Development of Earthquake Early Warning System

Using ADXL335 Accelerometer

Yoga Priyana1, Folkes Laumal2, Emir Husni3
1,3
School of Electrical Engineering & Informatics, Institut Teknologi Bandung, Bandung 40132, Indonesia
2
Electrical Engineering, Politeknik Negeri Kupang, Kupang, Indonesia

Indonesia is an archipelago located at three earthquake belts. This condition cause an earthquake can occur anytime and
threaten human life. A quick and accurate early warning system by using the seismic wave data processing is required so, the
number of victims affected by the earthquake can be shortened. Here, ADXL335 accelerometers are used as seismic sensors
with an Arduino minimum system. The results show that when the first earthquake’s vibration occurs, P wave data detected
by the ADXL335 sensor is successfully buffered, calibrated, transmitted and displayed on the server. When there are errors
on the transmission, server will request for retransmission. The alarm of the earthquake early warning system will be
activated if there are at least three sensors from different locations successfully transmit P wave data with the same scale.
This is needed to prevent fake seismic waves.

Keywords: Early warning system, earthquake, ADXL335.

1. INTRODUCTION information given, the more loose the time for the
population to respond.
Indonesia is a confluence of three major tectonic Currently, Indonesia has seismograph earthquake
plates, there is an Indo-Australian, Eurasian and Pacific. sensors. Seismograph has sensitive instruments that can
Around the meeting point is the accumulation of plate detect seismic waves generated by earthquakes. Seismic
collision energy and when the layer of earth not able to waves that occurred during the earthquake displayed as a
hold energy anymore, it will release during energy named wavy line on the seismogram.
earthquake (Primary wave and Secondary wave). The With the evolution of IC technology, many
experience of the earthquake in some areas has provided microcontroller are produced. The microcontroller is
thousands of lives and properties, so its necessary to equipped with internal memory blocks, accumulator,
measures of mitigation to reduce the earthquake risk. In ALU, RAM, I/O ports and interrupt ports that support
short time when the earthquake happens, can be used to package as a medium for reading, processing and writing
provide early warning to the community through the data from and to other components, so there are some
development of early warning system based on the arrival microcontroller-based sensor systems being made with
of the P wave, the first wave that arrives before the real the more complete working principle1. One of a circuit
earthquake happens (S wave). integrated developed with microcontroller is Arduino.
*
Email Address: [email protected]

