Skip Navigation


Huawei VDSL2 cabinets give better speeds than ECI
Thursday 19 January 2017 15:58:53 by Andrew Ferguson

On average looking at broadband speed tests for people connected to VDSL2 cabinets there is a 2.1 Mbps download benefit to being connected to a Huawei cabinet versus an ECI one based on December 2016 speed test results. We have run this analysis on speed test results going all the way back to the start of 2012 and right until October 2015 there was very little in it.

Speed test Comparison for VDSL2 tests on ECI and Huawei cabinets
 ECIECIHuaweiHuawei
DateMedian Download
Mbps
Mean Download
Mbps
Median Download
Mbps
Mean Download
Mbps
Dec 2016 27.8 30.5 31.4 32.6
Nov 2016 28 30.5 31.3 32.5
Oct 2016 27.5 30.1 30.8 31.9
Sep 2016 27.2 29.5 30.2 31.4
Aug 2016 27.7 30.2 30.4 31.7
Jul 2016 28.7 31.2 31 32.2
Jun 2016 28.6 30.5 30.6 31.5
May 2016 28.2 30 30.5 31.3
Apr 2016 27.8 29.6 29.7 30.5
Mar 2016 27 28.7 28.9 29.8
Feb 2016 27 28.9 29.3 30.2
Jan 2016 27.1 28.8 29.1 29.9
Dec 2015 27.5 29 29.1 29.8
Nov 2015 27.4 29 28.9 29.9
Oct 2015 26.1 27.3 26.6 27.5
Sep 2015 26.3 27.5 26.5 27.6

The pattern is roughly followed by upload speeds, with Huawei areas recording better upload speeds compared to ECI areas since summer 2015, though the difference is smaller with a gap of 0.2 Mbps to 0.8 Mbps. The Huawei cabinets also recorded a 3ms lower small TCP packet latency average, and while the quality metric measured during the download test is very close, Huawei has been slightly ahead since July 2016 after years of being neck and neck.

Unfortunately mere mortals do not get to choose which cabinet is installed in your area, but looking at the work for the BDUK areas ECI only accounts for 6.3% of the cabinets installed and outside that 42% of the commercial cabinets deployed are ECI based. This suggests ECI cabs are not deployed very often if at all these days.

One possibility for the better speeds is if Huawei cabinets are used in the more densely populated locations and we crunched the numbers again to learn that for Huawei cabinets our coverage model tells us 94.1% will get 30 Mbps or faster, but ECI users had a better chance at 97.3%. So this means that even though Huawei cabinets appear to be favoured in areas with more people at significant distances from the cabinet better speeds are being experienced in the real world.

Distribution premises covered by ECI and Huawei VDSL2 cabinets across the UK regions
 ECIECIHuaweiHuawei
Region% VDSL coverage% superfast
Over 30 Mbps
% VDSL coverage% superfast
Over 30 Mbps
East of England 33.9 32.9 55.1 51.7
East Midlands 44.8 43.8 47.5 45.3
London 15.8 15.5 70.8 69.4
North East 46.5 45.8 35.1 33.9
Northern Ireland 10.5 10.2 84.3 65.9
North West 36.1 35.3 55.9 53
Scotland 39.3 38.3 45.9 41.9
South East 29.2 28.5 63.2 60.4
South West 38.1 36.3 47.5 42.9
Wales 31 30 59.3 54.8
West Midlands 39.1 38.3 50 47.4
Yorkshire and Humber 43.8 42.6 35.7 33.5

So why the better speeds? The enhancements of G.INP and Vectoring and the possibility that the core VDSL2 chipset in Huawei cabinets is simply performing better are largely to explain.

So why spend time running this analysis? Well it highlights that not all VDSL2 kit is identical, and highlight the importance of using hardware that supports G.INP and Vectoring for your VDSL2 modem so you can benefit from the Huawei effect. Also looking to the future March 2017 should see much wider use of lower 3dB, 4dB and 5dB noise margins for lines that are stable enough, thus boosting some peoples speeds further and by doing the leg work now it will be easier for us to spot whether the new low margin roll-out is having an effect. The difference between ECI and Huawei performance once combined with the low margin roll-out may mean we adopt different models based on type of cabinet deployed where this is known for deciding the coverage levels across the UK - hence keeping a keen eye on things.

Comments

Posted by philipd 3 months ago
Vectoring isn't used except in some very rare cases so I wouldn't think that is providing any advantage to the vast majority of FTTC customers.

Other factors not touched on are probably the CPE chipset used, which mostly seem to be Broadcom based on most kit supplied or purchased, which pair better with the Huawei cabs, losing a little advantage when used on ECI cabs.

G.INP is going to make a difference, and it will be interesting if the gap closes again when BT re-enable G.INP on ECI cabs, if they ever do.
Posted by andrew (Favicon staff member) 3 months ago
The divergence seems to be around the time G.INP was rolling out, but did not expect it to be so visible, hence the length drop of data on the observations.

Posted by WWWombat 3 months ago
First ... great analysis. Thanks!

Good to set this baseline prior to lower SNRM margins being experienced.

Intuitively, I'd agree that the difference is likely to be explained by G.INP.

The comparison of latency really helps back that up... an average of 3ms might mean that 30-40% of premises managed to shift from old-style INP to new-style G.INP protection.
Posted by WWWombat 3 months ago
Can I add that it looks like Huawei improved in 2 steps - in Nov 15 and May 16.

While Huawei's might have gained G.INP in 2015, I think the DLM algorithm's shifted in 2016 (coinciding with the 2nd attempt at G.INP for ECI). Certainly the banding process became worse about then, but perhaps other things improved.
Posted by adslmax 3 months ago
I was lucky to have Huawei cabinet
Posted by 2doorsbob 3 months ago
I wonder if the 55/10 profile also helped speeds in 2015/16 as this is also still quite new
Posted by chrysalis 3 months ago
Am surprised this data has been released, as openreach wont want people moaning about ECI cabinets.
Posted by adslmax 3 months ago
Openreach should replaced all ECI to Huawei cabinets instead for the better for everyones.
Posted by AndyCZ 3 months ago
This data has nothing to do with Openreach.
Posted by Diggory 3 months ago
"Well it highlights that not VDSL2 kit is identical" 'Not all', perhaps?

Also FYI your Delicious links are 404.

Great site though.
Posted by AndrueC 3 months ago
Huawei?

Much better thanks.

(I'm here all week, folks) :)
Posted by andrew (Favicon staff member) 3 months ago
Delicious problem fixed
Posted by Mince1978 3 months ago
How can I tell which kind of cabinet it is in the village.?
Do they look different?
Thanks
Posted by WWWombat 3 months ago
Take a look:
http://www.kitz.co.uk/adsl/fttc-cabinets.htm
You must be logged in to post comments. Click here to login.