Ultra-Wideband: Multimedia Unplugged

PART 2 OF THIS SPECIAL REPORT DEALS WITH A NEW MODE OF VERY HIGH-SPEED, LOW-POWER NETWORKING CALLED ULTRAWIDEBAND

A funny thing happened to ultrawideband wireless technology on its way from the laboratory to market. It changed from a unique carrierless radio system into something far less exotic.

When the technology is finally standardized, it will be a carrier-based system most likely incorporating frequency hopping and orthogonal frequency-division multiplexing (OFDM). Its purpose, however, will remain unchanged: to replace almost every data cable in your home, even the ones going in and out of your television set, a job that requires moving hundreds of megabits of data per second. That's faster than all but the speediest of wired networks. The speed is achieved, however, over distances of only 10 meters or so [see "UWB Over Cable?"].

But it's the speed that has so many companies excited. Such heavyweights as Hewlett-Packard, Infineon, Intel, Microsoft, Mitsubishi, Panasonic, Philips, Samsung, and Texas Instruments all want a piece of the impending action. They formed the MultiBand OFDM Alliance in June and have come to dominate the IEEE's ultrawideband (UWB) task group that's writing a UWB standard.

Ultrawideband's new clothes have two tailors. The first is the growing recognition of the technology's commercial potential. The other involves the surprising limitations placed on it by the U.S. Federal Communications Commission (FCC, Washington, D.C.) in February 2002, which made the carrierless approach less attractive.John McCorkle, chief technical officer [see “FCC (Finally) See Little Noise From UWB.”] Now the large companies are vigorously promoting a technical approach different from the ones developed earlier by the smaller companies that started the commercial UWB movement.

John McCorkle, chief technical officer of XtremeSpectrum, smiles because he no longer needs to use cables like those he's holding. A plasma flat-panel monitor displays a wireless transmission produced using his company's ultrawideband set of chips.

Despite this change, UWB promises to revolutionize home media networking, taking over such tasks as downloading images from a digital camera to a computer, distributing HDTV signals from a receiver to multiple TV sets around the house, connecting printers to computers, and potentially replacing any electronic signal (not power) cable on the premises [see figure [PDF download]].

Indeed, UWB could be embedded in almost every device worthy of the use of a microprocessor. Products as disparate as toys, thermometers, and clocks could all benefit. By paying a modest fee, users could upgrade their electronic ”personality” toys to do more interesting things. They could download the upgrade and communicate between the computer and the toy via UWB devices in each.

Readings from electronic medical thermometers could automatically be entered into the electronic chart that records vital statistics of a patient being examined. Perhaps UWB technology could even get us all to work on time by continually checking our clocks against a network signal; it might also signal when a clock's battery is low.

High, wide, and deep

According to the FCC, ultrawideband is any signal that occupies at least 500 MHz of bandwidth in the 7.5-GHz chunk of spectrum between 3.1 GHz and 10.6 GHz. That definition also includes some rather strict limits on radiated power and power density: it's a lot less than the 3 mW allowed for a cellphone.

As originally developed by several start-up companies that borrowed the technology from U.S. military research, like XtremeSpectrum Inc. (Vienna, Va.) and Discrete Time Communications (now Staccato Communications Inc., San Diego, Calif.), UWB involves transmitting low-power streams of extremely short pulses--on the order of 10-1000 picoseconds.

Since such pulses intrinsically occupy a huge amount of bandwidth, their energy is spread thinly over a large swath of the radio frequency spectrum--from a few hundred megahertz to several gigahertz. These frequencies are so high that they can be transmitted directly, without first being modulated onto a carrier, as is done with conventional radio systems like AM and FM broadcasts, cellular telephony, and Wi-Fi.

The program information, be it a movie, song, or text message, is impressed onto the pulse train by varying the amplitude, spacing, or duration of the individual pulses in the train. This is different from the more conventional modulation techniques used in most digital wireless systems, which typically encode information in the form of changes in the phase of the radio wave. But the concept is the same: changes in some parameter of the transmitted signal--be it a pulse's position or a sinusoid's phase--carry the transmission's information.

The advantages of UWB are several. For one thing, it works well in crowded and noisy radio environments. Before the FCC put limits on the spectrum that a UWB signal could occupy, developers thought in terms of spectra that began somewhere in the vicinity of 100 MHz and stretched up to several gigahertz. Such a broad signal is quite resistant to interference because any interfering signal is likely to affect only a small portion of the desired signal. Also, if some frequency components of the signal have trouble penetrating the walls of a building, the theory goes, the majority will probably get through.

As far as causing interference is concerned, the pulses are at such low power levels that any equipment along their path cannot hear them. In fact, their power must be less than the level permitted for the incidental electronic noise generated, for example, by appliances or the switching power supplies in computers.

Not only do weak pulses prevent UWB systems from interfering with other wireless systems, they also make UWB inherently short range, a benefit when you want to operate multiple independent links within the same house. Best of all, for indoor applications, a UWB system deals well with multipath interference, which results when a signal bounces off one or more reflective objects on its way from transmitter to receiver.

