See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/310821467

An IoT-cloud Based Wearable ECG Monitoring System for Smart Healthcare

Article  in  Journal of Medical Systems · December 2016

DOI: 10.1007/s10916-016-0644-9

CITATIONS READS
67 5,688

5 authors, including:

Zhe Yang Qihao Zhou

Beijing University of Posts and Telecommunications Beijing University of Posts and Telecommunications
8 PUBLICATIONS   371 CITATIONS    3 PUBLICATIONS   67 CITATIONS   

SEE PROFILE SEE PROFILE

Lei Lei Kan Zheng

James Cook University Brisbane Beijing University of Posts and Telecommunications
102 PUBLICATIONS   1,951 CITATIONS    284 PUBLICATIONS   3,721 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Open Source X-LoRa View project

IoT and Healthcare - improving efficiency and reducing wait times View project

All content following this page was uploaded by Kan Zheng on 06 February 2018.

The user has requested enhancement of the downloaded file.

 An IoT-cloud Based Wearable ECG Monitoring System
for Smart Healthcare

Zhe Yang∗ , Qihao Zhou∗ , Lei Lei‡ , Member, IEEE, Kan Zheng∗ , Senior Member, IEEE, Wei Xiang† ,
Senior Member, IEEE
∗
Intelligent Computing and Communication (IC2 ) Lab
Key Lab of Universal Wireless Communications, Ministry of Education
Beijing University of Posts & Telecommunications
Beijing, China, 100088
‡
State Key Laboratory of Rail Traffic Control and Safety
Beijing Jiaotong University
Beijing, China
†
College of Science and Engineering
James Cook University
Cairns, QLD 4870, Australia
Contact email: zkan@bupt.edu.cn

Abstract

Public healthcare has been paid an increasing attention given the exponential growth human population and
medical expenses. It is well known that an effective health monitoring system can detect abnormalities
of health conditions in time and make diagnoses according to the gleaned data. As a vital approach
to diagnose heart diseases, ECG monitoring is widely studied and applied. However, nearly all existing
portable ECG monitoring systems cannot work without a mobile application, which is responsible for data
collection and display. In this paper, we propose a new method for ECG monitoring based on Internet-
of-Things (IoT) techniques. ECG data are gathered using a wearable monitoring node and are transmitted
directly to the IoT cloud using Wi-Fi. Both the HTTP and MQTT protocols are employed in the IoT cloud
in order to provide visual and timely ECG data to users. Nearly all smart terminals with a web browser
can acquire ECG data conveniently, which has greatly alleviated the cross-platform issue. Experiments
are carried out on healthy volunteers in order to verify the reliability of the entire system. Experimental
results reveal that the proposed system is reliable in collecting and displaying real-time ECG data, which
can aid in the primary diagnosis of certain heart diseases.

Index Terms– Health monitoring, ECG, wearable sensors, IoT cloud.

 2

I. I NTRODUCTION

With a rapid growth in human population and medical expenditure, healthcare has become one of
most significant issues for both individuals and governments. Meanwhile, according to a report from the

World Health Organization (WHO) [1], the problem of population aging is becoming more serious. Health
conditions of aged people usually need to be checked more frequently, which poses a greater challenge to
existing medical systems. Therefore, how to identify human diseases in a timely and accurate manner with

low costs has been paid an increasing attention. Due to the dominance in the diagnosis of heart-related
diseases, electrocardiogram (ECG) monitoring has been widely applied in both hospitals and medical
research [2].

Traditionally, the ECG is detected through large and stationary equipment in professional medical
institutions. The kind of equipment usually employs twelve electrodes to collect ECG data due to their

good performance in short-term measuring. However, the equipment is unlikely to be portable, which
means that patients’ activities are severely limited during the period of data collection. Moreover, as
these devices are usually too expensive for home use, patients have to go to hospital frequently, which

will inevitably increase the burden of hospitals. Therefore, a portable system for a long-term ECG signal
detection with low costs is highly desired.

