No, 3G cell phones aren't available yet. But we're getting close. The first 3G phones will show up in Japan this year as NTT-DoCoMo initiates its WCDMA system in May. European countries will no doubt be next as they're already on the upgrade path from the current Groupe Speciale Mobile (GSM) systems to WCDMA. The U.S. is lagging due to the many competitive standards and the large number of carriers, manufacturers, and others with investments in specific technologies. Moreover, distractions caused by 2.5G standards further delay real 3G. Most carriers and equipment manufacturers believe that 3G won't hit the U.S. in full force until 2003 or 2004.
Today, the world is firmly in the midst of 2G cell phones or digital cell phones. Europe and most of the rest of the world supports GSM, a TDMA system. In the U.S., the older analog AMPS system is still widely used along with a variety of 2G digital systems. One of the most widely used systems here is the CDMA system invented by Qualcomm Inc. Formerly called CDMAOne, it's based on the IS-95A/B standards.
The IS-136 TDMA system also is popular throughout North America. But GSM is additionally used in the higher frequency, 1900-MHz PCS band in the U.S.
Characteristics of 3G phones were spelled out by the International Telecommunication Union's (ITU's) IMT2000 guidelines announced last year. In general, a 3G phone should permit global roaming and have packet-based rather than circuit-switched data-transmission capability. The data rate should be at least 144 kbits/s from a moving vehicle, 384 kbits/s at pedestrian speed, and 2 Mbits/s from a fixed location. This makes such applications as e-mail, Internet access, and even video transmissions possible.
Carriers have several options for migrating from 2G to 3G systems (Fig. 1). For GSM, it's best to go to GPRS and/or EDGE. GPRS is a 2.5G technology that works with GSM and steals time slots from voice to support data in the channel. It's essentially a software overlay in the basestation that works with GSM and provides a data rate of up to 115 kbits/s. GPRS isn't 3G, but it provides a high-speed packet-based system that speeds up available slow circuit-switched data approaches. While GPRS has yet to be deployed, several carriers have tested it, and some will implement it in 2001 as an interim solution until they decide on a firm 3G solution.
A Cutting EDGE Solution
The most promising TDMA/GSM data enhancement is termed EDGE for Enhanced Data Rates for Global Evolution. This system builds on the GPRS time-slot-stealing packet-data structure, but it adds 8PSK modulation for a data rate of up to 384 kbits/s. GSM/GPRS adopters can quickly and easily upgrade to EDGE with another software overlay and by adding new radios capable of demodulating the 8PSK signals. As far as IS-136 TDMA users are concerned, EDGE is their only hope for near-3G solutions. Motorola Inc. has an EDGE solution for IS-136 carriers with its Aspira product, which provides up to a 476-kbit/s data rate.
In Europe, the ETSI and ITU encouraged all GSM carriers to adopt the UMTS WCDMA system. This system uses frequency-division-duplexed (FDD) 5-MHz paired channels to provide a data rate of up to 2 Mbits/s. UMTS WCDMA requires lots of spectrum space, which may not be available.
Spectrum is a precious commodity in the U.S., where UMTS WCDMA might not be practical. Fortunately, a time-division-duplex (TDD) version of WCDMA has been developed. It maintains use of a single 5-MHz channel, and time-shares send and receive functions while maintaining compatibility with the FDD WCDMA. This permits widespread roaming.
In the U.S., 2G CDMA systems are being expanded to meet 3G specifications, via the cdma2000 upgrade path, which takes the IS-95A CDMA standard and enhances it to meet 3G goals. IS-95A supports 14.4-kbit/s circuit-switched data. Though IS-95B takes this to 64 kbits/s, it has never been widely adopted, nor will it probably ever be, given the other CDMA developments.
Cdma2000 supports 1X and 1XRTT packet data up to 153 kbits/s within one standard 1.25-MHz IS-95 channel. Up to 307 kbits/s is possible with a recently released upgrade version of 1X. A 3X version using three 1.25-MHz channels raises the data rate to about 460 kbits/s, which places it nearly in 3G territory. All of this is backwards-compatible with existing IS-95A systems, making the upgrade more affordable than going to a new WCDMA system. In the meantime, Motorola came up with an enhancement to 1X. Called 1X Plus, it has a data rate of up to 1.38 Mbits/s in a single 1.25-MHz channel.
