Building a Fibre NBN on a Copper budget

On 17th July 2013 I delivered a talk at the CommsDay Wholesale and Data Centre Summit in Sydney about the NBN called “The Ideal Wholesale NBN Market”.

This talk proposes just a few of the many ways in which the build cost, build time, and operating cost of the FTTH NBN could be lowered (perhaps dramatically lowered) by auditing the entire existing design and by applying the fruits of some lateral thinking about what is really needed to make the network work – and what the network can do without.

Here is the PDF slide pack for my talk:

FIbre on a Copper Budget

Here (below) is an embedded video that ‘plays’ the presentation slides synchronised with the audio recording of the presentation ‘as delivered’.

This is a much better way to appreciate the content, so if you can spare 20 minutes, please click on the video below – and I hope you enjoy it:

10 thoughts on “Building a Fibre NBN on a Copper budget

  1. Wow, I’ve been following the NBN/UFB stuff a bit from finnish perspective and history in the internet business since early 90′s, but I never dig deep enough to realise there’s really a battery backup and POTS port included. I never expected that as the NBN project looked pretty good from the outside.

    Meanwhile up here in the north I’m paying 14 AUD worth of euros for a 100/10 interweb in the wall, provided as a single CAT6 RJ45 ethernet port (terminated to a switch nearby, there’s also fiber to the wall but it’s there for future proof). There’s other carrier alternative behind the CAT6 as well in case I get fed up with the current one.. The number of POTS is going down very fast, I haven’t had one for ~10 years, mobile calls are way cheaper.

  2. Pingback: Simon Hackett fixes the NBN cost problem | techAU

  3. Hi Simon, it sounds like a sensible solution, how would your scenario handle alarms, health alerts and other systems that require the 24*7 (almost) availability that POTS provide. Also how would it speed up the rollout timetable, I am in area of Adelaide (Flagstaff Hill) that is not even on the NBN rollout map and even though I can see the CBD from my lounge window I find my home is 7500 meters from the exchange in Brighton with woefully inadequate download speeds. If one member of the family watches an SD video, the other members of the family are back to dial up modem speeds, if the internet gods are smiling. Note: our download speed is around 2,500 mbps. As someone who has been in the IT industry all my working life (I am 57) and pay close attention to technology developments in this area I find it disappointing that with the NBN’s massive cost to the taxpayer, I will likely be passed caring before it comes to my home. Cheers

    • Hello Phil,

      There are multiple ways to handle alarms, health alerts and other systems that have tended, in the past, to use the PSTN.

      The approach is the same as that for PSTN voice emulation – as per the side on that issue in the talk.

      Specific examples:
      - You can use a fixed wireless PSTN emulator that carries voice and data over the mobile network instead (as I said in my talk – think in terms of a mobile phone with no screen and an RJ11 analog jack on the side instead of speaker and microphone). Carriers around the world have been deploying these things for a long time.

      - In the case of any modern device in the list you provided, most of them now support either the direct use of a mobile SIM in a mobile network access module built into the device, and/or they have either ethernet or WiFi and can simply reach out over the Internet instead of using the PSTN at all.

      - Alarm systems, as a specific case, are increasingly available with ethernet jacks for Internet access. And if your alarm doesn’t support that, just ask your alarm company to install an emulator module on the phone line port of your existing alarm. A small module that pretends to be the phone network as far as the alarm is concerned, and that converts the resulting ‘phone calls’ to the monitoring centre into an https connection over the Internet. You can also get those modules with a SIM slot to use the mobile network either as primary or backup to an Internet link.

      [Thats how I do it at my house, and if your alarm company doesn't understand the question, find a better alarm company. I'd recommend, in my own home state of South Australia, that you try MSE Alarms, for being an exemplar of doing this stuff really well - they've done a great job for me :) ]

      - Health alarms – likewise, they can work in the same way as noted above. If your monitoring company doesn’t ‘get’ this, find a better one. Just call it evolution in action.

