International Journal of Advances in Science Engineering and Technology, ISSN: 2321-9009, Vol-4, Iss-3, Spl. Issue-2 Sep.

-2016

FALL DETECTION SYSTEM FOR ELDERLY PERSON USING

EMBEDDED SYSTEM
1
SHILPA R. KAWALE, 2G.M.MALWATKAR
1,2
ZEAL College Of Engineering And Research, Narhe, Pune, India
E-mail: [email protected], [email protected]

Abstract— In this paper proposed fall detection system is designed for elderly person monitoring that is based on various
sensors and operating through indoor and outdoor tracking using embedded system. With thresholds, accidental falls can be
detected in the tracking healthcare section positions. The readings from pulse rate sensor, temperature sensor and
accelerometer sensor is compared by ARM9.If any of the sensor reading cross the set value then fall is detected. Now a days
there are many technical solutions to care of the elderly person are implemented. The proposed system is ready to be
implemented on an elderly person.

Index Terms— Accelerometer, Pulse rate sensor, GPS, GSM, ARM9(LPC2929), RFID reader.

I. INTRODUCTION results; it will be found that the proposed fall

detection system achieved by a accuracy high of 90%
We know that the population of world is growing above, and the specificity and sensitivity are 95% [8].
very fast and there are lots of technical devices are
present in the market which gives solution to care of II. RELATED WORKS
elderly person in recent years [1]. An indoor and
outdoor tracking using embedded system contains In the field of medical, many research projects has
various electronics devices such as accelerometer, been developed for the health care of elderly person.
sensors, RFID reader, LPC 2929,GSM,LCD etc to Here we use accelerometers and different sensors.
improve the life of the elderly people[2].Today their With the purpose to detect successfully typically falls
are many technologies are available in the market by applying threshold to accelerations position. It has
which gives solution to help for the longer life of the following methods.
elderly person [3].Why detection fall is one of most
important topics in elderly person care of reasons are A. Wearable Detection Methods
fear and fall falling are important and detection fall Bagale.et al [4], presented an algorithms which is
can be apparatus fairly effectively with the today based on accelerometer, of fall detection on fall of
current technology available. The accelerometer is real-world.
used to detect the accurate movement of the person.
The accelerometers are accurately light in weight and Cho, Nguyen and Lee [7], used an accelerometer
it also inexpensive and limitations are safe results which is worn on the waist, by applying thresholds to
even in high acceleration [4].The population of the accretion, they detect accurate fall.
elderly people has been increasing from the last few
decades. This population is not self-determining B. Audio, Video Surveillance Detection Methods
which results in large number of accidents takes Lustrek and Kaluza [9], perform well and it tests and
place[5].Among these accidents, falls is a major can be found that more comfortable than worn
public health problem that affects every ten year of sensors body. However,the sensors which is worn on
millions of worldwide elderly people, yielding the body are most widely used method for fall
psychological and physical consequences. As we detection.
know that some of the elderly people is not able to lift
themselves on their own after falling; it is important Diraco and others [10], separated movement area
to detect such such types of events as soon as from image by using bayesian segmentation method
possible. Hence such type of problems problem is based on the threshold of distance of the 3
related to researchers in the field of health dimensional centroid human from the floor plane and
technologies [6]. as they move furher found that the fall had happened
when the human centroid is less than the specified
In this paper, fall detection system for elderly person height and situation are maintained more than 4s. The
with tracking of indoor and outdoor using embedded human threshold centroid value of 0.4m, accuracy
system environment proposed that based on global was 97.3% and the recognition rate of 80.0%. Human
system for mobile communication, accelerometer body data in stooping, standing, laying and sitting
sensors and radio frequency identification down analyzed them they collected and concluded
reader[7].This proposed system will be deployed in a groups three of threshold centroid and the recognition
system of prototype and experimentation of 30 trails rate & credibility.

Fall Detection System For Elderly Person Using Embedded System

93
International Journal of Advances in Science Engineering and Technology, ISSN: 2321-9009, Vol-4, Iss-3, Spl. Issue-2 Sep.-2016

C. Vision Based Methods the base station which is used for the tracking of the
Vision based methods denoted to the posture and person.
shape change.
Rougier et al.[11] analyzed deformation human
shape during a video sequence which is used to the
person’s silhouette.

D. Ambient Based Methods

Ambient based methods are usually depend on even
passive motion infrared sensors, acoustic sensors and
pressure sensors which is placed around the
caretaker’s houses.

Yan et al,[12] proposed a perceived nature invasive

method. They developed such a system in which it is
not necessary that the user should waer the sensors,
although was able to detect the location of the user
which is based on interaction observation with the
network of sensor home-installed.

