Electronic Design

3G Cell Phones: Still To Come

Economic, technological, and regulatory issues must be resolved for further progress.

Disraeli said, "Everything comes if a man will only wait." Well, 3G cell phones still aren't here, and certainly not due to a lack of effort. Handset and basestation manufacturers and carriers are struggling with a variety of cost, technical, and regulatory problems to reach the 3G nirvana. At the same time, they're trying to maintain and upgrade existing 2G systems to continue making a profit. You can be sure that 3G won't arrive on a specific date. In fact, 3G already exists in a few places around the world, but not on any grand scale.

It will be a few more years before we see 3G rolled out in the U.S. and Europe because it's complex and costly, and there's a real concern that consumers may not actually want or need it. We certainly won't see it in the U.S. until the spectrum shortage problem is addressed. The indecision of the U.S. Federal Communications Commission (FCC), the National Telecommunications Infrastructure Association (NTIA), and the U.S. Congress isn't helping U.S. deployment.

The lure of 3G technology is its potential. 3G phones promise improved digital voice communications, greater subscriber capacity, and fast packet-based data services like e-mail, short message service (SMS), and Internet access at broadband speeds. Most carriers also expect consumers to want location services, interactive gaming, streaming video, home monitoring and control, and who knows what else, while being fully mobile anywhere in the world.

But will subscribers really want or need 3G? What data services will subscribers actually lust after and pay for? Professor Nalin Kulatilaka at Boston University's School of Business says that carriers should focus on applications for mobile enterprise workers such as salesmen, claims adjusters, and other road workers.

The International Telecommunications Union (ITU) defines 3G, known as the Universal Mobile Wireless System (UMTS), as a packet-based system that achieves speeds of 144 kbits/s while fully mobile, 384 kbits/s during slow movement, and 2 Mbits/s from a fixed position. It implies full global roaming. In effect, 3G is a mobile broadband connection where even voice is Internet Protocol (IP).

The ITU/UMTS standards additionally define a specific wideband code-division multiple-access (WCDMA) radio. Other technologies also qualify as 3G under the ITU's definition. But overall, 3G means CDMA, which is clearly superior to the other digital standards, namely the GSM (Global System for Mobile) communications standard used worldwide, and the IS-136 time-division multiple-access (TDMA) standard used primarily in North America.

CDMA puts all users into a single wide-bandwidth channel and sorts them out by a pseudorandom coding process. This technique is far more spectrally efficient, as it permits more users and higher data rates in a smaller amount of spectrum than TDMA. Also, it's far more tolerant of the multipath and fading problems so common in mobile operation. The downside of CDMA is its incredible complexity.

CDMA also is a second-generation (2G) digital technology. Developed by Qualcomm, this CDMA standard is known as IS-95, or CDMAOne. It is widely employed in the U.S. but is not as widespread as TDMA.

The question is, how does a carrier move from current TDMA, or existing CDMA systems, to the new WCDMA? Three paths are emerging.

Three Paths To 3G
The first option is the path from existing IS-95 CDMA systems to Qualcomm's cdma2000 standard, which is backwards-compatible with IS-95. Available now, this new version of CDMA has been deployed in Korea and Latin America. The first systems are expected to show up in the U.S. early this year.

When used with the 1X en-hancement, cdma2000 permits a data rate of 153 to 307 kbits/s in the currently assigned 1.25-MHz voice channels. This can later be upgraded to 1X-EV-DO, which promises up to 2.4 Mbits/s in a 1.25-MHz channel. Because this technology is available now, you must wonder why more carriers aren't taking this relatively fast, easy path to 3G. Verizion, Sprint PCS, Bell Mobility, and Leap are just beginning to go this way.

Most carriers opt for the second path to 3G of working toward ITU/UMTS WDCMA. This standard uses 5-MHz wide paired channels in a frequency-division duplexed (FDD) scheme to provide a 2-Mbit/s data rate. A compatible time-division duplexed (TDD) version is available for those with insufficient spectrum.

