Blind faith: While leading a digital standards holy war, prompted mostly by economic nationalism, U.S. carriers and vendors dismissed and ignored facts about the real complexity and cost of IS-

A banner three stories high lobby hangs in the spacious lobby of Qualcomm's headquarters in La Jolla, Calif. "We're building the wireless world," it reads. Like many things at Qualcomm, this statement is true only in the broadest sense.

There is no diminishing Qualcomm's role in the development of the current narrowband standard for code division multiple access (CDMA), which has become the dominant digital wireless technology in North America.

Commercial CDMA networks are also operating in South Korea and Hong Kong. And CDMA has been purchased, or selected for field trial, by wireless carriers in Argentina, Brazil, Chile, China, Germany, Israel, Peru, the Philippines and Thailand.

But behind the banner's sweeping statement are some overlooked details.

The wireless world, in truth, is being built by many. Of the numerous manufacturers of CDMA, Qualcomm is among the smallest. The number of people in the world with CDMA phones is dwarfed by the millions who are using other types (Table 1).

But at Qualcomm, broad strokes matter most.

To this day, Qualcomm markets the version of CDMA that it pioneered, designated IS-95 by the Telecommunications Industry Association, as superior to all other digital wireless formats in terms of quality, capacity, economics and ease of implementation. The company claims that IS-95 is as mature as GSM, the digital standard used throughout Europe, most of Asia and by selected wireless carriers in the U.S.

"CDMA's 10- to 20-fold capacity increase will provide much-needed relief for wireless service carriers, especially large service providers in crowded urban areas," reads a marketing brochure Qualcomm distributed in March at Wireless '97, the Cellular Telecommunications Industry Association's annual convention.

"CDMA reduces the need for investment in new network facilities. The spread spectrum signal used by CDMA has a much better range than [advanced mobile phone service], GSM or [time division multiple access]," the pamphlet assures. "Because of this only one-half to one-fifth the number of cell sites are needed to provide the same or improved network coverage. CDMA cell sites are comparable to those of other digital technologies and lower than AMPS."

For years, these claims were made by Qualcomm and other vendors, including Motorola, Lucent Technologies and Northern Telecom, and by many of their carrier customers, particularly AirTouch Communications.

Irwin Jacobs, chairman and chief executive officer of Qualcomm, still stands by the claims today. "As far as I know, generally speaking, things are performing as anticipated," he says. When addressing the issue of cell site and base station requirements for CDMA against other formats such as GSM, Jacobs declares, "I would know of no situation where the numbers of base stations would be comparable."

But after considerable investigation of CDMA markets worldwide, and interviews with service providers deploying the technology in a number of frequency bands, it has become more than evident that, in practical applications, IS-95:

* does not offer capacity that is 10 times AMPS.

* can be difficult to optimize amid rapid subscriber growth.

* cannot be overlaid on AMPS in a 1-to-2, let alone 1-to-5, ratio with any expectation of quality coverage.

* is far from being as mature as other digital wireless technologies.

Unfortunately, any meaningful dialogue among operators and vendors to solve inherent problems in the standard has been muted by a crusade to establish CDMA as a viable technology at all costs. Discussion of ways to solve endemic problems of the IS-95 system have been intentionally suppressed while the benefits have been hyped beyond all reason.

Thus, it has taken as long as five to six years to get CDMA out of the trial phase into commercial implementation; it is reportedly costing carriers, especially the newer personal communication services providers, millions in extra costs for cell sites and delays due to RF optimization; and, perhaps in the greatest irony of all, U.S. vendors are falling behind South Korea and Japan in terms of perfecting the very technology designed to keep the U.S. competitive in the digital wireless future.

A new RF model CDMA turns on its ear the traditional RF notion that only one radio channel can carry one conversation. Instead of assigning time slots for each call the way phone networks--wired and wireless--have been engineered for decades, CDMA codes multiple conversations on a single 1.25 MHz carrier (Figure 1).

This fact alone accounts for some of the resistance against the technology, says Ira Brodsky, a wireless industry consultant and CDMA advocate. CDMA is tough to grasp, especially for professionals who have spent a lifetime living by the rule that the way to properly engineer a system is to isolate users from each other.

"The idea of putting them on the same [carrier] is foreign," says Brodsky. "It drives them nuts. They said this wasn't going to work at all."

On the contrary, CDMA does work, and it can be implemented and optimized for quality service. There is even a consensus emerging that CDMA may well be the basis for all future wireless systems.

Although Qualcomm has hitched its name to the CDMA star, and often seems to claim every CDMA patent that comes down the pike, considerable work in CDMA technology has been done for years by a slew of other companies, including Ericsson, Interdigital, Nokia, Motorola and Lucent, not to mention a handful of smaller companies that see enormous potential for CDMA in the wireless local loop. The European RACE program has been performing active research on CDMA for almost 10 years.

But serious issues, first raised several years ago about IS-95 CDMA, refuse to go away. And as current deployments of IS-95 are bearing out, early warnings and predictions that were dismissed back then have indeed come true. Any carrier considering CDMA, especially the IS-95 version, must take into account the birthing problems the technology has had worldwide.

First and foremost has been the capacity issue. CDMA advocates started out in the 1980s claiming CDMA could accommodate as much as 40 times the number of subscribers of a typical analog AMPS system. By the early 1990s, that claim was down to between 10 and 20, the numbers still used by Qualcomm.