An earthquake early warning is personalized in alarm

and running text that executed on a sensor system.
Preparedness and speed of response are necessary
because of narrow time from passing information to
alleged actual of arrival earthquake. The earlier
There are three types of accelerometers: a single axis and pounding supersonic aircraft. There are two types of
acceleration to detect movement of an object in one seismic waves, namely Primary waves (P) and Secondary
direction, 2-axis acceleration to detect object motion in 2- Waves (S). Primary wave that occurs when the initial
directions and 3-axis acceleration to detect motion 3- vibrations occur and propagate as longitudinally.
directions. Secondary wave propagates like as a wave of water that
ADXL is 3-axis accelerometer sensor with 3.3 volt moves up and down with speed about 4 km/Sec and
analog input. In that respect are several series of these slower than Primary. Secondary wave has a large
accelerometers, including ADXL330, ADXL335 and amplitude so it has the power to destroy buildings or
ADXL345. The accelerometer has ± 3g sensitivity of each cliffs6.
axis (x, y, z) and can be used to measure the static and
dynamic acceleration, object motion, collisions and 2.3 UDP Socket
vibrations3,4.
This research aims to develop an earthquake early UDP socket is the TCP / IP protocol that
warning system using ADXL335 accelerometer through connectionless and unreliable. Connectionless meaning
several problems, namely: that data sent without to negotiate between two hosts and
1. How to design an earthquake early warning system unreliable meaning that the message sent without
using the ADXL335 accelerometer, accompanied by an acknowledgment. Application layer
2. How to make calibration of ADXL335 in order to protocol that works with UDP socket must perform the
obtain the reference as an indicator of the vibration recovery of the message that is lost during transmission7,8.
occurrence, UDP have 32 bit data format as indicated in Figure 1
3. How to make transmission of data from sensor to with maximum data transfer is 65 KB every time send.
client and client to server using communication
medium,
4. How to handle loss/error occurred when data sent to
the server,
5. How to declare the earthquake early warning.
The research is expected to contribute in seismology
for giving early warning when an earthquake will come,
machinery vibration analysis, object tilt and collision
analysis. Fig.1. UDP Data Format7
3. DESIGN
2. SURVEY AND METHOD
In general, early warning system in this research
2.1 Early warning system provides information based on the data primary waves
The Early Warning System is a series of systems to that first received by the system so that rescue efforts can
inform the public about the emergence of a natural event be done as early before the actual earthquake arrives (S
(natural signs). Early warning of disaster was act of wave). P waves propagate in longitudinal and S wave in
providing information in communicating with easily transfersal. Arrival times of S wave typically ranging
understood language. In critical condition, manifested in from seconds to minutes depending on sensor distance
the form of an early warning is alarming. The earlier from the earthquake epicenter6.
information submitted, people experience more time to Here, the design of earthquake early warning system
respond and rescue of their lifetimes. divided in two parts, the Base Station and the Server. In
Early warning information obtained in two ways, the both parts, there are 2 users with an identity and
conventional and modern. Conventionally, disastrous responsibility for observing data processing. Specifically,
introduction is done by the introduction of natural in the server, the user has an authorized to make sure the
phenomena that prior to the disaster, which is tailored to truth of early warning.
the characteristics of disaster. Modern ways, disastrous The Base Station consists of 2 parts and placed near
introduction are done with the monitoring atmosphere to the earthquake points, namely:
activity periodically by satellite or high-tech equipment. 1) Seismic sensors which are placed against on the
People who living in prone areas should be empowered cement floor and consists of Arduino as minimum system,
and respond to natural disasters toll system so that ADXL335 as vibration sensor for detects vibration in 3
number of victims can be slimmed down5. axes3 and EM411 as a position sensor for detects the
distance from GPS satellites and convert it using the
2.2 Seismic waves NMEA 0183 standard9. All of that are connected through
I/O pins on each element. On the Arduino equipped with
Seismic waves are waves that propagate inside or 32 KB ATmega328 microcontroller, the block amplifier
surface on earth from seismic sources such as from and ADC to strengthen and data convert.
annearthquakee. Volcanic eruptions, landslides, storms 2) Client with a complete database system to
2
accommodate all sensor data send by an Arduino.
Communication from earthquake sensor to client using
1.5 meters of USB cable.
The design of Base Station can be seen in Figure 2.

Fig.2. Design of Base Station

The client then connected to server using 13.5 Mbps
max internet connection without the barrier. Location of
server placed in the government building that handles of
the seismic. The server also has a database system that
will accommodate data from all clients that connected to
it. Communication client to server using UDP socket to
accelerate data transmission. Format data applied to the
system are 32 bits because the architecture of
microcontroller supports of this format.
Full design of the sensor system is shown in Figure 3.

