A MEMS-based commutation module with vibration sensor for wireless sensor network-based tunnel-blasting monitoring

Abstract

The authors have developed a tiny, low-cost accelerometer that utilizesa semiconductor fabrication technology called MEMS (Micro Electro-mechanical System), for building a sensor network which would have a large quantity of such sensors deployed all over the host structure to be monitored. Due to the small dimension and extremely low power consumption, the sensor device is well-suited for such networks, where the supply of external power is very constrained. The authors have designed both a sensor device, and a sensor module having wireless communication capability built around it, and tested them in real-world tunnels, as well as in test labs. The test results showed that our prototype performed adequately for its intended use.

Full text

KSCE Journal of Civil Engineering (0000) 00(0):1-10
DOI 10.1007/s12205-013-0108-4
− 1 −
www.springer.com/12205
Information Technology
A MEMS-based Commutation Module with Vibration Sensor for
Wireless Sensor Network-based Tunnel-blasting Monitoring
Jungyeol Kim*, Soonwook Kwon**, Seunghee Park***, and Youngsuk Kim****
Received Feburary 27, 2012/Accepted January 21, 2013
··································································································································································································································
Abstract
The authors have developed a tiny, low-cost accelerometer that utilizesa semiconductor fabrication technology called MEMS (Micro
Electro-mechanical System), for building a sensor network which would have a large quantity of such sensors deployed all over the
host structure to be monitored. Due to the small dimension and extremely low power consumption, the sensor device is well-suited
for such networks, where the supply of external power is very constrained. The authors have designed both a sensor device, and a
sensor module having wireless communication capability built around it, and tested them in real-world tunnels, as well as in test labs.
The test results showed that our prototype performed adequately for its intended use.
Keywords: vibration sensor, tunnel, structural health monitoring, MEMS, sensor network
··································································································································································································································
1. Introduction
Blasting work using explosives produces extensive noise and
vibration that can negatively affect nearby residents and facilities.
For that reason, many jurisdictional authorities require monitoring
of such works. To monitor the effects of the explosion, the
vibration of bedrocks and artificial structures in the nearby area
is measured, and the measurement value is analyzed for assessment
of possible effects from the explosion. Currently such measurement
job relies on very expensive sensors, making it uneconomical
when considered for building a sensor network.
Our research that is described in this paper has two goals:
first, to develop an inexpensiveyetaccurate accelerometer based
on MEMS (Micro Electro-Mechanical Systems) technology for
small dimension and cost reduction; second, to develop a
wireless communication module for easier deployment of the
sensor network based on our MEMS sensors.However, the
communication module can also be configured for other
sensor types.
This paper is structured as follows: first, it briefly reviews the
measuring process of tunnels, and the available technology for
tunnel measurement-specifically for wireless sensor networks,
and the current state-of-the-art regarding sensor technology
suited toa network approach, featuring low power consumption,
small physical dimension, etc. Second, the authors describe a
technological overview, with test result of our wireless com-
munication node designed for the sensor network monitoring
vibration of a tunnel. Third, the authors present a sensor device
designed by the authors using Micro Electromechanical System
(MEMS) technology, together with test results of our prototype.
Finally, alternative power supply methodology, which has been
the subject of active research, is discussed, together with
preliminary testing of available technologies.
2. Analysis of (Tunnel) Measuring Process
To provide a technical review of current practices in measuring
tunnels, the authors collected various manuals and guides for
measuring tunnels from published sources, such as the national
standard specification of civil engineering projects, Korean
Standard for testing and measurement, tunneling manuals from
several private firms, etc.
The authors also tested existing technological elements for
measuring (Fig. 1). The elements were evaluated using the following
criteria: application area, sensor type (conventional versus MEMS),
communication method (by-wire vs. wireless), and measuring
method (dynamic vs. static).
