---

Presenter: Samanvay Upadhyay

Faculty Sponsor: Jeremy Gummeson

School: UMass Amherst

Research Area: Electrical and Computer Engineering

Session: Poster Session 4, 2:15 PM - 3:00 PM, Auditorium, A19

ABSTRACT

The goal of this thesis is to examine a hybrid energy harvesting system for battery-free wearable devices used for health and environmental monitoring. Three Python-based simulation pipelines have been developed for photovoltaic (PV), thermoelectric (TEG), and kinetic (KEH) EH methods, as well as an energy-aware scheduling model to mimic realistic conditions for testing different sensing and transmission methods utilizing a small capacitor as a buffer for energy storage. Publicly accessible datasets have been used in conjunction with transient and steady-state models as previously discussed to determine session-level power consumption as well as employing a bootstrap distribution to test scenarios. An analysis of different scheduling methods such as periodic, threshold, opportunistic, and predictive scheduling was carried out in terms of device availability, quality of data, and minimum buffer sizes necessary for energy storage. Simulation results show that the hybrid EH system is able to generate varying amounts of power for battery-free wearable health monitoring devices. The KEH generates short pulses of high power levels (4.8 mW average), thermoelectric EH generates a constant level of power (1-4 μW), and photovoltaic EH generates 10-30 mW levels of power outside, whereas it is negligible inside. The amount of power being harvested is in line with expectations for each EH method (29.16 mW for outdoor exercise, 13.03 mW for KEH alone, and 0.03 mW for resting TEG alone). In summary, the hybrid EH system is able to accommodate a variety of sampling methods while reducing buffer sizes significantly.