Thanks to the development of mobile Internet and wireless sensor networks (WSNs) [22], wearable
ECG monitoring systems have emerged which are able to detect ECG signals using a non-intrusive sensor
and transmit the signal to the smart phone through wireless transmission techniques, such as Bluetooth

or Zigbee [3] [4] [14] [18]. For the sake of portability, electrodes of the WSN-based ECG monitoring
systems are usually less than traditional methods. At the expense of accuracy, it is sufficient to collect the

basic information of the heart. These portable sensors are usually embedded into some wearable textiles,
which have little impact on the user’s daily activities. With the aid of these systems, long-term ECG
can be monitored in a cost-effective manner. However, to the best of our knowledge, nearly all existing

systems cannot work without a smart phone, which is used as a receiver and processor of the ECG data
[5] [6]. Due to limited power and computational capabilities, the complex tasks of data transportation and
processing may have a great impact on the daily use of the smart phone. Furthermore, in order to support

all the OS platforms of smart terminals, great efforts are required for the cross-platform development of
 3

the mobile application.

In this paper, the architecture of an ECG monitoring system based on the Internet-of-Things (IoT) cloud
is firstly proposed. Based on this architecture, we design and implement a wearable ECG monitoring

system. The ECG data gathered from the human body will be transmitted directly to the IoT cloud using
Wi-Fi without the need of a mobile terminal. Compared with Bluetooth or Zigbee, Wi-Fi can provide
higher data rates and wider coverage areas. In order to provide convenient and timely access to ECG

data for users, both the HTTP and MQTT servers are deployed in the IoT cloud. The gathered data are
stored in a non-relational database, i.e., Redis, which can greatly improve the speed and flexibility of data

storage. A web-based graphical user interface is implemented so that it provides ease of access for doctors
and patients alike using smart phones of different OS platforms to access to the data services provided by
the IoT cloud. The proposed system has been successfully deployed and fully tested with demonstrated

effectiveness and reliability in ECG monitoring.

The reminder of this paper is organized as follows. Section II presents the architecture of the IoT-

based ECG monitoring system. The system implementation is introduced in Section III, which includes
a monitoring node, the IoT cloud and a graphic user interface (GUI). In Section IV, we conduct several
tests on a healthy volunteer in order to verify the reliability of the proposed system. Finally, Section V

concludes this paper.

II. A RCHITECTURE OF I OT- BASED ECG MONITORING SYSTEM

The architecture of the IoT-based ECG monitoring system is illustrated in Fig. 1, which mainly consists
of three parts, i.e., the ECG sensing network, IoT cloud, and GUI.

A. ECG Sensing Network

The ECG sensing network is the foundation of the entire system, which is responsible for collecting
physiological data from the body surface and transmitting these data to the IoT cloud through a wireless
channel. Wearable ECG sensors are usually adopted in this system, which have little impact on the user’s

daily life. Through this means, ECG data can be recorded over long hours or even days. Then, the ECG
 4

ECG Sensing Network Internet-of-Things Cloud Graphical

User Interface

Data storage

Sensor
Access Point

Data cleaning Data analysis Mobile App

Bluetooth GPRS/LTE

Mobile phone
Sensor Disease warning

Base Station Web

Zigbee
Sensor GPRS/LTE
Mobile phone

Hospital

Fig. 1. Architecture of the IoT-based ECG monitoring systems.

TABLE I
C OMPARISONS AMONG TYPICAL ECG SENSING NETWORKS .

WiFi-based Bluetooth-based Zigbee-based

ECG sensing network ECG sensing network ECG sensing network
Protocol IEEE 802.11 IEEE 802.15.1 IEEE 802.15.4
Coverage 20-200 m 20-30 m 2-20 m
Data rates 11-54 Mbps 3-24 Mbps 10-250 kbps
Power consumption Medium Low Low
Smart terminals are needed for Smart terminals are needed for
Data collection is independent
Terminal dependency receiving and forwarding receiving and forwarding
of smart terminals.
sensed data. sensed data.
High, nearly all houses and public Medium, often supported by Low, only supported by specific
Popularization
places provide Wi-Fi access points. smartphones. devices

signals are processed through a series of procedures, such as amplification, filtering, etc., to improve the
signal quality and to meet the requirements of wireless transmission.

