Tuesday, 12 March 2013

NBN: Biased reporting at the ABC, or the Coalition shooting the Questioner?

Media Watch roasted an ABC Technology journalist whom I've always found considered in his reporting, well researched and unbiased. To me, this is a result of Turnbull's expertise as a Barrister, attacking people and ignoring pertinent, but inconvenient questions...

The journo has had the temerity to do two things:
  • ask the right questions, and keep asking them, and
  • not lie down and accept the Turnbull-Fletcher party line.
Here are how I would phrase some of the key issues on the Political discourse over the NBN.
  • Why is the NBN important? For Economic Productivity (businesses), not watching videos on-line at home. That's a secondary benefit of relatively low-value.
The Libs purposely conflate the NBN with the economically unimportant residential deployment.
  • We have a perfect model for a large-scale green-fields VDSL-FTTN deployment as well as Telstra/OPTUS HFC, and they neither is economic: Transact (Canberra where I live).  Eventually the FTTN network written down 75% and acquired by iiNet last year.
Even with perfect copper, in the richest, most savvy market in the country, is NOT economic by Telco standards.
  • Mr Turnbull always avoid the central question, necessary for a Cost Benefit Analysis:
What's his Demand Forecast model?
If he disputes CISCO's VNI model, what model will he use?
The VNI model forecasts both average peak rate demanded and aggregate download volumes.
  • Turnbull also avoids the reciprocal question, also necessary for a Cost Benefit Analysis:
        What's his Supply-side Forecasts for each of his technologies?

Both of these forecast models, with hi/low ranging, are required for his Cost Benefit Analysis.
Turnbull knows this stuff from his time in Merchant Banking and Technology investment, he just chooses to ignore the questions, choosing to attack & deride anyone who asks them.

There's a great line by Scotty in StarTrek, "you canna break the Laws of Physics, Captain".

For fixed line services, you can use "transmission line" (aka wave-guide) technologies or "Free Air", wireless, either RF or laser or sub-space carrier from StarTrek.

There are three fundamental laws of Physics that apply:
  •  Shannon's Law (IIRC) says for a given SNR and bandwidth, there's a maximum symbol rate (not sure how they derive the error rate, but its related to the Noise Margin. More noise, more errors).
  •  Transmission lines have a "bandwidth-distance product". If you want double the bandwidth, you have to halve the distance. Which means at least four times as many nodes (in an area). I estimated 350,000 nodes to meet an 800m rule for 93% premises. Look to 1.25M for a 400m rule.
A senior manager involved in the TransAct rollout confirmed to me that most of the cost for any node is for the site works, permits and on-site crew time. The electronics were a small part of the cost. Also, for copper conductors, RF flows only via the surface, not in the bulk conductor. The higher the frequency, the thinner the depth of the flow, and so all joints become impedance changes and reflect back part of the xmit energy. Exactly what you don't want in old, corroded cables. As an added bonus, you get RF signal leakage from twisted pairs. Adding a shield lowers the leakage, improving the distance and reduces interference. [Why Cat 5 won't handle 1Gbps, you need Cat 6 and distances are shorter] This is the basis for "Coaxial" cable: a central conductor with high dielectric constant and a good shield. [Bells Labs invented it] Someone expert in this (e.g. Malcolm Moore, along term Telstra Engineer) will tell you a whole lot more: this is a very complex science, especially when you scale up to large Distribution Networks maintained by ordinary guys in white vans in a rush.
  • Free Air systems [wireless, laser or quantum] suffer much more from
    • fading, external interference, dispersion, path loss, propagation delays/echos, multi-path fading, physical barriers or transceiver misalignment.
    • drop-outs & signal loss [atmospheric, vegetation, buildings, vehicles, ...]
    • terrain and shadows and d) lightning strikes on (necessarily) exposed external transceivers.
Using RF, you have to be very careful about channel allocation and spill-over. Then you have signal interception... These problems will never go away, its why point-point is always {better, cheaper, faster, secure, reliable} than Free Air for the same technology. But we don't have Free Air laser systems that will go 500km! Trivial with Fibre, and no Curvature of the Earth problem, just a little group-delay.
Eavesdropping on Free Air systems, especially RF, is really, really easy, even for 'spot' beams in point-to-point configs.
If you happen to be the USA, you can place a satellite or high-altitude plane a few thousand km's away and intercept all you want...

Robert Morris Snr, ex-Ball Labs and Chief Software boffin @ NSA ("you never really retire there") shared this bon mot at a Unix conference in Sydney in the mid 1990's... He also said no codes were unbreakable, it was about cost. He quoted $5-10M for an intercept... The spooks are very pragmatic about some things  I'm sure you know that the 'secret' A4 cipher used in GSM was broken around 10 years ago...

Turnbull-Fletcher know the business inside-out and know that they need to model both demand and service/supply sides over 25-50 years to design and dimension the network and to calculate Breakevens, Discounted Cash Flows, NPV's and the all-important Free Cash Flow. Yet they won't even acknowledge the questions, which is suspicious in itself.
In Telecomms two rules apply "Horses for Courses" and "There's NO Free Lunches". For mobile users, only wireless works. But you can't use wireless for everything: the Laws of Physics limit what you can do.

No comments:

Post a Comment

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