Qualcomm recently an-nounced the 1XEV enhancement to cdma2000: one 1.25-MHz channel with a 2.4-Mbit/s data rate. This is clearly a 3G-level system that bears serious consideration in the U.S.
No, there won't be a single 3G standard here. We will continue to live with analog, 2G, and two or three variants of 2.5G and 3G. As a result, the real design challenge will be multimode, multiband cell phones as well as basestation solutions that support both the legacy and the newer standards.
The reasons for 3G are many and varied. But as always in business, the bottom line rules. Carriers expect to increase revenue and profits with 3G applications and expanded capacity. Users want and need the wireless data applications.
If 3G means broadband data rates, then what will be the primary applications of a 3G phone? Most believe that the 3G "killer app" will be...ta daa...voice. This should be no surprise. After all, a cell phone is first and foremost a telephone. Just because it can handle packet data doesn't mean that data applications will dominate its use.
Messaging May Exceed Voice
The most promising application is instant messaging or short-message service. This is already very popular in Europe and increasingly used in the U.S. Some even think text-to-text messaging will exceed the use of voice many times over in the coming years. E-mail is an obvious use, but most believe it will be overshadowed by short messaging.
Internet access supposedly is the most common 3G application. It's now on WAP-enabled 2G phones, of course, but it's pretty miserable. Yet things will get better as more sophisticated wireless browsers are developed and the screens get bigger and go to color.
Some other applications envisioned by carriers and so-called wireless applications service providers (WASPs) include MP3-like music downloads, games, and mobile (m)-commerce, better known as shopping. With full 2-Mbit/s 3G, two-way video becomes a reality using MPEG4. Videoconferencing is a real possibility, but will we actually do it? As usual, you will see 3G applications evolve into what customers accept and find useful.
Finally, a key application that we shouldn't overlook is the convergence of the cell phone and the personal digital assistant (PDA), such as the Palm Pilot. PDAs have gone wireless, and cell phones are incorporating PDA capabilities, so it makes sense to merge the two. This is happening right now.
As if 3G equipment wasn't complex or expensive enough, the FCC has mandated the inclusion of enhanced 911 (E911) services in all cellular systems. The FCC requires all cell-phone carriers to report the location of a cell-phone user making a 911 call to the local Public Service Answering Point (PSAP). Carriers must have systems in place by Oct. 1. One-fourth of all new cell phones must have E911 capability by Dec. 31, 2001, 50% by June 2002, and 100% by the end of 2002. This means carriers will need some kind of location system, and cell-phone users may or may not need additional hardware, depending on the carrier system used.
There are three basic ways to meet this requirement. Handset manufacturers can include a GPS receiver, which can then send latitude and longitude information to the carrier via the data path. Alternatively, the carrier can create a unique proprietary system based on a triangulation scheme. But the ultimate solution appears to be a hybrid system that involves both a GPS and a special carrier system. Vendors working on triangulation systems include Allen Telecom, Cell-Loc Inc., TruePosition Inc., KSI Inc. (TruePosition and KSI recently merged), and Xypoint Inc.
Additionally, chip vendors are actively developing what they hope will be the ultimate solution. Qualcomm recently bought SnapTrack, a company offering a GPS-based solution in the form of a GPS receiver and processor that can be incorporated into handsets. Other GPS receiver manufacturers include SiRF Technology Inc., Lucent's recent spinoff SyChip, Conexant, Infineon Technologies, and STMicroelectronics.
While the downside to carriers and handset manufacturers is extra costs, the carriers have the potential to recover some of their investment by selling the location information to companies offering location services. Examples of this include navigation directions or the locations of nearby ATM machines, restaurants, or parking lots. But will users really buy these services?
Two critical issues that must be dealt with before 3G becomes a reality are spectrum space and safety. It's clear that the available spectrum in the U.S. won't meet the projected demand or the needs of 3G services.
The FCC's primary effort has been to find a way for the U.S. to comply with its agreement made at the World Radio Conference (WRC) held in Turkey last spring. The U.S. said that it would work toward freeing up the 1710- to 1855-MHz and 2520- through 2670-MHz bands for 3G use, frequencies that coincide with those set aside by other countries for 3G. Obviously, common frequency bands are essential for global roaming. A common equipment standard would be nice as well.
In the U.S., these bands are allocated, primarily to the military and government. The FCC has been meeting with the various agencies and the National Telecommunications and Information Agency (NTIA) to resolve the conflict.
