CRII: CSR: Systems and Tooling Enabling Adaptive Intermittent Computing

Northwestern University, Evanston IL

Investigators

Abstract

Tiny energy harvesting computers tap into the sun, radio waves, vibration, and other sources to power all their sensing, computing, and communication tasks. By leaving batteries behind, these devices can function maintenance-free, potentially for decades, enabling numerous applications in the Internet-of-Things. These battery-free devices can lose power at any point because of fluctuations in energy availability, which makes programming, debugging, and deployment challenging. The goal of this project is to develop the systems and tools that make these devices resilient to power failures, dependable, easier to program, and able to function in unpredictable environmental and energy conditions. The intellectual merits of this project are focused on improving adaptivity, programmability, reliability, and applicability of intermittent computing devices. Three complementary research thrusts are pursued: 1) development of a rigorous classification of energy harvesting environments built on identifying unique energy modes, aiding in application design; 2) prototyping a novel recognition circuit for anticipating changes in the energy harvesting environment or energy mode based on remanence decay; 3) development of a task-based programming language with built-in adaptation semantics leveraging energy modes and the recognition circuitry with accompanying low overhead runtime system. The work in this project develops adaptive systems and tooling for intermittent computers that will enable even novices to build reliable and long-lived sensor devices. By making application development easier, and improving availability (and reliability), these systems will positively impact applications in healthcare (wearable and body sensor networks), defense, ecology, horticulture, infrastructure, wildlife tracking, space exploration, and many other areas where long-term, low cost, massive scale sensing is essential. All hardware, software, and collected environmental profiles and energy traces will be made freely available, along with documentation and tutorials. These project artifacts will be stored in a dedicated repository for scientists, educators, hobbyists, and industry to access and use. This repository will be maintained for at least three years after the duration of the project. (https://gitlab.com/ka-moamoa/battery-free-sensing) This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.