The prophet & the project manager: How two guys from McCaw tackled AT&T's data dilemma

Does the most recent generation of wireless industry acolytes even know McCaw Cellular? From the youngest, baby-faced blue-collar technicians to the white-collar business school grads still paying off their student loans, McCaw is a name they vaguely associate with Seattle's other billionaire, that resident of the occasional corporate board.

But do they know about McCaw Cellular? The company that invented roaming? The company that built and acquired properties that created the first national network? Do they know that McCaw was the epitome of an entrepreneurial and aggressive corporate culture a decade before anyone went to work for a dot-com?

“It was a pretty strong, risk-taking culture,” said Rod Nelson, executive vice president and chief technology officer at AT&T Wireless, who started working at McCaw in 1985. “We were the first ones doing automatic roaming. We may forget, but people didn't have it in 1989. We thought they should, and we knew we could get it done.”

What the new kids on the block think they know is that this kind of innovation and success ultimately meets a dishonorable demise. Industry titans always fall — that's what they've been reading about in news magazines for the last two years. Companies that are hugely successful ultimately go bankrupt. The CEO ignobly tenders his resignation. Scandal follows.

But that never happened at McCaw. It went out on top. As the industry's biggest national carrier, it accomplished the amazing technical feat of stitching together that network from various acquired spectrum, migrating the whole thing to digital and setting up the systems that allowed users from one city to use their cell phones in another. And its reign ended scandal-free and without a trip to bankruptcy court when it was acquired in 1994 by AT&T in what arguably stands as the most significant merger in the wireless industry's 20-year history.

And though Craig McCaw, the company's eponymous founder, settled into a new career of entrepreneurial investing and corporate guidance, many of the people who made up the backbone of that company are still in the industry. Several of them still work for AT&T Wireless.

They are people like Nelson and Greg Slemons, executive vice president and chief network officer at AT&T Wireless, who has been working alongside Nelson since 1986. They were the real architects and builders behind McCaw's network, and over the last three years, they have begun constructing the network of the future at AT&T — the architecture that is ushering the company into the age of multimedia networking.

The project that began three years ago grew from a moment of critical decision. In the fall of 2000, the 3G future was beckoning, and there was tremendous pressure on carriers to put their strategies in place. AT&T had realized that its current digital voice network, founded on TDMA technology, would not carry it well into an approaching data evolution. Also, the data network facilities the company had been operating for almost 10 years, based on CDPD technology, were lacking in scale and designed more for short data bursts than the more continuous data experience the Internet had begun to encourage.

“It wasn't so much that CDPD was a mistake. It's just that the Internet changed everything, including what we would come to expect from the mobile data experience,” Nelson said.

“We still operate that CDPD network,” Slemons added. “It covers 98 million pops and has served us well. We just needed something that would provide a more graceful evolution to data, and allow us to move voice and data to a single network.”

Leave it to Slemons to put it in practical terms. In many ways, he is the pragmatic project manager to Nelson's philosophical seer. It might sound like a put-down to some people, but Slemons said it himself: “Rod's the visionary.”

What AT&T Wireless needed in 2000 was both a philosopher and a practitioner. Its choices were to stay with TDMA a bit longer; build a CDMA network from scratch and become — in the mold of competitors Sprint PCS and Verizon Wireless — another national CDMA carrier; or migrate to a GSM/GPRS network. Furthermore, in late 2000, the pressure was on to get it done, whatever the decision. Industry expectations were that 3G was just around the corner, with data subscribers in the millions by 2003 or 2004.

Though CDMA was prospering in the U.S., Nelson and Slemons voted for the move to GSM/GPRS. Its broad use as the global standard was just part of the appeal, according to Nelson. “Because GSM has been around for so long, it's considered to be an old technology that is not up to high-bandwidth performance standards,” he said. “We found that to be unfounded criticism.”

In terms of data performance, GSM/GPRS allows for a very quick software upgrade to higher-bandwidth EDGE technology, but it also has a voice advantage, providing a much-improved vocoder ratio over TDMA.

While these technical capabilities have long been a subject of healthy debate in different technology camps, the global acceptance of GSM cannot be argued. Slemons said it is not just that GSM carriers have more subscribers worldwide, but that vendors — particularly handset makers — devote more of their research and development resources to GSM products because they feel more secure with the bigger market. “It's simply economics,” Slemons said. “The GSM market is where the devices and infrastructure are. We had the benefit of seeing some products used in Europe first, and for us, it was just a frequency band switch.”