While this might be true in the theoretical sense, the best capacity increase seen thus far has been in South Korea, where there are more CDMA subscribers than anywhere in the world. With phones that use an 8 kb vocoder, SK Telecom, formerly Korea Mobile Telecom, is getting between six and eight times AMPS capacity in the densest areas of Seoul, says Joo Sik Lee, general manager of the digital CDMA business division at the carrier.

In the U.S., where carriers, seeking better voice quality, opted for phones that use 13 kb vocoders, most say they are seeing capacity gains that are about six times AMPS.

In fact, Lucent, which is banking heavily on CDMA business worldwide, began quietly breaking ranks from Qualcomm earlier this year on capacity claims. Since the early summer, the company has been openly telling operators not to expect more than six times AMPS when using the 13 kb vocoder.

The only operators that claim to be seeing the capacity benefits as originally claimed are Sprint PCS, PrimeCo Personal Communications and AirTouch Communications. Keith Paglusch, chief executive officer of Sprint PCS, says the carrier is routinely seeing eight times AMPS. Lowell McAdam, president of PrimeCo, says his systems are seeing eight to 12 times analog.

Craig Farrill, vice president of strategic technology with AirTouch, perhaps the most influential of all CDMA operators, hedges a little. "We've seen somewhere between six-and-a-half times to 18 times [AMPS], depending on the situation," he says. "Unfortunately, you'd like the 18 times to be in the center of Los Angeles and you'd like the six-and-a-half times to be in the middle of the desert. But it kind of works out the other way around."

Other companies such as Ameritech Cellular and 360 Degrees Communications say they are getting only six times analog. When asked about these reports, Jacobs suggests that they may be saying so for legal reasons only. "[They are] putting contractual terms into agreements," he says. "And contractual terms are going to be on the conservative side."

Brodsky says Qualcomm's capacity claims must be seen in context of the company's background. "Qualcomm is a computer company," he says. "They think of themselves as a computer company, unlike Motorola or Ericsson." Brodsky extends the analogy into the world of dial-up modems. Modems might be marketed and sold as offering 38.4 kb/s. Line conditions, data traffic and congestion, however, might mean the user gets an average throughput of only 17 kb/s. That doesn't make Qualcomm wrong, he says.

But when applied to wireless, it makes carrier business plans incorrect by several orders of magnitude.

Because of its unusual characteristics, CDMA cell sites can shrink and expand during the course of a single day, depending upon how many callers are in a given cell at one time.

More alarming than the capacity issue, however, has been the realization that IS-95 systems are difficult to optimize for coverage. If anything, it has been this, not capacity concerns, that have resulted in delays in getting CDMA commercialized. In fact, debate over the capacity issue, which essentially involves easy-to-grasp numbers, may have drowned out more critical reports about coverage problems, which involve much more esoteric engineering measurements and formulas.

One factor that affects CDMA coverage, and that has tended to obscure some of the problems, is the tendency of CDMA cells to change in size, or "breathe" (see sidebar at left).

As more handsets enter a cell, explains Chris Demange, director of network technology at 360, ambient RF noise and link loss increase, which causes the signal to degrade. This problem can be further exacerbated if there are other RF sources adding noise throughout the area.

"Your cell radius gets smaller if you add subscribers rapidly," says Demange. "You have to plan ahead."

CDMA specifications typically include a 3dB margin for cell breathing. The net effect is to increase the number of cells required to meet specifications.

The number of cells required appears to be larger than anticipated, however, as several carriers, including 360, Ameritech and SK Telecom, have learned. All these carriers have routinely had to install more cells than CDMA specifications require, even after allowing the margin for cell breathing.

A jihad is launched What few know is that almost four years ago, AirTouch engineers identified and quantified other range-limiting problems that may be driving the requirement for more cell sites, according to documents obtained by Telephony.

The carrier, itself a shareholder in Qualcomm, had bet heavily on the success of CDMA. It was the first carrier to commit to the technology. It also had been persuasive in getting carriers such as U S West, Bell Atlantic and Nynex to sign on. U S West and Bell Atlantic, along with AirTouch, became equity partners in PrimeCo, the 1900 MHz PCS provider, which also adopted CDMA.

But the push for CDMA was more than a standards battle, although that's how it appeared at the time. The real motive behind the crusade to establish CDMA was to get U.S. industry back into the digital wireless game.

Even as the first analog AMPS systems were being deployed in the mid-1980s, capacity issues were arising in major markets such as New York, Los Angeles and Chicago. The U.S. had a digital upgrade strategy, namely to move to the TDMA system that the TIA had designated IS-54. At the same time, the Europeans were developing their own slow frequency-hopped TDMA-based digital format, a more robust version that becameGSM.

By the late 1980s, it was becoming clear that IS-54 was a disappointment. Customer surveys and focus groups routinely found that AMPS quality was better. Many U.S. carriers had soured on TDMA. AT&T Wireless remains the only large-scale U.S. user of TDMA, which has since evolved to a new version, IS-136, sometimes called digital AMPS (D-AMPS).

But unlike IS-54, early reports from GSM operators were extremely positive. The European nations managed to work out the frequency coordination and, by 1990, GSM's success was more or less assured.

This put Ericsson and Nokia in the pole position to be worldwide leaders in digital wireless infrastructure. It also put Motorola and Lucent (AT&T Network Systems at the time), and their customers, behind the digital eight-ball.

Perhaps the biggest gambit GSM developers made was that, in pursuing the European and Asian markets, they sacrificed the chance to make the GSM air interface backward-compatible with AMPS. This left an opportunity--and an economic reason--for the U.S. to explore a digital alternative.