Although reliable and precise, existing sensor technologies
revealed shortcomings when it came to the communication criterion,
as they demonstrated inconvenience, due to cable connection
TECHNICAL NOTE
*Senior Researcher, Korea Institute of Construction Technology, Koyang, Korea (E-mail: jrkim@kict.re.kr)
**Member, Associate Professor, Dept. of Civil, Architectural and Environmental System Engineering, Sungkyunkwan University, Suwon 440-746, Korea
(Corresponding Author, E-mail: swkwon@skku.edu)
***Member, Assistant Professor, Dept. of Civil, Architectural and Environmental System Engineering, Sungkyunkwan University, Suwon 440-746, Korea
(E-mail: shparkpc@skku.edu)
****Member, Professor, Dept. of Architectural Engineering, Inha University, Incheon, Korea (E-mail: youngsuk@inha.ac.kr)
우편번호 보내 주세요 ..

Jungyeol Kim, Soonwook Kwon, Seunghee Park, and Youngsuk Kim
− 2 − KSCE Journal of Civil Engineering
problems. The test led us to decide that all sensors used in our
system must utilize wireless communication (Zigbee) for
transmitting sensor readings; also, for the sensors, existing
sensors were not to be replaced with MEMS counterparts, if
they offered better results than the newer, less matured
MEMS breed.
3. Review of State-of-the-art Researches
Even for MEMS, there are many technologies that can be
used for making vibration sensors. Available MEMS sensors
for measuring acceleration, force and deformation include
piezoelectric, piezo-resistive, capacitive, resonance, optical
and magnetic.The authors surveyed research papers published
in international journals and conference proceedings from
2005 onwards that cover various MEMS sensors, and the
wireless sensor networks that employ them. The authors also
reviewed some non-MEMS sensor technologies applied to
health monitoring of construction structures, including
tunnels (Table 1).
For the sensor technologies, the authors used performance
criteria (e.g. accuracy, measurable range, etc.) for making a
comparison between them.
Many of the currently available MEMS sensors are based on
the electro-mechanical type, which measures the varying capacitance
of a moving mass; to increase sensitivity, the mass is shaped like
a comb. Such sensors can have various application areas other
than accelerometers, such as inclination sensors (Yu et al., 2009),
etc. As their comb-like shape is very vulnerable to external
Table 1. Review of Sensor Technologies for the Tunnel Measurement System
Source Tech highlights What is sensed Sensor type
Sensor performance
Power
characteristics
Measurable range Accuracy
Alfado, Weiss
et al. (2009)
Bone-implantable
multi-axis CMOS-MEMS
stress sensor
Stress
(force)
Piezo-resistive ~ 250 kPa
100 Pa, w/ 1Hz
interval, 15.2dB SNR
3.3v,
775 uW /
measure-
ment
Azevedo et al.
(2007)
Silicon carbide (SiC)
MEMS resonant strain
sensor-silicon comb-driven
double-ended tuning fork
(CDDETF) type
Strain
(deformation) -
measuring resonant
frequency, which
varies over applied
strain
Mechanical-Capac-
itive
Survivable up
to 10000 g
acceleration
Sensitivity: 66 Hz/ue,
Resolution: 0.11ue
over 10-20k Hz,
-109 dBc/Hz noise
floor
N/A
Kon and
Horowitz
(2008)
High-res MEMS
piezoelectric strain sensor
using Zinc Oxide (ZnO)
layer
Strain
(deformation)
Piezoresistive N/A
40.3 ne at 2140 Hz,
28.7 ne at 10 KHz
N/A
Yu et al.
(2009)
Wireless inclination
sensor system, which
uses VTI’s SCA100T
MEMS inclinometer
Inclination
Mechanical-capaci-
tive, using over-
damped elements
to avoid vibration
±30 deg
(SCA100T-
D01), ±90 deg
(SCA100T-D02),
Survivable up to
20000G
Resolution: 0.0025
deg at 10 Hz
Sensitivity: 70 mV/deg
(D01), 35 mV/deg
(D02)
5v
Ioppolo,
Ötügen et al.
(2008, 2009)
Micro-optical wall shear
stress sensor using whis-
pering gallery mode
(WGM) resonator - optical
MEMS
Shear stress
Optical - using non
moving deformation
of light-transmitting
material, sensing its
optical resonance
Up to 1 kHz 0.01Pa N/A
Kang, Schulz
et al. (2006)
Polymer-reinforced
carbon nano-tube (CNT)
strain sensor - a resistive
type electrical strain sensor
Strain Resistive
±6000
micro-strain
N/A 20v
Leng et al.
