Ultrawideband Wireless: Not-So-New Technology Comes Into Its Own

Wireless designers now have yet another option to add to their bag of tricks. Called ultrawideband (UWB), the basic technology has actually been around as long as wireless. Marconi's original spark transmissions and all early wireless telegraphy were UWB. Such wideband transmissions were banned in the 1920s but rediscovered in the 1960s as new methods of secure radar and radio.

Out of the early days—where UWB was called impulse, baseband, or carrierless radio—has emerged the most modern version. Thanks to semiconductor manufacturing advances and the Federal Communications Commission's (FCC) recent approval, practical UWB circuits and products are becoming available.

Most conventional radio transmissions use the data signal to modulate a sinewave carrier on some specific frequency by the FCC. The usual goal is to select the most spectrally efficient form of modulation to get the maximum data rate through the usually restricted assigned bandwidth channel (see "How UWB Works," below).

In UWB, the serial data is translated into very short pulses of a unique shape, then applied directly to an antenna. Fourier fans know that very short pulses, regardless of their specific shape, produce an extremely wide bandwidth signal. One definition of UWB is that the signal has a bandwidth of at least 25% of the center pulse frequency, or 1.5 GHz, whichever is larger. Figure 1 compares several conventional spectrum uses to UWB. Because of the wide bandwidth needed by UWB and the potential for interference with other services, the FCC has placed UWB in the 3.1- to 10.6-GHz range.

A UWB signal starts as a high-speed rectangular pulse train that is shaped into unique pulses called monocycles (Fig. 2). These aren't one cycle of a sinewave, but instead pulses derived with a Gaussian filter. The desired transmission rate determines the pulse frequency, and the duty cycle is always very low—several percent or less. The pulse width sets the center frequency of the signal bandwidth. This center frequency f_C is roughly the reciprocal of the pulse width 1/t_W. A 200-pS pulse will produce a center frequency of 1/200 * 10^­12 = 5 GHz. The general rule of thumb for a 3-dB bandwidth is 1.16 times the center frequency, or in this example, 5.8 GHz.

This technique spreads the signal so that it overlays any other signals in its bandwidth. But the key to the technique is its very low power level, which makes it appear as noise to most other narrowband or spread-spectrum equipment. The low power level also severely limits the range of the signal to the general vicinity of the transmitter.

To transmit data from one point to another, you must modulate the pulses. UWB has two common types of data modulation: pulse-position modulation (PPM) and binary phase-shift keying (BPSK) (Fig. 3). Biphase is the easiest to implement and gives the best spectral efficiency. On-off keying (OOK) and pulse amplitude modulation (PAM), which are simpler for many applications, can be used as well. Multilevel PAM offers the possibility of increasing data speed for a given bandwidth and range.

If the application calls for sharing the spectrum with many users, multiple access methods are used. For instance, the data to be transmitted can be combined with a pseudorandom code as in spread-spectrum communications to permit channelization of the bandwidth. The unique codes allow individual signals to be identified and recovered at the receiver. The key to successful UWB is coding optimized for the application. Then the encoded serial data is sent to the pulse-forming circuit where the monocycles are generated.

Today, CMOS ICs can easily create such pulses very inexpensively at low power. The pulse is usually initiated by a narrow rectangular pulse from a CMOS logic circuit that drives some type of pulse-forming network that shapes the rectangular pulses into the desirable Gaussian monocycles de-scribed earlier.

Because the power levels permitted are so low (50 to 200 µW), no power amplification is needed. As the pulse radiates, it's naturally differentiated on the way to the receiver. The received pulse looks something like Figure 4. At the receiver, a broadband, low-noise amplifier (LNA) increases the signal level, after which a multiplier and integrator autocorrelate to recover the data.

A major challenge in UWB systems is the antenna. A typical wireless antenna is a resonant l/2- or l/4-wavelength device with a relatively narrow bandwidth. By using thicker or physically expanded elements, the bandwidth can be widened. But in UWB, the antenna needs even greater bandwidth.

