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

net/publication/315915442

A simple design of wearable device for fall detection with accelerometer and
gyroscope

Conference Paper · November 2016

DOI: 10.1109/ISESD.2016.7886698

CITATIONS READS
11 1,144

4 authors, including:

Mochamad Irwan Nari Sena Sukmananda Suprapto

Politeknik Negeri Jember Institut Teknologi Kalimantan
6 PUBLICATIONS   13 CITATIONS    7 PUBLICATIONS   31 CITATIONS   

SEE PROFILE SEE PROFILE

Widyawardana Adiprawita
Bandung Institute of Technology
83 PUBLICATIONS   429 CITATIONS   

SEE PROFILE

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

Auditory Neuroscience View project

Cooperative Control on multi-robot systems View project

All content following this page was uploaded by Sena Sukmananda Suprapto on 16 July 2019.

The user has requested enhancement of the downloaded file.

 2016 International Symposium on Electronics and Smart Devices (ISESD) November 29-30, 2016

A Simple Design of Wearable Device for Fall

Detection with Accelerometer and Gyroscope

M. Irwan Nari, Sena S. Suprapto, Ilman H. Kusumah Widyawardana Adiprawita

School of Electrical Engineering and Informatics School of Electrical Engineering and Informatics
Institut Teknologi Bandung Institut Teknologi Bandung
Bandung, Indonesia Bandung, Indonesia
[email protected], [email protected], [email protected]
[email protected]

Abstract—Falling is one of the most common injurious The simplest algorithm is using a threshold value [3]. There are
accidents in the elderly which could result in pain, disability, and also several algorithm modifications to optimize the system such
even death. About 33% of the elderly have falling accident each as adaptive threshold [10], lower fall threshold (LFT) and upper
year. According to the Statistics Indonesia data in 2014, the total fall threshold (UFT) [9]. However, the more complicated
population of over-60-year-olds increase to 8.03% of the total algorithms are using Support Vector Machine (SVM) [7],
population in that year. In this research, we use the accelerometer Markov Model [8], Bayesian Classifier [11] and Hybrid Model
and a gyroscope sensor for fall detection. The device is placed in [12]. However, the appropriate algorithm for fall detection
the position of the waist. This study uses 6 type activity system has to be able to distinguish the state of a falling person
comparators such as walking, running, upstairs, down stairs, sit
under any circumstances in a real environment.
down quickly and stand up quickly. These tests are also used for 3
falling accidents such as falling forward, to the right and to the In this paper, we use the accelerometer and a gyroscope
left. By comparing the results of these tests, the falling threshold sensor for fall detection system. The accelerometer and
values can be acquired. The threshold values of A, B and C are 0.6, gyroscope sensor provide inertial body change and rotation
1.7 and 175 respectively. Results show that the system can detect speed of the body data respectively. The developed system can
sensitivity and specificity up to 90% and 86.7% respectively. detect fall accident in daily activities. These study results are an
easy-to-use wearable device design and a fall algorithm
Keywords— Fall Detection, Accelerometer, Gyroscope, Wireless
development. The system is also equipped with a push button
Sensor System, Elderly
that used to ask for help when the person who fall still conscious.
In addition, the system can also give an alert to a related person
such as family member or emergency department.
I. INTRODUCTION
The remainder of this study is organized as follows: Section
The elderly population in the world is increasing every year. I describes the introduction and some algorithms of the fall
Based on data from The United Nations (UN), the total detection system in the previous study. Section II describes the
population of 60-year-olds and beyond is estimated to reach 1.4 proposed method of fall detection system including the design
billion people in 2030 [1]. The largest population is in Africa of a device, system architecture and fall detection
which in 2015 amounted to more than half a billion people. algorithms. Section III discuss the experimental results. Finally,
According to The Statistic Indonesia survey (BPS, Badan Pusat in Section IV describes the conclusions and future work of the
Statistik) in 2014, the elderly population in Indonesia is also an study.
increase to 8.03% of the total population in that year [2]. The
increasing elderly population is caused by increasing life II. METHOD
expectancy and declining birth rates.
A. Systems Architecture
Falling is one of the incidents most often occur in the elderly.
In general, the fall detection system consists of a sensor, a
About 33% of the elderly have experienced at least one or more
microcontroller, a power supply, a Bluetooth and a receiver
falls each year [3]. The mechanism falls usually occur through
slipped. This happens because of several factors such as a module. The embedded device system architecture and the
physical injury and unstable emotional factor [4]. This situation receiver module using a laptop are shown in Fig. 1 and Fig. 2
would be very harmful to the elderly. Especially if the elderly respectively.
living alone at home. The impact of falling such as pain, 1). Sensor Module
disability, even death and will increase healthcare costs [5][6]. MCU-6050 [13] is a motion detection sensor with 6 axis
In previous studies, there are several fall detection method sensor which 3 axis accelerometer and 3 axis gyroscope sensor.
selections of sensor and algorithm. The sensor selections have The sensor dimensions are 4x4x0.9 mm. The working voltage is
several axis variations which are one, two, or three axes. Some 3.3 volts DC. These sensors are connected to the microcontroller
studies use one of them, and others are using its combination.

