Ultra Wideband's Ultrawide Ambition

We've been hearing about Ultra Wideband (UWB) for years. Once hyped as a way to achieve much higher data rates than today's Wi-Fi networks, it's the most controversial technology in wireless networking, thanks to the way it uses spectrum. Still illegal in most of the world, the FCC has approved it for use in the United States, and the first products are on the way.

But don't get too excited. Because UWB broadcasts on almost all radio frequencies at once, rather than sticking to an assigned band like most wireless technologies, the FCC has limited its power output to reduce interference. This in turn reduces its range, with the exact figure depending on the data rate. If you only need few Kbps, it can go up to 100 feet. If you need a full Gbps, it can only travel a tenth as far.

Even worse, squabbling vendors have been unable to agree on a standard. The Intel-led WiMedia Alliance looks set to be the main standards body, as its specification has been adopted by most of the PC industry. But there's also a rival group, the Motorola-led UWB Forum, which is targeting cell phones and could be important internationally. Both have demonstrated hardware and announced products due later this year, but these won't be interoperable with each other.

The first applications for both standards will be consumer-focused--for example, beaming photos from digital cameras, music to MP3 players, and home movies from cell phones. But many vendors in the WiMedia Alliance also plan enterprise applications, including peer-to-peer file transfer and wireless links from laptops to overhead projectors. It even opens up the prospect of interoperability in new areas, such as laptop docking stations. Instead of needing a proprietary interface customized to a particular laptop make or model, users will be able to connect to any WiMedia monitor and port replicator.

BAND WAGON
The rival specifications are very similar. Both offer the same trade-offs between data rate and distance, though the Forum's version extends to longer ranges and higher speeds (see figure). Unfortunately, seemingly arcane differences at the Physical layer make interoperability impossible. The Alliance's technology, known as Multiband Orthogonal Frequency Division Multiplexing (MB-OFDM), divides the radio spectrum into 14 channels, each for use by a different network. The Forum's, known as Direct Sequence UWB (DS-UWB), divides it into only two and uses codes to allow each one to be used by several networks at once.

Some devices may eventually support both standards, and the Forum has even proposed a specification known as the Common Signaling Mode (CSM), which would enable devices to adapt based on which standard is most appropriate. However, none of the Alliance members have accepted this, and there's little overlap between the two groups' membership. Siemens is the only major vendor to have joined both.

Of the two specifications, the WiMedia Alliance's is the one that most enterprise applications will use, thanks to its strong support from the PC industry. However, even standardizing around WiMedia may not be enough. The protocol that the Alliance has specified covers only the Physical and MAC layers, leaving vital functions such as security and the actual data transfer to higher-layer protocols. Four of these are under development so far, and more may be on the way.

USB OVER UWB

The higher-layer protocol closest to fruition is a wireless version of the Universal Serial Bus (USB), the port that links PCs to most peripherals. Although some of these devices can already connect wirelessly, they do so via Bluetooth, Wi-Fi, or infrared, none of which are ideal. The sub-1Mbps data rates of Bluetooth and infrared restrict them to keyboards and mice. And while Wi-Fi is faster, its complex security and high power consumption are overkill for cameras and MP3 players. Even setting those factors aside, Wi-Fi's 30Mbps doesn't compare well to the 480Mbps of a USB 2.0 cable.This is the same as the maximum data rate of the WiMedia UWB variant, and that's not a coincidence. Intel is also the main force behind Wireless USB, and it developed the protocol with that in mind. The rest of the Alliance was only too happy to go along, and not just because many of them are start-ups dependent on Intel's venture capital.

The theory is that USB is already the world's most popular communications standard. It has an installed base of more than 2 billion devices, with around 600 million more shipping each year. The Alliance hopes this will provide UWB with a ready-made mass market.

Most of the 2 billion USB plugs won't be replaced by UWB, of course. A USB port provides power as well as communications, which a wireless link can't. Everything will need either its own electrical connection or a battery, which will likely mean that flash drives and other devices intended to be small and cheap will still need to be physically plugged into PCs. The same applies to input devices, though researchers are working on ways to generate power from the motion of a user pounding on a keyboard or clicking a mouse.

However, the lack of power isn't a problem for devices such as printers, PDAs, and cameras, which are likely to be UWB's earliest adopters. And it can even open the way for new applications, such as peer-to-peer file transfer between laptops. Unlike the legacy parallel and serial cables it replaced, regular USB can't be used for direct PC-to-PC links. According to the USB Implementers Forum, which oversees the standard, some vendors have tried. Unfortunately, their cross-cables can link the power supply of one machine to the data bus of another, destroying both.

Wireless USB has other advantages over its wired predecessor. The present USB architecture requires a hub, just like traditional Ethernet. Most PCs make this invisible by including a small hub of their own, but it becomes an issue when additional ports are needed. The wireless version uses the airwaves as a hub, allowing ad hoc networks of up to 128 devices.FIRE WIRELESS

Although USB is important, it's far from UWB's only proposed Application layer. The specification is extensible to support almost any protocol, and so far the WiMedia Alliance has specified three more: native IP data, the consumer-focused Universal Plug and Play (UPnP), and wireless IEEE 1394 (FireWire), a protocol designed by Apple for transferring large multimedia files.

The precedents for this aren't encouraging. When cell phone vendors developed Bluetooth, they standardized nine different Application-layer protocols, several of which were never adopted. With the first WiMedia hardware supporting only Wireless USB and the other protocols not due until next year, they too might never see widespread use.

Respond to Chief Technology Editor Andy Dornan at http://blog.networkmagazine.com.Lowdown

Promise: No more wires. UWB will first enable high-speed file transfers between any two devices in close proximity to each other, then replace the cables linking PCs to monitors. It will eventually enable self-organizing meshes.Players: Intel and Motorola are leading rival groups, corresponding roughly to the computer and telecom industries. The two groups have similar aims, but their specifications are incompatible. While neither specification is yet complete, the chances of interoperability are slim.

Prospects: UWB gadgets will be in shops in time for this year's holiday season, but they'll likely just be for consumers--mostly digital cameras and MP3 players. Enterprise equipment, such as overhead projectors and laptop docking stations, are unlikely to ship until at least mid-2006.