They found one in CDMA, a high-end radio technology that until then had applications mostly in the military. Founded in 1985, Qualcomm, then known as Quality Communications, had sufficiently developed the technology for commercial mobile communications that it had been awarded several patents.

By 1990, the hype machine had geared up big time. CDMA supporters said phones would be on the street by 1992 (Telephony, Jan. 14, 1991, page 8) and that CDMA would be a dominant standard by 1994 (Telephony, Dec. 9, 1991, page 3).

After PacTel committed to CDMA in February 1992, Bell Atlantic Mobile Systems and American Personal Communications followed suit. In June of that year, CDMA proponents won their biggest battle to date when the CTIA, which had been resisting the tide in favor of TDMA, voted to order up standards for CDMA. IS-95 was born.

The next crusade was to block encroachment of GSM, which was ramping up for deployment in Europe, into the U.S. That was predicated on establishing that CDMA was not only more advanced technologically but less expensive and easier for carriers to deploy.

It also didn't hurt to wave the flag every now and then to remind the industry, and the U.S. government, that CDMA was American born and bred. Even though Motorola and Lucent had tremendous stakes in CDMA's success, Qualcomm was regularly trotted out as the little American company that could. The feisty California enterprise was Main Street's last defense against the Viking hordes from Scandinavia.

Amid this holy war climate, anything that detracted from the message that CDMA was better, cheaper and easier was unwelcome.

Technology trials in San Diego and Seattle were hailed as grand successes. Reports that cast the slightest doubt on the workings of the technology were denied.

"There was a technology decision: Prove CDMA doesn't work or shut up," recalls George Schmitt, president of Omnipoint Communications Services in New York, who was with AirTouch at the time.

But as time went on, stories of tumult leaked. Field tests showed that, on 800 MHz systems, CDMA was prone to interference from neighboring AMPS channels. Overseas in South Korea, Shinsegi Telecom, an AirTouch joint venture, grew impatient with delays and sought to build an AMPS system despite a government mandate to develop the CDMA format. The revolt was quashed, reportedly when the U.S. Commerce Department stepped in to smooth out problems between Qualcomm and the Korean vendors.

Specific details to this day remain sketchy, and no one will comment on the record. In Korea, Shinsegi officials could not be reached. In the U.S., AirTouch would not provide Shinsegi contacts for this article.

Others involved in early CDMA deployment recall raising questions about system optimization and the complexities of both infrastructure and software, neither of which seemed to be ready for deployment in the near term despite all the predictions of imminent CDMA dominance.

The other issue was the sheer lack of CDMA handsets. Engineers employed at the time by carriers choosing CDMA say management was refusing to listen and, at worst, dismissing all questions as irrelevant.

"CDMA is a religion. You don't get facts, you get 'I believe,'" says Don Cox, a former engineer at Bell Labs and Bellcore and now a professor of electrical engineering at Stanford University.

The AirTouch test One of IS-95's unique attributes is the autonomous power control performed by the handset. Unlike any other cellular format, CDMA phones adjust their power output very fast. RF power--the energy radiated by the handset--increases and decreases automatically in order to smooth the rapidly changing signal levels received at the base station. This makes it easier for the base station to receive the signal.

>From the start, CDMA critics were raising questions about how power control aspects of IS-95 handsets would affect base station range performance, particularly when manufacturers were claiming you could get quality coverage with one-half to one-fifth the number of cell sites.

In wireless system design, range performance is gauged by a cell site's link budget. A link budget, measured in decibels, expresses the maximum amount of degradation a radio signal can withstand. As such, the link budget directly affects the number of cell sites and base stations that an operator needs in order to provide quality coverage.

Engineers determine a link budget through a mathematical calculation that factors in a number of variables, including the mobile's transmit power, data rate and receiver noise density.

A critical variable in the link budget equation is the ratio of bit energy to noise, abbreviated as Eb/No. This ratio is directly analogous to the signal-to-noise ratio used in analog systems.

CDMA vendors claimed that the handset power control aspect of IS-95 allowed the Eb/No value for CDMA base stations to be set as low as 7 to 8 dB; recent Qualcomm documents suggest an even lower 6.7 dB. These values are about 5 to 7 dB less than the values in the IS-95 technical specification.

While a 5 dB reduction may not seem like much, it actually can reduce the cell site requirement by 50% or more. From this calculation sprang the Qualcomm claims that CDMA systems require 50% to 80% fewer cell sites than other digital systems.

Since the total cost of hardware, software, site acquisition and legal fees for a single cell site can reach $400,000 to $500,000, the entire economic argument for IS-95 CDMA hung on Qualcomm's Eb/No values holding true in the real world.

And here's where things got tricky.

Most engineers agreed that IS-95 Eb/No estimates of 7 to 8 dB were sound, assuming calculations were made well within the cell site. But some warned that link budgets for range performance should be calculated based on more rigorous conditions. They warned that it was extremely chancy, if not incorrect, to base an entire system design on the assumption that handsets would never have to transmit at maximum power.

Specifically, engineers questioned what would happen if many CDMA handsets in a given cell were transmitting from close to the edge of the cell and thus at peak power output. With phones transmitting from the edge, link budgets based on an Eb/No of 7 dB with power control might be too low, and the cell base station would begin dropping calls. The 7 to 8 dB Eb/No value, while good for capacity, would not hold up under limited power conditions.

In the fall of 1994, engineers at AirTouch Communications set out to answer the question. With a Qualcomm observer on hand, the engineers performed a laboratory test of the range performance of CDMA handsets at the edge of a cell site. What they found was that, as handsets approached the edge, frames began to clip, noise increased and calls dropped.

