Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier

https://orcid.org/0000-0002-7373-6655

AccessType

Open Access Dissertation

Document Type

dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Electrical and Computer Engineering

Year Degree Awarded

2022

Month Degree Awarded

May

First Advisor

Deepak Ganesan

Second Advisor

Daniel Holcomb

Third Advisor

Trisha Andrew

Fourth Advisor

Yeonsik Noh

Fifth Advisor

Jeremy Gummeson

Abstract

An exciting new direction for ubiquitous computing that has emerged in recent years is the ability to leverage functionalized fabrics as sensing elements, thereby potentially converting any textile that we wear into a sensing substrate. There are three reasons for this trend: first, the familiar feel of fabrics as a material we interact with daily makes it a natural surface for enabling continuous and unobtrusive sensing, second, the flexible nature of fabrics which makes them suitable for applications where the use of rigid components are not suitable, and third, the relatively large real estate on a textile makes it possible to use multiple sensors in the form of a sensor array, thereby improving our ability to sense on a larger scale.

Textile sensors can be useful for a range of human-centered applications including monitoring of physical activities, physiological signals, and sleep patterns using everyday sleepwear. While rigid electronics sensors require tight contact with body to reliably gather information, textile sensors can transform a loosely worn shirt into a robust tool for physiological and activity sensing. Additionally, fabric-based sensors can be used as sensing elements for various other applications as well: smart toys, bed sheets, occupancy sensing carpets, face and sleep masks, and many other human-centered applications.

However, sensing with loose-fitting textile sensors present several challenges in terms of signal quality, noise absorption, and robustness to environmental changes. In contrast to rigid electronic sensors, the quality of the absorbed signal for textile sensors depends on dynamic characteristics of the textile sensor such as whether it is flat or folded as well as the base pressure applied on the sensor. In addition, high impedance and relative large surface area of the fabric-based sensors result in greater electromagnetic noise that need to be dealt with. Finally, textile sensors are much more sensitive to environmental changes such as humidity, temperature, pressure, and motion, all of which affect the signal and effectively, reducing signal to noise ratio compared to conventional electronics sensors.

On the other hand, the availability of real estate and abundance of sensing locations surrounding fabrics in everyday life, enables the potential of using numerous fabric sensors to form an array of sensors to detect physical changes in the environment. However, as the number of sensors increases, so does the challenges we face in terms of engineering and signal processing while finding a balance between power consumption and accuracy. This optimization gets more complicated as we consider transmission of raw data versus embedded processing.

The focus of this thesis is on addressing these challenges and dealing with low signal to noise ratio to translate the stream of data coming from multiple textile-based channels into actionable insights. My work has three key contributions. First, I explored the possibility of using a triboelectric textile sensor to monitor joint dynamics in a loosely-worn shirt. Second, I designed loose-fitting sleepwear to track physiological signals as well as sleep posture by taking advantage of the relative large surface area of sleepwear to place multiple textile sensor to opportunistically sense cardio-respiratory ballistics. To enable the above two solutions, I developed several signal processing techniques to transform the noisy streams of data and fuse the information from sensors to obtain physiological parameters, including heart rate, respiration rate, and sleeping posture and movement parameters including elbow/knee flexion and extension and joint angular velocity. Third, I used textile sensors in monitoring interactions, in particular, recognizing fine-grained interactions with objects, such as toys. Interactive toys are the most important tools in improving children's cognitive abilities. In this work, I studied the challenges arise when dealing with large number of fabric sensors in close proximity and proposed solutions and proved their validity by means of extensive evaluations. The focus of this work is on optimizations in hardware and software to minimize power consumption for longer battery lifetime. I show that we can robustly sense a child's interaction with toys by using textiles as sensors and provide real-time feedback by processing the raw data either locally or by offloading and processing it in a more powerful nearby device.

DOI

https://doi.org/10.7275/28184621.0

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 4.0 License.

Share

COinS