Stateful Energy Management for Multi-Source Energy Harvesting Transient Computing Systems

• DOI: 10.23919/date56975.2023.10137108
• Publication Date: 2023-06-02T19:32:57Z
• Authors: Sergey Mileiko; Oktay Cetinkaya; Rishad Shafik; Domenico Balsamo
• Keywords: Transient (computer programming)

The intermittent and varying nature of energy harvesting (EH) entails dedicated management with large storage, which is a limiting factor for low-power/cost systems small form factors. Transient computing allows system operations to be performed in the presence power outages by saving state into non-volatile memory (NVM), thereby reducing size this storage. These are often designed task-based strategy, requires storage sized most consuming task. That is, however, not ideal since their tasks have requirements, i.e., operating voltage. Hence, overcome issue, paper proposes novel unit (EMU), tailored multi-source EH transient systems, that selecting voltage next task at run-time, optimizing task-specific needs startup times based on application requirements. For first time literature, we adopted hybrid NVM+VM approach allowing our EMU reliably efficiently retain its internal state, stateful EMU, under even severe conditions. Extensive empirical evaluations validated operation proposed overhead (0.07mJ update <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\simeq 4\mu\mathrm{A}$</tex> static current consumption EMU).