SleepSmart: Smart Mattress Integrated with e-Textiles and

IoT Functions for Sleep Apnea Management

Gozde Cay
Supervised by Kunal Mankodiya
Electrical, Computer & Biomedical Engineering
University of Rhode Island
Kingston, USA
[email protected]

Abstract—Obstructive Sleep Apnea (OSA) is one of the most

serious sleep disorders. People who is suffered from OSA may
not be aware that their upper airway is blocked and they have
difficulty to breathe. For this reason, real-time sleep monitoring
in daily life is important. In response to this need, I aim to develop
an unobtrusive, wireless sleep monitoring system called
SleepSmart. SleepSmart, is a smart mattress topper, which is
integrated with textile pressure sensors and allows people to
monitor their sleeping habits and respiration rate in real-time in
their own bed with its IoT functions. The overall research has
three milestones: (1) designing the mattress topper, (2)
performing signal analysis on the pressure data to extract
respiration rate, and (3) establishing an IoT infrastructure to
provide services. This paper demonstrates the promising ongoing
research results and the goals for future milestones.

Keywords—obstructive sleep apnea, e-textiles, smart textiles,

IoT, embedded system

I. INTRODUCTION
People who have sleep apnea experience shallow breath or
even stop breathing during their sleep. When OSA occurs, the
upper airway is blocked during sleep and people may not be
aware that they have difficulty to breathe [1]. Monitoring of
Figure 1: General Diagram of the System
respiratory, sleep and cardiac parameters are used to diagnose
OSA [2]. In this project, a research to develop an unobtrusive II. RESEARCH GOALS
sleep monitoring system, SleepSmart that is made of smart
textile pressure sensors connected to a wireless embedded The overall system aims three milestones.
system with IoT functions, is pursued. The aim of SleepSmart a) Research Goal 1: Designing textile pressure sensors
is to develop a fabric mattress topper, which is embedded with embedded into a mattress topper: This milestone aims to
textile pressure sensors, which can track sleeping habits and fabricate a smart textile mattress topper integrated with
respiration rate to monitor the probability of OSA. SleepSmart piezoelectric conductive materials to monitor the pressure
topper consists of a fabric sensor grid, resistor networks, and changes.
an embedded system with IoT and wireless capabilities.
b) Research Goal 2: e-Textiles data analytics for OSA
Embedded algorithms will be developed to detect the episodes
of OSA during sleep and also monitor the sleep quality. IoT detection: The signal coming from the pressure sensors will
framework will be established to offer person-centered be analyzed and features of sleep stages and OSA will be
visualization of the sleep quality and OSA data on a tablet app. extracted. Several signal processing methods and machine
learning techniques will be applied.
c) Research Goal 3: SleepSmart IoT infrastructure for
OSA management: To provide services to individuals with
sleep and OSA challenges, a SleepSmart IoT infrastructure
will be established. The Sleep Labs in the region will be
colloborated with for the pilot study.

This research is supported by NSF CAREER CPS#1652538. Gozde Cay

is supported by Turkish National Ministry of Education for her PhD research.

Authorized licensed use limited to: SUNY AT STONY BROOK. Downloaded on August 10,2020 at 14:31:42 UTC from IEEE Xplore. Restrictions apply.
 III. ONGOING RESEARCH
A. Designing Textile Pressure Sensors Embedded Into a
Mattress Topper
1) Cushion Cover Design
A 4x3 textile pressure sensor grid (equal to 12 sensors) was
designed to perform initial testing. The textile pressure sensors
were designed with piezoelectric conductive materials Figure 4: Monitoring Sensor Location and Pressure Changes
responsive to subtle pressure changes such as respiration cycle
or changes in sleeping posture. To monitor the pressure C. Bluetooth Communication
changes, the sensor grid was connected to embedded The data collected from the sensors was sent to a tablet for
computing unit, MSP432 (TI). While each sensor row was displaying and further processing via Bluetooth. The sensor
connected to digital pins to power up the sensors; each sensor grid was connected to Arduino Nano 33 BLE Sense board and
column was connected to analog pins to collect analog pressure its built-in BLE communication protocol was used. The data
data. An embedded computing code was written with Energia coming from the sensors were displayed by LightBlue App in
IDE (Wiring and Arduino framework for Texas Instruments an Android tablet.
MCU boards) to drive the sensors and to collect pressure data.
Then the sensor grid was expanded to be built on a cushion IV. DISCUSSION AND FUTURE WORKS
cover. The pressure sensors were connected to each other with Promising results were achieved from initial tests for
conductive tape. A conductive thread was used to make designing the mattress with textile pressure sensors. However,
connections between sensor array and analog/digital pins. The there are still disadvantages that need to be addressed. The
size of each pressure sensor was 1 sq. inch. As the next step, connection between textile sensors and the electronics is the
the conductive tape was changed with the conductive thread biggest challenge. Also, the repeatability and linearity of the
and the topper was placed on top of a cushion foam, covered sensors are the big issues. To solve those problems, we are
with a bed sheet to simulate the bed. The process is shown in working on designing fully knitted textile pressure sensors. The
Figure 2. sensors will be knitted with fabric sensor and conductive thread
and/or silver plated knitted jersey, therefore they can be easily
sewed onto cushion cover or any other fabric.
For the Research Goal 2, designing a printed circuit board
(PCB) which includes all electronics by using their main chips
is aimed. Thus, the system can be controlled from a small
embedded system and connected to the environment via
Figure 2: The Design Process of Cushion Cover Bluetooth communication for displaying. In addition, an
embedded algorithm will be developed for further signal
B. Monitoring the Pressure Changes and Sensor Location processing and machine learning.
To monitor the pressure changes, the data coming from For wireless communication and real-time monitoring
sensors were stored into a .csv file and was plotted from that (Research Goal 3), a custom Android app will be developed to
.csv file on Matlab. The pressure was applied using bare fingers show the user interface, collect the data and store it in .csv file.
tapping on to the sensors. Figure 3 shows the pressure changes Moreover, to plot the data MQTT communication protocol will
when the finger was tapped. be used. A custom Python script will be written and run to take
the data from sensors and plot the data in the client computer.
Thus, we will be able to monitor the pressure changes in real-
time.
ACKNOWLEDGMENT
This research is supported by NSF CAREER
CPS#1652538. Gozde Cay also thanks Turkish Ministry of
National Education for their financial support, Kunal
Mankodiya for his supervision and Nick Constant, Shimra J.
Fine, Vanessa Kamara and Michaela Bellisle for their valuable
help.
Figure 3: Monitoring the pressure changes REFERENCES
[1] “Obstructive Sleep Apnea Explained”, n.d. Accessed on: Nov. 5, 2019
The system was also programmed with Processing IDE to [Online]. Available: https://www.webmd.com/sleep-
show the sensor location on graphical user interface (GUI). disorders/guide/understanding-obstructive-sleep-apnea-syndrome#1
According to the pressure changes coming from MSP432, the [2] L. Spicuzza, D. Caruso, G. Di Maria, "Obstructive Sleep Apnea Syndrome
GUI showed the location of sensor and changed its color from and Its Management" in Therapeutic Advances in Chronic Disease, vol. 6, no.
white to black. The results is shown in Figure 4. 5, pp. 273-285, September 2015. doi: 10.1177/204062231559031

Authorized licensed use limited to: SUNY AT STONY BROOK. Downloaded on August 10,2020 at 14:31:42 UTC from IEEE Xplore. Restrictions apply.