AirTouch engineers concluded that the correct Eb/No value for CDMA base stations should be 14 dB, not 7 to 8. This 5 to 6 dB error, the report said, results in an approximately 100% overestimation of the area covered by a CDMA base station. The correct number of base stations, in absence of the link budget error, would double the number of cells--and significantly increase the cost--for a CDMA network.

It is critical to remember that the engineers did not say that CDMA wouldn't work. What they concluded, however, was that Qualcomm's economic claims about IS-95 were highly questionable. Given the cost of cell sites, any carrier that committed to CDMA and budgeted its infrastructure using the suggested link budget could easily find itself deeply underfunded. Furthermore, with the link budget margins included in the system for cell breathing, typically 3 dB, this deficiency would not have surfaced until a reasonable number of users were on the system.

AirTouch's range performance test attracted a flurry of responses from the highest levels of Qualcomm, Motorola and Nortel's cellular infrastructure organizations. While the vendors did not dispute the test's methodology, they claimed the results were interpreted incorrectly. Qualcomm was especially vociferous in defense of its own technology.

The general argument was that the Eb/No values would hold because the soft handoff feature, unique to CDMA, allows phones to hand calls off to other cells long before they reach the edge.

Upper management at AirTouch faced a critical decision. In the end, with both an investment and a reputation on the line, the company chose to listen to its vendors rather than its own people.

After that, the jihad mentality prevailed. As in any religious war, the atmosphere was charged with emotion and suspicion. Since many of the engineers involved in the test had previous experience with building GSM systems, they were deemed heretics and liabilities to the cause. Mathematical realities took second place to the demand for unquestioned submission to a doctrine that preached one only needed to believe in CDMA deeply enough and every inherent technology problem would solve itself.

Employees who performed the test were locked out of the labs for a time. In due course, most left the company. The range performance findings, say two sources involved in the test, were buried.

AirTouch's Farrill denies that there was any attempt to limit the availability of information about early CDMA experiences. "We did our own testing with no help from Qualcomm. We brought in a number of our peers--Bell Atlantic, Ameritech, GTE, U S West and many others--to verify the testing we were doing and that, in fact, we could produce the kind of performance, capacity, quality, coverage and reliability with the technology."

In a paper presented to the GSM World Congress in 1996, there is a section on range performance problems that supports the conclusions of the AirTouch test, although the author and industry consultant Richard Russell does not connect it with the carrier.

Farrill maintains that neither he nor AirTouch agrees. "The material in this paper does not reflect AirTouch's views at the time or now. It is, in our opinion, not factual information."

Predictions prove out Theoretical aspects aside, most carriers, even AirTouch, say they have to raise their base station link budgets and install more cell sites to maintain quality service, especially in high-traffic areas.

AirTouch blames it on the surrounding RF environment. "There's so much noise in parts of west L.A. that we have to raise the amount of power in the CDMA system in total, which means you have to put a few more sites out there," Farrill says. "But, on an average, it has not hurt our economics at all."

PrimeCo's McAdam says the company is doubling the number of cell sites systemswide to improve coverage and accommodate rapid growth, but he stops short of saying PrimeCo has experienced the range performance problem. "You run into edge effects in all formats," he says, echoing Farrill. "Our propagation is as good as GSM, if not better."

Nonetheless, after a fairly successful launch in Chicago in November 1996, the quality of PrimeCo's service began to deteriorate rapidly.

In an April article posted by the Newsbytes News Network, an on-line wire service for trade publications, Bob Woods reported,"I couldn't even get a signal in parts of the city. And we're not talking about the downtown area here--several times in various Chicago neighborhoods the phone could not connect with the PrimeCo network."

PrimeCo has acknowledged the coverage problems to customers in Chicago and, in an apologetic letter, promised infrastructure improvements and better service in months to come.

Although Ameritech--which in late July began competing with PrimeCo in Chicago--is pleased with its choice of CDMA, it is not shy about discussing its complexity. Optimization took more than two years, says Tony Muscato, Ameritech's vice president of network planning, in part because much work had to be done in the evening hours to prevent the buildout from interfering with the live AMPS network. But that wasn't the only factor complicating the process.

Muscato says CDMA is extremely complex and that power management, frequency reuse and forward error rates are all keys to optimization. Ameritech is overlaying CDMA on AMPS cells in downtown Chicago and Detroit at a 1-to-1 ratio, says Muscato. "Initial models said 1-to-3, but that did not meet our satisfaction levels," he says.

Other operators overlaying CDMA on analog systems report similar experiences. SK Telecom is running at a ratio of 4-to-5 in its largest cities in South Korea, says Lee. And even in a relatively small market such as Las Vegas, 360 is overlaying CDMA on AMPS cells on a 1-to-1 ratio in the densest areas, according to Demange.

Only when pressed about these findings from CDMA markets did Qualcomm concede that there is a coverage/cost/capacity tradeoff. To get the capacity gains of CDMA, says Paul Sarraffe, manager of network planning at Qualcomm, carriers have to add more cell sites or add more blocking algorithms. Companies like PrimeCo, he says, are adding subscribers so fast they are overloading their systems past the 3 dB margin that typically allowed for cell breathing.

For quality coverage in dense areas, CDMA must be overlaid 1-to-1 on AMPS, says Sarraffe, adding that

1-to-2 is doable in less dense areas. Sarraffe directly contradicts his company's marketing materials by saying a 1-to-5 ratio is not feasible except in the most rural areas.

