PhD Dissertation Proposal: Charles Cao, Ultrasonic Sensing Systems: From Hardware Design to Interactive and Physiological Applications
Content
Speaker:
Abstract:
This research investigates the design and implementation of ultrasonic sensing systems that allow everyday devices to perceive and interpret signals beyond the audible spectrum. By integrating advancements in hardware design, firmware optimization, signal processing, and machine learning, the work advances ultrasonic sensing from foundational system architectures to high-level applications in interaction and human physiology.
The first part presents a six-layer HDI PCB designed for a battery-powered wearable device featuring an 8-microphone array. The accompanying firmware supports real-time ultrasonic beamforming and signal analysis, enabling high-resolution spatial sensing within a compact and energy-efficient form factor.
Next, the thesis examines how commodity smartphones can be repurposed to support enhanced ultrasonic sensing. Through modifications to the Android audio stack that increase system sampling rates, the platform achieves improved accuracy and unlocks new sensing applications—using only the device’s built-in microphones and speakers.
Extending the exploration of consumer audio platforms, a multi-modality fusion model is developed for headphones, combining audible and ultrasonic features. This model enables whisper-level and silent-speech keyword detection, providing a privacy-preserving and noise-robust interface that broadens the possibilities for future auditory computing devices.
Finally, the research introduces a through-skin ultrasonic sensing system implemented on smartphones, capable of detecting subsurface physiological signals such as blood flow. This demonstrates the feasibility of transforming mobile devices into accessible, non-invasive biomedical sensors.
Together, these projects establish a unified framework for ultrasonic sensing systems, bridging hardware design, software adaptation, and intelligent signal processing to support both interactive and physiological applications.
Advisors:
Jeremy Gummeson and Jie Xiong