Re: [8023-10GEPON] 10G-1G coexistence
Dear all,
Since the topic of discussion is becoming lengthy, I will comment inline ...
Best wishes
Marek Hajduczenia (141238)
SIEMENS Networks S.A. - IC COM D1 R
Rua Irmãos Siemens, 1
Ed. 1, Piso 1
Alfragide
2720-093 Amadora
Portugal
* Marek.Hajduczenia@siemens.com
http://marekhaj.easyisp.pl/index.php
(+351.21.416.7472 4+351.21.424.2082
-----Original Message-----
From: Frank Chang [mailto:ychang@VITESSE.COM]
Sent: quarta-feira, 10 de Janeiro de 2007 6:33
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: Re: [8023-10GEPON] 10G-1G coexistence
Frank E;
Thanks for the input. Quick comment is that I donot disagree with what you said, actually there existing multirate analog front-end parts off-the-self specially in the 1G/2G range or even up to 4G. Several vendors incls my employer are making alot of those, mostly for fibre channel. And PON ONU side can be multirate as well because of its inherent driver. That's not what I meant for the feasibility study.
For 1G-10G coexistence, we are talking about much bigger rate gap, so far not heard any such part exists.
<Marek-10-01-2007> I must agree here with Frank. Doing 1G/4G is one thing, doing 1/10G with different line coding and incompatible clock rates is another. This may pose a significant challenge for the construction and implementation of the hardware, especially the MAC client layer. I know we are not supposed to deal with the hardware implementation - it has to work with the state machine, but still ... </Marek>
Any analog 10G part like TIA should be able passing low rate traffic, but its perf. will be significantly tolerated. This requires re-tweaking both the low- and high- side of the freq. response to deal with Rx sens. and burst to burst variation in levels. Also to trigger such switch, it may need some kind of extra feature for "rate indicator" if such signal isnot issued by MAC.
<Marek-10-01-2007> Frank, such a signal would have to originate from the MAC Client which would cause violation of the stack - I do not believe we will be able to include such type of signalling in the standard. The big question is then whether we want to go for parallel optimized 1G/10G reception path or try to squeze everything into a single receiver and separate the signal paths only in the electrical domain. </Marek>
Typically these are hard to squeeze in due to constraints from to-can type packages, which will reside in compact module. I feel this is one of the challenging part.
Beyond that, FEC data sync typically works after low rate following some kind of data recovery and deMuxing function in the PMA layer. The extra feature for "rate indicator" could be added, but multirate range typically much smaller because of jitter issue. My understanding there are few parts even for small rate gaps.
Bursting switching and speed-sensing could be quick enough if done in anlog way. But any 10ns timescale for the bursty Rx response is questionable. The current burst part quite much limited to rate below 2.5G.
Another consideration as Maurice point out, FEC is rarely used in 1G today, but may be a necessary option for 10G. This will probably further complex the scenarios to perform the FEC burst data sync.
<Marek-10-01-2007> Frank, correct me if I am wrong but the 1G/10G stacks will be implemented independently from what I gather from Glen's presentation (and common sense type of approach) - in that case, isn't it natural that 1G signal with or without FEC is dealt with in the 1G stack the way it is dealt with (and works) in the current standard compliant 1G EPONs? We do not need to change anything in here since the 1G and 10G signal paths must be already separated before they reach the FEC sublayer. </Marek>
Regards
-Frank
<Marek-10-01-2007> Thank You for Your comments </Marek>
-----Original Message-----
From: Frank Effenberger [mailto:feffenberger@huawei.com]
Sent: Monday, January 08, 2007 8:21 AM
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: Re: [8023-10GEPON] 10G-1G coexistence
Frank C.
You might find the following link interesting:
<http://www.patentstorm.us/patents/6862322-description.html>
Evidently, this typewriter company has an idea for a "switchable bandwidth
optical receiver". The basic idea is that you start with a low-rate/high
gain TIA, to which you add a FET shunt impedance. When detecting low rate
signals, you leave the TIA alone. When detecting high-rate signals, you
activate the FET so as to load down the TIA, increasing bandwidth albeit
reducing gain.
Of note, the application area in mind for this invention was classical PTP
links. To transform this to PON, there are two potential issues:
1. Burst mode: We need to change our bandwidth quickly. In my opinion, I
don't think this is a problem, because the device being used to change the
impedance is a simple electronic component - if we can't switch a FET on the
10ns timescale, then we have bigger problems.
2. Speed-sensing: In classical systems, auto-sensing/negotiation is often
the goal. But, in PON, the OLT is the master, and the OLT knows when a 1
Gb/s burst is coming in, and when a 10 Gb/s burst is coming in. So, with
that a-priori knowledge, it shouldn't be a big deal.
"What about ranging?" you might ask. Well, the simplest solution to that
would be to have two ranging opportunities, one for low speed and one for
high speed, with the TIA set appropriately. Since ranging doesn't happen
very often, it is efficient enough.
One other data-point on this area: I've heard that Calix (formerly Optical
Solutions) has a commercial system that supports both 622/155 B-PON and
2.4/1.2 G-PON in the same ONU. I'm not sure about their OLT side. Anyway,
it is an interesting approach to network evolution support, and a glimpse of
what might be possible.
Regards,
Frank E.
-----Original Message-----
From: Frank Chang [mailto:ychang@VITESSE.COM]
Sent: Sunday, January 07, 2007 2:30 AM
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: Re: [8023-10GEPON] 10G-1G coexistence
Glen;
I support the initiative to investigate this approach. One concern is that
there exist no common OLT RX o/e frond-end can handle 1G and 10G
simultaneously before able to enable one of the 2 separate sync circuits.
This is in addition to the complexity of OLT sync circuit, we need to
explore its technical feasibility as well.
Regards
-Frank
-----Original Message-----
From: Glen Kramer [mailto:glen.kramer@teknovus.com]
Sent: Friday, January 05, 2007 12:21 PM
To: STDS-802-3-10GEPON@listserv.ieee.org
Subject: [8023-10GEPON] 10G-1G coexistence
Dear Colleagues,
Based on one of straw polls taken at the November meeting, there seems to be
a strong support for using 1310 nm wavelength for upstream 10G transmission
(see Straw Poll "F" in November *unapproved* minutes). Many people have
indicated they would support this approach if coexistence can be achieved
with 1G upstream.
I believe there is a way that 10G upstream can coexist with 1G upstream,
both using 1310 nm wavelength. Please, comment on the attached slides, and
also please let me know if you'd like to be included as a supporter for this
approach.
Best regards,
Glen