CAREER: System Support for Renewable Energy-driven Devices

University Of Arkansas, Fayetteville AR

Investigators

Abstract

Given the rate of fossil fuel depletion, alternative energy driven systems are poised to determine our planet's future. However, there is still a lack of computer systems that can intelligently use harvested energy for a wide range of applications. To fill this void, this CAREER research is developing techniques for designing renewable energy systems spanning small devices (powered by indoor light), disaster-relief sensors (solar powered), and large systems such as green homes. This project meets the above goal through a four fold research agenda. First, to better understand the variability in energy harvested this research uses low power side channel measurement infrastructure to profile energy generation and compression techniques to determine appropriate model granularity for harvesters. Second, this project uses multi-tiered software and hardware architectures that can function under variable power inputs. Third, to tame the uncertainty associated with energy harvested, this research is designing a multi-scale self-adaptive optimization framework tuned to the computational capability of the associated device. Finally, this research is designing replayer systems that is used to playback renewable energy traces for rapid development of such systems in the lab. With target applications such as healthcare, emergency relief, and green homes, this CAREER research has profound societal impact. Additionally, the project meets other broader goals, including undergraduate research and education and maintaining student diversity in Arkansas. Through Green Computing courses, this research is instrumental in percolating renewable energy device design into the graduate and undergraduate curricula at the University of Arkansas, and through distance education to the rest of the world. Also through support from industrial partners, the research prototypes are being made publicly accessible.