Electrical Engineering and Computer Science

Dissertation Defense

Implementations of Low-power μProcessor System for Miniaturized IoT Applications

Jeongsup Lee
WHERE:
Remote/Virtual
SHARE:

Miniaturized sensing devices enable the IoT for many practical applications. Due to the small form factor, these types of devices have limited battery size, and therefore efficient energy consumption is a key requirement. However, since they are used in a wide range of operating conditions that vary across time, it is challenging for them to achieve energy efficiency.

In this context, this dissertation focuses on achieving energy-efficiency in miniaturized sensors. The first part presents a dynamic power management technique for the μprocessor, which enables on-chip closed-loop minimum-energy-point tracking and hence guarantees energy-optimal operation at all times. The next part presents a µprocessor system designed for high temperature applications. It features a deep sleep-mode that allows the complete system to retain full 16-kB SRAM contents with 0.54µW at 125˚C, which is 26× lower than the baseline design. The last two parts present essential sub-blocks required for low-power μprocessor systems: a wide-range level converter and a switched-capacitor DC-DC converter. The proposed level converter offers robust operation across a wide range of low and high supply voltages as well as PVT variations. The switched-capacitor DC-DC converters are automatically generated from the given input specifications. Based on the theoretical analyses, the proposed DC-DC generation tool directly finds the optimal design parameters.

Chair: Professor Dennis Sylvester