(....cont'd)

I don't know how long ago you were a ham radio operator, but today's radio's bear very little relationship to what a radio was 20 years ago. 20 years ago they were turners, amplifiers and what not. Nowadays the principle components are: an antenna, a digital to analogue converter and a analogue to digital converter, and a very powerful specialised computer.

The squeezing on the sponge is done by the CPU. The faster the CPU, the more complex coding we can use, the faster the data rate. So oddly the driving force behind todays better radio's is: Moore's Law. The other oddity is what is limiting this process. It's power, ie Volts x Amps. The limitations come from two avenues.

In a computer where power is unlimited, the problem is literally the CPU melting. The first 1 Gbit/sec ethernet controller ran cool, the first 10 Gbit/sec card dissipated 45 watts - roughly 1/2 the thermal limit of air cooled silicon.

In a phone the problem is somewhat different. That power must come from a battery. Nonetheless, 4G phones still run hot: http://www.anandtech.com/show/4450/lg-revolution

@Antiseptic: I hope that when the Gillard mob is booted, that Conroy hasn't managed to cast this in stone.

The Liberal currently policy is to re-negotiate the deal with Tesltra. (Labor didn't buy copper or infrastructure, just the customer base.) If that fails the fallback position is the original: kill the NBN. Since Telstra gets to keep its current monopoly if re-negotiations fail it seems likely they will.

@Antiseptic: It seems to me that the main reason for the NBN was/is as a Keynesian stimulus measure

After it is built, assuming they succeed in converting all the broadband customers (seems likely), the thing will produce a profit over and above interest payments.

@Antiseptic: I doubt it will last as long as the PSTN has.

You know something that is better than fibre? I don't.

@Antiseptic: will leave a huge amount of dark fibre

What is the problem with that? It doesn't rot, and the expensive bits are hole it lies in, and the lasers at the end.

rstuart, I'm suggesting an arrangement whereby a series of small cells (call them what you will) communicate with each other and with a local base station via directional wireless. If the cells are arranged on light poles, it is simply a matter of relaying the signal through the intervening cells to the base, where it can pick up a cable. surely you'd agree it's much cheaper to only haul cable along main roads, in large ducts that will be vastly more capacious than needed once the existing copper is removed. the rest of Telstra's network would be Telstra's problem to either leave in the ground or salvage.

5.8GHz is used for some cordless phones (inefficiently) and gives me about 100m of range from around 200mW, even through a couple of timber and plaster walls and a steel one. I'd reckon that's adequate or what I'm proposing and it could be enhanced further by the homeowner installing a suitable router Here's a proposal for the very thing. It was the first link when I googled 5.6GHz band just now.

http://focus.ti.com/docs/toolsw/folders/print/tsw5005evm.html?DCMP=hpa_rf_general&HQS=Other+IL+tsw5005evm

No, I'm not getting encryption or compression confused with modulation. I was referring specifically to the fact that compression on the fly is a trivial task these days (Moore's Law, as you say). As the largest bandwidth demand is going to come from such things as streaming video and downloading large files, this means that more useful data goes on a given piece of spectrum.

What I find most amusing in this discussion is that we're arguing the toss about something that is not at issue - the relative ultimate capacity of fibre vs wireless. Since it is extremely unlikely, I suspect, that the majority of subscribers to the NBN will pay for anything more than the cheapest package or perhaps the next one up, then the ultimate capacity of the cells I propose is unlikely to be challenged. (say) 10 subscribers at 50MB/s max is just 500Mb/s, which leaves a full 50% of even 4g bandwidth unused.

rstuart:"You know something that is better than fibre? I don't."

No, nor do I at present, but I'm also sure that there will be something that is as good or better and doesn't require a physical wave guide to make it work. The PSTN has lasted about 130 years. Do you really think that the next 130 years will see fibre unchallenged for the of distributing comms to the home? I'd have to say that was extremely optimistic.

On the subject of dark fibre, when I was with Telstra I learnt a few of the many ways in which empty or disused conduit and its contents degrades. Fibre may not rot, but it can be damaged in very many ways while in the ground. I also can't see a compelling reason to drag the cable down every street, with the cost and disruption associated, especially since the majority of the capacity will simply be unused.

I should say that I'm not at all sure of the cost of the proposals I've kited, but if it could be done at say half the cost, or if a variety of tech could be used on a fit for purpose basis, it really is a disgrace not to at least consider it.

