 
> I guess American page numbering has the title slide labelled as page 1, where European numbering has the first content slide labelled as page 1, just like a book J
> 
> 
> 
> Ian
> 
> 
> 
> Ian Dedic
> 
> Chief Engineer
> 
> Communications Business Unit
> 
> Fujitsu Semiconductor Europe GmbH
> 
> 3 Concorde Park, Concorde Road
> 
> Maidenhead SL6 4FJ, UK
> 
> Tel : +44 (0) 1628 504 711
> 
> Mob : +44 (0) 7795 534 253  *** NEW ***
> 
> 
> 
> From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx] 
> Sent: 20 August 2014 22:02
> To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: Re: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> While the technical results in Ian’s July presentation are impressive, his page numbering is off. Or we have the same problem as with how floors are numbered in a building, with Ian following the European convention.  
> 
> 
> 
> Below is the correct link to the page showing the additive effects of TX and RX quantization:
> 
> 
> 
> http://www.ieee802.org/3/bs/public/14_07/dedic_3bs_01a_0714.pdf#page=15
> 
> 
> 
> 
> 
> Chris
> 
> 
> 
> From: Chris Cole [mailto:chris.cole@xxxxxxxxxxx] 
> Sent: Wednesday, August 20, 2014 7:43 AM
> To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: Re: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Hi Sudeep,
> 
> 
> 
> An elegant measurement showing the additive effects of TX and RX quantization is shown in Ian’s July presentation:
> 
> 
> 
> http://www.ieee802.org/3/bs/public/14_07/dedic_3bs_01a_0714.pdf#page=14
> 
> 
> 
> Chris
> 
> 
> 
> From: Chris Cole 
> Sent: Tuesday, August 19, 2014 10:31 PM
> To: 'Sudeep Bhoja'
> Cc: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: RE: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Hi Sudeep,
> 
> 
> 
> Here is a nice discussion on the  subject:
> 
> 
> 
> http://en.wikipedia.org/wiki/Quantization_(signal_processing)
> 
> 
> 
> Chris
> 
> 
> 
> From: Sudeep Bhoja [mailto:sbhoja@xxxxxxxxx] 
> Sent: Tuesday, August 19, 2014 10:25 PM
> To: Chris Cole
> Cc: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
> Subject: Re: [STDS-802-3-400G] FW: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
> 
> 
> Chris, 
> 
> 
> 
> I am still struggling to understand what you mean by quantization noise penalty in TX and RX. Can you please explain? 
> 
> 
> 
> Thanks,
> 
> 
> 
> Sudeep 
> 
> 
> On Aug 19, 2014, at 9:23 PM, Chris Cole <chris.cole@xxxxxxxxxxx> wrote:
> 
>    Hi Matt,
> 
>     
> 
>    Thank you for the questions and comments. Please see detailed responses inserted in the text of your email below.  
> 
>     
> 
>    More generally, the PMD specifications presented on today’s SMF Ad Hoc call are idealized and scrupulously apples-to-apples. As we quantify more penalties, the specifications will get harder not easier. Before fractions of a dB are challenged piecemeal, we should first have a comprehensive set of understood penalties and design limitations, which can then be looked at in total. A detailed contribution from you moving us in this direction would be welcome. 
> 
>     
> 
>    Chris
> 
>     
> 
>    From: Matt Traverso (mattrave) [mailto:mattrave@xxxxxxxxx <mailto:mattrave@xxxxxxxxx> ] 
>    Sent: Tuesday, August 19, 2014 11:12 AM
>    To: Chris Cole
>    Cc: STDS-802-3-400G@xxxxxxxxxxxxxxxxx <mailto:STDS-802-3-400G@xxxxxxxxxxxxxxxxx> ; Marco Mazzini (mmazzini)
>    Subject: Few questions on cole_02_0814_smf & clarification on updates from cole_3bs_01a_0714
> 
>     
> 
>    Hi Chris,
> 
>     
> 
>    Thanks for the contributions.  You may have addressed some of these questions at the microphone (on cole_3bs_01a_0714 <http://www.ieee802.org/3/bs/public/14_07/cole_3bs_01a_0714.pdf> ) or verbally, yet as someone who has been unable to attend the calls/meetings, I’d appreciate your patience in addressing the questions.
