Power Optimization of a Continuous Glucose Monitoring (CGM) System

August 2024 – May 2025

Problem

CGM battery life lasted approximatley 3 hours. This is not sufficient for individuals who must monitor their glucose levels at various points throughout the day without re-charging the battery every few hours.

Solution

Power the implantable glucose sensor on a lower duty cycle to reduce power consumption and extend the life of the system. However, switching of the subcutaneous electrochemical sensor can affect sensor signal output without a proper stabilization period. To meet specifications, the signal output had to be within a 5% margin of error.

Contributions

Formulated various pulse width modulation (PWM) algorithms in C++ to test operatation of the system at various lower duty cycles.
Formulated a C++ algorithm to extract necessary parameters to process sensor output.
Built a custom breakout PCB to enable repeatable hardware-software integration testing.
Created a multi-modal data acquistion system to acquire raw sensor data, process and vizualise preformance in real time.
Designed and executed verification tests to ensure both device and measurement system consistancy.

Outcome

While all duty cycle variations were under the 5% margin of error, lower duty cycles led to higher error most likely due to insufficient time for the sensor to stabilize. Lower duty cycles extended battery on average 75%, while higher duty cycles extended battery life around 15%.

Figures

Figure 1. Energy consumption calculations per hour for each duty cycle tested. Lower duty cycle consumed less energy.

Duty cycle vs battery life and percent error. — Figure 2. Decreasing duty cycle significantly extends battery life (days), but generally increases percent error.

Figure 3. Verifying system sensor output for each pulse width modulation algorithm.