[LINK] National Broadband Network - except it's not national.

Richard Chirgwin rchirgwin at ozemail.com.au
Sat May 8 20:17:53 AEST 2010


David Boxall wrote:
> On 7/05/2010 10:31 AM, Tom Worthington wrote:
>   
>> ...
>> If the primary way to access the Internet is from a wireless mobile
>> device, then most of the need for fibre to the home disappears.
>> ...
>>     
> That's a big if. Is nothing conceivable that can't be achieved "from a 
> wireless mobile
> device"?
>   
Lots.

Stil pointed out the Shannon-Hartley theorem, and the clear outcome of 
that, given enough demand for bandwidth, is that the cell-to-households 
ratio would eventually have to approach parity. At which point, what 
good is served by having a high-energy, low-efficiency wireless channel 
for the last few metres when it's cheaper to just run the fibre?

Wireless is also limited by spectrum availability, something that's 
glossed over by those in the industry with an interest  in pushing 
wireless as the solution that makes fibre obsolete. To get (for eg) the 
full rated performance of LTE+ you must have contiguous wireless 
channels - something which would require a very long process of 
reorganisation of huge slabs of the national radio spectrum (working 
from memory at an Ericsson briefing, you need contiguous 20 MHz channels 
in whatever spectrum you're using. That sounds easy - but spectrum at 
the moment is very fragmented).

And you need more and more of those contiguous channels, the more users 
you're trying to serve.

Not just that. While I agree that some cellular prices represent 
rent-seeking, there is a real extra cost in base stations. Imagine, for 
example, that you have a 160 Mbps-capable system (coming up in the LTE+ 
standards, I think).

If that's supposed to support (say) 20 households, that leaves 8 MBps 
per household. Now, different people will have different duty cycles, 
but it is absolutely inevitable that there will be peak periods, in 
which those households get what amounts to ADSL1 performance.

And at 20 households per base station, we need 400,000 base stations - 
not to mention the absolutely deafening community outrage if base 
stations start getting installed on both sides of every street. To match 
the current NBN performance - not allowing for fibre's upgradeability - 
we would need close to four million base stations, which is just ludicrous.

Having been stung by the "wireless will make fibre obsolete" lobby quite 
recently, I should be more circumspect, but actually I don't care.

In the real world, people will use wireless for convenience, fixed 
(fibre) to serve fixed locations.

RC
> We need a network with enough headroom for future developments. 
> Developments beyond what we can foresee. Otherwise, we risk ending up 
> with something that's obsolete before it's completed.
>
>   




More information about the Link mailing list