New wireless networking standards commonly present design challenges in the physical layer, which in turn present circuit simulation and testing challenges. The WiMedia MB-OFDM UWB signal with its low power, hopped, 528 MHz wide OFDM format is a current example, which will have broad impact as the base technology for certified wireless USB and next-generation Bluetooth for personal computer, consumer electronics and mobile devices. This article reviews WiMedia’s MB-OFDM PHY layer structure and discusses various MB-OFDM testing topics, including transmitter analysis, receiver simulation and evaluation, and using simulations during circuit design to improve device immunity to interference. Finally, simulated and measured results from industry leading solutions are presented.

With the blessing of the WiMedia Alliance, the stage is set for widespread implementation of certified wireless USB and next-generation Bluetooth 3.0 products, using multi-band orthogonal division modulation (MB-OFDM)-based ultra-wideband (UWB) physical layer (PHY). Wireless USB supports data rates of 53.3, 80, 106.7, 160, 200, 320, 400 and 480 Mb/s. Support for transmitting and receiving data rates of 53.3, 106.7 and 200 Mb/s is mandatory (see Figure 1). The UWB spectrum is divided into 14 bands, each with a bandwidth of 528 MHz (see Figure 2). The first-generation products are focusing on the first three bands (3.1 to 4.8 GHz), also known as Band Group no. 1, for simplification in the design of radio and analog front-end circuitry. The WiMedia standard specifies a MB-OFDM scheme to transmit information. A total of 110 sub-carriers (100 data carriers and 10 guard carriers) are used per band to transmit the data. In addition, 12 pilot sub-carriers allow for coherent detection. Frequency-domain spreading, time-domain spreading and forward error correction (FEC) coding are used to vary the data rates. The coded data is then spread using a time-frequency code (TFC).

The standard specifies two types of time-frequency codes:

• Time-frequency Interleaving (TFI): where the coded information is interleaved over three bands.

• Fixed Frequency Interleaving (FFI): where the coded information is transmitted on a single band.

Support for both TFI and FFI is mandatory in the WiMedia standard.

Figure 3 shows one possible realization of a transmitted RF signal using three frequency bands, where the first symbol is transmitted at a center frequency fc = 3432 MHz, the second symbol at fc = 3960 MHz, the third symbol at fc = 4488 MHz, the fourth symbol at fc = 3432 MHz and so on. A zero-padded suffix (ZPS) is appended to the OFDM symbol to provide a mechanism to mitigate multi-path effects and allow sufficient time for the transmitter and receiver to switch between different center frequencies. For data rates of 200 Mb/s and lower, the binary data is mapped onto a QPSK constellation. For data rates of 320 Mb/s and higher, the binary data shall be mapped onto a multi-dimensional constellation using a dual-carrier modulation (DCM) technique.

MB-OFDM Benefits

UWB channels are highly dispersive and therefore pose significant multi-path challenges. MB-OFDM receivers are much more efficient at capturing multi-path energy (over 90 percent). In addition to capturing energy more efficiently, a MB-OFDM system also possesses several other desirable properties: high spectral efficiency, inherent resilience to narrowband RF interference and spectral flexibility. With an OFDM system, the transmitted spectrum can easily be shaped by nulling out tones or turning off bands in order to protect sensitive or critical bands. Several aspects of MB-OFDM specification are chosen to reduce implementation complexity:

• Constellation size limited to QPSK; required precision of digital logic (specifically IFFT and FFT), as well as reduced ADC and DAC.

• Large sub-carrier spacing: Phase noise requirements on PLL relaxed and improved system robustness to synchronization errors.

Transmitter

At the transmitter, the input bit stream is scrambled. A forward error correction (FEC) code (convolutional) is applied to provide resilience against transmission errors. The encoded sequence is then interleaved (TFI or FFI) and mapped to frequency bins of an OFDM symbol (QPSK or DCM). An IFFT is used to transform the frequency-domain information into a time-domain OFDM symbol. These are converted into continuous time-domain analog waveforms by a DAC, which is up-converted to the appropriate center frequency and transmitted. Figure 4 shows the UWB spectrum, simulated in Agilent’s Advanced Design System (ADS) WiMedia Library. What do the designers need to test in order to meet Federal Communications Commission (FCC) and WiMedia standard compliance for UWB transmitter? The most important tests are error vector magnitude (EVM), adjacent channel power ratio (ACPR), power control and packet error rate (PER).

