http://delimiter.com.au/2011/07/23/enough-with-the-nbn-pricing-hysteria-already/
Shadow Minister this link is for you.
I am sure you are just as happy as I am, that you taunted me until I started to post links to prove I had been reading such.
I stand forever amazed, that any one, including your self, has any faith in anything you post.

@Antiseptic: If your friend is having problems, he may need to install an amplifier

He is using one.

@Antiseptic: setting peak speed requirements for 4G service at 100 Mbit/s for high mobility communication (such as from trains and cars) and 1 Gbit/s for low mobility communication ... Sounds like a great deal more than 1.5Mb/s to me.

Look closely at the peak figure. 1Gbit/sec. Sound familiar? I hope so - it is the one I used.

In the same article they say: "Peak link spectral efficiency of 15 bit/s/Hz in the downlink, and 6.75 bit/s/Hz in the uplink (meaning that 1 Gbit/s in the downlink should be possible over less than 67 MHz bandwidth)."

So lets see 15x67M=1005, so they do indeed get 1GHz. But only if the entire cell is devoted to 1 user. And 67MHz is larger than what is typically handed out. In Australia you only get 45MHz. Clearly an oversized cell devoted to one user bears no relationship to reality, but 1GHz looks great on sales brochures and hey, you swallowed it.

The reality bites in the next point in the Wikipedia article you link to:

"System spectral efficiency of up to 3 bit/s/Hz/cell in the downlink and 2.25 bit/s/Hz/cell for indoor usage."

Again I was being generous. I allowed for multiple cells in my previous post I divided by 4. 15/3 means I should have divided by 5.

So lets look at the figures again. 3 bits/sec/Hz/cell means a 50MHz spectrum allocation yields 150 Mbit/sec, which has to be shared by all users. A minimum 1Km cell radius (see http://en.wikipedia.org/wiki/Cellular_network#Frequency_reuse ), gives us approx 3 square kilometres. Assuming a minimum housing density of 8 hours per hectare (http://en.wikipedia.org/wiki/Medium-density_housing ) yields 2400 houses. (Recall I said 2000). Assuming a contention ratio 40:1, that means each user gets 2.5 M bits/sec.

Again I am being generous. No one is going to use a minimum size cell in a minimum density broad acre development. You might find in it a city, but its household density is 4-5 times as high, so your data rate drops to 600kbits/sec.

@Antiseptic: Dear me, rstuart, you do get upset easily.

It appears so. Intelligent people behaving like fools has that effect on me.

This wasn't so much you. I suspect you have had a good experience with wireless, and are just extrapolating this to a larger scale. Sadly wireless doesn't scale. If it did, Foxtel wouldn't have needed to build that expensive cable network, they would have sent the additional channels over the air, encrypted.

What I did above is basic engineering. It is no different to any other engineering task - you find a model, plug in the figures and out pops the answer. You will be surprised to hear it is the sort of thing engineers are called upon to do before governments commit an a $27 billion in a brand new network. You know - look into a basic things like "what would do better job - wireless or fibre", and "are we building a white elephant which will be worth nothing in a decade". I imagine the bankers that are giving them the additional $8 billion would have wanted something similar. I'm not a banker of course, so that's just a guess.

The point I am leading to is unlike you, Shadow Minister claims to be a practising Engineer. You know, one of those 4 year tertiary trained people the public spent 1000's educating. This is the person you get to crunch the figures and be ruthlessly objective - the sort of person we can and do trust to build big complex things that won't kill us.

The thing that nearly drives me to tears if you put a Labor policy in front of Shadow, he is about as objective as a 1st years arts student confronted with a GM potato.

rstuart:"But only if the entire cell is devoted to 1 user. "

One of the important points I made and in fact something that is part of the 4g specification is the use of low-power femtocells. These serve only a few users and hence their utilisation is unlikely to be anywhere near capacity for much of the time. At peak times on the exchange i'm on with my ADLS service the problem is backhaul congestion, not saturation of the PSTN link. The backhaul problem is a common factor.

There are a couple of other aspects. The first is the use of frequency-hopping and the availability of multiple channels per connection (diversity, as in 802.11n, for example), which can be dynamically switched to optimise spectrum usage. This means that the issue of leakage between cells is much less of a problem. On top of that, we're still talking about 1800MHz band, when 5.6GHz is available and entirely suitable for the short-range femtocells in question. Higher frequencies still, which have inherently more bandwidth, will no doubt be made available as well, if the model were changed from a cell of several km to one of several tens of metres.

