2018 IEEE International Symposium on Smart Electronic Systems (iSES) (Formerly iNiS)

Three Tier Architecture for IoT Driven Health

Monitoring System Using Raspberry Pi
Neel Kamal, Prasun Ghosal
Indian Institute of Engineering Science and Technology, Shibpur, India
Email: [email protected], p [email protected]

Abstract—Wireless sensor networks (WSNs) have witnessed A. Novel Contributions of the Paper
advancement in medical services from real-time tracking and
computer-assisted machine to alert response systems. Due to
The paper provides the details of a proposed system efficient
a tremendous shortage of trained manpower and a huge cost to continuously monitor and generate different reports of the
for setting up state-of-the-art facilities, it is often not possible patient. The developed system is robust and economical. The
to deliver proper health care services in the rural and remote system can inform the wellness status of the inhabitant to the
areas. Lack of accurate and timely information further adds health service provider in advance. The system provides real
complexity and challenges to the problem. The proposed system
uses a three-tier architecture that can be generally applied to
time monitoring of the patient. The system processes various
WSN based healthcare systems. The proposed model monitors samples of data and extracts the information through prob-
the patient body temperature, heartbeat, and body position abilistic analysis according to the proposed decision model.
movements constantly, and sends this information to site pages The system generates predictive decision in the implemented
and crisis centres/services from the remote location. WSNs are hardware prototype depending upon several monitoring health
composed of low power consuming sensors. Raspberry pi is a
credit card sized board that uses 5V power supply. The proposed
parameters.
and implemented system prototype uses raspberry pi that is
driven by Internet of Things (IoT) connected through different B. Overall Organization of the Paper
sensors DS18B20, ADXL345, ADC1015 and heartbeat sensor. The
framework additionally gives crisis warning to a specialist and The remainder of the paper is organized in the following
sends the information on a web server. The framework utilizes manner. Section II discusses related research work and motiva-
DS18B20, heartbeat sensor, and accelerometer. The system is tion for the paper. Section III provides overview for proposed
designed secured by providing a mechanism to authenticate the system having three tier architecture. Section IV describes the
user to get access to patient data. Implemented system hardware
mechanism of decision making model for hardware prototype.
prototype consists of two controllers that provide a mechanism
to bring personal health status in the normal range in case of Section V gives brief description on hardware prototype con-
emergency. sisting of stepper motor. Results are discussed in section VI.
Finally section VII consists of conclusion and future work.
Index Terms—IoT, Health monitoring system, Raspberry Pi
and IoT, Architecture.
II. R ELATED R ESEARCH AND M OTIVATION
Implementation of IoT (Internet-of-things) in healthcare
services provides data consistency and reduces duplication of
I. I NTRODUCTION data related to patient [1][3]. Electronic Health care system
provides tracking, tracing and monitoring of patients with
In India, nearly three quarters of 1.2 billion population live accuracy where a traditional health care system falls short [5]
in rural areas. Providing efficient health care services is a [10]. IoT technologies are incorporated with various sensors
significant challenge for them. Due to tremendous shortage connected to processors through various open sources like
of trained manpower and huge cost for setting up state-of-the- AWS (Amazon Web Services) [7] [14]. The use of fog com-
art facilities, it is often not possible to deliver proper health puting in IoT provides time sensitive health care applications
care services in the rural and remote areas. Lack of accurate [11]. Fog computing brings cloud closer to cease-customers
and timely information further adds to the complexity of the and records sources by allowing computation at the side of the
problem. The efficient use of technology can improve the network. Low latency is the essential gain [6]. IoT applications
overall performance of the health care system and its reach are often latency-sensitive [9]. Such applications can also
ability amid the citizens of India. be provisioned as factor-based totally in a hybrid cloud/fog
A survey was conducted by NCBI to observe the function- surroundings with elements spanning cloud and fog [4] [13].
ality and working conditions of different types of health care This permits placing some of its additives in the fog region
delivery to observe the improvement in health care services closer to the IoT devices, and therefore reduces the latency [6].
with the introduction of information technology in health care Healthcare utility development in FIWARE FI-superstar uti-
field. The survey revealed that the data related to patient are lizes cloud computing and net of things (IoT) idea to assemble
not kept properly and were maintained manually that leads to innovative future net services for healthcare provisions in the
duplication of data of patients and inconsistency in data. information and conversation era [2]. This encompasses quite

