Hype and hysteria have characterized most discussions about ultrawideband (UWB) technology until recently. But for now, much of the breathless rhetoric around UWB seems to have dissipated in the wake of an FCC ruling earlier this year that has wide-ranging implications for high-speed data communications, radar and imaging. The ruling, handed down in February, defines a spectral "mask" for UWB that the FCC believes will appease opponents of the technology long enough to allow the development and realization of UWB devices that can be used to examine its potential and effects in real-world scenarios.
For UWB developers, the ruling has set the clock ticking in the race to be first to market a practical, commercially deployable device that meets at least some of the expectations. But it's not so easy. Despite the scaled-back level of those expectations, and despite many years gestation at Darpa, UWB isn't proving to be as straightforward or intuitive as many had hoped.
The technical difficulties are legion. Many designers are confounded by the mind-boggling difficulties of actually detecting and acquiring a high-data-rate, low-level signal that resides below the noise floor in the presence of multiple users, multipath interference and interference from incumbent wireless devices. And, by the need to do it fast (sub 1 second), at low cost, low power and with a small footprint.
This has generated reams of research papers on everything from signal processing to antenna design, including modulation and coding schemes, modeling and simulation and coexistence.
And the problems aren't all about technology. Facing myriad wireless incumbents, UWB has an uphill battle ahead of it if it is to find a home as a communications alternative. However, if it leverages its inherent advantages, including location resolution and see-through wall capabilities, it may well succeed over time.
So what is UWB?
UWB isn't exactly new and some may argue that the technology goes all the way back to Marconi and his early impulse transmissions using spark gaps. Realistically, however, the roots of UWB research reach back to the early 1960s when tools and techniques became available to study electromagnetic pulses in some depth.
Alternatively referred to as impulse, carrier-free, baseband, nonsinusoidal and time-domain signaling, the term UWB wasn't actually coined until 1989 by the U.S. Dept. of Defense (DoD). The DoD developed UWB for radar, location and communications applications. UWB's ability to operate below the noise floor which makes for a low probability of detection at low data rates was particularly attractive for clandestine communications.
UWB technology is based on the generation of extremely short digital pulses in the subnanosecond range (1 to 1,000 picoseconds). To transmit information, such pulse trains can be modulated any number of ways, including time, phase, amplitude and voltage. However, it is the fact that the pulses can be modulated directly by the baseband signal instead of using a high-frequency carrier that gives UWB radios their much-hyped simplicity of design. No longer are expensive, complex, large-footprint analog circuits required for carrier-signal generation on the transmit side and for carrier stripping on the receive side.
Because it is so short, the pulse's associated coherent frequency spectrum can be multiples of gigahertz wide, thereby dispersing the pulse's energy across many narrowband systems (such as GPS, cellular, PCS, satellite radio and the various wireless-network bands). The bandwidth and the center frequency of the pulse are determined naturally by its length, but typically some kind of filtering is used to limit the bandwidth to keep it within the FCC mask as defined by the ruling.
This wide relative bandwidth allows UWB to penetrate walls and other obstacles, making possible capabilities such as through-wall imaging, while at the same time endowing it with a large degree of immunity to multipath interference relative to narrowband systems, a characteristic particularly appealing to the communications industry.
Also, because the amount of information any signal can carry is a trade-off between bandwidth, power and distance (with various modulation and coding schemes used to optimize the communications channel), the wide bandwidth of UWB signals has the allure of potentially very high data rates of up to 500 Mbits/s. It's important to note that the information-carrying capacity scales linearly with bandwidth, and logarithmically with power, making it much more attractive to designers to scale the bandwidth to achieve higher rates.
The FCC ruling in February was designed to curtail the interference this wide-bandwidth signal might have on GPS, military, ground/air navigation or cellular/PCS applications. The ruling specifically defines UWB as a signal with a bandwidth of 500 MHz or bigger, or 20 percent fractional at the -10 dBm point. The ruling limits UWB power to Part 15 limits (-41.25 dBm) operation over the 3.1- to 10.6-GHz band, though other bands are allowable below 900 MHz and roll-offs around the limits vary according to the application and whether it's for indoor or outdoor use.
