Features

Why Intel is looking at Soft Radio and UWB ...

Is it the prospect of a personal area network running at 500 megabits per second? Is it the ability to see through walls? Or is it the temptation of developing something like "ultra-spread spectrum"? Intel Fellow Kevin Kahn, explains why he's studying ultra-wide band radio technologies.

"Most of the interest Intel has in UWB radio, is capacity," said Dr Kahn in a briefing session at the Intel Developer Forum today, "but it has lots of potential, in several areas. It is particularly interesting to us, because of the research we're doing into soft wireless; configurable radios on CMOS silicon."

Ultra-Wide Band radio has been given astonishing popular hype recently, because of lay magazine which revealed that it can "see through walls" and pinpoint bodies - and so might be useful to military. It is, actually, something which is at least three years away from any possible commercial exploitation. And there are several reasons why it is hard to do, says Kahn, and several others which make it very attractive.

"For example, it's called pulse radio. It's based on very short sharp pulses; so you get very accurate time measurements between transmitter and receiver; which means you can do very good positional estimates. That interests a lot of people. It would work inside warehouses, for example, for locating things. People following that route aren't so much interested in data rate; they'd turn data rate down and increase range; and of course, you have a choice. We're interested in high data rate, over short ranges of up to 5m, no more."

Kahn was very keen that nobody should confuse this with USB (universal serial bus) version 3.0. "This has nothing to do with it, apart from the fact that theoretically, we think we can get up to 500 megabits, which is very similar to USB 2.0 at 400 megabits. So you'd be able to do the entire interconnect between components of a computer using UWB instead of USB, if you could make it work."

The system works on the basis that if you spread these pulses with very low power indeed. you generate a sort of ethereal buzzing which is indistinguishable from background radio noise. "The FCC has approved it for experimental work between 3.1 GHz to 10.6 GHz; and in Europe, it is still illegal, but ETSI is looking at a very similar spectrum mask. Etsi is being a bit more realistic, in that it doesn't have the sharp cut-off that the FCC has asked for, which isn't really feasible. And while it's not legal in Europe yet We got a waiver for the demo here at the Forum, pretty easily."

Kahn's team demonstrated 100 megabit data transmissions, using something you wouldn't want in your home; two big circuit boards with funny pixie-hat antennae, huge copper coils, and lots of discrete circuitry. It caused a stir, despite the fact that it's a trick he's pulled off several times since February, when the FCC licensed it. But although Kahn was happy to talk about it, it's clear it's just one of many technologies he is interested in.

"There are still huge areas not decided. Nothing comes free. Receivers are tricky; they involve detecting very, very short pulses, arriving rapidly. It's spread over a huge amount of bandwidth; there are lots of radio systems living in that. So you'll see narrow-band "jammers" - people who look like jammers to you, but who are legitimate users of their own licensed frequencies! - and you have to cope with that. Then, as you go to higher data rates, and pulse repetitions go up, you're prone to inter-symbol interference, which you have to deal with too.

But primarily, it can't become commercial, because "there are no standards for how you'd deploy it. One question is 'how do you arbitrate the channel?' You need to ensure that if several UWB users are in the same area, they can work out who is talking to whom. That has not been worked out. Then there's issues with antenna design; most antenna design is typically for narrow band; it has to be a different kind of antenna; and it has to operate with huge amounts of bandwidth."

But mastery of this technology could open a lot of other gates.

Intel is convinced it can put a "soft" wireless onto a CMOS chip; one which operates at several frequencies. And this could be very useful, if it could ever be negotiated. "The interesting thing about spectrum is how much unused spectrum there is. You go out into the street with a spectrum analyser, and you'll be amazed how much spare spectrum there is. But it's not the same this morning as this afternoon; it's not the same here as in Munich city; it's not the same here or in Japan. If you want to use unused spectrum, it's a very complex task. It's not just a technical challenge; it involves economists. We had a meeting the other day; Michael Powell, chairman of FCC, several high level economists, politicians, and techno-geek types like me, all spent a day and a half, all discussing how you might begin to move towards it."

He looks wistful: "It's an alluring idea, but a very long-term prospect; and IF you went there, you'd want much more flexible radios. And if we could build soft wirelesses ... "