10 1 1 126 6482 PDF

The Shock and Vibration Digest
http://svd.sagepub.com
A Summary Review of Wireless Sensors and Sensor Networks for Structural Health Monitoring
Jerome P. Lynch and Kenneth J. Loh
The Shock and Vibration Digest 2006; 38; 91
DOI: 10.1177/0583102406061499
The online version of this article can be found at:

http://svd.sagepub.com/cgi/content/abstract/38/2/91
Published by:
http://www.sagepublications.com
Additional services and information for The Shock and Vibration Digest can be found at:
Email Alerts: http://svd.sagepub.com/cgi/alerts
Subscriptions: http://svd.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations (this article cites 65 articles hosted on the

SAGE Journals Online and HighWire Press platforms):
http://svd.sagepub.com/cgi/content/refs/38/2/91
Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

© 2006 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.
Articles
A Summary Review of Wireless Sensors and Sensor

Networks for Structural Health Monitoring
Jerome P. Lynch and Kenneth J. Loh
ABSTRACT—In recent years, there has been an increasing community is aggressively pursuing novel sensing technolo-
interest in the adoption of emerging sensing technologies for gies and analytical methods that can be used to rapidly iden-
instrumentation within a variety of structural systems. Wire-
less sensors and sensor networks are emerging as sensing tify the onset of structural damage in an instrumented structural
paradigms that the structural engineering field has begun to system (Liu and Tomizuka, 2003a, 2003b). Called structural
consider as substitutes for traditional tethered monitoring health monitoring (SHM), this new paradigm offers an auto-
systems. A benefit of wireless structural monitoring systems mated method for tracking the health of a structure by com-
is that they are inexpensive to install because extensive wir- bining damage detection algorithms with structural monitoring
ing is no longer required between sensors and the data
acquisition system. Researchers are discovering that wire- systems.
less sensors are an exciting technology that should not be Structural monitoring systems are widely adopted to mon-
viewed as simply a substitute for traditional tethered monitor- itor the behavior of structures during forced vibration testing
ing systems. Rather, wireless sensors can play greater roles or natural excitation (e.g. earthquakes, winds, live loading).
in the processing of structural response data; this feature
can be utilized to screen data for signs of structural damage. Structural monitoring systems can be found in a number of
Also, wireless sensors have limitations that require novel common structures including aircrafts, ships, and civil struc-
system architectures and modes of operation. This paper is tures. For example, some building design codes mandate that
intended to serve as a summary review of the collective structures located in regions of high seismic activity have
experience the structural engineering community has gained structural monitoring systems installed (International Con-
from the use of wireless sensors and sensor networks for
monitoring structural performance and health. ference of Building Officials, 2002). The monitoring system
is primarily responsible for collecting the measurement out-
KEYWORDS: wireless sensors, structural monitoring, dam- put from sensors installed in the structure and storing the
age detection, smart structures, decentralized computing measurement data within a central data repository. To guar-
antee that measurement data are reliably collected, structural
monitoring systems employ coaxial wires for communica-
tion between sensors and the repository. While coaxial wires
1. Introduction provide a very reliable communication link, their installation
in structures can be expensive and labor-intensive. For exam-
Structures, including bridges, buildings, dams, pipelines, ple, structural monitoring systems installed in tall buildings
aircraft, ships, among others, are complex engineered systems have been reported in the literature to cost in excess of
that ensure society’s economic and industrial prosperity. To $5000 (USD) per sensing channel (Celebi, 2002). As struc-
design structures that are safe for public use, standardized tural monitoring systems grow in size (as defined by the total
building codes and design methodologies have been created. number of sensors), the cost of the monitoring system can
Unfortunately, structures are often subjected to harsh load- grow faster than at a linear rate. For example, the cost of
ing scenarios and severe environmental conditions not antic- installing over 350 sensing channels upon the Tsing Ma sus-
ipated during design that will result in long-term structural pension bridge in Hong Kong is estimated to have exceeded
deterioration. For example, recent seismic events, including $8 million (Farrar, 2001). The high cost of installing and main-
the Loma Prieta (1989), Northridge (1994), Kobe (1995), and taining wires is not restricted only to civil structures. Others
Chi-Chi (1999) earthquakes, reveal civil structure vulnera- have reported similar issues with respect to the costs associ-
bility to damage and failure during natural catastrophes. To ated with monitoring systems installed within aircrafts, ships,
design safer and more durable structures, the engineering and other large structural systems (MacGillivray and God-
dard, 1997).
Damage detection methods provide engineers with auto-
Professor J. P. Lynch (jerlynch@umich.edu) and K. J. Loh, Department of mated tools that can be used to screen response data for signs
Civil and Environmental Engineering, The University of Michigan, 2350 of structural distress. Over the past decade, a large number
Hayward Street, Ann Arbor, MI 48109-2125, USA. of damage detection methods have been proposed, as
The Shock and Vibration Digest, Vol. 38, No. 2, March 2006 91–128 reported by Doebling et al. (1998) and Sohn et al. (2004).
©2006 SAGE Publications Damage detection methods can generally be classified as
DOI: 10.1177/0583102406061499 one of two types: local-based or global-based damage detec-
Figures 1, 3–5 appear in color online: http://svd.sagepub.com tion methods. Local-based damage detection methods attempt