This road to WCDMA is by way of GSM systems. All of Europe and a huge percentage of the rest of the world, including the U.S., uses GSM TDMA systems. UTMS WCDMA has been defined as the natural upgrade from GSM. Virtually all GSM carriers will take this route with multimode phones and basestations.

Along the way, most carriers are adopting an intermediate solution called General Packet Radio System (GPRS). This so-called 2.5G system adds a packet-based, always-on feature to existing GSM systems. GPRS steals time slots from the eight available in a GSM frame to carry packet data using the existing GMSK modulation. It's a fast, cheap, and easy to implement software upgrade. It can achieve practical data rates in the 28- to 56-kbit/s range with a maximum theoretical rate of 112 kbits/s--not bad for e-mail and short messages, but marginal for Internet access, browsing, gaming, or video.

Another 2.5G enhancement to GSM, enhanced data rates for global evolution (EDGE), gives an even faster data rate by adding eight-level modulation (8PSK) to achieve up to 384 kbits/s. Many carriers plan to follow the GSM-GPRS-EDGE path to WCDMA. Some even plan to skip the slower GPRS phase and go directly to EDGE. Conversion is relatively easy in most systems by replacing cards in the basestations.

Another option is to convert from the existing IS-136 TDMA systems to GSM, then follow the GSM-GPRS-EDGE path to WCDMA. The GSM system is simply overlaid with the IS-136 system, so spectrum can be shared until full conversion. This seems like an expensive and time-consuming proposition, but both AT&T Wireless Systems (AWS) and Cingular are well on their way toward this conversion.

Today is clearly the age of 2.5G systems, specifically GRPS, EDGE, and cdma2000 1X. Carriers seem to be using 2.5G to give them breathing room and find out which data services will sell best. But 3G is here now. Korea and several Latin American countries use Qualcomm's cdma2000 system. Japan's NTT DoCoMo launched its WCDMA Freedom of Mobile Access (FOMA) 3G system last October. A WCDMA 3G system is also in full operation on the Isle of Mann.

But 3G is brand new and everyone is on a steep learning curve. Lots of testing is going on. So, heavy deployment of 3G in the U.S. and Europe isn't expected until 2003 or 2004. Some project that 3G won't be fully rolled out until between 2005 and 2007, as cost, profit, spectrum, and technological problems are solved. Most project about a 50/50 split between WCDMA and cdma2000 in the U.S., with a much greater percentage of WCDMA worldwide.

It appears the technology alone is driving us toward 3G. However, Brian Rodrigues, director of product management for Qualcomm's CDMA Technologies, says applications should be the driver. Qualcomm sees location services as the leading 3G applications, like the FCC-mandated E911 service. Using Qualcomm's gpsOne hybrid GPS/network-based location system not only fulfills the E911 mandate, but also can lead to many other useful and profitable location services.

E911 permits pinpointing the location of a cell-phone user when the phone is dialed. While its full phase-in isn't expected until 2005 when it's required, carriers hope to reclaim some of their investment with so-called "location services" based on 3G.

While the technical challenges of 3G are great, designers should be happy that many new products are available to address the gnarly problems of getting to real 3G. First, 2G GSM continues to grow, says Tyson Tuttle, product manager for the wireless products division at Silicon Laboratories. Perhaps unexpectedly, adoptions of its new Aero GSM RF chip set (Si4200 and Si4201) keep growing. This three-band radio features a low IF of 100 kHz with a DSP filter to eliminate the problems of direct-conversion devices. It's compatible with both GPRS and EDGE designs.

The Silicon Labs Si4133W synthesizer also enjoys wide usage, as in Panasonic's WCDMA 3G phone used by many DoCoMo subscribers. Two synthesizers are needed in WCDMA systems that use variable rather than fixed TX-RX frequency-spacing FDD that's permitted in WCDMA systems.

It's difficult enough to design and build a 3G handset or basestation, but almost impossible without some kind of valid test signal. Luckily, those working on UMTS WCDMA products can rely on UbiNetics for test solutions. Bob Thomas, vice president for business development, says that the TM100 Test Mobile, which emulates a UMTS 3G handset for basestation testing, is very popular. It operates in the ITU 3G 1920- to 1980-MHz (TX) and 2110- to 2170-MHz (RX) bands. It's fully current with and tracks 3GPP standards updates. The CS100 is UbiNetics' emulator of a Node B terminal (basestation) for handset designers. UbiNetics also offers its WCDMA protocol stack software for handset and base station designers wanting to speed up and decrease the cost of design.

