In this lengthy discussion of the “National Broadband Network”, it was generally assumed that the existing copper network had essentially reached its speed limit, and that it could never provide the kind of internet performance that Australians will want in the medium-term future, necessitating the construction of the “Rolls-Royce” fibre-to-the-premises solution announced by Kevin Rudd a couple of days ago. But are these two assumptions correct?
Most Australians with “broadband internet access” at home use ADSL. This technology essentially uses your old phone line to send high-speed data. The latest incarnation, which these days most Australians have access to after Telstra finally turned it on a couple of years ago (after deliberately leaving it dormant in their exchanges for some time) is ADSL 2+, which tops out at 24 Mbit/s under ideal conditions. That is, an ADSL 2+modem can carry about 3 megabytes of data per second. That’s the rough equivalent of three books’ worth of text per second, or or a three-minute pop song in MP3 format. Theoretically, it’s enough to stream live about 1.5 HDTV channels, or, if you like, one HDTV and one SDTV channel. The practical capacity is somewhat lower; something closer to one HDTV channel is probably the practical limit of ADSL2+ under good conditions.
But is that the limit of what the copper network, with a few upgrades, might realistically provide? In a word, no.
The government’s original proposal for “fibre to the node” would have involved replacing only part of people’s phone lines with fibre optic cable; the last few hundred metres would have been left alone, and each “node” would have connected to individual buildings using a of a souped-up version of ADSL called VDSL.
So, how fast is VDSL? Considerably faster than ADSL2+, if you put in fibre links to an intermediate point, and just use VDSL for the last bit to individual homes and businesses. The latest version – VDSL2 can support speeds of around 100 megabits per second over a half-kilometer distance to the nodes. Yep, that’s as fast as the government is promising for fibre to the home, and it’s fast enough to deliver four simultaneous high-definition TV channels.
That’s with existing technology. What might we be able to do in the future? this academic paper suggests that, in many cases, it might be possible to do a whole lot better, by borrowing some technology, amusingly enough, from the latest generation of wireless devices. In some places, there are multiple unused wires in the “local loop” to people’s houses – provided in case people want to have second or third phone lines. Those lines are essentially sitting dormant. At a naive level, if you’ve got 4 lines available, you’d think you could increase the speed by a factor of four, right? Wrong. Through magic which I do not claim to understand but is now fairly standard practice in electrical engineering, you can take advantage of the interaction betweenthe signals on the lines to transmit much more information than you’d think – you can add one and one and get more than two. According to their calculations, it’s quite feasible to squeeze ten times the data on to four lines. That’s enough throughput to carry 40 HDTV streams simultaneously.
But there’s a further complicating factor. The unit of measurement I’ve chosen is a high-definition digital television stream, because that’s the most demanding type of data most people will ever try to send or receive on the internet. But, with video, what’s important isn’t how much data is transmitted, it’s that the picture you see looks good, regardless of how much data is actually transmitted. And depending on how you represent the information in the video stream, the amount of information required to make a good-looking picture varies greatly.
No practical digital video system (except for some scientific applications) tries to transmit exactly what the camera sees to the end viewer. What you see on the screen is an approximation. The methods for making those approximations – known as codecs – have improved a lot over the years, and the latest ones can squeeze a video that looks just as good into a lot less space. The standard used for terrestrial digital video transmission known as MPEG-2, is 15 years old. Later schemes, such as MPEG-4, can squish a picture that looks just as good to you, the viewer, into half the space. So our standard ADSL2+ can carry two HDTV streams using MPEG-4, VDSL2 can carry eight, and the super-DSL of the future might conceivably carry 80 or so.
So, in a nutshell, I reckon that we could have gotten enough capacity out of a moderately souped-up copper network, in large parts of Australia, for the medium term future. And we could have done it at much less than 43 billion dollars.
The experts advising the government presumably know this, but they’ve still recommended replacing Telstra’s copper network entirely with fibre optic. And, in my view, the key reason isn’t technical. The expert report on the broadband tenders is concerned the upgrade path from the kind of FTTN network I’ve examined here to the Rolls-Royce FTTP, but the key objection is as follows:
The proposals have also demonstrated that rolling out a single fibre-to-the-node
(FTTN) network is……likely to require exclusive or near-exclusive access to Telstra’s existing copper sub-loop customer access network (CAN), the so called ‘last mile’, thereby confirming that strong equivalence of access arrangements would be essential. As well, providing such access to a party other than Telstra runs a risk of liability to pay compensation to Telstra. The Proposals have this risk remaining with the Commonwealth but they have not addressed the potential cost to the Commonwealth of any such compensation. In any event, the Panel considers that no Proponent could accept the cost risk and continue to have a viable business case.
Of course, if Telstra had never been floated, or the network itself had remained in government hands and just the retail part had been sold, this wouldn’t have been an issue. How many billions of dollars will that little mistake cost Australia?