Another 3G option is the 700-MHz band now occupied by several dozen UHF TV stations on channels 60 through 69. The FCC has been trying to get broadcasters to abandon these channels, but they don't legally have to until the end of 2006, when all stations should have converted to digital TV on other frequencies. The FCC regularly issues reports on spectrum actions. (Go to www.fcc.gov for the latest word.) Even if a decision is forthcoming this year, an auction for these frequencies isn't scheduled until Sept. 30, 2002.
Waiting for a decision by 2006 or even 2002 is unacceptable. If we're going to design new handsets and basestations, we need to know now. Meanwhile, the U.S. could fall way behind Europe and Asia in offering 3G equipment and services, putting us at an economic disadvantage. This might give interim 2.5G services a fair shot. GSM/GPRS, TDMA/EDGE, and cdma2000 1X or 1XEV could turn out to be our 3G for the immediate future.
Safety and health issues also continue to impact the cell-phone industry. These include the ongoing discussion about RF causing brain cancer and an increasing accident rate caused by cell-phone use in moving vehicles.
Studies to date have indicated no concrete proof that talking with a cell phone at the ear for extended periods of time or being exposed to RF from nearby basestations is harmful. Still, people want an answer. Until then, manufacturers will have to continue reassuring customers that all is safe.
As for driving distractions, these are real. A study by the University of Toronto in 1997 stated that talking on a cell phone while driving quadrupled the risk of an accident. The real threat is that the government—local, state, or federal—will enact unpopular laws to protect us. Over 23 states in the U.S. have proposals or actual bills that will restrict cell-phone use while driving.
The design of a 3G handset—a mobile data terminal in IMT2000 terminology—is a world-class project. The challenge will be to design units that can handle a slew of data and yet be small enough to satisfy user de-mands. Basestations have the even bigger challenge of developing designs that support multiple standards for years to come. The key to all of this lies in the software and the processors used. With Internet access and multiple data applications, e-mail and messaging, location services, streaming video, music, and other multimedia applications, a robust operating system will be needed to manage things—a stable, open platform.
One choice, from Symbian Ltd., is the EPOC product family of operating systems (OSs) for cell phones and pen-based and keyboard-based PDAs. The Pearl OS is designed especially for data-enabled 3G phones. This reference design allows designers to focus only on the specific custom software that's necessary for their phone.
Microsoft Inc. has spent years trying to make its popular Windows CE software run on portable devices. But because it's large, it was never practical to use on something like a cell phone. Microsoft finally developed software strictly for cell phones. Called Stinger smart phone software, it may compete directly with Symbian EPOC. Like Windows, it could become the de facto standard for cell phones, much as Windows has been for PCs. Microsoft also has a cell-phone browser called Microsoft Mobile Explorer that supports both WAP and HTML.
Internet access also requires a browser, like the popular WAP, that works with a tiny screen as well as no keyboard or mouse. Developed by phone.com (now Openwave Systems), WAP is not only a browser but also a complete development system with its own language known as the Wireless Markup Language (WML). Like HTML, it allows Web-page designers to craft Web pages for cell phones with just four or eight lines of display. WAP is already widely used on 2G and 2.5G phones. The WAP Forum, the organization that sponsors and promotes WAP and serves as a sounding board for its more than 300 members, has a road map that includes upgrades for larger color screens and other capabilities.
The biggest gripe about WAP is its use of its own special language, WML. Everyone else uses HTML or XML to create Web pages. If an organization wants to offer a Web site for cell phones, it must recreate its pages with WML. But the WAP Forum is working on that, so eventually WAP will use HTML, XML, or some other hybrid derivative.
Perhaps the most successful browser so far is NTT-DoCoMo's i-mode software, presently only used in Japan. It supports color, graphics, animation, and other features not yet available in WAP. You can expect i-mode to be widely adopted by U.S. and European manufacturers in the near future. Because of DoCoMo's recent $9.8 billion investment in AT&T Wireless Services (AWS), you can expect to see i-mode implemented in phones for AWS' networks.
Creating one's own baseband and controller software is another major project. Most companies use their past designs and update them as needed. But if you're starting from scratch, good luck. One possibility is to go to a company like International Wireless Technologies LLC, which has a complete set of software to take care of almost any baseband or control function. Its Software Re-Programmable Transceiver Platform lets you quickly get to market with a complete software suite of programs. The company's flagship product is the MC2K embedded controller chip that's preprogrammed as you need it. Using the MC2K along with the IWT software library, you can quickly put together a product for existing or emerging wireless standards.
