Intermittent computing is an emerging computational model that allows software to operate reliably on devices that harvest energy from their environment. Energy-harvesting frees devices from the need for batteries, battery replacements, tethered power, and regular maintenance, enabling deployment to far remote installations, such as in civil infrastructure and outer space. This does not come for free: power failures impede progress and can leave a system’s memory and execution state inconsistent. Unpredictable future energy availability, and the difficulty of precisely characterizing device power consumption makes building a reliable intermittent system a challenge. I describe our experience with intermittent computing. I describe our recent efforts to mathematically formalize the behavior of intermittent software execution, and lessons learned from these formal modeling efforts. I discuss concrete incarnations of our intermittent and energy-harvesting computing results: a terrestrial, long-range batteryless camera system with the ability to do on-device machine learning and transmit over 10km distances with no batteries and two tiny, intermittent computing nanosatellites that sense, compute on sensor data, and communicate with Earth.
Brandon is a Professor at Carnegie Mellon University working on hardware and software computer systems to answer the question: “How capable can a computing system be, given physical restrictions on its form factor, input power, energy storage, or other resource constraints?” His research has been recognized with 8 best paper awards, the 2019 IEEE TCCA Young Computer Architect Award and the 2015 Bell Labs Prize.
Thu 19 Nov Times are displayed in time zone: Central Time (US & Canada) change
|11:00 - 11:40|
Brandon LuciaCarnegie Mellon University