978-1-5090-3840-4/16/$31.00 ©2016 IEEE 88

 MPU-6050 Bluetooth
Accelerometer + Gyro HC-05

= ( ) + ( ) + ( ) (2)
Arduino Nano Board
(Microcontroller)

Power Supply
Gyrox, Gyroy, and Gyroz are the angular velocity of the x-
axis, y-axis and z-axis. In terms of getting a difference of fall
Fig. 1. Architecture module of device embedded in the body with another activity, it needs threshold values. In this study,
there are 3 thresholds for detecting a fall accident. These
thresholds are:
Inbuilt 1) Threshold A: Minimum limit for the acceleration of fall
Bluetooth
in the early phase.
2) Threshold B: Minimum limit for the acceleration of fall.
Fig. 2. Receiver module using a computer 3) Threshold C: Minimum limit for the angular velocity of
Fall.
through I2C communication (inter-integrated circuit). We use it
to measure the acceleration and angular velocity to detect the
man while falling accident. C. Fall Detection Algorithm
We measured parameters of sensitivity and specificity to test
2). Microcontroller Module the reliability of created system. Both of these parameters
Arduino Nano Microcontroller is used to control the system. can be calculated using Equation (3) and Equation (4).
This type of Arduino is ATMega328P chip-based
microcontroller which has dimensions of 45mm x 18mm and
weighing about 5 grams. This microcontroller has 14 digital pins
I / O, 7-12 volt input voltage, the clock speed of 16 MHz, and = (3)
the memory of 32 KB of flash. With these features, this board is +
suitable for embedded devices [14].
3). Power Supply
= (4)
External Power Supply is 12 volts battery which has small +
dimension. This battery is connected to Vin pin which has 6 - 20
volts input voltage and to ground of Arduino Nano.
True Positive (TP) is a condition where the system can detect
4) Bluetooth Module the fall when the fall occurred and False Positive (FP) is a
Bluetooth is a wireless communication protocol that works condition where the system detects a fall but the fall does not
on a radio frequency of 2.4 GHz. Bluetooth HC-05 [15] is a occur. True Negative (TN) is a condition where the system does
Bluetooth module that is common module in the market and the not detect the fall and the fact fall accident not occur and False
price is relatively cheap. This module is connected to the Negative (FN) is a condition where the system can not detect the
microcontroller via the TX and RX pin. This module can be fall when the fall occurred.
operated with the baud rate default. The advantage of using this
communication is a high speed of data flow and ease of interface
with other devices.

B. Fall Detection Algorithm

The parameters which are used to determine the total vector
acceleration and angular velocity similar to a previous study [9].
The total acceleration vector or commonly called the signal
vector magnitude (SVM) can be calculated by Equation (1).

= ( ) + ( ) + ( ) (1)

Where Accx, Accy, and Accz are the gravitational Fig. 3. Fall detection device and device placement
acceleration of the x-axis, y-axis and z-axis. Then to calculate
the angular velocity can be used in the Equation (2).