Though being careful not to fault the technology, infrastructure vendors are rolling out equipment that seems to respond to demands from wireless carriers using CDMA to fill coverage holes--the more economically the better.

At Asia Telecom in June, Motorola's Cellular Infrastructure Group exhibited a small CDMA base station and a microcell designed to be implemented cost-effectively and in small places. The microcell, dubbed the SC 601, is especially compact and, according to Mike Malone, director of marketing for Motorola's Personal Communications division, can be configured directionally to act as a macrocell, covering seven to nine miles. Pricing starts at $80,000. At the same show, Motorola introduced its Nadine network designer that uses the intelligent network to assist carriers in modeling network growth capacity and evaluating the effect of loading.

Nadine can identify hot spots and employ "what-if" scenarios so carriers can plan moves and changes in advance of field work, says Tony Kobrinetz, vice president and general manager for the advanced products division at Motorola CIG. Nadine can be configured for all digital formats.

This week at the 1997 Personal Communications Showcase, Repeater Technologies, formerly Peninsula Wireless Communications, will introduce a repeater specifically aimed at CDMA operators. Dave Bolan, vice president of marketing, plays the economic card from the start. The repeaters can be used to fill coverage gaps, particularly on highways. Bolan is encouraging carriers to look at hybrid deployment of base stations and repeaters, especially as carriers find urban centers draining most of their resources.

AirTouch and PrimeCo are already said to be trial customers.

The Korean experience The most information about the complexity of IS-95 CDMA has come from SK Telecom. With more than 1.5 million customers, SK Telecom is indisputably the world's largest CDMA service provider.

Much like the U.S., Korea tied national economic interest to developing CDMA. With the Europeans firmly established with GSM and the neighboring Japanese lined up behind their own national wireless technology, the personal handy-phone system (PHS), Korea's Electronics and Telecommunications Research Institute, with heavy lobbying from the U.S., got the national industry to back CDMA.

ETRI, with Hyundai, Samsung, LGIC and Maxon, made an intense, focused effort to make CDMA work to its fullest potential, with little regard for the up-front capital costs.

Starting with licenses purchased from Qualcomm, the Koreans brought their renowned work ethic to the task. Even then it took more than three years, plus the near-defection of Shinsegi, to commercialize IS-95 in South Korea.

Throughout the process, SK Telecom confronted the same problems that CDMA carriers are finding elsewhere today. "It's not easy to optimize CDMA," says Lee. "CDMA cell boundaries change with the growth of subscribers."

He strongly advises that every carrier perform extensive field engineering studies on propagation during every portion of the buildout. SK Telecom employed hundreds of personnel day and night on the optimization process, he says. Vigilance must continue during the subscriber ramp-up phase as well.

In Seoul, SK Telecom has overlaid about 200 CDMA cell sites on its original 250-cell AMPS system. Some IS-95 cells are as small as 500 meters in radius in high-traffic areas, but radii can be as large as 10 kilometers on the open road, Lee says.

While SK Telecom might exemplify a cautious, eyes-open approach to CDMA, Hutchison Telecom in Hong Kong might typify a carrier that banked on CDMA living up to its claims of being fast, easy and robust. Hutchison's second wireless license in Hong Kong required that it choose between D-AMPS or CDMA, plus it faced a time limit to get the system on-line or else lose the allocation. With all this at stake, all signs suggest that Hutchison's experience with CDMA in Hong Kong has been an ordeal, if not an outright disaster--so much so that the company refuses to discuss its experience with the press, having turned down two on-site interviews with Telephony in the past 10 months.

The only information Hutchison did provide Telephony was a faxed response on Hutchison letterhead, but attributed to no one by name, that stated the company's CDMA subscriber base is "well over 100,000," network coverage is "comprehensive with more than 170 cell sites and 50 repeaters," including coverage in vehicle and Mass Transit Railway tunnels, and that four phone models are available from some 200 retail outlets.

Beyond Hutchison's comments, this is what's known to be true:

* Hutchison was the first carrier in the world to turn up CDMA commercially.

* Hutchison is a player in the world's most competitive wireless market.

* Hutchison operates a GSM system in Hong Kong and could draw on immense in-house engineering experience.

* Hutchison capped the system at about 47,000 subscribers last year, and it began loading more customers only this spring.

Based on information from sources who worked with Hutchison in Hong Kong, it appears that the carrier ran into serious RF design problems during its ramp-up phase. Unlike the Koreans, Hutchison was in a competitive situation from the get-go and time was of the essence. When the company capped the system, a decision believed to have been made in the second half of 1996, it may have been a response to the RF complexities of shrinking cells, range performance problems and interference from other RF sources, all further compounded by the treacherous RF terrain of Hong Kong and the Kowloon peninsula.

According to Masoud Bassiri, vice president and managing director of Mobile Systems International, a consulting firm that helps carriers, including Hutchison, deal with RF-related engineering issues, Hutchison found it had to add cell sites and recalculate Eb/No values.

Though Hutchison appears to be committed to CDMA buildout, it still faces other problems that seem to be endemic to IS-95, including a shortage of phones.

Some at Hutchison have come away from the experience with a bitter taste in their mouths, particularly over the way CDMA was marketed. Says Colin Tucker, who worked on the Hong Kong system and is now group technical and operations director at Orange, Hutchison's GSM PCS operator in the U.K.: "CDMA is a very good technology, but Qualcomm overhyped it."

More frequency assignments Last March at CTIA's Wireless '97, Qualcomm's Irwin Jacobs, along with several carriers using CDMA, said the format had closed the technology gap between itself and GSM and was as mature as other digital formats. While CDMA certainly had achieved a degree of market power, another look at the facts reveals a different story.