The ECG data gathered from sensors are transmitted to the IoT cloud via a specific wireless protocol,
e.g., Wi-Fi, Bluetooth, Zigbee, etc [17]. All the three protocols can provide enough data rates for
transmitting ECG signals with satisfying energy consumption. However, due to the limited communications

ranges of Bluetooth and Zigbee, a smart terminal (such as a mobile phone) is usually needed to receive
the ECG data and then send the data to the IoT cloud through the wireless protocols of the General
Packet Radio Service (GPRS) or Long Term Evolution (LTE). Comparisons among various types of ECG

sensing networks are given in Table I.

 5

B. IoT Cloud

Thanks to the development of the advanced IoT techniques, ECG data can be stored and analyzed
effectively and efficiently. With the aid of an IoT cloud, computation-intensive data process and analysis

tasks can be carried out in powerful servers, which greatly eases the burden of smart devices [20].
Generally speaking, an IoT cloud for ECG monitoring usually consists of four functional modules, i.e.,
data cleaning, data storage, data analysis, and disease warning.

• Data cleaning
Significant features can be extracted from ECG signals so as to detect potential heart diseases.
However, during the processes of data collection and transmission, noise may be introduced into the

ECG signal, which would adversely affect the diagnosis accuracy. Therefore, the ECG signal needs
to be cleaned at first. Commonly, a properly designed filter is employed to remove the noise outside
the band of the ECG signal. Furthermore, the procedure of data auditing is usually employed to

detect data anomalies and contradictions;

• Data storage
ECG data plays a vital role in the diagnosis of heart diseases. Thus, historical data are needed to be

stored in the database for further analysis. The ECG data often include the time and digitized signal
amplitude. In addition, at least one copy of the data needs to be stored for disaster recovery;
• Data analysis

Making the full use of data is one of the most important functions of the IoT cloud. Therefore, the
IoT cloud often provides a data analysis platform to extract useful information from the ECG signal.
Specific data mining or machine learning approaches can be applied to these data. For example, after

extracting the significant features of the ECG signal, a support vector machine can be established to
diagnose certain heart diseases [8] [21]; and
• Disease warning

Sudden heart attacks seriously threaten the lives of cardiac patients, especially when patients are
alone. Therefore, disease warning on the IoT cloud has become important for protecting patients
from being injured. Based on the results of data analysis, the IoT cloud is able to understand the

real-time health conditions of the patient. In the event of any suspicious readings, the IoT cloud will
 6

Controller Podule

Sensor WiFi
Sampling ADC Packetizing
Podule Buffer Podule

Power Podule

Fig. 2. Main components of the ECG monitoring node.

notify the family of the patient and the doctor in time.

C. GUI

The GUI is responsible for data visualization and management. It provides easy access to the data in
the IoT cloud. Users can log onto the cloud to acquire visualized ECG data in real time. Generally, two
kinds of GUIs are available for users to visualize ECG data, i.e., mobile apps and webpages. A mobile

app can provide an immediate response to user input, while webpages are more convenient in terms of
maintenance and upgrade.

III. I MPLEMENTATION OF THE I OT- BASED ECG MONITORING SYSTEM

Based on the architecture proposed in Section II, an IoT-based ECG monitoring system is implemented

using the advanced techniques of mobile sensing, cloud computing and Web. Details about the monitoring
node, IoT cloud and GUI are introduced as follows.

A. ECG monitoring node

The ECG monitoring node is responsible for collecting ECG data from the human skin and then sending

these data to the access point via a wireless channel. As depicted in Fig. 2, the ECG monitoring node
in our system mainly includes: 1) Sensor module; 2) Controller module; 3) Wi-Fi module; and 4) Power
module. A photo of the ECG monitoring node is shown in Fig. 3.

1) Sensor module: The sensor module is the foundation of the monitoring node, which is responsible
for collecting ECG data from the human body. With the aid of the AD8232 ECG sensor and certain

peripheral circuit, weak ECG signals can be detected with satisfactory accuracy [9]. Since the typical
 7

Dry electrode Controller module and

Wi-Fi module

Sensor module

Fig. 3. Photo of the ECG monitoring node.

frequency of the ECG signal lies between 0.5 Hz and 100 Hz [10], a band-pass filter is used in AD8283
to remove the noise outside this frequency band. After that, the filtered signal is amplified using an
operational amplifier. Finally, with the help of the sensor module, ECG signals from 0 v to 3.3 v are

gathered.