0-7695-6618-9/18/$31.00 ©2018 IEEE 167

DOI 10.1109/iSES.2018.00044
a number of use instances such as monitoring, rehabilitation IV. D ETAILS OF D ECISION M AKING M ODEL
and others of sufferers making use of cloud offerings [12] [8].
Detailed description of the decision making process in the
proposed three tier architecture may be provided as follows
III. P ROPOSED S YSTEM (shown in Figure 2).

The proposed system provides continuous monitoring of

a patient. The system uses raspberry pi and various sensors
like DS18B20, ADC1015, ADXL345 and heart beat sensor.
The system provides assistance that includes automation in
systematic monitoring of patients, transmission of medical data
at real-time, saving those data in database, and manipulation
data for future use. The proposed system operates in two
modes. The mode depends upon number of samples collected
that is 50 samples (fast mode) and 200 samples (high precision
mode). The system uses a web page that requires authentica-
tion to access patient data. The proposed system consists of
three tier architecture. The tier one consists of wired sensor Fig. 2. Decision model.
network that have low power consumption to collect various
samples of patient. Tier two consists of processing unit that
includes processing model, probabilistic model and decision
model. The system generates live graph of various samples A. Data processing
of data that are collected through GPIO. Tier three consists
of web page for remote access for patient condition and The tier one consists of group of sensors connected from
hardware prototype that provides mechanism to control body various GPIO (gpio4, SDA and SCL) to collect the samples
temperature and heart beat of patient in case of emergency. of body temperature, heart beat and body position of patient.
This is shown in Figure 1. These are carried to the processing model. The system collects
various samples from patient and processes these samples
through various models to convert a raw data to information,
maintaining the Integrity of the Specifications.

B. Decision making model

Decision making model with its two controllers at tier three
provides a mechanism to bring person health status in nor-
mal range. Decision making model makes decision regarding
hardware prototype that are room temperature controller and
external pace maker consisting of stepper motor. Stepper motor
used is Model 28BYJ-48 and they are driven by ULN2003
that provides required current to drive steppers. Hardware
prototype consists of two controllers that provide a mechanism
to bring person health status in normal range in case of
emergency condition.

Fig. 1. Three Tier Architecture of proposed system.

C. Protocols used in proposed system

The system does processing on data received from sensor
to determine whether data is in normal range or unusual The proposed system uses various protocols at different tiers
range to determine various cases like heatstroke, hypothermia, of three tier architecture. Tier one consists group of wired
bradycardia and tachycardia. The system does probabilistic sensors having interfaces with raspberry pi using protocols one
analysis of data collected from sensor through GPIO. Hard- wire protocol for temperature sensor and I2C for accelerometer
ware prototype consists of two controllers that provides a and ADC. I2C can communicate with one controller and total
mechanism to bring person health status in normal range 127 slaves at same time having 7 bit address. Tier two does
in case of emergency condition. Decision making model the processing of data and visualization of data using mat plot.
makes decision regarding hardware prototype that are room Tier three does the wireless communication between system
temperature controller and external pace maker (in emergency and web page (iothms.com) using get and post method and
case) consisting of stepper motor. IEEE802.11 and provides decision to hardware prototype.

