Inside UWB design: A tutorial

With features combining flexibility, robustness, and high precision ranging capabilities, UWB is emerging as a particularly interesting wireless transmission method for applications requiring high data rate over a relatively short range. Multimedia traffic and cable replacement (wireless USB) are examples of the different types of application considered.

In April 2002, the Federal Communications Commission (FCC) approved and defined the emission of UWB signals in the United States. The FCC definition of UWB is quite simple. UWB signals must occupy a bandwidth greater than 500MHz or a bandwidth at least 20% of the carrier frequency.

Associated with this definition, the FCC opened up a new spectrum for UWB transmissions, with one of the bands from 3.1GHz to 10.6GHz having a maximum power emission limit of -41.3dBm/MHz. Outside of the United States, the European Telecommunication Standards Institute (ETSI) is working on its own standardization through the TG31A group.

The FCC regulation of UWB emissions raised the interest of major chip manufacturers, and triggered discussions around the advantages and disadvantages of the original Impulse Radio (IR) scheme versus a more traditional, carrier-based continuous transmission alternative.

This discussion is reflected in the current IEEE802.15.3a task group status supporting two different proposal for a physical layer: a Multi-Band (MB) frequency hopping approach with Orthogonal Frequency Division Multiplexing (OFDM) and a Direct-Sequence (DS) approach in line with IR.

Although outside of the scope of this paper, it is also interesting to note that the Medium Access Control (MAC) is another interesting area with plenty of protocols willing to support the UWB PHY solutions. In practice, an UWB PHY comes in three implementation standards.

In the first, Direct Sequence UWB, the content data stream is applied to a very short duration wavelet, either as bi-orthogonal keying or as pulse position modulation, resulting in a "carrier-less" spectrum. In the second, TD / FDMA, this string of pulses is stepped over a sequence of center frequencies to mitigate any multi-path that would appear on one channel. In the third, OFDM, channel bounding of two or more OFDM signals provide the necessary throughput rates.

UWB radio requirements
The targeted applications and the current UWB spectral mask highlights some initial considerations for radio designers; link budgets, multipath and propagation models, antenna design, silicon technology selection to comply with the targeted consumer market (Bluetooth type costs), power consumption to fit with wearable devices and low power transmit power, etc. Interferer management, Rx linearity and RF filter complexity are three major radio issues to be considered when designing a UWB radio. The capacity of the UWB system occupying a bandwidth B is a function of the signal-to-noise ratio (SNR), which is dependent on the distance between the transmitter and the receiver. C=B.log2(1+SNR). Through the bandwidth term, this equation shows how UWB can enable higher data rates than narrowband standards.

Direct sequence amd TD/FDMA UWB
The first two approaches for UWB appear, at first glance, to be remarkably simple. They do not require any analog up-conversion or down-conversion, and avoids all interference rejection filtering.

In principal, all that is needed is to apply the data stream to a 3 ns long wavelet, either as bi-orthogonal keying or as pulse position modulation, resulting in a "carrier-less" signal occupying the spectrum from 3.1 GHz to 5 GHz, or from 6 GHz to 10.6 GHz. Coding can be applied in which all combinations of a string of consecutive bits are assigned to a member of a family of orthogonal codes.

When a certain string of bits is to be transmitted, its code is selected to bi-orthogonal modulate the wavelet. This is called M-BOG modulation, or Multilevel Bi-Orthogonal Keying. With this approach, all that is needed for the radio portion of the transceiver is broadband amplification, broadband antenna radiation (for the transmitter and receiver), broadband low noise amplification, and decoding to recover the original data stream.

While this may seem straight forward, there are difficulties in realizing the RF circuitry, such as wide-band non-dispersive amplifiers and antennas. Further, there are major challenges for the DSP engines used to recover the data over a highly corrupted channel.

The following discuss some of the radio and baseband challenges that must be faced in the design of a direct sequence spread spectrum ultra-wide band system. Basically, the design complexity has been shifted from the radio to the baseband DSP.

Antenna
In a narrowband radio, the antenna is relatively small and highly efficient, with an easily achieved high Q. For an UWB radio, a highly efficient antenna is required with a flat frequency response over a wideband, forcing the LNA and antenna to be co-designed.

For IR-UWB, the antenna plays a significant role in shaping the transmitted pulse. The antenna design options include: (1) increasing the antenna loss, (2) operating a monopole below resonance (resulting in poor efficiency), (3) using a tapered slot microstrip, and (4) employing a multi-layer ceramic antenna.

RF Front End
In a narrowband radio, the RF front end requires low power consumption, usually constrained by a tough transmit mask and linearity specification (specifically for OFDM). The narrowband radio front-end blocks use reactively tuned circuits to deliver selective gain.

With UWB radios, the design of a wideband LNA is a "must have", but high power consumption is often associated with such requirements. RF linearity and filtering requirements must be considered together given the limited filtering opportunity available between the antenna and the ADC. On top of this, stability could be an issue considering the large amount of gain required in the receiver.

In general UWB radios use passive or active resistive circuits to provide frequency independent gain.