2) Controller module: As the core of the ECG monitoring node, the controller module is used to process

the gathered ECG signal and to send them to the Wi-Fi module. All the signal processing functions are
implemented in a high-performance Microcontroller Unit (MCU), i.e., STM32F103RC [11]. This MCU
is widely used in signal processing applications due to its powerful hardware, e.g., the 32-bit ARM

Cortex-M3 micro-controller operating at 72 MHz, a Flash memory up to 512 Kbytes, and an extensive
range of enhanced I/Os and peripherals connected via two APB buses. Furthermore, for the convenience
of developers, this MCU also provides some popular functional modules, such as the analog-to-digital

convertors (ADCs), PWM timer and communications interface. The ECG signal is processed using this
MCU, which consists mainly of four procedures, i.e., sampling, ADC, buffering, and packetizing. Firstly,
the analog ECG signal is digitized through sampling and ADC. Then, the ECG data are temporally stored

in the buffer before being packaged in accordance with a certain format. Finally, the data is transmitted
to the Wi-Fi module through a Universal Synchronous/Asynchronous Receiver/Transmitter (USART).

• Procedure 1- Sampling

The ECG signal collected from the sensor is in analogue format, which needs to be sampled at first.
 8

According to the Nyquist theorem, the sampling rate is set to be 200 Hz, given that the bandwidth

of the ECG signal is usually between 0.5 Hz and 100 Hz [10]. Using a 16-bit timer, the control
module can enter a timer interrupt every 5 ms for sampling and storing. In addition, the priority of
the timer interrupt is made the highest so as to avoid the effect of the other interrupts, such as the

external interrupt or USART interrupt;

• Procedure 2- ADC
The MCU has three built-in 12-bit ADCs, each of which shares up to 21 external channels, performing

the analog-to-digital conversion in the single-shot or scan modes. Only one channel of an ADC is
needed to digitize the analog ECG signal;
• Procedure 3- Buffering

The ECG data are temporarily stored before being packetized. The MCU has 48 Kbytes of embedded
SRAM, which is enough for buffering the data. When the ECG data accrue to 1 Kbytes, they will
be packetized in accordance with a certain format. Meanwhile, newly arrived data are also stored in

the remaining space of the buffer; and

• Procedure 4- Packetizing

All ECG data are packetized before they are sent to the Wi-Fi module through the USART. A complete
data packet usually consists of a series of data fields, e.g., the header, flags, payload, checksum, etc.
These data fields are designed to improve the efficiency and accuracy of data transmission;

3) Wi-Fi module: Through the USART, the Wi-Fi module can obtain the ECG data. The Wi-Fi module

provides a fast and convenient access to the Internet, which is able to transmit real-time ECG data to the
IoT cloud. Due to the use of the powerful MCU, data are packetized and transmitted according to certain
communications protocols.

4) Power module: The power module provides a reliable energy supply to every module in the ECG
monitoring node. Two modes of power supply are provided for users, i.e., the USB and the lithium battery,

due to their long lifetime and high portability.

 9

ECG monitoring node

Step 4- Publish

Database
IoT Cloud
Storage server

Step 5- Storing

HTTP server MQTT server

Step 3- Subscribe

Step 1- Request
Step 2- Response

GUI

Web Mobile Web

Patient Family member Doctor

Fig. 4. Architecture of the IoT cloud for ECG monitoring.

B. IoT Cloud for ECG Monitoring

After the system obtains large amounts of ECG data through the monitoring node, an IoT cloud is in
demand to provide a speedy and convenient way to store these data in a database, and to display the

ECG signal when required. Therefore, based on the state-of-the-art techniques of web services, cloud
computing and data storage, an IoT cloud is implemented which is illustrated in Fig. 4.

1) Servers: Three types of servers of various functionalities are used in the IoT cloud, i.e., the storage
server, HTTP server and MQTT server. Thanks to the development of virtualization technologies, these

servers can be established on visual machines so as to make full use of physical resources, e.g., CPU,
 10

memory, etc. The servers deployed in the IoT cloud are elaborated as follows.