      - Finally, you can just do *exactly* what the Alcatel NBN NTU does – you can use a PSTN to VoIP adaptor. These are what the VoIP industry calls an ATA, there are a huge plethora of them on the market, and this is simply the PSTN emulator piece of the Alcatel NTU as a separate box. So just buy one of those, plug it into your ethernet network, and plug the analog device into the analog port.

      [Note also that most of the better VoIP-capable Internet routers do this very well as well - certainly the Fritz!Box is excellent for this sort of thing - including such legacy applications as plugging in a fax machine].

      The bottom line is that the PSTN ports in the Alcatel NTU aren’t magical. You can just use another ATA instead, in the rare (and becoming rarer) instances where the device you want to work with can’t simply ‘speak TCP/IP’ natively. And I can tell you that over the next 5 years, the proportion of such devices that *will* just ‘speak TCP/IP’ is only going to rise.

  4. Pingback: Building a Fibre NBN on a Copper budget | PK's Blog

  5. Simon,

    That was fantastic and spot on. I gave up trying to talk to NBN Co crowd and got tired of being vilified. We have been doing just what you suggested and a lot more in other parts of the world and it is with sadness that we watch Australia’s efforts at fibre sinking into the inevitable mire.

    We are doing designs that bring the home connected cost right down, and use a single port ONT. We also try and involve the local communities by making the solution low tech and are even getting phenomenal value add, and I might mention a far more robust network by using old tyres for example.

  6. I’ve seen various comments via other social media paths that move me to write down a few more things here.

    I only had a 20 minute speaking slot for that presentation (its how that conference works) so I had to be very selective with what I could say (and what I had time to say) in that period. This meant that I had to assume the viewers of the talk (a telco conference) understood the environment at least a little, and that I could rely on them understanding some points that seem to be new to some of those who have responded subsequently (and that have lead to some substantially incorrect contentions as a result of that lack of domain knowledge).

    So – Here are some points that didn’t make it into the talk but that are relevant to some of the comments about the talk that I’ve read to date (elsewhere) – in the hope that this might help a little more…

    * The fibre infrastructure being built is GPON. The fibre that appears in your home/office isn’t dedicated to you – its shared between (typically) 32 premises in your local area. The way this works is that one upstream fibre port run by the NBN delivers data via an intermediate point where a passive optical splitter literally shares the beam out to 32 (or more) separate single fibres, each of which then lands in a destination home/office. The downstream data rate shared across those premises is (today) 2.4 gigabits/second downstream and 1.2 gigabits per second upstream. This is upgradeable later.

    [and this is quite technically comparable to the shared-last-mile nature of HFC networks]

    A good treatise on this stuff for those interested can be read here:

    Those who find the concept of fibre being shared between multiple house to be a surprising thing should really have a read of the link above – please. If you don’t really have that straight in your mind, then much of the rationality of optimising the network termination approach might be lost on you.

    * My proposals that the ONT can be cut down from a 6 port device to a 1 port device (option 1) or that it can be cut down to no NBNCo-supplied ONT at all (option 2) aren’t exactly radical, but I’ve seen contentions made that I’m somehow trying to lock customers into a single provider – which is just wrong:

    - If a site really needs multiple independent delivery ports, then just ship and install the NBNCo standard NTU, but do this *by exception*. Nothing stop this happening. But for the other 99.9% of sites, why not do it faster, cheaper, and more flexibly by letting you plug the fibre straight into a GPON-port equipped customer router?

    - Alternatively, the GPON network supports there being passive split points at two levels between the upstream port and the consumer. For instance, it should be possible to use a 1:2 or 1:4 or 1:whatever PON splitter in the premises to allow multiple separate services to be delivered, too (though in practice ,just delivering an NBNCo standard box in such situations is both pragmatic and simple, given that the box already exists).

    - There is no lock-in to a single provider implied in any of this, because (exactly as for ADSL for the last decade), if you want to churn providers, you just do that. And a GPON router works exactly like a DSL router – just enter the login details for your new provider and get on with it… The industry has got a decade of experience in doing this already. No lock in, nothing particularly hard happening here.

