With digital TV crawling slowly towards reality, broadband advocates and companies are eyeing the freed spectrum to launch Internet services over the freed spectrum, so it's worth understanding the potential applications and limitations of the so-called “whitespace” spectrum.
The reason for the enthusiasm is simple: analogue TV uses a lot of spectrum. TV transmissions reach from a few tens of megahertz at the lowest frequencies, up into the UHF where TV stations bump their elbows against mobile phone transmissions. The broadcast allocations have two functions: they separate TV stations from each other, and they separate TV stations geographically.
To avoid interference, analogue TV stations in Australia are spaced 7 MHz apart, and TV signals travel a long way (especially in the VHF spectrum), so signals in adjacent areas have to be separated from each other. If there are four broadcasters in a town, they'll occupy 28 MHz of spectrum. All of the towns nearby – close enough that their signals might reach consumers' TVs – will need their own 28 MHz of spectrum for their broadcasters, and those 28 MHz will need to be separated from each other.
So if we think of a simplified model, with one area in the centre surrounded by five surrounding, overlapping areas, we would need nearly 170 MHz of spectrum to serve all of them with analogue TV.
Digital TV reduces the bandwidth needs because compression schemes such as MPEG reduce the bandwidth required for a single transmission (stream) – and here's where digitisation creates the possibility of the “digital dividend” beloved of governments, since broadcasters can either offer new channels within their spectrum allocations or they can use their spectrum allocation for high-definition signals, but at the same time, spectrum becomes available for the government to auction to other users (America raised nearly $US20 billion in its spectrum auction earlier this year).
But the auctioned spectrum doesn't count as “whitespace”. Like broadcast spectrum, mobile spectrum, and in Australia, the wireless broadband spectrum held by Unwired, these radio frequencies are licensed for the exclusive use of their owners.
Where's the Whitespace?
Let's return to our hypothetical town with its four broadcasters using 170 MHz of spectrum. In the next town, the broadcasters will be using different frequencies, and further away, different frequencies still – and here's where the whitespace exists.
If, for example, a transmitter in Wagga Wagga is using around 720 MHz as its frequency (UHF 55 – 718.5 MHz, to be precise), then that channel will be kept free for quite a distance – places like Cooma use it, but there are some mountains between Cooma and Wagga.
So if you're in a place where there is no nearby transmitter on Channel 55, the spectrum is owned by the broadcaster, but not used in your location. Whitespace proponents advocate creating wireless broadband networks that use whatever TV frequencies are free in a particular geographic region.
The idea looks attractive for several reasons, but the most important is that the frequencies used for TV transmission have good propagation characteristics: the signals propagate long distances, and are good at passing through walls.
As a result, a number of vendors – including Microsoft and the ubiquitous Google, but also companies more conventionally associated with radio transmission such as Motorola and Philips – are proposing whitespace broadband devices, and with the FCC's decision in America to permit the use of whitespace for broadband, such devices could begin shipping next year.
Where's the Catch?
The gotchas in whitespace broadband fall into three categories: regulatory, standardisation, and technology.
Let's start with regulation. To date, the US is the leader in the whitespace market – other countries will be following developments in America with interest, but so far, Australia's ACMA doesn't have any plans for whitespace broadband on the table. Since spectrum planning is a core activity of ACMA, you can bet that the issue is on the table, but our digital TV migration has a longer window than in America, so there won't be any rush to use whitespace for rural broadband in this country until the spectrum becomes available.
Nor is there any guarantee that Australia will follow the same regulatory path as America. That will depend on whose voices hold sway after the digital TV changeover is completed.
Standardisation is the second issue. At the moment, there are few standards covering the whitespace market, with the sole exception of a proposal to add 700 MHz transmission to the WiMAX suite of standards.
Part of the problem is newness. Although TV frequency-based wireless data systems have been researched for some time (look at Australia's own “BushLAN” experiments at the Australian National University, for example), with no strong market driver for adoption of these frequency bands, vendors have not seen any need to drive whitespace standards.
Without standards, however, both operators and consumers face the prospect of a vendor lock-in, something which history tells us will hold back the market.
Standardisation, however, needs the technologies themselves to be more settled than they are at the moment.
The Whitespace Technical Issues
There's no real secret to the business of sending data over radio waves: it's been happening for ages. What matters in the world of whitespace is the desire to create broadband networks in the available frequencies.
Interference is the first and most prominent issue. The development of the whitespace market will depend on proposed “cognitive radios” that are able to work out which frequency bands are available for use, wherever they happen to be deployed. This involves two pieces of “smarts” in the radio systems: the ability to detect other systems using the same channels (well understood in radio systems design), and the ability to work out which spectrum is allocated to TV broadcasting in a particular area. Most proposals accomplish this by combining GPS (to tell the radio system where it is) with a database lookup (listing the TV frequencies in use in a particular area).
A second issue has to do with how many radio channels are available. As I mentioned before, TV frequencies start down in the “VHF” range (well below 100 MHz) and reach up towards those already in use by mobile telephone networks (with a gap between VHF and UHF transmission that's allocated to other applications). However, not all of the television band is suitable for broadband networks.
Look, for example, at 802.11n: to achieve its best performance, it needs a 40 MHz radio channel. That's no problem up in the gigahertz range where WiFi operates, but the low VHF doesn't have room for such large channels.
Even with systems that don't have to operate at 802.11n speeds, the spectrum has to support a reasonable number of broadband transmissions in as small a radio channel as possible.
We can't get the same spectral efficiency as 802.11n, which needs highly efficient signals, but what if we matched 802.11G, which fits 54 Mbps into a 20 MHz channel and delivers around 11 Mbps to the end user?
To give a broadband customer (say) 5 Mbps maximum throughput at WiFi efficiency would require more than 20 MHz channels, and even using a WiMAX scheme (which is much more efficient) we would need perhaps 10 MHz to give a user that throughput (discounting the overhead of the radio system). If a given area has 100 MHz of “whitespace”, it would support ten connections per “cell”; or perhaps 100 users per cell at 10:1 contention ratio.
A final challenge has to do with the radio waves themselves. One reason higher frequencies are preferred for cellular applications is that they lend themselves well to being “beamed”. Up in the GHz it's much easier to create cells that don't leak, and the antennae can be kept smaller.
At lower frequencies, directionality is much harder to achieve, which restricts the usefulness of whitespace frequencies to the upper end of the TV spectrum.
And finally, although these signals can travel long distances, like all radio systems, they suffer from increasing noise – and a corresponding drop in user capacity – the greater the distance between transmitter and receiver.
Thus for the worst-served users in Australia, those who live the longest distance from all infrastructure including TV transmitters, radio-based broadband will be at its slowest and least attractive. However, at least those without any TV signal at all will have plenty of whitespace available to them.
The conclusion to be drawn from all of this is not that whitespace is not useful – although it is over-hyped in the current debate. It is not a panacea, and in particular, it won't solve the broadband problems of the worst-served 2% of Australia's population. It will be a useful option for those who live beyond ADSL but not beyond their local TV station.