Re: [8023-10GEPON] [POWER_BUDGET] Min code gain requirement
Raymond,
The power budget ad hoc understands that they need to provide the required
coding gain to the FEC group. However, it would take them some time to
analyze various proposals and find the proper balance between what they can
achieve in the optical domain and what help they need from FEC.
The basic organizational question here is do we do everything serially
(like wait for PMD group to give us the required FEC gain before we start
working on FEC format, and after that start thinking about sub-rating vs.
super-rating)...
... or try to parallelize activities to the extent possible.
There are some things that we can decide without having the exact gain
requirement set. For example, we may discuss format of FEC codewords and
synchronization procedures (even though ratio of parity to payload may
change as a result of choosing a different coding scheme later).
We also can make a fairly reasonable assumption that FEC will result in
additional overhead (in other words, that we would need something stronger
then what backplane specified). Then, independently of the other tasks, we
may discuss whether in general the group prefers sub-rating MAC or
super-rating PHY.
I treat it as iterative process where we would continuously refine the
features as we go forward.
Regards,
Glen
> -----Original Message-----
> From: Raymond Leung [mailto:wkleung@HUAWEI.COM]
> Sent: Wednesday, January 31, 2007 6:59 PM
> To: STDS-802-3-10GEPON@listserv.ieee.org
> Subject: Re: [8023-10GEPON] [POWER_BUDGET] Min code gain requirement
>
> Dear All,
>
> Until now, it seems to me that almost all discussions on FEC concentrate
> only on how to match with the MAC rate, line rate etc. There are very
> little discussions on FEC to help the design to fulfill the system power
> budget.
>
> Is it possible for the power budget ad hoc group to define a minimum
> required coding gain of FEC, say 4dB optical gain or ~6dB electrical gain.
> (RS(255,239) can achieve almost 6dB electrical code gain at BER=1e-12 and
> so
> we may define the minimum coding gain based on this reference.) After
> defining the minimum value of code gain, it become possible to examine the
> maximum code rate, minimum code length and FEC algorithm all together to
> achieve the target code gain. With this kind of figures (max. code rate,
> min code length, etc), we can then go back to examine whether it is
> possible
> to do with the MAC and line rates. If it is possible, then we can go
> ahead,
> otherwise we can feedback that the required min code gain is not a
> feasible
> requirement.
>
> FEC, which is the most flexible module in the system, allows us to make
> any
> change on its parameters easily compared with other optical device. We
> can
> adjust the code rate and code length to obtain difference coding gain
> (with
> penalty on the bandwidth efficiency and complexity). Unlike other optical
> devices, FEC no need to wait for the technology improvement to achieve a
> better gain.
>
> If we need a high power budget to achieve a higher split ratio (say 64,
> 128.), FEC is a good choice to solve this problem.
>
> Regards,
>
> Raymond