Fig.4. Communication Sensor to Arduino

From calculations, If there is no vibration,
ADXL335 gives the voltage value 0.000, 0.000, 0.000 for
x-axis, y-axis and z-axis and Acceleration is 0.000.
Fig.3. Full Design of Earthquake Sensor Vx_axis, Vy_axis and Vz_axis values, then used as
reference for chart display, Vy_axis value and its changes
Design of earthquake sensor system consists of as a reference for determining scale whenever an
hardware and software. Hardware design includes of earthquake occurred. When no vibration, 3-axis voltage
interconnection sensor to Arduino, Arduino to client and value will stabilize at the calibration value. But when
client to server. trigger given (vibration), the voltage value will be
The design of software includes of communication changed.
data sensors (AXL335, EM411) to Arduino, Arduino to A communication client - server designed with UDP
the client and node to server. Communication sensor to socket protocol in MYSQL. The selection of this protocol
Arduino as in Figure 4. because it is faster, does not require back and a
There are 2 data taken from sensor, namely position handshake7.
data from EM411 and vibration data (x,y,z) from Data send as serial start with latitude, longitude, x-
ADXL335. Foremost, the Arduino will taken position axis, y-axis and z-axis with maximum data 216-1 – 20
data and putting into port and second, Arduino has taken byte header – 8 byte UDP header = 65535 bytes (65 KB)
vibration data (x, y, z) and store in buffer for calibrating. every transmitter. Identity of client and server using local
The Arduino will calibrate data in buffer with IP with port 3306 which support mysql.
specified zero_G and scale values. Zero_G is the axis Handling of loss/error data is designed in two phases:
value when the sensor is in a stationary position without detection and correction error. Detection phase performed
reversed and the scale is the average of change in each with comparing vibration data and its duplicate. If it is
axis. Based on calculations, the value of zero_G are 384, same, the true status will appear on the web. Only if one
383 and 473 for each axis and scale is 102.3. One vector is empty, and so the condition of loss data appears.
formula used to generate x, y, z voltage and Acceleration Correction phase is performed by requiring the client to
(A)10,11.
resend the correct information.

Fig.6. Earthquake Sensor.

Fig.5. Determine earthquake early warning

Fig.7. Sensor, Client and Server Connections
Figure 5 shows the determining of early warning
Data from the sensor are sent to the Arduino every
status. The server need 3 vibration values from the 3
0.01 seconds. The information will be sent to buffer for
sensors at different locations. Each sensor will sends
calibration. Three axis calibration results show that
vibration data (x-axis, y-axis, z-axis) to database and
vibration values are not always stable when no vibration
server will check and match the first vibration data on a
condition (0.000, 0.000, 0.000). Sometimes changes to
specified range of scale. If all sensors give the same scale,
0.010 or -0.010. But from characteristics of ADXL335,
the server will declare the earthquake early warning and
Noise Density founded for Xaxis, Yaxis at 150µm and
activate of alarm. This method is used as avoiding a fake
Zaxis at 300 µm3, so Peak Noise Density for Xaxis and
signals.
Yaxis are 0.084 and 0.064 for Zaxis. This result is still
Scales that used to determine of an early warning
under Peak Noise Density. When there is no vibration,
system as indicated in Table 1.
data is not taken.
Table.1. Conversion of 3-axis value from Zerro_G to The next test is shaking the sensor to ensure the
Scale. communication process among sensors, Arduino, client
Before After Multiply End Scale and server are correct and graphs vibration appears. The
calibration calibration factor in use result shows that vibration data are successful sent to
383 – 390 0.000 – 0.010 100 1 client every 0.01 second and protocol functions is
390 – 397 0.010 – 0.020 100 2 successful. Next, the first vibrations of data successfully
397 – 405 0.020 – 0.030 100 3 send to the server every 0.01 second and displayed in a
405 – 412 0.030 – 0.040 100 4 graph as Figure 8 and 9
412 – 419 0.040 – 0.050 100 5
419 – 426 0.050 – 0.060 100 6
426 – 433 0.060 – 0.070 100 7
433 – 440 0.070 – 0.080 100 8
440 – 448 0.080 – 0.090 100 9
448 – 455 0.090 – 0.100 100 10
455 – 463 0.100 – 0.110 100 11
463 – 473 0.110 – 0.120 100 12

4. RESULT AND DISCUSSION

With 1.5 meters USB cable, the earthquake sensor Fig.8. Vibration display in the client
will connect to the client. After the client turn on, the
internet Wifi will be active. The earthquake sensor
connecting with other devices is shown in Figure 6 and
Figure 7.

