3G is Poised to Take Over the Cellular World
After much hoopla, third-generation (3G) hardware and software are finally preparing to become a reality at a global scale. Unresolved design and protocol questions are rapidly being addressed — but does anyone want a hand-held unit for the Internet?
By the time 2004 rolls by, there will be one billion mobile phone subscribers worldwide. Industry observers predict that by then the wireless phone population will equal — or exceed — that of wired phones. This trend seems to be confirmed by the fact that, by 1997, more people were signing up for mobile than fixed telephone services.
Once yuppie toys, mobile phones are now an essential part of how a majority of people live and work. Increasing volume is reducing the cost of what was once considered an expensive service, stimulating further growth. Meanwhile, network operators are multiplying services to provide added value to customers. These services range from mass-market services such as voice mailboxes to specialist services such as using mobile phones to receive stock market share prices.
These advanced services, as well as others being contemplated, require an increase in the capability of handsets and networks to manage extensive information traffic. According to Nokia, cellular service providers look to 3G to enable them to offer a number of Internet-based and Internet-centric services, as shown in Figure 1 . The reality is that, although the necessary bandwidth will probably become available, display and battery technologies have a lot of catching up to do, while portability considerations are limiting display size. Internet content providers like Yahoo, Excite and CNN, however, are beginning to tailor some of their offerings for the cellular market.
The Coming of 3G
3G is the term given to wireless hardware and services that will allow mobile users to make video calls while simultaneously accessing a remote database or receiving E-mails and phone calls. The technology is expected to significantly expand the range of options available, allowing the delivery of communication, data and entertainment services.
Leaders in the 3G field, such as Motorola’s PCS operation, Lucent Technologies, Nokia and Ericsson, agree that the foundation already exists in the shape of today’s digital mobile phone networks. What is needed to support these advanced multimedia services is expansion of the wireless links’ bandwidth and the shoehorning of the hardware and software for all these minor miracles into an adequate ergonomic form factor.
A significant aspect of 3G is the market pull for multimode and multiband handsets. In Europe, for example, GSM already exists in two bands, and 3G is being licensed for operation in yet a third. The expectation in Europe and other parts of the world is that 3G also means 2G GSM capability. Motorola has led the way in marketing dual-band GSM capability. There is now an expectation that GSM 3G equipment will support all three bands: 900 MHz, 1800 MHz and IMT-2000. The goal of IMT-2000 is to upgrade the world’s 2G systems — including CDMA, TDMA, and GSM — to comply with a common set of 3G requirements. Figure 2 shows Motorola’s two-liter prototype wideband CDMA (W-CDMA) unit, developed for DoCoMo in Japan. Size and weights are key for development of commercial handsets, which are measured in terms of cubic centimeters and grams. New technology is difficult, so the first prototype units are quite large. In this case, the two liters is a volume displacement measure of the prototype’s size. Future test units will decrease significantly in size.
IMT-2000 is expected to provide global roaming and improve data access. It features circuit-switched capabilities for voice and packet-switched capabilities for data. 3G systems will offer up to 384 kbps data speeds initially, increasing later to 2 Mbps. An advantage of 3G is that it uses only as much bandwidth as is necessary. For a voice call, the system allocates 8 kbps; for videoconferencing or the Internet, more bandwidth is assigned.
Puzzling Out the Alphabet Soup
AMPS (Advanced Mobile Phone Service): Analog cellular phone standard used in the US and other countries.
CDMA (code division multiple access, also known as spread spectrum): Digital communication technology that uses specific modulation codes used to provide PCS capability.
EDGE (Enhanced data rate for GSM evolution): Air interface expected to provide a high speed data solution that can be deployed in limited spectrum blocks of 1 MHz in each direction.
ETSI (European Telecommunications Standards Institute)
GPRS (General Packet Radio Service): A packet-switching technique to optimize network resources for the efficient transmission of data.