92 The Shock and Vibration Digest / March 2006
to identify damage based on screening structures at their com- motivator for considering the installation of wireless sensors
ponent or subcomponent length-scales. Many non-destructive in structures, the fact that wireless sensors are a new sensing
evaluation (NDE) technologies, including ultrasonic inspec- paradigm offering autonomous data processing is fueling recent
tion, can be classified as supporting local-based damage excitement. Specifically, wireless sensors proposed for SHM
detection. While local NDE is suitably scaled to the structural will be responsible for screening their own measurement data
damage phenomena (e.g. cracks, yielding), local-based inspec- to identify the possible existence of damage. Already, many
tion technologies generally require a trained professional to data processing algorithms have been embedded in wireless
operate in the field, thereby raising their costs. Furthermore, sensors for autonomous execution.
the operator must have knowledge of potential damage regions With wireless sensors rapidly evolving in multiple engineer-
to prioritize inspection of the complete structure. For exam- ing disciplines, there currently exist a large number of dif-
ple, post-Northridge structural inspections discovered severe ferent academic and commercial wireless sensor platforms.
fatigue cracking of steel moment frame connections. As a In the first half of this paper we provide a detailed summary
result of this discovery, all steel moment frame connections of the current inventory of wireless sensors that have been
in the Los Angeles region have been inspected using ultra- explored by researchers for structural monitoring. This sum-
sonic NDE; the cost of inspection is reported as $200 to $1000 mary is delineated into two parts: academic and commer-
per welded connection (Hamburger, 2000). cial platforms. The majority of the wireless sensors described
Global-based damage detection refers to numerical methods herein are passive wireless sensors. Similar to traditional
that consider the global vibration characteristics (e.g. mode cabled sensors, these passive wireless sensors only measure
shapes, natural frequencies) of a structure to identify damage. structural responses due to static and dynamic loadings. This
Global-based damage detection was initially proposed as a is in contrast to active sensors that can interact with or excite
result of the availability of structural monitoring systems that a structure when desired.
could be installed in a structure to collect response time his- As costs continue to decline and field deployments of wire-
tories. However, with tethered structural monitoring systems less sensors are defined by ever higher nodal densities, local-
expensive to install, the nodal densities of most systems have based damage detection is becoming increasingly attractive.
been low (often, only 10–20 sensors are installed in a single Active sensors, such as piezoelectric pads, are proving to be
structure). Such small numbers of sensors are poorly scaled a powerful sensing technology that is ideally suited for local-
to the localized behavior of damage, often rendering global- ized SHM (Park et al., 2000, 2003; Wu and Chang, 2001).
based damage detection difficult to implement. Particularly To take full advantage of the benefits of active sensing, some
for structures exposed to widely varying environmental and wireless sensors are now being designed with actuation inter-
operational loadings, such as civil structures (e.g. bridges, faces to which active sensors can be attached. Wireless sen-
buildings, dams), damage detection using global vibration char- sor prototypes that are capable of achieving active sensing
acteristics is even more challenging (Doebling et al., 1998). are also described in detail in this review paper.
To address the limitations current sensing technologies place Recognizing power consumption to be a major limitation
on both local- and global-based damage detection methods, of wireless sensors operating on batteries, some researchers
the research community is actively exploring new technolo- are exploring the development of power-free wireless sen-
gies that can advance the current state-of-practice in structural sors known as radio-frequency identification (RFID) sen-
monitoring and SHM. In particular, wireless sensors repre- sors. RFID sensors are a passive radio technology, which
sent one potential sensing technology that can help advance capture radio energy emanated from a remote reader so that
the structural engineering field’s ability to economically it can communicate its measurement back. RFID sensors
realize SHM. Interest in wireless sensors was initially moti- explicitly developed for structural monitoring are also
vated by their low-cost attributes. The eradication of exten- included as part of the paper’s scope.
sive lengths of coaxial wires in a structure results in wireless With wireless sensors offering impressive computational
systems having low installation costs. These low costs prom- resources for processing data, hardware only represents one-
ise wireless monitoring systems defined by greater nodal half of the complete wireless sensing unit design; software
densities as compared to traditional tethered monitoring sys- embedded in the wireless sensor represent the second half.
tems. With potentially hundreds of wireless sensors installed With computational power coupled with the sensor, wireless
in a single structure, the wireless monitoring system is also sensors are capable of autonomous operation. Without a phys-
better equipped to screen for structural damage by monitoring ical link existing between individual wireless sensors and the
the behavior of critical structural components, thereby imple- remainder of the wireless sensor network, wireless sensors
menting local-based damage detection. must know when to act autonomously or collaboratively. Soft-
Wireless sensors are not sensors per se, but rather are auton- ware embedded in the wireless sensor’s computational core
omous data acquisition nodes to which traditional structural is responsible for its autonomous operation including the
sensors (e.g. strain gages, accelerometers, linear voltage dis- collection and storage of data, interrogation of measurement
placement transducers, inclinometers, among others) can be data, and deciding when and what to communicate to other
attached. Wireless sensors are best viewed as a platform wireless sensors in the wireless sensor network. Embedded
in which mobile computing and wireless communication software can be classified as one of two types: the operating
elements converge with the sensing transducer. Perhaps the system (OS) and engineering analysis software. The OS takes
greatest attribute of the wireless sensor is its collocation of control of the operation of the unit and is intended to serve as
computational resources with the sensor. Such resources can an abstraction layer that hides the implementation details of
be leveraged to allow the sensor to perform its own data hardware from upper engineering analysis layers. The second
interrogation tasks. This capability is particularly attractive layer is where algorithms designed to autonomously interro-
within the context of SHM. So while cost has been an early gate structural response data are stored. In this paper, the var-

