Kitz Forum
Broadband Related => Broadband Technology => Topic started by: tickmike on October 05, 2014, 01:18:11 AM
-
Of interest
< note BT copyright
SIN498 Issue 6.0
September
2014
Generic Ethernet Access Fibre to the Cabinet (GEA-FTTC) Service and Interface Description
I found it from
http://www.sinet.bt.com/sinet/ > Newly added documents > Accept copyright < > SIN 498 Issue 6.0
September 2014
-
Direct link should work - as long as the document is the most recent. *
http://www.sinet.bt.com/sinet/sins/pdf/498v6p0.pdf
*tested in browser where Ive not been to SINET before in
-
Did anyone notice the new stuff in this version?
As discovered on TBB: TBB linky (http://forums.thinkbroadband.com/fibre/f/4361752-fttc-based-adsl2-from-bt-openreach.html)
A new product: "GEA over ADSL2+"
I'll repeat my analysis:
Having read through the test specifications for GEA over ADSL2plus (in annex B), I get a feel for how they intend the product to work.
- BT specifies that GEA over ADSL2+ is intended for long lines only, so only specify tests with range options of 2km, 2.5km and 3km.
- For comparison, the corresponding tests in Annex A (for GEA over FTTC) have range options from 100m to 2.2km.
- The default line profile in the DSLAM would appear to set maximum speeds of 12Mbps down and 1.5Mbps up.
- Looks like they want to support multicast too - so presumably aims to deploy TV this way too.
- Use of PSD masks would be needed, as I described in the earlier post.
The SIN also specifies that the modem has to support SRA and PHYR (G.INP).
-
A new product: "GEA over ADSL2+"
I could be wrong on this, but Im sure BT retail already offer this as a service. Ive seen someone post on a forum a couple of months ago about this and how they were on a long D side and because their VDSL speed was less than 'x', BTr had put them on a special product and they get charged accordingly. I cant for the life of me recall what it was called now, but I do remember looking at the product on the BTr website. I'll try see if I can rack my brains where I saw it and or do a search later. I wish I could remember what the product was called :-[
The SIN also specifies that the modem has to support SRA and PHYR (G.INP).
That has definitely been there for a while... because I quoted some of SIN498 to TPLink a while back and it was one of the requirements I noticed when I sent them the link to an earlier version of SIN498.
----
grrr. cant find anything on it... and still cant recall where I saw it. I dont normally visit the BT forums, but if it was on there then I would have been looking for something else. I do remember thinking at the time - Ive not seen that before and I quickly glanced at the product.. it was something stupid though like 'Super broadband' which is absolutely no help when it comes to searching. I have a vague recollection at the landing page I first went to, and it listed Infinity 1, Infinity2 and this third product. Because I was trying to look for something else... I didnt really pay much attention at the time.
-
I could be wrong on this, but Im sure BT retail already offer this as a service. Ive seen someone post on a forum a couple of months ago about this and how they were on a long D side and because their VDSL speed was less than 'x', BTr had put them on a special product and they get charged accordingly.
Would that be something like "Total Broadband Option 3 with Fibre", or "Faster Option 3 with Fibre"? I recall BT Retail introduced something like this a few years ago, when they had a strict minimum threshold for the Infinity product line of 15Mbps; at the time there were 2 options within the "standard broadband" product line (hence the "option 3" part), and there was no such thing as "Infinity 2" as 80/20 hadn't started at the time.
There were the usual screw-ups when the product was first made available (sales reps mistakenly ordering a standard ADSL-based product instead); once sorted, it looked like subscribers were treated identically to Infinity subscribers after the order. They had engineer installs, with SSFP and the standard Openreach modem.
BT retail would make the discrimination based on the estimate, so (of course) some people got considerably higher speeds than estimated - and I recall seeing some actual download speeds higher than 30Mbps. That would preclude the use of any form of ADSL2+.
Unfortunate that we were subject to locked-out modems, as that means we have little evidence of what these people actually synced with.
