Kitz ADSL Broadband Information
adsl spacer  
Support this site
Home Broadband ISPs Tech Routers Wiki Forum
   Compare ISP   Rate your ISP
Please login or register.

Login with username, password and session length
Advanced search  


Author Topic: BT's Rate -vs- Reach for G.Fast  (Read 1358 times)


  • Kitizen
  • ****
  • Posts: 1674
BT's Rate -vs- Reach for G.Fast
« on: October 09, 2017, 11:19:51 PM »

I found this graph (attached below), which I think can be used as the best approximation for what BT believes to be the rate -vs- reach for G.Fast.

Source? A NICC document, "Guidelines on the Use of DSL Transmission Systems in the BT Access Network", ND1405 rev 4.1.1. Not quite a formal BT document, but you can bet that this content was sourced via BT's Kevin Foster.

I found the document while investigating UPBO for VDSL2, but it also includes a section for G.Fast added in July 2016. Unfortunately, it may or may not represent the speeds available via the amendment 2 improvements - it makes no mention one way or another.

The graph is of interest, as it shows that BT intend to use frequencies from 19MHz (not 23MHz, as a lot of speed graphs are based on). However, there is also a section on locating G.Fast nodes ... which probably interests many:
For, the ANFP specifies what is required to operate at a G-SLCP either co-located with an existing PCP (where VDSL2 may have been deployed) or closer to the customer premises.

It is recognised that ubiquitous ultrafast coverage is not possible using equipment solely located at existing PCPs, and that co-ordinated group size limitations may impose an upper bound on the level of service take up from locations supporting large numbers of lines. For these reasons, deployment closer to the customer premises is likely to be necessary.

The complex optimisation of new injection points in the D-side network would need to account for length distribution of sub D-side (affecting achievable performance), deployment cost, and target service offer. This includes the number of lines servable from a given location or co-ordinated group (affecting maximum service take up), deployment cost, local siting and power issues, self-install and equipment chosen. Such practicalities will complicate real deployments. The following illustrates some of the factors in the trade-offs involved in optimising availability, service speed, and cost:

Figures 4 and 5 in [4] include an aggregated view of the D-side distribution of the Openreach network. Figure 33 shows an indicative rate versus length curve for operating starting from 19MHz, although actual performance will vary greatly depending on the chosen equipment and deployment locations.

The reference to "Figures 4 and 5 in [4]" relate to the Sagentia document on line lengths, and fig5 is what I based my well-posted annotation (also attached below). I guess this makes for a quasi-official acknowledgement that the Sagentia results aren't wrong.


  • Global Moderator
  • Senior Kitizen
  • *
  • Posts: 22582
  • Over the Rainbow Bridge
    • The ELRepo Project
Re: BT's Rate -vs- Reach for G.Fast
« Reply #1 on: October 09, 2017, 11:47:28 PM »

Very interesting. Thank you for finding that document and providing an analysis thereof.
:cat:  100% Linux and, previously, Unix. Co-founder of the ELRepo Project.

Please consider making a donation to support the running of this site.

Black Sheep

  • Helpful
  • Kitizen
  • *
  • Posts: 4787
Re: BT's Rate -vs- Reach for G.Fast
« Reply #2 on: October 10, 2017, 11:52:52 AM »

Interesting they've used CAD55 as the cable model. Unless I'm missing something, CAD55 in my world is a 4-pr drop-wire .... actually it is classed as 'Aerial cable', due to the multi-core aspect.


  • Kitizen
  • ****
  • Posts: 1674
Re: BT's Rate -vs- Reach for G.Fast
« Reply #3 on: October 10, 2017, 01:15:21 PM »

That's what I've taken it to mean too, with 0.5mm copper.

I think this cable is used in some of the graphs because it is one of the cable models formally included by the ITU-T in G.Fast. A lot of the original research work to model cables came from BT and TNO (Dutch), so the models tend to have included cables that they have been working with.

I think the Dutch cables tend to come in quads rather than pairs, so their models have very much focused on the interaction between pairs. From what I have read, it seems the new models (over and above the older DSL models) take into account more capacitance between pairs, so it isn't a surprise that BT would want to model a cable with a few pairs.

The models date back to when G.Fast was for the DP, so aerial drop wires make sense. With their refocus, perhaps they've got more models for underground D-side cables.