Re: [EFM] Interesting headline / copper / ITU-T specs
Thank you, Bob,
this explanation helps to understand how "non-copper" people see the
issue. I mostly agree with your arguments. Regarding new coding schemes, several
modern (turbo-coding, LDPC, and come else) have been actively studied in ITU to
enhance the current ADSL performance. Neither one was accepted, in my opinion
mostly because the improvement they bring is rather small (another 1.5-3 dB of
coding gain depending on the noise conditions) but require very significant
raise of complexity. We are rather close to the theoretical capacity bound, so
all these coding and technology improvements, whatever they are, will not
increase the distances by more than about 10%, I think. Another issue is
crosstalk compensation techniques. Here we can improve the distance more
substantially, and overcome some spectral compatibility issues, however special
conditions are required to implement these algorithms, plus complexity and
latency. Intensive study of these techniques (which are actually also not new)
is going now under Dynamic Spectral Management (DSM) project in T1E1.4.
Bob Barrett wrote:
> There are many 'meanings' in my message here ;-). My main interest is 1GE
> optical, so I have no axe to grind as a vendor here. I sat in the copper
> track because I wanted so learn what was going on there. My votes in the
> optics and EPON tracks were not going to matter, and OAM was not having a
> One of the meanings is that I am sympathetic to Ethernet to copper modem
> implementations based on ITU-T PHYs, but I don't think that they need to be
> totally standardised within 802 and branded as EFM in order to succeed in
> the market. The reference case for this would be a router with a 10M
> Ethernet interface on one side and an ITU or ANSI circuit interface on the
> WAN side, say T1 or E1 historically. Such products succeeded in the market
> without all of the functionality being totally standardised within 802. I
> see the same being true for lower cost copper based circuit interfaces that
> are standardised in other bodies. These 'new' systems are just cheaper
> versions (by reason of being more integrated) of the routers of the early
> 1990s that can run over some other coding scheme rather than T1 / E1.
> Cannibalising the T1 market for data is now less of an issue to the SPs,
> because the business T1 market now needs to move to higher data rates.
> Another is that I am sympathetic to the Ethernet 'old guard', and like
> 'them' I do not want to see a data rate of less than 10M branded as
> 'Ethernet', just for the marketing benefit. I am also sympathetic to the
> needs of the SPs for reach, and to Geoff Thompson and his respect for the
> laws of physics. I would like to see a new PHY defined that can do 10M to
> 22k feet so that all system vendors at least can start a next generation of
> product from a 'level commercial playing field'. That's the only way I see a
> greater than 75% vote coming out of the copper PHY / coding track. Enhanced
> / bonded PHYs based on ITU-T specs is one way of addressing the issue.
> Personally I would like to see EFM take advantage of some of the newer
> coding schemes that are rumoured to be working in the university research
> labs at this time. That is why I think Scott was on the right track to
> request chip vendors to bring forth the results of their research projects
> in July.
> I don't think there is any chance of EFM getting 75% for a current copper
> coding scheme, so I don't think it matters if it takes longer to go through
> the process for a brand new coding scheme. I think a new coding scheme will
> stand more chance of being accepted in EFM if it is not already part of the
> work of another standards body.
> Best regards
> -----Original Message-----
> From: firstname.lastname@example.org
> [mailto:email@example.com]On Behalf Of Vladimir
> Sent: 23 May 2002 21:09
> To: bob.barrett@xxxxxxxxxxxxxxxxxx
> Cc: firstname.lastname@example.org
> Subject: Re: [EFM] Interesting headline
> does it mean you have a sympathy to the idea of EFM based on standard
> ITU-T PHYs? This is what I concluded from your message.
> Bob Barrett wrote:
> > Those that remained in Edinburgh until the bitter end of the interim at
> > seven PM Wednesday, would have woken to see the headline on the business
> > section of The Scotsman on Thursday morning reporting:
> > EFM future in doubt .....
> > The EFM in this instance is the Edinburgh Fund Managers. If I was at home
> > would scan the headline and post it on my web site, not post it as a zip
> > file :^). I'll bring it to the July meeting.
> > However, given the deadlock in the copper track these may be prophetic
> > words.
> > It all comes down to rate and reach. So to cut to the chase what is
> > is a copper PHY and coding scheme that can support 10Mbit/s and a reach of
> > 22k feet, and preferably a new specification, so that no major
> > industry body or other standards body has an axe to grind.
> > So I guess I am supporting Scott's call for the chip vendors to bring
> > finest forward at the July meeting.
> > There are plenty of Ethernet over copper products out there today based on
> > existing non-IEEE PHYs that work at sub-rates with long reach. There is
> > little point in trying to shoe-horn one of these into EFM. The fighting is
> > all about not letting the other camp win, because we all know that none
> > get 75% unless everybody bar one camp quits, and that will not happen. So
> > can we let them go as first generation, or whatever, and get on with the
> > next generation please?
> > These products can already be labelled Ethernet as they have a conformant
> > Ethernet port, so the marketing argument doesn't fly, and I don't buy the
> > 'buy it at Frys' argument either. I have never seen a Fry's outside of
> > California ;-). Seriously, the copper to your home and your service
> > will define what type of product you can use. The nearest that we will get
> > to an open market is the SP providing a list of products that will work on
> > the end of their copper.
> > Best regards
> > Bob