NBN: Why Wireless is Impractical

Broadband in Australia is a hot issue right now. Do we spend $43b on Labor’s National Broadband Network (NBN), $6.5b on the Coalition’s hybrid wireless/satellite/”optimised” xDSL/HFC solution, or do we do nothing at all?

Nothing is not an option – our broadband capacity is a joke – and for me it isn’t even a political argument about Labor vs Coalition policies. It is purely and simply a technology vs cost argument. Both $43b and $6.5b is a lot of money, but either way it is money that is more than worth investing for the future of this country.

Given that the Labor government has been returned, at this time the Coalition plan is effectively dead in the water anyway.

The Labor plan is obviously a mainly optical fibre-based solution, and the Coalition plan didn’t resolve to do much more than “optimise” the existing xDSL network in major centres, so once the limit of xDSL technologies is reached, there is very little scope to do much more with the ancient copper network than to replace it with wireless or fibre.

To analyse how something like that might all come together, let us look at the numbers for one of the major cities in Australia, Melbourne.

Melbourne is made up of a population of approximately 4,000,000 people, covering an area of approximately 8,806 square kilometres. Using some simple mathematics, that means that on average, there are 454 people living in each square kilometre of Melbourne. Based on 2.8 people per household – (from 2001 figures of 1,200,000 households across a population of 3,366,542) – this means that for every square kilometre of Melbourne, there are roughly 162 households.

If you were to deliver a guaranteed baseline of 12Mbps to every single one of these 162 households, you would require approximately 1.95Gbps of backhaul to support those 162 households, for each and every one of those 8,806-square-kilometre sized areas. That is a total of around 17,171Gbps of backhaul into and out of Melbourne, to support a guaranteed 12Mbps for every single one of the approximately 1,435,378 households in Melbourne.

That’s a lot. Lets also remember that 12Mbps is only 1/83rd of 1Gbps – the potential speeds promised under the Labor NBN plan. Multiplying that against 17,171Gbps, you’ll need 1,425,193Gbps of backhaul to guarantee everyone 1Gbps. An astronomically high figure. And this doesn’t cover the large number of business premises within Melbourne, so the numbers blow out even further.

However, nobody is guaranteeing anyone 1Gbps. In fact, most people won’t be seeking to buy a connection anywhere near that wide. Realistically, most people might only seek to get a maximum of 100Mbps on the NBN, which when allowing for reasonable contention ratios, might see people getting around an average of 60 to 70Mbps in practice.

Still plenty fast for most applications, and obviously not requiring 1,425,193Gbps of backhaul. Some homes and businesses will want – (or even need) – 1Gbps, but in the immediate future that will be the exception, rather than the rule.

It is important to acknowledge, that no matter what the average speed per household settles out to be in a completed national network, the amount of backhaul required remains the same, no matter if you build it using wireless or fibre optics. You still have to do the heavy lifting into the core of the network, away from the edge of the solution.

The amount of backhaul required around the nation into any section of the network is nothing more than a simple mathematical equation – you just have to decide and/or model how much bandwidth each user will need to get on average, to scope out the capacity of the various links in the chain.

Once we have kitted out the backhaul to deliver the capacity we require on a national level, we now have to lay the delivery model across the so-called “last mile” – the connection from the various households and businesses to the concentration points on the backbone. The two options in the debate are wireless and fibre optics. So lets return to the mathematics.

If you need to build out 1.95Gbps per second to every square kilometre of Melbourne using wireless, you might put a wireless tower in each of those one-square-kilometre areas. Wherever you stand, there would be a tower somewhere within a one kilometre radius. If you stand at the base of any single tower, you should be able to see the towers in the adjacent eight square-kilometre-sized areas, and possibly those beyond. That’s a lot of towers, and nobody will want a tower built in their backyard.

If you decided to halve the number of towers, and build two-kilometre-square areas, you’d need to pump 3.9Gbps into each of those towers. However, those towers would need to be taller and emit more powerful radio signals to effectively cover the bigger area by line-of-sight, or risk end customers receiving more reflected and therefore weakened signals. Halve the number of towers again, and build four-square-kilometre areas, and you’ll need to kit that tower out with 7.8Gbps, and make it even taller and more powerful, once again.

You can see the pattern emerging. Do we really want to live within a forest of “internet towers”? Towers that might be quite high initially to keep the numbers down, but might need to be “front filled” around as capacity requirements grow.

Reduce the number of towers, and you increase the number people relying on each tower, creating larger outages in the event of a tower failure, affecting more people and businesses. There would be so much public debate as to exactly where to locate the towers, that jumping through all the red tape when the NIMBYs get into the debate, that construction delays will be significant and costly.