> 
>     
> 
>    1)      Quantization (*) noise: I count three penalties in your draft budget containing quantization noise.  I am concerned that you are overestimating by putting it into three line items.  (a) modulation penalty <adding 0.2dB for “exact levels”>; (b) TDP- on slide 4 of cole_02_0814_smf, you indicate that TDP includes quantization <unclear constituent pieces of TDP>; (c) TIA & Quantization penalties on the RX.  These are all reasonable sources of penalty to further investigate, yet I’m concerned that your comparison table probably mixes some cells which are well known with others that have more rigorous experimental basis such that the reader may inadvertently  draw conclusions which are not beneficial for the industry.  In particular, what is the basis of the RX Quantization penalty?  Do you have a reference?
> 
>     
> 
>    *** The total of all quantization related penalties in the TX and RX specifications is ~1 dB which is not an overestimation. 
> 
>    RX quantization and other penalties were calculated through simulations. Quantization effects are well understood and extensively published on. Estimating this penalty is complicated by dependence on relative contribution of all other noise sources. We are open to refining it, although any update is likely to be small.
> 
>     
> 
>    2)      FEC Coding Gain on KP4 vs. KR4: (this refers to cole_3bs_01a_0714 <http://www.ieee802.org/3/bs/public/14_07/cole_3bs_01a_0714.pdf> ): I believe you used the 1e-15 input BER for the KP4 and the 1e-12 input BER for KR4 (leveraging zhai_400_01_0713 <http://www.ieee802.org/3/400GSG/public/13_07/zhai_400_01_0713.pdf#page=10> ).  Also, I noted on on slide 8, you took a 0.5dB OMA relaxation to accommodate a ~3% rate increase in KP4 vs. KR4 – this seems excessive.  I didn’t notice KR4 mentioned in cole_02_0814_smf <http://www.ieee802.org/3/bs/public/adhoc/smf/14_08_19/cole_02_0814_smf.pdf> , is there a reason that you are selecting KP vs. KR?
> 
>     
> 
>    *** All FEC gains are referenced to BER = 1e-12 so that a comparison can be made to LR4.
> 
>    The penalty in the presentation for 3% KP4 FEC overhead is not 0.5dB but 0.064dB (rounded to nearest 1/10 of a dB). It is equally applied to all PMDs.
> 
>    KP4 was used because that was proposed in bhat_3bs_01a_0714 <http://www.ieee802.org/3/bs/public/14_07/bhatt_3bs_01a_0714.pdf> , of which you are a supporter, and is minimum required for higher order modulation PMDs. KR4 was used for some 50Gb/s lane PMD alternatives in the July presentation, but changed to KP4 to enable a fair comparison. KP4 could not be used for DMT because it requires a stronger FEC, as was noted during the SMF Ad Hoc call.
> 
>     
> 
>    3)      TDP: On slide 8 of cole_3bs_01a_0714 <http://www.ieee802.org/3/bs/public/14_07/cole_3bs_01a_0714.pdf> , you have a 2.0 dB penalty for PAM4 2x λs DML KP4 FEC, yet on slide 5 of cole_02_0814_smf <http://www.ieee802.org/3/bs/public/adhoc/smf/14_08_19/cole_02_0814_smf.pdf> , this has shifted to a 1.5dB penalty.  Why did the TDP change?  Conversely, the “4x100G KP4 FEC PAM4 MOD” column in your ad hoc preso has 2.5dB versus a 2dB penalty in the July presentation – why the change?  Any references you can point to?  (seems strange that the 8x got easier yet the 4x got harder?)  As you can tell from mazzini_01a_0814_smf <http://www.ieee802.org/3/bs/public/adhoc/smf/14_08_19/mazzini_01a_0814_smf.pdf> , we are trying to understand the behavior & best characterize penalties.
