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

RE: [EFM] FEC needs a good discussion home!




Larry,

Your message relates to several issues, so I will try to parse them
here.

The effect of FEC can be most visible as an improvement in optical link
performance, and within that context, I agree with Frank - it's
reasonable to propose that FEC's home should continue to be the Optical
PMD group.

FEC is currently low on the priority list in the Optical PMD group's
agenda because we first have to make architectural decisions on P2P (1G
and 100M) and choose a baseline PMD set.

From the perspective of optical link performance improvement, I can't
think of any reason why we should not continue to study and debate the
inclusion of FEC in PHY.

But to be fair, FEC raises several system level questions when you
mentally place it somewhere in the PHY - say, between MAC and PCS - and
run through a checklist of questions. Can we maintain backward
compatibility with legacy 1000BASE-X PHY? Can FEC operate within the
rules of 8B10B and yet provide sufficient performance improvement in
optical links to pass the cost/benefit analysis of system implementers?
Can legacy PMA devices operate acceptably well in the presence of lower
SNR, or will we have to abandon the wish of leveraging legacy PMA
devices?...You get the picture. Sure, it also raises potential system
benefits - Moore's law means FEC could some day cost very little, and as
Carlos pointed out, we can predict link degradations with more
sophistication.

Therefore, I would argue that FEC should start building a second home,
while continuing to keep its home in the Optical PMD group. I don't
think it means creating yet another formal group, yet. I think it means
targeting the whole EFM group, answering system level questions to their
satisfaction, and putting to vote a specific technical proposal to add
FEC sublayer in the PHY. You should do this AFTER you have established
feasibility with the Optical PMD group. The support of optical folks is
a necessary but not a sufficient condition.

Regards,
Vipul

vipul_bhatt@xxxxxxxx
408-857-1973


-----Original Message-----
From: owner-stds-802-3-efm@majordomo.ieee.org
[mailto:owner-stds-802-3-efm@majordomo.ieee.org]On Behalf Of larry
rennie
Sent: Wednesday, May 08, 2002 2:39 PM
To: FEffenberger@xxxxxxxxxxxxxxxxx; Vipul Bhatt; Wael Diab
Cc: stds-802-3-efm@ieee.org
Subject: Re: [EFM] FEC needs a good discussion home!


Frank, et al,

I am willing to start such a discussion group on the best way to
implement FEC.  To date, FEC implementation, although discussed during
specific FEC presentations, has not been discussed in a formal
discussion group because most of the effort to date has gone into
"proving" that FEC, for the cost,  will indeed be worthwhile for EFM.

I believe there is now enough technical information to predict with
confidence the performance improvement expected from FEC and that it is
worth the cost.  I plan to summarize this information at the Edinburgh
meeting. Hopefully, there will be enough support in Edinburgh to pass a
motion in support of adding FEC (or at least support to go forward with
FEC) and then we can get down to the nitty gritty implementation
details.   Vipul, Wael, is this a reasonable approach to take?  I would
expect such a discussion group to be populated with individuals that
can, and are willing , to contribute positively to the FEC
implementation details.

Regards,
Larry