Even if all these obstacles are overcome, what happens when 12Mbps is not enough? The numbers – (and the costs) – get multiplied by orders of magnitude when you have to go back and increase the capacity.

There are several other factors to consider with wireless. Weather conditions DO make a difference, and sometimes a very large difference. Topography is also a stumbling block, as the land mass of Australia, particularly in urban areas, is not flat. While radio signals do get around corners, between buildings, over hills, and into valleys, the power requirements increase to provide a guaranteed minimum amount of signal to the entire cell. This is why your mobile phone drops in and out of signal coverage, sometimes within only a few metres of travel. The physical surroundings and transient/ambient conditions can have a huge effect on the level of service you get.

Or don’t get.

The other thing to remember is that the RF spectrum that would be needed to implement a national wireless network of this magnitude will not become available until the start of 2014 at the earliest, when all of the analogue television broadcasts in Australia cease, and this assumes that the timetable for this cessation does not slip at all.

So even if we decided to build a national wireless broadband network TODAY – it would be 2014 at the very earliest before any really significant progress could be made in its construction.

Now lets look at fibre. Remember, the backhaul requirement per square kilometre doesn’t change.

Fibre will mostly be underground, thanks to the Telstra/NBN Co agreement, so it will get around all of the corners, between all the buildings, over all of the hills, and into all of the valleys that become potential question marks when using a wireless solution.

It will be able to deliver the exact same speed to all customers – whether you are next door to the exchange, or some kilometres from it. With existing xDSL solutions, the further you are away, the slower speeds you get. At home, I can get ADSL2 which can theoretically get up around the 24Mbps mark, but given my distance from the local exchange, my modem only trains up to the DSLAM at around 4Mbps – far below the proposed 12Mbps minimum of the NBN and the Coalition plan.

The fibre being installed – (and this is underway) – is capable of carrying 40Gbps, leaving plenty of room for future growth, without the need to rip it out of the ground and start again. The Coalition plan speaks of $6.5b – but I seriously doubt the need to significantly upgrade the infrastructure far more rapidly is factored into that. Suddenly, the $43b plan of the Labor government doesn’t seem so expensive.

Upgrading wireless capacity would likely require more towers and/or transmission power, new transmission standards – (and therefore new modems for everyone every few years) – new spectrum to be found and subsequently allocated, more arguments about where to put towers, and significantly more cost.

Look at the public stoushes that get underway now when a mobile phone carrier wants to build one single new tower – people want the service, but not the tower. Not in their backyard.

So how do you give people the service – the quality service they will demand for the money – without the tower?


UPDATE (12:13pm): I’ve had a mathematical error pointed out to me – of course increasing the size of the cells to two-square-kilometres does of course “quarter” the number of cells, and increases the individual cell requirements to 31.2Gbps, and approximately 64,286Gbps across the city. The same sort of scale change is also required for then increasing to four-square-kilometres. I blame the almost week-long headache I’ve had and finally shaken for a small dropping of the ball! Of course, correcting the error further counts against the argument for wireless!

Thanks to @rkjobling for tipping me off to my error – people who know me do know how much I hate to make errors!

  • Pingback: Delinker()

  • Convinced me

  • This topic raises a lot of issues, technicalities. We can only understand the numbers and amounts associated with each of the deal. I wonder if where can we read an article that somehow uses simple non-technical stuff on the 2 options. –Jinx

    • There are a great many articles on the subject – but the debate itself is hardly a “simple, non-technical” one. If we as a nation want to assess which is “value for money”, we need to address the technicalities of each option, so it is important that these are presented.I don’t think it is too much of a stretch to derive from my article that a the Coalition “solution” is initially cheaper, but also less reliable, and that potential future costs to update it as more capacity is required is far less predictable than a fibre “solution”.The Labor “solution” is undoubtedly more expensive initially, but delivers far more predictable outcomes in that the edge infrastructure is significantly less likely to require updating in the life of the network to increase end user capacity. In fact, given that the Coalition plan would lock us into an expensive upgrade cycle, over the same time period – and lets say 30 years for arguments sake. The Labor plan might cost $50b over that time, but the Coalition plan might cost more than that due to continual upgrade costs.In my opinion, dollar for dollar, the Labor plan is more effective in the long term – and it is long term needs that such infrastructure needs to support. The Coalition is locked into government-term-by-government-term thinking, and that is dangerous for our future.For the record, my general political leanings are towards Coalition politics – the NBN is just too good an opportunity for Australia to pass up.

  • Pingback: All Aboard the NBN | Sam's Blog()

  • Pingback: The Sheep Is Dead – Long Live the Sheep()