Energy-Aware Synthesis of Embedded Systems on Multiprocessor Platforms

University Of North Carolina At Chapel Hill, Chapel Hill NC

Investigators

Abstract

Project Summary: Energy-aware Synthesis of Embedded Systems on Multiprocessor Platforms Sanjoy Baruah James H. Anderson Two evident trends in the real-time embedded systems domain are (i) the proliferation of multiprocessor platforms, and (ii) the proliferation of applications for which energy consumption must be minimized. In the years ahead, it is reasonable to expect these two trends to converge: indeed, one major advantage of multiprocessor implementations over uniprocessor ones is that the former tend to be more energy-efficient. To date, however, the little work that has been done on energy-aware multiprocessor designs has been limited to relatively simple table-driven scheduling schemes; a precise characterization of the energy savings realized by moving to a multiprocessor implementation is not known. These observations point to a significant need for research to be conducted on the tradeoffs that exist when designing energy-aware multiprocessor systems. This project proposes to fill this need by developing and experimentally evaluating design methodologies and scheduling algorithms that facilitate the synthesis of embedded real-time systems upon multiprocessor platforms, with the primary goal of minimizing energy consumption. Broader impact. Based partially on the results of this project, two new courses will be developed: an introductory graduate course on embedded systems, and a graduate seminar on multiprocessor embedded designs. The introductory embedded-systems course will be created as a follow-on to current real-time systems and advanced operating systems courses. The seminar course on multiprocessor designs will be a follow-on to the introductory course. A concerted effort will be made to involve underrepresented groups in our research. Results from the algorithmic research will be disseminated through publications. Any software of general utility that results from this project will be made publically available on the web.