For many, the ruling by the FCC, though considered conservative, was a landmark event in wireless signaling that kept the cellular industry and military at bay until UWB could be properly evaluated. The ruling itself could be re-evaluated, probably within a year to 18 months from last February.
"What they really needed was a proposal to let the industry go forward, while meeting the demands of the other incumbents," said Jim Meyer, vice president of marketing and business development at Time Domain Inc. (Huntsville, Ala.), a developer of UWB technology since its founder Larry Fullerton filed his first UWB patent in 1987. "While we're comfortable with the mask, as is, for wireless personal-area networking (WPAN), we're hoping the FCC will give us more spectrum and power [in the next review] though we don't necessarily need it," added Meyer.
For Carl Panasik, a manager in Texas Instruments' wireless advanced architectures division, the ruling was as significant in its implications for the FCC itself as it was for UWB. "This is one of the first times that a technology has been enabled by regulatory means," he said. "This is a big shift for the FCC as it has always been focused on spectrum and allocating spectrum for services. They used frequency and power to do their regulation for them."
But many are not so enthusiastic. "My feeling is that they were pressured by Congress to get something out sooner rather than later," said Robert Fontana, president of Multispectral Solutions Inc., and an 18-year veteran of the UWB industry. "Instead of going to a further Notice of Proposed Rule Making, which they should've done, they went right to an R&O (Report and Order) that was thrown out rapidly and now it's coming back to haunt them. I'm betting that as silicon appears, the rules will have to be changed," he said.
Fontana's critique is based on his belief that the FCC misunderstood the behavior of UWB signals radiating from shock-excited UWB antennas, a misunderstanding that could lead to UWB interfering with satellite digital audio radio services (SDARS) at 2.2 GHz or PCS at 1.85/1.95 GHz.
Fontana believes that the FCC assumed that a UWB's power spectrum would be evenly distributed across the 10-dB down bandwidth points, when in fact it will more likely have peaks at the 3-dB down points that will contain up to 90 percent of the pulse's power. That assumption by the FCC, said Fontana, is based on a fundamental lack of knowledge on the radiation pattern of a shock-excited antenna.
Those power peaks will naturally increase with the pulse-repetition frequency as the data rates go up, causing a comb-like continuum of spectral lines. "I think SDARS will have problems and PCS even more so," said Fontana. He places special emphasis on PCS as the whole network can be compromised if a single user gets affected by UWB. "It's been overlooked by the FCC that UWB can cause the link budget to change for a given user, and if the basestation has to compensate for that one user it affects the whole system," he said.
Fontana recommends that the rules be changed such that everything below 3.1 GHz would be 35 dB down from Part 15 (equivalent to a little over -76 dBm). "This extra margin will protect SDARS and PCS transmissions, and those are the ones that have lost the most from these proceedings."
Others have found the ruling very satisfactory, including XtremeSpectrum Inc., one of the more visible advocates for the adoption of UWB in the commercial communications space. "Almost 1,000 filings were sent it, so someone was asleep at the wheel if something was left out," said Veronica (Roni) Haggart, the company's vice president of strategic relations
Good or bad, the FCC guidelines are forming the template for global regulatory rulings, with Canada, Europe, Japan, Korea and Singapore all interested in adhering closely to the ruling. "They [the European Commission] want to follow the FCC rules though not perfect and that seems to be right," said Gerard Fargere, Advanced System Technology Access Network Director at STMicroelectronics.
Each of Europe's many countries has its own regulatory body, with its own airwaves to protect. "This technology needs a large bandwidth, they have to accept that," said Fargere. "I can't see how one country can block it, as they'll be blocking the whole technology."
Indeed, close cooperation between global regulatory bodies on UWB is being seen as mandatory to avoid a recurrence of the wireless LAN debacle of the mid-to-late '90s where the United States, Europe and Japan all introduced incompatible, noninteroperable standards.
The UWB effort is being overseen in Europe by the Conference European of Post and Telecommunications (CEPT), with support from the European Regulation Organization (ERO), with the main impetus coming from the United Kingdom, France and Germany.
