UWB Takes Flight

Periodically, a new communications technology becomes the darling of literature, conferences, seminars and startups. Relevant examples include MPLS, trellis-coded modulation, Bluetooth and turbo codes. The latest technology generating this type of buzz is ultrawideband, which arrives to compete with multiple in-home wireless-networking initiatives.

My childhood TV watching occurred in pre-Nickelodeon (not prehistoric) years. Thus, there was no SpongeBob SquarePants, no Rugrats, no Hey Arnold! or any of the other cleverly bizarre series that both adults and kids can find amusing. Back then, both animated and nonanimated series still involved superheros and similar "good guys" such as Batman, Underdog, Mighty Mouse and my childhood favorite, Hong Kong Fooey.

There were shows I really disliked as well—Racer and Lost in Space come to mind, not to mention Ultraman. I'm not sure why, but I just thought it was a dumb premise—a giant heroic robot who lived in a missile silo and looked exactly like a guy dressed up in aluminum foil, beating up other guys in stuffed monster costumes.

Adding to the distaste for all things "ultra" was the fact that, when I was in elementary school, the use of "ultra" in general was very not cool. It was, in fact, derisively sarcastic to call something ultra anything.

Thus, you can see why it has taken me a long time to warm up to writing about ultrawideband (UWB). The other conundrum to be faced when using hyperbole like ultra is, what do we call the "next" generation of UWB—U2? Anyway, for the sake of CSD, I will nonetheless bravely plow ahead with an overview of this appealing new technology.

Fundamental concepts
So, what is ultrawideband? Well, for that, we can go right to the UWB networking group's Web site and find the FAQs, where we learn that "Ultra Wideband Radio is a revolutionary wireless technology for transmitting digital data over a wide spectrum of frequency bands with very low power." There is a bit of marketing in that answer (i.e., "revolutionary"), as there is apt to be in a promotional type of Web site. Nonetheless, there are a couple of very different things about the way UWB systems signal-things that make the "innovative" synonym available in Word for the word "revolutionary" appropriate, if not the "world-shattering" one.

First, UWB, by definition, implies a signal that has a bandwidth of at least 25 percent of a center frequency, or wider than 1.5 GHz. Different literature places moderately different numbers here, with this particular one based on the most recent paper I reviewed in conjunction with an IEEE Symposium (Leeper). Nonetheless, it is the theme that is important-the power spectral density extends across significantly more bandwidth than any commercial technology used in communications today.

A second key difference is the absence of a traditional carrier signal that takes some baseband digital information and modulates it onto a frequency at which the system actually operates. This type of RF carrier is a staple of both narrowband and existing broadband communication systems. The well-understood sinusoidal carrier signal approach to communications is thus cast aside-one reason why understanding UWB systems can be fitful for some. The built-in grasp of an RF communication channel as a frequency is difficult to overcome.

Rather than implement sinusoidal carriers, UWB uses extremely narrow pulses as the "carrier" for modulation by payload data. These pulses are on the order of a nanosecond and possibly as low as tens of picoseconds. Pulses of RF "carriers" to definitively create a "center" frequency have also been used. However, given the duration of the pulse, to call it a carrier is misleading because of the very few cycles of a sinusoid that fit within a pulse duration.

Figure 1 and Figure 2 compare the time- and frequency-domain versions of carrier-based RF signaling and UWB signaling. The nature of the signal is that of the mathematical, but fictitious in real life, impulse function. Because of this, UWB is often also referred to as "impulse radio." The impulse function, if you recall from your undergraduate signal-processing work, is a signal of infinitesimal duration and infinite amplitude. It is, therefore, somewhat difficult to implement in practice. To add to the mystery, the "area under the curve" if you were to integrate the impulse function is unity for a unit "weight" impulse. This can be somewhat confusing for those used to seeing a similarly flat spectrum and associating it with a noise spectrum. The important difference here is that the impulse function spectrum is frequency content of a deterministic signal. On the other hand, the noise spectrum—which is actually a power spectral density (PSD)—is flat because of the statistical properties of noise (and thus a PSD description), and specifically due to the fact that the autocorrelation function of ideal additive white Gaussian noise is an impulse. In other words, a noise sample is completely uncorrelated with all samples other than itself.

Nanosecond-type pulses result in baseband pulses extending from near dc to a few gigahertz, creating a spectral density of a few microwatts per megahertz. Of course, antennas are not dc devices, so it's immediately clear that UWB pulses invariably undergo some distortion to be reckoned with by the receiver. Both low-frequency distortion and high-frequency propagation effects impact the fidelity of the pulse received.

Another important item to recognize is the interference issues from the regulation standpoint. UWB signals encompass licensed frequency bands, resulting in spectral interference to signals occupying the bands. By keeping the pulse durations very short and correspondingly implementing appropriate amplitudes, UWB essentially creates a multiple-access system beneath the required noise floor of these regulated frequency bands. Much compatibility testing will continue to go on to ensure that signals in these bands can coexist with UWB signals.

Multiple access
While we have discussed standalone UWB signals, a last key item to recognize is that UWB enables multiple access through use of the time domain, in a way similar to time-division multiple access. Within a TDM frame, each user in a TDM system has an allotted time slot within which they can send their data. Ultimately, the TDM stream is demultiplexed into its constituent parts for delivery of a particular sender's part of the frame to the destination. UWB instead uses time hopping for multiple access. The concept of a frame within which users can each signal still exists, as does the idea that each user gets one "time slot" in the frame. But there are two significant differences with UWB signaling in a multiple-access environment.

First, not the entire frame is filled with pulses. The same can be said of TDM, which involves implementing some information-bearing overhead. But in the case of UWB, empty slots of time that could be carrying payload are left that way for time-hopping and overall robustness purposes.

Second, the assigned time slot for each user will vary from frame to frame. In one frame, user No. 1 may be in the 10th pulse slot, while in the next frame he may be in the 30th. Typical numbers for the number of pulse slots available in a frame slot are 100 to 1,000. So, doing some math even I can handle without a calculator, one frame of 100 pulses, each of 100 picoseconds, means frame times (zero guard intervals) of 100 MHz. If each user's pulse delivered one symbol per pulse-not necessarily the case or even desired, since pulse propagation distortions may drive spreading the payload over more than a single pulse-then 100-Msample/second rates (100 Mbits/s or greater) could be achievable. While things are not this rosy overall in general at this point in UWB development, you can see the potential that has created excitement over UWB given switching speeds now in reach.

Another simple way to look at the data rate capability is to recall that Shannon capacity, under additive white Gaussian noise (AWGN)-limited assumptions, is given by C = BW Log2 (1 + SNR). Clearly, narrowband systems are limited in a direct fashion by the channel bandwidth (BW), and rely on increases in signal-to-noise ratio to enhance capacity in an inefficient mathematical manner.

By contrast, UWB systems have bandwidth that's orders of magnitude higher. Since the capacity is related linearly to bandwidth, the throughput potential is much greater. Developers have actually begun to refer to the capabilities of UWB by advertising its spatial capacity (Leeper), meaning its throughput as a function of coverage area. This has come about because of the clearly targeted applications for in-home networking as a competitive offering to 802.11 and Bluetooth technologies. Both of those technologies use traditional signaling methods and currently have a specified rate of up to 54 Mbits/s—though proprietary implementations ratchet that up to over 108 Mbits/s. However, with the simultaneous ability to provide high data rates and multiple access, UWB offers about 10 times more spatial capacity—bits per second per square meter—than projected 802.11a systems, and 30x more than Bluetooth (Leeper).

Meanwhile, UWB also has significant implications in other applications, including radar and tracking/location systems, such as automotive and geological systems. In fact, before it was called UWB, this technology was a well-studied art in radar spread-spectrum signal design because of its ability to provide both location and speed information to very fine increments of accuracy.

Modulation of UWB pulses is typically simple in format. The complexity is brought about by the extreme nature of the rise and fall times of the pulses. However, it is precisely some of the advances made in high-speed switching and integrated semiconductor technology that have made UWB more practical. And, recall, there is no carrier signal necessary (some approaches used RF pulses rather than baseband to establish a legitimate "center" frequency). Basically, the hardware capability of today is finally now catching up with the theory that has been relegated to academia for so long.