(2005)
Fiber optic sensors using
extrinsic Fabry-Perot
interferometric sensors
and fiber Bragg grating
sensors (as a part of a
fiber-optic sensor network)
strain optical
±500
micro-strain,
when up to 14
MPa force is
applied
N/A N/A
Fig. 1. Test of Existing Sensor Technology

A MEMS-based Commutation Module with Vibration Sensor for Wireless Sensor Network-based Tunnel-blasting Monitoring
Vol. 00, No. 0 / 000 0000 − 3 −
shock, the capacitive-type sensors have inherent drawback in
terms of durability, which have been addressed in (Azevedo et
al., 2007).
Another type of MEMS sensor uses the piezo-resistive property
of a metal oxide layer. For such sensors, Kon et al. (2008)
demonstrated a simply shaped (compared to the comb-shaped)
piezo-resistive sensor that was sensitive to higher frequency
vibration. Alfado et al. (2009) took a similar approach, which
implemented multi-axis sensors on a single device. In general,
those sensors are able to measure physical deformation (i.e. not
accompanying movement) that should be limited to a microscopic
scale.
An optical method is seen in two series of studies (Ioppolo et
al., 2008; Ioppolo et al., 2009), which measure the varying optical
resonance of a light-transmitting mass caused by its deformation,
thus sensing strain. The varying intensity of a laser beam
transmitted through an optical fiber is also used for a strain
sensor (Leng et al., 2006), though optical fiber is vulnerable to
damage caused by external force, so a protection mechanism is
necessary.
Though not classified as MEMS, Kang et al. (2006) showed
that Carbon Nano-Tube (CNT) can be used for sensing external
force, which can be embedded into the building structure.
Although there are numerous methods known for MEMS
sensors, sensing movement (and thus acceleration) is best
implemented with a varying-capacitance type sensor; however,
there are several drawbacks if we consider such sensors for
sensing vibration: first, the sensor is vulnerable to excessive
shock, which can be caused from blast, so it must be designed
for durability. Second, a durable design may decrease the
sensitivity of the sensor when measuring smaller movement;
to address this issue, the authors designed a cantilever-type
opto-mechanical sensor that provides both durability and
sensitivity.
4. Development of MEMS-based Vibration Sensor
The authors developed two different MEMS sensor types-a
conventional, comb-type, electro-mechanical sensor, and a
cantilever-type opto-mechanical one. The latter was designed
after two iterations of the former studies (Kim et al., 2005; Kwon
et al., 2006), because our electro-mechanical design didn’t achieve
sufficient performance for measuring micro-acceleration. The
developmental details of those sensor types are described in the
following sections.
4.1 Design and Fabrication
The authorsfirst designed a single axis MEMS accelerometer
capable of sensing planar acceleration. It is a differential type,
using comb-shaped electrodes for applying the area-variation
method. The mass is suspended with a folded-beam spring,
which is illustrated in Fig. 2. Our design was evaluated with
ANSYS software for stresses applied to the spring and the
pendulum mass, and the resonance mode of the sensor as well
(Fig. 3). The result was used for determination of the size of the
micro-components and operational frequency range of the sensor,
respectively.
The authors fabricated our sensors several times, as we revised
the design. Later revisions have addressed performance issues in
earlier ones, such as linearity, sensitivity, noise. For the last
(fourth) revision, three-axis vibration sensors utilizing three discrete
single-axis sensors were developed, accompanied by improved
circuitry and communication performance; as a result, the authors
had a mass of 120 µg, with springs whose dimensionsare 450 µm ×
4 µm, with maximum detectable frequency of 1 kHz. Fig. 4
shows a picture of our sensor sample taken with a Scanning
Electron Microscope (SEM).