Lynch and Loh / A SUMMARY REVIEW OF WIRELESS SENSORS AND SENSOR NETWORKS 93
Figure 1. Functional elements of a wireless sensor for structural monitoring applications.
ious software options for wireless sensors are described. An sors into a digital representation that can be understood and
emphasis is placed on embedded engineering analyses, includ- processed by digital electronics. The quality of the sensor
ing damage detection algorithms, which have already been interface is a function of the conversion resolution, sample
embedded in the computational cores of wireless sensors. rate, and number of channels available on its analog-to-digital
A true test of a new emerging sensing technology is its converter (ADC). Selection of an appropriate sensing inter-
performance in the field. The research community has installed face must be done in consultation with the needs of the mon-
wireless structural monitoring systems upon a diverse set of itoring application. For most structural monitoring applications,
structures to assess the performance of wireless sensors within an analog-to-digital conversion resolution of 16-bits or higher
the complex and challenging field environment. In the liter- is preferred. Ordinarily, low sampling rates (e.g. less than
ature, a large number of validation tests have been performed 500 Hz) are adequate for global-based structural monitoring.
on laboratory structures as well as upon bridges, buildings, However, wireless sensors are increasingly explored for use
aircraft, offshore oil platforms, naval ships, among many oth- in acoustic and ultrasonic NDE; as a result, there has been a
ers. In this paper we provide a detailed description of the growing need for higher sampling rates in excess of 500 kHz
current state of experimentation with wireless sensors in the (Grisso et al., 2005; Lynch 2005).
laboratory and the field. Once measurement data have been collected by the sensing
We conclude this summary review with our outlook upon interface, the computational core takes responsibility of the
the future directions of wireless sensors and sensor networks data, where they are stored, processed, and readied for com-
for SHM. With wireless sensing technology still in its infancy, munication. To accomplish these tasks, the computational core
much work remains for bringing this promising technology is represented by a microcontroller that can store measure-
to widespread use in all types of structures. In particular, ment data in random access memory (RAM) and data inter-
future research studies are needed on challenging issues such rogation programs (such as damage detection routines) in
as power consumption, time synchronization, multiscale net- read only memory (ROM). A broad assortment of microcon-
work topologies, decentralized data processing within large- trollers is commercially available. A major classifier for
scale networks, and formulation of power-efficient data driven microcontrollers is the size (in bits) of their internal data bus
usage strategies. with most microcontrollers classified as 8-, 16-, or 32-bits.
While larger data buses suggest higher processing throughput,
2. Hardware Design of Wireless Sensor both cost and power consumption of these microcontrollers
Platforms for Structural Health Monitoring are also higher (Gadre, 2001). An internal element of every
microcontroller is a clock. The speed of the clock is a direct
The fundamental building block of any wireless sensor measure of how fast embedded programs will be executed
network is the wireless sensor. Selection of an appropriate by the microcontroller. Again, as the speed of the microcon-
wireless sensor is necessary because the performance of the troller increases, there is a linear increase in power consumed.
entire wireless structural monitoring system is dependent If the size of the internal RAM and ROM memory is inade-
upon the individual wireless sensor. As shown in Figure 1, quate, additional external memory can be added to the com-
all wireless sensors can generally have their designs deline- putational core design.
ated into three or four functional subsystems: sensing inter- To have the capability to interact with other wireless sen-
face, computational core, wireless transceiver and, for some, sors and to transfer data to remote data repositories, a wire-
an actuation interface. less transceiver is an integral element of the wireless sensor
Wireless sensors must contain an interface to which sens- design. A radio transceiver is an electrical component that
ing transducers can be connected. The sensing interface is can be used for both the transmission and reception of data.
largely responsible for converting the analog output of sen- Similar to microcontrollers, a plethora of radios are readily

