Openreach Pause UK Trial of G.INP Fix for ECI FTTC Broadband Cabs UPDATE2
Openreach has paused a trial of Physical Retransmission ReTX (G.INP) technology for their troublesome ECI based FTTC “fibre broadband” street cabinets. The technology works fine on their Huawei estate but ECI has long been the naughty child, although their trial is expected to continue next year.
G.INP (ITU G.998.4) is essentially an error correction solution that is designed to help resolve spikes of Electromagnetic Interference (impulse noise), which can impact the stability and performance of VDSL2 lines (FTTC). The introduction of this technology can, on some lines, also produce a small increase in service speed.
In June 2017 we received some good news after Openreach announced that they may have finally found a solution to the long running problem of getting G.INP to play nice with their ECI estate (here) and we understand that the trial finally started on 25th September 2017. This was due to run until December 2017, although a new announcement states that the trial has been “paused” (here).
The related briefing is private but we hope to get a bit more background from our contacts in the near future. Suffice to say, ECI has long been a problem area for new FTTC upgrades and so it wouldn’t surprise us if Openreach were still having trouble getting it to work.
UPDATE 1:53pm
We understand that this pause is temporary and will last until sometime in early 2018, which is largely attributed to the Christmas and New Year break.
In other words the trial should continue and Openreach are currently examining the initial rollout to 30,000 lines. Once resumed the plan appears to be for this to be extended to 570,000 lines, which suggests more of a wide-scale pilot than a limited trial.
UPDATE 1st Feb 2018
We understand that so far the ECI Retransmission Trial has migrated around 600,000 lines and the official rollout is about to commence across their estate. The xdB (3dB profile) has also been successfully applied to some ECI lines.
Mark is a professional technology writer, IT consultant and computer engineer from Dorset (England), he also founded ISPreview in 1999 and enjoys analysing the latest telecoms and broadband developments. Find me on Twitter, Google+, Facebook and Linkedin.
« Ofcom UK Moots High Frequency Spectrum for Fixed Wireless Broadband
Privacy Notice: Please note that news comments are anonymous, which means that we do NOT require you to enter any real personal details to post a message. By clicking to submit a post you agree to storing your comment content, display name, IP, email and / or website details in our database, for as long as the post remains live.
Only the submitted name and comment will be displayed in public, while the rest will be kept private (we will never share this outside of ISPreview, regardless of whether the data is real or fake). This comment system uses submitted IP, email and website address data to spot abuse and spammers. All data is transferred via an encrypted (https secure) session.
NOTE 1: Sometimes your comment might not appear immediately due to site cache (this is cleared every few hours) or it may be caught by automated moderation / anti-spam.
NOTE 2: Comments that break our rules, spam, troll or post via known fake IP/proxy servers may be blocked or removed.
Hmm not good. Just checked and the one my flat connects to is an ECI cabinet.
How can I check that for my cabinet?
I connect to a Huawei cabinet but checking addresses for my neighbours between me and the cabinet speeds are very strange some closer get good speeds with some further away and houses between them getting slower speeds. I am in sync at almost fastest speeds at cabinet on VDSL but if I moved to G.nip I get slower speeds than now. This exercise was a waste of money as it only helps those within a certain distance from the cabinet. The majority will lose speed, only way it should have gone was finishing the fibre to the home. You can read the intermittent increase of speeds in the forums where I have posted them
That sounds awfully like the kid-with-a-new-toy, if I can’t have one nobody else should either. Most of life is swings and roundabouts we’ll find when we grow up. We all really do need to get out of it and start growing up.
G.nip works best around 300m from the fibre port. Originally the fibre cable and eqpt, was meant to go nearer DP’s than cabinets. But
The cost out weighed the return on investment. So solutions are in hand. Just to finish Huawei = Chinese, ECI = Isreal, who have always
had poor eqpt on the network. I say poor but mean absolutely Crap, nightmare working with it.
@dragoneast go look at the expected speeds houses 4 doors from cabinet to get less than some further away I don”t want it don’t need it but … You don’t expect a home 4 doors away from cabinet to get less than a Semi 6 doors away. Figures do not add up. As for ever needing 330 I do not think any person needs that speed I never max out my 80..
https://www.ispreview.co.uk/talk/showthread.php/34624-G-fast-waste-of-effort
At the moment G.Fast is not a fully rolled out product and as such the checker results may not be accurate. It would be better to compare sync speeds of normal FTTC to the sync speeds G.Fast
Openreach have a blackout period from mid-Dec where they don’t roll out firmware and software upgrades.