Fig. 2. Design Sketch of Our Single-axis MEMS Accelerometer
Fig. 3. ANSYS Simulation of the MEMS Sensor Design: (Left)
Stress Analysis for the Mass and the Spring; (Right) Analy-
sis of Its First-order Mode of Resonance (999 Hz)
Fig. 4. Scan Electron Microscope (SEM) Image of our MEMS Device

Jungyeol Kim, Soonwook Kwon, Seunghee Park, and Youngsuk Kim
− 4 − KSCE Journal of Civil Engineering
4.2 Making a 3-axis Vibration Sensor and a Communica-
tion Module
Using the single axis MEMS accelerometer we developed, the
authors built a 3-axis vibration sensor, by combining three sensor
modules with newly-developed circuitry (Fig. 5). One of the
issues in combining the sensors was accommodating the three
sensor channels into the limited bandwidth of the communication
module.
A communication module consists of multiple transmitters
embedded in MEMS vibration sensors, and a receiver that
collects data from the transmitters. Both transmitters and the
receiver use Zigbee communication protocol (based on IEEE
802.15.4), which enables error-free data transfer from multiple
transmitters over the 2.4 GHz radio band.
Performance of a wireless communication module is a
constraining factor for sampling rate of an individual sensor,
number of available transmitters, power consumption, etc. To
extract maximum performance from the communication module,
each transmitter node was set to send data in a 200 ms interval,
with the help of internal memory for buffering sensor readings
during the interval. Also, the sensor was set to measure at least
300 Hz of vibration frequency (i.e. sampling rate), as the burst-mode
data transmission used in our system enabled us to squeeze out more
bandwidth by reducing overheads in the data packet-which, at a
300 Hz sampling rate, allowed simultaneoustransmission of ten
single-axis sensor readings, or three triple-axis sensor readings.
For operation of overall sensor network (dubbed as USN,
ubiquitous sensor network), the authors developed internal
measuring software and network software on TinyOS 2.0, which
also contributed to low-power performance.
Our wireless data transmitter consists of a sensor module
(which itself consists of a MEMS sensor and a driver circuitry), a
signal processing circuitry, a Zigbee communication module and
a power supply. Fig. 6 shows an actual transmitter module.
Input voltage from the MEMS sensor (i.e. sensor output signal) is
amplified first, then is fed to an embedded low-pass filter for
removing noise, An embedded analog-to-digital converter generates
a stream of 16-bit word data from the amplified, noise-free
analog signal. The data stream is then to be transmitted via a
Zigbee communication module (which is built around a CC2420
chip from Texas Instruments). The module is configured to
buffer the data in its embedded memory for allowing burst-mode
transmission, which allows the achieving of a 300 Hz sampling
rate over the rathernarrow bandwidth of Zigbee.
A receiver module communicates with multiple sensor modules
over 2.4 GHz Zigbee protocol. When it receives data from the
sensor modules, it passes the collected data to a host computer
via USB. A data receiver can be connected up to three transmitters,
each of which has three data channels assigned for X, Y, and Z
axes.
4.3 Laboratory and Field Tests of the Sensor Module
The authors analyzed performance of our prototype sensor by
comparing its performance with existing sensors. For the
comparison, a commercial ICP accelerometer (capable of measuring
-0.5G to +0.5G) and a MEMS accelerometer manufactured by
Analog Devices (capable of measuring -1.7G to +1.7G) were
Fig. 5. 3-axis Vibration Sensor Module Built with Three MEMS
Accelerometers
Fig. 6. A Wireless Communication Module, with Open Protective
Housing
Fig. 7. Test Result: Triangular Dots Show Readings from Our Pro-
totype, Square Dots Show Readings from a Commercial
MEMS Sensor (for Comparison), Small Diamond Dots Show
Readings from the ICP Sensor (as a Reference)

A MEMS-based Commutation Module with Vibration Sensor for Wireless Sensor Network-based Tunnel-blasting Monitoring
Vol. 00, No. 0 / 000 0000 − 5 −
used. The ICP sensor was selected as a reference, because it has a
very low noise level, and is very sensitive. For the test, the
authors installed these sensors on the same elastomeric test bed;
then, the authors produced artificial waves, having variable
magnitudes ranging from 0.5 mm up to 48 mm, for measurement.