GSM operators can still offer a more robust packages of services, including e-mail and, through a more diverse array of subscriber equipment, Internet access and more varied interactive data services. The infrastructure equipment is on the order of two to three times less expensive.

On the infrastructure side, many think IS-95 won't truly achieve its potential until carriers can open additional 1.25 MHz frequency assignments on a cell-by-cell basis within their franchise areas. But to do that, the system also would have to perform reliable hard handoffs between these frequency assignments.

CDMA's soft handoff feature is well-known. In essence, as a call moves from one cell to another, the conversation's digitally encoded frames are received by base stations at two or more nearby cell sites. The system then selects the stronger of the two. Because frames are always being received by more than one base station, the odds that they will be lost or dropped, or that errors will be inserted, are less than that of TDMA systems that use hard handoff.

But CDMA's tradeoff was operation on multiple frequency assignments. As discussed, CDMA encodes conversations on the same frequency. Even when a system uses two or more frequency assignments, right now there is no way for calls to switch between them.

The solution will come from software that can tell a CDMA system how to prioritize handoffs. For example, one way is to program the system to tell the phones to look first for room on different frequencies within the same cell, then look for a soft handoff to other cells, then a hard handoff to other cells.

Carriers and vendors are racing to make this work, for systemwide deployment of additional carriers is tremendously expensive. There is talk that Motorola, Lucent or one of the Korean vendors may introduce a package along these lines as early as this week's PCS Expo, but there was no official word at press time. Either way, the company that does so will gain an immediate and significant competitive advantage among infrastructure suppliers.

More to the point, former CDMA engineers such as Omnipoint's Schmitt say that early in the process, CDMA carriers glossed over the complexities of the second carrier problem and refused to believe it would take this long to solve.

"They'll be splitting cells forever," says Schmitt. "With CDMA, you were never supposed to have to split a cell."

As of press time, only the Asian carriers had begun to use additional frequency assignments. SK Telecom confirmed that it is using four in Seoul and as many as three in other cities. Sources close to Shinsegi say that it, too, has opened four in Seoul. Hutchison reportedly has deployed at least two in Hong Kong.

However, in all cases, the second carriers were deployed systemwide, and it appears that no one has done hard handoffs between frequencies within one market. In the U.S., Sprint PCS, using Qualcomm equipment, is doing hard handoff between frequencies at the boundaries of the Oklahoma City and Tulsa, Okla., metropolitan statistical areas. SK Telecom's Lee says the Korean carrier is working on an algorithm of its own.

Although it can be argued that CDMA makes more efficient use of the radio spectrum, the current state of the technology permits carriers to use only a small portion of their spectrum allocation. While established carriers still have AMPS users on these frequencies, new PCS carriers--like PrimeCo and Sprint PCS--that spent millions to win spectrum auctions are using just a fraction of what they paid for. They may have bought a PCS country estate, but current technology has them stuck in the butler's pantry.

"Most of IS-95's problems were understood a number of years ago," says Robert Rowe, director of engineering at Aerial Communications, a GSM operator in the U.S. and former director of engineering for AirTouch's PCS project. "However, people had so much personal credibility invested in IS-95, and the manufacturers had so much R&D invested in the system, that no one wanted to hear about problems."

CDMA--the next generation Although carriers seem more willing than in the past to discuss the problems they've had in optimizing CDMA, the holy war has not entirely abated.

A new battle is brewing over wideband CDMA, the next step in the development of wireless.

One of the biggest endorsements of CDMA came this spring when NTT Docomo announced it would begin rapid development of W-CDMA, in essence conceding that Japan's PHS would not be suitable for third generation wireless in one of the world's most crowded countries.

But instead of congratulating them on their wisdom, the U.S. vendors sent Docomo the message that they were doing it all wrong.

In June in Singapore, Qualcomm, Lucent, Motorola and Nortel united behind a W-CDMA standard that would be separate and incompatible from the International Mobile Telecommunications-2000 work being done by the Europeans. Beating the drum was the CDMA Development Group--chaired by none other than AirTouch's Farrill--which will act as conduit to worldwide standards bodies.

"Everyone seems to agree that CDMA is the future. The question now is what is the optimum bandwidth for CDMA," says Brodsky, the wireless consultant. Qualcomm may have chosen to design a relatively slim 1.25 MHz channel because for a while it looked as if CDMA in the U.S. was going to operate on the narrower 800 MHz reserve band. But that constraint no longer exists.

The CDG appears to have the interests of current CDMA network operators at heart because it is seeking a standard that would be backward-compatible with IS-95. But while the IMT-2000 effort is designed to evolve GSM into a standard that uses code division techniques, there have been no noises about shutting out IS-95 compatibility either.

"There's no reason why you can't take IS-95 and get to a wideband system," argues Brodsky. "Ericsson chose not to buy into IS-95. Now they are saying, 'We want CDMA, but we're going to have a better CDMA.'"

The wild card will be what the Koreans choose to back. Should they elect to develop CDMA in conjunction with IMT-2000, things might end up coming full circle--with Qualcomm, the very company that pushed the U.S. into CDMA leadership, being a chief reason for its loss. This is not an unlikely scenario.

For one, Qualcomm has lost much of its stature in South Korea. Shinsegi proved a disaster, and there seems to be no love lost with SK Telecom.

