Consumers now take wireless technology for granted, but standards, power, and other issues remain in the quest for universal cutting-edge functionality.
When Qualcomm wanted to trademark the term “ubiquity” in the 1980s, the notion of always-available mobile voice service seemed a bit far-fetched—so much so that a Motorola executive responsible for the company’s new cellular communications product group, in response to an industry analyst’s curiosity, said that “Motorola will not sell more than 1 million cellular phones.”
Today, of course, many of the more than 2 billion cellular subscribers worldwide are upset when they can’t get sufficient download speeds from the Internet to watch streaming video. Wireless communications—voice, data, and now video transport—has become an “anytime, anyplace” service. We no longer hope for wireless connectivity. We expect it.
Yet the evolution of wireless as a utility comparable with cable, satellite, and broadcast for access and service quality has really just begun. Today’s wireless operators focus on specific applications and capabilities—targeting customer types, providing national and international access, and using different combinations of transport technologies tailored to provide individualized services.
Two fundamental technologies, digital communications and integrated signal processing, lie at the core of wireless communications’ evolution into the vision of “ubiquity” voiced by Qualcomm nearly 25 years ago. These two technologies are the foundation for the future viability of high-quality anytime, anyplace wireless.
The ever-growing demand for wireless bandwidth, driven by the integration of Internet access, increased voice capacity, and quality of service, has rapidly driven wireless transport technologies from analog FM into the complex domain of digital signaling. In addition, Internet access has shifted cellular technology from balanced uplink and downlink data rates to a downlink-heavy model.
One result of this change is the current battle for broadband wireless supremacy between a new generation of networks optimized for data transport based on 802.16e (mobile WiMAX) and the next-generation cellular standard known as LTE.
Interestingly enough, both LTE and mobile WiMAX use OFDMA in the device downlink, which optimizes spectral efficiency, enables high-capacity downloads from the Internet, and, as a result, creates a critical path for the evolution of broadband wireless services to the status of “utility.”
OFDMA lets system designers take advantage of state-of-the-art high-data-rate baseband technology developed for wireless local-area networks, or WLANs (today’s dominant 802.11g/n signaling technology, which also uses OFDMA). Improvements in signal processing power, physical footprint, and baseband power consumption (including its impact on device battery life) have enabled the levels of broadband performance that consumers have come to expect from their existing mobile devices.
LTE and WiMAX differ in their uplink multiple access approach, with WiMAX continuing to use OFDMA for the uplink to provide maximum data rates. LTE adopts a single-carrier approach, called SC-FDMA, that suits the evolution of cellular networks in their effort to provide voice and data services. The crux of this decision by LTE designers lies with the technical challenges that broadband wireless presents to the device uplink’s RF physical layer.
Uplink and the RF Physical Layer
The most popular cellular phone in the early 1990s was Motorola’s StarTac. Its clamshell design defined a fundamental advance in phone portability. And, unfortunately, it could burn your ear in a call that lasted longer than 10 minutes. The excessive heat was a result of the low operating efficiency of the RF PA. In fact, transmit RF PA performance has been one of the most longstanding challenges for mobile device design.
The operating efficiency of the RF PA essentially defined mobile-phone battery life until the introduction of the GaAs HBT in the late 1990s. The GaAs HBT PA eventually elevated PA efficiency into the 60% range, meaning the device converts 60% of the energy it consumes into RF energy, with the other 40% converted into heat. This advance enabled the 8 to10 hours of talk time now common in GSM/GPRS 2G mobile phones.
EDGE and WCDMA, on the other hand, introduced other technical challenges for mobile-phone PAs. Both of these mobility standards put a premium on optimizing the tradeoff between ACP, transmit power, and operating efficiency. As a result, today’s 3G phones typically feature three hours of talk time for WCDMA and four hours for GSM/GPRS/EDGE, which is in large part due to increasingly complex RF power amplifiers.
Bringing Together LTE and WiMAX
Crack open any 3G phone and you’ll find multiple RF PAs. The fundamental differences between WCDMA, EDGE, and GSM/GPRS signaling demand architectural variations in PA design that limit the practicality of functional integration. To date, there are no integrated 2G/3G PA devices.
However, the evolution of broadband wireless communications to WiMAX and LTE brings, for the first time, some level of commonality in RF signaling. This invites the possibility of a single PA architecture supporting functional integration while attaining the output power and ACP performance required by high-capacity, high-data-rate wireless networks.
The ultimate challenge is to achieve convergence in small, power-efficient PA devices with performance meeting the expectations of broadband wireless consumers. The first step in designing a converged architecture is a full understanding of LTE and how it is similar to and different from WiMAX, especially in terms of frequency, duplexing, power, linearity requirements, and modulation.
Many different bands are currently available for LTE operation. The table shows frequency bands for uplink operation defined in 3GPP TS 36.101. (For the mobile terminal, the uplink band is the band of interest, since this is the transmit frequency.)
While many bands are available, and both FDD and TDD options are shown, the main bands of interest for North America are bands 13 and 14 (700-MHz bands) and band 4 (1710 to 1755 MHz). In Europe, band 7 is expected to be widely used, with operation from 2500 to 2570 MHz. In Japan, it is likely that band 1 (1920 to 1980 MHz) will be deployed first for LTE.
In terms of mobile WiMAX, worldwide regulators are allocating spectrum in the 2.3- to 2.4-GHz and 2.5- to 2.7-GHz regions, as well as the 3.4- to 3.6-GHz band. India, for instance, recently allocated 2.3- and 2.5-GHz bands for mobile WiMAX.
