With this months approval by the Federal Communications Commission of an Ultra-Wideband chip set, device designers and users can enjoy new options for robust, high-bandwidth wireless connections with minimal power consumption.
The XS110 chip set, now available in sample quantities from Freescale Semiconductor Inc., enables a claimed data rate of 110M bps over distances of 10 meters; the company also claims bit-error rates on the order of 1 in a billion at a power draw of three-quarters of a watt. (Full specifications are at www.freescale.com/files/
Freescales announcement calls the XS110 the first FCC-approved UWB chip set. This is not strictly correct—a chip set from Time Domain Corp., of Huntsville, Ala., was certified in September 2002. However, Time Domains offerings have so far focused on data rates of less than 10M bps, with applications in high-reliability communications and high-resolution radar and other somewhat-exotic realms.
Freescale does appear to be first out of the gate with a package aimed at mainstream consumer and enterprise desires for wireless connectivity in personal networks and LANs, with bandwidth sufficient for the most demanding multimedia data streams.
UWB radio technology is so different from prevailing radio practice that it almost seems like an entirely new engineering discipline. Conventional radio techniques treat bands of radio frequencies as almost the equivalent of real estate. The assumption thats built into almost all current radio equipment, and into national and international regulations, is that spectrum must be viewed as a collection of channels. Any given channel, within a vicinity whose size depends on a combination of frequency range and transmitter power, can be used for only one task at a time.
Some channels are quite narrow—for example, those used by radioteletype transmissions—while others, such as TV broadcasts, are much wider. An amateur radio operator using Morse code can occupy as little as 80Hz of spectrum. On the other hand, a U.S.-standard HDTV signal needs an 8MHz swath—enough to accommodate several tens of thousands of Morse code conversations, even if empty guard bands are left between adjacent pairs of signals.
To think of this in everyday terms, the conventional spectrum allocation process is like passing out differently colored filters with matching sets of colored eyeglasses to every person or group that wants to use flashlights. If only one group of cooperating parties illuminates objects of interest with a particular shade of red and only their glasses allow that single color to pass, then they can focus attention on any desired part of an otherwise-darkened landscape—without being distracted by anyone elses interests.
If no one is using an assigned color at any given time, thats just too bad. No one else sees any more than they would under worst-case conditions of all channels being active. Multiple user groups could agree to share a single channel, taking turns in the same way that multiple parties using handheld radios say “over” to indicate that its someone elses turn to talk, but parties have a strong motive to acquire spectrum rights based on their expectation of worst-case need. Economists cringe at the resulting inefficiencies, with a scarce resource winding up overpriced and underused.
A UWB approach violates all conventional notions of good radio engineering practice, using much more bandwidth than information theory requires for a given amount of content.
To put this in terms of nightscapes and flashlights, every UWB user has the same kind of white-light flashlight, but each is assigned a different distinctive timing pattern to use with a blinker switch—and is given a pair of glasses whose lenses can switch between being opaque or transparent, controlled by the same coded pattern.
At any given moment, a person will see glimpses of things that his or her communication partners didnt intend to illuminate, but, over time, that person will mostly see the targets that his or her own coded pattern is picking out. If other users or groups are idle, an active user will see a less confused signal or will be able to get a larger number of clear and distinct views over any given period of time.
As activity by other users increases, each user will be able to make an independent choice about trading bandwidth for quality. By settling for fewer independent views per unit of time, users will be able to use statistical techniques to elevate what they want to see above the noise. This phenomenon is called processing gain, in contrast to the amplification gain of a conventional volume control.
This is where UWB techniques start to defy common sense. A simple spreadsheet, using random-number formulas and a mix of logical and statistical functions, is all thats needed to demonstrate that a signal can be detected—at any desired accuracy level—even when that signal is actually weaker than the random noise of static and the other interfering signals in the same spectral band.
If the story stopped here, at the level of mathematics and information theory, then UWB proponents would therefore be justified in using phrases such as “spectrum glut”—as did the normally restrained international journal, The Economist, on its Aug. 14 cover. When the math meets the metal of actual radio hardware, though, things get much more complicated.
Strong nearby signals, for example, induce a behavior called front-end overload, which reduces the sensitivity of a radio receiver across its entire range. Moreover, some radio applications, such as radio astronomy, cant depend on Moores Law to make processing gain more affordable as channel activity grows; they have to detect the signals that they find.
An elevated noise floor, with ever-more users transmitting growing numbers of coded sequences, could drive astronomical users literally to the far side of the moon in search of a quiet radio shadow within which they could make their observations.
Freescales XS110 approval lets the UWB genie out of the bottle, even if the other messy contents of that bottle are so far being spilled only in the band between 3GHz and 10GHz—a spectral region of limited application because basic physics limit the range of signals in that band.
Powerful interests will almost certainly contend, however, that the benefits of UWB should be combined with the greater range and versatility of radio frequencies that are currently used by other established user communities. The contest that follows will be unlike any other in the history of technology—not a duel between alternative uses of a resource but one between essentially different ideas of what the resource is.
Technology Editor Peter Coffee can be reached at [email protected]