Thanks Chris. Also thanks to Vipul Bhatt ,Will Bliss and Jonathan King for reviewing my emails.
As the result of the review, I am fixing some of the definitions and resending the email below.
As for the name of the BiDi specific Tx to near-end RX impairment, lets hold on to the term Echo (as opposed to NEXT) to be consistent with the copper world and the rest of the literature. In a BiDi, the near-end TX can reflect back to the RX through a single reflection point (ie a connector or the BiDi element itself) on the same channel. This is echo. If the RX channel was not the same as TX channel, then we could call it NEXT. (Keep in mind this is all naming and it does not really matter much).
Here is the corrected email:
The MPI path is defined to be the result of two reflections effect involving traversal of three segments. This is defined in page 5 of
http://www.ieee802.org/3/bm/public/nov12/farhood_01_1112_optx.pdf .
Both BiDi and Full-Duplex channels suffer from MPI in the same way and the upper bound calculations in farhood_01_1112_optx applies to both.
HOWEVER, the BiDi is also creating the Echo problem in addition to MPI. This is because the fiber is lit up on both ends and it is very possible that the transmitted signal reflects back to your own receiver through SINGLE REFLECTION (as opposed to two reflections). Lets call this single-reflection impairment Echo not to be confused with MPI. So
1- The MPI analysis in farhood_01_1112_optx only covers MPI and not Echo.
2- Similar upper bound analysis could be carried out for the Echo channel. However, I would imagine upper bounding the echo and leaving all of it uncancelled would be a huge problem since only a single reflection point is involved and the echo interference will be much stronger than MPI.
3- I could not find any echo cancellation circuitry in Jiangwei presentation. I do understand that *Circulators* can have good isolation capability and the very first *echo tap* is probably addressed here to certain extent. I am not sure how the other farther echo taps on the echo-tail are addressed. This is when the signal goes out of the TX, hits a connector and reflects back into the near-end receiver on the same channel.
5- The Echo impairment is a side-effect of the BiDi architecture and is not a problem in so-called Full-Duplex architecture. Also a BiDi with two different wavelengths has a much smaller Echo problem compared to a single wavelength case (I believe Jiangwei presentation covered single wavelength case).
Regards,
Arash Farhood
Director System Eng
Cortina Systems
Date: Wed, 14 May 2014 14:13:03 -0700
From: chris.cole@xxxxxxxxxxx
Subject: Re: [802.3_400G] MPI presentation in 802.3bm standard
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Arash,
We are now in agreement.
The case that limits bi-di receiver performance is reflection from a connector near the transmitter (i.e. NEXT). For a 10km link and standard fiber plant, -14dB interferer is reasonable to use in the analysis.
Chris
From: Arash Farhood [mailto:arash_farhood@xxxxxxxxxxx]
Sent: Wednesday, May 14, 2014 9:13 AM
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_400G] MPI presentation in 802.3bm standard
Hi Chris,
Thanks for your email. In our Q&A during Jiangwei presentation we both made a few technical mistakes when describing the problem (okay I made a lot of mistakes and you made a few :) ) and below I try to frame the question properly. Let me know if I missed anything.
The MPI path is defined to be a triple-reflection effect involving two reflecting points. This is defined in page 5 of
http://www.ieee802.org/3/bm/public/nov12/farhood_01_1112_optx.pdf .
Both BiDi and Full-Duplex channels suffer from MPI in the same way and the upper bound calculations in farhood_01_1112_optx applies to both.
HOWEVER, the BiDi is also creating the Echo problem in addition to MPI. This is because the fiber is light up on both ends and it is very possible that the transmit signal is reflected back to your own receiver through SINGLE REFLECTION (as opposed to triple reflection). I am going to call this single-reflection impairment Echo not to be confused with MPI. So
1- The MPI analysis in farhood_01_1112_optx only covers MPI and not Echo.