Winkley, Jiang and Jiang [13] proposed a verity of

two components, base station and direct monitoring
Fig.1: Block diagram of fall detection system for elderly person
device. They find that the proposed classifier using embedded System.
outperforms the conventional classifiers with higher
distinguishing capability. In a particular system, they A. Real –Time Fall Detection
measured a real time skin/ambient temperatures .. A fall detection algorithm [3] suitable for real time,
by using the micro-controller of the accelerometer,
The video detection methods are better than the various sensor and various devices. The acceleration
ambient based methods and wearable based methods component along with the three axes as X, Y and Z
[11].These systems are often suffer from high risk of respectively [9].
privacy and also very high cost of implementing
cameras. So they used wearable sensor based By using information from an body temperature and
methods in such types of research. We successfully pulse rate and monitor the health parameters as well
detect fall with accuracy above 90% by applying as fall detection using sensors such as accelerometer,
threshold to acceleration of accelerometer which is the impacts of these falls can successfully be
worn on the body of the person. The accelerometer distinguished from activities of daily lives reducing
also resulting in 100% accurate fall detection using the false detection of falls. From the experimentation
various accelerometers and sensors. results of 30 trials, it is found that the proposed fall
detection system achieved a high accuracy of above
III. SYSTEM IMPLIMENTATION 90%, and specificity are 95%.
The block diagram of proposed fall detection system B. ARM9 LPC 2929
shown in fig.1.It consist of ower supply, pulse rate ARM9 is the heart of our system. Here we are
sensor, accelerometer sensor, temperature sensor, operating all the components using ARM9.It is 144
RFID reader, GPS, GSM,LCD display and buzzer. pin IC. The LPC2927/2929 combine an ARM968E-S
Accelerometer sensor is used to locate the exact CPU core with two integrated tightly coupled
movement of the person in X and Y direction. Pulse memory blocks, which operates at the frequencies of
rate is used to find the persons pulse .The temperature 125 MHz.It has three analoge to digital converters
sensor can be used to measure the exact body which is used for signal condishning. Ech of 10-bit
temperature of the person. The readings from all ADC. ARM7 is one of the widely used micro-
these sensors is given to the ADC converter which controller family in embedded system application.
converts the analog voltage into digital voltage. These section is humble effort for explaining basic
features of ARM9. The ARM is a family of
This ADC circuit is inbuilt in ARM9(LPC2929).The instruction set architectures for computer processors
readings from all these sensors continuously display based on a reduced (RISC) architecture. It also has a
on LCD and if any of the sensor readings goes flexible Clock Generation Unit(CGU) which provides
beyond the set value or if fall is detected then buzzer dynamic clock gating and scaling. It has104 general-
is automatically on and message is send to the server purpose I/O pins with programmable pull-up, pull-
and to the concerned person using the GSM modem. down, or bus keeper. As compared to ARM7 which
Also we are showing the google map using GPS to has only one power supply, CPU operating voltage

Fall Detection System For Elderly Person Using Embedded System

94
International Journal of Advances in Science Engineering and Technology, ISSN: 2321-9009, Vol-4, Iss-3, Spl. Issue-2 Sep.-2016

