A Chip on the Shoulder of Bell Labs

Everybody talks about driving the cost out of the network. Often that starts with work force reductions and a good facilities scrub. But maybe that's because so few bean counters understand the potential that a turbo/Viterbi decoder architecture — complete with a logMAP algorithm and a programmable logsum correction table — has for driving network infrastructure costs through the floor.

Chris Nicol, head of Bell Labs at Australia's North Ryde facility, and his teams of researchers and developers do understand. They have come up with a new chip design for Lucent Technologies base stations that combines voice and data decoding in a single architecture and complies with 3GPP standards. While the new development won't do much for existing networks, it could increase capacity for operators deploying tomorrow's 3G UMTS networks by an estimated 10%.

At the heart of Bell Labs' advancement in receiver efficiency is an interleaver address generator. That technology reduces the need to rely on a host processor and a variable width bin lookup table, which aides in table size correction accuracy. It is one of an increasing number of innovations that graduate directly from the lab to the production line.

“We started out by asking ourselves, ‘What's the best algorithm that we think we probably can't implement?’ and then set about developing it,” Nicol said. “We pulled people into the project who didn't know what they couldn't do.”

And they pulled people in from all over the world, including France, the U.K. and the U.S. The Australian team (pictured on page 33) developed the algorithms, worked on the channel decoder and produced the silicon chip. The French team worked on the receiver portion of the chip, while the others developed other aspects. A hundred researchers worked in groups of 20 to 30 in a virtual project that, in addition to technical challenges, had to overcome a barrier bigger than the Great Reef: time zones.

“The whole idea about taking a project and rotating it around the world so it is always being worked on almost simultaneously is a myth,” Nicol said. “When we get teams from three continents in on meeting, one of the teams has to give.”

It's usually the Australians, but they're used to it. “Fortunately, research is the same around the world, and that's not a myth,” Nicol said. “Researchers just love being in the lab.”

In terms of network performance and efficiency, a 10% increase in capacity is huge, said Christine Laredo, senior analyst at The Strategis Group. “The question is, how much additional capacity will there be in a real deployment rather than in a test environment?”

Time will tell, but this scenario shows that when the going gets tough, technology companies turn to their scientists. And scientists love a challenge.

No longer the full-time researcher he was when he joined Bell Labs in 1992, Nicol heads up both the research and production teams, a structure he credits with speeding the process of getting technology out of the lab and into production.

“Research teams want to come up with the most elegant solution ever so they can stand next to it and point to it and say, ‘That's my best work,’” Nicol said. “The production team is focused on delivering quality products that work across all parameters and can be delivered by a certain date.”

By heading both departments and managing them locally, Nicol gets his production team geared up even as the research progresses. “Luckily, they are driven by a passion for what they love to do. All I can really do is not hinder them.”

The technology will be used in Lucent's Flexent One BTS base station for UMTS operators. The company claims it is the first to digitally decode voice and data signals within a single channel in compliance with 3GPP standards.

Laredo said performance leadership within the 3G infrastructure space is still up for grabs. With the cost of implementing 3G wireless networks scaring operators into dragging their feet, Lucent is betting that its efficiency focus will provide a competitive advantage.

“Putting the most advanced channel coding algorithms into our base stations could directly translate into capacity improvements for the whole network,” Nicol said.

Lucent claims a 12% increase in link sensitivity over other logMAP solutions, saying it increases the coverage area of the base station and the number of users it can handle by 10%.

“All cellular systems are interference-limited,” said Jim Gunn, Ph.D. and senior analyst at Forward Concepts. “But these are advanced techniques and algorithms that mitigate interference more effectively than has been done traditionally, which was to let [channels] interfere with each other and assume that you could decode them.”

With the efficiencies Lucent will theoretically get with the new decoder, the technology could go a long way toward addressing the problems of cell site density in 3G network. “You'd have a better chance at installing wideband CDMA or cdma2000, for that matter, without having to re-engineer your cell site design layout,” Gunn said. “Knowing them and the Bell Labs tradition, I have no doubts they can do what they say they can do.”