168
D. Process flow of system
Raspberry Pi communicates with sensors through GPIO.
The system Controls the devices and sensors from GPIO by
using raspbian OS. Raspberry pi is connected by laptop using
Ethernet port and uses PUTTY to make connection between
laptop and raspberry pi. Each system has a unique user name
and password. The samples are collected and processed by
processing model. The data are stored in buffer for probabilis-
tic analysis and live graphs. The output of probabilistic model
is input to decision making model to take decisions for the Fig. 4. Interfacing diagram of DS18B20 and ADXL345.
hardware in case of emergency.

V. H ARDWARE P ROTOTYPE
Stepper one (room temperature controller) module controls
the temperature of surroundings according to normal body
temperature. It moves in clockwise direction to increase the
temperature of surrounding in case person body temperature
falls below the normal temperature. Similarly it moves anti-
clockwise to increase the temperature of surroundings in case
the person body temperature goes above the normal tempera-
ture. Stepper two (pacemaker) module controls the heart beat.
Stepper two moves in clockwise direction to increase the heart
beat in case person heart beat falls below the normal heart beat
rate and moves anticlockwise to increase the heart beat in case
person body heart beat becomes more than the normal range.
Fig. 5. Screen shots of result for body temperature of patient.
VI. E XPERIMENTS AND R ESULT D ISCUSSION
This section consists of detailed description of the interfac-
ing done for the experimental purpose, experiments performed,
and discussion on results obtained thereby. Figure 3 shows
interfacing of heart beat sensor(along with ADC) and figure 4
shows interfacing od ADXL345 and temperature sensor with
raspberry pi using SDA, SCL and GPIO 4 . The result of body
temperature is (shown in Figure 5).The result of heartbeat and
unexpected body movement is (shown in Figure 6) and data
visualization of heart beat, body temperature and unexpected
body movement is (shown in figure 7).

Fig. 3. Interfacing diagram of ADC1015 with heartbeat sensor.

Fig. 6. Screen shots of result for heartbeat and unexpected body movement
A. Results and Discussion of patient.

The proposed system works more efficient way as compared

to previous work. The previous proposed system does not hypothermia, bradycardia, and tachycardia. The system does
provide any processing of data or any kind of authentication probabilistic analysis of data collected from sensor through
access. The proposed system does processing on data received GPIO. The previous work does not provide any mechanism
from sensor to determine whether data is in normal range to control patient condition in case of emergency. Detailed
or unusual range to determine various cases like heatstroke, comparisons have been presented in Table I.

169
 precautionary measures in case of emergency.