The modulation measurements, such as EVM, provide great insight into the transmitter performance (much like doing an eye-diagram or eye-mask on digital data). EVM sheds light on “How is the overall system performing?” Starting at a system level with simulation tools, such as Agilent’s Advanced Design System (ADS), allows the designer to simplify very complex designs. The WiMedia design library in ADS (supporting all bands) is used as a starting point to experiment with system and architectural tradeoffs. The design library provides a “Golden Radio” as a starting point. A designer can add radio impairments such as transmitter amplifier gain and compression characteristics, LO phase noise, fading channel and filtering effects. All these impairments affect EVM and simulations are used to predict the sensitivity of all impairments on the EVM. Figure 5 compares some EVMs.

The primary purpose of the ACPR test is to guard against the possibility that in-band spurs may drive the 0 dBr level of the spectral mask test, which may hide potential failures. Therefore, it is desirable to construct a test that is insensitive to spurs, providing a good measurement of how much relative energy is leaking into the adjacent channel. The goal of the spectral mask test is essentially to measure how much interference is allowed to be transmitted into neighboring bands in a controlled way. The ACPR test measures the ratio of the in-band signal power to the out-of-band signal power and ensures that this ratio is at least 20 dB. ACPR can be simulated in ADS and measured with vector signal analyzer (VSA) software.

Receiver

The received signal is amplified using an LNA and down-converted to the complex baseband using I and Q mixers. The time-frequency kernel provides the sequence of sub-band center frequencies based on appropriate TFC. Next, the complex baseband is low pass filtered to reject out-of-band interferers. The signal is then sampled and quantized using a 528 MHz ADC to obtain the complex digital baseband signal. Baseband processing begins with the packet detection and FFT operation follows to obtain the frequency-domain information. The output of the FFT is equalized using a frequency-domain equalizer (FEQ). A phase correction is applied to the output of the FEQ to undo the effect of carrier and timing mismatch between transmitter and receiver. The pilot tones in each OFDM symbol are used to drive the digital PLL. The output of the FEQ is demapped and deinterleaved before passing on the Viterbi decoder. The error-corrected bit sequence is now descrambled and passed on to the MAC. Two important considerations in WiMedia PHY receivers are receiver sensitivity and receiver PER. These results will vary depending on the receiver complexity. The challenge is to find the right balance between low cost (less complex receivers) and high end (more complex receivers). System-level simulations in ADS help quantify the tradeoffs. For example, designers need to understand the effect of using low precision ADCs (4 bits) versus high precision ADCs (6 bits). High precision ADCs obviously increase the cost and complexity of the system. Most importantly, the bit precision of the ADC largely determines the power consumption of a UWB receiver. As shown in the simulation results, the requirement for the type of ADC used can vary depending on the data rates. Figure 6 compares the simulated results (using ADS WiMedia Library) of different precision ADCs for data rates of 80 and 320 Mb/s. Just looking at the constellation, it can be seen that whereas a 4-bit ADC can be used for lower data rate applications that use QPSK modulation, it will probably not meet the requirements of high data rate applications above 200 Mb/s, where system nonlinearities require higher precision to meet packet error rate (PER) specifications. Advanced Design System (ADS) design and simulation software, from Agilent Technologies, can also be used to simulate receiver PER versus range, using additive white Gaussian noise (AWGN) and fading channel models. A receiver PER measurement is required for the WiMedia Specification. Figure 7 shows a comparison between simulations for PER of an AWGN channel and a simple fading channel. For very high data rate video services, multiple input multiple output (MIMO) techniques can be used to further enhance link reliability. The results shown in this article are all from Band Group no. 1, since most of the current developments are in this band. However, select companies are already looking at higher bands (Band Group no. 3 in particular) in order to develop radios that support international (Japan, Europe and Asia) regulatory standards. The ADS WiMedia library supports all bands. Figure 8 shows the current status on regulatory activity around the world for UWB.

