Wireless raises the final standard

Rupert Goodwins: The strangest wireless system has become more mainstream, but may still be the last thing you need.

When it first materialised, ultrawideband radio -- UWB -- looked like alien technology gone astray. Since Marconi first put finger to Morse key, radio has worked in the same way. First generate a pure radio wave on one frequency -- the carrier -- and then wobble it up and down (AM) or from side to side (FM) to get the message across. UWB had no carrier. It had no frequency. It had a stream of tiny pulses, little chirps of energy that had a shape and a time. The receiver knew when to look for them, and if it could count enough in the right place it knew it had a signal.

Such an outré scheme -- and others that shared the carrierless concept -- had many advantages. They needed very little power. They could be very fast. Most excitingly, they could exist alongside existing radio transmitters without either side interfering with the other: it's as if someone had invented magic fairy wings you could strap to your sneakers that let you zoom above roads choked with cars. But there were still fears that by reusing frequencies, it would interfere with systems that rely on weak signals from afar -- such as mobile phones, satellite navigation and radio astronomy.

After much heated discussion the American regulator, the FCC, decided to let UWB out of the lab, provided it followed a very strict set of guidelines. It could use most of the frequencies it wanted, but at levels lower than the noises non-transmitters make. In other words, a UWB transmitter had to emit less radio power than, say, an ordinary CD player.

With a whisper of delight, the UWB radio industry set to work. It soon became clear to a number of companies operating within the constraints of the FCC envelope that while the original idea had a lot going for it, it wasn't a good fit for the mixture of PCs, consumer electronics and mobile phones on the market. Last week, an industry consortium largely formed from members of the IEEE 802.15.3a committee in charge of such things collapsed 16 different proposals into one -- leaving six or seven others outside the magic circle. By the end of this week, the chances are good that these too will have been adapted, adopted or deleted, leaving a fast path to full approval.

The new proposed standard looks very different from the original idea. In place of a sea of pseudo-random pulses, there are tightly regimented bands 528MHz wide, each populated by hundreds of individual carriers at 4 MHz intervals. Broadband geeks may think there's something familiar about this picture, and there is. OFDM, Orthogonal Frequency Division Multiplexing.

OFDM is already in use everywhere -- it's the Ford Transit of the data communications world. As well as wireless networking, it's at the heart of digital radio and television, DVDs and ADSL. It's well understood by designers, and there's a lot of working silicon. The consortium -- Multiband OFDM -- has combined that with very wide radio bandwidths and low powers to get a system that can use from three to 13 bands, offering a mix of power consumption, speed and resistance to noise that it thinks will fit most needs.

By the end of next year, expect to see product that runs at around 500Mbps over a few metres -- and expect to see that built into all manner of consumer electronics. The big vision is that you buy a new wall-hanging 40" LCD TV, for example, and just run power to it. Everything else -- signals from your satellite receiver, an IP connection to your AVI server, streamed video from the mobile phone acting as a babycam in the nursery -- is wireless. If the figures pan out, OFDM UWB will do much better than 802.11a or g on less power.

So if UWB is so good, why will we want 802.11? It's not possible to up the power on UWB to give it a decent range without breaking the FCC rules, which would seem to give existing networks a long lease of life. Newer technologies that are yet to become established, such as Zigbee, will have the choice to absorb UWB ideas or be bypassed, but older stuff will have to depend on the protection of standards and regulation to fend off the upstart.

Even here, UWB has a trick up its sleeve. The full 13 bands stretch from 3.4 to 10GHz, and any radio that covers such a huge sweep of spectrum will be able to do other things too. UWB could see the first widespread deployment of software radio, where the hardware is flexible enough to be reprogrammed to any standard and set of frequencies you can think up.

At that point it doesn't matter whether your radio chip comes out of the box singing 802.11, UWB, Zigbee or the Girl from Ipanema. It'll be smart enough to listen to the conversations going on around it before launching in -- and once enough people have got soft radio networks, new standards can be rolled out over the Net. And that will be that for the problem of what to buy and when to buy it. UWB could be the end of IEEE number madness: perhaps the greatest advantage of them all.