ISSCC Speakers Tout Ubiquitous Networking

SAN FRANCISCO — Researchers at the 2005 International Solid State Circuits Conference agreed that wireless networking will become ubiquitous and intelligent, but differed on how the networks will be rolled out.

In pre-conference interviews, plenary speakers Daeje Chin of South Korea and Hugo De Man of Belgium both said intelligent wireless networking would consist of building blocks based on nanotechnology or CMOS devices reflecting extreme variations on Moore's Law.

Chin, South Korea's Minister of Information and Communications, said network proliferation will be limited by regulation and consumer sentiment. De Man, a professor at the Katholieke Universiteit of Leuven, Belgium, and a senior research fellow at IMEC, saw obstacles in the device scaling required for intelligent networks.

"The ubiquitous network will give information, rather than require users to search for it," Chin said. He cited RFID tags as an example of the embedded intelligence delivered by ubiquitous wireless networks. In manufacturing and commercial warehouse distribution applications, RFID tags would register the origin, destination and ownership of packages.

Chin also asserted that the spread of "invisible silicon" tracking mechanisms must not violate privacy standards.The ministry overseen by Chin is the equivalent of a combined U.S. Federal Communications Commission and Commerce Department. The former Samsung CEO is the first high-level government official to deliver a plenary address at ISSCC.

Seoul's "839 Strategy" seeks to enhance cooperation between service providers, infrastructure builders and device developers. The plan calls for the development of eight services, three major infrastructures, and nine "growth engines." Services in development include W-CDMA and terrestrial digital TV along with wireless broadband and digital multimedia broadcast networks which will reach 10 million Korean homes (60 percent of Korean households) by 2007.

De Man echoed Chin's vision of an ubiquitous intelligent network, but envisions a high level of computational intelligence distributed throughout the network. The medical and healthcare communities will be big users of personal networked devices, he predicted. "New wearable devices will give you the ability to communicate with whatever you find in your neighborhood."

The concept of "ambient networks" borrows from work at the University of California's Berkeley Wireless Research Center. These networks of wireless sensors steal their power from heat and energy in their environment. "Your body could be the battery for an ambient device," De Man said.

Apart from the enormous embedded software issue, De Man said the proliferation of autonomous wireless networks is contingent on the extreme application of Moore's Law. Several device-level problems and critical scaling issues must still be overcome, he insisted.The first is a computational issue: For the wearable device to communicate readily with its environment, it must be capable of performing 100 million operations per second. Despite increased computational performance, power consumption must be reduced by several orders of magnitude. The only way to implement this with a battery-powered device embodying billions of transistors is with "nanoscaling," De Man said.

"Unfortunately, the days of 'happy scaling' are over," he added. "Down to 0.5-micron geometries, things got better as they got smaller: More features, more functions, higher clock speeds and even reduced power consumption."

Nanoscale technology will require new approaches. "You have transistors that leak," he said, "and you don't know anymore what is a 'one' and what is a 'zero'." How do you ensure reliability, ease of use and low cost at these dimensions? he said

FinFETs and other transistor structures will likely proliferate in the relentless march toward smaller semiconductor geometries. "The CMOS we have in 2010 will not be like the CMOS we have today," De Man said.

De Man said problems of process scaling, infrastructure and platform development are beyond the means of a single company or research organization. "We need big alliances, worldwide in scale."0