Mobile WiMax is receiving a great deal of attention, in both the popular and the trade press as well as in the educational/technical sphere. Service providers and infrastructure manufacturers regularly announce advanced wireless services, capabilities and equipment contracts. A successful launch of mobile WiMax services will depend greatly upon the service provider's investment, the infrastructure equipment quality and the availability of affordable customer equipment.
The performance, functionality and cost of equipping a mobile device with WiMax are greatly influenced by the RF subsystem. Designing a cost-effective subsystem that meets the power budget and satisfies customer expectations can be challenging. Fortunately, by selecting the right transceiver architecture and using techniques to enhance efficiency, it is possible to implement WiMax in mobile equipment cost-effectively.
Mobile WiMax overview
Mobile WiMax, based on the 802.16e standard, is largely expected to be used for next-generation broadband networks. While it is still an evolving standard, the WiMax Forum has done a comprehensive job of defining the main specifications in a number of "system profiles" that specify which features of the standard are mandatory and which are optional. This flexible system ensures interoperability, but leaves room for innovation and differentiation.
The 802.16e standard encompasses a wide range of frequency ranges, fast Fourier transform (FFT) sizes and channel bandwidths. When manufacturers seek to qualify a product, they specify which band classes they are targeting.
For economy of scale, most transceivers and RF front-end modules will cover multiple band classes--a requirement that can prove challenging for RF designers. The channel bandwidth can range from 3.5 to 10 MHz, which makes it necessary to integrate variable filter functionality within the transceiver. In addition, the transceiver must cover a fairly broad range of operating frequencies. These requirements demand a high level of integration, necessitating careful, and often novel, design techniques on the part of the IC designer.
New design parameters
Since many designers are already familiar with the requirements of fixed WiMax, it may be valuable to describe the design parameters of mobile WiMax as compared to its fixed counterpart.
A key feature of mobile WiMax is scalability; the FFT size is proportional to the channel bandwidth. In terms of the RF subsystem, this means that the subcarrier spacing and symbol time are roughly constant, and are independent of channel bandwidth (in early versions of the standard, these were identical for all channel bandwidths). For fixed WiMax, the symbol time varies from 11 to 128 microseconds. For mobile WiMax, it varies between 102 and 144 microseconds. A longer symbol time provides much more resilience to multipath fading. Since fading is to be expected in a mobile environment, the longer symbol time is very important for the success of 802.16e.
Early performance parameters
While fixed WiMax is based on orthogonal frequency-division multiplexing (OFDM), mobile WiMax is based on orthogonal frequency-division multiple access (OFDMA), where each user is given a transmit-and-receive slot in both frequency and time. The difference here is significant. For example, a fixed WiMax transmission will typically use all of the available subcarriers. In mobile WiMax, a user will normally use only a subset of available subcarriers, and these are used at a preassigned time. OFDMA allows for greater flexibility in assigning resources to optimally use the available bandwidth. In addition, using only a subset of the available subcarriers allows the transmit power for each individual subcarrier to be higher, and allows for an increase in transmit range. This technique, called subchannelization, is a key feature of mobile WiMax.
Although some may draw comparisons between Wi-Fi and WiMax, as both are OFDM-based standards, the implementation of WiMax is much more difficult from an RF point of view. The most significant difference between wireless LANs and mobile WiMax is the subcarrier spacing. For WLAN, subcarrier spacing is fixed at 312.5 kHz, while for mobile WiMax, subcarriers are spaced only about 10 kHz apart. The tighter subcarrier spacing means that close-in phase noise generated by the synthesizer must be significantly better in WiMax than in WLAN. This requirement alone precludes the use of currently available WLAN chip sets in WiMax products.
The major design constraints with mobile WiMax are size, power and cost. They in turn drive architecture choices. With tighter power and cost constraints, direct-conversion RF transceivers are proving to be the only architecture for mobile WiMax. In addition, high-efficiency power amplifiers (PAs) will be required to deliver RF power with reasonable dc power consumption. One likely way forward is to use an integrated transceiver/PA device that is optimized for high efficiency.
Multiple-input multiple-output (MIMO) technology is required for mobile WiMax. The current systems profile requires 2 x 2 MIMO in the downlink (BS to MS) and 1 x 2 SIMO in the uplink (MS to BS). This means that the mobile station will need to have a single transmit chain and two full receive chains. This puts further constraints on size, making an even stronger argument against a superheterodyne architecture. It also requires a high level of integration in the RF subsystem
Low power consumption places heavy demands on the PA design. Because of the relatively high peak-to-average-power ratio and linearity requirements, it will be difficult to achieve high efficiency and low power consumption with a standard Class A or even a Class AB power amplifier.
