|
PHILIP KOOPMAN
Associate Professor
Electrical and Computer Engineering and Computer Science
 www
Distributed Embedded Systems:
The vast majority of the billions of processors manufactured yearly
are used for embedded applications rather than desktop computing.
Increasingly, these embedded processors are being incorporated into "smart" sensors
and actuators, and are connected over a real-time network to form
distributed embedded systems. Such systems have significantly different
requirements and trade-offs than conventional computing systems,
encompassing the areas of interdisciplinary design optimization,
ultra-high dependability, very low cost, real-time performance, safety,
low power consumption, and extended duration life-cycle support.
Many of the needs of distributed embedded systems can be met via the
use of inherently robust architectures. Recognizing that even the highest
quality building block components cannot be made failure-free, robust
architectures incorporate techniques such as graceful degradation,
automatic incorporation of upgrades, and acceptance of non-exact replacement
parts. The result can be a system which exhibits highly dependable
operation in real-world scenarios over a profitable and long product
life-cycle.
Professor Koopman's areas of interest include teaching distributed
embedded system design techniques, developing quantitative system robustness
measurement tools, creating inherently robust architectural frameworks,
and exploring the fundamental issues involved in understanding dependable
distributed embedded systems. His current research effort is the RoSES
project on self-configuring embedded systems that degrade gracefully
when components fail. He has recently completed work on the Ballista
project to measure and improve software robustness.
|