"The current timeline is to finish compatibility studies this year and roll it into an EU-harmonized standard by the middle of next year," said Mahesh Balakrishnan, vice president of technology and business development at Philips Semiconductors. "Very possibly they'll work with people in the IEEE 802.15.3a [U.S. WPAN standards body] as there's a history of HiperLAN2 and 802.11 to fall back on," he said. Ben Manny, director of residential communications at Intel's Architecture Laboratory is actively engaged with European Telecommunications and Standards Institute's [ETSI] Task Group 31 to facilitate that. "We don't want separate standards," said Manny.
The European and Japanese regulatory rulings are both expected to be slightly more restrictive than the FCC in terms of rolloff at the lower frequencies to protect incumbents such as satellite operators.
UWB floodgates opened
With the FCC ruling now in place, the floodgates have been opened for a host of UWB solutions that are expected to be announced this year and throughout 2003, with much of the activity coming from the communications side for now. With target data rates of up to 500 Mbits/s, UWB is being targeted at WPANs, audio/video distribution within the home, as well as a cable replacement option for USB and FireWire.
According to Philips' Balakrishnan, the plethora of UWB proponents that have emerged from the woodwork over the last four or five years is a result of a number of factors. Chief among those were the rise in data rates beyond voice and the decision by the DoD in the mid-1990s to opt for dual-use technologies. "Also, the improvements in semiconductor processes allowed their implementation, versus discrete implementations with its limited market." The emergence of WLANs in the late '90s made any form of wireless connectivity important, he said.
Whatever the reasoning, those that have jumped on the UWB bandwagon are finding that UWB is a bit more problematic than expected. These proponents now face, head on, the difficulties of quickly and accurately detecting a signal below the noise floor at high data rates and in the presence of multiple users and multipath interference, along with interference from the wireless incumbents. The problems have led to some innovative antenna designs, wide-dynamic-range circuitry and low-power implementations as well as advanced detection, modulation and coding schemes. Much of the work to date remains tightly under wraps as companies protect their pending patents.
"The advantage of UWB is that the transmitter is quite simple but the receiver, on the other hand, is quite complex and power hungry," said Eric Janson, vice president, Cambridge Silicon Radio (CSR) North America, a Bluetooth proponent. "It has to pull signals in the presence of powerful interferers that will inevitably be present. Good dynamic range is needed and that's costly." Janson limits UWB's applications to asymmetric services, such as in-store security tags, mice, keyboards and joysticks. "But not for LANs or PANs as it'll suffer from the classic near-far problem," he added, though he does believe it'll work well over short distances of up to 2 m. "Claims of UWB for cellular seem farcical to me."
"This is an important technology, but it doesn't change the laws of physics," said Andy Rappaport, a partner with August Capital, a Menlo Park, Calif., venture capital firm. "The biggest misconception is that it's more energy efficient than other radio technologies. In most cases, that's just not true. There are some cases that take advantage of the features of UWB, some that don't, but all are governed by Shannon's Law anyway," Rappaport said.
Rappaport does agree with UWB's penetration and location advantages, but takes issue with its purported data-transmission advantages, with much of his skepticism based on the susceptibility of a wideband signal to interference from other wireless transmissions including other UWB signals. "The second misconception is that UWB radios will be easier to design. This isn't necessarily true," he said. He believes it would be more accurate to say that "radios that take advantage of the unique attributes of UWB will be easier to design. RF is always a tradeoff, and as you increase the distances dynamic range becomes more of a factor and you can run into problems." Limiting UWB to very short distances is the key, said Rappaport, as it isolates the wideband transmission to a very localized spot. "At distances of 1 foot I've seen Gigabit transmissions I can't think of a more energy-efficient way to do this."
Short-range applications of UWB will be one focal point at the IEEE 802.15.3 Working Group meeting in Monterey, Calif., this month. There, the covers will come off much of the research to date. Communications-focused UWB players will put their cards on the table in an effort to have their technology selected as the foundation for the IEEE 802.15.3a's physical-layer for WPANs. The "a" stands for alternate, as a QAM-based PHY has already been defined for data rates up to 55 Mbits/s.
