Artificial Intelligence Edge Computing Technology Based on Energy Harvesting

Overview

This project, supported by the Basic Public Welfare Research Programme of Zhejiang Province, focuses on future green and autonomous AIoT devices to achieve sustainable DNN reasoning in energy harvesting (EH) systems.

Methodology and Results

Experimental results show that our approach improves the persistence and DNN inference quality of the energy harvesting system, achieving a 15.22% reduction in energy consumption for a single DNN inference, while comparing favourably with existing state-of-the-art methods in terms of accuracy and latency.