ISPReview has this old story:
http://www.ispreview.co.uk/story/2011/07/01/bt-launch-uk-faster-total-broadband-option-3-deal-for-sub-15mbps-fttc-lines.html
The SIN also specifies that the modem has to support SRA and PHYR (G.INP).
That has definitely been there for a while... because I quoted some of SIN498 to TPLink a while back and it was one of the requirements I noticed when I sent them the link to an earlier version of SIN498.
True - though those requirements were a subset of the whole set for VDSL2 modems, including the vectoring ones.
The section I read (and copied those requirements from) were specifically for the "GEA over ADSL2+" modems, so didn't include anything like vectoring. I thought it was worth noting, because we don't see either SRA or PHYR as part of today's DLM for ADSL2+ (which would be BT Wholesale's DLM); it gives another suggestion that BT Openreach's DLM may be headed in a new direction.
For the sake of completeness, the test specification does include tests with FEC and interleaving turned on (a lower variant with 1 symbol of INP and 8ms delay, and a higher variant with 2 symbols of INP and 16ms of delay) - so we should expect those to continue. Conversely, all profiles mentioned had a fixed 6dB target margin - suggesting that we would not see the Openreach DLM performing stepwise increases of the SNRM to 9, 12 or 15dB even on ADSL2+.
-
Thinking about it, this would make for an interesting installation experience.
If the estimate for VDSL2 ends up less than, say, 15Mbps, then the installation could end up with best performance on either VDSL2 or ADSL2+ ... so the engineer ought to turn up to try out both. Which might mean using two different modems.
Of course, the idea of "best" changes over time - particularly when crosstalk is taken into account. And the switch-on of vectoring could change the ise of "best" yet again.
It could end up with a very fluid choice...
-
Has this new article from TBB any connection?
http://www.thinkbroadband.com/news/6654-asdl2-from-street-cabinets-to-solve-2-mbps-usc.html
-
Has this new article from TBB any connection?
http://www.thinkbroadband.com/news/6654-asdl2-from-street-cabinets-to-solve-2-mbps-usc.html
Yes. Essentially Saffy has read the forum thread and decided to create a News Item based on the information contained therein. ;)
-
Has this new article from TBB any connection?
http://www.thinkbroadband.com/news/6654-asdl2-from-street-cabinets-to-solve-2-mbps-usc.html
Yes. Essentially Saffy has read the forum thread and decided to create a News Item based on the information contained therein. ;)
cheeky saffy :D
-
Would that be something like "Total Broadband Option 3 with Fibre", or "Faster Option 3 with Fibre"?
Im afraid I cant recall. :( It wasnt 'Total' for sure.. but could possibly have been 'Faster'.
Conversely, all profiles mentioned had a fixed 6dB target margin
Hmm, so adsl following the same as VDSL. A lot of gamers would prefer a higher SNRm than interleaving. BToR's current solution of not being able to turn off interleaving -nor reset the DLM. But atm we on VDSL dont seem to have any choice in the matter, and its not something that BToR seem bothered about.
I saw a BT document once that strongly implied the reason why they dont use SNRm steps is because they can be tweaked and over-ridden using DMTtool (which they mentioned by name). ???
-
There was a story about it here http://www.ispreview.co.uk/story/2011/07/01/bt-launch-uk-faster-total-broadband-option-3-deal-for-sub-15mbps-fttc-lines.html
-
Thinking about it, this would make for an interesting installation experience.
If the estimate for VDSL2 ends up less than, say, 15Mbps, then the installation could end up with best performance on either VDSL2 or ADSL2+ ... so the engineer ought to turn up to try out both. Which might mean using two different modems.
Of course, the idea of "best" changes over time - particularly when crosstalk is taken into account. And the switch-on of vectoring could change the ise of "best" yet again.
It could end up with a very fluid choice...
I was searching for something earlier for my Zyxel router, when I stumbled across this link (http://www.businesswire.com/news/home/20080616006095/en/ZyXEL-Introduce-VDSL2-IP-DSLAMs-ADSL22-Fall#.VDPspBawTd4). Not much info there, but it would appear that Zyxel have been manufacturing VDSL line cards with ADSL2/2+ fall-back for 6yrs.. so Ive no doubt the likes of ECI and Huawei wont be that far behind.