range of 3.0 V to 3.6 V.But ARM9 has dual power display the co-ordinates on LCD. We are sending
supply, CPU operating voltage: 1.8 V ± 5 % and I/O these co-ordinates to the concerned person and also
operating voltage: 2.7 V to 3.6 V. on server using GSM which displays the position on
google map.
C. Temperature Sensor
Using targeted circuit, skin temperature measurement IV. PROPOSED ENHANCED FALL DETECTION
is done. The temperature sensor gives an analog ALGORITHM
output and by using the ADC converter which is
inbuilt in ARM9 micro-controller the analog voltage The proposed enhanced fall detection algorithm is
is converted into digital voltage [14]. This sensor based on placing an accelerometer sensor on the
allows it to measure the external temperature of the user’s body. The person detection falls the X, Y and
skin, positioned in such a way that it is in contact Z direction of acceleration readings. The micro-
with the person’s skin. Hence, from the skin controller compares the readings and when these
temperature, the body temperature is measured. There readings cross threshold, then the fall is detected in
can be different method to estimate the exact body X, Y and Z direction [7].
temperature from skin temperature [15],but usually The tracking section determine the exact position of
the body temperature is 5.1 C which is higher than user indoor scenario by using an RFID reader and
skin temperature. RFID tags. Using RFID reader in a micro-controller
to detect the RFID tags and position on base station
D. Pulse Rate Sensor display.
A custom pulse rate sensor was designed to read the The outdoor tracking is done by using GPS receiver
patient’s pulse rate. The designed sensor is very small which reads the latitude and longitude co-ordinates to
and its cost is less. From the pulse rate of the person, micro-controller in every seconds and micro-
we can also measure the persons heart rate whose controller will read and display the co- ordinates on
technique is based on near infrared spectroscopy [10]. the LCD. Using GSM modem sending these co-
The pulse rate sensor is low cost method of ordinates to base station. The algorithms following
measuring the pulse rate. steps and the proposed method [3] for fall detection is
based on monitoring the magnitudes of three axial X,
E. Accelerometers Y and Z components of the accelaertion, magnitudes
To successfully detect different falls, we use 3-axial in the three planes, and the total magnitude. The fall
accelerometer[16].The accelerometer sensor find the detection is based on the observation that a fall is
movement of the person in X,Y, and Z direction. Two associated
methods of accelerometers are threshold and
orientation, both method gives accurate movement of Start
the person during fall. Init LCD,ADC AND Timer to 10 msec
Dispay project name on LCD
F. Health Monitoring “A”
Health monitoring can also be done for elderly person Read pulse rate
using different sensors. The readings from pulse rate “C”
and temperature sensor will continuously displayed Is time 10 sec→Y→”E”→Pulse rate*6→Store and
on LCD and as soon as any parameter crosses set Display Pulse rate on LCD→”B”
point the arm9 micro-controller will send an sms to N→Is pulse pin = high→Y→”C”
base station using GSM Modem [12]. “D”
Is time 10 sec →Y→”E”
G. Tracking Section N→Is pulse pin = Low→Y→”D”
In tracking section, we are doing indoor and outdoor N→Pulse rate +1→”C”
tracking. Read ADC0 channel 1
For indoor tracking we are using RFID reader and Store and display temperature on LCD
RFID tags. By using the RFID reader we can find the Delay ADC0 channel 1
exact position of the elderly person inside the home. Store and display temperature on LCD
Here the micro-controller detect the RFID tags using Delay 1 sec
the RFID reader and display the position to base Read the accelerometer reading
station using GSM modem. Is serial interrupt detected?(GPS) N”B”
Y Read the GPS coordinates and display on LCD
Here we are using GPS receiver for tracking the “B”
elderly person in outdoor scenario. The GPS can find Is fall detected? N “C”
the exact movement of the person by sending the co- Y Display fall detected Send SMS to base
ordinates in the form of latitude and longitude for station”A”
every 2 seconds. The micro controller contineously “C”
read these latitude and longitude co-ordinates and Is RFID detected? N”D”

Fall Detection System For Elderly Person Using Embedded System

95
International Journal of Advances in Science Engineering and Technology, ISSN: 2321-9009, Vol-4, Iss-3, Spl. Issue-2 Sep.-2016

Y Compare the RFID number and display the Above figure 2 shows the hardware module used in
indoor position on Visual basic (PC server) this project with all the components connected. As
“D” shown in the above figure the ARM9(LPC2929) is
Is time= 120 Sec N”A” entrench with temperature, pulse rate and
Y send SMS to Mobile No 1 (Temperature and accelerometer sensor, the readings are continuously
pulse rate and GPS coordinates) displayed on LCD. If any of the parameter exceeds
send SMS to Mobile No 2 (Temperature and pulse the set point then an sms is sent to the concern
rate and GPS coordinates) authorities.
“A” Figure3 shows the google map of GPS co-ordinates in
terms of latitude and longitude. When the fall is
The algorithm start then initial LCD, ADC and timer detected the GPS co-ordinates sends to the base
to 10 msec. Display the project name on LCD. station and shows the google map on the pc server.
Display the pulse rate on LCD. Read ADC0 channel1
then store and display temperature on LCD then delay Table1 shows the readings for indoor tracking and
1sec after read accelerometer reading. Then display table2 shows the readings for outdoor tracking as per
fall detect on LCD and send sms to base station and the monitoring system for elderly person using pulse
after 120sec send sms to mobile no.1 and mobile no.2 rate and temperature and accelerometer sensor. The
with temperature, pulse rate and fall detection table shows readings of 10 different elderly persons.
message.
Table 1: Readings for indoor tracking.
V. EXPERIMENTAL RESULTS

Table 2: Readings for outdoor tracking.

Fig.2: The fall detection system for elderly person using

embedded system

The implementation of realization “Fall detection

system for elderly person using embedded system” is
done successfully. The hardware module of the
proposed fall detection system for elderly people as
shown in figure2. CONCLUSION

In this paper, a fall detection system based on sensors

which is worn on the body of the elderly person are
implemented that successfully detected falls. By
making the use of the information from an body
sensors, we can monitor the health parameters of the
elderly person as well as successfully detect fall by
using an accelerometer sensor. Using GSM the
concerned person will receive the sms on his mobile
phone, in case any of the parameter goes beyond the
normal range or if fall is detected. Also by using GPS
the concerned person know the exact position of the
user. From the experimentation of 30 trials, it is
found that the proposed fall detection system
achieved a high accuracy of above 90%, and the
Fig.3: Google map of GPS co-ordinate. sensitivity and specificity are 95%.

