Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

ENERGY HARVESTING, WIRELESS,

STRUCTURAL HEALTH MONITORING and REPORTING SYSTEM

S.W. Arms, C.P. Townsend, D.L. Churchill, J.H. Galbreath, B. Corneau, R.P. Ketcham, N. Phan*

MicroStrain, Inc., 310 Hurricane Lane, Williston, VT, USA

* Branch Head, Rotary Wing/Patrol Aircraft, NAVAIR Structures, Naval Air Systems Command, Lexington Park, MD

ABSTRACT

Energy harvesting, combined with wireless sensors, could greatly improve our ability to monitor and maintain critical
structures. This paper reports on the development of an integrated structural health monitoring and reporting (SHMR)
system for use on Navy aircraft. We have previously reported on energy harvesting, wireless sensor modules which
have passed MIL-STD-810F tests for vibration, shock, humidity, and temperature extremes. These modules were
integrated into the pitch link of a Bell M412 helicopter. In February 2007, the first successful flight test of an energy
harvesting wireless sensor was achieved. Pitch link loads were recorded and periodically transmitted into the cabin
during flight. The flight test proved that our e-harvesting load sensor can operate continually without battery
maintenance. The direct measurement of operational loads enables accurate fatigue tracking of these critical rotating
structures.
But not all sensors are wireless – “less wire” is often appropriate, particularly when relatively high sample
rates and/or power levels may be required. Additional inputs to an integrated SHMR system may include data from the
vehicle’s hard-wired bus. Therefore, there is a need for a scalable, time-synchronized sensing system, capable of
supporting both wireless and hard-wired sensor networks. Our goal was to develop and test a versatile, fully
programmable SHMR system, designed to synchronize and record data from a range of wireless and hard wired sensor
networks.
Wireless sensors included strain gauges, accelerometers, and thermocouples. Hard-wired sensors included
gyroscopes, accelerometers, and magnetometers. Data from an embedded Global Positioning System (GPS) provided
position, velocity, and precise timing information. The inertial sensing suite provided vehicle orientation (pitch, roll,
and yaw) data. These data were collected at multiple sampling rates and time stamped and aggregated within a single
scalable database on a base station, termed the wireless sensor data aggregator (WSDA). The WSDA’s processor
supports a Linux server, web interface, eight (8) Gigabytes secure flash memory, CAN, IEEE 802.15.4, Ethernet,
RS232/422, and mobile phone. The data may be relayed over mobile phone networks to a secure server. Software to
access the aggregated data over the internet was developed, using the time stamp as a unifying reference for the various
types of sensor information.
The WSDA, in addition to providing a central location for collecting wireless and wired network sensor data,
also provided a beaconing capability to synchronize each sensor node’s embedded precision timekeepers. For testing, a
saw tooth analog voltage waveform was provided as an input to two wireless nodes to provide a reliable means of
determination of the system’s timing accuracy. With the synchronization beacon provided at the start of a 2 hour long
test, and with two wireless nodes exposed to multiple temperature cycles of -40 to +85 degrees C, the system
demonstrated a 5 millisecond timing accuracy.

1. INTRODUCTION

Recent developments in combining sensors,
microprocessors, and radio frequency (RF)
communications holds the potential to revolutionize the
way we monitor and maintain critical systems.1 In the
future, literally billions of wireless sensors may
become deeply embedded within machines, structures,
and the environment. Sensed information will be
automatically collected, compressed, and forwarded for
condition based maintenance.
But the problem for the end user (and for our
environment), is that wireless sensors need energy to
operate, and batteries are a pain to maintain. A
solution to this problem is to harvest and store energy
from the environment – using strain, vibration, light,
and motion to generate the energy for sensing and
communications. Combined with strict power
management, smart wireless sensing networks can
operate indefinitely, without the need for battery
maintenance.
To extend the life of today’s rotary wing aircraft,
dynamic component removal, refurbishment and
replacement must be optimized. To accomplish this,
an accurate and up-to-date system must be developed
1
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
to establish the current and past history of each fatigue
critical aircraft component. By directly tracking
loading histories, component fatigue rates are known,
enabling individual aircraft maintenance to be
optimized based on actual usage - rather than on
indirect estimates based on flight hours and flight
regimes.
El-Bakry of Airbus proposed the use of
passive RFID tags to facilitate aircraft component
tagging and tracking.2 Maley et al of the US Navy
emphasized the need for embedded, wireless sensors
capable of detecting and tracking the “precursors” to
crack initiation.3 Examples included strain sensitive,
printable, two dimensional bar codes4 and active
wireless strain sensors.5 The Navy’s long term vision
is to deploy distributed wireless sensor networks along
with RFIDs and bar codes to provide a wealth of usage
information about an entire aircraft structure.
Operating at microwatt power levels,
MicroStrain’s sensing nodes support a variety of
sensors, including conventional strain gauges, load
cells, pressure sensors, accelerometers, and
thermocouples. Overcoming the limitations of
batteries through ambient energy harvesting has been a
natural evolutionary step in wireless technology
development. Using both piezoelectric and
electromagnetic energy harvesters, we have
demonstrated that sufficient energy could be harvested
to power a wireless strain sensor transceiver.6
These energy harvesting sensor systems have
been adapted to track damage on the rotating structural
components of helicopters, using wireless strain
gauges. An example of a critical component is the
control rod, or “pitch link”. Pitch links are traditionally
only monitored with slip rings during instrumented
aircraft flight testing (on one or two aircraft). Pitch
link loads in the Sikorsky H-60 have been found to
2. OBJECTIVES
Our objectives were to develop and test a versatile
SHMR system, designed to synchronize, record, and
report data from a range of wireless sensors (strain
2.1 Concept of Operations
Wireless technologies for tracking the load history of
helicopter rotating components, combined with inertial
and global positioning system (GPS) information, can
be used to compute structural loads with improved
accuracy. The integration of these sensor systems will
lead to reduced cost flight testing, improved safety, and
enhanced condition based maintenance (Figure 1).
Ideally, the integrated structural health
monitoring system would report aircraft load history
data without human intervention. Data collected
during flight would be automatically recorded on
 
vary strongly with flight regimes: during pull-ups and
gunnery turns, the loads were measured at
approximately eight times that of straight and level
flight.7 Therefore, pitch link loads are a good indicator
of the rotating structure’s usage severity.
Wireless strain gauges placed strategically on
the pitch link enable the direct measurement of static
and dynamic axial loads, while canceling out bending
loads and thermal influences. We have demonstrated
that the operational strains in the pitch link will
generate enough power to allow continuous, wireless
operational load monitoring of this critical structure,
even during conditions of straight and level flight.5 In
the spring of 2007, the first successful flight test of our
energy harvesting wireless pitch link was performed on
a Bell M412 helicopter.8
In this paper, we provide a summary review of
our energy harvesting efforts, and we report on new
systems which expand on our previous work by
combining both wireless and hard wired data
acquisition. In order for these systems to accurately
track structural usage severity, it is critical that the data
from various wireless and hard wired sensors be
accurately time stamped. Le Cam has previously
reported on using an integrated GPS module on each
smart sensor to accomplish an absolute precision of 1
microsecond.9 Rather than include GPS modules (and
their antennas) at each node of the SHM system, we
chose to incorporate GPS within a central node, herein
referred to as the Wireless Sensor Data Aggregator
(WSDA) node. To maintain accurate time on the
remotely distributed sensor nodes, we chose to develop
a low power, temperature-compensated timing engine.
We also report on the capability to send data
to a remote server, using mobile phone networks to
enable automated data collection and reporting.
gauges, accelerometers, thermocouples, etc), along
with data from GPS and an inertial sensing suite.
board, without wireless communications, since each
wireless load tracking node would be capable of
recording data within its local non-volatile memory.
After the aircraft lands, the on-aircraft base station
would query the network of wireless load tracking
nodes, and prepare data files for remote transmission
over the cellular or satellite connection. Data would
then be analyzed and maintenance instructions sent
back to aircraft technicians. Figure 2 provides an
illustration of this concept of operation.
2
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

Figure 1. Concept
for wireless energy
harvesting systems
for helicopter
flight test and in
service structural
loads tracking

Figure 2. Concept
of operations for
automated
helicopter
condition based
maintenance

3. METHODS

3.1 Wireless Strain Sensor Nodes

Each wireless sensor node includes an instrumentation
amplifier, a programmable gain amplifier with
programmable gain and offset adjust, and anti-aliasing
filter, 12-bit successive approximation analog to digital
(A/D) converter, embedded microcontroller
(processor), 2 MB of non-volatile memory (EEPROM),
and a programmable 2.4 GHz transceiver chip. Each
individual node is assigned a unique identification code
(RFID), which enables specific nodes on the network
to be address by the WSDA base station. Our current
wireless nodes system support both 16 bit addresses
and 96 bit RFID codes. Broadcast addresses are
reserved for network wide command and control. A
block diagram of a strain sensing node is provided in
Figure 3.

3
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
The embedded firmware within each node supports an
open architecture command structure that enables
wireless control over each node in the network, or of
all nodes in the network. Node specific commands are
typically for wireless adjustment of gains, offsets, or
shunt calibration of specific sensor channels with a
node. This flexibility enables support of a wide range
of Wheatstone bridge type sensors, including strain
gauges, accelerometers, pressure transducers, torque
sensors, load cells, and magnetometers. Energy is
conserved by multiplexing a pulsed, regulated
excitation voltage to the sensor bridge.6 The use of
high resistance bridge sensors is preferred, as this also
conserves energy by reducing the current drawn by the
sensors.
Temperature coefficients for these wireless
strain sensor nodes have been measured in our
laboratory at -.007%/degree C (offset) and
.015%/degree C (gain) when tested over a temperature
Figure 4.
Integrated
wireless shear
pin (Shear-
Link) node,
shown at right,
adjacent to a
US quarter
dollar for size
 
Figure 3. Block diagram of MicroStrain’s
wireless sensing node and base station
transceiver
range of -40 to 125 degrees C. During these tests, the
strain gauge bridge was simulated by a bridge of fixed
precision resistors, with this resistor bridge located
outside of the environmental testing chamber.
An example of a fully integrated wireless
shear pin is provided in Figure 4. This wireless node
features strain gauges bonded within a tiny aperture,
which are connected internally to a miniaturized
wireless sensing node. This type of sensor can be used
to monitor shear loads on many structural connections
on and within an aircraft, including the landing gear.
3.2 Energy Harvesting
A key barrier to the widespread adoption of wireless
sensing networks has been the problem of keeping the
node’s batteries charged. MicroStrain, Inc. has been a
leader in adaptive energy harvesting electronics for
wireless sensor networks. Our electronics feature
smart comparators – switches which consume only
nanoampere levels of current – to control when to
permit a wireless sensing node to operate. This insures
that the energy checkbook is balanced, in other words,
the system waits until there is sufficient energy to
perform a programmed task. Only when the stored
energy is high enough will the nanoamp comparator
switch allow the wireless sensor to draw current.10
This is critical for applications where the ambient
energy levels may be low or intermittent. Without this
switch, the system may never successfully start up,
because stored energy levels may always remain
insufficient for the task at hand.11
Fully integrated energy harvesting wireless
sensors have been developed for the helicopter control
rod, or pitch link. Piezoelectric materials bonded
directly to the pitch link were used to harvest strain
energy for operation during flight. This work
demonstrated, for the first time, that an energy
harvesting wireless sensor for rotating helicopter
components could be operated indefinitely, using only
the strain energy of operation for power.
We have also reported on vibration energy
harvesters, which use piezoelectric materials and tuned
resonant structures, which were designed to resonate at
4
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
the predominant frequencies of the machine to which
they were affixed. These harvesters were
demonstrated, in Navy shipboard applications, to be
effective under ambient conditions of very low
vibration amplitude (30 milliG’s) and low vibration
frequency (53 Hz).12
Embedded software was used to balance the
requirements of sampling and transmitting data against
the amount of energy that the harvester may be
generating in a given application. The embedded
software supported various operating modes for the
wireless sensor, which enabled the sensor node to adapt
its power consumption depending on the amount of
energy that was being generated.8
Solar energy harvesters have also been used
for civil structural monitoring applications. We
describe these applications in more detail in the
sections that follow.
3.2.1 Tracking Helicopter Component Loads with
Energy Harvesting Wireless Sensors
The US Navy, through its SBIR program, has
supported MicroStrain’s development of wireless
sensor nodes that use piezoelectric materials to convert
cyclic strain and vibration into power. One compelling
application is in monitoring the critical rotating
structures of helicopters. The direct measurement of
loads on these structures will allow enhanced
maintenance, which will greatly reduce operational
costs. Better loads tracking also has the potential to
save lives through improved performance and safety.
We have focused our initial efforts on the
helicopter control rod, or “pitch link”. The pitch link is
responsible for controlling the rotors’ angle of attack as
the rotor rotates through the air. As mentioned
previously, pitch link loads vary strongly with aircraft
flight regimes, reaching much higher loads (6X) during
maneuvers as compared to straight and level flight.7
Therefore the pitch link is an excellent indicator of
vehicle usage severity, and can provide critical data for
improved condition based maintenance. A photograph
of the microelectronics module developed under the
Navy SBIR program for the pitch link application is
provided in Figure 5.
Figure 5. Microstrain’s Energy Harvesting Wireless
Pitch Link Load Sensing Node (patent pending)
 
Figure 6. MicroStrain’s Energy Harvesting, Wireless
Loads Tracking Pitch Link installed on Bell M412
Working in collaboration with Bell
Helicopter/Textron (Fort Worth, Texas), the first
successful flight test of our energy harvesting wireless
sensor node was performed in March 2007. We
instrumented the pitch link of a Bell Model 412
helicopter with our energy harvesting sensor node,
along with piezoelectric materials and a full strain
gauge bridge, which cancelled thermal and bending
influences while amplifying tension and compression
loads (figure 6).
The flight test showed that our energy
harvesting strain and load sensor will operate
continually, without batteries, even under low energy
generation conditions of straight and level flight. By
continuously monitoring the strains on rotating
components, our wireless nodes can record operational
loads, compute metal fatigue, and estimate remaining
component life. These techniques can also be applied
to other applications, such as monitoring large civil
structures.
3.2.2 Monitoring Large Bridge Spans with Solar
Powered Wireless Sensors
Recently, sudden structural failures of large bridge
spans, such as the Interstate 35W Bridge in
Minneapolis, and the Chan Tho Bridge in Vietnam
have resulted in the tragic loss of lives and of loved
ones. Three years ago, the Federal Highway
Administration reported that ~20 percent of the US
interstate bridges (nearly 12,000 bridges) were rated as
deficient. Developing and deploying cost efficient
methods for monitoring bridges - and for determining
which bridges require immediate attention - should be
an important priority for the United States and the
world.
Wireless sensor networks have the potential to
enable cost efficient, scalable monitoring systems that
could be tailored for each particular bridge’s
requirements. Eliminating long runs of wiring from
5
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

each sensor location greatly simplifies system the entire network. Photographs of this bridge and the
installation and allows a large array of sensor nodes to wireless G-LINK® nodes as installed in Corinth are
be rapidly deployed. provided in Figures 7 and 8.
We have recently supported two major The second solar powered installation is on
wireless installations which are actively monitoring the the Goldstar bridge in New London, Connecticut, in
structural strains and seismic activity of major spans. collaboration with John DeWolf, Ph.D. of the
Leveraging energy harvesting technology supported by University of Connecticut. This system monitors not
the US Navy, these wireless sensor networks are only vibration, but also the strains and temperatures
powered by the sun, and therefore do not require from key structural elements of the span Intended for
battery maintenance. long term monitoring, these new installations overcome
MicroStrain has previously described battery the limitations of older types, which required that the
powered wireless strain sensors for structural health wireless node’s batteries be replaced or recharged
monitoring.13,14,15 One example is the Ben Franklin periodically. Maintenance of batteries is simply not
Bridge, which spans the Delaware River from practical on bridges, where sensor nodes must be
Philadelphia, PA to Camden, NJ. The monitoring placed on, under, and within the structure in locations
system was accessed remotely over commercial which may be extremely difficult to access.
cellular telephone networks, and sensor data were The Connecticut Goldstar bridge program is a
provided to the customer via secure access to a web- long term project developed to learn how bridge
based server. The wireless nodes measured structural monitoring systems can be used for evaluation of in-
strains in the cantilever beams as passenger trains service behavior, for long-term structural health
traversed the span. Measurements taken over several monitoring of each bridge, and for assisting the
months’ time were used to document the bridge’s Connecticut Department of Transportation to manage
cyclic structural strains under contract from the the State’s bridge infrastructure.17
Delaware River Port Authority (DRPA).
For the Ben Franklin Bridge, at the locations
tested, the measured strains and calculated stresses
were far below the endurance limit. Repeated cyclic
stress above the endurance limit results in the gradual
reduction of strength, or fatigue. Therefore, bridges are
designed to operate at stress levels below this limit.
From the information automatically collected by the
wireless strain sensors, DRPA engineers concluded that
cyclic stress fatigue due to train crossings was not a
problem.16
MicroStrain’s first solar powered bridge
installation was recently made in Corinth, Greece.
This system uses arrays of wireless tri-axial
accelerometer nodes to monitor the span’s background
vibration levels at all times. Each node and solar panel
are packaged within watertight enclosures for outdoor
use. In the event that seismic activity is detected at any
one of the nodes, the entire wireless network of nodes
is alerted, and data are collected simultaneously from

Figure 8. Solar powered wireless G-Link ® seismic

sensors on Corinth Bridge, Greece. High gain antenna
(top left) was used to provide a wireless communications
range of 150 meters.
Figure 7. Solar powered wireless G-Link ® seismic
sensors on Corinth Bridge, Greece
6
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
3.3 Inertial Sensing Suite
The inertial sensing suite (ISS) consists of orthogonal
arrays of microelectromechanical systems (MEMS)
based angular rate sensors, accelerometers, and
magnetometers. Each of these MEMS sensors is fully
temperature compensated to minimize their individual
gain and offset errors, with the compensation
coefficients burned into the unit’s on-board non-
volatile memory (EEPROM). During factory
calibration, each ISS is also corrected for its unique
mechanical misalignment among the nine MEMS
sensors as well as its unique magnetic signature to
correct for distortions of Earth’s field as detected by
the three magnetometers. These alignment and
magnetic correction matrices are also burned into each
unit’s EEPROM. Sculling and coning errors are
minimized by using a dedicated, 16 bit resolution,
sigma-delta A/D converter. The dedicated A/D
converters sample each MEMS sensor at a rate of
2 KHz, with the data are recorded at a rate of 200 Hz,
which removes unwanted dynamic transients - such as
may arise from vehicle vibration - from the orientation
calculations.
The data from all nine sensors are used to
compute orientation, which may be reported in Euler,
Quaternion, or Matrix form. The structural monitoring
3.4 Wireless Sensor Communications
Radio communication between the wireless sensor
nodes and base stations is based on the IEEE 802.15.4
standard in the 2.4 GHz band. One of the major
advantages of this frequency band is that it may be
used, license free, worldwide. The transceiver we are
currently using is available commercially through
Texas Instruments, model CC2420 (Austin, Texas).
There are 16 distinct communications channels in this
band, which may be software selected. As shown in
Figure 10 below, channels 15, 20, 25, and 26 are non-
overlapping with 802.11b in North America, and
channels 15, 16, 21, and 22 are non-overlapping in
Europe.
The CC2420 radio transceiver has features
which are well suited to our energy harvesting wireless
sensor node applications. This transceiver can be
powered up very quickly (~0.7 milliseconds) which
conserves energy, since the board’s main
microprocessor powers up the radio transceiver only
when appropriate.
As we have previously described (8), the
energy consumed by the radio can be further conserved
by reducing the number of packets sent per second.
For our wireless helicopter pitch link applications,
strain data were logged at a specified rate (64 Hz), and
once 100 samples were acquired, the system
transmitted these data. By buffering time stamped data
and transmitting these data in a single packet, the
“digital process overhead” associated with the framing
and checksum bytes was minimized.
 
Figure 9. Inertial Sensing Suite includes orthogonal
arrays of temperature compensated gyros,
accelerometers, and magnetometers, along with an
embedded microprocessor.
system records and time stamps these data using a
digital serial interface. A photograph of our inertial
sensing suite is provided in Figure 9.
Our protocols for wireless communications
are scalable. By combining time division multiple
access (TDMA), carrier sense multiple access
(CSMA), and frequency division multiple access
(FDMA), we can support a large number of wireless
sensor nodes. For example, assuming a network of
wireless strain nodes were configured to sample a tri-
axial strain gauge rosette at a rate of 5 samples per
second, this system will support up to 600 distinct
wireless nodes (1800 strain gauges) using only TDMA
and CSMA techniques on a single radio
communication channel. By adding radio transceiver
chips within the WSDA base station, the system will
theoretically support 16 of these strain sensing
networks, or as many as 1800 strain gauges*16 radio
channels = 28,800 individual strain gauges.
An important component on each wireless
node is an independent, precision, nanopower real time
clock (RTC), with a +/- 3 part per million (PPM) time
reference. The real time clocks on all wireless and
wired sensor nodes are synchronized at the beginning
of a test to the base station’s time reference, using a
wireless beacon to communicate that reference. The
WSDA base station uses a built-in hard-wired Global
Positioning System (GPS). In case GPS is not
available, the WSDA uses its internal +/- 3 PPM real
time clock as the timing reference to insure
synchronization of all the remote sensor nodes to the
WSDA’s clock. A detailed description of the timing
engine methodology is provided in the next section.
7
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
Figure 10. IEEE 802.15.4 and IEEE 802.11b channel selections in the 2.4 GHz frequency band
3.5 Sensor Data Aggregation
The wireless sensor data aggregator (WSDA) is
responsible for data collection from several different
buses and networks, including the CAN bus, USB, and
802.15.4 wireless networks. A photograph of this
device is provided below in Figure 11.
Figure 11. MicroStrain’s Wireless Sensor
Data Aggregator (WSDA)
 
Important design criteria for the WSDA were:
• Open architecture operating system
• Provides time synchronization platform
• Data saved in scalable sensor database,
allowing multiple sensor types to be archived
in a single file
• Data collected at multiple rates aggregated
into single data base
• Data sorted based on parameters such as time
stamp or sensor type
• Multiple bus interfaces supported:
• CAN
• IEEE802.15.4
• Ethernet
• RS232/RS422 (HUMS)
One of the primary challenges for this
distributed multi-network topology is synchronizing all
data acquisition points, hereafter referred to as nodes,
throughout the entire system. Figure 12 provides a
block diagram of the WSDA and associated hard-wired
and wireless nodes. The WSDA’s functional blocks
include the GPS receiver, timing engine,
microprocessor core running Linux 2.6, CAN bus
controller, and wireless controller. Time
synchronization for hard wired and wireless nodes was
accomplished by periodic beaconing of a reference
timing signal, as described in the sections that follow.
8
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

GPS
Receiver CAN
CAN Nodes
Nodes
CAN CAN
CANNodes
Nodes
Controller

Timing
Wireless Wireless
Wireless Nodes
Engine WirelessNodes
Nodes
Controller Wireless Nodes

Wireless Wireless Nodes

Wireless
Wireless Nodes
Controller WirelessNodes
Nodes
µP Core
running
Linux 2.6

USB Node
SYNCH CLK and TRIG

USB Node
CAN Synch Mechanism

RS-232 Node
Wireless Synch Beacon
RS-232 Node

Figure 12. Block Diagram for wireless and hard-wired (networked) SHM System

3.6 Temperature Compensated Timing Engine

The data aggregator and wireless sensor nodes feature
a highly-accurate, temperature compensated timing
engine. The temperature compensation method relies
on an onboard temperature sensor, calibrated software
look-up tables, and an oscillator tuning mechanism to
maintain accurate frequency vs. temperature. The
timing engine is used on the data aggregator to provide
a stable 1 Hz reference for the synchronization beacon.
On the wireless sensor nodes, the same timing engine
is slightly modified to provide adjustable output from
1 to 4096 Hz, which is used to drive a sensor-sampling
interrupt on the host processor.

3.7 Wireless Synchronization Beacon

Timing drift among wireless sensor nodes can be
corrected by a centrally broadcast synchronization
beacon. There are several requirements for practical
implementation: First, the wireless network controller
must be capable of transmitting a single broadcast or
multi-cast addressable packet that can be received by
all nodes simultaneously. The IEEE 802.15.4 radio
offers a broadcast addressing feature that satisfies this
requirement. Second, the timing for the
synchronization packet transmission and reception
should be as fixed and deterministic as possible. On
the data aggregator, packet transmission is initiated by
a hardware interrupt, which is driven by a precise
timing source (GPS or aforementioned timing engine).
The 802.15.4 carrier sense multiple access (CSMA)
function is disabled to ensure that the beacon packet is
transmitted immediately. On the receive side, the
chosen 802.15.4 radio uses a hardware state machine to
automatically process, decode, and error check all
incoming packets before sending an interrupt request to
the host processor. With the packet transmission and
reception functions handled in hardware, the
communication latency is minimal and relatively
deterministic.

9
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008
3.8 Wireless Digital Communications Latency
To characterize this single packet communication
latency, a digital oscilloscope was used to measure the
time between the completion of packet transmission on
the wireless network controller, and completion of
packet reception on the wireless sensor node. Due to
the relatively short (<70 m) communication range of
the 802.15.4 radio, it is assumed that differences in the
3.9 Timing Synchronization Test Methodology
To determine the timing accuracy of the beaconing
method, both long term tests (13 hour) and short term
(2 hour) tests were performed. The first long term test
was designed to compare data synchronization without
any form of beacon correction, while the second long
term test was designed to compare data
synchronization with the beacon correction enabled.
For the beacon-enabled test, each node was
configured to resynchronize to the beacon every
60 seconds. The testing hardware consisted of one
MicroStrain WSDA, and two MicroStrain G-Link
wireless accelerometer nodes, each G-LINK with
integral precision timekeepers, mounted next to each
other on a rigid aluminum plate as shown in Figure 13.
The wireless accelerometer sensor nodes were
configured to sample at a rate of 128 Hz, with timing
sourced from the embedded temperature-compensated
timing engine.
The data aggregator was tested at room
temperature (25 degrees C). The wireless
accelerometer nodes were placed in a temperature
controlled environmental chamber for both long term
tests, with the node electronics exposed to open air
within the chamber, and no thermal insulation of any
form. Each 13 hour test was run while thermally
cycling the nodes between -20 and 60 degrees C at a
ramp rate of 10 degrees C per minute. After 50 to 60
thermal cycles, the aluminum plate was tapped and
shaken in a manner to provide common inertial
excitation to each wireless accelerometer node.
The short duration (two hour) tests used two
independent wireless strain sensing nodes (SG-LINKs,
MicroStrain, Inc.) with integral precision timekeepers.
In the each of these tests, the timing beacon was sent
only once – at the start of the test. In the first short
term test, the WSDA and SG-LINK strain sensing
nodes were exposed to room temperature only. In the
second short term test, multiple cycles of extreme hot
and cold temperatures were used in an attempt to
destabilize the SG-LINK node’s embedded
timekeepers.
Each SG-LINK node was provided with a
common analog input voltage, provided by a rack-
mounted laboratory function generator (B and K
Precision, model 4011A). A saw-tooth waveform of
10 Hz frequency and 2.8 volts magnitude from the
function generator was used as direct input to each of
 
propagation delay are negligible (<300 ns) and
therefore ignored. The test was performed with the
node and the wireless network controller separated by a
distance of 2 meters. Measured latency values fell
within the range of 3.8 to 4.0 microseconds, and jitter
never exceeded 200 nanoseconds as tested at
25 degrees C.
Figure 13. Set of two wireless accelerometer nodes
(G-LINK®), mounted on a rigid aluminum plate
the SG-LINK nodes. This type of waveform was
selected because the inflection point (peak) of a saw-
tooth waveform may be identified even in the case that
the data recorded by the sensor nodes were to miss the
precise instant that the peak occurred. The wireless
strain sensor nodes were configured to sample at a rate
of 128 Hz, with timing sourced from the embedded
temperature-compensated timing engine.
For the second (elevated temperature) test, the
wireless SG-LINK nodes were both placed in a
temperature controlled environmental chamber, with
the node electronics exposed to open air within the
chamber, and no thermal insulation of any form. Each
test was run while thermally cycling the nodes between
-40 and 85 degrees C at a ramp rate of 12.5 degrees C
per minute, for 2 hours duration, which provided ~6
complete thermal cycles.
The Data Aggregator a transmitted a periodic,
programmable time interval synchronization beacon,
with timing sourced from an embedded, temperature
compensated timing engine. The Data Aggregator
collected data from the wireless nodes and stored it on
a Structured Query Language (SQL) database hosted
on a 4GB compact flash card. All sensor data were
downloaded in Excel (.csv) format from the data
aggregator and plotted for analysis.
10
 Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

3.10 Remote Data Reporting

To enable report data reporting, the WSDA includes an
internal cellular telephone modem and, will, in the
future, support an external satellite modem. The
remote cellular (GSM/GPRS) terminal interface uses a
PPP protocol for secure transfer of data to a remote
server. We selected a cellular modem based on a trade
study, and chose the GSM/GPRS modem because it
offered the widest global cellular coverage. We have
also identified an embeddable satellite modem card that
utilizes the Inmarsat Broadband Global Area Network
(BGAN). This network provides the best combination
of worldwide coverage and high data rate at reasonable
cost per byte of data transferred.
Data sets stored on the WSDA were
transferred to our web server for development, test, and
demonstration purposes. A flash and HTML web
interface was created to enable ongoing, completely
independent structural monitoring by MicroStrain’s
OEM customers. In order to provide secure access to
the data stored on our remote server, a password
protected user management and authentication process
was implemented. A dedicated server hosting
company was selected to provide a scalable, reliable,
and secure dedicated server to support the Navy in the
future.

4.0 RESULTS
For the long term (13 hour duration) tests, and subject
to 60 cycles of temperatures ranging from
-20 to 60 degrees C, with a timing beacon sent at the
onset of the test only, the synchronization accuracy was
measured at 39 milliseconds (Figure 14). This timing
offset constitutes a relative (node-to-node) drift rate of
0.8 parts per million (ppm), where drift =
0.039 seconds / 48000 seconds.
In contrast, the 13 hour test with 60 second
timing beacon produced no discernible timing offset
between the two wireless nodes (Figure 15). In theory,
with a beacon interval of 60 seconds, the maximum
timing offset should be 60 seconds multiplied by the
maximum drift rate. Using the relative drift rate
measured from the first test, this would amount to
60 seconds x 0.8 ppm = 48 microseconds. With a
sample rate of 128 Hz, the timing resolution of the test
was inherently limited to 7.8 milliseconds. In this case,
we were unable to detect so small a magnitude of the
timing offset, which we estimate to be less than
50 microseconds.

Synchronization Test without Beacon

39 ms drift between Node #1 and Node #2
Test Duration = 13.3 hours
Temp Cycled from -20 C to 60 C, 50 times
4
Node #1 Z-axis
Node #2 Z-axis
3.5

2.5

1.5

1
)
n(g

0.5
tio
ra

0
c
acle
e

-0.5

-1

-1.5

-2

-2.5

-3

-3.5

-4
49499 49499.25 49499.5 49499.75 49500 49500.25 49500.5 49500.75 49501 49501.25 49501.5
time (s)

Figure 14. Timing synchronization without beaconing and with exposure to cycles of extreme
temperatures of -20 to 60 degrees C; timing accuracy result: 39 milliseconds over ~13 hours
11
 Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

These results prompted us to perform the saw
tooth waveform tests, in order to better characterize the
system’s true timing accuracies that could be expected
with beaconing. For the first 2 hour room temperature
test, with beacon correction provided only at the start
of the test, and with the strain nodes exposed only to
room temperatures, yielded a timing offset of
325 microseconds between the two wireless strain
nodes. This timing offset is barely visible in Figure 16;
constituting a relative (node-to-node) drift rate of 45
parts per billion (ppm), where drift = 325 microseconds
/ 7200 seconds.
For the second 2 hour test, again, with beacon
correction provided only at the start of the test, and
with the strain nodes cycled from -40 to +85 degrees C,
produced a timing offset of 5.71 milliseconds between
the two nodes, as shown in Figure 17. This constitutes
a drift rate of 793 parts per billion.
An additional 2 hour test was performed (over
temperatures of -40 to +85 degrees C) using a lower
frequency saw-tooth wave voltage input of 1Hz, in
order to check our prior result. This additional short
term test yielded a timing offset between the two SG-
LINK nodes of 5.04 milliseconds, or 700 parts per
billion, which agreed closely with our prior result.

Synchronization Test with Beacon

Beacon Resynch Interval = 60 s
Test Duration = 13.3 hours
Temp Cycled from -20 C to 60 C, 50 times
1.5
Node #1 Z-axis
Node #2 Z-axis

0.5
ration(G)

0
accele

-0.5

-1

-1.5
50060 50060.5 50061 50061.5 50062 50062.5 50063 50063.5
time (s)

Figure 15. Timing synchronization with beaconing and with exposure to cycles of extreme
temperatures of -20 to 60 degrees C; timing accuracy result: no discernible timing difference

12
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

Figure 16. Timing synchronization w/ beacon sent once –

at start of 2 hour room temperature test: 325 microseconds

Figure 17. Timing synchronization w/ beacon sent once –

at start of 2 hour exposure to -40 to +85 degrees C: 5.7 milliseconds

13
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

5.0 CONCLUSIONS synchronization may be adequate for many SHM

Combining advanced microelectronics with sensors,
energy harvesting methods, power management, data
collection, and web-based data distribution is a
powerful solution for structural health monitoring
(SHM) problems. Energy harvesting represents an
exciting development, with the potential to eliminate
battery maintenance for hard to access sensor nodes.
A wireless sensor data aggregation node
(WSDA) has been developed which is capable of
collecting time stamped data from both wired and
wireless sensor networks and sending these data to a
secure server on the internet.
Precision time keepers, embedded within each
sensor node, are synchronized to the WSDA using a
beaconing method to provide a periodic timing
reference. Synchronization of the sensor network
provides several key advantages, including enhanced
scalability of wireless communications, and the ability
to store all sensor data in a single time stamped
parametric database.
With no beacon except at the start of a 13
hour long test, the system maintained a timing
synchronization of 39 milliseconds, while exposed to
continuous, 8 minute duration thermal cycles ranging
from -20 to 60 degrees C. This level of timing
applications. The system achieved a timing
synchronization accuracy of ~5 milliseconds with a
timing beacon sent every 2 hours. This accuracy
improves when the thermal environment is stable. The
timing accuracy is also improved by sending the
beacon more frequently. For flight tests that require a
synchronization of sensor data to sub-millisecond
accuracies, a conservative approach would be to
provide a synchronization beacon every 20 minutes.
For those applications where the timing
synchronization may not need to achieve sub-
millisecond accuracies, all wireless sensor radio
communications may be turned off completely during
flight. The wireless nodes would then be used in data-
logging mode only, which conserves power and
eliminates all propagation of 2.4 GHz radio energy
from the nodes during flight. Time stamped data can
be collected from the wireless nodes after the flight is
completed.
The system is capable of remote reporting
using mobile phone networks, with satellite reporting
currently under development. These capabilities,
coupled with appropriate wireless security methods,
will enable critical structural sensor data to be managed
remotely, securely, and automatically.

6.0 ACKNOWLEDGEMENTS

MicroStrain, Inc. gratefully acknowledges the support

of the US Navy’s SBIR program and the Office of
Naval Research BAA program.

7.0 REFERENCES
1
The Economist, Special Report on Telecoms, April 28th-May 4th 2007, pages 3-18
2
El-Bakry, M., Proc. 5th International Workshop on Structural Health Monitoring, Stanford University, Stanford, CA,
Sep 13th, 2005, pages 521-528
3
Maley, S., Plets, J., Phan, N.: Proc. American Helicopter Society 63rd Annual Forum, Virginia Beach, VA, May 1-3,
2007, CONF 63; VOL 2, pages 1456-1467, ISSN 0733-4249
4
Vachon, W., Hovis, G. et al. “Crack Detection and Monitoring Crack Growth in Fastener Holes Using DMI Optical
SR-2 Strain Measurement Technology” , Direct Measurement Inc. Press Release, Atlanta, GA, 2006
5
Arms, S.W., Townsend, C.P., Churchill, D.L., Moon, S.M., Phan, N., Proc. American Helicopter Society Annual
Forum 62, Health and Usage Monitoring Systems (HUMS), Phoenix, AZ, May 9-11, 2006, CONF 62; VOL 2, pages
1336-1341, ISSN 0733-4249
6
Arms, S.W., Churchill, D.L., Townsend, C.P., Galbreath, J.H.: Proc. SPIE’s Symposium on Smart Structures and
Materials 2005: Smart Electronics, MEMS, BioMEMS, and Nanotechnology, San Diego, CA, March 2005, Vol. 5763,
page 267; DOI:10.1117/12.600302
7
Moon, S., Menon, D., Barndt, G.: Proc. American Helicopter Society Forum 52, Washington, DC, June 4-6, 1996
8
Arms S.W., Townsend, C.P., Churchill, D.L., Augustin, M., Yeary, D., Darden, P., Phan, N., Proc. American
Helicopter Society Forum 63, Virginia Beach, VA, May 2007, CONF 63; VOL 2, pages 934-941, ISSN 0733-4249
9
Le Cam, V., LeMarchand, L., Cottineau, L.-M.: Proc. Third European Workshop on Structural Health Monitoring,
July 5-7, Granada, Spain, July 5-7, 2006, pages 1339-1347
10
Hamel et al, “Energy Harvesting for Wireless Sensor Operation and Data Transmission”, US Utility Patent number
7081693, filed March 2003
14
Energy Harvesting, Wireless SHM and Reporting System              2nd Asia‐Pacific Workshop on SHM, Melbourne, 2‐4 December 2008 

11
Churchill, D. L. Hamel, M. J. Townsend, C. P. Arms, S. W.: Proc. SPIE’s Symposium on Smart Structures and
Materials 2003: Smart Electronics, MEMS, BioMEMS, and Nanotechnology, San Diego CA, 3-5 March 2003, Volume
5055, pages 319-327, ISSN 0277-786X
12
Arms, S.W., Townsend C.P., Churchill, D.L., Hamel, M.J., Augustin, M., Yeary, D., Phan, N.: Proc. International
Workshop on Structural Health Monitoring, Stanford, CA, 11 September 2007, pages 1741-1748
13
Townsend C.P., Hamel, M.J., Arms, S.W.; Proc. Sensors for Industry Conference, 2002. 2nd ISA/IEEE
Publication Date: 19-21 Nov. 2002, pages 172- 178
14
Galbreath J.H., Townsend, C.P., Mundell, S.W., Arms, S.W: Proc. International Workshop for Structural Health
Monitoring, Stanford, CA, September 2003, pages 1215-1222
15
Arms, S.W., Galbreath, J.H., Newhard, A.T., Townsend, C.P.: Proc. Structural Materials Technology VI, NDE/NDT
for Highways and Bridges, Buffalo, NY, 16 Sep 2004, pages 331-338
16
Rong, A.Y. and Cuffari, M.A.: Proc. Structural Materials Technology VI, NDE/NDT for Highways and Bridges,
Buffalo, NY, 16 Sep 2004, pages 327-330
17
DeWolf, J.T.: Proc. Third European Workshop on Structural Health Monitoring, Granada, Spain, July 5-7, 2006,
pages 372-378

15