Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Bowdon]

BT Update on FTTPoD Stop Sell Removal, 3db Trial and Long Range VDSL

BT has published some important updates on their broadband products and future UK technology trials, which reveals a plan to test 3db Signal/Noise margins, the first location for their Long Range VDSL (FTTC) trial and removal of BTWholesale’s “stop sell” for Fibre-on-Demand (FoD) lines.

As usual most of this information has found its way to us via BTWholesale’s most recent ISP Forum event and attendees, which is often used to brief Internet Service Providers (ISP) on the operator’s plans for the forthcoming year. In order to save time we’ll cut right to the chase and link in past articles for those who seek some deeper context.

[WWWombat]

3dB trial:
- Proof of concept on 150 lines; 84% saw improvement. No-one worse
- Current trial is up to 40k lines
- Expect 100% to gain an average of 10% speed
- Expect 90% to gain an average of 15% speed
- Expect 50% to gain an average of 17% speed

No mention of what happened to error rates.

(Aside: I wonder how they'll achieve this expectation of 100% of lines to gain. Surely it means they'd have to lift all package speeds too? 44Mbps, 60Mbps, 88Mbps?)

My old rules-of-thumb would suggest that 3dB could buy 11Mbps if you get 80Mbs; 10Mbps if you get 70Mbps; 8Mbps if you get 50Mbps; 6Mbps if you get 25-35Mbps.

Those value do suggest that, if they are seeing a median improvement of 17%, then they are achieving a full 3dB gain in most places ... not just 2dB or 1dB.

[gt94sss2]

(Aside: I wonder how they'll achieve this expectation of 100% of lines to gain. Surely it means they'd have to lift all package speeds too? 44Mbps, 60Mbps, 88Mbps?)

I presumed they meant an increase in the attainable rate not the speed of the package purchased..

[niemand]

Due to some delays on provisioning of any other operators considering options with regards to FoD.

Emphasis on considering. It's still ridiculously costly compared with even EADs.

[S.Stephenson]

If only the costs could drift back down to the level it was when it was first announced.

[Chrysalis]

I assume now the 3db trial will be a hauwei only thing unless they prepared to try it without g.inp.

[S.Stephenson]

Well there are quite a few lines left on ECI with g.inp active just only a few on DSLWS.

I'm sure that lines would still be stable even without g.inp at 3db, my non-g.inped line is sitting at 4.3db quite happily.

[N0STIE]

Mine is currently at 4.3 dB DS without G.INP and it seems perfectly stable

[Chrysalis]

hence my unless comment, it may be that they still prepared to try it without g.inp.  But obviously it will be much harder for a line to have a low enough error rate for DLM to change a line over.

[kitz]

3dB trial:
- Proof of concept on 150 lines; 84% saw improvement. No-one worse
- Current trial is up to 40k lines
- Expect 100% to gain an average of 10% speed
- Expect 90% to gain an average of 15% speed
- Expect 50% to gain an average of 17% speed

No mention of what happened to error rates.

I too would have liked to have seen error rate info.   As you say 100% of lines wont see improvement unless they raise the 40/55/80 sync rate limits.
 
As mentioned by me in another thread, the 150 lines was phase one trial last year.  I was told last month's (phase 2) trial was open for up to 50k of lines. 
The amount actually on the trial would depend upon the ISP participation.  - link.

---

I keep repeating that the 3dB trial was independent of G.INP and is not cab related.

[Chrysalis]

I have accepted the statement with a different meaning.

e.g. right now there is a 6db target on all lines.  Yet lines that are either banded or synced at the max line rate will have a higher snrm.  This change I expect is not targeted at these lines and as such they are excluded from results.  After all if you alter the target snrm of a line synced at 80mbit with say a 8db snrm, then it will stay at 8db snrm, it doesnt need the extra snr.

The data is probably from lines not in those 2 categories.

If I was openreach I wouldnt want to be raising the 80 cap to 88, that will only trigger more speed related faults.  What happens e.g. if someone initially gets 88 on the 3db target, then after some crosstalk drops to 81, they would still be at 80 on the original cap and not be bothered, but then would be more likely to raise a fault on the 88 cap.

Not to mention I think there is no marketing benefit from 88 over 80, VM already beat that headline speed.

What I expect has happened, is openreach after deciding to not do a blanket vectoring rollout (probably due to cost), then had a discussion how else could they bring speeds closer to the headline speeds, and dropping target snrm was a solution proposed with minimal cost.  I would also expect error levels to be pretty low to allow a line to lower its snrm, as there is no point e.g. lowering a line with 2500 ES a day, then at 3db it goes into red status, and what happens then? interleaving whilst on 3db? or move back to 6db and start jumping back and forth.  More likely there will be an amber like on the current profiles, and it will only move lines down in green status.