WiMAX advantages bring about new challenges
WiMAX, which stands for Worldwide Interoperability for Microwave Access and is a form of broadband wireless access, is based on the IEEE 802.16 standard for wireless metropolitan-area networks (MANs). Widespread deployment of the technology is expected over the next three to five years, driven by WiMAX''s ability to deliver affordable "Last Mile" broadband Internet services. Many of the companies entering the WiMAX market include those that have dominated the WLAN arena. As engineers schooled in WLAN tackle this emerging standard, they face many different system considerations, especially in terms of RF requirements and architectures.
In a typical 20-MHz channel bandwidth deployment scenario, WiMAX Forum certified products will support downlink data rates of 65 Mbits/s at close range to 16 Mbits/s at distances of 9 to 10 km, which is enough bandwidth and transmission range to deliver high-speed simultaneous access to voice, data, and video services to hundreds of businesses or thousands of residences. WiMAX''s extended range is driving a significant market opportunity. In addition, it''s proving useful in delivering broadband services to rural areas where it''s cost-prohibitive to install landline infrastructure.
Understanding the standards
How does the 802.16 WiMAX standard compare to the 802.11 WLAN standard? To start, both are based on orthogonal frequency division multiplexing (OFDM), use multiple pilot tones, and support modulations ranging from BPSK to 64 QAM.
But there are some major differences as well. For instance, rather than a fixed 20-MHz bandwidth with 52 subcarriers as in 802.11, WiMAX systems can use variable bandwidths from 1 to 28 MHz with 256 subcarriers (192 data subcarriers) in either licensed or unlicensed spectrum. The first WiMAX rollouts are expected to use 3.5- and 7-MHz channel bandwidths.
WiMAX supports subchannelization, meaning that instead of transmitting on all 192 data subcarriers, you can transmit on just a subset. In this scenario, by using the same amount of power over fewer carriers, the system achieves greater range. As WiMAX CPE evolves into in-building devices, it''ll be necessary to make up for the power loss incurred when transmitting the signal outside the building. Because CPE is typically limited in power, concentrating the power over fewer subcarriers in the uplink can balance the power in the uplink and downlink, and enable greater range.
While the larger number of subcarriers gives WiMAX an advantage over 802.11, the resulting challenge to the system design is that the subcarriers are spaced more closely together, so there are tighter requirements for phase noise and timing jitter. This translates to a need for higher-performance synthesizers.
WiMAX also uses a variable-length guard interval to improve performance in multi-path environments. The guard interval is a time delay at the beginning of the packet to compensate for multi-path interference. With a very clear channel, the guard interval can be shortened, increasing the throughput. With more subcarriers, and with a variable-length guard interval, a WiMAX system''s overall spectral efficiency will be 15 to 40% higher than a WLAN system. For instance, WiMAX achieves a spectral efficiency ranging from 3.1 to 3.8 Mbits/s/MHz, compared to only 2.7 Mbits/s/MHz for 802.11a/b/g (see the table).
Error-vector magnitude (EVM) requirements for 802.11 are specified at -25 dB, which is required to achieve a 10% packet error rate. For 802.16, EVM is held to -31 dB, which is based on a 1% packet error rate. This lower error rate helps contribute to WiMAX''s longer range. Also contributing to the longer range is the receiver noise figure, which is more stringent for 802.16. Specifically, 802.11''s maximum noise figure is 10 dB, while 802.16 operates at 7 dB.
802.11 only supports time division duplexing (TDD), where transmit and receive (Tx/Rx) functions occur on the same channel, but at different times. In comparison, the 802.16 spec offers more flexibility, supporting TDD, frequency division duplexing (FDD), and half-duplex FDD (H-FDD). FDD uses simultaneous Tx/Rx on different frequencies; H-FDD transmits on different channels at different times. The approach that designers select affects cost, footprint, and design time. For example, an FDD system will cost more because simultaneous Tx/Rx requires two complete radios. However, FDD will allow greater throughput, as bandwidth is dedicated for receive and transmit, and this bandwidth is used simultaneously.
Another significant difference between WiMAX and 802.11 is ranging and transmit dynamic range. In 802.11, the output power is virtually fixed, and systems typically transmit at the same power all the time. However, for WiMAX, a ranging process determines the correct timing offset and power settings. This process ensures that transmissions from each subscriber station arrive at the base station at the proper time and at the same power level. As a result, the 802.16 standard requires that subscriber stations have a 50-dB transmit dynamic range. This allows systems that are close to the base station to back off their transmit power, while those far away can transmit at maximum power. This is significant because WiMAX supports transmit ranges of several kilometers, and transmitting at maximum power near the base station would be disastrous.
