Thursday 2 May 2013

NBN: Why Fibre?

In answer to "Why do you think Fibre is better than Copper for 'last mile' connections?"

Short version:
Fibre was born Digital, like Ethernet over cat-5 copper. DSL fakes digital onto low-spec phone wiring with modems.
Explaining what DSL modems have to do is close to:
To supply 110V from a car-battery: first convert 12V DC to 240V AC, they covert the 240V AC to 110C AC with a very big, heavy and expensive transformer. It's almost right, at 50Hz not 60Hz. Just sometimes you'll get bitten.
Some first thoughts on why Fibre to the Premises:
  • not subject to lightning strikes
    • I've had a phone blown off the wall in a country home.
  • better life & lower maintenance than copper,
    • including limited interference.
  • start of its product life, far from mature. 
  • not subject to Distance-bandwidth product of coax or twisted pair.
    • has Group-Delay & phase problems.
  • maintains good bandwidth over distance, if loss budget OK:
    • 20km works vs 8km for phone.
But Fibre isn't Only Good Things:
  • fragile. Lead-ins need to be buffered & protected.
  • splicing it is tricky & expensive. Not for amateurs & DIY.
    • Terminating needs skill/tools as well.
  • the sort of power needed in residential areas is dangerous to the eyes.
    • DIY fiddling will be problematic.
  • you really don't want to run out of cores & need to run more cables.
  • I'm guessing with copper, the material::labour ratio was less about labour, upgrades cheaper.
  • when it stops, it stops dead, like all things digital.
There is one completely unsolved problem with Fibre:
 Central Office powering of Customer Premise Equipment. [local batteries are needed]
I remember seeing a paper in the early 1980's about remotely powering CPE with light delivered down the fibre. It seemed infeasible to me then and I think 30yrs on, as nobody has made it work, the physics don't work.

The 2003 fires in Canberra showed a weakness with "nodes". I had friends in Chapman who lost their TransACT landline after 24 hours and didn't get it back for weeks (until the 240V came back). We saw the weakness of mobiles after Hurricane Sandy as well: without power to charge batteries, they are useless in a few days.

The most salient point is a micro-Economic argument:
We can only afford to maintain one universal network for the same task.Cable TV over Coax survived while it didn't compete directly with phone/copper. It's future in a fully digital customer network is bleak.
Someone has to pay for the maintenance of the copper, even the last 400-800m.
Corrosion, (car) accidents above gnd, diggers, lightning, termites/insects/bird pecking, flooding, etc don't stop if just one person is using a network segment.

Telco networks are dominated by Fixed, not Variable, Costs for the services:
costs must be shared amongst the customer base.
A 'death spiral' can result with a service becoming less popular, increasing the per-customer costs, making it less popular, increasing costs more, driving customers away. Such a "tipping point" quickly leads to the commercial failure of the less popular network.

Having two parallel networks that are perfect subsitutes for one another (DSL vs Fibre) means the more desirable will eventually displace the other, if customer utility-value supports it & Fixed Costs aren't too different.

Is their some sort of Economics Equation like a Desirability/Utility ratio? I've not researched this. [but other questions like this that I've asked have drawn blanks from Economists]

When two services have roughly the same Cost/Benefit ratio and approximately the same Fixed costs, then the one with a larger customer base will displace the other through the 'death spiral' effect.

There are a couple of services that need good upload speeds (e.g. stream SD/HD video or disk backups) where Only The Best Will Do: the 1-4Mbps upload limit of DSL (haven't checked that) will force some customers to Fibre, despite the price, because of functionality.

This is another technical feature of Fibre:
  • it can be upgraded to active Ethernet, symmetric service
  • SOHO/SME's may want 4-fibres to run dual GBIC/SFP's
    • are there optical transmit/receive multiplexors for GBIC's to allow single-fibre operation? I've not heard of them, but haven't looked.
So, why do I like Fibre?
  • in the long-term, economics dictates just one universal network can survive, if costs are similar.
  • FTTP allows active ethernet or dark-fibre paths to be sold: symmetric data paths are sometimes necessary

No comments:

Post a Comment

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