And all that is without even considering the rapid advances being made in compression and encryption, which allow more information to be carried per packet. Yes, the same tech applies to fibre, but the overhead on a residential fibre means there's no compelling reason for it's use.

You see, I do understand that the ultimate capacity of a fibre network is greater than a broadcast network. All I'm saying is that committing to a one size fits all fibre to the premises model is needlessly expensive and will leave a huge amount of dark fibre in the ground. A multiplicity of 4G cells using narrow-cast, directed signals to communicate with relay stations that are in turn connected via fibre links would be a much more cost effective plan, it seems to me. Especially since the mobile network is going to replicate a great deal of the NBN coverage anyway.

On the subject of Foxtel, I have some experience, as well as with Optus, both of which I have contracted to. HFC was a fairly cheaply available tech that was already obsolete O/S when the rollout began. Foxtel chose to go with the cheapest option for their network, since they didn't provide critical services over it and chose a simple dendritic layout with no redundancy. Optus, since they were providing phone services and had to meet service-level commitments, installed a ring-topology network with built-in redundancy. Even so, on more than one occasion they have had both sides of a ring cut and had their network go down. Queensland was affected on one occasion and Melbourne on another. The mobile network at the time was a very crude AMPS system that was not even an option for consideration for data and simply used a 3kHz channel for plain voice. We didn't even get a GSM network here until 2000 and lets not forget that dial-up using a 28.8kb/s modem (max, 28.8) was still the norm.

As far as wireless not scaling, take a look at that spectral eficiency link you provided. Bitrate has expanded from that 3kbit/s of AMPS to the 1Gb/s of 4G and the subscriber base has similarly expanded

http://upload.wikimedia.org/wikipedia/commons/thumb/1/19/Mobile_phone_subscribers_per_100_inhabitants_1997-2007_ITU.png/250px-Mobile_phone_subscribers_per_100_inhabitants_1997-2007_ITU.png

To say wireless doesn't scale is simply not true.

I was also involved as a ham operator with packet radio on the 2m band, before there was any digital network of any kind. We counted ourselves lucky to get more than a few baud! I think there's still an active PR group in Brisbane, but I suspect they're probably concentating more on private wireless networking these days.

I'm glad you understand I'm not taking the Mick. I would have hoped to see more discussion of the pros and cons of the distribution model chosen before the rollout began. I hope that when the Gillard mob is booted, that Conroy hasn't managed to cast this in stone. It needs to be revisited.

@Antiseptic: One of the important points I made and in fact something that is part of the 4g specification is the use of low-power femtocells.

A Femtocell covers a house, maybe. (5.6GHz will be lucky to get through a couple of plaster walls.) How on earth do you think the data gets from the Femtocell to your ISP? Femtocells and their cousins use the customers land lines.

I suspect you are saying picocells could go on every pole of something. Fine. But what's the point? You end up with fibre running down every street anyway. If you put picocells on the poles, the telco has to supply power, maintain it and its battery. If you run the fibre to the home all this becomes to the customers problem and things like broken microwaves, welders, lightening aren't a problem.

@Antiseptic: rapid advances being made in compression and encryption, which allow more information to be carried per packet.

I don't know what your talking about here, and believe me I should. There have been some "advances" in encryption - principally new ciphers to cope with Moore's law. Without Moore's law DES would be still as good as it was 20 years ago, and DES3 is still robust. But encryption has _nothing_ to do with the speed data is sent.

But compression - no. Nothing.

I suspect you are getting your terminology confused with modulation and http://en.wikipedia.org/wiki/Modulation coding techniques. Yes, there have been advances. But this is not like water pump, where you can get any pressure you want by just pumping harder. Spectrum efficiency is like a sponge. There is a fixed amount of water in there, and squeezing the sponge twice as hard only gets you a little bit more. The Shannon–Hartley theorem tells us how much water is in the sponge. You can see how close we are to that limit here: http://www.wirelesse2e.com/index.php/category/lte/ We are already squeezing pretty hard, aren't we?

(cont'd....)