Interference

Interference is an extremely important subject that will determine the technical and market acceptability of UWB systems, as shown in Figure 9. FCC regulations require that UWB must not cause any harmful interference to licensed services. Specific interference mitigation techniques need to be used for in-band and out-of-band (OOB) interference.

There are two scenarios that need to be considered:

• UWB transmitter is an interferer and a licensed band receiver is a victim.

• UWB receiver is a victim and a licensed band transmitter is acting as an interferer.

A UWB transmitter can be prevented from interfering with licensed OOB services such as GPS, PCS, ISM and WLAN bands by meeting the FCC mask emission specifications (different mask requirements for Japan, Europe and Asia). If a possible victim is present in-band (for example, if a WiMAX signal at 3.5 GHz is present in the Band Group no. 1 UWB transmitter spectrum), the prominent technique used to mitigate interference i detect-and-avoid (DAA). This technique can be classified into two methods:

• Detect and Course Avoidance: the system will run the detection algorithm and effectively shut off the band containing the victim services and continue using the bands that are free of victim services. Going back to the WiMAX example, Band 1 (3.4 to 3.9 GHz) can be shut-off and the UWB transmitter can continue to transmit in Band 2 and Band 3.

• Detect and Fine Avoidance: specific avoidance is achieved by frequency notching (also known as tone-nulling) the narrowband signal and using the rest of the spectrum. This involves inserting zeros at the FFT stage to achieve notches in the transmit spectrum. Theoretically, a five-bit DAC would achieve a 30 dB notch; practically, it is 15 to 20 dB, due to system nonlinearities. ADS can be used to simulate DAC precision versus notch depth, nonlinearities, etc. These techniques increase power consumption, add hardware complexity and cost, due to more computation at the transmitter.

UWB receivers can reject OOB interferers by simply using an appropriate filter. For example, for a Band Group no. 1 receiver, a WLAN interferer at 5.19 GHz can be filtered out using a bandpass filter centered at 3.96 GHz, with an approximately 1.5 GHz bandwidth. UWB receivers are also likely to encounter in-band interference from narrowband systems such as WiMAX signals (interfering with Band Group no. 1) that may be quite powerful, operating in an uncontrolled manner and in close proximity. A recent article by David Leeper from Intel (available at www.wimedia.org) studied this very behavior using Agilent’s SystemVue simulator. The results indicate that switching to fixed frequency interleaving TFC mode (TFC7 for Band 3, as per the WiMedia Standard) can provide protection against a powerful interferer like WiMAX. With FFI, the receiver performance will still degrade because of the interferer, but not “break” unless the interferer is so close (less than a meter) to the WiMAX receiver that front-end nonlinearities dominate the performance. The actual result will vary depending on the exact receiver implementation (ADC precision, etc). Other imperfections like LO leakage that may downconvert WiMAX interferer energy into UWB baseband can further limit receiver BER. Even then, additional filtering can be used to reduce the impact of WiMAX interferers. Figures 10 and 11 give measured data for some example devices.

In conclusion, filtering techniques in conjunction with DAA mechanisms can be used to protect WiMAX and UWB receivers from possible interference and to meet emerging international regulations.

Acknowledgment

The author would like to thank Staccato Communications and Wisair for their support in the writing of this article by providing valuable feedback and measured data from their UWB devices.

Amolak Singh Badesha received his BS degree in electrical engineering from the University of California at Davis in 2003. He is an application specialist with Agilent Technologies’ EEsof EDA division, supporting advanced design system (ADS) simulation software. He has been with Agilent for two years, working on various RF applications including handset PAs, FBAR duplexers, MMICs (all now part of Avago Technologies) and since August 2005 with Agilent EEsof EDA. His interests include RFIC design and system integration, emerging standards such as WiMAX, WiMedia and MIMO, high speed digital channel design for signal integrity and advancement in system simulation technologies.