I guess they feel that 30k lines is enough to analyse over the holiday period to make sure there are no issues.
Openreach are a waste of time! Always does! Never learn!
Network change freezes are a very common thing. Operators, resellers, distributors and in turn the vendors that support them tend to have people on leave so leaving things alone to try and ensure stability is a really good idea.
Could they be using G.Vectoring instead? It’s recently been applied on my line, although Huawei
Very unlikely in most cases.
No. ECI hardware has co vectoring capability.
It can be retrofitted and Openreach have done this in trials but, as you said Mike, the hardware installed now can only vector on the same line card which is no good.
Definitely feels like a two tier system if your unfortunate to be on an eci cabinet like me there’s little chance of getting speed improvement when I got fttc back in 2013 I was getting 70 down & 20 up now I’m down to 53 with interleaving applied most of the time! Where as if your lucky to be on a Huawei cabinet you get everything g.inp,vectoring, 3db profile so wish I was on a Huawei cabinet, i would just be happy to have g.inp back on eci was lucky to have it last year when it was rolled out and what a difference it made nearly a 10 meg gain in download speed and hardly errors on the line, let’s hope Openreach get it sorted and it’s back early next year hopefully
Same here, it’s sad that those on ECI are paying the same price as those on Huawei with either G.INP and/or Vectoring. I honestly have doubts we’ll see a G.INP rollout for ECI next year, more a matter of when Openreach will give up with G.INP on ECI I guess. Pretty frustrating.
Oh, forgot to mention that there’s also possibly a two tier system within ECI as well. Line cards running on version b206 versus those running d086, at least for my line I’ve found that my line syncs quite a bit faster on d086 vs my other line on b206. The power output shows as 13 dBm or higher on d086 where on b206 it showed as around 6 dBm if I recall correctly. My line on b206 was something like 55/13 I believe, where my new line on d086 (after some further adjustments by SFI engineers due to a fault I was having) is 82/24 currently (or around 90/24 probably if I wasn’t using Draytek which applies reed solomon coding ‘R’ on the downstream even on fastpath). Of course I could always try probably reduce my target SNRM to 3 dB on the DrayTek for more speed but right now there’s no need as I’m already syncing at 80/20 :).
I think Openreach should replaced all ECI cabinets to Huawei cabinets and removed all ECI cabinets to scrapyard.
I totally agree,but I doubt it
You never know,no idea as to how many ECI cabinets would need replacing,but if they can’t sort ECI issue,then they might say enough is enough & decide to have all Huawei cabinets.
There was a ECI cabinet here,which got flattened,& was replaced with a Huawei.
@Optical flattened you say, replaced with a Huawei you say…mmm [rubs chin thoughtfully, remembering his local ECI cabinet is right next to a main road] 😉
Oh, look, trying to squeeze every last drop of speed out of unbalanced copper pairs by using frequencies they were never designed to carry, leading to them radiating like b*****y, exhibiting crosstalk through the roof and picking up every bit of external RF turns out to be a nightmare.
Who’d have thought it ?
RF engineering 101. VDSL should never, ever have been contemplated.
Unbalanced copper pairs? You don’t know what you’re talking about. If they’re unbalanced even by a few ohms they fail a pair quality test when tested.
Hmm. Not sure who doesn’t know what they’re talking about here.
Ohms is a measure of DC resistance.
I’m talking about balanced in terms of the pair having a constant impedance throughout its length. Whilst this impedance will (obviously) vary with frequency, at any given frequency within the range in use, the impedance needs to be constant.
Otherwise radiation and crosstalk will occur.
The bundles of wires used for telecoms do not exhibit this behaviour. This is why all these techniques to reduce the impact are required.
And yet those techniques persist, and thrive.
It seems that Moore’s law does a great job in making outdated all those limitations that things were “never designed for”.
Hence, for example, the existence of 2.5GBASE-T for Cat-5e cable, putting 2.5Gbps down something that “was only designed for” 1Gbps.
Using the same example space, the fact that 5GBASE-T will work on lengths of cat-5e less than 100m is also a fine example of what is going on with the current telecom TP and what it was “originally designed for”. Its capability was only for voice frequencies, but for length of multiple kilometres. Using shorter lengths, as an FTT{N,C,S,B} deployment necessarily does, is one way of changing the requirements underpinning what something “was designed for”.
In short, harking back to what something “was designed for” is a terribly luddite attitude.
RF engineering 102 will soon bring in technology that will start the question of whether FTT-Home is every going to be cost-effective.