Fall Detection System For Elderly Person Using Embedded System

96
International Journal of Advances in Science Engineering and Technology, ISSN: 2321-9009, Vol-4, Iss-3, Spl. Issue-2 Sep.-2016

REFERENCES [9] Lustrek and Kaluza, “Fall detection and acivity

recognition with machine learning,
”Informatica;2009;33:205-12.
[1] Liyang Zhau,Zhen Liu,Jian Guo, “A survey of fall
[10] Diraco G,Leone A, Siciliano P, “An active vision system
detection algorithm for elderly health monitoring,” in
for fall detection and posture recogniation in elderly
Proc. 2015 fifth International IEEE Eng.pp, 270-274,
helthcare,” [C] // Proceeding of the Conference on Design,
2015.
Automation and Test in Europe. [S. 1.]: CDATU,
[2] Wang Ye, Bai Xiang-yu, “Research of fall detection and
2010:1536-1541.
alarm applications for the elderly,” in Proc. 2013
[11] C. Rougier, J. Meunier, A.S. Arnaud, and J. Rousseau,
International IEEE Eng. pp, 615-619, 2013.
“Robust video surveillance for fall detection based on
[3] Praveen Kumar,Prem C. Pandey, “A wearable inertial
human shape deformation,” IEEE Trans. Circ. Syst. for
sensing device for fall Detection and motion tracking,” in
Video Tech., vol. 21, no. 5, pp. 611-622, May 2011.
Proc. 2013 annual International IEEE Eng. pp, 978-1-
[12] H.R. Yan, H.W. Huo, Y.Z. Xu, and M. Gidlund, “Wireless
4799-2275-8, 2013.
sensor network based E-health system: implementation
[4] Bagala et al., Becker C,“Evalution of accelerometer-based
and experimental results,” IEEE Trans. Consumer
fall detection algorithms on real-world falls,” PLoS ONE
Electron., vol. 56, no. 4, pp. 2288-2295, Nov. 2010.
7(5):e37062.doi:10.1371/journal.pone.0037062.
[13] Chin-Feng Lai, Member, IEEE, Sung-Yen Chang, Han-
[5] Jin Wang, Zhongqi Zhang,Bin Li,Sungyoung Lee,and
Chieh Cha, “Detection of cognitive injured body region
R.Simon Sherratt, “An enhanced fall detection system for
using multiple triaxial accelerometers for elderly falling,”
elderly person monitoring using consumer home
IEEE Sensor Journal., vol. 11, no. 3, pp. 763-770, March
networks,” in Proc. 2014 International IEEE Eng. pp,23-
2011
29, 2014.
[14] Can Tunca, Sinan Isik, M. Yunus Donmez, and Cem
[6] Paulo Armando Cavalcante Aguilar, Jerome Boudy, Dan
Ersoy, “Distributed mobile sink routing for wireless sensor
Istrate, Bernadette Dorizzi, and Joao Cesar Moura Mota,
networks: A survey,” IEEE Communications surveys and
“A dynamic evidential network for fall detection,” IEEE
Tutorials. vol. 16, no. 2, pp. 877-897, 2014.
Journal.in Biomedical and Health informatics, vol. 18, no.
[15] Karandeep Malhi, Subhas Chandra Mukhopadhyay, , Julia
4, pp. 1103-1113, July 2014.
Schnepper, Mathias Haefke andHartmut Ewald, “A
[7] Nguyen,T. Cho and Lee,“Automatic fall detection using
zigbee-based wearable physiological parameters
wearable biomedical signal measurment terminal,”In
monitoring system,” IEEE Sensor Journal., vol. 12, no. 3,
Proceedings of 31st Annual international conference of
pp. 423-430, March 2012.
IEEE EMBS,5203,5206,2009.
[16] Mitja Lustrek,Hristijan Gjoreski,Simon Kozina,Bozidara
[8] Q. Zhang, L. Ren, and W. Shi, “HONEY: A Multimodality
Cvetkovic.Violeta Mirchevska,Matjaz Gams,“Detecting
Fall Detection and Telecare System,” Telemed J E Health,
falls with location sensors and accelerometers,” Artificial
vol. 19, no. 5, pp. 415-429, 2013.
intelligence conference,pp,1662-1667,2011.



Fall Detection System For Elderly Person Using Embedded System

97