Wideband antennas have been developed for government and military use. Some are just beginning to reach the commercial market, like antenna maker SkyCross' meanderline antenna (MLA) technology. According to CEO Alan Haase, SkyCross has a new product designed for the forthcoming UWB applications in the 3.1- to 10.6-GHz range. Frank Caimi, the CTO, indicates that this new antenna maintains a VSWR of 2:1 or less and has a linear phase response across the entire band.

Regulatory Issues: UWB has been used in government and military applications for years. It hasn't generally been approved for commercial use. The primary exception is approval given to companies making ground-penetrating radar devices that provide "x-ray vision" to public service entities for detecting flaws in roads or bridges, or for search and rescue. The FCC offered special licenses to companies making these devices, which are still widely used.

With semiconductor technology making UWB practical, companies wishing to address potential opportunities, especially in data communications and networking, began to push the FCC in 1998 for a ruling that would permit UWB operation. On Feb. 14, 2002, the FCC announced its Report & Order permitting UWB operation from 3.1 to 10.6 GHz with a power level not to exceed the Part 15 rule (15.209) for unintentional radiators. This is 500 mV/m measured at 3 m over a 1-MHz bandwidth for frequencies over 960 MHz. That translates to an emitted power spectral density of ­41.25 dBm/MHz.

Operation isn't permitted below 3.1 GHz due to the potential for interference with GPS satellites in the 1.2- and 1.6-GHz ranges, PCS cell phones in the 1.9-GHz range, and other government systems operating in the L and S microwave bands. Operation below 900 MHz has potential (Fig. 5). Right now, the U.S. is the only country with approved UWB service, although Europe and Japan are currently working on regulations and standards.

The Killer App: The FCC categorizes UWB applications as radar, location, or data transmission. The first and still largest application of UWB is radar. The military investigated it for years before developing a very high-resolution radar it also hoped would be stealthy.

As noted, several companies make a UWB transceiver for ground-penetrating radar devices to find faults in roads, bridges, and other concrete and asphalt structures. Variations of these products have been developed for public service personnel, like firefighters, police officers, and rescue workers looking for people buried in rubble. Such devices can also see through walls, a feature of interest to the military and police. Radar imaging methods can extend to devices for locating pipes, electrical wiring, construction studs, and steel reinforcements. UWB has also been developed for medical imaging.

One very promising application is short-range collision avoidance for cars. DaimlerChrysler already has prototypes of such radars working in the 26- to 29-GHz range for automatic braking systems. Look for these as an option on high-end vehicles in the near future. There are many other radar-like applications, like intrusion alarms, where proximity detection is the goal.

Data communications is the most undeveloped area of UWB. The most likely killer app of this unique wireless method is in short-range personal area networks (PANs). UWB is being developed as a physical layer (PHY) option in the IEEE 802.15.3 PAN standards. It's expected to support data rates of up to 110 Mbits/s over a range of 10 m. Such PANs are ideal for making wireless interconnections to consumer entertainment equipment.

Because of its very high data-rate capability, UWB is the PHY of choice to carry video data. Other common wireless choices, like 802.11b and Bluetooth, are too slow for video. The 54-bit/s 802.11a local-area network (LAN) standard handles video, but it's costly. UWB does it faster, simpler, and less expensively, without compression.

Many consumer electronic systems can benefit from a UWB connection that eliminates wires. Connecting your cable TV output to multiple sets or set-top boxes is a common requirement. Connecting camcorders to TV sets or communicating between a DVD player and a large-screen TV set some distance away is another possibility. Look for wireless versions of USB 2.0 and IEEE 1394 interfaces using UWB soon.

Good News And Bad News: UWB offers some unique advantages to potential wireless applications.

Spectral efficiency: UWB overlays an existing spectrum yet theoretically doesn't interfere with other narrower-band services occupying that spectrum.

Very high data rates: Rates of 20 to 100 Mbits/s up to 10 m are easy. Rates to 500 Mbits/s are within range, and a 1-Gbit/s rate is potentially achievable. Of course, the rate is limited by distance.

