|
Frank,
Thanks
for the nice analysis of this issue.
Here
are four assumptions:
1)
This is an issue only in the OLT in the upstream direction. (While
someone might want to make a dual mode ONU, it will always be running at one
speed or the other.)
2)
Most of the optical link budget proposals have had APD in the OLT
upstream.
3)
Most people like the idea of having the 1G and 10G upstream links share the same
band.
4) To implement a dual-mode 1-10G Amplifier,
you will have to optimize it for 10G
Given
those four assumptions, the 2-3 dB penalty that you calculated will be
present for the upstream 1Gbps link over the existing 802.3ah budget in 1G-10G
dual mode OLTs.
Given
that result, we seem to have two options w.r.t. the dual MII
proposal:
A) Retain the dual MII proposal and require that
all 1G-10G dual-mode OLT upstream implementations be sensitive enough to handle
the 2-3dB upstream loss over the 802.3ah 1G budget due to this issue.
(There is some possibility that particular 10G-specific losses that will
counter-balance some of this, i.e. chirp, scattering, etc.)
B) Implement a control structure in the MII
Management Interface that allows the Dynamic Bandwidth Allocation circuit to
communicate its expectation of which-rate-to-expect-when in a timely manner
to the PMD to allow it to switch the receiver in real time. (The latency
of that management interface may have to be constrained, and there might need to
be a "calendar" structure to the registers.)
There
may be other possibilities, but they are not obvious. Solution A has some
appeal.
Thanks, Brian Holden, PMC-Sierra
_______________________________________________
Brian Holden
PMC-Sierra, Inc.
3975 Freedom Circle, Santa
Clara CA USA
+1.408.239.8123 Fax +1.408.492.9462
brian_holden@pmc-sierra.com http://www.pmc-sierra.com
All,
Please find attached
my first draft of what I call the “PMD101” presentation.
This is basically
taking the highlights out of Chap. 4 of Agrawal’s textbook.
Comments,
improvements, etc are all welcome. Please chime in.
You will note that
the last page touches on the FEC superating topic (hence the subject line
above). From the PMD side, the answer is very simple: if you want to go
7% faster, you indeed need 7% more light power to get the same SNR. At
least, this is true if the receiver is extensively optimized.
Sub-optimal receivers might exhibit smaller penalties, since they don’t
attempt to get the last ounce of goodness at the lower rate.
I will start a
separate thread on the FEC issue, so hold discussions on that topic until
then.
Sincerely,
Frank
E.
From: Frank
Effenberger [mailto:feffenberger@huawei.com]
Sent: Thursday, January 18, 2007 12:18
AM
To:
STDS-802-3-10GEPON@listserv.ieee.org
Subject: [8023-10GEPON] [Dual-Rate-PMD]
Call for interest.
All,
This is just a quick note to alert
everybody to our work item on dual rate 1G-10G PMD for EPON.
The group will use the exploder
for all correspondence, but we will try to include the tag [Dual-Rate-PMD] on
all Emails.
If you couldn’t care less about
the topic, you can set your Email filters appropriately.
To the matter at hand, I think the
first order of business is to prepare a short tutorial slide-set that covers
the key issues of PMD design and link rate.
Those considerations are at the
base of the problem that we face. Stay tuned for more details.
On a side issue, we also had a
question on the issue of modifying the line rate to accommodate FEC. The
tutorial mentioned above will also inform that debate, in that it will
quantify the PMD sensitivity penalty that comes from link rate increase.
A bonus, free of charge. J
Sincerely,
Dr. Frank J.
Effenberger
Huawei Technologies USA
1700 Alma
Drive, Plano TX 75075
Office (732) 625
3002
Cell (908) 670
3889
|