The recent buzz about ultra-wideband (UWB) has spurred reactions ranging from fear and uncertainty about interference problems, to excitement about the possibility of high-speed wireless connectivity for the desktop. For most sensor applications, the gigabit promises of UWB offer a questionable benefit for devices that typically generate small amounts of data. However, UWB offers other, less obvious advantages, such as precise ranging and operation in hostile RF environments. Although this technology looks promising, hurdles remain for quelling interference and developing a communications standard for UWB devices.
The Current State of UWB
Ultra-wideband involves modulation techniques to spread a signal over a bandwidth >500 MHz. This is in contrast to conventional modulation formats that typically have modulated bandwidths of no more than a few megahertz. Since spectrum congestion is a problem, where could such a wide signal be used without interfering with existing systems? In 2002 the FCC allocated limited use of spectrum between 3.1 GHz and 10.6 GHz for UWB applications. To address interference issues UWB devices are limited to very low peak transmit power levels. Between 3.1 GHz and 10.6 GHz, UWB device limits are the same as those for devices classified as unintentional radiators, a term used to describe devices that weren't designed to be transmitting radios but may yet radiate EMI.
Concern still exists: Will the proliferation of UWB devices interfere with existing communication systems such as GPS, PCS, and Nexrad, as well as sensitive aviation guidance systems? The ability of UWB to coexist with these systems, GPS in particular, has been the topic of several recent studies (www.darpa.mil/ato/programs/netex.htm). Meanwhile, competing standards proposals and manufacturers using different approaches to UWB signal generation are fighting it out in studies of interference characteristics and how different UWB methods perform. It's also uncertain what the FCC's final ruling will be with regard to UWB.
Current UWB radios typically use one of two methods to generate the signal: pulse-based and modulated carrier. In pulse-based radios, the transmitter generates small-duration, high-energy pulses which are then modulated with data by varying either the amplitude or the positioning of the pulses. Modulated carrier-based systems use conventional modulation techniques and radio architectures, albeit with much larger modulated bandwidth. The two proposed techniques in this category are direct-sequence spread-spectrum (DS-UWB) and multiband orthogonal frequency division multiplexing (MB-OFDM).
Pulse-based UWB radios could be considered the classical approach to this technology, which was once termed "impulse" or "carrier-free" radio. A primary advantage to pulse-based systems is the simplicity of transmitter design; it's easy to design pulse transmitters to consume very little power and to use very little circuitry. A challenge of the design is to efficiently use the spectrum while obeying FCC limits. The spectral content of pulse waveforms is highly dependent on the shape of the pulse generated. In a simple pulse generation circuit, the output power is concentrated in the center of the desired band, rolling slowly off at the edges, as shown in Figure 1. In order to fit the signal within the FCC spectral mask (shown in black) the edges of the band contain little power, leaving much of the available spectrum unused. Using more advanced pulse shaping can overcome this, but then the advantage of a simple transmitter is lost.
Figure 1. This figure shows the spectral mask of a simple pulse-based UWB transmitter (shown in red) compared to the spectral mask allowed by the FCC (shown in black). Most of the energy is concentrated in the center of the band, and this means the available spectrum at the edges of the band is not efficiently used.
The receiver complexity is also increased, particularly when the radio needs to operate in a high-interference or high-multipath environment. Since the pulses are often very narrow (on the order of 100 ps) and run at very low duty cycles, the synchronization of two radios can be very difficult. To eliminate background noise, the signal must be sampled in sub-nanosecond windows of time and integrated. The high-accuracy time bases, high-speed correlation circuits, and accurate integration circuits required to perform these operations typically leads to increased circuit complexity and power consumption when compared to a more traditional radio architecture.
Modulated carrier UWB systems rely on existing modulation techniques such as DS-UWB and MB-OFDM for signal generation. These architectures resemble those of typical radio systems since the majority of signal modulation and demodulation takes place in high-speed digital circuits. DS-UWB resembles a pulse-based approach from a signal standpoint but allows more efficient use of available spectrum. This approach to UWB is championed by the UWB Forum (www.uwbforum.com). The MB-OFDM approach (www.multibandofdm.org) divides the available spectrum into 500 MHz channels. The radio constantly hops through the channels as it modulates data. The smaller channels allow tighter spectral control at the band edges while also providing some level of redundancy should one channel contain interference (see Figure 2).
Figure 2. This plot, captured from a spectrum analyzer, shows the MB-OFDM signal. The sharp dropoff at the edge of the signal allows it to be placed close to the edges of the band allotted by the FCC.
The current IEEE 802.15 working group has two task groups evaluating UWB as a physical layer. UWB provides an attractive physical layer for the 802.15.3a task group because of the high data rates that are achievable. In terms of sensor applications, UWB is a candidate physical layer for the 802.15.4a task group because of the ranging performance attainable with UWB frequencies.
What UWB Offers
Regardless of radio architectures or modulation methods, UWB has some other inherent advantages. Since the energy of the signal is spread over such a wide bandwidth, there is an inherent low probability of interception and a low probability of detection, particularly if snooping is performed with a narrow-band receiver. The power limits imposed by the FCC dictate a signal that is difficult to detect beyond close proximity to a transmitting device. In order to achieve reasonable range, these low power levels have forced UWB designers to implement radios with coding gain to permit reception of signals beneath the noise floor. Without knowledge of the coding techniques or spreading sequences involved, signals with a peak power lower than background noise power are very difficult to detect and more difficult to intercept.
The wide-bandwidth UWB signal creates a great deal of spectral diversity. In environments where signals may be obstructed by some objects and reflected off others, operation over a very wide bandwidth increases the likelihood of partial signal reception. In imperfect transmission path situations, parts of the UWB signal will be impaired while it is possible that other portions of the signal may be enhanced. Provided the modulation/demodulation techniques are tolerant to the slightly imperfect signal (and in some cases, capable of taking advantages of the enhanced portions), the integrity of the UWB link can be maintained (see Figure 3.
Figure 3. This time plot shows both an expected and an actual signal received in an office environment. Reflections off items such as a filing cabinet can result in strong, late-arriving pulses that can interfere with the anticipated waveform.
The high frequencies and large bandwidth of the signal allow precise range estimation between two communicating UWB radios. Both MB-OFDM and DS-UWB groups can achieve targeted ranging with <10 cm accuracy. For sensor applications in particular, this ranging information, when combined with triangulation techniques and range data from multiple devices, opens the door to accurate positioning.
With all the concern about UWB's interference effects on other communication systems, the fact that other devices have a more profound effect on the UWB radio performance is often overlooked. Since UWB spreads its energy thinly across a wide swath of spectrum, only a small percentage of the total radiated power will fall into the band of a conventional narrowband radio. Since the UWB receiver must have such a large bandwidth, incumbent technologies occupying the same spectrum effectively place all their available energy within the UWB receiver bandwidth, creating a great deal of co-channel interference. This interference is a formidable problem and a primary motivation for using multiband frequency selection techniques.
En route to creating a standard for high-data-rate UWB communications, these issues—interference with and from existing systems—must be resolved. The solutions to these problems will yield advances in signal processing and potentially open the door to more applications using UWB. For sensor applications the future may well hold low-power, robust data links with added functionality for ranging/location and possibly built-in radar and imaging capabilities. Independent of standards battles and architecture tradeoffs, the high level of industry interest and investment in UWB would certainly lead one to believe that it is here to stay. ?