In an effort to relax linearity requirements, the current systems profile requires only 16QAM operation in the uplink (64QAM is optional). This relaxes the error vector magnitude (EVM) requirement by 6 dB (16QAM requires a –24-dB EVM, vs. –30 dB for 64QAM). However, it is widely expected that 64QAM operation will be required, especially for laptop operation. To meet this requirement, PAs will likely start to use advanced efficiency-enhancement techniques, and a standard Class A or Class AB power amplifier will be used less often.
To illustrate this point, suppose that the desired transmit power is +24 dBm at the antenna port. There will likely be a 2-dB loss between the PA and antenna, so the transmit power from the PA must be +26 dBm, or 400 milliwatts. If a standard state-of-the-art Class AB PA is used with 15 percent efficiency, the power consumption from the PA alone will be 2.6 W. The transceiver, D/A converters and baseband chip set will increase power consumption to over 3 W, and battery lifetime will be a serious issue. If the PA efficiency is increased to 25 percent with efficiency-enhancement techniques, the PA power consumption will be reduced to 1.6 W, and the power consumption of the WiMax radio will be reduced to about 2 W. Note that 2.5 W is approximately the power consumption for a GSM transceiver used in many handheld devices today.
Another issue that can often be overlooked in mobile WiMax is the interaction between WiMax and Wi-Fi networks. For example, WiMax band Class 3 operates from 2,496 to 2,690 MHz. Since Wi-Fi operates from 2,412 to 2,462 MHz (in the United States), the frequency separation between Wi-Fi and WiMax radios is very small, and there is a significant chance of harmful interference occurring between the two networks. Because the two bands are so extremely close, RF filtering is not practical, and a WiMax radio must be designed to be able to operate in the presence of a Wi-Fi radio. This requirement primarily affects the WiMax receiver, since it must be able to receive weak WiMax signals in the presence of potentially very strong Wi-Fi signals. These Wi-Fi blocker signals may be much stronger that any potential WiMax blockers.
Mobile WiMax networks are being deployed now, and they are expected to be built out quite rapidly, particularly in the United States. This means that manufacturers will need to find innovative ways to bring products to market in the shortest possible time.
To minimize risk, it is imperative that manufacturers are involved both in standards development and with the WiMax Forum to ensure that they have access to design requirements as early as possible. Having a fully functioning RF reference design that can pass the WiMax Forum's radio conformance test specification will speed integration. It also helps to have built-in test and diagnostic functions in the RF chip set.
Once an RF chip set is available, it must be integrated with a baseband chip set. This integration can take a long time, and this is an area where significant time savings can be achieved. To be successful, RF and baseband manufacturers should start interacting early. To minimize integration time, the degree of interaction between the RF and baseband chip sets themselves should be minimized. For example, if the RF chip set can implement fully autonomous automatic gain control and calibration routines, software development time is shorter and overall integration time is reduced.
Another way to reduce integration time is in the baseband-RF interface itself. In the past, analog interfaces were used to transfer data between the RF and baseband chip sets. Each baseband chip would have slightly different requirements for this interface, resulting in increased integration time.
For mobile WiMax, new standards are being developed for an all-digital baseband-RF interface. These standards define the format for data transfer between the baseband and RF chip set, as well as the format of the serial control interface. In addition to allowing for easier integration, a standardized digital interface also results in a lower-cost design. Since the baseband transceiver is all-digital CMOS, it scales well with state-of-the-art CMOS processes, enabling smaller and lower-cost baseband chip sets.Unfortunately, neither RF nor analog converters can be easily scaled. It therefore makes good sense to keep the converters with the RF, where both can be implemented in optimal technology without affecting the CMOS baseband die.
Mobile WiMax represents significant challenges to RF designers, but armed with the necessary information, design techniques and devices, they have a good chance at meeting their time-to-market, power, size and cost goals.
Darcy Poulin (firstname.lastname@example.org), senior systems engineer at SiGe Semiconductor, has more than 10 years of experience in RF engineering and IC design. Poulin has significant expertise concerning regulatory issues and also holds several patents key to RF front ends. He holds a bachelor of science degree with honors in engineering physics from Queen's University at Kingston, Ontario, and a PhD in applied physics from McMaster University in Hamilton, Ontario.
Andrew Parolin (email@example.com) is director of wireless data products at SiGe Semiconductor. Parolin manages the development of the company's wireless products, which are used in wireless local-area network (WLAN) systems, WiMax equipment, Bluetooth devices and cordless telephones. He holds a bachelor's degree in engineering from the Technical University of Nova Scotia, a master of science degree from Queen's University in Kingston, Ontario, and a master of business administration from the University of Ottawa.
See related charts