Saturday, 25 May 2013

NBN: Coalition Fibre to the Node isn't pure-digital broadband, isn't secure, isn't meant to last

Not only is the Coalition's Fibre to the Node (FTTN) plan to share two mutually-interfering networks, over existing copper, more complex and expensive that it need be, it also flags they aren't designing it for longevity. Implicit in this design choice is "we're building it to throw away, soon." i.e. with a 10-15 year, or less, economic life.

The network design can never be optimum for either phone or digital/broadband, the combination is more complex, expensive and lower reliability than pure-digital and is missing two critical network design element: it doesn't follow the existing NBN design (standard device interfaces, end-end control & L2 Bitstreams) but ignores that engineering designs can only be optimised to for one thing.
The Coalition's FTTN is not pure-digital broadband, but a hybrid analogue phone/digital broadband, just like the Telstra network's 8800 RIMs. "RIM"s, Remote Integrated Multiplexors, were designed as remote elements of telephone exchanges, initially dubbed Fibre-to-the-Kerb, in the early 1990's. ADSL was added slowly and with a lot of difficultly, ending with the current "Top Hat" conversions: a DSLAM bolted on top. Again a mini-phone exchange, full of complex, expensive compromises.

If the analogue-phone+broadband-on-shared-copper Node approach was good & efficient, Telstra and every other Telco/ISP would've already rolled it out extensively. That even Telstra, the largest FTTN operator in Australia, haven't seen a way to ramp up hybrid-shared-copper FTTN to wide-scale when their strategy in 1995 was to fully complete the transition by 2010, says it is deeply flawed in both Engineering and economic dimensions. The savings in line maintenance must be offset by additional complexity and cost. Telstra are very good at identifying savings and network/operational improvements.

When Telco engineers do pure-digital broadband over copper, they design them to carry Telephony digitally (VoIP: Voice over IP), not as analogue signals, a 100 year-old standard. There are multiple Australian examples:
  • OPTUS HFC Cable, Sydney, Melbourne, Brisbane [1995-7]
  • TransACT, Canberra, VDSL [2002]
  • Adam Internet, Adelaide CBD, 'The Precinct' [2010]
In Engineering Design there is an Ironclad rule: for something to be the best at what it does, it can be designed to only do one thing. Designs can only be optimised for single factors, whereas phone and broadband have diametrically opposed electrical characteristics and the service quality/capability of both must be compromised for them to work together at all.

There's another element of good Digital Network Design missing in the Coalition's FTTN: remote control, provisioning, management and monitoring of every network device, especially including the Customer Premises Equipment (CPE).

The Coalition isn't including a standard VDSL2 modem in its design. Turnbull, when quizzed on this, says, "Consumers can buy a $50 VDSL modem if they want". There is so much wrong with this (price, features, compatibility, certification, filter/splitter, install & line-test) that it needs to be dealt with separately.

Any competent Telco or Network Engineer will say that this "BYOD" (bring your own device) approach is "mad, bad and dangerous" for multiple reasons:
  • remote control, management, administration, diagnosis and monitoring is not possible.
    • This is central to efficient Telco operations these days.
  • DIY installs are notorious for being incorrectly, poorly or incompletely done.
    • This is exactly why only trained/certified technicians are allowed to work on the Telco side of any service.
    • The VDSL2 modem becomes the edge of the Telco network in the Coalition's FTTN, a fact they seem to overlook, but will have severe engineering & maintenance ramifications.
  • To be done properly, a Central Splitter needs to be installed in-home for every FTTN service.
    • Because the Telco, not premises, wiring has to be cut, this has to be done by a registered cabler, so any DIY advantage of BYOD is gone.
    • Done as an "on-demand" retail service, this will cost householder directly 10-100 times what NBN Co could do in a co-ordinated mass deployment. This is why after its trials, NBN Co elected to run Fibre to every premise in mass rollouts. "Do it right, Do it once".
The Coalition FTTN design also violates very strong design principles already at the heart of the NBN design, compromising the ability of NBN Co to efficiently and effectively run its own network:
All network devices look identical to RSP's & have identical capabilities, all connect identically at the Points of Interconnect (PoI's), all can be provisioned, managed & monitored remotely by NBN Co's central NOC (Network Operations Centre), all are secure and updatable/controlled only by NBN Co. Customers can never access the owner password of any network device on the NBN, only their own attached devices.
All the high-speed DSL work in labs is for pure-digital, not hybrid analogue phone+broadband. The necessary filters/splitters reduce high-frequency performance, increase noise, add multiple points-of-failure and provide a source of hard to trace "unforced errors". This includes compromising "Vectoring", the great white hope of Turnbull for reaching reasonable broadband speeds.

