CAREER: Enabling Low-Cost, Maintenance-Free, Long-Term Embedded Sensing Without Batteries

点击翻译按钮获取中文摘要

Unlimited, stable, and reliable power is an assumption upon which most traditional computing systems and abstractions are designed. This project explores new techniques and tools for developing sophisticated sensing applications that do not rely on batteries and are powered by a variety of environmental energy sources. Replacing batteries with capacitors has many advantages for system designers. Capacitors have smaller size, lower cost, and the potential for deployments lasting decades longer than batteries - which wear out after two to five years. But even with energy harvesting advances, current capacitor-based devices are difficult to program, test, and deploy due to fluctuating energy supplies, limited energy storage, and frequent loss of power. This project will explore and prototype hardware techniques, software systems, programming languages, and tools that will enable application designers to quickly prototype sophisticated applications, focusing on application goals instead of avoiding and recovering from power failures. This project focuses on a research agenda to: 1) design and develop a novel computing platform specifically for transiently-powered sensing devices; 2) prototype techniques for managing harvested energy in hardware in order to simplify software design; 3) develop a new programming model that allows programmers to represent applications in a simple disruption-tolerant form; 4) develop runtime techniques that schedule computing tasks in order to work effectively towards achieving application goals in spite of unpredictable power outages; 5) create novel tools that help developers predict how their transiently-powered applications will perform when deployed; and 6) evaluate these contributions in the context of real pilot deployments. The project, which includes broad k-12 outreach, will provide experiences conducting innovative research and working with biologists and other domain scientists for graduate and undergraduate students around the development of devices that provide useful data for decades in spite of frequent power failures while providing.