GSM (Global System for Mobile communications): Digital communication technology used to provide PCS capability
IMT-2000 (International Mobile Telecommunications-2000): Third generation wireless standards defined by the International Telecommunications Union (ITU) to provide universal coverage and enable seamless roaming across multiple networks.
IS-95 CDMA : The broadest band and only digital technology to be deployed at multiple spectrums worldwide: 850 MHz, 1.9 GHz, and 1.85 GHz
PCS (Personal Communications Services): Wireless communications services using the 1.9 GHz band which generally use all-digital technology for transmission and reception.
PDC (Personal Digital Cellular): Digital system used in Japan.
PHS (Personal Handyphone System): Japan's cordless telephone standard
TDMA (time division multiple access): Digital communication technology that uses allocated unique time slots to provide PCS capability
The Design Tightrope
Frequencies assigned by regulators around the world make up a daunting list. These are the cellular service bands that potentially could be required (or desirable) to be included in products. They, in turn, force design engineers to increase their products’ flexibility in the RF area and move toward the capability to adopt different modulation types, frequency bands and modes of operation where air interfaces allow independence between receiver and transmitter. This required capability means no overlapping receiver and transmitter operation such as US-TDMA and GSM, while others may require simultaneous, duplex operation, such as CDMA.
Incorporating these characteristics into a global mode handset is, well, challenging. Manufacturers like Motorola and Lucent are looking at the technologies needed to support these requirements, such as wideband tunable RF circuits. A related technology is software-defined radio (SDR), once strictly a military option, but now, thanks to Moore’s Law, a commercial possibility relevant to cellular products — base stations, not necessarily handsets. Direct-conversion receiver architectures and direct-launch transmitters also are considered because they allow some selectivity functions to be moved into an implementation that is not fixed-tuned or limited. The challenges are many, such as filters and VCOs. Obviously, in this environment, it is impossible to have one antenna and one power amplifier per frequency band.
A central dilemma is whether to allow retuning or design for broadband operation. There is no need to outline here the difficulties associated with very broadband designs of antennas and circuits (such as power amplifiers), which are dependent on narrowband techniques for efficiency. Linear modulation — associated with 3G systems and many 2G systems — is another hurdle for the amplifier to provide tightly controlled and accurate high fidelity modulation. Amplifiers must be designed to provide reasonable efficiency while still meeting performance characteristics.
Linearization techniques have natural limitations on their applications. One area is the limitation of the carrier frequency over which the system operates. Another is related to the modulated signal’s bandwidth. Beyond the desired modulated signal’s bandwidth are the linearization systems’ distortion products. Adapting to both wideband signals and the very wide 3G cellular service bands presents specific and difficult technical problems. There are fundamental limits that must be considered. For instance, linearization techniques that rely on summing signals with a specific phase relationship become difficult when that relationship must be controlled over a very large bandwidth.
Fitting It All In
The form factor adds another problem layer. Phones are getting smaller and, ergonomic questions aside, RF factors also are becoming a hindrance. Providing antennas that operate in three or four frequency bands as a single unit, tunable to different bands with reasonable efficiency, is a tall order.
Antenna diversity with effective system gain is difficult to achieve because of how the hand smothers one antenna or the other in the handset. Fortunately, 3G systems are adopting a basestation transmitter diversity technology that allows the phone to operate with a single antenna and additional processing in the baseband to achieve the same link diversity gains.
Another 3G design nightmare is the push for more features. Because the phone cannot grow, the RF portion is left with a smaller volume. Component miniaturization brings circuits closer, with the accompanying problems. As Duane Rabe, vice president of the technical staff and director for Motorola’s Personal Communications Sector put it, “In the old days of AMPS we’d analyze the frequencies generated in the radio portion and carefully select the IFs for the related oscillator frequencies to avoid certain combinations. We’re now well past the point where we can effectively avoid those kinds of self-interference problems. The days of tweaking are over and sophisticated engineering techniques are required to control, instead of dodge that energy.”