    * The pricing of second and subsequent ports on the NBNCo termination box is exactly the same as the first port

    This seems to be lost of some critics of the concept of optimising the cost and rollout speed of the network.

    It means, for instance, that no… you won’t wind up with pay tv services selling you a box on a separate port, because you won’t want to pay another $30-$40 per month for that second port to be lit up, just to get exactly the same outcome as you would get by buying that service over the Internet link supplied on port 1. Thats just non-sensical.

    The same thing extends to other single-purpose uses of other ports. Sorry, but smart-metering happens over either (a) the Internet path or (b) a mobile network enabled device with a SIM in your meter box. Your power company isn’t going to pay an extra $30 per month to NBNCo in order to use a dedicated port to read your meter once a quarter, nor will they pay to run new wiring from your NBNCo box to your meter box up front. This really doesn’t make sense.

    Think about the real world you live in right now – what you do in these regards, you do ‘over the Internet’ or over the cellular network. The demand for installing multiple totally independent, and ‘each-fully-priced-from-scratch’ links in one physical household are extremely rare.

    Again – if they really are needed, then put an NBNCo NTU in ‘by exception’. Each. But otherwise, even in (say) a shared household – are two people in that household really going to want to buy separate fully priced Broadband services in parallel, when both people wind up paying at least $30-$40 per month more (in total) for doing so, over and above their usage based costs? Really?

    * I’ve seen some contentions in other social media that delivering ‘just the bare fibre port’ somehow locking a customer in to a single ISP (false). If you change providers, just change – and churn – exactly as for DSL. And put the new ISP details into the GPON customer router you already have, or swap that box out for another one. Not exactly hard, is it? And we’ve been doing this with ADSL for more than a decade already.

    * I’ve seen contentions that this sort of approach doesn’t make much difference to the cost because the major it cost is in other things (like rollout labour). However:

    - NBNCo is already deploying using multiple pass rollouts – with a fibre team doing the fibre drop, and a separate team coming past later to do the electrical wiring and installation of the NBNCo ONT. I discussed this very thing straight after the talk with one of these rollout companies!

    And then… the ISP does a delivery pass of their own to deliver whatever hardware they supply with their service.

    Getting rid of one of these two NBNCO truck rolls is a highly non-trivial labour cost and time saving. The latter especially since NBNCo contractors (then) don’t need to deal with/resolve the electrical wiring and power supply requirements of the NBNCO NTU any more. Let alone the battery and its associated costs.

    Ask any contractor whether the think the rollout would be cheaper if they could halve the number of truck rolls nationally… and they’ll all give you the same answer: “Duh”.

    - *any* cost for the NBNCo ONT and battery, when multiplied by millions of premises, is a hell of a lot more expensive than $0.

    And I suppose a last set of points for people who may disagree with me, and here it is…

    Disagreement is just great – and in general debate is a good and healthy thing.

    Please just be prepared to do two things:

    - First, don’t attack me, attack the idea if you don’t like it – ‘ad hominem’ is pretty lame and un-necessary. It also tends to be a shield against the second point;

    - Second: If you have a better way – please – pretty please – stand up and do your own presentation about it, and I’d love to take a look at it. Don’t just attack me for standing up and trying to help :)

    This presentation was just meant to be a conversation starter. The real cost optimisations can only happen from the inside – from a ‘soup to nuts’ deep audit done with full access to every aspect about the manner in which NBNCo are operating – which is clearly impossible from the outside.

    Thats why I limited my suggestions to things that were obvious from the outside – but still managed (I believe) to provide a sense of the savings that are there to obtain if the will is there to do so.

  7. Hello Alan,

    My sincere apologies. The image was handed to me by a third party and I hadn’t figured out the original source at the time that I delivered the talk.

    I am only too happy to acknowledge the source of the image – as per your post above.

    I hope that its ok that I used it, not withstanding that I now find myself asking forgiveness after the fact target than having obtained permission ahead of time.

    Best wishes,
    Simon Hackett

Comments are closed.