• HTTP server
Based on the traditional request-response mechanism, the HTTP server is able to accept users’

responses and respond accordingly. In order to access to the ECG data, users need to send a GET
request to the IoT cloud via a URL. Then a file written in the Hypertext Markup Language (HTML)
is transmitted to the browser through the HTTP protocol. The browser is able to convert the HTML

file into a user friendly graphical interface for users to securely log into the server. After successfully
gaining the access, the HTTP server sends another HTML file which is used to present a graphical
interface for displaying the ECG signal;

• MQTT server
The HTTP server only provides a graphical interface to the ECG signal. However, the transmission of
ECG data from the monitoring node to the webpage is implemented based on the Message Queuing

Telemetry Transport (MQTT) protocol. Unlike the traditional HTTP protocol, the MQTT protocol is
designed to maintain a long-lived connection between the device and the client [12]. Therefore, the
latency of data transmission can be reduced to a certain extent. Furthermore, the MQTT protocol

requires less communications overheads compared with the HTTP, which helps save bandwidth for
data transmission. This advantage is useful especially when the data volume is massive. Due to the
above advantages, an MQTT server is more suited for providing real-time ECG data to users. The

mechanism of the MQTT protocol is based on the notion of “topic”. Two types of operations are
used in the communications between the device and the client, i.e., subscribe and publish. With the
help of the MQTT server, messages published on a certain topic by the device can be transmitted to

the clients who have subscribed the same topic. Therefore, the real-time performance of the server
is improved; and
• Storage server

The storage server is responsible for storing the ECG data into the database, which plays a vital role
in the diagnosis and early detection of heart diseases. Thus, timely and reliable storage of the ECG
data is regarded as one of the most significant functions of the IoT cloud. Traditionally, IoT clouds

employ relational databases to store sensed data, such as MySQL and Oracle. However, with a rapid
increase in both data volumes and types, the read-write speed of relational databases has become
 11

a bottleneck for the performance of the cloud. Thanks to the advent of non-relational databases,

IoT clouds can store data in a more flexible manner. As one of the most promising non-relational
databases, Redis is employed in our proposed system. Redis stores all key-value data in the memory.
As a result, both the flexibility and the I/O speed of the database can be greatly improved [13].

2) Main procedures of data transmission through the IoT cloud: With the aid of the IoT cloud, the
ECG signal of a certain patient can be easily obtained via a web browser. Apart from the patient himself
or herself, anyone related to this person has access to these private data, such as family members and

doctors. Thus treatment for a potential heart disease is more likely to be carried out in time. The main
procedures of data transmission from the ECG monitoring node to the webpage are depicted in Fig. 4,
which are also detailed blow.

• Step 1- Request: The user sends a request to the HTTP server to access the webpage;
• Step 2- Response: The HTTP server sends an HTML file back to the user, which can be converted
to a webpage by the web browser;

• Step 3- Subscribe: Using the Application Programming Interface (API) of the IoT cloud, the webpage
is able to subscribe certain topics related to the ECG monitoring node;
• Step 4- Publish: The ECG monitoring node publishes data to the MQTT server on a certain topic.

These data are forwarded to all the webpages that have subscribed the same topic; and
• Step 5- Storing: ECG data are stored into the database managed by the storage server.

C. GUI for ECG Monitoring

Independent of additional mobile applications, users can login the IoT cloud and acquire ECG data

by just visiting a certain website using a web browser of any OS platform. Therefore, nearly all smart
terminals, including desktop PCs, laptops and smart phones, are able to obtain the service of the IoT
cloud. The web-based GUI is shown in Fig. 5. Apart from displaying real-time ECG signal, users can

also obtain the historical data by selecting the start time and end time in the control panel.
 12


 
 





    

       









     
     
 
 

 
 
 
 

  
 
 
 
  

 
 
 
  
 


Fig. 5. Example of the GUI for the ECG monitoring system.

TABLE II
K EY PARAMETERS OF THE ECG MONITORING SYSTEM .

