Intel has publicly demonstrated its first working ultrawideband wireless link recently at an engineering session of the Intel Developer Forum in San Francisco.
The system, as raw an engineering prototype as it is possible to build, transferred data at 100 megabits per second (mbps): the company has set itself the goal of demonstrating it working at USB-2 speeds, or around 500mbps. Regulatory approval for such systems in the US was only given on 14 February this year, a fortnight before the demonstration, and regulatory concerns continue to be as thorny as technical considerations, if not more so.
"The FCC (the US communications regulator) received over 900 submissions concerning its approval of UWB," said Ben Manny, director of residential communications at Intel's Architecture Labs. "We think ultrawideband has the potential to be a classic disruptive technology."
UWB is a radical radio technique that transmits a stream of very low power pulses without a central frequency. Theoretically, these could be broadcast anywhere in the entire radio spectrum, but concerns about interference with existing services means that they have to avoid some areas such as those used by satellite navigation systems and cellphones, with most of the transmitter power falling between 3GHz and 10GHz where existing services are fewest. Spectrum owners are concerned about UWB both because of the potential of interference, and because UWB is in effect using parts of the radio spectrum that the existing owners have paid for.
Two breadboards festooned with copper coils -- "we call it the still: everyone else is very suspicious about what we're really up to," said one engineer -- were placed a few feet apart and hooked up to a pair of computers. Seconds later, an on-screen meter showed 100 megabits a second pouring across the ether. It can be difficult to tell whether a demonstration is really showing what it purports: in this case, the look of relief and delight on the faces of the engineers left the crowd in no doubt. Another on-screen meter showed that the bit error rate (BER) was in the region of three or four errors per thousand bits: impressive for raw wireless, even more so from hand-built prototypes that wouldn't have looked out of place on Marconi's workbench. As one antenna was shielded from the other by hand, the BER went up and the throughput went down. It was real.
Because the system was so new and untested, the engineers couldn't say how much power per bit it was using to transfer that 100mbps. However, one of the chief theoretical advantages of UWB is that it needs far less power than existing systems: under the current restrictions, UWB systems will have to radiate no more power than current electronic devices that are not designed to transmit at all.
There are many practical problems to overcome before UWB is ready for products, according to Manny. One is the antenna: most antennae are tuned to a particular frequency, but UWB needs precisely the opposite. The Intel rig used two mushroom-shaped devices known to radio engineers as solid discones; these have a very wide bandwidth but radically new designs will have to be developed for commercial uses.
UWB has many advantages over carrier-based systems. It is much more immune to multipath distortion, where a signal from a transmitter goes through two different paths before reaching the receiver; the different time delays on the two paths shifts one out of phase with the other and causes distortion when they meet. A UWB receiver receives the main pulse from the transmitter first, and can ignore subsequent reflected pulses: alternatively, it can work out the timing difference and look for the reflected pulses to improve throughput. UWB is a very low energy system, so every little extra picked up helps.
UWB behaves differently in other ways. While carrier-based systems such as UWB slow down just about linearly as over distance, UWB falls off exponentially. According to figures shown by Intel, a 500mbps connection at five metres will fall to 10mbps at 20 metres. However, it is possible to get further distances at low speeds with UWB: amateur experimenters using similar techniques have established world record distances for low-power communication. Also, one of the other unique attributes of UWB is that it knows where it is: by measuring precisely the transit time of the pulses, the relative location of transmitter and receiver can be known to around a centimetre. One idea is to use this information for beamforming, where the transmitter uses steerable beams of power to concentrate transmissions in the direction of the receiver.
Standards bodies are already moving in on UWB. The American IEEE 802.15.3 -- high-speed personal area networks -- study group got going in November, and the UWB physical layer task group should be going by July. This will be an industry consortium effort led by the various UWB start-up companies who've been with the technology the longest, with Intel, Sony, TI and Motorola involved. Meanwhile in Europe, ETSI TG31 is working on the issue, with good liaison between the groups. Nobody wants a repeat of the geographical problems that have dogged nearly every wireless standard for the past 20 years.
The standards bodies face some unique challenges. First, UWB is illegal in most countries: radio regulations are exclusively designed to cope with systems that fit neatly into bands. Another question is whether standards should cope with both main uses of UWB, location and communication. Different pulse techniques and powers might be required for the two aspects, but they share the same band. In effect, all UWB devices are on the same channel so they can't be partitioned as classic radio systems can, by frequency. And interoperability is an issue: how must the pulse be defined so that every vendor wanting to build stuff to the standard knows how to do it? None of these problems seem to be likely to be showstoppers, but they do make UWB a rather different proposition than any previous system.
Intel sees UWB as being a good cable replacement technology, and is also considering things like a wireless digital video link -- laptop to projector, home theatre cabling, or a future Bluetooth. It expects initial uses to be where existing cable is hard to install, tethers equipment that would otherwise be mobile or is just plain ugly.