Universities played a key role in the development of such technologies as semiconductors, the Internet and distributed computing. Yet today, they find themselves unable to afford tools and equipment, while industries have to retrain graduates, said Steve Liu, associate professor of computer science at Texas A&M University.

Liu will tackle this dilemma at a panel he's moderating on Thursday that calls for a "win-win" partnership between industry and academia. The panel itself is an exercise in industry-academia cooperation; it's a joint activity of the Embedded Systems Conference Silicon Valley and the associated, research-oriented IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS 2006).

"In the 1970s universities led the industry, but these days technology is so complicated that it becomes quite expensive for universities to acquire equipment," Liu said. "Even textbooks are a major burden for teachers and students." While industry complains that it has to retrain students, "universities are frustrated because we can't get up-to-date technology," Liu said.

Liu himself has worked extensively with industry, including ARM, Xilinx, Microsoft, Sun Microsystems and National Instruments. Such companies have helped with financial, hardware or software support for Liu's classes--and that has led to an unusual approach to university instruction.

"We've stopped using textbooks," Liu said. "We use data sheets directly. And we require students to have open projects, rather than some cookie-cutter type of thing." Liu's classes thus require students to undertake research projects, such as implementing an embedded system in an FPGA for a hypothetical target system. If the project is good, Liu will reimburse the student from his own research budget.

"The students don't like it, because they are used to the comfort zone of textbooks, homework and quizzes," Liu said. "But in my class, it is not so. You have to come up with a project on your own."

Liu hopes the panel discussion will dispel stereotypes that exist on both sides. "Some think academia couldn't care less about the real world, and some think companies don't care about research. "I want to break the ice, to help everybody understand better."

One panelist from the industry side is Greg Bollella, distinguished engineer and director of real-time Java strategy at Sun Microsystems Inc. Bollella noted that he has strong partnerships with the Center for Efficiency Oriented Languages in Cork, Ireland, and the Real-Time Garbage Collection group at Lund University in Sweden. Sun supports university research financially, so that university groups can research topics that might interest Sun, he said.

"You have to have the correct expectations," Bollella said. "You're not going to ask an academic department to produce product-level-quality stuff on a schedule." Meanwhile, he said, it's time for the practitioner community to start looking at academic concepts, such as real-time scheduling--and time for academia to go beyond raw math and provide some useful application programming interfaces and implementations.

Also on the panel is Douglas Stuart, software engineer at Boeing Co. and co-chair of RTAS 2006. "I think we've had some successes in working with academia, and there are also some challenges," he said. "Universities are looking toward their own research agendas, and sometimes they don't align themselves [with us] as well as we'd like."

Now in its 12th year, RTAS (see www.rtas.org) is co-located with ESC Silicon Valley for the second year in a row. The shared locale is an attempt to provide "outreach" to the practitioner community, Stuart said. RTAS typically attracts an international audience of 100 to 200 people, with papers from both industry and academia, he noted.

This year, Stuart said, RTAS has four areas of emphasis: real-time and embedded-systems theory, industrial applications, development tools and hardware/software co-design. RTAS 2006 opens tomorrow with tutorials and workshops. The technical program begins on Wednesday with a keynote by Douglas Schmidt, professor of computer science and engineering at Vanderbilt University, on the challenges of ultralarge-scale real-time systems.

Win-Win Partnership of Academia and Industry:

Why should we care? Where is our common future?

Location:
Fairmont Hotel, San Jose, Calif.

Moderator:
Steve Liu, associate professor of computer science, Texas A&M University

Panelists:
Frank Mueller, associate professor of computer science, North Carolina State University

Douglas Schmidt, professor of computer science and engineering, Vanderbilt University

Helen Gill, program manager, Computer System Research Program, National Science Foundation

Greg Bollella, distinguished engineer, Sun Microsystems Inc.

Douglas Stuart, software engineer, Boeing Co.