Just when you’ve mastered the complexities of Wi-Fi standards like IEEE 802.11b and 802.11g, and after you’ve figured out that Bluetooth can work if you perform the steps just right, a new wireless player ambles into town. Make room for ultrawideband (UWB), which will probably be the death knell for cable USB connections in 2007.
Yes, yes, you’ve heard it before: Bluetooth has been on the verge of killing USB for a few years now – but that was always hype. Bluetooth typically runs at just 1 Mbps (the latest Bluetooth 2.0+EDR version, which ships on new Macs, is 3 Mbps). In comparison, the original USB 1.1 operates at 12 Mbps, while the current USB 2.0 flavor carries 480 Mbps. Bluetooth’s reach may number in the tens of millions of devices, but it didn’t kill USB.
So why should you pay attention to another wireless contender? UWB is a radically different approach to wireless data exchange that boasts the raw speed and flexibility necessary to become a peripheral replacement with less of the irritation that accompanies Bluetooth pairing.
UWB: Short, Fast, Low — UWB is a relatively recent wireless networking approach that turns Wi-Fi, cellular, and other wireless networks on their heads. Existing standards typically use very narrow slices of radio spectrum and pump as much signal power as possible through that band to get the greatest range and highest throughput. UWB uses a literal ultra-wide band – a swath of spectrum that’s several gigahertz wide, hundreds of times wider than almost any existing wireless technology.
Since UWB devices use extremely low-power signals, their chatter is more or less undetectable by other equipment using the same range of spectrum. UWB pulses are very brief as well. Because of the low power and desire by the United States Federal Communications Commission (FCC) and other international regulators to avoid trampling on existing uses, current UWB standards can send usable signals only about 100 feet (30 meters). (The one technology that can faintly hear UWB? Wi-Fi operating in the unlicensed 5 GHz band; UWB has to "notch" or avoid transmitting across part of that range by FCC rules.)
Within 100 feet, the near-term versions of UWB that will hit the market can exchange data typically at the full rate of 480 Mbps – the same as USB 2.0. Beyond 100 feet or if obstacles are in the way, UWB rapidly drops in throughput. Signals can be detected at hundreds of feet, but reports indicate that only a few Mbps would be possible at that range, which is one reason why UWB won’t replace Wi-Fi or Ethernet. Some newer flavors in testing can top 1 Gbps. There’s more room in the technology in the future, too, as regulators may allow higher signal levels or even wider swaths of spectrum to be used, while companies become more clever about encoding data.
As in so many areas of computing, the standards process has reared its ugly head. An IEEE committee – named 802.15.3a for reasons you likely don’t want to know – deadlocked about two years ago about what form of UWB to use as a short-range network standard. The only thing the group proved capable of agreeing upon was to disband last month.
Two opposing alliances that formed within that standards group remain: One comprises a few companies, foremost among them Freescale. Freescale is the semiconductor spin-off of Motorola and incorporates the Motorola acquisition of UWB pioneer XtremeNetworks. Their form of UWB is now seen as classical, using the entire stretch of FCC-allotted spectrum.
The other organization, the WiMedia Alliance, has Intel and a host of other chip and electronics firms as members. The group’s approach encompasses both radio technology and higher-level applications, such as hard-disk mounting over UWB and TCP/IP networking over UWB. The WiMedia Alliance merged with the Multi-Band OFDM Alliance, which divides the FCC-allotted spectrum into a few pieces and then uses orthogonal frequency division multiplexing (OFDM), in each piece. OFDM, also used in Wi-Fi flavors 802.11a and 802.11g, divides up spectrum into smaller subchannels, each of which sends data quite slowly compared to the overall bandwidth to make it easier to reconcile signal reflection and cope with interference that may exist in only a small part of the band in question.
The two UWB versions are incompatible and both claim performance, spectrum utilization, and manufacturing-cost benefits.
UWB’s Place in the Ecosystem — UWB by itself is just radio technology that sends data among compatible devices. Networking devices succeed or fail by the layers on top of the raw physical part. Ethernet’s success came in part from an ecumenicalism that allowed many different protocols like TCP/IP, AppleTalk, and NetBEUI to run seamlessly on the same medium.