Development systems for DSP have improved, too. Most DSP makers offer various systems to speed up and simplify software development. John Eckard, C6000 tools product manager at Texas Instruments Inc. (TI), says the new eXpressDSP development system provides a massive base of usable routines but supplies new tools to facilitate new code design with C, a standard algorithm format, a DSP/BIOS with RTOS kernel, and a standard scheduler that facilitates code development, reuse, and transferability. Third-party support also is available.
For 3G phones, the most important components are the processors. All 3G cell phones will contain at least two processors, the embedded controller and a DSP chip. The controller takes care of the display, keyboard, and other usual housekeeping duties. Additionally, it deals with some data-handling functions, such as the protocol stack in certain phones. The GPRS, EDGE, 1XRTT, and other data protocols are managed by the controller too. As more data functions are added, the controller is expected to contribute more to the processing load.
The most widely used embedded controller in cell phones is ARM's popular 32-bit RISC ARM7 and ARM9 cores. Virtually every cell-phone manufacturer has adopted it as the controller standard. Even Motorola, which has used its own proprietary MCore/Dragonball processors, has now licensed ARM for future designs.
For basestations, other larger, more powerful but higher-power-consumption processors can be used. Motorola's PowerQuicc series is ideal for many communications applications.
A new player in this arena is the new fabless semiconductor company, Alchemy Semiconductor Inc. It recently announced its Au1000 Internet Edge Processor (Fig. 2). Based on the popular MIPS32 instruction set, it features a 32-bit core with a five-stage pipeline, 16-kbyte data and instruction caches, a multiply-accumulate (MAC) processor, and an R4000-class MMU that accommodates SRAM, SDRAM, and flash. The chip was designed for very high-speed, low-power operation. At a 400-MHz clock, it consumes less than 0.5 W. Alchemy believes the AU1000 will be a contender in the embedded area.
Another potential contender in this field is Intel Corp. and its powerful StrongARM processor. A newer and even more powerful and low-power processor based on the StrongARM is XScale. It's ideal for handsets.
As for DSP chips, they're primarily used for baseband functions related to the air interface. A DSP handles baseband encoding and decoding, compression, decompression, filtering, and a variety of other processing chores that are unique to the particular standard. In some software-defined radio phones, the DSP may also handle IF filtering, downconversion, demodulation, and modulation functions.
Most major handset manufacturers, like Nokia and Ericsson, have adopted TI's DSP chips. Dennis Barrett, TI's C5000 platform manager, says that the company's C5000 chips are in over 66% of all digital cell handsets shipped today. The TMS320C54 has been a popular device, but the newer C55x has faster speeds, lower power consumption, and code compatibility. It offers performance of up to 600 MIPS with a power consumption in the 0.25-mW/MIPS range.
The C55x DSP chip and an ARM9 controller combined on a single chip form the hardware basis for TI's new Open Multimedia Applications Platform, or OMAP (Fig. 3). When combined with a subroutine library, interfaces, and development tools, OMAP offers a complete solution for handling virtually any multimedia application in a 3G handset. TI's goal with OMAP is to provide an open-architecture platform with APIs to facilitate third-party applications development. This package is designed to work with any OS that includes Microsoft, Symbian, and Handspring. Already, Nokia, Ericsson, Sony, and Handspring of PDA fame have selected OMAP for future designs.
TI's more powerful TMS320C6000 platform is additionally gaining inroads in the cellular market. Henry Wiechman, TI's Worldwide C6000 strategic marketing manager, indicates that eight of the top 10 wireless basestation OEMs have adopted one of the C6000 models for new designs. These devices can achieve performance levels of 1200 to 2400 MIPS, critical for basestation radios that must support multiple standards as well as high-speed data applications, like video.
Another key player in the DSP arena is Motorola. The company's DSP56600 series Onyx processors have been a standard for years in many GSM and TDMA phones. Paul Marino, vice president and director of Motorola's DSP Core Technology Center, indicates that the cell-phone market is clearly driving the company's DSP growth.
The StarCore DSP chip family, developed jointly by Motorola and Lucent, carefully balances code size, clock cycles, and power dissipation for a given application. The SC140, which is similar in architecture and performance to TI's C55x, runs at 300 MHz and consumes less than 0.1 mA/DSP MAC at 1.5 V.
