The Universal Serial Bus (USB) makes it easy to use just about any kind of device on a PC. You just connect your gizmo with a cable and it works.
But a group called the Wireless USB Promoter Group wants to make USB even easier to useby eliminating the cable. This group, with member companies Agere Systems, Hewlett-Packard, Intel, Microsoft, NEC, Philips Semiconductors, and Samsung Electronics, is developing a wireless version of USB. A specification for Wireless USB should be complete by the end of this year, and products should be on the market sometime next year.
So why another wireless technology when we already have Bluetooth and various flavors of 802.11? For one thing, higher bandwidth. Bluetooth and 802.11 are good for sending data and documents over the air, but they can choke on very large files, such as audio and video. Wireless USB, with transfer rates up to 480 Mbps, doesn't have that problem. Another Wireless USB advantage is lower power consumption. Bluetooth is a low-power technology itself, but not nearly as low as Wireless USB. And Wireless USB's low power consumption is particularly attractive just where its high bandwidth is attractive, in consumer devices like portable music players and video players.
Works Like Wired USB
Besides higher bandwidth and lower power, Wireless USB's big appeal is simply that it's compatible with and works like the nearly ubiquitous wired USB. It doesn't implement a local-area network, like Wi-Fi, or a personal-area network, like Bluetooth, because a lot of people and a lot of consumer devices don't need networks. Wireless USB simply connects devices point-to-point, like wired USB, and transfers data between them. In other words, it does just what a lot of people want to do, and the way they're accustomed to doing it, but without wires.
Wireless USB also allows many devices to engage in wireless transfers at the same time. It can wirelessly link as many as 127 devices to a host (usually a PC), with each device getting its own time slices for transfers (Figure 1). Even while sharing Wireless USB's 480-Mbps transfer rate, several devices can operate virtually simultaneously and still transfer data very quickly.
Figure 1: As many as 127 Wireless USB devices can connect to a host.
The upcoming Wireless USB spec also lets some devices work without a PC host. Much like the technology of the upcoming USB-To-Go, it gives certain devices limited hosting ability so that they can link to each other. A digital camera, for example, might link directly to a printer instead of through a PC.
Underlying the technology of Wireless USB is Ultra Wideband (UWB), in particular a version of UWB known as MB-OFDM and promoted by the Multiband OFDM Alliance (MBOA). MB-OFDM employs multiple frequency bands and orthogonal frequency-division multiplexing (OFDM) and is one of two primary contenders for the technology to be approved for the IEEE 802.15.3a Ultra Wideband physical-layer (PHY) and media-access control (MAC) specification (Figure 2).
Figure 2: Wireless USB's MB-OFDM Ultra Wideband technology is a candidate for approval as IEEE standard 802.15.3a.
Ultra Wideband, as its name implies, operates over a very broad spectrumfrom 3.1 to 10.6 GHz. MB-OFDM, the MBOA's version of UWB, divides this spectrum into 14 528-MHz bands (Figure 3) and allows the selective use of bands. This approach allows it to skip over any bands that regulatory agencies in any country might decree to be off-limits.
Figure 3: MB-OFDM, Wireless USB's Ultra Wideband implementation, divides a 7-GHz spectrum into 528-MHz bands.
Because UWB transmits energy over such a wide spectrum, it must transmit at very low power to avoid interfering with numerous other wireless applications. In the US, the FCC has specified a maximum transmit power density of -41 dBm/MHzessentially the level of electronic noise radiated by common consumer devices. That low power works well at short ranges, however, so it's well suited to Wireless USB. Wireless USB's maximum data rate of 480 Mbps applies at ranges of less than 2 meters. At 10 meters, the specified rate drops to 100 Mbps, and beyond 10 meters, the signal level and data rate fall off quickly.
Low Power Consumption
Fortunately, low-power transmission means low power consumption, so Wireless USB is very kind to batteries. Wireless USB devices use far less power than 802.11 (Wi-Fi) devices, for example, although that's not surprising considering that their maximum range is much less. But Wireless USB also provides longer battery life than Bluetooth does, even though it operates at a higher maximum power level. The difference is due primarily to Ultra Wideband.
One power advantage for Wireless USB and Ultra Wideband is simply that UWB is very power efficient per bit of data transferred. Stephen Wood, an Ultra Wideband technology strategist at Intel, provides an example that shows how. If you assume about 800 mW for UWB and 100 mW for Bluetooth, says Wood, and if you assume 480 Mbps for Wireless USB and 1 Mbps for Bluetooth, then UWB is 60 times more power efficient per bit.
But Ultra Wideband's real power advantage, however, comes not during its "on" (transmission) time, but in its "off" time. "In real life," says Wood, "it's the quiet time that dominates the power consumption equation, and UWB is an extremely fast technology that gets on and off the air very rapidly, so Wireless USB devices can sleep a lot." Wood says that an Ultra Wideband radio in an MP3 player can provide several times the battery life of a Bluetooth radio.
An additional benefit of Ultra Wideband's low power output is that it helps keep Wireless USB devices from creating electronic interference in other devices. Because the radiated power from UWB radios is no higher than noise that devices normally radiate anyway, interference becomes noticeable only at very close ranges, such as a few inches. If you put a Wireless USB device right next to a cordless phone, you might cause a problem, but you could probably cause the same problem by putting any other electronic device next to the phone.