The Killer App
At a recent wireless show, the author received a demonstration of the latest cellular marvel. The tiny handset’s looks and features would have turned Captain Kirk green with envy. Then, after highlighting the phone’s capabilities, the sales rep leaned over and said in a confidential tone, “You wanna know what the real killer app is?” He looked around conspiratorially to make sure no one was close enough to hear, then whispered, “Voice!”
Voice still dominates the cellular industry. Inevitably, data will grow, but how quickly uses will be found to consume 384 kbps or above on a sustained basis is anyone’s guess. The one sure thing is that with response time, faster is always better, particularly in an interactive situation. Impatience is no stranger to the Internet.
Standard setting is less of a challenge than once thought. There are many so-called harmonization groups successfully working around the world. Standard setting will not be a serious obstacle to the deployment of 3G. The obstacles that do exist are related to the legacy of past national or regional standards — in frequency assignments that are different in Europe, the US and Korea. This challenge requires much consideration of how to accommodate the right flexibility in products to balance functionality in a cost-effective way in the right size with power consumption efficiency.
A distinction exists between global and US cellular markets. The global market will be largely GSM based, so considerable 3G technology will use W-CDMA as defined by ETSI. Japan will operate with a variation of it, and China has opted for a GSM variance, making it likely the country will also go W-CDMA. Japan, for example, has run out of PDC and PHS capacity and needs 3G for additional voice capacity. The European carriers have not reached capacity yet, but the writing is already on the wall.
Most of the Americas’ marketplace (the US and, to some extent, Latin America) has consolidation issues with GSM, TDMA and CDMA. This consolidation is already occurring between GSM and TDMA with the advent of EDGE technology. Thus, the US is likely to see two dominant factors: EDGE and CDMA-2000 (essentially, the 3G variation of IS-95).
Spectrum allocation issues make US deployment of W-CDMA a problem. The 1900 MHz spectrum is allocated to PCS while in Europe, Asia/Pacific and the rest of the world it is largely reserved for 3G. Technically, we lack green field spectrum 3G deployment. Most US carriers will have to backfill their existing spectrum with some form of 3G, whether CDMA-2000 or EDGE.
US carriers do not have capacity problems. CDMA carriers would like to see improved performance on their networks, particularly since IS-95 has delivered 6- to 10-to-1 capacity improvement instead of the promised 40-to-1 improvement over AMPS. GSM provides 8-to-1 improvement, so CDMA networks are not much of an improvement. TDMA is challenged because, although it uses less spectrum and has more efficient equipment, it has capacity issues. It does not offer the same upward evolutionary path that GSM and CDMA do. EDGE conversions will probably provide the path being sought.
Nokia firmly believes the Internet will be tailored to offer services to a cell phone. Every major Internet content provider understands the issues well and is developing content optimized for cell phone display.
Standards activities seem to be coming slowly to a consensus. For instance, the Operator Harmonization Group has reached consensus on a framework for a much more harmonized version for code division. There has also been success on the TDMA side, as well as GSM, in jointly migrating toward GPRS and EDGE. Customers in the 2G world now have a good 3G path. The TDMA Group has a path through EDGE and GPRS, which leads them to 3G, and the CDMA world can evolve to TDMA-2000 or the direct-sequence version.
For CDMA, the product plan is here. “For years we’ve anticipated adding 3G equipment to the current base station infrastructure to support evolution to 3G,” said Brian Bolliger, wireless strategy director for Lucent Technology’s Wireless Networks Group. “We have a platform rolling out early next year that will support both 2G and 3G radios. With the commonality being proposed between 2G and 3G, concerns appear more manageable now. We’re comfortable with being able to offer incremental 3G capabilities on top of the current platform.”