Another player battling for 3G DSP market share is Analog Devices Inc. (ADI). This company's Tiger SHARC ADSP2100 series DSPs are ideal for many 3G applications. ADI combines its ADSP218x class DSP on a chip with an ARM7 controller, plus memory and I/O, to form the AD6522, which is part of the Softfone chip set. When combined with the AD6521, which contains all of the ADCs, DACs, and codecs, a complete baseband solution is available for GSM/GPRS phones.
An interesting DSP development is the joint effort of both ADI and Intel to create a powerful low-power DSP core. While little is known about it, the device is most likely targeted at the potentially voluminous handset market.
Another baseband processor option for CDMA is LSI Logic's new CBP-3.0, targeted at IS-95A/B designs. It contains an ARM7 controller, two OAKDSPCore processors, mixed-signal circuitry, and standard-cell logic to provide a complete, low-power baseband solution.
While a DSP dominates, some manufacturers are still opting for an ASIC or logic solution to baseband processing in basestation design where size and power consumption are less of an issue. In complex CDMA receivers or those that must translate multiple protocols, an ASIC or FPGA is the only thing fast enough to do the job. For example, PLD manufacturer Altera has chip sets for WCDMA designs.
Some additional technologies to be aware of and factor into your 3G plans are speech recognition, voice response, smart antennas, and Bluetooth. Speech recognition seems to be a natural technology to include in any phone. Why not use it for control purposes too? This isn't new. It has already been tried with cell phones, and a few products are now available. But users report unreliable translation.
A number of companies are working on next-generation systems and services. Some that you should pay attention to include SpeechWorks International Inc., Nuance Communications Inc., Talk2 Technology Inc., phone.com and software.com (they recently merged), and Vocal Point Technologies. The real breakthrough may result from VoiceXML, a voice XTML that's de-signed for writing voice-accessible Internet content.
Another development is text-to-voice software that takes e-mails or messages and converts them to voice, a feature that will help the cell-phone distraction problem.
Smart antennas are basestation arrays that employ military-style phased arrays with DSP computing power to automatically adapt and reconfigure their radiation patterns to fit the traffic. By becoming more directional and by being able to control beam width as well as azimuth on both the uplink and downlink, these antennas permit carriers to increase their capacity without adding new equipment, simply be-cause carriers can implement a more efficient frequency reuse plan.
While not specifically a 3G technology, many carriers are expected to employ adaptive antennas in their systems at about the time they implement 3G. As the cost of new cell sites skyrockets, smart antennas seem to be a smart choice for increasing the capacity of minimal additional expenses. Furthermore, it might figure into a location-detection scheme. Some companies working on smart antennas are AirNet Communications Corp., Ericsson, and Metawave Communications Corp.
Bluetooth Headsets Available
Bluetooth is the standard for low-power radio transceivers that can be built into other electronic products to make them wireless. Designed as a wireless cable replacement, low-cost Bluetooth ICs make it possible for virtually any electronic device to communicate with any other device over a short distance. Bluetooth operates in the 2.4-GHz unlicensed ISM band up to a distance of about 10 meters. Bluetooth's founder, Ericsson, is currently making Bluetooth-enabled headsets for cell phones and cordless phones.
Cost and interference problems are the biggest issues when incorporating Bluetooth in a handset. These will certainly arise when two complete wireless transceivers, including antennas, are packaged in the same small housing.
Most crystal-ball gazers see 4G systems with data rates of up to 20 Mbits/s. Some also think that a few of the 3G applications, like video, may not be implemented until 4G comes around. Perhaps even voice communications will be packetized via the VoIP standard.
Furthermore, we can probably look for an even newer air interface. Orthogonal frequency division mulitplexing (OFDM) keeps coming up as a possibility. This super-complex method is now practical thanks to inexpensive DSPs. Plus, it's far more robust in terms of weak-signal and multipath reception. But who knows? Maybe the enigmatic ultra-wide-bandwidth (UWB) pulse technology will mature and come forward to reveal some true benefits.
|Companies Mentioned In This Report|
Analog Devices Inc.
fax (781) 326-8703
Ericsson North America
fax (908) 822-9866
LSI Logic Corp.
fax (512) 891-0318
Openwave Systems Inc.
fax (858) 658-2100
Talk2 Technology Inc.
Texas Instruments Inc.
fax (281) 274-4296