Example modulation schemes include pulse amplitude modulation (PAM), on-off keying or pulse-position modulation. PAM, for instance, can be binary (signaling a one or a zero for every UWB symbol) or M-ary, meaning M different amplitudes, and thus with more bandwidth efficiency from the symbols. If M = 4, then each symbol represents two bits; if M = 8, then three bits are represented and so on. On-off keying is a form of binary PAM, where one of the bits—typically zero—is assigned the signal level of zero, or "off." As it turns out, the added bandwidth efficiency of M-PAM is somewhat wasteful, because the very narrow pulse shapes ultimately are the determinant of the spectrum regardless of how efficiently symbols are encoded. Throughput is gained more efficiently by changing the pulse repetition frequency.

Increasingly popular with UWB systems is pulse-position modulation. PPM systems represent the information by placing the pulse within a particular slot of a time slot window that has been sliced into some number of increments. If a time window of 500 ps exists for a particular transmitter, diced into four 100-ps intervals (and assuming 100 ps is reserved for margin), then depending on where in the 500 ps the pulse was detected, two bits of the signal would be represented.

Why UWB?
Because of our traditional mind-set in the analog and RF world, this all sounds a little odd. Why do UWB? Much of the impetus behind this technology is short-range wireless connectivity, such as in the home or office. This includes the desire for broadband access from increasingly mobile, or portable, devices. Another obvious reason for UWB's draw is the scarcity of spectrum left to access, and the regulatory obstacles to its use. UWB combats this by "flying below the radar" with regard to its spectrum. The power spectral density of the UWB signal may be below the natural noise floor associated with channel noise and spreading. Finally, there has been a rapid ramp-up of wired high-speed data access through DSL modems and cable modems. This wired high-speed access, and its potential to see higher and higher pipe capacities, means that home networks must keep pace to avoid being the bottleneck. How unfortunate would it be to have high-speed Internet traffic circle a metropolitan transport ring over 10-Gbit Ethernet, subsequently get handed off to modem technologies that have overcome fantastic broadband barriers to deliver high-speed access to the residential subscriber—only to have a bottleneck occur at the last stage of delivery to the end device?

In addition to the properties that make it consistent with the growth in broadband access and other practical issues, there are some other technical observations we can make. First, because a UWB system is time-based—and very short-time-based—for a given Tx/Rx pair, multipath issues associated with typical narrowband wireless systems are significantly mitigated. The gating of the receiver around the short pulse duration ensures that multiple reflected signals arriving outside the very short-duration window do not degrade the signal or drive sophisticated signal-processing receivers, as in more traditional wireless systems. The result is a loss of the dreaded "fade margin" term in link budgets, allowing the low-power transmissions—essentially resulting in AWGN-limited systems.

Another interesting result of the broadband nature of UWB pulses is their penetrating properties. This is due to the lack of concentrated energy spectrum that can be dispersed or attenuated as a function of largely uncontrollable material properties, and because of the significant low-frequency content whose propagation properties are not as affected by physical barriers. Finally, in terms of hardware manufacturing, the removal of the carrier generation and recovery subsystems, of narrowband filter requirements and of sophisticated digital processing algorithms in receivers has significant implications on system designs. Carrier generation and recovery and filter design represent typically inconvenient hardware elements in a digital communication system design, often placing limits on size and constraining the ability to build an integrated solution, while the signal-processing needs drive development as well as IC cost.

It will be interesting to see how UWB plays out. Significantly, there are alternative markets for UWB systems, such that, should the communications market not work out, the technological development will nonetheless not have stagnated, offering UWB a solid chance of eventually leaving a market impression.

Related Article
"Multiband OFDM: Why it Wins For UWB"; http://www.CommsDesign.com/story/OEG20030624S0006

References

1. Leeper, David, "Ultrawideband: The Next Step in Short-Range Wireless," IEEE International Microwave Symposium, Philadelphia, June 2003.
2. Ray, Saikat, An Introduction to Ultra Wide Band (Impulse) Radio, October 2001.

Rob Howald (rob.howald@motorola.com) is the director of systems engineering in the transmission network systems group of Motorola's Broadband Communications Sector in Horsham, Pa. He has a BSEE and an MSEE from Villanova University and a PhD from Drexel University.