Parameters Values
Type AD 8283
ECG sensor Power voltage 3.6 v
Output voltage 0-3.3 v
Wireless transmission protocols IEEE 802.11 b/g/n
Wi-Fi module Power voltage 3.3 v
Power consumption < 71 mA
CPU Intel Core i7-3632QM 2.2GHz / i7-3770 3.4GHz
Operation system Ubuntu 12.04 LTS, 64 bit
Server
HTTP server Apache Tomcat 8.0.27
Database Redis 3.2
Gender Male
Age 22
Volunteer
Height 179 cm
Weight 65 kg

IV. E XPERIMENTAL R ESULTS AND A NALYSIS

A. Experimental Configurations

Conventional 12-lead systems are widely applied in medical institutions, which can capture accurate
ECG signals for professionals. However, too many electrodes may have a negative effect on the portability
of the system and comfortability of the patient. According to [14], a 3-lead placement is sufficient to

capture the primary features of the ECG signal. In order to best sample the ECG signal, the electrodes
need to be placed around the heart and form a triangle. The 3-lead placement employed in our system is
illustrated in Fig. 6. Several experiments are conducted on a healthy volunteer using the 3-lead placement

for the purpose of verifying the reliability and accuracy of the proposed system. Key parameters of the
ECG sensor, Wi-Fi module, server, and volunteer are listed in Table II.
 13

R L

Fig. 6. 3-lead ECG electrode placement.

QRS
complex

R RR interval R

T T
P P

PR interval

Q Q
S S
QT interval

Fig. 7. Standard ECG signal.

Typical ECG signals mainly consist of five types of waves, i.e., the P wave, T wave, Q wave, R wave,
and S wave, as illustrated in Fig. 6. The intervals of these waves are usually used to diagnose a variety of

heart diseases. Among all the features of these waves, four are most commonly used in medical diagnosis,
i.e.,

• RR interval:

As one of the most conspicuous characteristics, the R wave is often used to identify the period of
an ECG signal. RR interval indicates the time interval between two adjacent R waves, which may
become irregular in the event of some heart diseases, e.g., the arrhythmia;
 14

TABLE III
N ORMAL VALUES OF KEY ECG PARATERMS .

Features Normal range (Second)

RR interval 0.6- 1
PR interval 0.12- 0.20
QT interval 0.32- 0.44
QRS duration < 0.12

• PR interval:
The PR interval measures the time between the beginning of the P wave and that of the QRS complex.
It indicates the time the impulse takes to reach the ventricles from the sinus node;

• QT interval:
QT interval represents the time between the start of the Q wave and the end of the T wave, which is
related with the ventricular depolarization and repolarization. There is an increased risk of ventricular

fibrillation or even sudden cardiac death if the QT interval exceeds the normal value; and
• QRS complex:
QRS complex is mainly associated with the ventricular depolarization, which consists of three

important waves, i.e., Q wave, R wave and S wave. Through analyzing the morphology and the
duration of the QRS complex, certain diseases are likely to be detected, e.g., electrolyte imbalance

or drug toxicity.

According to [15] [16], the normal values of ECG paratmeters are listed in Table III.

B. Results and Analysis

Fig. 8(a) plots part of the ECG data collected from the healthy volunteer. It is evident that the
intervals between adjacent R waves (RR interval) are nearly the same, which shows no risk of developing

arrhythmias. In order to demonstrate the key features of the measured ECG signal, two cycles of the signal
are chosen as examples, which is illustrated in Fig. 8(b). Through comparing the values of these features
with the normal ranges listed in Table III, certain heart diseases can be detected preliminarily. As can

be seen from Fig. 8(b), the interval between two R waves is 0.68 s, which indicates a healthy heart rate
of the volunteer. The QT interval and QRS duration are measured to be 0.32 s and 0.11 s, respectively,
indicating no abnormality in the processes of ventricular depolarization and repolarization. Finally, the PR
 15

2.6
ECG Data
RR Interval

2.4

2.2

Output Voltage (V) 2.0

1.8

1.6

1.4
0 0.8 1.6 2.4 3.2 4.0 4.8 5.6
Time (s)
(a) Portion of the measured ECG signal.