With WiMAX operating at 2.5 GHz and LTE using 1.7 or 2.5 GHz, it is certainly possible to create a single PA module to handle these frequencies. This type of design would be analogous to what is being done with GPRS/EDGE/GSM quad-band PAs (which are currently offered by numerous vendors, such as RF Micro Devices, Skyworks, and TriQuint), where the PAs required for each band have the same basic calibration scheme and voltage.
While both TDD and FDD are possible in LTE and WiMAX, it is widely expected that initial LTE networks will be FDD systems while WiMAX will initially use TDD. For a converged PA architecture, this means the design will need to be more complex, including additional filtering for FDD and the ability to switch on and off without incident for TDD.
From a PA design perspective, the greatest complexity lies with an FDD system. Since it transmits and receives simultaneously with Tx and Rx on different frequencies, significant filtering is required to ensure that the transmit energy does not leak into the receiver, desensitizing it. This requires large and lossy duplex filters. (Typical losses are 2.5 to 3 dB.)
In contrast, there is no duplexer in a TDD system. Only a switch is required to move between Tx and Rx modes, and switch losses are much lower than those from duplexers, at approximately 0.5 dB.
Another challenge with FDD is that the uplink and downlink ratios are fixed and cannot be easily adjusted. Note in the table that the FDD-paired spectrum is always symmetric, with equal bandwidth allocated to uplink and downlink. Since data usage is often highly weighted toward the device downlink, the FDD spectrum may not be used efficiently.
In contrast, a TDD system can change the uplink to downlink ratio easily. To ensure that adjacent cells all remain synchronized, though, all cells in the network would normally standardize on one particular ratio.
With WiMAX and LTE systems diverging on their choice between TDD and FDD, it would certainly be desirable to develop a handset that could support both TDD and FDD operation. However, significant advancements will need to be made in TDD/FDD coexistence for this to occur.
On the plus side for integration, LTE and WiMAX will likely be similar in both output power and linearity requirements. We expect required transmit power to be on the order of +23 to +24 dBm for mobile WiMAX and LTE devices. EVM requirements for LTE systems are still being defined, but they will likely be similar to the WiMAX EVM requirements of –30-dB EVM for 64QAM and –24-dB EVM for 16QAM.
EVM is a measure of the distortion present on the signal, and it is often degraded by linearity limitations in the PA. For TDD technologies, achieving good EVM can be challenging because the characteristics of the signal path must stabilize quickly at the beginning of each transmit burst.
In terms of modulation, as noted above, both technologies use OFDMA for downlink. In the uplink, LTE technology uses SC-FDMA to reduce the relatively high PAPR found in OFDMA. SC-FDMA shares several characteristics in common with OFDMA. For example, they are both block-based modulation schemes where channel equalization is performed in the frequency domain. In some ways, SC-FDMA could be regarded as “DFT-precoded (discrete Fourier transform)” or “DFT-spread” OFDMA.
While both of these modulation technologies divide the transmission bandwidth into smaller subcarriers, a major difference between them is how they use data symbols within these carriers. OFDMA uses a number of narrowband subcarriers, extending for the entire duration of the symbol (Fig. 1). SC-FDMA has a shorter symbol time, but it occupies a wider bandwidth per data symbol.1
Figure 2 shows the effects of this lower PAPR on EVM, where the measured PAPR for SC-FDMA is about 2 dB lower than for OFDMA. This means that, for the same sized PA, an LTE system can transmit approximately 2 dB more power than a WiMAX PA, meeting a –28-dB EVM specification.
However, there are other considerations. For instance, the OFDMA system will likely perform better in fading environments and can use receivers that are less complex. And while SC-FDMA reduces PAPR, which can result in more efficient PAs, the duplexer losses from using FDD in LTE will likely negate any gains from SC-FDMA. As a result, the two systems will be approximately equal from a transmit point of view.
While the idea of a converged PA architecture for WiMAX and LTE is still only a possibility, technologies are in place to make it a likely path. Multiband PAs are already in production for WLAN and cellular, so it is possible to develop them to cover the fragmented spectrum for WiMAX and LTE.
BiCMOS PAs are in development, and they can enable adaptive bias control, which will drive higher efficiency for PAs, as will digital predistortion and analog linearization techniques. In addition, biCMOS can be used for a variety of other control functions, making it possible to include a serial interface for features such as multiband PA selection, tunable filters, and switch control with fewer control lines.
Combined, these existing technologies lay the groundwork for true WiMAX/LTE coexistence in a single PA. When that occurs, the opportunity for wireless voice, video, and data ubiquity may be truly in our grasp.
- Agilent Technologies, “3GPP Long Term Evolution: System Overview, Product Development, and Test Challenges,” http://cp.literature.agilent.com/litweb/pdf/5989-8139EN.pdf, 2008.
John Brewer Jr. is vice president, Corporate and Business Development, with SiGe Semiconductor. He holds a bachelor of science degree in electrical engineering from Santa Clara University.
Peter L. Gammel is the chief technical officer and vice president of engineering at SiGe Semiconductor. He holds a PhD in physics from Cornell University and bachelor of science degrees in physics and mathematics from the Massachusetts Institute of Technology.
Darcy Poulin is a principal systems engineer at SiGe Semicondcutor. He is a WiMAX Forum Certified RF Engineer. Also, he holds a bachelor of science degree with honors in engineering physics from Queen’s University at Kingston and a PhD in applied physics from McMaster University in Hamilton, Ontario.
- If You Want To Save The Environment, Start By Recycling Your Cell Phone
- NTIA Unveils National Broadband Map