(...cont'd)

What they have done, and with the NBN will do is use wireless in low very density (rural) areas. http://www.theaustralian.com.au/australian-it/government/ericsson-seals-1bn-nbn-wireless-deal/story-fn4htb9o-1226067115500 This will be fixed wireless with dedicated capacity for each customer so it doesn't suffer from the "collapsing cell" syndrome. Ie, new towers, new infrastructure. As you can see from the link, Ericsson won the contract and it looks like it will be the first LTE network in Australia.

That probably gives us a hint as to why it isn't done. Once the density gets to a certain point, it is simply cheaper to run the fibre. In fact my guess this is exactly the calculation the NBN performing. They are obligated to provide 12 Mbit/sec to everyone, and they use the cheapest method to do so in any given situation. So they aren't really ignoring wireless. They are just using it when it makes sense to do so.

You don't want a wireless NBN connection. Or at least I would not. Unlike fibre, which as rolled out gives you up to 1G bit/sec, they are only guaranteeing 12Mbit/sec for wireless.

@Antiseptic: The PSTN has lasted about 130 years.

I didn't realise it had been 130 years. You are almost certainly right, but I doubt anything beyond 30 years matters.

@Shadow Minister: Your calculations are based on one antenna using one wifi frequency slot.

No, they don't Shadow. You obviously don't realise it, but LTE just like WiFi LTE divides the allocated spectrum chunk into a number of channels, typically 5MHz in width. In order to achieve the speeds it does LTE the bonds a number of those channels together. That 150MHz you get close to the centre of the cell is when you use all available channels simultaneously on a 67MHz spectrum allocation.

@Shadow Minister: Garbage in = garbage out.

Too true, as you have so ably demonstrated here from your very first post. In your case if you could reduce the amount of garbage in by just ignoring those Liberal press releases you seem so keen on regurgitating here.

Rstuart,

I have a degree in electrical engineering with majors in communications and field theory, (i.e. how radio and optic communication works) and am staggered at the rubbish you are posting.

Using a TCPIP 1GB/s network, it is entirely possible to be connected to 1000 people at 1GB/s, and as long as not everyone is trying to download vast amounts of data simultaneously, an intelligent network will manage, or even slow down the data transfer, giving each user an effective 1MB/s connection, while still being connected at 1GB/s.

If you look at the available frequency spectrums:

450-862, 2300-2400, 2700-2900, 3400-4200, and 4400-5000 MHz

as per

http://en.wikipedia.org/wiki/IEEE_802.11y

You will see that the total available spectrums are far higher than your posts indicate. Just at the higher spectrum there is 600MHz to play with.

Wifi is in its infancy, and to simply point at what is presently available in Australia now and to determine its limitations based on this is extremely naive.

@Antiseptic: 5.8GHz is used for some cordless phones (inefficiently) and gives me about 100m of range

The words "inefficiently" and "100m range" are two sides of the same coin. The actual throughput would be under 100 kbit/sec, the raw throughput at least 100 times that. When you only need 1 in 100 bits to make it through loosing a few because of distance doesn't matter.

@Antiseptic: the latency would be limited to that caused by the internal path between Tx and Rx in each relay.

It wasn't the routing latency I was thinking about, as you often find yourself going through 10's of routers now. It is the bit error rate. The bit error rate of fibre is 10e-10, wireless literally a million times worse than that and fluctuates wildly. http://www.scribd.com/doc/6991411/Ebook-Wireless-Implementation-Of-An-Ieee-80211-Wireless-Lan-Model-Using-Opnet I guess you could call it the cheap Chinese drill effect. Handling it over 1 hop is hard enough. Handling the cumulative effect over many hops sounds hard.

There are several ways to overcome a high BER, the one you use depends on the type of data you are sending. Voice and video must be timely, and you don't mind the odd bit of data corruption producing noise, so the usual technique is forward error correction - ie putting redundant data in the stream to recover the few bits that are normally lost, and just live with the error bursts. For computer data, like your bank statement, must either either arrive intact or not at all, and you are willing to wait. In that case you use explicit ACK's, timeouts and retries that may take forever in video terms.

So yes, you are right, you can optimise for video. But what if this is the NBN, meaning its IP everywhere and you don't know the type of data you are carrying?

@Antiseptic: I thought the smart meters injected a signal onto the power cable?

It was the idea once. Turns out making our power lines giant RF noise injectors pissed more than a few people off.

@Shadow Minister: giving each user an effective 1MB/s connection, while still being connected at 1GB/s.

For gods sake Shadow, try reading what has been said. Hint: this point has already been addressed, not once but several times.

@Shadow Minister: Just at the higher spectrum there is 600MHz to play with.

So as per the post above, 67 MHz gives us 600 kbit/sec in the city. You are claiming we have 10 times that spectrum. Righty oh, so that would mean we would get 6 Mbit/sec, still assuming unrealistic conditions. You realise that the NBN now is contracted to provide a _minimum_ of 12 Mbit/sec, and that 100 Mbit/sec plans are being offered now. That's now, not in a decade or two's time. In two decades time a 0 could easily be added to both figures without stressing the fibre they are laying now in the slightest.

@Shadow Minister: I have a degree in electrical engineering with majors in communications and field theory, (i.e. how radio and optic communication works) and am staggered at the rubbish you are posting.

If you want to be treated like an engineer start behaving like one. The working behind the all figures given has been posted. All assumptions have been referenced. If you think you have found an error just point it out, calmly and logically - like an engineer does.

Rstuart,

Your maths is out again, you left out 3 zeros.

67Megahertz bandwidth gives 600Mb/s

600Mhz gives 6Gb/s according to your method.

@Shadow Minister: Your maths is out again, you left out 3 zeros.

No Shadow I did not. It was all explained above. But never mind. You evidently don't read it, and I have decided repeating isn't is going to progress the discussion.