Incidentally, the use of the phrase “balance” in telecom twisted pair is normally a reference to the use of differential signalling. What matters there is that the balanced pair have the same impedance to ground. Twisting the pair usually happens too.
https://en.wikipedia.org/wiki/Balanced_line
Every college kid eventually realises that the 101 course only gives them a basic understanding of a topic. And that, just perhaps, a whole industry has had a century of engineering advancement that a 101 can only scratch the surface of.
But you’re not wrong on the problem – it is indeed a crosstalk nightmare.
No surprise, then, that a big chunk of the research behind G.Fast went into understanding the changes to crosstalk at higher frequencies, and the changes to the models for crosstalk behaviour on our cables.
With better models, and the theory of vectoring now in the “understood” column, it becomes a matter of allowing Moore’s law to catch up, and give us silicon that can vector first at 106MHz, then at 212MHz, and later at 424MHz.
So, pray tell me, what help is vectoring in mitigating external RF being picked up and impacting speed ? And how will it mitigate other external RF devices being rendered useless by the radiation from the cables ? Or is it being luddite to suggest that there should be other uses of the spectrum up to a few hundred megs other than VDSL and Ethernet over Power ?
Cat5e is a whole different piece of engineering. Maximum cable run of 100m. Not to be used outside of a single building. Single “circuit” in single bundle. All in all, an apples and oranges comparison with telco external plant. RF engineering 103 🙂
Openreach just closing down operations for Xmas as usual….
As far as ECI estate goes, I’ve had only one experience of monitoring or using a connection via one of its DSLAM’s ….
That line being one of a fair distance but has shown great stability and reliability, good speeds too…
My experience with Huawei 96/288’s has been somewhat annoying as I’ve found the line switching between various profiles during certain periods such as Xmas (noise) but ECI even though errors have been detected has stayed put and inevitably no issues have arose from it doing so.
So whilst Huaweis estate might be more advanced it don’t necessarily mean those with ECI tech on the end of the line are doomed.
However a common ‘excitement’ from this topic is always that extra speed, I’m not a speed freak, I like stable, low latency and a decent upload…. I prefer my line fast path, something barely existent with Huawei cabs when I left the coverage zone.
Those on any VDSL kit Should always remember…..
You could be forced into being a Virgin Media Cable Customer
*throws up in mouth*
Well it’s now Feb. – Anyone know if the trials have resumed, it’s all gone very quiet. The least ECI could do is supply redesigned cards for the cabinets that actually work properly on the grounds that the old ones ‘are not fit for purpose’. Also BT/Openreach could consider rolling out G.Fast to areas served by ECI cabinets first.
Why did OpenReach use ECI cabinets know they where shit. Where I live it’s a mixture of Huawei and ECI Cabs.
Digital Durham part of BBUK all fitted Huawei Cabs.
I’m on a ECI Cab when I joined BT I was 70Mb/s now because of noise has fallen too 55Mb/s. On a night it fall too 20Mb/s or less.
love to know where the supposed 600,000 lines are. been waiting nearly 2 years for its elusive return…
let’s hope it actually works this time properly.
There’s dozens of ECI lines being monitored live on MYDSLWEBSTATS but absolutely no new G.INP activations on any the ECI lines recently.
It’s a very small sample of lines but spread right across the country.
When this rollout really gets going it will show up there. Worth keeping an eye on.
I know. I am one of them :(……still no sign of it apart from the 2 lines that were enabled in April 2016 and are lucky enough to have held on to g.inp..
@mark.
any chance you can ask for an update?? Still not seeing any new activations on mdws after another month on from your article update. thanks
Anyone seen news on this – seems to have gone quiet. Wondering if it will happen after all. All open reach site says is the trial was paused no mention of resuming.
Surely isps are fed up with more and more users complaining of speed drops
still not active on my line. been waiting since April 2016…..
lol.. just noticed the picture shows a huawei cab.. need to change pic to an eci cab…
Is the author of this article able to check his source for an update on G.INP on ECI lines?
Bad news guys… this update was posted on kitz forum supposedly from an openreach source:
“we have paused the trial of ReTx whilst we work with ECI and their partners to fix a bit loading algorithm problem in one of the chip sets which was causing anomalies on a subset of lines on the trial. We do not have a date for any next steps for this trial as we are concentrating our efforts on working with our supplier and their partners to find a fix”.
Openreach should do the right thing and rip this rubbish kit out and replace with decent stuff, this really does now seem like a two tier service and its just pot luck which side of the fence you are on.
Do you have a link to that please?
the link to the post is http://forum.kitz.co.uk/index.php/topic,21320.0.html