The baseline 15.3a requirements call for rates of 110 Mbits/s at 10 m with up to four channels and a power consumption of no more than 100 mW nominal (for the PHY). The rates will be scalable up to 200 Mbits/s at 5 m, with rates increasing down past 4 m.
Leading the charge to UWB for WPANs is XtremeSpectrum with its XSI100 four-chip Trinity offering for rates up to 100 Mbits/s. Announced in late June, the XSI100 is based on the company's biphase-modulated UWB technology with an integrated, direct-driven antenna.
In direct rebuttal of the radiation-control problems associated with a direct-driven antenna, as described by MSSI's Fontana, Martin Rofheart, CEO and co-founder of Xtreme, commented, "We generate the transmit signal electronically so that it meets the FCC specification before it reaches the antenna." In other words, according to Rofheart, "the antenna is not the primary mechanism for shaping the spectrum, though it is involved. The transmit signal is then coupled into an antenna whose frequency response is matched to it."
But the XSI100 won't be alone for long. According to Jim Baker, executive vice president of commercial products at Time Domain, that company is on the cusp of announcing its own FCC-compliant chip set possibly by the time this article appears. STMicro's Gerard Fargere expects to have silicon for WPANs and home content distribution in 2003, with product in 2004. Texas Instruments, Philips and Intel are all at varying stages of readiness, ranging from core research to product definition.
"Our interest is in HDTV at 12 to 17 Mbits/s," said TI's Panasik. "We're always looking for ways to connect our DLP [Digital Light Processor] technology with a high-quality connector at low power." Philips' Balakrishnan also looks to UWB for HDTV distribution within the home, quoting rates of 20 Mbits/s and above.
Panasik has little confidence in current wireless home networking technologies (namely 802.11a) as a means to get to the data rates required for the low bit-error rate needed for HDTV on its DLP. That's why the company was one of the initial investors in XtremeSpectrum (XSI).
"XSI has shown operation beyond 15 m at 50 Mbits/s," said Panasik. "They demo'd four DVD screens across a large room and our DLP people experts in video said it looked like wire. And they know what to look for."
But Panasik doesn't discount 802.11 altogether. "We've done the calculations of UWB data rates as a function of distance compared to 802.11a. We found there's a crossover point at 10 m where 802.11a dominates in data rate versus distance."
Time Domain's UWB solutions to date have mostly been based on pulse-position modulation (PPM), whereby the pulses are "dithered" in time according to the baseband information. But this scheme runs into serious problems as data rates go up past 40 Mbits/s, where the timing required to accurately predict and acquire the pulses becomes too demanding.
To go after the 500-Mbit/s market the company has derived modulation and coding schemes, as well as "innovative capabilities. . .though we can't discuss them now as we're looking to get them accepted into 802.15.3a and it's very competitive," said Baker. However, Baker does admit they're moving from PPM to a form of biphase modulation, with the eventual goal of an all-CMOS, single-chip solution, though it will start with a SiGe front end, followed by a separate digital baseband and a MAC.
Intel, which set the 500-Mbit/s bar for UWB, sees the technology as an enabler for fully integrated, CMOS digital radios that will eventually allow wireless USB 2.0 transmissions at the full 480 Mbits/s. After recruiting Sumit Roy of the University of Washington to lead its research, the company focused its UWB efforts entirely on communications.
Intel's priority is scalability and integration (for ubiquitous connectivity) and so is pursuing a sub-band approach whereby bands are added to increase the total bandwidth of the radio. "But this is proving costly, as it's almost like adding separate radios," said Manny, though he does see advantages in terms of receiver design and modulation schemes. The company demonstrated its capabilities at a spring Intel Developer's Forum, where it showed 100 Mbits/s at 250 microV (0.5 W) over 3 m.