Another UbiNetics product offering is complete modules that provide a fast, easy way for manufacturers of laptops, PDAs, and other mobile devices to add plug-in UMTS WCDMA capability.

When the full implementation of 3G phones occurs, specifically the UMTS 3GPP version of WCDMA, its operation will feature the Internet Protocol, specifically IPv6 that provides a sufficient number of addresses to handle all of those new 3G subscribers. All data will be fully packetized, but keep in mind, so will voice.

When testing such systems, developer companies have found that the long IP headers used ahead of voice packets slow the connection and transmission times, use up too much bandwidth, and suffer degradation from dropped packets due to noise in the channel. Key to overcoming this problem is to use header compression, which occurs in the layer-3 IP stack (Fig. 1).

Special software has been designed to significantly compress the headers by a factor of 4 to 1 or more depending on the size of the voice payload compared to the header. This is such a superior method that it has been incorporated into the protocol standards for UMTS 3GPP 3G phones. The standard is designated RFC 3095 by the Internet Engineering Task Force (ITEF), the standards organization for Internet communications.

In the 80-byte packet, 60 bytes are for the header. This is reduced to one byte before transmission. Although data isn't affected, it may be compressed in a separate operation, depending on the application. Compressing video with the MPEG-4 standard is an example.

The overall result of this reduction in packet size is that you only use 5%, instead of 75%, of the available bandwidth for the packet. Overall total efficiency goes up, connection time goes down, and errors from noise are significantly reduced by smaller packet size.

Sweden's Effnet Group AB now offers this header compression software as a standard product. Called EffnetEdge robust header compression (ROHC) for cellular radio links, it implements the IETF RFC 3095 standard in C code. EfferNet IPHC, or IP header compression for data, implements the IETF RFC 2507 standard.

Software is now available for UMTS WCDMA releases 4 and 5. A version for cdma2000 is expected shortly. According to Rich Stamm, Effnet's director of marketing, both offer handset and basestation manufacturers faster time-to-market, lower development costs, shorter development times, and a positive performance boost.

Because CDMA systems are wideband, basestation power amplifiers must have sufficient bandwidth and be highly linear to accurately amplify the complex CDMA signal. Class-AB amplifiers in the 10-W to 100-W range are used now, but they do cause signal distortions. To correct this problem, most basestations use a feed-forward amplifier. This form of broadband linear amplifier uses two power amplifiers in a system of level adjustment, delay, and cancellation that subtracts the distortion content from the output to achieve high linearity over a wide bandwidth. But the amplifiers are complex, expensive, and highly inefficient.

PMC-Sierra has developed an alternate solution that uses predistortion and an ordinary class-AB linear-power amplifier (Fig. 2). The baseband signal is oversampled and fed to the Paladin-15 DSP chip. This chip is preprogrammed so that the distortion is opposite to the distortion produced by the power amplifier. The DSP output is converted to analog form by a digital-to-analog converter (DAC) and then upconverted to the final output frequency. It is then amplified by the power amplifier. The predistorted signal cancels the distortion added by the amplifier, producing a highly linear result.

Since the predistortion is not perfect, the power-amplifier output is sampled, downconverted, digitized, and fed back to a DSP. The DSP uses the output samples to compute new corrective compensations parameters for the Paladin chip. The predistortion is fine-tuned using the Adaptive Control Processor Compensation Estimator firmware running on the DSP to further reduce the final output distortion. This feedback also helps compensate for any other variations such as temperature, power-supply drift, mechanical shock, aging, and any other dynamic characteristic that can produce nonlinearities.