Fig. 7 illustrates the test result. Since ICP sensors cannot measure
accelerations greater than ±0.5 g, only two MEMS sensors were
used for measurements over that value.
From the test result, it is shown that the measurement value
from our prototype is closer to the reference value (from the ICP
sensor) than the commercial MEMS sensor. The average error
ratio of our prototype was 8.2%, while that of the commercial
MEMS sensor was 10.1%. Both MEMS sensors behaved worse
as the acceleration increased, yet the commercial sensor had
more error than ours. On the other hand, those MEMS sensors
showed difficulty in measuring acceleration less than 0.1 g, due
to their inherent noise level.The full comparison of commercial
MEMS sensor and our prototype in Fig. 8 shows similar trends
among them.
The authors conducted a test of our prototype sensor module at
an actual blasting site located in Daesan, South Korea (see Fig. 9
for pictures of the actual test scene). Sensor device and its
communication module showed their performance and robustness
in the actual site.
5. Opto-mechanical Vibration Sensor Design
In order to improve our MEMS sensor, the authors iterated our
design and fabrication effortsfive times. For the first three iterations,
the authors tried to improve various performance aspects, such as
linearity, sensitivity, and noise rejection.For the fourth iteration,
the authors developed a 3-axis accelerometer, utilizing our
single-axis sensor devices. For the fifth and final iteration, the
authors designed a new sensor type for measuring micro
acceleration less than 100 mg, which was not possible with our
varying capacitance type design, due to higher noise floor. Such
a limit was considered unbeatable for that type of MEMS
accelerometers, including commercial devices.
5.1 Designing the Optic-MEMS Accelerometer
For measuring the micro acceleration (< 100 mg), the authors
developed an opto-MEMS type sensor design. This sensor
design is coupled with a laser displacement-measurement module
that allows measurementof the actual vibration. The laser
measurement module has a simple construction, is less prone to
noise issues, and is very precise, which has been proven in
various applications. Cost reduction is also viable, by using LED
light sources.
For our prototype, the module consisted of a laser, a photo
diode, signal processing circuitry and power supply. The authors
used a P15 laser module from 3 Laser Tech Inc., of output power
35 mW at 635 nm single wavelength. The diameter of the beam
output was adjusted to 200 µm, using an optical lens. The light-
sensing photodiode was a Hamamatsu Model S3979, a photo
diode for single-axis displacement measurement; its nominal
resolution of displacement is 0.1 µm, and maximum sensitivity is
obtained at 920 µm wavelength. At 635 nm, its performance
degradation is negligible, and even shows better resistance to
temperature changes, according to the datasheet. The power
supply and signal processing circuitry were connected to the
laser and the photo diode respectively. Fig. 10 shows a simple
diagram of the signal processing module.
In our optical sensor module, the laser beam is projected to a
MEMS-fabricated cantilever, which will create movement of the
reflected beam due to the vibrating cantilever, which is captured
by the photo diode, and is converted to electronic signals to be
Fig. 8. The Full Comparison of Commercial MEMS Sensor and
MEMS Prototype
Fig. 9. Snapshots of the Testing of the Sensor Module at an Actual
Blasting Site

Jungyeol Kim, Soonwook Kwon, Seunghee Park, and Youngsuk Kim
− 6 − KSCE Journal of Civil Engineering
processed by downstream modules. Fig. 11 is a diagram of the
entire optical measurement module, and Fig. 12 shows the cross-
section diagram of the module.
5.2 Design Variables of the Accelerometer
Three variables significantly affect the resolution of the MEMS
accelerometer: these are resonant frequency, vertical displacement,
and linearity.
5.2.1 Resonant Frequency
Resonant frequency of a vibrating object is determined by its
shape and material. It was necessary to analyze the shape of our
accelerometer mass, in order to decide the measurable frequency
range. Since our accelerometer is cantilever-shaped, the authors
used a simple beam-like shape for modeling our accelerometer to
simplify the analysis. Using the simple beam model, its shape is
determined with three parameters of length(L), width(W), and
thickness (T). For such a beam-shaped elastic structure, Eq. (1)
shows how the resonant frequency is related to other variables.