2- Similar upper bound analysis could be carried out for the Echo channel. However, I would imagine upper bounding the echo and leaving all of it uncancelled would be a huge problem since only a single reflection point is involved and the echo interference is much stronger than MPI.
3- I could not find any echo cancellation circuitry in Jiangwei presentation. I do understand that *Circulators* have good isolation capability and the very first *echo tap* is probably addressed here. I am not sure how the other longer echo taps are addressed. This is when the signal goes out of the TX, hits a single connector and reflects back into the near-end receiver.
4- I am not sure how widely the term *echo* is used by the optical folks but it is important to not mix this up with MPI.
Date: Wed, 14 May 2014 03:47:16 -0700
From: chris.cole@xxxxxxxxxxx
Subject: Re: [802.3_400G] MPI presentation in 802.3bm standard
To: STDS-802-3-400G@xxxxxxxxxxxxxxxxx
Hi Arash,
As per discussion during the Jiangwei’s 56G PAM-4 SMF bi-di 400GbE presentation, you may consider extending your analysis to calculate the impact of MPI on bi-di SMF receiver performance.
In this case a deterministic calculation is required. The case of interest is a 26dB connector RL for near-end TX with worst case link loss for the far-end TX and worst case power difference between the near-end TX and far-end TX, i.e a classic NEXT calculation. For a 10km link, 6dB of channel loss and 6dB power delta are reasonable values to use in the analysis giving a -14dB interferer.
Chris
Dear friends,
Looks like there are a lot of concerns around the sensitivity of PAM to MPI. This issue was investigated in 802.3bm by a few folks. Below is the latest bm contribution that discussed the *Statistical MPI* analysis.
http://www.ieee802.org/3/bm/public/nov12/farhood_01_1112_optx.pdf
The term *Statistical Analysis* was used to provide a more accurate *upper bound* compared to a *non-statistical* method that was more pessimistic and was presented earlier by Cisco.
As an example, using the statistical upper bound method using PAM4 modulation with 4 connectors , 26dB return loss per connector and ROSA/TOSA and ER=6dB:
Using non-statistical upper bound method based on bhatt_01_0512, MPI penalty=6.2dB
Using Statistical method based on farhood_01_0112_optx, MPI penalty=3.81dB
(For the definition of the *MPI penalty*, pls refer to either of the presentations)
The zhu_3bs_01_0514.pdf is showing 5.6dB MPI penalty. This seems too high.
Assuming the opical link is not under our control but ROSA and TOSA for 400Gb/s PMD is new and the RL is under control, then with Connector RL=25dB and ROSA/TOSA RL of -35dB:
Using non-statistical upper bound method based on bhatt_01_0512, MPI penalty=2.5dB
Using Statistical method based on farhood_01_0112_optx, MPI penalty=1.78dB
Now, even 1.78dB looks like a large number. Please note the following:
1- Even though the statistical method is more accurate than the simple upper bound method, it is ignoring channel IL and assumes everything is perfectly and coherently aligned (pls see farhood_01_1112_optx.pdf). In reality even the statistical method is a stretch. The traditional way that 802.3 has modeled this is through a *discount factor*. Even though I would consider this method not really scientific but in reality the MPI penalty is even lower than 1.78dB . (with discount factor of 0.6, the MPI penalty is 0.98dB.). I will consider bringing in more comprehensive analysis that address some of these pessimism.
2- 1.78dB MPI penalty does not really mean 1.78dB degradation in sensitivity. The MPI impairment is not Gaussian and similar method to COM model in 802.3bj can be developed to identify the actual operating margin loss (which would be less than 1.78dB). This is even more relevant when there is a FEC in the PMD.
3- The Statistical method assumes NO CANCELLATION/MITIGATION of the MPI in the receiver. This is not necessarily the smartest choice for 400Gb/s PMD.
4- Assuming 26dB connector return loss and 35dB ROSA/TOSA RL, then MPI penalty is 1.41dB. I am not sure why the 25dB connector RL was chosen.
Regards,
Arash Farhood
Director System Eng
Cortina Systems