Overall system design challenges
When designing a new WiMAX system, the first question is whether the system will be TDD, FDD, or H-FDD. Many countries, such as Canada and much of Europe, are generally adopting an FDD structure. In the U.S., if the system will be used in licensed spectrum, then the duplexing will already be specified. If the system will be FDD, two complete radios (including synthesizers) operating simultaneously on different frequencies will be required.
This type of system will need extensive external filtering to prevent the transmit power from leaking into and interfering with the receiver. In addition to cost, the dual radios and filtering required become a significant concern for board space. Many industry leaders expect that base stations will use full FDD mode due to its higher throughput, while the subscriber stations will use lower cost H-FDD or TDD.
Given a choice, H-FDD can be an attractive alternative because it has a single radio (and single synthesizer), and similar costs to TDD. The key concern with H-FDD is that the synthesizer must be able to switch between the transmitter and receiver within 100 Μs. Because the system isn''t simultaneously transmitting and receiving, the filtering requirements are relaxed significantly compared to FDD.
Perhaps the 802.16 specification that has the greatest impact on system design is EVM, because the EVM must be 6 dB higher for 802.16 than for 802.11. This has a number of implications. First, all the system blocks must be more linear. Second, phase noise must be considerably better than in an 802.11 design. Tighter phase noise requirements have implications for the synthesizer, which result in a longer settling time. Third, if an I/Q interface is chosen, then I/Q balance must be tighter as well, and will likely require I/Q calibration.
The biggest impact of a tighter EVM requirement is on the power amplifier (PA). In addition to having to meet a -31 dB EVM, target, there are two other significant factors that conspire to make PA design challenging, First, the peak-to-average power ratio (PAPR) of 802.16 is higher than 802.11. Because 802.16 has more subcarriers, the PAPR is about 10 dB, which is 2 dB higher than 802.11''s 8-dB PAPR. Second, an 802.16 system typically transmits at a higher power than an 802.11 system. Hence, the PAs in WiMAX systems must deliver more power, they must be more linear, and they must be able to handle a higher PAPR than 802.11 PAs.
The end result is that the PAs will consume more power, and they''ll be less efficient. Considerable effort must be made to develop higher efficiency, more linear PAs, especially for mobile applications where power consumption is critical. It''s also possible that adaptive predistortion will be needed to achieve high linearity with high efficiency.
RF architectures
When selecting an RF architecture for a WiMAX design, the basic choice is between a superheterodyne or direct-conversion architecture. In terms of satisfying the stricter transmitter regulatory requirements, a superheterodyne architecture is advantageous because of off-chip filtering of unwanted emissions.
There are two different kinds of superheterodyne baseband interfaces: IF and I/Q. With an IF interface, the signal at the baseband processor is at a low (but not zero) frequency. Typical IF frequencies range from 10 to 50 MHz. With an I/Q interface, the signal at the baseband processor extends to dc. In this case, any I/Q imbalance will result in images that fall directly on top of the desired signal and appear as noise. Therefore, I/Q balance is critical for an I/Q interface, and it''s likely that external I/Q calibration will be required. For this reason, the IF interface is preferable because it doesn''t require any external calibration.
A direct-conversion transmitter architecture, on the other hand, takes the two I/Q inputs at baseband and directly modulates them up to the RF. This architecture is attractive because it leads to a smaller and less expensive radio design. It removes the need for an IF local oscillator, and it eliminates the surface acoustic wave (SAW) filter. The challenge with this approach is that performance is harder to maintain. For instance, any small dc offsets that occur will degrade system performance. I/Q balance is also critical. Therefore, both dc and I/Q calibration will be required. In addition, without SAW filtering, spurious transmissions may result in spectral mask emission failures.
IC options
In most current 802.11 designs, the RF, physical layer (PHY), and media access controller (MAC) are all integrated on one IC. Because WiMAX is new and production volume levels are relatively low, WiMAX ICs aren''t yet as integrated. Hence, IC designers must decide how to partition functionality. While it''s possible to integrate the transmitter and receiver, with both the RF and IF sections on one chip, this approach is not common today.
For a superheterodyne architecture, it''s common to partition the chip. The partitioning can be done either at RF/IF (where transmit and receive are on the same chip, but with separate chips for RF and IF), or at Tx/Rx (there''s a separate Tx and Rx chip, but both include RF and IF chains).
An RF/IF partitioning is a better alternative, as one synthesizer can be shared between both ICs. This technique uses one fully-programmable synthesizer located on the IF chip that produces all the required local-oscillator signals to drive both the transmit and receive paths. To achieve the best performance at the lowest cost, IC makers can use different process technologies for the two ICs. For instance, it''s possible to use a silicon CMOS process for the IF chip, and SiGe or GaAs for the RF device.