2.8
ECG Data
2.6 QRS Duration= 0.11 s

2.4
RR Interval= 0.68 s
Output Voltage (V)

2.2

2.0

1.8

1.6

1.4
QT Interval= 0.32 s
PR Interval= 0.15 s
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2
Time (s)
(b) Key features of the measured ECG signal.
Fig. 8. Data collected by the wearable ECG monitoring system.

interval is also within the normal range, which implies that the process of atrial depolarization is normal
as well.

Based on the above analysis, it can be concluded that this volunteer has no obvious heart diseases,
which accords with the foregone conclusion. Therefore, the data collected by the proposed system satisfy
both the accuracy and reliability requirements.

It is well known that the ECG signal is related to the motion state of the volunteer, attributed to the
 16

variation of the heart rate. To reveal the impact of movement on the ECG signal, the volunteer is monitored

under two motion states, i.e., the stationary state and moving state. The comparison between the ECG
signals under these two states are illustrated in Fig. 9. As can be observed from the figure, the cycle
of the ECG signal in the stationary state is greater than that in the moving state, as a person in motion

usually has a faster sinus rhythm. However, there is no obvious difference between the amplitudes of the
two signals, because the amplitude of the ECG signal relates mainly to the resistance of the human body
as well as the distance between each electrode, which does not vary too much in the two states.

2.6
Stationary State
2.4
Output Voltage (V)

2.2

2.0

1.8

1.6

1.4
160 320 480 640 800 960
Time (S)
2.6
Moving State
2.4
Output Voltage (V)

2.2
2.0
1.8
1.6
1.4
1.2
160 320 480 640 800 960
Time (S)

Fig. 9. Comparison of the ECG signals in two distinct states.

V. C ONCLUSION

We designed and implemented an ECG monitoring system based on cutting-edge IoT techniques. The

architecture of the ECG monitoring system was presented at first. Typical ECG sensing networks including
Wi-Fi, Bluetooth, and Zigbee were introduced and compared. Based on the proposed architecture, an
IoT-based ECG monitoring system was implemented. Through a wearable monitoring node with three

electrodes, real-time ECG signals can be collected with satisfactory accuracy. The gathered data were
transmitted to the IoT cloud using Wi-Fi, which supports high data rates and wide coverage areas. The
IoT cloud is responsible for visualizing the ECG data to users and storing these valuable data for further

analysis, which is implemented on the basis of three servers, i.e., the HTTP server, MQTT server, and
storage server. Eliminating the need of mobile applications, the web-based GUI provides a versatile means
 17

independent of any mobile OS platform for users to access to the ECG data. Further studies on ECG

monitoring are still needed in the future. For example, the accuracy of diagnostic results based on the
ECG signal needs to be improved so as to provide a more reliable disease diagnosis. It is believed that
long-term and user-friendly ECG monitoring can greatly help mitigate existing healthcare problems to a

certain extent.

VI. ACKNOWLEDGMENT

This work was supported by the National High-Tech R&D Program (863 Program 2015AA01A705), the

China Natural Science Funding under the grant 61331009, the National Key Technology R&D Program of
China under the grant 2015ZX03002009-004, the Fundamental Research Funds for the Central Universities
under the grant 2014ZD03-02.