R EFERENCES
[1] Jigar Chauhan and Sachin Bojewar. Sensor networks based
healthcare monitoring system. In Inventive Computation Tech-
nologies (ICICT), International Conference on, volume 2, pages
1–6, 2016.
[2] Maria Fazio, Antonio Celesti, Fermı́n Galán Márquez, Alex
Glikson, and Massimo Villari. Exploiting the fiware cloud
platform to develop a remote patient monitoring system. In
Computers and Communication (ISCC), 2015 IEEE Symposium
on, pages 264–270. IEEE, 2015.
[3] M Surya Deekshith Gupta, Vamsikrishna Patchava, and Virginia
Menezes. Healthcare based on iot using raspberry pi. In
Green Computing and Internet of Things (ICGCIoT), 2015
International Conference on, pages 796–799. IEEE, 2015.
Fig. 7. Data visualization of body temperature , heart beat and body position. [4] Kirak Hong, David Lillethun, Umakishore Ramachandran,
Beate Ottenwälder, and Boris Koldehofe. Mobile fog: A
TABLE I programming model for large-scale applications on the internet
C OMPARISON WITH PREVIOUS WORK of things. In Proceedings of the second ACM SIGCOMM
workshop on Mobile cloud computing, pages 15–20. ACM,
Paper [5] Paper [1] Proposed 2013.
ICCSP ICICT System [5] Abhilasha Ingole, Shrikant Ambatkar, and Sandeep Kakde. Im-
2015 2016 plementation of health-care monitoring system using raspberry
Heart Beat YES YES YES pi. In Communications and Signal Processing (ICCSP), 2015
Body Temperature Sensor YES YES YES International Conference on, pages 1083–1086. IEEE, 2015.
Unexpected Motion of Body NO NO YES
[6] S Jayanth, MB Poorvi, and MP Sunil. Raspberry pi based
Detection of Abnormal Con- NO NO YES
ditions in Body Temperature
energy management system. In Green Engineering and Tech-
Detection of Abnormal Con- NO NO YES nologies (IC-GET), 2016 Online International Conference on,
ditions in Heart Beat pages 1–5. IEEE, 2016.
Processing of Data Locally NO NO YES [7] JeongGil Ko, Chenyang Lu, Mani B Srivastava, John A
Data Available on Web page YES NO YES Stankovic, Andreas Terzis, and Matt Welsh. Wireless sensor
Authenticated Access NO NO YES networks for healthcare. Proceedings of the IEEE, 98(11):1947–
Live Streaming of Samples NO NO YES 1960, 2010.
Hardware Prototype NO NO YES [8] Liang Li, Yunzhou Li, and Ronghui Hou. A novel mobile
edge computing-based architecture for future cellular vehicular
networks. In Wireless Communications and Networking Con-
ference (WCNC), 2017 IEEE, pages 1–6. IEEE, 2017.
VII. C ONCLUSION AND F UTURE W ORKS [9] Tanmaya A Onkar and PT Karule. Web based maintenance for
The proposed system is efficient to generate different reports industrial application using raspberry-pi. In Green Engineering
in different modes i.e. precision mode (200 samples) and fast and Technologies (IC-GET), 2016 Online International Confer-
ence on, pages 1–4. IEEE, 2016.
mode (50 samples). The system does the processing of data
[10] Vivek Pardeshi, Saurabh Sagar, Swapnil Murmurwar, and
locally before transmission of data to web page. System does Pankaj Hage. Health monitoring systems using iot and raspberry
detection of hypothermia, heatstroke, bradycardia of patient pi—a review. In Innovative Mechanisms for Industry Applica-
by analyzing (200 samples/sec) of body temperature and tions (ICIMIA), 2017 International Conference on, pages 134–
heart beat. System provides the Authentication mechanism to 137. IEEE, 2017.
[11] Mamta Puppala, Tiancheng He, Xiaohui Yu, Shenyi Chen,
get access of patient health data from health care provider.
Richard Ogunti, and Stephen TC Wong. Data security and
The system provides continuous monitoring and takes proper privacy management in healthcare applications and clinical data
actions through controllers. The developed system is robust warehouse environment. In Biomedical and Health Informatics
and is possible to develop at a very low cost. The system can (BHI), 2016 IEEE-EMBS International Conference on, pages
inform the wellness status of the inhabitant to the caregiver 5–8. IEEE, 2016.
[12] Shahid Raza, Hossein Shafagh, Kasun Hewage, René Hummen,
in advance. The proposed system generate real time graph of
and Thiemo Voigt. Lithe: Lightweight secure coap for the
data that are body temperature, heart beat and body position internet of things. IEEE Sensors Journal, 13(10):3711–3720,
that are collected through GPIO for better visualization of 2013.
patient data. The system provides cost effective and real [13] Mahadev Satyanarayanan. The emergence of edge computing.
time monitoring of a person. The system does processing Computer, 50(1):30–39, 2017.
[14] Shree Krishna Sharma and Xianbin Wang. Live data analytics
of various group of samples and extracts the information
with collaborative edge and cloud processing in wireless iot
through probabilistic analysis. The system generates decision networks. IEEE Access, 5:4621–4635, 2017.
for hardware prototype that are room temperature controller
(stepper one) and external pacemaker (stepper two) in case
of emergency. Future works may be extended to consider
and analyze more numbers of health parameters and making
the decision engine more accurate to analyze and predict
real time health conditions and suggesting proper necessary

170