Ultra-Low Power Multi-Modal Sensor Interface Circuits and Systems for Personalized Physiological Monitoring

Author: ORCID icon orcid.org/0000-0003-3243-6362
Wang, Peng, Electrical Engineering - School of Engineering and Applied Science, University of Virginia
Advisor:
Calhoun, Benton, EN-Elec/Computer Engr Dept, University of Virginia
Abstract:

Personal healthcare devices are developing towards multi-parameter physiological tracking for a more comprehensive assessment of users’ health conditions. Based on two critical physiological signals, Photoplethysmography (PPG) and Electrocardiography (ECG), various vital parameters, such as heart rate, blood oxygen saturation (SpO2), and blood pressure, are derivable. The emerging energy harvesting technologies can potentially enable self-sustainable, free-of-interruption healthcare devices that track users’ health parameters continuously. However, the available power density (∼20μW/cm2) from the ambient environment does not match the high power consumption of the existing physiological sensing devices. Thus, the pursuit of a fully self-powered and sustainable sensing system requires reducing the load power consumption, especially the PPG analog front-end (AFE) and LED power which currently limits the overall system. In addition, the user’s biological characteristics, such as skin tone, vary the power and performance of the physiological sensing hardware. How to maintain a reasonable sensing performance without consuming excessive power remains an open question.

This thesis makes contributions that address the challenges stated above. First, an end-to-end PPG sensor interface model provides an analytical solution for realizing < 20μW AFE and LED total power across all skin tone types when measuring PPG at finger. The model points out that lowering the LED current, AFE bias current and supply voltage can greatly reduce the AFE and LED power without performance degradation. Second, a PPG-sensing AFE implements the optimized design choices from the model and shows competitive noise and DC offset cancellation resolution on silicon. Measurement results from a 65nm CMOS test chip demonstrate 532nW AFE power and 10.3μWtotal power, which are 37x and 5-40x lower than prior work, respectively. Third, a highly-flexible PPG AFE design shows reconfigurable sensing operation for monitoring tri-modal parameters: heart rate, SpO2, and pulse transit time (PTT). The programmable-gain transimpedance amplifier along with the flexible LED drivers greatly extends the operating space for fitting the user’s biological variance and sensing location variance. While a 65nm CMOS test chip demonstrates a minimum 7.7μW total power for sensing heart rate, it also shows 18.9μW and 43.7μW total power for the SpO2 and PTT sensing, respectively, which are event-triggered modalities for assisting the primary heart rate tracking. Fourth, a system calibration approach based on the flexible AFE chip described earlier and a custom microcontroller chip provides a solution for automatically matching the hardware operating point with the user’s biological characteristics. System measurement results demonstrate that the proposed approach can sort out the optimized operating point for different users and sensing locations. Finally, two versions of ECG AFE designs provide 165nW and 3nW heart rate sensing solutions. Fabricated on the same die with the PPG AFEs, they serve as alternative sensing options for enhancing the system availability under extremely poor harvesting conditions.

Degree:
PHD (Doctor of Philosophy)
Keywords:
photoplethysmography (PPG), electrocardiography (ECG), analog front end, sensor Interface, personalized healthcare, ultra-low power
Sponsoring Agency:
NSF NERC ASSIST Center (EEC-1160483)
Language:
English
Rights:
All rights reserved (no additional license for public reuse)
Issued Date:
2021/12/12