rstuart, I have to say I'm disappointed. The paper you referenced was in relation to the original 802.11 specification and technology has moved on a long, long way.

As a simple experiment, I took my laptop which uses 802.11n and walked away from my router while testing ping and jitter using www.pingtest.net and running a large test download, while simultaneously d/l on two other machines on the same router/link. Ping/jitter didn't increase from the 27/1 that I measured beside the router with no load until I was almost at the limit of range of the signal. As we discussed earlier, processing and discrimination have improved markedly in line with Moore's Law and developments in signal-processing.

LTE, WIMAX and of course 4G all improve on this even more.

http://www.hindawi.com/journals/jcnc/2010/628657/

The take-home message is that wireless doesn't work well at the boundaries, but hey, neither does fibre, which is why they install boosters.

I guess what I'm saying is that once again, just as with HFC, the country is going to be saddled with technology that is essentially obsolete ,for purely political rather than technical reasons, with all due respect. LTE is a kludge to extend the capacity of existing spectrum, but can still deliver 12Mb/s at 7km range and fibre will become redundant for domestic services as soon as 4G becomes available. Future developments will only further weaken the case for fibre.

I think that was made clear by the 20 year non-compete clause for wireless that Telstra had to agree to.

http://www.theaustralian.com.au/business/opinion/non-compete-clause-exposes-flaws-of-nbn/story-e6frg9k6-1226081624053

"The relevant clause in the deal is that Telstra can't promote "wireless services as a substitute for fibre-based services for 20 years"."

Once again, noone is arguing that fibre is not superior to wireless in terms of bandwidth and latency, just that it is unnecessarily expensive and overspecified. It's akin to buying this year's V8 instead of waiting till next year and buying the new hybrid. Act in haste, repent at leisure.

@Antiseptic: LTE, WIMAX and of course 4G all improve on this even more.

No. LTE gets 150 Mbit/sec near the antenna, 70 Mbit/sec average, which obviously means less than 70 Mbit/sec near the edges. That is not the picture you are trying to paint - good performance over most of the link, then a sudden drop off at the end. The real effect shown clearly by the graphs in these early Ericson trials of LTE. The speed goes down as the SNR goes down, as you would expect. http://www.ericsson.com/ericsson/corpinfo/publications/review/2008_03/files/LTE.pdf

@Antiseptic: Ping/jitter didn't increase from the 27/1 that I measured beside the router with no load until I was almost at the limit of range of the signal.

802.11n is designed to operate at its rated speed within 100m, line of sight, with 10's of people using it. Is that what you where doing? Then congratulations - it was within spec. I have had the reverse experience. I've been using 802.11n, happily with in line of sight with 20m or so of the AP. But it was within a lecture theatre with 60 other people and then the speaker suggested the room look at something online. I got disconnected.

Oh, and you know my long suffering wireless user this all started with. He has now realised while it worked real well when he first got it going late one night, it isn't always like that. He says it is possible to tell if someone is using a mobile phone in the house - but that's a minor thing. His real issue is it drops to 50Kbit/sec during peak hour. Turns out his neighbours are using the same solution.

@Antiseptic: the country is going to be saddled with technology that is essentially obsolete

Didn't you just say in a post or two ago you can't think of anything better than fibre?

@Antiseptic: unnecessarily expensive and overspecified.

Compared to what? Not doing anything? Because it is sure as hell is cheaper than wireless now. And given internet use is rising exponentially, I doubt it will be looking overspecified in a decade or two.

rstuart:"Didn't you just say in a post or two ago you can't think of anything better than fibre?"

I also said it's a question of fitness for purpose.

That's why there's that non-compete clause. Wireless is adequate for all but the most demanding domestic applications today and will become much better in the near future, let alone over the next 20 years. The only member of my household who ever complained about 3G was my son, who's a 13 year old mad keen on first person shooters.

Rstuart,

Your maths falls apart at the first hurdle:

" pretend you can get 1Gbit/sec per cell.

The smallest mobile phone cell is 1Km radius (in reality they're much bigger), covering 2000 houses, all of whom must share the 1Gbit/sec. So each house gets 500 kbit/sec, which is under 1/3 of what you are guaranteed to get with the worst ADSL line right now!"