89
 III. RESULTS
This study uses 6 activity comparators such as walking,
running, up stairs, down stairs, sit down quickly and stand up
quickly. These tests are also used for 3 falling accidents such as
falling forward, to the right and to the left. By comparing the
results of these tests, the falling threshold values can be
acquired.
Fig. 5 shows the accelerometer and gyroscope resultant from
each activity such as walking, running, sit, stand, upstairs, down
stairs and fall. Based on these observations we determine the
limits of fall detection systems. The threshold values of A, B and
C are 0.6, 1.7 and 175 respectively.

Fig. 4. Fall Detection Flowchart Fig. 6. Alert received at the computer

Fig. 6 shows the COM port output in realtime. If all of these

thresholds of the fall detection system are acquired, then system
will give an alert information on laptop.

IV. CONCLUSION
In this paper, a simple design of a wearable fall detection
system is presented. The subject in this study is male with a
height of 179 cm. The device is placed in the position of the
waist. Results show that the system can detect sensitivity and
specificity up to 90% and 86.7% respectively. Hence, our future
work is focused on making this system connected to android for
data acquisition and emergency alarm.

REFERENCES
[1] World Population Ageing Report 2015. Available: http://www.un.org
[2] Statistik Penduduk Lanjut Usia 2014. Available:
http://www.bappenas.go.id
[3] Yujia Ge and Bin Xu, “Detecting Falls Using Accelerometer by Adaptive
Thresholds in Mobile Devices,” Journal of computers, 2014.
[4] Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, and Dong Xuan,
“PerFallD: A Pervasive Fall Detection System Using Mobile Phones,”
Fig. 5. Display of Accelerometer and Gyroscope data for various activities Journal of computers, 2010.

90
[5] Shumway-Cook, A Ciol, MA Hoffman, J Dudgeon, BJ Yorkston and K [10] Yujia Ge and Bin Xu., “Detecting Falls Using Accelerometer by Adaptive
Chan,“Falls in the Medicare population: incidence, associated factors, and Thresholds in Mobile Devices,” Journal of computers, Vol. 9. 2014.
impacton health care,” Phys, 2009. [11] L. Atallah, B. Lo, R. King and G. Yang, “Sensor Positioning for Activity
[6] Shih-Hau Fang, Yi-Chung Liang, and Kuan-Ming Chiu, “Developing a Recognition Using Wearable Accelerometers,” IEEE Trans. Biomed.
Mobile Phone-based Fall detection System on Android Platform,” Circuits Syst., vol. 5, no. 4, pp. 320-329, 2011.
Seventh International Conference on Intelligent Environments, 2012. [12] J. Parkka, M. Ermes, K. Antila, M. van Gils, A. Manttari and H.
[7] T.Zhang, J.Wang, L.Xu and P.Liu, “Fall detection by wearable sensor and Nieminen, “Estimating Intensity of Physical Activity: A Comparison of
one-class SVM algorithm,” Intelligent Computing in Signal Processing Wearable Accelerometer and Gyro Sensors and 3 Sensor Locations,” 29th
and Pattern Recognition: International Conference on Intelligent Annual International Conference of the IEEE Engineering in Medicine
Computing, 2006. and Biology Society, 2007.
[8] P. Jayachandran, T. F. Abdelzaher, J. A. Stankovic, and R. K. Ganti, [13] MPU-6050 Module. Available: https://www.invensense.com/wp-
“SATIRE: a software architecture for smart attire,” in Proceedings of the content/uploads/2015/02/MPU-6000-Datasheet1.pdf
4th International Conference on Mobile Systems, Applications and [14] Arduino Nano Module. Available:
Services (MobiSys ’06), pp.110–123, June 2006. https://www.arduino.cc/en/uploads/Main/ArduinoNanoManual23.pdf
[9] Huynh, Quoc T., Uyen D. Nguyen, Lucia B. Irazabal, Nazanin [15] Bluetooth hc-05 Module. Available:
Ghassemian, and Binh Q. Tran, “Optimization of an Accelerometer and https://www.rcscomponents.kiev.ua/datasheets/hc_hc-05-user-
Gyroscope-Based Fall Detection Algorithm,” Journal of Sensors, Vol instructions-bluetooth.pdf
2015, 2015.

91

View publication stats