Iain Gillott, principal at iGillott Research, agreed that economics were at the forefront of AT&T's decision, saying that vendors convinced the carrier that GSM could be even cheaper to deploy and maintain than TDMA.

Ultimately that decision influenced other TDMA carriers — including a fledgling RBOC joint venture called Cingular — to do the same, Gillott said. It was a difficult realization for some small companies that had recently finished their TDMA buildouts. “They had to follow if they wanted to be roaming partners with one of the biggest carriers in the world,” he said.

With the commitment to GSM made, Nelson and Slemons estimated a breakneck pace for deployment. They set a three-year project schedule that called for the GSM/GPRS overlay to cover 167 million pops by the end of 2002, with the third year of the schedule devoted to a rapid EDGE upgrade covering the same landscape. “It's the steepest mountain I've ever seen in front of us, and that's what our vendors were telling us, too,” Slemons said.

AT&T's primary vendors on the project were Ericsson and Nortel for infrastructure and Nokia for handsets. If Slemons felt the pressure of facing his biggest professional challenge, he also felt a strong measure of confidence because AT&T had already worked with these vendors for several years, and had tackled a different but not much less daunting deployment project years before.

“We had to replace one vendor's 2G equipment with another's, and that had taught us how to get in sync with our suppliers on a big schedule, to communicate and to work under time constraints,” he said. “It was sort of a trial run for this project.”

As the work started, the deadline seemed to be the biggest pressure point, but then the economy tanked. The economic downturn hurt AT&T Wireless the same way it hurt so many other telecom companies — the carrier was forced to cut costs sharply, and it endured several rounds of layoffs that eliminated thousands of jobs. Even in the summer of 2003, AT&T is still reportedly set to cut up to 1000 additional jobs.

But the network migration continued, and it turned out that AT&T was well prepared to work under increasing economic burdens. “Cost leveraging was really in the plan from the beginning,” said Slemons. “It wasn't really a reflection of economic pressure. My team had it in mind for 10 years before that to always leverage our existing assets to whatever degree possible.”

The cost-saving measures AT&T undertook included integrating many of its existing 850 MHz TDMA antennas with 1900 MHz GSM technology. The carrier also realized some cost efficiencies in installing its GSM overlay vs. its previous TDMA network. It used the same command center facilities and systems for the GSM network as it did for TDMA. Also, because GSM was more prevalent worldwide, there was a high level of vendor competition that pushed equipment prices down. Moreover, AT&T had more leverage to negotiate equipment prices with its incumbent vendors because a large number of new vendor “prospectors” were lured into the hunt by AT&T's ambitious project.

“Our incumbent vendors definitely had to transform their thinking,” Slemons said. “They couldn't win this business on their incumbency and market share alone.”

However, AT&T also had to transform its philosophy about the extent to which it was willing to cooperate with its own competitors in the interest of saving money. In the McCaw days, carriers never would have looked at teaming up with their competitors to reduce costs by sharing antenna sites, but the more frugal spending attitudes of recent years made it a viable option.

“We partnered with other carriers [such as Sprint and T-Mobile] because it makes no business sense for us to have separate sites in the same area,” Slemons said. “It's a transformation in competitive thinking, but also in how network teams from different carriers work together. It's been a positive and constructive experience for us.”

Among other expense moves, AT&T took advantage of declining fiber prices, and rather than leasing, the carrier decided to buy some dark fiber to be used as transport between cell sites. It also has used wireless transport facilities to a small degree, and the carrier is looking at increasing its usage of these systems.

By the end of 2002, the GSM/GPRS migration had cost AT&T about $2.5 billion or about $15 per pop — a pretty big number, but less than pursuing a greenfield strategy and building a completely new network. Despite job cuts and other economic distractions, AT&T also finished this first phase of its three-year plan ahead of schedule — with three months to spare, in fact. (An addendum to the completion of this part of the project is that AT&T's network team now has another — albeit smaller — GSM/GPRS migration ahead of it. The company acquired its Telecorp PCS affiliate in 2001 and recently started moving its TDMA facilities in 10 states to GSM/GPRS.)

