Battery lifetime optimization for AI @ Edge devices

Optimizing energy consumption of Artificial Intelligence of Things (AIoT) devices is mandatory and challenging. Energy Harvesting (EH) from RF, solar, thermal, wind, and kinetic energy sources can be a good substitute for traditional batteries. EH is expected to have abundant applications in future AIoT and self-powered micro-systems such as wireless sensor networks and wearable devices. The combination of different sources of energy with harvesting capabilities is considered as a viable option to develop autonomous or AIoT devices with reduced dependency on batteries. Since the output voltage of the different energy harvesting mechanisms varies significantly depending on the strength of the ambient power source, they need to be converted into a suitable and stable voltage level using a power conditioning circuit. However, power conditioning of the harvested energy is very challenging. On the other hand, stringent requirements for low-power intelligent devices at the edge require the study and development of Artificial Intelligence (AI) algorithms and suitable hardware support to mitigate the computational demand of these systems.
In this research proposal, our main objective is to develop autonomous devices featuring AI algorithms with reduced battery dependence.

Faculty Supervisor:

Yvon Savaria

Student:

Partner:

Intégration Dolphin Inc

Discipline:

Engineering

Sector:

Manufacturing

University:

Polytechnique Montréal

Program: