Ieeepro techno solutions ieee embedded project - low power wireless sensor network for building monitoring

R
S S S S
A
AR
A
A
A
A
AR
A
S strain sensormodulestrain sensormodulesatbasement
A accelerometermodule
AR accelerometer-router
R roof router
B
B base station
directional link
>1km line of sight
Figure 1. Architecture of the sensor network for building monitoring
Low Power Wireless Sensor Network
for Building Monitoring
Tom Torfs, Tom Sterken* Steven Brebels, Juan
Santana+
, Richard van den Hoven+
, Chris Van Hoof
IMEC, Leuven, Belgium;
*IMEC / Ghent University, Ghent, Belgium;
+
IMEC / Holst Centre, Eindhoven, The Netherlands
Contact e-mail : torfst@imec.be
Nicolas Saillen, Nicolas Bertsch*, Davide Trapani+
,
Daniele Zonta+
, Paris Marmarasº, Matthaios Bimpasx
Thermo Fisher Scientific, Enschede, The Netherlands
*MEMSCAP, Crolles, France;
+
University of Trento, Trento, Italy;
ºNetScope Solutions, Athens, Greece; x
ICCS, Athens, Greece
Abstract— A wireless sensor network is proposed for monitoring
buildings to assess earthquake damage. The sensor nodes use
custom-developed capacitive MEMS strain and 3D acceleration
sensors and a low power readout ASIC for a battery life of up to
12 years. The strain sensors are mounted at the base of the
building to measure the settlement and plastic hinge activation
of the building after an earthquake. They measure periodically
or on-demand from the base station. The accelerometers are
mounted at every floor of the building to measure the seismic
response of the building during an earthquake. They record
during an earthquake event using a combination of the local
acceleration data and remote triggering from the base station
based on the acceleration data from multiple sensors across the
building. A low power network architecture was implemented
over an 802.15.4 MAC in the 900MHz band. A custom patch
antenna was designed in this frequency band to obtain robust
links in real-world conditions.
I. INTRODUCTION
Buildings can progressively accumulate damage during
their operational lifetime, due to seismic events , unforeseen
foundation settlement, material aging, design error, etc.
Periodic monitoring of the structure for such damage is
therefore a key step in rationally planning the maintenance
needed to guarantee an adequate level of safety and
serviceability. However, in order for the installation of a
permanently installed sensing system in buildings to be
economically viable[1], the sensor modules must be wireless
to reduce installation costs, must operate with a low power
consumption to reduce servicing costs of replacing batteries,
and use low cost sensors that can be mass produced such as
MEMS sensors. The capability of MEMS and wireless
networking for monitoring civil structures is well documented
[2][3][4]. The presented work addresses all of the above
requirements.
II. SYSTEM ARCHITECTURE
A. Network architecture
The monitoring system consists of two types of sensor
modules: strain sensing modules and acceleration sensing
modules. They are placed in the building as shown in Fig. 1.
The strain sensor modules are mounted at the lowest level of
the building, to estimate the vertical column loads and any
variation due to settlement. Horizontal acceleration is
measured by two 3D acceleration sensing modules (where
only the two horizontal axes are really required) at each level
during an earthquake, allowing analysis of the seismic
response of the whole structure. A typical 7-story, 24-column
building requires approx. 72 strain sensors (3 per column) and
14 accelerometer modules (2 per floor).
The data is wirelessly transmitted to a nearby base station
using a line of sight link with a range of >1km. The line of
sight link uses directional antennas to improve the link budget,
but not so directional that alignment is required, which could
pose a problem during seismic events. The receiver base
station can store and process the data or forward them,
immediately or later, using classical wide area network
connection technology. In this way, provided all modules as
well as the receiver base station have battery back-up power,
the data acquired during seismic events can be properly
recorded even in case of outages of the electric power and/or
communication networks.
In order to form a robust wireless link from all modules,
including the strain sensor modules at the basement of the
building, towards the receiver base station, a multi-hop
network architecture is used as shown in Fig. 1. On the roof of
Work performed in the MEMSCON project, funded by the European
Commission (FP7, contract number 036887), website: www.memscon.com

Radio
control
(ATMega)
patch
antenna
MEMS 3D
accelero‐
meter
Sensor
read‐out
ASIC
Sensor
control
(MSP430)
ADC
(16‐bit)
Memory
64Kx16
Radio
802.15.4
900MHz
Battery
(lithium 3.6V)
wireless module
Power
regulation
circuit
MEMS
strain sensor
4‐wire cable
(max length: 1.5m)
Sensor
read‐out
ASIC
concrete‐embedded module
OR
Figure 2. Block diagram of the sensor module with either a built-in
accelerometer or a separate strain sensor module embedded in the concrete
(a) (b) (c)
5 cm
Figure 3. Picture of the wireless sensor module:
(a) antenna, (b) electronics, (c) battery
the building a dedicated router module (without sensor) is
placed to forward the data between the sensor network and the
receiver base station. Some accelerometer modules on
intermediate floors can be configured as additional
intermediate routers when required to obtain a robust link
from all sensor modules in the building towards the roof router
module. As shown on Fig. 1, it is recommended to place the
router modules in or close to the stairwell for improved
vertical floor-to-floor propagation through the building.
For lowest power consumption in the sensor modules, the
network is implemented using indirect data transfer using
polling on top of a standard 802.15.4 MAC. In this way, the
end nodes’ radio is powered down most of the time. Only the
routers and base station have their receivers constantly on. To
avoid rapid battery depletion, the modules with router
functionality are mains-powered through an AC/DC adapter,
with the battery serving only for back-up power in case mains
power is interrupted. The end nodes are battery powered only.
B. Sensor module architecture
The block diagram of the sensor modules is shown in Fig.
2 and a picture in Fig. 3. Both the accelerometer and strain
sensing variants of the module use the same core components.
For installation into the building these components are placed
into a standard off-the-shelf plastic casing that can be
conveniently mounted on the floor, wall or ceiling using
screws, and offering access for sporadic battery replacement if
needed. The core components are:
1) A custom-developed low power capacitive sensor
read-out ASIC [5]. This ASIC can be matched to either
MEMS-based comb finger capacitive accelerometers or strain
sensors in a half-bridge configuration. Its gain can be set by a
number of integration pulses N, optimizing signal-to-noise
ratio and bandwidth with power. In addition, the architecture
suppresses residual motion artifacts. In combination with the
MEMS strain sensor, it can measure a range of ± 20,000με
with a resolution of 10με and non-linearity <0.6%. In
combination with the MEMS accelerometer it can measure an
acceleration range of ±2.5g with a resolution of 80dB (13-bit)
for vibrations between 10-100Hz and a non-linearity <1%.
2) A low power 16-bit successive-approximation analog-
to-digital converter (Analog Devices AD7683).
3) A low power microcontroller (TI MSP430) to control
the sensor data acquisition and temporarily store the data in a
64Kx16bit SRAM memory (Cypress CY62126) .
4) A low power wireless IEEE 802.15.4-compatible
module (Atmel ATZB-900) operating in the 900MHz band.
This frequency band was chosen in preference to the more
common 2.4GHz band because it offers a larger propagation
range for the wireless communication. The wireless module
includes a radio chip (Atmel AT86RF212) and a baseband
microcontroller (Atmel AVR) which needs to be active only
during wireless communication events.
5) A custom patch antenna was designed for the modules.
The patch antenna is tuned for 868MHz operation with an
efficiency of 51% using standard FR4 material as the
substrate. Its size is 5 x 5 x 1.3 cm3
. Its shape and radiation
pattern is optimized for wall-, floor- and ceiling-mounting in
the building.
6) The modules are powered by a an 8.5Ah C-cell long
operating life primary Lithium Thionyl Chloride battery
(Tadiran SL-2770), suitable for 10 to 25 years of operation.
The accelerometer consists of 2 transverse comb finger
structures for the X and Y axis and a pendulating one for the Z
axis and was fabricated with a surface micro-machined
process from a 85μm thick SOI wafer. It has 78 fingers with a
total sensitivity of 2.02pF/g. The Z sensor has an area of
2.17mm2
per plate. Innovative cap through connections were
used. The main tradeoff in the design of the accelerometer is
the sensitivity-bandwidth-linearity in all three axes, a
challenge for the design given the different used structures.
The XY and Z accelerometers are packaged together with the
readout ASIC into a system-in-a-package and then mounted
onto the printed circuit board as can be seen on Fig. 3.
The MEMS strain sensor is a longitudinal comb finger
capacitor. The strain sensor fabrication procedure starts with a
SOI wafer with a 500μm thick handle, 50μm thick fingers and
2μm thick oxide layer with 400 fingers in the sensor and it has
a sensitivity of 0.133fF/μe. Two anchors were etched-out of
the surface to create the necessary clamps to attach the sensor
to the rebar of a pillar. The fingers are protected with a
borosilicate class cap.

(a)
(b)
PDMS
Strainsensor
Wire bonds
Printed circuit board
Components
Cable (I/O)Polyimideor
steel carrier
Figure 4. Strain sensing front-end module
(a) picture of a realization on a steel carrier (b) diagram
0.000
0.200
0.400
0.600
0.800
1.000
1.200
1.400
1.600
1.800
Strain sensor Accelerometer
Power consumption (mW)
Standby
Sampling &storage
Sensor& readout
Processing
Radio transmission
Radio polling
Power conversion
loss
Figure 5. Sensor module power consumption breakdown
The MEMS strain sensor is packaged together with the
readout ASIC into a special front-end strain sensing module
(Fig. 4) which is to be embedded inside the reinforced
concrete onto the reinforcing bar, preferably prior to the
pouring of the concrete. The sensor is mounted on a polyimide
or steel carrier which in turn is glued (or in case of steel, could
also be welded) onto the reinforcing bar. The module is
molded in PDMS silicone to protect the components from the
environment during installation and pouring of concrete, while
remaining a mechanically compliant package to avoid
distorting the strain sensor measurement.This front-end strain
sensing module is connected to the rest of the module with a
small 4-wire cable with a maximum length of 1.5m.
The use of custom-developed MEMS sensors and read-out
ASIC allows to meet the specific requirements of the building
monitoring application and differentiates the presented system
from the earlier prototype system presented in [6] and [7].
C. Measurement initiation and radio polling
1) Accelerometer modules
The main trigger for the recording of an acceleration
measurement is the detection of the start of an earthquake. The
detection can happen in two ways:
a) The output of the built-in accelerometer in a selected
number of monitoring nodes exceeds a certain minimum
threshold, during a certain minimum time; these monitoring
nodes alert the base station which will decide whether to
wake up the entire network of acceleration sensing nodes
over the radio The monitoring nodes are selected based on
their location and amount of environmental noise; ground-
level nodes may be suitable candidates, provided they are
sufficiently far removed from disturbance sources such as
heavy traffic. The selection of monitoring nodes can be done
dynamically from the base station. This allows for example to
disable the monitoring function on nodes that report
unusually high numbers of false alarms. To that purpose, the
hardware and software of the monitoring nodes are identical
to that of the non-monitoring nodes. The monitoring function
is an optional function which can be enabled or disabled
during operation by the base station.
b) An external source of earthquake detection can be
coupled to the base station, which can use such information in
addition to possible information from monitoring nodes to
decide whether to wake up the entire network or not.
In both of these cases the wake-up of (most of) the
acceleration sensing nodes to initiate measurement has to be
done over the radio link. This also implies that it is possible to
wake up the nodes via the base station over the radio link at
any chosen time independent of the presence of an earthquake,
which is a desired functionality for testability and monitoring
of the system. It also means that all modules in the network
will be woken up during a detected event, even if the
accelerations locally at some modules have not (yet) reached a
value exceeding the trigger threshold.
It is required to be able to record the early onset of an
earthquake event, even before and certainly no later than 1 ms
after it reaches a pre-set trigger threshold. In order to do this,
the accelerometer is constantly running at 3 x 200Hz sample
rate with the measurements recorded in a 54-second loop
buffer., This requires an ultra low power sensor and readout.
The power consumption of the 3D accelerometer and 3-
channel readout operating continuously is 125 µA at 3V.
The node must be woken up within 54 seconds after the
start of the recording of interest to avoid the loop buffer
overflowing which would lead to data loss. To respond timely
to an event triggered from the base station, the radio polling
interval of the accelerometer modules is set to 15 seconds.
Once the event trigger is reached the loop buffer contents are
preserved and once the buffer is full recording will continue in