(1)
F
r
= Resonant frequency
K = Elasticity coefficient
M =Mass
From Eq. (1), it is revealed that a higher elasticity coefficient
and lower mass is necessary for achieving a higher resonant
frequency. For the elasticity coefficient (K) and mass (M), the
following Equations are known:
(2)
E = Young’s Modulus
W =Width
L = Length
t = Thickness
(3)
r = Density
From Eqs. (2) and (3), the following Equation regarding the
resonant frequency can be derived:
(4)
From Eq. (4), it can be seen that the resonant frequency of a
beam-shaped elastic object is determined by its thickness (t)
and length(L). Therefore, such an object should have a high
elastic coefficient with low mass, if it needs a high resonant
frequency.
5.2.2 Vertical Displacement
Vertical displacement of an accelerometer is caused by elastic
movement; therefore, the following Equation can also be applied:
F
r
π
2
---
K
M
-----=
KEW
t
L
---
⎝⎠
⎛⎞
3
⋅⋅=
M ρ LWt⋅⋅ ⋅=
F
r
1
2π
------
E
ρ
---
t
L
2
-----
⎝⎠
⎛⎞
=
Fig. 10. Circuit Diagram of the Signal Processing Module
Fig. 11. Concept of the Optical Vibration Measurement Module
Fig. 12. Cut-away View of the MEMS-based Opto-mechanical Vibra-
tion Sensor

A MEMS-based Commutation Module with Vibration Sensor for Wireless Sensor Network-based Tunnel-blasting Monitoring
Vol. 00, No. 0 / 000 0000 − 7 −
(5)
K = Elasticity coefficient
x = Vertical displacement
F = Driving force
This implies that lower elasticity coefficient and higher driving
force are necessary for achieving more vertical movement, and
that the shape of the sensor should be designed to lower its own
elasticity coefficient for sufficient displacement.
5.2.3 Linearity
Although an elastic mass deforms in a linear manner, the
frequency of the driving force adversely affects damping of the
movement. This issue is discussed in a later chapter.
5.2.4 OtheR Design Factors
Minimum area
Since the displacement of the accelerometer is measured
optically using a laser, it should have a reflective surface whose
area is sufficient for such measurement. Since we used a circular
laser beam whose diameter is 200 µm, the reflective surface
should be larger than that.
Q-factor
Q-factor represents the response of a resonant mass inside a
fluid. To increase the factor, the area must be small.
5.3 Shape of High-resolution Accelerometer
Our MEMS accelerometer is made of Silicon with 50 µm
thickness due to the fabrication process. It is designed to satisfy
two conflicting goals: to have a large reflective area, while
having minimal overall size for faster response and better
sensitivity. Figs. 13 and 14 show two shapes we have designed.
The DT type cantilever (which has dual spring beams) is
designed to have a high Q-factor, by reducing the area of the
springs while maintaining its width; the ST type, which has a
large single spring beam, tries to minimize the loss of Q-factor,
by reducing the sensor area while keeping its displacement
high.
5.4 Optimization of the Sensor Design
As demonstrated in Eqs. (1) and (5), the accelerometer mass
must have both a higher resonant frequency and a larger vertical
displacement; the latter property requires lower elasticity,
whereas the former requires a higher one. To choose the optimal
design parameter, the authors conducted numerical analysis on
the sensor shape against its Length(L), Width(W), resonant
frequency and vertical displacement (design variables). Fig. 15
illustrates the relationship between design variables and their
result (resonance/deformation). The authors used pre-determined
design variables, such as minimum width, target resonant frequency,
FKx⋅=
Fig. 13. Shape of the Accelerometer Mass - cantilever ST Type
Fig. 14. Shape of the Accelerometer Mass - cantilever DT Type
Fig. 15. Numerical Analysis of Bending Mode with Respect to Shape

Jungyeol Kim, Soonwook Kwon, Seunghee Park, and Youngsuk Kim
− 8 − KSCE Journal of Civil Engineering
target displacement, etc., then ran the analysis program, using
MATLAB.