With GSM/GPRS deployed, AT&T Wireless has quickly turned its attention to EDGE upgrades, which promise to be the easier phase of the three-year project. EDGE, which could supply as much as three times the data bandwidth of GPRS, is added to the existing overlay by installing new software at each cell site. Cingular Wireless recently became the first U.S. carrier to launch EDGE in a single market, but AT&T hopes to top that by having all of its sites throughout the nation ready at year's end. “It's easier, without a doubt,” Slemons said.

Yet in some sense, the job won't even be done with EDGE. Not only will the carrier eventually migrate further up the bandwidth scale to UMTS, but also it is constantly testing and tweaking the network already in place.

“The job is never over,” said Slemons. “The focus now is optimizing the new network. It takes time to find all the defects, but performance improves every day.”

Slemons and his team are trialing a much different network than the voice-centric one they worked on at McCaw, back when drive tests were the primary service quality-testing tool. But with data-centric GPRS and EDGE, the user experience is just as important as the connection itself. “The RF issues are the same with these new data technologies, but beyond the connection, everything changes,” Slemons said. “You have to test the browser, and how long it takes pages to download. We perform different stationary tests rather than mobile [drive tests]. That's new to us.”

It's another aspect of AT&T's new data network that is a long way from what Slemons dealt with at McCaw. The heaviest burden he carried then was the cumbersome mobile phone that bulged from the breast pocket of his suit.

“I carried around a Motorola 8000 that was the size of a brick,” he said. “I was walking around with a brick under the lapel of my suit. This was back when we all wore suits.”

With much of the hardest work in the three-year plan behind him, Nelson — the architect of that plan — has a little time to reflect on what the carrier has accomplished. But he also has a better perspective on how much pressure really exists to deploy 3G, and how much of the forecast from 2000 has not proved accurate.

“Data acceptance has gone a little slower than what everyone expected,” Nelson said, in what many industry watchers would consider an understatement. However, AT&T and other companies didn't have much of a choice in late 2000 other than to plan for an aggressive buildout. Though the earlier predictions were “unrealistic,” he added, the growth of data usage also has been hampered by later-than-expected availability for data handsets, as well as user indecision regarding handset purchases.

Now, the wide variety of data-enabled mobile phones and other types of data devices presents carriers with the challenge of determining what will catch users' eyes and what won't. But Nelson doesn't fear the increasing fragmentation of the device market. “The broader industry is experimenting with different devices, and it's a pretty exciting time because we don't know yet what the prevailing form factor will be.”

While the widening variety of devices doesn't concern Nelson, he is more interested in network operators coalescing around as few network standards as possible for the sake of data users. The wireline Internet would not have progressed so rapidly had companies not united on the IP standard, he said; likewise, even though CDMA camps will remain, allowing a technology like TDMA to rest in peace will speed general user acceptance of the mobile Internet.

“We moved to the most common standard in wireless — GSM,” Nelson said. “The rest of the TDMA community has followed our lead from three years ago, and in the next six years, we'll see the market effect of that.”

In the more immediate future, AT&T is dealing with the last part of its three-year plan: the EDGE upgrade. While, as Slemons indicated, it's a quick and simple upgrade for the network operator, Nelson said that it represents a very significant step forward in the mobile data experience for users.

“In the network bandwidth continuum, there's a dotted line that separates what qualifies as utility on the left from what qualifies as very comfortable usage on the right,” he said. “EDGE is definitely on the right side of that line.” In reality, EDGE will bring the mobile Internet experience much closer to that of the broadband wireline Internet, and AT&T Wireless and other carriers will have made up the distance on the last thing still keeping wireless inferior to wireline.

After that, the future is wide open. UMTS, which AT&T already is planning to launch in four markets late next year, will be the next step, putting users further on the right side of the dotted line. There will also be an opportunity to further integrate mobile data technology with wireless LAN technology at the network level to improve the user experience even more, Nelson said. “After that, what's next? 10 Mb/s?”

Yet at the same time AT&T propels itself through new thresholds in the bandwidth continuum, other carriers are doing the same, and Slemons said the company can't rest on the success of its most recent major project. “With EDGE, we'll go a leg up on the [CDMA] 1XRTT crowd,” he said. “This is a horse race. Right now, we're a little bit ahead.”

AT&T's Evolution by Numbers

Time allotted for GSM/GPRS build: 24 months
Time it actually took: 21 months
Time allotted for EDGE upgrades: 12 months
Time it actually took: TBD
Cost of GSM/GPRS build: $2.5 billion (within budget goal)
Cost of EDGE upgrades: $334 million (estimated in 2000)