|Thread Links||Date Links|
|Thread Prev||Thread Next||Thread Index||Date Prev||Date Next||Date Index|
The first weekly FEC conference call was held this morning. The meeting notice is attached below. A summary of the call follows:
Larry Rennie (Chair)
Mike Parket (EFMA)
Michelle Olesiejuk (EFMA)
1. Talked about the relevance of the FEC technical motion passed by
.ah at the Edinburgh meeting. This was discussed because there have
been recent FEC issues raised that are really non-issues when measured
against this motion. It was agreed that we would abide by the motion
and move forward, i.e.
b) Backward computability (Stream approach would use FEC On/Off and Frame approach transparent)
2. Briefly discussed the applicability of existing test equipment if an FEC On/Off capability is implemented. Some disagreement on whether or not new test equipment would be needed anyway. Decided to put this item as a metric in the metric comparison list (See AI 2 below)
3. FEC baseline proposals will assume the existing 8B10B line code.
4. Discussed specifying FEC to enable leveraging into other optical standards (FC, 10GE). Agreed that this would be considered in the FEC spec but would not be a driving force. It will be added to the metric list to be used in comparing FEC baseline approaches (currently stream and frame).
6. Ajay Gummalla is working on Stream approach details.
7. Briefly discussed the impact of the 10E-4 BER on the PMA (symbol alignment, etc.). Frank E. said that Vipul's optics group is working this issue (see action item 1).
8. Discussed that if FEC is optional, then two PMD parameter sets for both TX and RX would be required (a set with FEC and one without FEC). For example, the TX PMD spectral width could be relaxed with FEC and the RX PMD BER would be 10E-4 with FEC.
9. Agreed that FEC meetings will be the same time every week.
1. Frank E. will contact Vipul to find out status of Optics group's work on impact of 10E-4 BER.
2. Larry R. will put together a list of metrics to compare FEC baseline proposals. Will send out for comment before next meeting.
Wednesday, August 7 from 1-2 PM EST.
larry rennie wrote:
Several of today's emails on FEC has expanded the list of issues. Most of these new issues are really issues that need to be decided before we talk about specific FEC implementation proposals (currently the frame and stream based approaches). In fact, they are issues that, in a perfect world of standard development, would have have been decided before any FEC design implementations were undertaken. The issues that are really "pre-FEC design" I tried to list first:
1. Should FEC be mandatory or optional for P2MP? What about P2P.
2. If mandatory or optional for P2P should it be backward compatible with legacy PMDs?
3. Do we want the FEC we design to have the capability to be leveraged beyond this EFM application? For example, leverage into 1GE, 10GE, 1GFC, 10GFC.
4. Maintain the 8B10B line code or change to accommodate a "better" FEC implementation (for example, 64B66B)? It is argued that a new line code would require a new PCS and, probably, a new and tougher CDR design.
5. Effects on symbol alignment and reliability at a pre-FEC line BER of 10E-4? Other effects?
6. Comparison of stream-based vs. frame-based technique. Establish a list of comparison metrics. What work is needed to come up with these metrics. Note that the remaining issues listed would be part of the comparison metrics.
7. Impact on existing Clauses.
8. Impact on new Clauses (EPON timing/delay, etc.). OAM error reporting.
9. Complexity vs. cost concern vs. performance enhancement. Is the benefit worth the cost? Silicon impact, throughput impact. Experience of prior and existing uses of FEC in a "similar" environment (G.709, G.975, etc.).
Mike Parket wrote:Dear EFMA Members and Invited Guests:
You are invited to attend a weekly FEC Conference call. The first conference is scheduled for:
Date: Wednesday, July 31, 2002
Time: 1:00 p.m. (Eastern)
Duration: 1 hour
Chair: Larry Rennie, National Semiconductor
Dial in: 1-800-331-8016 (toll free), 1-816-650-0666 (direct)
Participant code: 744004
Proposed Agenda/Discussion: see below
A weekly FEC conference call is being established to come up with one EFM solution that will have a good chance of being accepted by the EFM TF at either the September or November meetings. The EFMA is interested in encouraging consensus on one FEC proposal and has volunteered to host these weekly calls. In order to reach consensus by September or November, there are a lot of issues that need to be discussed and resolved. At the Vancouver meeting, the P2MP track came up with the following FEC issues:
- Comparison of stream-based vs. frame-based technique
- Performance analysis
- Impact on Clauses
- Impact on EPON timing/delay
- Backwards compatibility issues
- "Optional" or "mandatory" - is it really optional?
- Complexity vs. cost concern
I have re-arranged this list in what I believe is the best order of priority and added some comments:
1. Backwards compatibility. We need to decide if this is a hard goal, a desired goal or not a goal at all.
2. Comparison of stream-based vs. frame-based technique. Establish a list of comparison metrics. What work is needed to come up with these metrics. Note that the remaining issues listed would be part of the comparison metrics.
3. Impact on existing Clauses.
4. Impact on new Clauses (EPON timing/delay, etc.)
5. "Optional" or "mandatory" - is it really optional?
6. Complexity vs. cost concern vs. performance enhancement. Is the benefit worth the cost? Silicon impact, throughput impact. Experience of prior and existing uses of FEC in a "similar" environment (G.709, G.975, etc.).
The following is the suggested agenda for FEC conference call 1:
1. Introductions. EFMA's role.
2. Purpose of conference calls. Goal(s). Meeting minutes (need volunteer).
3. Establish a list of issues we need to discuss and their priorities (see suggested list above).
4. Lets talk per agreed upon prioritized list of issues.
Note: These conference calls are not to replace reflector discussion but are intended to be a forum for issues and discussions that cannot be resolved via reflector email and can best be handled in a real time environment the conference calls provide.
Michael Parket / Coordinator / Ethernet in the First Mile Alliance
39355 California Street, Ste. 307 / Fremont, California 94538
Phone +1.510.744.4003 / Fax +1.510.608.5917 / Mobile +1.408.209.6719
Managed by Association Management Solutions (AMS);
Forum Management, Meeting and Event Planning