4
 condition. The 3-axis value is processed with other
variables in the database. Communication from the sensor
to the client is performed by the USB cable and from the
client to the server by internet using the UDP protocol.
Handling of loss/error is done with detection and
correction. Error detection is done by duplicating data
before sending and error correction by asking the client to
resend the correct data. An earthquake early warning will
be declared if there are 3 Base Stations (sensors) sending
first vibration data (P wave) with the sama scale.
Fig.9. Vibration display in the server
When first vibrations detected, this system only
If the loss/error data event is occurred, the server needs 3.03 secs to activate early warning status and make
will display the loss status on the web and will be alarm ON. This time is measured in conditions as follows:
repaired after client resend the correct data. connection sensor to the client using 1.5 meter USB cable,
On the server, status of earthquake early warning is connecting client to server using 13.5 Mbps speed
active with the reference of the first vibration data from 3 Internet network.
different base stations. First vibrations data is P-wafe data.
This data will be stored in the database. The server does
not directly declare the warning status because it still REFERENCES
have to wait for the arrival of the P-wave data from 2 base
stations. There were 3 seconds time delay given to [1] Raj Kamal, Microcontrollers: Architecture, Programming,
Interfacing and System Design, 2nd Ed., India : Pearson
adjusting the data to other sensors. If 2 other sensor data
Education, 2011.
show the same P-wave value with the same scale, then the [2] Ashish G.Bajaj and Vikran R. Bajaj, Build IEEE 802.15.4
server will give early warning and activate the alarm. But Wireless Sensor Network for Emergency Response Notification
if one of the sensors is different, then the server assumes for Indoor Situations, International Journal of Engineering
nothing happened. Total time for sending the first Trend in Electrical and Electronics (IJETEE), vol. 1, Issue. 2,
March. 2013.
vibration data from the sensor to the server and alarm [3] ___________, Small, Low Power, 3-Axis ± 3g Accelerometer,
activated is 0.01s + 0.01s + 0.01s + 3 s = 3.03 s. ADXL335, Norwood, MA 02062-9106, USA : Analog Devices,
Location of all sensors can be viewed through the Inc. All rights reserved, 2009.
map. When the sensors indicate warning, the marker will [4] Jonathan Bernstein, Low-Noise MEMS Vibration Sensor for
Geophysical Applications, Journal of Microelectromechanical
be red, but will be green if nothing happened. Display of
System., vol. 8, Dec. 1999.
sensor location on the map is shown in Figure 10. [5] Sarwidi, Upaya Pengurangan Dampak Bencana Gempa Bumi,
Lokakarya Nasional : Pengembangan Sistem Peringatan Dini,
PSBA UGM-PMI Pusat, Yogyakarta, 2001.
[6] Afnimar, Seismologi, edisi pertama, Laboratorium Seismologi
Institut Teknologi Bandung, ITB, 2009
[7] Behrouz A. Forouzan, DeAnza College with Sophia Chung
Fegan, Data Communication and Networking, 4th Ed. , Mc Graw
Hill Higher Education, 2007.
[8] Rajkumar Buyya, Thamarai Selvi Somasundaram and Xingchen
Chu, Object Oriented Programming with Java Essentials and
Applications, New Delhi, India : McGraw-Hill Education (India)
Pvt Ltd.
[9] ___________, EM411 GPS Receiver Engine Board, Globalsat
Fig.10. Mapping Status of Sensor Technology Corporation, Doc Version 1.2.
[10] Frederic Camps, Sebastian Harase and Andre Monin, Numerical
Calibration for 3-axis Accelerometers and Magnetometer,
5. CONCLUSION CNRS/LAAS-University of Tolouse-France, IEEE., 978-1-4244-
3355-1, Sept. 2009.
Development of earthquake early warning system [11] Tadej Bwravs and Others, Three-Axial Accelerometer
can be made with the ADXL335 accelerometer and an Calibration using Kalman Filter Covariance Matrix for Online
Estimation of Optimal Sensor Orientation, IEEE., 0018-9456,
Arduino UNO system. The first vibration data can act as 2012.
reference of Primary wave. Calibration process is done by
taking the 3-axis value of ADXL335 when no vibration