Thursday 1 August 2013

Debate: rebutting Turnbull - New Technologies


Business Spectator, part of News Ltd "The Australian", posted a review of the Kohler v Turnbull NBN Debate. The Debate is also available as streaming video.

This is a response to these comments of Turnbull on "New Technologies":

The shadow minister said there are new technologies to deliver broadband his party has not yet canvassed.
Five years ago, there was a "very big difference" in service level between ADSL2+ and fibre to the premises, he said.
But in recent years, "that difference has compressed".
"There is a point at which increases in speed cease to have any marginal utility."
Just where are these mysterious New Technologies that are constantly invoked by Turnbull?

If they aren't in the Labs and demonstrated as feasible, then they cannot appear in our homes in 10 years. But for every "great idea" demonstrated in a Lab, only a small fraction make it into production.

There are four things that Turnbull could be talking about, that are in the Labs and demonstrated:

  • Fibre transceivers, Terabit is currently demonstrated, 100Gbps in 96 WDM, for 9.6Tbps total, currently in production. Not that he's interested in reality.
  • Radio for "5G" Mobile. 1Gpbs demonstrated. Useable over what distance? What band? How many users per cell? What power and compute requirements?
  • Radio over Twisted Pair Copper: DSL variants faster than VDSL. 1Gbps has been demonstrated on good copper over modest distances. What power and compute requirements? What cost for both ends?
  • Magic or Psychic Transfer. Only demonstrated in the minds of loonies, making it possibly "credible" to some folk.
I also dispute his assertion that "In the last 5 years, the difference between ADSL2+ (24Mbps) and Fibre to the Premises has compressed". His comparison, not defined in his usual fashion, could be VDSL2 @ 100Mbps versus 1Gbps GPON, the standard used here for Fibre to the Premises. The comparison is very specifically selected to exclude the advanced in production of the majority of Fibre Optic transceivers current in use - through an invalid and unstated assumption: Only Passive Optical Networking (PON or GPON for gigabit PON) can be used to houses. This is doubly false.

VDSL2 vs GPON is an invalid comparison: he's been exceedingly selective in his choices, freezing one, Fibre (GPON), and not the other (VDSL2). The trouble with VDSL2 is that nobody knows how it performs in mass-use. Deploying VDSL2 so that it may be used by customers, is a far cry from the majority of people using it on a daily basis. Engineering is all about the gap between Theory and Practice.

If he were to compare volume production Twisted Pair Copper and Optical Fibre transceivers, he'd be absolutely wrong. Five years ago, 10Gbps was the fastest production Optical Fibre, with 10km and 40km transceivers available. That's gone up to 40Gbps and 100Gbps. All these are available in Wave Division Multiplexing (WDM) - or light of many colours - from 8 to 96 times the throughput.

Also, Turnbull fails to mention the critical factor, the other half of Engineering, price.
There's a very simple comparison that tells you everything about the two technologies:
For production systems, what is the price of an extra 1Mbps? For Copper, it's around $2 and increasing. For Fibre, it's one thousandth of that and falling.
What's the reality of the two technologies, Fibre and Twisted Pair Copper? Copper is slow and expensive and getting more expensive, Fibre is cheap and getting cheaper.

If Fibre Optic transceivers are in volume production for any product, then they can be adapted for the Fibre to the Premises (PON) system, or PON simply abandoned and WDM ethernet used.

His argument is also at odds with his previous statement that "new technologies" will redefine how we do things.

The only technology that is still progressing along at Moore's Law rates and the per-Mbps cost is still decreasing is Optics Fibre transceivers, and if fully informed, Turnbull would know this fundamental.

While Twisted Pair and Free Air Radio ("5G") are improving bandwidth, the absolute improvements are unremarkable, while the costs are increasing, significantly, not reducing.

Turnbull has created a false comparison and then failed to address the Engineering and Financial fundamentals: Accelerating increase in Speed & Error-Rate and accelerating decrease in Cost-per-unit.

1 comment:

  1. "Difference has compressed" LOL. So Tumball is now hyping 1gbps over copper, simply amazing. Remember before the last election NBNco announced 1gbps plans would be available. We had Abbott claiming it was "utterly implausible" and the rest putting it down to an election stunt.

    Now apparently 1gbps is the the new sexy everyone wants and it's not "implausible" "from something that isn't even built yet" at all... OR an election stunt...

    Of course the thing is now instead of having to justify wasting 30+ billion taxpayers money for "25-100mbps" speeds with no apps to take advantage of that extra speed over ADSL2+ Tumbull now has to justify spending 30+ billion taxpayers money for 1gbps speeds with no apps to take advantage of that extra speed over ADSL2+. This is quite a conundrum for copper zealots too.

    ReplyDelete

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