"Qualcomm invented an engine, not a jet aircraft," exclaims SK Telecom President Jung Uck Seo with an agitation rarely seen in Asian businessmen. "The success of CDMA is not all Qualcomm's. We had people trained to develop it. We cleaned the garbage out [of the technology]. Then Qualcomm used us in their propaganda!"

Vendors of CDMA technology in formats other than IS-95 also are trying to distance themselves from Qualcomm and the narrowband standard. "IS-95 has been hyped out of all proportion," says Ray O'Leary, senior vice president of Adicom, which makes CDMA-based wireless local loop equipment. "IS-95 didn't work well and there was a lot of disillusionment."

One of the first things O'Leary must explain to customers is that his version of CDMA is not Qualcomm's. As a wireless local loop system, it does not have the troublesome power control aspects, nor does it rely on softhandoffs. It doesn't really need them for smooth operation.Some of the enmity toward Qualcomm derives from the fact that, in addition to being a CDMA patent-holder, the company also aggressively competes against its licensees. On both phones and infrastructure, Qualcomm's policy is to sell rights only to the previous generation's technology.

For example, its joint venture with Sony manufactures and sells its current line of QCP phones. The smaller Q-phones, designed to compete with tinier models using GSM and D-AMPS technologies, such as Motorola's Star-TAC, will be made and sold by Qualcomm alone.

The Korean vendors have moved away from reliance on Qualcomm to the point where they are developing their own CDMA integrated circuits for phones and infrastructure. They are beginning to enjoy success. Samsung beat out its contenders for a major contract in Shanghai, China. And Samsung is manufacturing smaller CDMA phones and getting them to markets such as Hong Kong, Peru, Indonesia and China faster.

As other manufacturers and operators do more to perfect CDMA technology, Qualcomm has been known to battle on patent grounds. Motorola, for one, renegotiated its licensing agreement with Qualcomm, reportedly because it felt that it had contributed as much, if not more, of its own intellectual property into its CDMA products.

Others color the situation with humor. During a briefing on Repeater Technologies' new CDMA repeater, David Bolan deadpans, "We did our due diligence. We don't expect to be sued by Qualcomm."

Come January, Qualcomm may well be a household word. Millions of TV viewers around the world will watch the Super Bowl live from Qualcomm Stadium in San Diego. It is a major marketing stroke that projects the company's name on a grand scale. At the same time it is somewhat fitting, for more often than not, the Super Bowl game rarely lives up to the advance hype generated by the media, the sponsors and the National Football League. So it has been with IS-95.

One can't deny Qualcomm its part in building the wireless future, as its banner in La Jolla proclaims. But the technology is young, and despite more than 10 years of development as a commercial mobile radio technology, the potential and limitations of CDMA are only now becoming truly understood.

Today, Qualcomm carries the banner. But whether the vendor, or the IS-95 standard for that matter, will ever carry home the trophy of digital wireless champion is far from certain.

Many engineers will tell you that radio planning is as much an art as it is a science. Perhaps this is more so with IS-95 code division multiple access technology.

What makes CDMA load engineering difficult is that all users are on one frequency assignment, or carrier. The link loss, or decrease in signal quality, in a given cell is directly proportional to the number of users that can be loaded onto one 1.25 MHz carrier.

The maximum recommended load is 13 or 14 users per carrier per cell. If a carrier tries to load more users per channel, the link loss increases almost exponentially (see figure), and calls may be dropped.

Cell coverage can vary tremendously throughout the day. The same phone in the same precise location may be able to get coverage during low-traffic periods such as 10 p.m. on Sunday, but no service during high-usage times.

Thus, carriers who engineer CDMA coverage during off-peak hours--as many do when overlaying on an analog system--can end up with a flawed design.

The one claim about code division multiple access that has held up continuously is its superior voice quality. All CDMA carriers say they are immensely pleased with the voice clarity over CDMA phones. Indeed, when working well, a CDMA call can sound almost as good as a landline phone call.

CDMA systems operating in the 1900 MHz personal communication services band in Tampa, San Diego and Puerto Rico scored particularly high in a real-world analysis published by Prudential Securities last March. The analysis measured qualityin a number of cities where systems were operational or being ramped up. CDMA systems operating in the original 800 MHz cellular band also scored well, but in some areas finished marginally behind GSM.

CDMA handsets, because of the power control feature and battery technology, also provide longer talk time and longer standby between charges.

It is not clear from the data, but the timing of the measurements suggest that they were done before GSM carriers added enhanced full rate codec (EFR) to their systems. Other studies have demonstrated a noticeable improvement in the transmission quality with its use. If the Prudential tests were performed today, quality scores might improve for GSM in some markets.

The fog off the harbor mixes with the hot night air to throw a damp, sticky blanket of humidity over the narrow streets and alleyways that criss-cross ahead of me.

I am in Tsim Sha Tsui at the tip of the Hong Kong's Kowloon Peninsula, surrounded by beery pubs, tiny tailor shops and street corner food stands. And I'm looking for a CDMA phone.

The notoriously unreliable phone book lists some 20 Hutchison dealers; I haven't been able to find one, despite scouring the streets of Causeway Bay, Wanchai, Sheung Wan and Central, all the while waiting for an interview with Hutchison Telecom Ltd. that would never happen.

Since I was on company time, I decided to put it to good use and started my own "street" research.

Every one of the hundreds of electronics stores in Hong Kong has a section of its window devoted to portable phones. The one I am standing by now is no exception. There's a wide selection: Motorola, Ericsson, Philips, Nokia, NEC. All of them bear the GSM logo that the Memorandum of Understanding group adopted as a sign of solidarity earlier this year. I step in from the damp.