Intel has based much of its work on bipolar pulses that Manny said gets rid of the spectral lines associated with PPM. It modulated the pulses using differential phase-shift keying (DPSK) for the IDF demonstration. "However, DPSK requires a broadband analog delay loop and this is hard to develop in a CMOS process," said Manny. For the demo, the designers wrapped a 10-foot length of coax around a cylinder. "It's so tough that we're looking at other modulation schemes besides DPSK," he added. Manny said Intel will persist in its pursuit of UWB, "as it's a fundamentally lower-power RF technology than carrier-based radio, which is what we need for portable devices."
Uphill battle for UWB
In an ironic twist, the wide bandwidth of UWB, which brought the GPS, military and cellular communities to arms, may also prove to be its most onerous difficulty now that it has been tentatively accepted. According to David Furuno, director of advanced wireless at General Atomics, "it is almost certain that UWB will be affected by co-located 5-GHz narrowband transmissions (i.e., 802.11a-based WLANs), which are up to 20 dB higher in power."
The degree of that interference has yet to be determined, though interference from other UWB networks may prove to be an even bigger issue, depending on their relative pulse repetition frequency.
General Atomics itself is prepping its own single-chip sub-$10 offering that's based on a proprietary modulation technique called Spectral Keying "that's neither PPM nor biphase but meets the FCC mask with good control over the radiated spectrum," said Furuno.
While Craig Mathias, principal with the Farpoint Group, sees the antenna design as "a gating factor in the long run," in terms of how well the overall radio works, General Atomics sees public-domain antennas as working quite well, though it is working on its own proprietary design. "However, it's not clear that's necessary yet," said Furuno. Target data rates for General Atomics are in the 12- to 40-Mbit/s range, "as we feel there's more pull there," hr said.
While myriad questions remain about the technical viability of UWB for communications, that may not be the true battle that UWB faces at all. "I think it's one of the most significant technologies with the greatest potential and, if we were to start with a clean radio sheet, this is the technology I'd choose," said Mathias. "But I know some pretty smart people who are convinced it's not going to go anywhere commercially." Mathias is convinced that there are too many other radios out there that work better, such as those based on 802.11 standards, and with the restrictions the FCC will always put on it, it'll never reach its full potential. "It's a technology that came along a hundred years too late it's a case of what it can do versus what we're letting it do."
Mathias is much more enthused about UWB's location and radar capabilities, which are where companies like Aether Wire come into their own. "We're taking advantage of the high-resolution properties of UWB, and have pretty good accuracy with our 'localizers' as we call them," said Vincent Coli, an Aether Wire spokesperson. The company is operating at around 970 MHz and is below Part 15 limits due to its very low duty cycle pulsing. The system uses a 1-ns doublet pulse whose negative/positive polarity is modulated by the baseband signal. Target applications include people/asset tracking, with military personnel and firefighters potentially being its first applications.
When it comes to location/tracking, the faster acquisition and lower power of UWB localizers have been seen as advantages relative to GPS. But this may be somewhat misleading, according to Rappaport, who believes they serve two very different location needs. "UWB can do precise local locationing, close to the source [potentially to the centimeter], while GPS does very accurate absolute positioning globally though its resolution is limited to 10 m," he said. Given these operating parameters, he sees them more as complementary than competing. "I can see a device with a GPS receiver and a UWB transmitter. The GPS device would provide rough locationing, but would be the absolute reference for the UWB device," Rappaport said.
Other wireless incumbents such as Proxim and Intersil, both prominent for their WLAN efforts, have adopted a wait- and-see" approach to UWB overall. According to Jim Zyren, market manager at Intersil, "Our position is that if the technology is promising for the marketplace we serve, then we'll get involved. If not, we'll look elsewhere."
In the final analysis, Rappaport summed up the debate with an old adage: "New technologies solve old problems badly," he quipped. "The dumbest thing you can do is chase the problem that's already been solved." He sees UWB being successful only if it first targets applications that haven't already been mastered, such as localized locationing, see-through applications and even very-high data rates over short distances. "The cellular guys can't stop it and once its deployed it can take on more and more applications, much like we saw with wired Ethernet," he said. "Anyway," he added, the cellular guys have much more to worry about right now besides UWB."