R EFERENCES
[1] (2016) Ageing. In: World Health Organization. http://www.who.int/topics/ageing/en/. Accessed 18 Sep 2016
[2] Abadi M, Subramanian R, Kia S et al. (2015) DECAF: MEG-based multimodal database for decoding affective physiological responses.
IEEE Transactions on Affective Computing 6:209-222. doi: 10.1109/taffc.2015.2392932
[3] Kim H, Kim S, Van Helleputte N et al. (2014) A configurable and low-power mixed signal SoC for portable ECG monitoring
applications. IEEE Transactions on Biomedical Circuits and Systems 8:257-267. doi: 10.1109/tbcas.2013.2260159
[4] Tseng C (2013) Coordinator traffic diffusion for data-intensive Zigbee transmission in real-time electrocardiography monitoring. IEEE
Transactions on Biomedical Engineering 60:3340-3346. doi: 10.1109/tbme.2013.2266373
[5] Miao F, Cheng Y, He Y et al. (2015) A wearable context-aware ECG monitoring system integrated with built-in kinematic sensors of
the smartphone. Sensors 15:11465-11484. doi: 10.3390/s150511465
[6] Tseng K, Lin B, Liao L et al. (2014) Development of a wearable mobile electrocardiogram monitoring system by using novel dry
foam electrodes. IEEE Systems Journal 8:900-906. doi: 10.1109/jsyst.2013.2260620
[7] Friedman R, Kogan A, Krivolapov Y (2013) On power and throughput tradeoffs of WiFi and Bluetooth in smartphones. IEEE
Transactions on Mobile Computing 12:1363-1376. doi: 10.1109/tmc.2012.117
[8] Alonso-Atienza F, Morgado E, Fernandez-Martinez L et al. (2014) Detection of life-threatening arrhythmias using feature selection
and support vector machines. IEEE Transactions on Biomedical Engineering 61:832-840. doi: 10.1109/tbme.2013.2290800
[9] (2016) AD8283 Datasheet(PDF) - Analog Devices. In: Alldatasheet.com. http://www.alldatasheet.com/datasheet-
pdf/pdf/418481/AD/AD8283.html. Accessed 18 Sep 2016
[10] Buenda-Fuentes F, Arnau-Vives M, Arnau-Vives A et al. (2012) High-bandpass filters in electrocardiography: source of error in the
interpretation of the ST segment. ISRN Cardiology 2012:1-10. doi: 10.5402/2012/706217
[11] (2016) STM32F103RC Datasheet(PDF) - STMicroelectronics. In: Alldatasheet.com. http://www.alldatasheet.com/datasheet-
pdf/pdf/231959/STMICROELECTRONICS/STM32F103RC.html. Accessed 18 Sep 2016
[12] Al-Fuqaha A, Guizani M, Mohammadi M et al. (2015) Internet of things: a survey on enabling technologies, protocols, and applications.
IEEE Communications Surveys & Tutorials 17:2347-2376. doi: 10.1109/comst.2015.2444095
[13] Phan T et al. Cloud databases for Internet-of-things data. In: Proc. IEEE 2014 International Conference on Internet of Things (iThings),
Taipei, Sep. 2014, pp. 117-124.
[14] Wang Y et al. Design and evaluation of a novel wireless reconstructed 3-lead ECG monitoring system. In: Proc. IEEE 2013 Biomedical
Circuits and Systems Conference (BioCAS), Rotterdam, Oct. 2013, pp. 362-365.
[15] Gertsch M (2009) The Normal ECG and its (Normal) variants. In: The ECG manual. Springer, London, pp. 17-36
[16] (2016) ECGlibrary.com: Normal adult 12-lead ECG. In: Ecglibrary.com. http://www.ecglibrary.com/norm.php. Accessed 18 Sep 2016
[17] Palattella M, Dohler M, Grieco A et al. (2016) Internet of Things in the 5G era: enablers, architecture, and business models. IEEE
Journal on Selected Areas in Communications 34:510-527. doi: 10.1109/jsac.2016.2525418
[18] Xiong X, Zheng K, Xu R et al. (2015) Low power wide area machine-to-machine networks: key techniques and prototype. IEEE
Communications Magazine 53:64-71. doi: 10.1109/mcom.2015.7263374
[19] Lei L, Kuang Y, Cheng N et al. (2016) Delay-optimal dynamic mode selection and resource allocation in device-to-device
communicationspart II: practical algorithm. IEEE Transactions on Vehicular Technology 65:3491-3505. doi: 10.1109/tvt.2015.2444791
[20] Hou L, Zhao S, Zheng K et al. (2016) Internet of Things cloud: architecture and implementation. Accepted by IEEE Communications
Magazine. arXiv: 1609.07712
 18

[21] Zheng K, Yang Z, Zhang K et al. (2016) Big data-driven optimization for mobile networks toward 5G. IEEE Network 30:44-51. doi:
10.1109/mnet.2016.7389830
[22] He S, Chen J et al. (2014) Mobility and intruder prior information improving the barrier coverage of sparse sensor networks. IEEE
Transactions on Mobile Computing 13:1268-1282. doi: 10.1109/tmc.2013.129

View publication stats