It looks like Calix (http://www.lightreading.com/cable-video/calix-ups-ante-on-vdsl-vectoring/d/d-id/691214) have been doing this for >3yrs
adding Packet Transfer Mode technology to the platform also helps because it allows a telco to support VDSL2 technology where the copper loop length supports it and fall back to ADSL2 where it doesn't. "Not only that, but we are actually improving the bandwidth on long-loop ADSL2 deployments by about 13 percent,
-
Hmm, so adsl following the same as VDSL. A lot of gamers would prefer a higher SNRm than interleaving. BToR's current solution of not being able to turn off interleaving -nor reset the DLM. But atm we on VDSL dont seem to have any choice in the matter, and its not something that BToR seem bothered about.
I've seen a thread or two over on the Plusnet forums that show they've been involved in some meetings between CPs and BT with the aim of adjusting the options regarding DLM tuning for FTTC. I suspect we might see some options to allow DLM to reduce speed in preference to adding latency, and allow easier control by the CP. Eventually.
Of course, use of SRA and G.INP might reduce the need for FEC & interleaving anyway, but I don't know if the latency implications of G.INP would be welcomed by gamers.
I saw a BT document once that strongly implied the reason why they dont use SNRm steps is because they can be tweaked and over-ridden using DMTtool (which they mentioned by name). ???
Ooh. Not heard that before!
The NICC document on DSM (found here (http://www.niccstandards.org.uk/files/current/NICC%20ND%201513%20(2010-01).pdf?type=pdf)) reports that TRA (tiered rate adaption, or banding) is preferable to AMA (automatic margin adjustment).
However, there's no doubt that the more modern DLM techniques all aim to share the spectrum equitably amongst users, and try to avoid unilateral power increases as a way to make an individual line perform better. I guess something like DMtool subverts that kind of altruism :police: vs >:D
-
Yes. Essentially Saffy has read the forum thread and decided to create a News Item based on the information contained therein. ;)
At least he referenced the forum thread from the article :angel:
-
it would appear that Zyxel have been manufacturing VDSL line cards with ADSL2/2+ fall-back for 6yrs.. so Ive no doubt the likes of ECI and Huawei wont be that far behind.
The Huawei DSLAMs certainly support ADSL2+, but the phrase "fall back" from Zyxel is interesting. I wonder if that is some form of automatic fall-back, so as to select for the best speed.
-
I think they will be using adsl2 rather than adsl2+ capped at 12mbit, the 12mbit coincides with adsl2's own limit.
Adsl2 is better than adsl2+ on long lines, some say it doesnt matter, but it did on my ukonline line, when I synced with adsl2+ I had a lower sync speed and worse stability than with adsl2.
-
Adsl2 is better than adsl2+ on long lines , some say it doesnt matter,
I recall having a telephone conversation with Azzaka, many years ago (He'd rung after reading my theory on here & it was also partially a courtesy type call as I'd just got out of hospital so it must have been one of the times in 2008). BT had just introduced ADSL2+ and we discussed the theory why adsl2 worked better on long lines. Zen had customers with long lines just having been moved to adsl2+ that were performing quite poorly and they'd seen an increase in fault reports from these customers. I still stick to the same theory I had back then, and have since repeated it countless times.
ADSL2 has improved modulation over ADSL1. It has much more efficient algorithms than ADSL1 in particular when it comes to overheads for error correction. The RS codeword is far more efficient and is how '8Mb' was extended to '12Mb' over the same frequency band to 1.1 MHz.
ADSL2+ uses ADSL2 technology, but with one major difference - it doubled the amount of bins available for use and utilises frequencies up to 2.2MHz
Its highly unlikely any line over 50db atten (its probably closer to 45dB) is ever going to be able to make use of the additional frequencies, but making all these additional tones available seemed to cause more background type noise.* despite the line not being able to make use of them.