5.5 Iteration of the Initial Sensor Design
The authors chose the ST type cantilever over the DT type,
because the former allowed a higher measurable frequency.
Based on it, the authors designed five different shapes that were
subject to finite element analysis for their dynamic behavior (see
Table 2 for detailed shape parameters of those sensors). The
analysis result shows that they generate displacements of 26 µm
~122 µm when 0.1 mg vibration is applied; for the secondary
resonant mode, all of them are above 250 Hz, which is our
design goal for maximum measurable frequency (see Fig. 16).
5.6 Fabrication of the Sensor
Our accelerometer is fabricated on a 4 inch SOI (Silicon on
Insulator) wafer. There are several issues to consider when designing
the sensor layout on wafer: we want to harvest as many sensors
as possible from a single wafer, yet spacing between individual
sensors must be arranged so that the completed sensors can be
detached easily from the wafer. Also, the fabrication process
needs to be carefully planned, so that they can be produced
economically.
The layout design also features supporting structures, which
prevent the fabricated sensors from damage when various forces
are applied to the wafer during the fabrication process. To
remove the remaining area, a deep silicon etching process is
used. Fig. 17 shows the photomask layout of the wafer.
5.7 Evaluation
To evaluate our MEMS-based optical sensor, the authors
measured the same vibration using two sensors - a reference
sensor (Wilcoxon Research 731A, Fig. 18), and our prototype.
The readings from both sensors were fed to a Dacron Photon 24-
bit dynamic signal analyzer (Fig. 19) for further analysis. To
reduce difference in attenuation of the input vibration, both
sensors were located as close to each other as possible, and their
readings were time-synchronized. The g-value of the reference
sensor was calculated using a linear conversion equation provided
by the sensor manufacturer, whereas our module didn’t apply
any conversion for linearity compensation.
From the test result, among five design alternatives, the STF3
model was closest to the reference sensor with respect to its
Table 1. Five Design Candidates for Our Opto-mechanical Sensor Mass (Unit: µm)
Type Lspr Wspr Lsei Wsei
STF1 3000 1000 500 1000
STF2 3000 1000 1000 1000
STF3 5000 1500 1000 1000
STF4 5000 1500 1000 1500
STF5 5000 1500 1500 1500
Fig. 16. FEM Analysis of Five Design Candidates (Example, STF2) - 0.1 mg is Applied. From Left to Right: (a) Vertical Displacement, (b)
Resonant Mode 1, (c) Resonant Mode 2
Fig. 17. Photo Mask Layout of Our Accelerometer
Fig. 18. Wincoxon Resarch 731A Vibration Sensor and Its Specifi-
cation

A MEMS-based Commutation Module with Vibration Sensor for Wireless Sensor Network-based Tunnel-blasting Monitoring
Vol. 00, No. 0 / 000 0000 − 9 −
dynamic characteristics. STF1 and STF2 fell short of sensitivity,
whereas STF4 and STF5 didn’t behave well in terms of signal
attenuation. Fig. 20 was taken from the test scene.
After calibrating our sensor module, the authors acquired the
final measurement values from our module, which is shown in
Fig. 21. The graph shows variable magnitude of the test vibration,
up to 500 mg.
For the 0~100 mg range, the standard deviation of the difference
with respect to the reference sensor was 2.77, showing that our
sensor was able to measure reliably in that range; for 100 mg~
500 mg, it was 4.52%.
6. Conclusions
For development of a new breed of sensors, the authors
investigated two different sensor types, both of which were
based on MEMS technology. The first type uses a comb-shaped
suspended mass that causes varying capacitance while the mass
is moving, due to external vibration; the second type, our latest
iteration, combined optical measurement using reflectors mounted
on a MEMS-fabricated cantilever, which enhanced sensitivity in
small-acceleration, e.g. less than 100 mg.
Basically, the result of this research can be applied to the
ground vibration monitoring due to blasting for tunnel construction.
It can measure vibration and acceleration of the blasting, and its
data can be utilized for the analysis of nearby ground condition.
Also, the sensor module can be used for the impact analysis of
buildings and civil infrastructures close to the blasting point.
