Home > News > Building Change into Real-Time Computing

Building Change into Real-Time Computing

From UVA Explorations
April 17, 2002

For Jack Stankovic, solving a difficult problem in computer science is like picking a lock. It’s not until you have aligned and ordered your assumptions about the problem that the solution can be released. “You start by selecting a problem that seems interesting and important,” he says. “The next step is to think about your assumptions to give you a way into the problem that will reveal a solution. This is where the creative leap occurs.”

Stankovic, the BP America Professor and chair of the Department of Computer Science, has been highly successful at making just this kind of creative leap. He is an internationally respected specialist in real-time computing, building machines that interact with their environment within very specific time constraints.

Real-time systems are everywhere, from nuclear power plant control and medical monitoring to antilock braking and burglar alarms. An example of a hard real-time system is a digital fly-by-wire system in an airplane; catastrophes can occur when such systems fail. A (fee-based) Web service is an example of a soft real-time system. If the service doesn’t perform in a timely manner, the worst that can happen is that you’ll gradually lose revenue and repeat customers.

One of Stankovic’s interests is creating real-time computer operating systems. “Traditionally, real-time operating systems had been built to work in well-defined environments,” he says. When the environment becomes complex and highly dynamic, this approach becomes both unwieldy and unworkable, especially if real-time interaction is your goal.

Stankovic challenged the assumption that real-time operating systems must be independent of specific information about the application. Instead, he built a reflective operating system that could incorporate changing information and application semantics. “Other people had begun to move in this direction, but I saw that this principle should be at the heart of any dynamic real-time system,”he says.

The tools that Stankovic uses to unlock solutions like these are inherently collaborative.“When I approach a problem, I look carefully at the ways other investigators have approached similar challenges,” he notes. “I read papers, identify their underlying assumptions, and see how they map to the problem I’m working on. Then, many times, by challenging key, almost implicitly accepted assumptions, a better solution emerges.”

He also finds it helpful to get out of the office and get together with a colleague or two to test ideas and brainstorm. “In a small group, you tend to listen to each other more intently,” he observes. “If you do this well, the excitement builds as you exchange ideas.”

Finally, Stankovic makes it a matter of intellectual discipline to try new approaches. “It’s all too easy to be a perfectionist and to focus on making incremental changes,” he says. “I try to encourage my students to invent, rather than to refine.”

Original Article | Local Copy


More news about John A Stankovic


Return To List