Nurani Saoda is a PhD candidate in the Department of Electrical and Computer Engineering at the University of Virginia. Her research is broadly focused on ubiquitous sensing using energy-harvesting sensors with an emphasis on sustainable computing. She designs hardware-software co-designed techniques to make applications reconfigurable, robust, and resilient when powered from harvested energy. Her work has been published in top sensing conferences and workshops including ACM MobiCom, IPSN, SenSys, and BuildSys.
Ubiquitous computing remarkably extends the reach of sensing by pushing computation into physical objects with multifarious forms. The result is sensing at a tremendous scale to enable industrial control, building management, citywide environmental monitoring, and mass inhabitant (i.e, wildlife and marine life) tracking. The energy and resource demands of these computers are lifelong and dynamic. Yet, the batteries powering them are short-lived and the hardware running them is rigid and unmodifiable. At scale and over long timeperiods, these factors significantly compromise the utility of Internet-of-Things (IoT) applications as batteries exhaust frequently and devices become obsolete. This contributes to two major sources of nondisposable components: expired batteries and obsolete electronics, which cumulatively contribute to millions of tons of harmful e-waste each year. My research addresses these challenges by realizing edge computers that do not fall short of energy (perpetual), support low-cost and low overhead sensing on repurposable components (ubiquitous), and allow reconfigurable functionality (sustainable). One key enabling factor of my research is to critically rethink the role of an energy-harvesting power supply in a sensor to truly enable sustainable computing. In my work, I proposed an architecture where the energy-harvesting power supply functions as an energy management “co-processor" that decouples managing energy from application execution both logically and physically. This lowers the barrier for new devices to utilize energy-harvesting for sustainable computing. For existing devices that require batteries, I leverage the co-processor architecture to transparently upgrade devices to energy-harvesting, while simultaneously enhancing the device’s functionality. Further, I demonstrated a technique to enable device reconfigurability by adding new sensing capabilities to the energy-harvester itself. The outcome is a class of self-powered green edge computers that introduce novel modular, reusable, and reconfigurable design points for sustainable IoT.