Furthermore, with the ease and speed of its deployment, the
sensor module can be employed for the emergency vibration or
acceleration measurement of any artificial structure.
However, the sensor module has several limitations. The first
limitation is its power consumption. Although the prototype
sensor module requires lower power than conventional sensors
and it can be operated several days with commercial batteries, its
power consumption still needs to be optimized for the long-term
battery operation. Solar cell or fuel cell can be an alternative
power source of this module at this point. The second limitation
is its size. The size and the arrangement of sensor jig, circuit, and
case are not optimized as other commercialized sensors. These
limitations should be overcome for the actual use of this module
during the commercialized process in the future.
Acknowledgements
This study was conducted as a base project of Korea Institute
of Construction Technology under the support of Ministry of
Knowledge Economy (Project Code: KICT 2009-061).
References
Alfado, F., Weiss, L., Campbell, P., Miller, M., and Fedder, G. K. (2009).
“Design of a multi-axis MEMS sensor for intraosseous bone stress
monitoring”, J. Micromech. Microeng, Vol. 19, No. 00, pp. 00-00.
Azevedo, R. G., Jones, D. G., Jog, A. V., Jamshidi, B., Myers, D. R.,
Chen, L. et al.(full name) (2007). “A SiC MEMS resonant strain
sensor for harsh environment applications.” IEEE Sensors Journal,
Vo l . 7 , No. 00, pp. 568-576.
Kon, S. and Horowitz, R. (2008), “A high-resolution MEMS piezoelectric
strain sensor for structural vibration detection.” IEEE Sensors
Journal, Vol. 8, No. 00, pp. 2027-2035.
Yu, Y., Ou, J., Zhang, J., Zhang, C., and Li, L. (2009). “Development of
wireless MEMS inclination sensor system for swing monitoring of
large-scale hook structures.” IEEE Transactions on Industrial
Electronics, Vol. 56, No. 00, pp. 1072-1078.
Ioppolo, T., Ayaz, U. K., Ötügen, M. V., and Sheverev, V. (2008). “A
Fig. 19. Dacon Photon 24-bit Dynamic Signal Analyzer
Fig. 20. Test Setup
Fig. 21. Comparison of the Sensor Output between the Reference
Sensor (Orange Circle) and Our Sensor (Blue Dot)

Jungyeol Kim, Soonwook Kwon, Seunghee Park, and Youngsuk Kim
− 10 − KSCE Journal of Civil Engineering
micro-optical wall shear stress sensor concept based on whispering
gallery mode resonators.” 46th AIAA Aerospace Sciences Meeting
and Exhibit, 8-11 January 2008 Reno Nevada.
Ioppolo, T., Ayaz, U. K., Ötügen, M. V., and Sheverev, V. (2009).
“Performance of a micro-optical wall shear stress sensor based on
whispering gallery mode resonators.” 47th AIAA Aerospace Sciences
Meeting and Exhibit, 5-8 Jan 2009, Orlando Florida.
Kang, I., Schulz, M. J., Kim, J. H., Shanov, V., and Shi, D. (2006). “A
carbon nanotube strain sensor for structural health monitoring.”
Smart Materials and Structure, Vol. 15, No. 00, pp. 737-748.
Leng, J. S., Barnes, R. A., Hameed, A., Winter, D., ATetlow, J., Mays, G.
C. et al.(full name) (2006). “Structural NDE of concrete structures
using protected EFPI and FBG sensors.” Sensors and Actuators A,
Vol. 126, No. 00, pp. 340-347.
Kim, J. Y., Kwon, S. W., Yoo, H. S., and Cho, M. Y. (2005),
Development of MEMS-based vibration sensor for tunnel system,
22
nd
International Symposium on Automation and Robotics in
Construction (ISARC), 11-14 September 2005, Ferrara, Italy.
Kwon, S. W., Kim, J. Y., Yoo, H. S., and Cho, M. Y. (2006). Wireless
vibration sensor for tunnel construction, 23
rd
ISARC, 3-5 October
2006, Tokyo, Japan, pp. 614-620.