On the CDMA side, there are two phases currently in progress for the CDMA-2000 or multicarrier version. One is an enhanced 1.50 or 1.25 MHz capability. Lucent will feature it as a channel card replacement on current fielded systems. It is expected to deliver 144 kbps packet data capabilities in addition to nearly doubling voice capacity over the current CDMA-1 network. This capability should satisfy many 3G requirements, providing more spectrum-efficient voice as well as data speed at approximately 144 kbps. The 5 MHz version of 3G will follow, which is expected to satisfy data requirements from 384 kbps up to 2 Mbps.
TDMA has been evolving and incorporating GPRS and EDGE. Some months back, Lucent announced its plan to build a common platform that would integrate SDRs with TDMA — as well as GPRS and EDGE — capabilities and, eventually, GSM. This transition is believed to be a more manageable way to move from 2G to 3G. On the GSM side, Lucent has made progress on its W-CDMA direct-sequence capability.
With hardware platforms maturing, almost all that remains is the software. Some standards groups have been considering taking these various access techniques and making them more Internet Protocol (IP)-centric. Some working standards committees are looking at the network and its interfaces other than the air interface. The search is on to incorporate packet technologies as either asynchronous transfer mode (ATM) or IP or some combination of the two.
The question is, now that an air interface capable for a high speed data option exists, what do you do with it? It is not going to be routed over the 64 kb circuit switch pipes of traditional telephony. It is the network, not just the interface, that requires attention. There is activity toward architecting systems around packet connectivity, both ATM and IP. Some picture the backbone network, like the Internet, becoming entirely IP.
It is possible to make a phone in a nice form factor, to push a button and make a call. However, data require more than just an over-the-air protocol. The marketing brochures from any mobile carrier show what is needed to do wireless data, making it obvious why it is only one or two percent of the market. A set of cables is necessary but, when the data cable is plugged in, the phone will not sit on the table. When laid sideways, the phone does not get a very good signal, so the user may be provided a briefcase with a pocket for the phone. After all this, the user gets a blazing 9600 baud rate! It will be some time before all the pieces of a true data offering are packaged simply and conveniently.
Bolliger believes there is an overfixation on air-interface nuances, chip rates and all the other activities around the air interface when, in fact, the far greater issue is how to build a network that supports the services envisioned. “Services requiring higher speed data almost demand a fundamentally different network — namely, more packet- than circuit-centric.”
Expectation and Reality
The problems facing a worldwide implementation of 3G will be resolved. However, there is a considerable disconnect between expectation and reality and it is related to the expected pace of technology. People — including most engineers — are used to dealing with the computer industry; Moore’s Law is expected to be applicable to all technology. Those of us who are part of the RF community understand that it does not apply to most of the circuits we work with. Analog, RF-type functions are on a very different (slower) development curve.
There are fundamental functionality limitations that do not exist with digital circuits. The power amplifier is a prime example. Because it must deliver a certain amount of power, power consumption cannot be brought down to zero. There are other areas, which are power-constrained in function, that people outside RF do not appreciate. This constraint relates to circuits like low noise amplifiers that, to achieve a certain intercept point and, hence, a certain linearity and dynamic range, must be biased and consume large amounts of power. Oscillators are another example: If a specific signal-to-noise ratio is desired, the signal must be brought up to a certain level above the noise floor.
For RF, it is architectural changes that yield the most significant improvements, for example, eliminating an oscillator rather than trying to design it for low power. Eliminating a filter by changing receiver architecture reduces size and cost more quickly than advancing the technology in which the filter is implemented.
However, all this having been said, anyone (in or out of RF) who doubts the potential applications for the high bandwidth channel 3G can deliver to users is mistaken. Historically, as the capacity of a communications system increases, people find unanticipated ways to use it. Telephone modems are a good example. The growth of the Internet was dependent on the enhanced data rate that could be delivered over a twisted pair. In the days of 1200-baud modems, the Internet would not have been practical. From this perspective, it is difficult to question the utility of the higher capabilities that the new 3G hardware and protocols will deliver.