Such a radio signal is received as multiple echoes, which cause distortion and fading. According to Martin V. Clark, a consulting communications engineer at The MathWorks (Natick, Mass.), and Moe Z. Win, the Charles Stark Draper Assistant Professor of Aeronautics and Astronautics at the Massachusetts Institute of Technology (Cambridge), if the delay spread of the echoes is much smaller than the system pulse width, the echoes can combine destructively, which leads to multipath fading. If, on the other hand, the delay spread is comparable to or larger than the pulse width, much of the multipath energy can be captured in the receiver, yielding superior performance.

For a transmitter-receiver separation of about 10 meters in an indoor environment, the delay spread is typically several nanoseconds--significantly more than a typical UWB signal's pulse width. In this kind of channel, a UWB signal is thus much more resistant to multipath interference than, say, an 802.11b (Wi-Fi) signal, with its much smaller bandwidth and much larger pulse width.

Interference worries

Yet those very benefits were undermined when the FCC, concerned about interference, especially to Global Positioning System (GPS) navigation receivers, set the 3.1-GHz lower frequency limit on the UWB spectrum. This severely reduced the amount of spectrum available and made the system less robust.

Other concerns were of possible interference with local-area networks based on IEEE Standard 802.11a, which operates in the unlicensed 5-GHz region of the spectrum. In fact, some members of the ultrawideband task group, IEEE 802.15.3a, proposed splitting the UWB spectrum into two parts, effectively avoiding the license-exempt 5-GHz region.

But creating the desired two-part spectrum by shaping the pulses is far from trivial. So the task group considered other routes. Actually, at least one company, Mitsubishi Electric Corp. (Tokyo), did develop a means for generating a two-part spectrum while preserving the original pulse-train nature of UWB, according to Andreas Molisch, associate professor at the Institute of Communications and Radio-Frequency Engineering at Vienna University of Technology in Austria. Just about everyone else in the task group, though, favored some sort of carrier-based solution.

Three companies (XtremeSpectrum, Motorola, and Partus-Cerva) preferred the direct-sequence code-division multiple access (CDMA) technology used by many cellular and personal communications systems. In fact, XtremeSpectrum has working silicon on the market today that delivers 100 Mb/s while consuming just 200 mW. But at least 20 enterprises supported a proposal by Intel and Texas Instruments that combined frequency hopping--jumping around from one part of the allowed spectrum to another at a rate of about three million hops per second--with OFDM. The technique uniting these two technologies is called MultiBand OFDM.

Orthogonal frequency-division multiplexing employs pretty much the same technique used in IEEE 802.11a and g. It is a carrier-based technique, but instead of using a single carrier, it uses multiple carriers, each of which may be controlled separately. Those that fall in noisy parts of the spectrum are modulated at low rates to make it easier to detect the modulation accurately. Those that exhibit a good signal-to-noise ratio can exploit their good fortune and carry more data. Individual carriers can also be completely disabled. Between disabling some carriers and controlling the system's overall hopping sequence, adjusting the shape of the occupied spectrum to avoid that ”forbidden” region around 5 GHz is fairly easy.

Getting the details right

For important aspects of UWB, such as security, automatic user recognition, authentication, and authorization, designers are paying attention to lessons learned in implementing previous generations of wireless. UWB will be able to incorporate a variety of next-generation security mechanisms developed for IEEE 802.11. Developers of UWB intend to leverage the considerable body of work on Plug and Play for the Universal Serial Bus (USB) and FireWire (IEEE Standard 1394).

One issue to which the IEEE 802.15.3a task group is specifically devoting a lot of time is authentication. Imagine two consumer devices such as a video display and a DVD player, both with UWB interfaces. How do you or they determine that it's O.K. for them to talk to each other while they are not to communicate with a similar UWB device a wall thickness away in the next apartment?

This is the kind of problem addressed, for example, by Bluetooth. Bluetooth was only recently developed to do much the same short-range, wireless low-power transmission as UWB, only at much lower data rates [see “Comparing Wireless Technologies (Roughly)”]. Now Bluetooth, with UWB nipping at its heels, could be dead, taking with it millions of dollars of investment and many R&D and standardization-body man-years.

But its authentication solutions will live on. UWB might require something like an ”authentication button,” which the consumer would press at the same time on the two devices, as a way of introducing them to one another.

Intel Corp. (Santa Clara, Calif.) is particularly interested in applying UWB to what would in effect be wireless USB connections. That interest has grown ever since company researchers realized that such systems could be built with standard CMOS processes, one of Intel's core competencies. (XtremeSpectrum's chipset includes one silicon-germanium chip.)

Intel isn't the only company working on UWB chips with standard CMOS. Staccato Communications has been working on UWB since 1996 and expects to unveil its first product sometime in 2005. ”CMOS is the dominant semiconductor process for very good reasons--size, cost, reliability, power consumption, and performance,” says G. Roberto Aiello, Staccato's founder, president, and CEO. ”We feel that those advantages are even more applicable to UWB devices.”