So why not give the line the best of both worlds? Better modulation & error correction algorithms of adsl2 yet not open the line to tones that they were never going to use.
Anyhow back to the call with Azzaka and he said that as an experiment, Zen were trying adsl2 on these longer lines (rather than taking them back to adsl1 which is what the ISPs previously did) and it did indeed seem to be proving successful.
afaik Zen were the first [BTw] ISP to try this, but since then Ive seen other ISPs follow suite and its a far more accepted these days that adsl2 may be best for some longer lines.
-----
*
1. I don't profess to know exactly why this would be, other than it does appear to cause some sort of 'seepage' to other tones. This similar theory is why complete tones may be blocked totally in the cases of say radio ham interference at the DSLAM or why certain rf filters are used. Blocking them completely is more effective than allowing them to remain in use and cause seepage to other frequencies.
2. We've all seen the increase in attenuation of appx 3dB that happens when a line is moved from adsl1 & adsl2 to adsl2+. Ive seen some state this is the reason why.. but I'm not totally convinced about this being a major factor because atten is just a measurement and line conditions are still the same... plus there are more tones for loop loss to be calculated over.
-
Thanks wombat, this has turned out to be an interesting discussion.
but the phrase "fall back" from Zyxel is interesting
Im not sure either as the details are scant, but the Calix mention of PTM to me is the key.
ADSL2+ will run over PTM. PTM also has a big advantage of far less (framing) overheads than ATM, so this will help throughput speeds a bit too.
There's no need for different modems either - take for example my Zyxel can be used on adsl1/2/2+ or VDSL from the one DSL socket and its not the only one that can do this Billion and TPlink to name a few.
Yes theres currently 2 different (software) interfaces:-
[1]An ATM interface for ADSL and
[2] a PTM interface for VDSL.
But if you ran ADSL2+ over PTM then surely the router should be able to auto-sense VDSL/VDSL2/ADSL2+ in exactly the same way that a router can auto-sense ADSL1/ADSL2/ADSL2+ over ATM?
Obviously there would still need to be a configuration change on the DSLAM but that happens anyhow now with adsl1/adsl2/adsl2+ etc.
Certain US ISPs are already upgrading their ATM networks to PTM (such as AT&T) and switching their standard ADSL users away from ATM and over to PTM framing.
With BT, on FTTC enabled exchanges a PTM network is already there. There's line cards available to support VDSL & ADSL2+. They'd perhaps have to rethink the PSDmasks but I really dont see that being too complicated.
The only fly in the ointment that I can think of, is LLU... and if it would cause the likes of Sky to spit out their dummies. :baby:
They are already unhappy about no direct access to the DSLAMs in the cab for VDSL so how much of a stink will they kick up when it comes to ADSL too?
There's already one line of thought that the LLU providers demands to provide access is what could be causing in part some delay to the roll out of vectoring.
-
...but making all these additional tones available seemed to cause more background type noise.* despite the line not being able to make use of them....
*
1. I don't profess to know exactly why this would be, other than it does appear to cause some sort of 'seepage' to other tones
This does make sense as the noise within a channel is more or less proportional to its bandwidth. Assuming the filtering to be correctly implemented, the extra bandwidth required for ADSL2+ entails the quid pro quo of higher noise figures. An ADSL2+ channel without the advantage of the additional ADSL2+ signal present would quite probably be less successful than the more bandwidth-restricted option of ADSL2 ordinaire.
Think of the obvious reductio ad absurdum: an early transatlantic cable carrying just one tone, detected by the human ear, could function adequately because most of the noise could be filtered out (or, in those days, the signal tuned-in – but much the same thing in practice).
"The wider you open the window, the more muck flies in"
-
Im not sure either as the details are scant, but the Calix mention of PTM to me is the key.
I would guess any VDSL2 enabled kit will run PTM over xDSL? I think Adtran, for one, does it.
They'd perhaps have to rethink the PSDmasks but I really dont see that being too complicated.
What's your thinking? I seem to remember reading the street side cabs were technically unregulated with regard to PSD masking; it's only the exchange which is 'regulated' if that's the right word?