Ultra Wideband is also fairly immune to interference from other sources. Many of the likely interferersradios in 2.4-GHz devices, for exampleconcentrate their energy in only a narrow portion of the Ultra Wideband spectrum, and MBOA's version of UWB can avoid using those bands when it senses interference.
Wireless USB transmissions can easily be intercepted within a 10-meter range, however, so an "association" process is necessary to ensure that a particular Wireless USB device connects only to an intended host. For example, if you bring an Wireless USB-equipped MP3 player into your office, it might be able to "hear" half a dozen nearby PCs, but you only want it to "talk" to your own PC. To ensure that, you have to associate the MP3 player with your particular PC. After that, any time the device hears your PC, even in the presence of others, it maintains that association and won't wirelessly connect to any others.
As of now, the pending Wireless USB specification doesn't say just how device association will occur. A simple but crude way, according to Intel's Wood, would be to temporarily connect a wire for association only. Other approaches being considered would involve an infrared link or near-field communication working at a range of a few inches.
In addition to device association, Wireless USB will use other means, including encryption, to protect wireless data from unauthorized use. Encryption is not a requirement of the Wireless USB spec, however, and most likely won't be included in some, perhaps many, WUSB-enabled devices. Printer manufacturers, for example, might well omit it or make it an extra-cost option in order to keep costs low. And given that most WUSB signals don't get far, if at all, beyond the room in which they originate, many consumers might not want to bother with encryption anyway. One area where encryption probably will be used, however, is in media devices, such as audio and video players, that receive copyrighted material over the air.
Wireless USB will have to be compatible with wired USB, obviously, and this has been the plan of the Wireless USB Promoter Group all along. To an application program, there will be no difference. Wireless USB will look just like wired USB, with details handled by device drivers.
Potential incompatibilities do arise in the airwaves, however. A USB cable works equally well in the U.S. and Europe, but wireless spectra are controlled by different regulatory agencies in different countries, so certain frequencies that Wireless USB can use in one country might not be available in another.
Does that mean, then, that a Wireless USB product made for U.S. consumption won't work in Europe? Maybe, maybe not. MB-OFDM, the MBOA version of Ultra Wideband that's used in Wireless USB, has the ability to selectively use different frequency ranges within a much broader UWB spectrum. In theory, then, a Wireless USB device could be implemented in a way to make it operable in different countries using different frequencies.
Considerations of compatibility must also take into account a version of Wireless USB that already exists, from Cypress Semiconductor. The Cypress version has been around for a year or two, but its technology is very different from that being developed by the Wireless USB Promoter Group. It operates in the 2.4-GHz band, and its data rate is only 62.5 kbps. Cypress does have a version with a 50-meter range, howevermuch longer than the 10-meter range now in development by the newer Wireless USB groupin addition to a 10-meter version. Cypress sells its chips for use in input devices like mice and keyboards, and they cost less than $2less than the faster Wireless USB chips now in development are likely to cost for some time.
The newer Wireless USB chips now in development do have the potential to be very inexpensive, though. For one thing, because they need to emit so little power, they can omit some circuitry, such as power amplifiers. They can also be implemented entirely in CMOS, providing yet another application for Moore's Law. And, of course, their broad applicability to a ready and waiting market makes economy of scale possible.
The versatility of the newer Wireless USB chips can also lead to savings in a new product's bill-of-material costs. A single Wireless USB port added to a PC, for example, provides multiple simultaneous connections to all kinds of devices, whereas wired USB connections require a separate port for each connection.
For now, though, work on the new Wireless USB is still ongoing. The spec is still in development, for example, and chipmakers are still striving to meet the requirement for 480-Mbps transfers. Some companiesIntel, for examplehave already demonstrated that rate; others are close and zeroing in. In the meantime, the Wireless USB Promoter Group is planning an even higher 1-Gbps rate that probably won't see silicon for at least a year or two.
One important outcome of the Wireless USB development could be a resolution of the deadlock between two competing Ultra Wideband approaches. Both the MBOA's MB-OFDM approach and the UWB Forum's DS-UWB approach have been in strong contention for acceptance as IEEE standard 802.15.3a. By early 2004, however, neither camp had been able to muster the two-thirds majority of votes required for approval. Consequently, the IEEE work basically came to a halt and two camps went their separate ways. "To prevent delay," says Intel's Wood, "the MBOA is basically taking all of its spec work and completing it offline. IEEE is jammed, but we're going to finish the job and release the spec and build the product."
The MBOA, of course, hopes that the anticipated broad success for Wireless USB will give its own MB-OFDM technology a boost in the IEEE standards process. In fact, Intela key MBOA playerled the formation of the Wireless USB Promoter Group.
MBOA also boasts the endorsement of MB-OFDM by the WiMedia Alliance, an industry association that promotes wireless multimedia connectivity and interoperability for various technologies, not just Wireless USB. The WiMedia Alliance is working closely with the 1394 Trade Association, for example, in the development of a 400-Mbps Wireless FireWire. In the future, then, wireless versions of USB and 1394 could both have the same MB-OFDM base.
About the Author
Gary Legg is a Boston-based freelance writer. He holds a BSEE degree and is a former editor and executive editor of EDN magazine. He can be reached at firstname.lastname@example.org