Actually, assuming you had a 1Gb/s connection rate, the 2000 people connected would be connected at 1Gb/s. Whilst their average download rate is 500kb/s, in reality only a tiny fraction would be transferring information at anywhere near that rate, as most of the time any particular user is idle. Only if everyone tried to download at any one time would the data rate drop to 500kb/s

The most likely scenario would be that 10% are transmitting, and those doing so would get 10Mb/s. But what you mention is a fantasy connection scenario.

Using a phased array would be a waste. What would be far more efficient would be a multiple of directional antennae on a single aerial each giving 1Gb/s to a particular portion of the area.

However, the point is that wireless is not competing head to head with fixed line, as its competitive edge is that it is mobile. And for the small users, which is 50% of users this far outweighs the comparatively slower connections.

All wireless has to do is to take 25% of the connections as is presently happening in the US, and the NBN is no longer financially viable.

@Shadow Minister: Actually, assuming you had a 1Gb/s connection rate, the 2000 people connected would be connected at 1Gb/s.

OK, once more.

Because the 1Gb/s is shared by many users it is effectively a backhaul http://en.wikipedia.org/wiki/Backhaul_%28telecommunications%29 . ISP's have to provision backhaul's now, so we can apply there existing provision rules. They even have a name. It's called the contention ratio. http://en.wikipedia.org/wiki/Contention_ratio A contention ratio of 40:1 is on the high side.

So all I did was apply this existing rule of thumb to the peak cell rate. But first we need a realistic peak cell rate, and 1Gb/s ain't one. Once you have multiple overlapping cells LTE drops to 150 Mbit/sec at the centre (see http://www.theaustralian.com.au/story-fn4htb9o-1226067115500 which agrees with spectral efficiency of 3bit/s/hz/cell from http://en.wikipedia.org/wiki/4G#ITU_Requirements_and_4G_wireless_standards ), averages 70 Mbit/sec, and as that is an average it is obviously smaller still near the edges. But lets assume 150 Mbit/sec.

Another thing we need is have to get a realistic estimate for houses that 150 Mbit/sec serves. If we use the smallest cell size (1Km, from http://en.wikipedia.org/wiki/Cellular_network#Frequency_reuse ) then we are obviously in the city, where density is high. I will use the 30 houses per hectare figure Wikipedia gives for medium density in Australia http://en.wikipedia.org/wiki/Medium-density_housing . That is 9,400 houses in the cell.

So the final speed each house sees is:

150 [Mbit/sec] / 9,400 [houses] * 40 [contention ratio] = 638 k bit/sec.

@Shadow Minister: All wireless has to do is to take 25% of the connections as is presently happening in the US, and the NBN is no longer financially viable.

Indeed Shadow. But as has been point out over and over again, since fixed land usage is currently growing that doesn't look likely, and looks even less likely for the NBN as it provides a faster service than ADSL for the same money.

@Antiseptic: That's why there's that non-compete clause.

Sorry, I didn't get your point last time around. I think you are reading into it more than exists. The NBN has collectively spent some $10 billion to get themselves a monopoly. If you want fixed broadband, you have to go to the NBN - end of story. You want to read "they are scared of wireless" into that. I don't think so. They are already rolling out wireless where it suits them. If wireless suddenly got better they would roll out more of it.

What they are scared of is competition. One of the great attractions of the NBN is it looks to be a profitable venture - something the government will make money on in the long term. But as Shadow keeps pointing out, they really do need that $24/mo from every existing fixed broadband user out there in order to make that a reality. The best way to make that happen is to buy yourself a monopoly. A normal corporation could not do that, but a government can. That is why it will take a government initiative like the NBN to pull off re-wiring the country.

@Anti: Wireless is adequate for all but the most demanding domestic applications today

No, it's not. It works for you in the sticks, because there are bugger all users. The whole point of my screaming match here was to make it plain today's wireless can't handle the load in the cities.

@Antiseptic: and will become much better in the near future

No, again wrong. The figures here weren't done with todays wireless. They were done with wireless from the near future - LTE. And it still didn't work in the cities, literally by over an order of magnitude.

@Antiseptic: let alone over the next 20 years.

So this is what your argument rests on. Something new, as yet unknown technological development will render fibre obsolete. As I pointed out it is unlikely to come from wireless, as we are getting near the Shannon–Hartley limit - ie the limits imposed by laws of physics.