> 
>     
> 
>    *** Generally, we have refined our analysis, and will continue to refine our analysis, so numbers changed and will change. In this iteration, we added jitter to 50G and 100G alternatives, and quantization to latter.
> 
>    Specifically, 50Gb/s NRZ TDP decreased because of use of stronger FEC (KP4 in Aug. vs. KR4 FEC in July). 100Gb/s PAM-4 TDP increased (“4x got harder”) because of use of a weaker FEC (KP4 in Aug. vs. BCH FEC in July). TDP decreases as pre-FEC operating BER increases because BER curves converge. The comment that “8x got easier” is incorrect; TDP for one 8x increased, for one 8x stayed the same, and for one 8x decreased. Coincidently, although arrived at differently, the 2.5dB TDP for 100G PAM-4 in today’s SMD Ad Hoc presentation happens to be the same as in bhat_3bs_01a_0714 <http://www.ieee802.org/3/bs/public/14_07/bhatt_3bs_01a_0714.pdf#page=9> , which you are supporting. The suggestion of bias is unfortunate. 
> 
>    Marco’s presentation is great. I appreciate his accuracy in describing the measurement environment and its limitations, and the fidelity of the data. Please support this effort so that we can better understand and quantify all penalties and design limitations.
> 
>     
> 
>    4)      FEC Optical Gain: Based on verbal conversations with IC folks, I do not believe the halving of NCG (page 6, row “FEC Optical Gain v. 1e-12 BER” cole_02_0814_smf <http://www.ieee802.org/3/bs/public/adhoc/smf/14_08_19/cole_02_0814_smf.pdf> ) is appropriate.  I am trying to find a better approximation.  If you have a good reference (beyond the generic intensity vs. E-field) I’d appreciate seeing it.
> 
>     
> 
>    *** A factor of approximately half is appropriate for going from electrical to optical domain. Exact values have been extensively discussed in 802.3bm, for example on the reflector with the subject line: “[802.3_100GNGOPTX] FEC and error ratios.”
> 
>    Detailed derivations can be found in:
> 
>    http://www.ieee802.org/3/bm/public/mmfadhoc/meetings/nov29_12/anslow_01a_1112_mmf.pdf <http://www.ieee802.org/3/bm/public/mmfadhoc/meetings/nov29_12/anslow_01a_1112_mmf.pdf> 
> 
>    http://www.ieee802.org/3/bm/public/jan13/anslow_03_0113_optx.pdf <http://www.ieee802.org/3/bm/public/jan13/anslow_03_0113_optx.pdf>
> 
>    We have used 2.6dB and 3.2dB as the optical gain for KR4 and KP4, respectively, as these have been generally accepted as correct. We are open to refining these gains, but any update is likely to be small. 
> 
>     
> 
>    5)      OMA Eye v. EyeSNR: Suggestion that you move to using Eye SNR (see slide 7 of mazzini_01a_0814_smf <http://www.ieee802.org/3/bs/public/adhoc/smf/14_08_19/mazzini_01a_0814_smf.pdf> ).  As we are moving into links that leverage FEC/DSP, I think it is appropriate that we adopt SNR terminology rather than eye openings.
> 
>     
> 
>    *** Since 802.3bs is chartered with writing optical specifications and many participants have an optics background, using optics terminology is the most helpful in promoting a broader understanding of different proposals. Many participants have a long history of using optics terminology for FEC, and more recently DSP, in telecom and ITU-T applications. Optics terminology was used to include FEC in 802.3bm 100GBASE-SR4 specifications, so we have a clear example of how to do this successfully in the IEEE.
> 
>     
> 
>    thanks
> 
>    --matt traverso
> 
>     
> 
>     
>