Strong immunity to multipath effects: Given UWB's very short pulse duration, the incident and reflected pulses will most likely arrive at different times. So, they won't overlap in time and won't interfere with one another. This means UWB works well indoors, a target-rich environment that produces considerable multipath effects.

Superior penetration: The short pulses easily pass through almost anything, making UWB about the only choice for looking beyond walls and seeing through structures.

High resolution in radar and other imaging: Again, the short pulses permit distances to be resolved in centimeters. Precise location and dimensioning are possible.

Little or no interference: The very low-power and spread-spectrum nature of UWB causes its signals to be below the noise level over its operating spectrum. It appears as noise to most other applications.

Power efficiency: UWB applications operate at microwatt levels or milliwatts at the most. The very low duty cycle of the pulses makes power consumption ultra low.

Simple circuitry: This wasn't true until recently. Thanks to submicron silicon, UWB circuits can be made with conventional CMOS. Transmitters are very simple. Receivers are more complex but easily implemented in CMOS.

Low cost: Because standard CMOS can be used and circuits are simpler than most other wireless systems, a transceiver's cost is potentially very low.

Inherently secure: Because the signal operates within the noise, it's difficult to detect. That plus any additional encryption encoding makes UWB the most secure wireless in existence.

Yet in the real world, not all news can be good. There are some disadvantages as well.

Low power: While it provides some benefits, it also severely restricts range. If higher power were permitted, higher speeds over longer distances could open whole new applications, like wireless broadband to the home.

Potential interference: Though the FCC's rules restrict power and frequency of operation, UWB may affect certain other wireless applications.

Complexity: Despite the apparent simplicity, UWB is a very complex wireless technology. Testing is a bear. Future chip sets and reference designs, though, will take some of the complexity out of designing with UWB.

What To Expect: As Eric Janson of Cambridge Silicon Radio of Bluetooth fame says, UWB is nothing more than just another wireless PHY. While that's true, its unique capabilities make it a niche of its own. The growth potential is in PANs for wireless communications in consumer entertainment applications. Those won't really compete with other PAN technologies, like Bluetooth and ZigBee. Some products will take advantage of the first wave of UWB silicon for such applications early next year.

Longtime UWB company Time Domain already has a chip set for its applications partners. Known as PulsON200, this generic chip set can hit data rates to 40 Mbits/s. It can be adapted to communications, location, or radar. According to Jim Baker, executive vice president for commercial products, the next-generation PulsON300 chip set, with 100-Mbit/s capability using the forthcoming 802.15.3 standard, will be available later in 2003.

Chris Fisher, vice president of marketing of XtremeSpectrum, says the demand for true consumer multimedia connectivity continues to grow. The XtremeSpectrum Trinity chip set is designed for this need (Fig. 6). It is based on the proposed IEEE 802.15.3 PAN standard.

The Trinity set delivers a 100-Mbit/s data rate and is designed primarily for consumer applications. It's available in 100,000-unit quantities for $19.95. Full production is expected in 2003.

Intel and Philips also are developing semiconductor products in this sector. Sony foresees wireless connectivity at the heart of most future entertainment systems. Ben Manny, Intel's director of wireless technology development, says Intel hopes to create a 500-Mbit/s transceiver that can serve as a wireless USB 2.0 air interface for PCs and consumer products. A wireless IEEE 1394 option could also become available.

Recently, nine leading technology companies formed the WiMedia Alliance, an open industry forum to develop and promote wireless PANs and multimedia devices. It will establish a certification program similar to the successful Wi-Fi group, which certifies 802.11b/a WLAN products. WiMedia will help spur the growth of the 802.15.3 UWB PHY and speed the use of wireless in consumer electronics by ensuring interoperabilty of the various consumer products, like set-top boxes, HDTV, still and video cameras, MP3 players, and other multimedia devices.

The FCC has promised to take another look at UWB within another year to see the effects of real applications on interference if any. Possible improvements could result. It is early in the game for UWB. The best is yet to come.