Broadband Network Operators interested in long term services, deploy pure-digital networks, not
hybrid analogue phone/broadband. These hybrids are "neither fish nor fowl". They aren't optimised for either analogue phone or digital services, nor can they be as they are opposites (DC+low-frequency AC, vs pure-AC high-frequency). They have all the problems of both and a brand new set of problems unique to making them work together. Their only justification is "sharing a single copper pair". In an age of very cheap VoIP chips and ATA's, the economic advantage for primitive filters/splitters has also dissolved.

Needless to say, the total cost of deployment, Network + Customer out-of-pocket costs, is much higher when all the extra work of cabling, patching and filters/splitters at both ends, is included.

This critique raises a question:
If hybrid analogue phone+broadband & BYOD is so flawed, why would the Coalition even suggest it?
Either they don't know better, they are slavishly following another design (e.g. UK's BT/Infinity), they are sacrificing durability and robustness to appear 'cheap' or for some political end.

We have to assume that Turnbull has recruited experts, even some competent in DSL & broadband. From his many public references, he is highly influenced by British Telecom's, as yet an incomplete experiment, though he likes to imply, or may even believe, it's a shining example of a working FTTN network. It definitely is far from that.

The Coalition made the NBN a political discussion in 2005, when Howard first rejected the best expert advice on the planet and ignored Sol Trujillo. It then backed this up with multiple sub-standard and unimplementable programs, such as OPEL. What ever happened to the billions from the Telstra sale promised for "broadband in the bush" to get Barnaby Joyce on-side? Country folk still live in a digital desert, 8 years after the T3 sale. We've got nothing to show for Barnaby's efforts.

In the 2007 and 2010 elections, the Coalition furthered this politicisation of basic broadband access with their costly, inadequate and unworkable policies. Unsurprisingly, this contributed to their failure to win those elections.

In 2013, the Coalition is continuing more than a decade of indifference and ineptitude in NBN policy.

The hybrid analogue telephone + digital service over single copper twisted pair FTTN solution proposed by the Coalition is nothing more than a "polished turd". It looks almost shiny, but is weak, smelly and nothing you'd personally want anything to do with.

This is a purely political design intended to seem "good enough" on brief inspection, but fails abysmally as good Engineering design at every level. The pity is that with a very small amount of effort, the cost could be lowered, the service improved and it brought up to acceptable engineering standards. It almost looks like they chose the worst design possible. Very, very strange.

The whole of this flawed design is driven by a single political objective:
Installation without disruption or technician visit, even though to get VDSL this is needed! This, very bizarrely, is to allow anybody with a 1925-compatible handset to not be affected by the rollout of a modern broadband network. Without exaggeration, this is as idiotic, imbecilic and perverse as insisting that only tyres suitable for ninety yaer-old Model-T Fords could be sold in Australia.
An NBN-design compatible FTTN would look like Adam Internet's VDSL2, but with a standard NBN Co Network Termination Device (NTD), not a commodity PPPoE modem. This does require, just like the FTTP, in-home work by technicians. It avoids installing filters/splitters at both ends, removes around two-thirds of the Node equipment (ATA's for phones are in the NTD) and allows technicians to optimise the copper pair for pure-digital broadband. And those 1925 rotary dial handsets will still work with the NBN Co NTD's. Cheaper, Better, Faster: what's not to like?