Terry Ngo, PMC-Sierra's product marketing engineer, indicates that using this system can considerably reduce basestation costs. Initial costs are significantly less, as the costly feed-forward amplifiers can be replaced by simpler, more efficient Class-AB linear power amplifiers. The resultant power savings is dramatic. Potentially, carriers can save megawatts and millions of dollars annually with this system. PMC-Sierra also recently announced its PM2360 reference design kit, which can be used to speed design time-to-market and lower development costs in new basestations for WCDMA and cdma2000.

Basestation design also is profiting from improved DSP chips. Doug Grant of Analog Devices says that the company's new TigerSharc DSP chip can potentially reduce basestation costs by up to 50%. The new ADSP-TS101S is now fast enough to process at CDMA chip rates of up to 4 Mbits/s, eliminating the need for fast special-purpose ASICs or FPGAs. Such an all-software solution fits well with the basestation's increasing need to deal with multiple radio standards and make future upgrades fast, easy, and inexpensive.

Texas Instruments (TI) is another company addressing the DSP needs of 3G radios. Its Open Media Applications Platform (OMAP), introduced over a year ago, has been adopted by many handset manufacturers and software vendors. OMAP provides a wide range of modem and baseband solutions for 2.5G and 3G phones. OMAP chips combine an enhanced ARM processor with one or two of TI's TMS32055x DSPs to perform modulation/demodulation and all baseband functions, regardless of the technology.

An example is TI's OMAP710 GSM/GPRS chip, which handles all functions for GPRS-enabled GSM 2.5G cell phones or PDAs. TI's marketing manager for its wireless components group, Richard Kerslake, reports that TI recently entered into a collaborative technology, product, and marketing agreement with Palm to provide OMAP hardware and software for Palm's future PDAs and other wireless products.

A PDA/cell-phone combo with a common keyboard and screen is a major new product category that will make suggested 3G applications possible. It may just end up being the 3G phone of choice.

In fact, PDAs and laptops are expected to compete with 3G phones for some data applications. PDAs now feature e-mail and Internet access. The same goes for laptops thanks to IEEE-802.11b wireless Ethernet developments. Wireless local-area network (LAN) ports are now showing up in many locations, such as airports, hotels, convention centers, and even restaurants and coffee shops.

Testing Poses A Major Challenge
The current status of 3G is clearly one of R&D and testing. Handset and basestation manufacturers are still developing products, and the carriers are busy testing and evaluating their performance. Test equipment is playing a major role in cellular phone systems. Testing 2.5G and 3G products involves many specific measurements to ensure compliance with current standards.

Agilent Technologies' variety of new test instruments and systems make testing and measurement much faster and easier. Agilent's new instruments automate these tests with single-button ease.

An example is the PSA II line of spectrum analyzers, which implement tests such as channel power, bandwidth, adjacent-channel power, harmonic distortion, third-order intercept, and others for multiple standards, including GSM/GPRS/EDGE, cdma2000, 3GPP WCDMA, and Bluetooth. The E4443A analyzer has a bandwidth of 6.7 GHz.

Agilent's E4438C Vector Signal Generator is designed to provide the signal inputs for testing handsets and basestations. Generating as high as 6-GHz signals, it supports all the major cell-phone standards, plus the Bluetooth and IEEE-802.11a and -802.11b wireless LAN standards.

Agilent's 8960 wireless communications test set combines a spectrum analyzer, a signal generator, and one-button operation for cell-phone production testing (Fig. 3). It supports all WCDMA, cdma2000, and 2G standards.

1. "Designers Face Tough Challenges In 3G Cell-Phone Specs," Louis E. Frenzel, Electronic Design, Oct. 2, 2000, p. 107.

2. "Driving Toward 3G Cell Phones: Are We There Yet?" Louis E. Frenzel, Electronic Design, Feb. 19, 2001, p. 113.

Need More Information?
Aglient Technologies Inc.
(800) 452-4844

Analog Devices Inc.
(800) 262-5643

Effnet Inc.
(650) 390-8701

NTT Communications Corp.
(+81) 3 6700 4010

(604) 415-6000

Qualcomm Corp.
(858) 587-1121

Silicon Laboratories
(877) 416-8500

Texas Instruments Inc.
(800) 336-5236

(+44) (0) 1763 285 265

TAGS: Mobile
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.