Abstract:

With the advances in VLSI and communication technology, embedded systems have been moving from board-level systems to System-on-Chips (SoC). In order to reduce the design and system cost, the design of embedded systems must move from craft to discipline. In this talk, I will mainly focus on the soft real-time embedded system design and present our recent research results in methodology, low power and quality of service (QoS).

For soft real-time applications, occasional deadline misses or data losses can be tolerated. By leveraging this feature and uncertainties in execution time, I will present a probabilistic design methodology in order to avoid over-designing systems while statistically meeting the performance requirements. As an example, I will present a set of energy reduction techniques by using dynamic voltage scaling (DVS) to exploit the slacks arising from the deadline missing for both single and multiple processor systems while meeting the user required QoS.

DVS has been widely accepted as one of the most efficient energy reduction techniques. For multiple-voltage DVS systems, different voltage settings (the number of voltage levels and the value of each level) may result in very different energy consumption. In order to find the best way to use DVS, I will formulate the voltage set-up problem and provide its practical solutions in order to minimize the system's energy consumption.

The system developed by using the probabilistic design methodology will meet the user required QoS statistically. As the traditional completion ratio metric can only be applied to independent tasks, I will present a new QoS metric that differentiates firm and soft deadlines and captures the task dependency as well. By using multimedia systems as an example, a set of on-line scheduling algorithms have been developed to enhance QoS, particularly for overloaded systems.

Short Bio:

Shaoxiong Hua is a Ph.D. candidate in the Department of Electrical and Computer Engineering at the University of Maryland, College Park, working in the Embedded Systems Research Laboratory. He received his B.S. and M.S. degree in Instrument Science and Engineering from Zhejiang University, China, in 1992 and 1995, respectively. From 1995 to 1998, he was an Assistant Professor with the Department of Scientific Instruments, Zhejiang University. His research interests include embedded and real-time systems, hardware/software co-design, distributed and fault-tolerant computing, and VLSI CAD.