On top of those protocols sit applications that make use of packet delivery and routing over a network medium. For instance, AppleShare Filing Protocol works over AppleTalk and TCP/IP using any network medium on which AppleTalk and TCP/IP operate (Wi-Fi, Ethernet, and so on). Safari uses HTTP, which runs over TCP/IP, which operates over Ethernet, DSL, and other physical media.
The WiMedia Alliance expects to release its own TCP/IP stack that works over its UWB flavor, and has partnered with the Bluetooth SIG, the USB Implementers Forum, and the 1394 Trade Association (the folks behind the specification underlying FireWire).
The Bluetooth SIG is trying to avoid being stuck with its single radio technology and will work with Freescale as well to make its applications available: object exchange (file transfer), dial-up modem (remote connection), fax, business card interchange, audio, and other features. Because programmers already know how to work with Bluetooth applications, it’s a simple matter to make those same features work over UWB.
The USB Implementers Forum has a Wireless USB specification that the WiMedia members’ adapter will work with. Likewise, the 1394 Trade Association has a set of digital video transfer protocols that will work as well over UWB as over a FireWire cable.
The ultimate goal of the WiMedia Alliance is that new computers will contain a single radio that will be able to handle Internet or local networking over TCP/IP, applications over Bluetooth, hardware over USB, and video over 1394 simultaneously. That’s a serious number of cables that could disappear.
But reaching this goal requires drivers and hardware integration, a point that’s seen as still many months away, and may require Microsoft, Apple, and Linux backers (via IBM, HP, and other companies and individuals) to integrate UWB support at the operating system level. The first phase, therefore, will be driverless.
Without drivers, UWB devices must emulate existing cable standards. The first wave to hit the market from several different companies will almost certainly be USB 2.0 via UWB. In this scenario, a dongle will plug into your computer’s USB 2.0 port while another dongle or a hub will be at the other end of the connection. They’ll package USB 2.0 traffic within the UWB connection, looking just like a USB cable to the computer.
This first wave will probably include sets of equipment that are locked to each other: only a dongle and its paired hub or dongle will be able to communicate. Later hardware will add generalized pairing between compatible devices. Freescale and WiMedia have talked about pairing devices by pressing buttons on the side of desired devices, using near-field communications (bringing the devices very close and pressing a button or using software), or software configuration.
Ultimately, the radio will just be built into most computers, like Wi-Fi is now, and the application and network protocol layers will require no extra work to support.
Where We’ll See UWB First — The first generally available device in the U.S. may be a USB 2.0 hub that Freescale has licensed for production to two companies familiar to Mac users: Gefen, which specializes in video interconnection and extension, and Belkin, makers of networking equipment, cables, and a host of audio and iPod accessories. (Freescale has a Chinese partner, Haier, which will incorporate its equipment initially only in domestic Chinese consumer electronics.)
This hub uses a USB dongle powered by a computer’s USB bus and a separate four-port USB hub that requires AC power. This first flavor runs at just over 100 Mbps, or above a fifth of the speed of UWB’s early potential. Belkin expects to ship their version in July 2006 for roughly $130, according to Ben Bamdad, a Belkin product manager. (The press release linked below mentions their original planned shipping of several months earlier than July.)
While USB 2.0 is useful for a variety of peripherals, such as printers, scanners, and hard drives, it’s likely that battery-powered portable electronics such as cell phones and MP3 players will eventually receive the biggest benefit from UWB because of its extremely low power usage. Wi-Fi is great, but even the lowest-powered chips designed for handheld devices will burn much more power than UWB radios.
UWB will also certainly find its way into consumer electronics because of the speed and potential simplicity. Imagine purchasing an LCD television/monitor, a DVD player, a digital home device (a Mac mini or a Microsoft media center), a stereo receiver, self-powered speakers, and a set-top cable/satellite receiver, all of which use UWB – you’d eliminate dozens of feet of different (and often wildly confusing) wires right there.
If the WiMedia Alliance’s vision comes true, which I expect to happen with Intel behind it, a single radio would enable communication among all categories of devices using all types of standards: a Sony camcorder would play via any brand of TV, but also push standard DV or HDTV video to a computer.
If someone could just get to work on practical wireless power, we could cut all cables. But that’s still science fiction.