[LINK] NBN white-elephant-to-be: better spend the $$$ on other things

Robin Whittle rw at firstpr.com.au
Tue Aug 17 23:43:47 AEST 2010

Thanks to the ALP's highly democratic facility for commenting on some
pages on their site, I was able to add a substantial comment to:


Here is a discussion of the NBN, independent of the ALP's Internet
censorship policy debacle which overshadows it.  Please see what I
wrote about this today in:

"There are 39 policies listed at this site, but the most significant
and unpopular is not mentioned - Internet censorship."

I have been writing about telecommunications since 1992, for
Australian Communications Magazine and more recently (until 2 years
ago) for Paul Budde Communications.  This is my viewpoint - unrelated
to Paul Budde's.

In an ideal world, we could have fibre to every home, school, office
and factory.  The way to do it is a passive optical network, such as
GPON, 10Gbps EPON or Wavelength Division Multiplexing WDM PON.  For
some up-to-date info on these:

Where this is not possible, 3G or other wireless technologies can be
used, but this is difficult over long distances, and there is very
limited data carrying capacity compared to fibre.

Where wireless is not possible, a geostationary satellite link can be
used.  However, this involves very high costs, limited total
bandwidth for the whole country, even lower upstream rates, and long
latency due to the distance to and from the satellite data in both
directions must take.  The satellite is 36,000km away, so that is
0.24 seconds delay each way for data between the NBN network's
satellite station and the user's satellite station - half a second
latency in total.  There are a very limited number of geostationary
slots.  The satellites cost billions and last 10 years or so.  So
this can never be a mass-market form of broadband connectivity.

PONs are clearly the way to go, and NBNCo has chosen GPON, with
2.8GBps downstream (towards the homes which share the fibre) and
1.2Gbps upstream (from the homes).  Multiple fibres branch out from
the extremely complex and expensive "Fibre Access Node" (FAN) and can
travel for a few tens of km to a splitter, where the fibre is
effectively split multiple times to create a 16 fibre, or perhaps 32
fibre tree - with one branch going to each home, office, etc.  There
are supposed to be 700 FAN sites for Australia

These long fibres, and most or all of the splitters and branches,
need to be installed all at the same time, not incrementally over the
years to come.  Extra fibres are needed to allow for denser
buildings, and so more homes and offices.  The fibres require no
power.  Each fibre carries downstream data to the ONTs in each of the
the 32 or so homes on a 1490nm infra-red wavelength, with encryption
per ONT.  The ONTs are coordinated so they take it in turns driving
1310nm light on the fibre back to the FAN, for upstream data.  ONTs
have Ethernet ports and perhaps a phone port.  For reliable phone
operation, each ONT needs a substantial a battery backup system.

Its a great way to do things, but this is not 1Gbps per customer.
There's 2.4Gbps downstream shared between 16 to 32 customers, and
1.2Gbps upstream.  10G-EPON and WDM improve on this, but at still
higher cost.

These fibres will generally not fit in existing street ducts, which
were built for a 50 pair copper phone cable, and frequently now have
a similar sized, but much stiffer HFC coaxial cable in them as well.
 Both these cables are about as thick as our thumbs.

So there will need to be directional boring along many suburban
streets, with costs of $1k to $2k per house passed, in ordinary 1/4
acre block densities.  The splitters need to be in pits or "fibre
distribution hubs":
 The branch fibres to each home can't necessarily be placed in an
existing duct, so there may be trenching or directional boring to
each house.  In the house, the fibre can't be bent sharply, because
the light leaks out - so it is tricky to get it to the location of
the ONT.

Data carriage in the "backbone" to each FAN can be made quite
reliable via redundant fibre paths.  However, from the FAN to the
homes, there is no redundancy, so be careful with backhoes.
Stringing these cables from power poles is possible, but that is less
reliable due to lightning, falling trees, vehicle collisions etc.

How can the cost of this, to 10 million homes, be under $43B, since
some fraction of that must go to the backbone, wireless and
satellite?  There's no proper basis for this figure.

It would be great if we could all have GPON, but we can't all get
1Gbps downstream.  Even if we could, there's no way any ISP could
offer services with such data volumes at affordable rates, for global
Internet access, since the cost of getting data across the Pacific,
and from Asia, is so high.

Its a mistake to think all this could be done for $43B.  This sort of
massive expenditure would have to be at the expense of other vital
priorities such as health, welfare, education, renewable energy
research etc.  Most people having fibre is a luxury we can't afford.
 DSL is fine for most purposes.  The only thing fibre does better is
support high quality video streaming.  Why spend tens of billions of
dollars for the benefit of couch potatoes?

So far, the NBN is an undemocratic white-elephant-to-be from the ALP
- an outfit which couldn't safely handle the installation of pink
batts!  If the ALP is voted in again, this will be official
government policy, along with a gravely undemocratic Internet
censorship regime.  Australia must avoid giving a mandate for this
Stalinist censorship and for this profligate waste of taxes on a
network which would be great, but which we don't need and can't afford.

Structural separation for Telstra and some government support for
broadband in the bush? Sure. Then we won't be wasting money we
desperately need for preventive health, education, youth support,
Medicare rebates for Social Workers and Family Therapists, and for
developing 24hr a day, heat-storing, solar thermal power stations on
a massive scale.

 - Robin Whittle

More information about the Link mailing list