Figure 6. (a) Building scale model in the lab with the wireless accelerometer modules and reference accelerometers installed
(b) Evaluation result comparing the proposed wireless accelerometer to the reference accelerometer output
a secondary 54-second buffer until the next event trigger.
2) Strain sensor modules
The main measurement scenario for the strain sensor is a
periodic readout. Samples are taken at a configurable sample
rate between 2 seconds and 18 hours. The strain sensor
modules use a radio polling interval of 60 seconds. This also
allows manual wake-up functionality from the base station,
again useful for monitoring and testability reasons. Unlike for
the accelerometers, in the case of the strain sensors the sensor
and read-out ASIC can be entirely shut down between
measurements. This results in a lower power consumption and
longer battery life. Since a typical building requires many
more strain sensors than accelerometer modules, it is useful
for the strain sensors to have the longest battery service life.
III. RESULTS
A. Power consumption
Fig. 5 shows the breakdown of the power consumption in
the sensor modules for strain sensor and accelerometer
modules. The total average power consumption is 0.274mW
for the strain sensor modules and 1.73mW for the
accelerometer modules. With the abovementioned C-cell size
battery this implies a battery life of 12 years for the strain
sensor modules and 2 years for the accelerometer modules.
B. Laboratory validation
Fig. 6 shows a lab setup on a scale model of the building
used to validate the accelerometer modules. For this test, the
custom patch antenna was not used, but a standard whip
antenna was used instead. The result shows a very good
correlation between the reference accelerometers and the
proposed wireless modules.
The strain sensor modules were preliminarily validated in
a calibration setup and show sensitivities between 10 and 20
με/mV, varying from sensor to sensor.
CONCLUSION
The presented wireless system for building monitoring
takes advantage of the unique features of custom-developed
MEMS sensors and read-out ASIC combined with an
optimized network and module architecture, to realize a
solution which offers long battery lifetime and potentially low
cost in manufacturing, installation and maintenance, while
providing high-quality sensor data at the right time.
REFERENCES
[1] Pozzi M., D. Zonta, W. Wang and G. Chen. 2010. “A framework for
evaluating the impact of structural health monitoring on bridge
management”. Proc. 5th International Conf. on Bridge Maintenance,
Safety and Management (IABMAS2010), Philadelphia, 11-15 Jul 2010.
[2] Lynch, J.P. and K.J. Loh. 2006. “A summary review of wireless
sensors and sensor networks for structural health monitoring,” The
shock and vibration digest, 38(2):91-128.
[3] Zonta, D., M. Pozzi, and P. Zanon. 2008. “Managing the Historical
Heritage Using Distributed Technologies,” International Journal of
Architectural Heritage, 2:200-225.
[4] Kruger, M., C.U. Grosse and P.J. Marron, 2005. “Wireless Structural
Health Monitoring Using MEMS”, Key Engineering Materials 293-
294: 625-634.
[5] J. Santana,R. van den Hoven,C. van Liempd, M. Colin, N. Saillen, C.
Van Hoof, "A 3-axis accelerometer and strain sensor system for
building integrity monitoring", Proc. 16th International Conference on
Solid-State Sensors, Actuators, Microsystems, Beijing, June 5-9, 2011.
[6] A. Amditis, Y. Stratakos, D. Bairaktaris, M. Bimpas, S.
Camarinopolos, S. Frondistou-Yannas, et al., "An overview of
MEMSCON project: an intelligent wireless sensor network for after-
earthquake evaluation of concrete buildings", Proc. "14th European
Conference on Earthquake Engineering (14ECEE)", Ohrid, FYROM,
30 Aug - 03 Sep, 2010.
[7] A. Amditis, Y. Stratakos, D. Bairaktaris, M. Bimpas, S.
Camarinopolos, S. Frondistou-Yannas, et al. "Wireless sensor network
for seismic evaluation of concrete buildings", Proc. " 5th European
Workshop on Structural Health Monitoring (EWSHM 2010)", Sorrento,
Italy, 29 Jun - 02 Jul, 2010.

Ieeepro techno solutions ieee embedded project - low power wireless sensor network for building monitoring

More Related Content

Ieeepro techno solutions ieee embedded project - low power wireless sensor network for building monitoring