There is a very simple way to test my assertion about hybrid analogue/digital being an unacceptable engineering solution:
In Enterprise networks, or even those of SME's, do we ever see anyone sharing analogue phones and digital signals (ethernet) on the same copper cables? No, not ever.
There was a brief period in the early days of twisted-pair ethernet (1993/4 with 10Mbps & Cat 3/4 cabling) when a few people flirted with sharing the 4-pairs in their cabling between telephones and digital networks. Mixing 5 volt and 50 volt services was really bad news (users plugged things in the wrong way around and blew up expensive gear) and when 100Mbps came along needing all 4-pairs and Cat 5 cabling, all this silliness ended. Everyone installed office buildings with "structured cabling", RJ-45 patch panels and multiple RJ-45 outlets per desk.

These days, Enterprises are moving away completely from analogue telephony. They are replacing their PABX's, handsets and phone wiring with VoIP phones, soft-phones on computers and "soft-switches", all run over their pure-digital network. Both single campus ethernet and their Wide Area Networks: Enterprise telephony is moving to "just another service" carried on their pure-digital network. Telcos know this and know they need to follow. The Coalition FTTN design is antithetical to this goal, it puts us further away, not closer, to a pure-digital network.

There is a widely-used standard that uses standard Cat 5 ethernet cabling for two purposes: Power over Ethernet (PoE). It adds around 20V-DC onto standard 4-pair cable for 10/100/1000Mbps ethernet without endangering standard equipment or affecting ethernet performance. You can plug your PC or laptop in 100% safely. CISCO, the largest Enterprise & Telco/ISP networking vendor, invented this technology to remotely power it's VoIP handsets. Modern digital telephony doesn't include any analogue 50V signals, a fact that seems to have escaped the brilliant engineering brains responsible for the Coalition FTTN design.

This is much more than the Coalition shunting considerable hidden costs onto unsuspecting householders, this is deliberately choosing a flawed Engineering design whose total deployment costs will be higher than a good design, will be much more expensive to run and maintain, is designed to be insecure and will deliver a much inferior service over copper than is necessary.

The Coalition FTTN network is solely driven by political considerations at the cost of good Engineering design and sound economic. It could be worse, but I'm not sure how...


  1. Hi Steve,

    Your article here got me thinking. I wonder if the Libs are using this as a bait-and-switch.

    They'll take FTTN to the election, then after their cost benefit analysis is done, 'due to overwhelming demand', they will switch to FTTH . They'll make a few small changes to make the rollout theirs, not Labor's, but proceed with FTTH.

    That could be why the Nationals aren't kicking up much of a fuss - they know they'll be getting FTTH.


  2. David,

    It could be bait-and-switch, but why would they undersell a policy/product designed to get them votes??

    Many people in the country are likely to get Fibre, unless you're in a town (FTTN) or a small town (nothing). It's the "Cost-Effective" clause.

    If the question is Fibre-to-the-Home or 4G mobile wireless for similar prices, I think most people would go for the Fibre.

    If the question is Copper-from-the-Node (FTTN) or 4G mobile wireless, I think most people would go for 4G.

    I can't see their FTTN solution being popular or long-lived. Telstra and Optus will move aggressively against it, undercutting NBN Co's revenues and profitability.

    Remember that the NBN single-price model only works when there's 90+% take-up overall and people in the City subsidise the Bush.

    That's been struck-out in the Coalition policy.

    So, yes I agree, the Nationals *think* they'll FTTH, mostly.
    But I can't see this half-arsed FTTN model having enough takeup for it to even break-even.


  3. I don't think it's bait-and-switch, but I do think that there's a chance they'll change their tune before the election. People like you are tearing their approach to shreds. On the other hand, it's also possible that they're starting with the "cheapest" option to see how people like it.

    It's not clear where you see fibre being installed. Not in towns, not in small towns. Out in the bush? That's where I am, and not even Labor NBN is planning fibre there. Instead we're getting fixed wireless, hopefully before mid-September.

    On the other hand, given only the choice between FTTN or 4G/LTE, I know that I'd choose FTTN. I've been connected via 3G for a few years now, and it's just not dimensioned to provide continuous reliable service. I don't see that changing in the immediate future.


  4. Greg,

    Good observation. Thanks.

    I was *assuming* that 4G would be engineered as an ADSL killer. Maybe not.



Note: only a member of this blog may post a comment.