While I'd hate to see nothing at all, or Telstra back in it's traditional place, I'm simply saying the game's not worth the candle at $43billion plus the inevitable massive cost expansion. Let's not forget that the original HFC rollout cost around 20% more than the original budget for the entire project and that was with Cable and Wireless holding the purse strings in Optus's case. I contracted to Optus during that rollout and the waste was simply staggering, as proper cost control was sacrificed for speed. I can't even hazard a guess at the additional cost of such a scheme under a Government with this one's record of mismanaging large projects, even if some other Minister were in charge.

The PR stuff was a long time ago...

Rstuart,

Your calculations are based on one antenna using one wifi frequency slot. In the 2.4GHz range there are 14 available 22MHz slots per antenna, and for the future far more available in the 2.5GHz and 3.3GHz slots etc. More low range antennas provide the same data available in smaller pockets. With future technologies, there is no reason why each antennae cannot support hundreds of slots, each with 100MB/s download capacity.

In short, most of the assumptions upon which you base your calculations are wrong. Garbage in = garbage out.

@Antiseptic: it is simply a matter of relaying the signal through the intervening cells to the base

You're describing mesh networking. I've never done it. There is one set up hobbists here in Brisbane called BrisMesh. A mesh networks theoretical capacity is surprisingly limited. As you could well imagine they have other nasty characteristics like high latency and large jitter that make them unsuitable for real time stuff like voice. I have only seen one real world application. Would you believe it's smart electrical power meters? http://en.wikipedia.org/wiki/Smart_meter

A smart meter requires an always up connection to the electricity producer. They don't want to have to insist on every customer having a working broadband connection, their density is high, and the amount of data the need to ship is small and it isn't real time. So some smart meters are wireless devices form themselves into a mesh that covers neighbourhoods, with some customers being paid a few cents to connect them to their internet routers. I thought it was brilliant.

@Antiseptic: As the largest bandwidth demand is going to come from such things as streaming video and downloading large files

All that stuff is already compressed. Try compressing a video some time. Believe it or not even web pages are compressed during transmission nowadays. There is an incentive for the large servers to do so, as it halves their bandwidth charges. And no, there are no huge improvements in compression in the pipeline.

@Antiseptic: which leaves a full 50% of even 4g bandwidth unused.

No. My 600 kbit/sec figure was optimistic for 4G as it would be deployed for mobile networks in a city. People use far more than that now. You would need to go to your picocell solution to have a hope of delivering better than ADSL. Effectively what you are suggesting is FTTN for a very small node, with the last leg wireless instead of copper. It does sound technically plausible (as in at least it doesn't break the laws of physics), but to my knowledge no one has ever done it.

(cont'd...)

rstuart:"You're describing mesh networking."

Not exactly. The upstream backhaul path through the relays/cells need not be IP-based at all, since it is inherently directional and doesn't require addressing or authentication until it hits the base. The downstream path would require addressing, but this could be done with a low-overhead, low latency distance-vector protocol, since the network topology between the relays is a simple and known one. There are no routers being added and subtracted on an unpredictable basis. IOW, the latency would be limited to that caused by the internal path between Tx and Rx in each relay. HFC already uses this method and fibre uses something similar. The relays/microcells may also be able to provide seamless handovers, but that would be a separate issue to wireless backhaul and is already part of the wishlist for the ITU.

I do understand that video streams etc are already compressed. They could be compressed more, but the limits of hardware constrain that, so as the hardware improves, the compression that can be achieved with a given latency/jitter spec does too. We've already discussed that. 4g is specified to be able to handle high-def video already.

I'm also sure that there are significant improvements to be made in compression and encoding techniques. Cryptology is a highly active field with some very bright people. I think it would be foolish to suggest that the field has already reached its pinnacular achievements.

I guess I'm suggesting a modern take on the hybridisation that produced HFC, with less fibre and no coax at all. It costs to put stuff in the ground and it costs to repair it, as well as taking considerable time. Replacing a small relay/microcell sitting on a light pole would be a trivial task by comparison and there would be no need for any third-party installation work at the premises.

I thought the smart meters injected a signal onto the power cable? Interesting nonetheless.

Thanks for the discussion, BTW. It's been a while since I've thought much on this subject