"Do you sell CDMA phones?" I ask the man behind the counter.

"No," he says. "You must go to Hutchison." He gestures in the general direction of down the street.

I pass more such stores, their windows stocked with teeny-weenie GSM phones. I stop in a few more and ask about CDMA. The answer is the same; the only place to get it is the Hutchison store.

After wandering up and down a few more back streets, I turn onto Cameron Road, a busy thoroughfare that divides Tsim Sha Tsui with the burgeoning TST East. After a day of scuffing my worn shoes on the city's cracked, uneven sidewalks, at last I find a Hutchison store.

The top two shelves in the window are filled with GSM phones. In a corner of the third shelf from the top, I hit pay dirt. Side by side are a Sony-Qualcomm QCP model, the same basic unit sold in the U.S., and a somewhat smaller Samsung SCH-200F.

I go in and ask about CDMA.

"CDMA is no good," the salesman says immediately in clipped English. "You want GSM," he says, referring to Hutchison's other wireless service.

"But CDMA's supposed to be better," I protest.

"GSM is better."

"But I hear I can use CDMA in China. I can use CDMA in the U.S." I really don't want to walk away empty-handed.

The salesman is insistent. "CDMA is not good. GSM is superior."

I remain adamant, asking to see a price list.

"The only thing better is that CDMA is cheaper," he concedes. I wonder if he is trying to steer me toward GSM for reasons relating to his sales commission.

He shows me a price list with a side-by-side price comparison. Most of the writing is in Chinese, but the numbers are clear enough in any language. CDMA service packages start at HK$190 ($25) a month and range up to HK$990. For the same package, GSM starts at HK$390 and goes up to HK$1450.

Back at the office, I call Ira Brodsky, an industry analyst and CDMA enthusiast. His read is that the salesman may have pegged me for a European who would get more utility out of a GSM phone. "If they see a Caucasian, they might automatically sell a GSM phone," he notes.

Although Hutchison ducks the face-to-face meeting, the company does try to explain the GSM sales pitch by fax. "Both GSM and CDMA have their own superiority," the fax reads. "GSM now has more value-added services--such as short messaging, information-on-demand, etc.--with the advantage of having a longer development period. Its automatic roaming coverage is reaching over 35 countries in Europe and Asia. CDMA also excels in its voice clarity."

Along with 6 million stories, there are almost 2 million cellular phones in Hong Kong. About 100,000 are CDMA, says Hutchison. My search goes on.

For competitive reasons, carriers are loathe to reveal the number of cell sites they are deploying in their markets. This can be troublesome when attempting to verify vendor claims that CDMA operators can cover the same area with fewer cell sites than those who choose other digital formats.

However, counties and municipalities often require operators to submit information about cell site deployment for environmental reasons. The city of Tampa required PrimeCo, GTE, AT&T Wireless and Aerial Communications to submit their total cell site counts, including towers, rooftop sites, utility poles and other structures such as water towers, providing a rare comparison of all three major digital formats.

The hype * CDMA offers capacity improvement 10 to 20 times AMPS and three to four times GSM

* Quality CDMA coverage can be acheived with one-half to one-fifth the number of cell sites (a 1-to-2 to 1-to-5 ratio of CDMA to AMPS cell sites)

* CDMA is as mature as GSM

* CDMA is easy to optimize

* CDMA is being adopted by carriers worldwide

The reality * CDMA, on average, offers six times AMPS capacity with a 13-bit vocoder; eight times with an 8-bit vocoder

* In dense urban areas or areas of high RF noise, a 1-to-1 ratio of CDMA to AMPS cell sites is necessary

* CDMA vendors and carriers are still working on hard handoff/soft handoff and the complexities of introducing multiple frequency assignments. CDMA service does not yet have short message service, subscriber identification module cards or the selection of compact handsets available with GSM and other formats.

* CDMA systems can be extremely difficult and time-consuming to optimize, especially during periods of heavy subscriber load

* CDMA is being built in pockets in Asia and South America, usually alongside other formats. The U.S. does not yet have a complete national footprint. Aside from wireless local loop or the occasional isolated system, chances for CDMA deployment in Europe are virtually nil.

There is a lot to say in favor of IS-95 CDMA. It offers superior voice clarity and shows every indication of being the digital wireless technology of the future. But, from what carriers report, it can be difficult to engineer and can be especially costly for carriers who put a premium on quality. Here's a list of hints to keep in mind for prospective buyers of IS-95.

Talk to as many CDMA operators as you can about implementation. Have your engineers grill their engineers. The most forthright sources can be found at SK Telecom in South Korea, Ameritech and 360 Communications in the U.S. Other carriers, somewhat leery of the press, may be more open among peers.

Have your engineers independently field test performance of CDMA systems in different markets.

Tune out the holy war. When talking to CDMA vendors, force them to keep the discussion on the merits and drawbacks of CDMA. Be wary of any discussion of CDMA that, within minutes, switches to bashing GSM or other digital formats.

Be skeptical of any claims from vendor marketing organizations regarding capacity improvement and system economics. Involve your engineers and RF designers in vendor presentations. See how well vendors answer their questions.

Demand specific guarantees from vendors on cost, delivery and optimization assistance, capacity and range performance. Read the fine print. Do not accept ad hoc explanations.

Encourage honest discussion. As an IS-95 carrier, your business growth in part depends on how well other CDMA carriers can deal with the same problems you might encounter. Report problems and solutions to industry groups and forums. Refuse to be intimidated into silence by vendors or larger carriers with more clout. Remember that you are the customer and this is your system and your service.