available for integration with a wireless sensor. Thus far, the ated by the microcontroller into a continuous analog voltage
majority of wireless sensors proposed for use in structural output (which can be used to excite the structure).
monitoring have operated on unlicensed radio frequencies. As simple as a wireless sensor may appear, many challenges
In the United States, 900 MHz, 2.4 GHz, and 5.0 GHz, have are associated with their design and use. In particular, their
been designated by the Federal Communications Commis- design requires a rational analysis to determine the trade-off
sion (FCC) as the unlicensed industrial, scientific, and medical between functionality and power consumption, with func-
(ISM) frequency bands. Many of today’s wireless technolo- tionality often coming at the cost of power. For example, larger
gies (e.g. 802.11, Bluetooth, Zigbee) operate on the same set communication ranges or greater computational power will
of frequency bands. If a wireless radio operates on the ISM result in greater electrical energy consumption by the wire-
frequencies, the FCC mandates the maximum power an less sensor. Since the integration of wireless communication
antenna can output is 1W, which effectively limits the trans- removes the need for transmitting data from one point to
mission range. another with cables, the lack of cables requires remote power
There exist two types of wireless signals that can be sent generation or portable power supplies to be coupled with
upon a selected radio band: narrow-band and spread spec- wireless sensors. Currently, batteries represent the most com-
trum signals. Narrow-band wireless transmission modulates mon portable power source for wireless sensors. However,
all of the data upon a single carrier frequency. Unfortunately, batteries only contain a finite amount of power; when batter-
naturally occurring phenomena such as multipath effects and ies are exhausted, replacement can be a difficult task, espe-
interference can diminish the performance of narrow-band cially when sensors are in locations where human access is
wireless signals (Mittag, 2001). To enhance the reliability of limited.
the wireless communication channel, spread spectrum wire- In this section, academic and commercial wireless sensor
less signals are preferred. Spread spectrum encodes data on platforms explicitly proposed for use in structural moni-
a number of different frequencies within a frequency band. toring and SHM systems are chronologically summarized.
By effectively spreading the signal energy over a broad spec- Tables 1 and 2 provide a comprehensive summary of the per-
trum, the probability of interference on the band is greatly formance features of the academic prototypes summarized,
reduced (Bensky, 2004). A number of methods for modulating while Table 3 summarizes commercial platforms. It should
data in a spread spectrum fashion include frequency-hopping be noted that the summary is not intended to be an exhaus-
spread spectrum (FHSS) and direct-sequence spread spectrum tive listing; rather, it highlights the state-of-the-art in wire-
(DSSS). less sensing up to March 2005.
Strong consideration must be given to the communication
range of the wireless transceiver. For example, to monitor a 2.1. Academic Wireless Sensing Unit Prototypes
large-scale civil structure, communication ranges in excess
of 100 m might be necessary, while monitoring an aircraft Realizing the need to reduce the costs associated with wired
structure permits the use of shorter range radios. The range of structural monitoring systems, Straser and Kiremidjian (1998)
the wireless transceiver is directly correlated to the amount have proposed the design of a low-cost wireless modular mon-
of power the transceiver consumes. As the wireless signal itoring system (WiMMS) for civil structures. Using commer-
radiates from an antenna in open space, it loses power in pro- cial off-the-shelf (COTS) components, a low-cost wireless
portion to the wavelength of the radio band and inversely sensor approximately 12 × 21 × 10 cm3 is produced. To con-
proportional to the square of the distance from the transmit- trol the remote wireless sensing unit, the Motorola 68HC11
ter (Rappaport, 2002). A direct result of transmission power microprocessor is chosen for its large number of on-chip
reducing inversely proportional to the distance squared is hardware peripherals and the availability of high-level pro-
that hopping data across a number of short-range radios is gramming languages (e.g. C) for embedding software. The
more energy efficient than using a single radio capable of 68HC11 is mounted upon the New Micros prototyping board
transmitting to longer ranges (Zhao and Guibas, 2004). When NMIT-0022 and features an 8-bit counter, a 16-bit timer,
radio waves encounter boundaries such as walls and floors, one asynchronous RS-232 serial port, and a 64 kB address
the signal’s power is reduced. Referred to as path loss, the space for data and program storage. In order to store embed-
amount of power lost by the wireless signal is dependent ded firmware for local data processing, 32 kB of addi-
upon the material through which the signal must penetrate. tional RAM and 16 kB of additional ROM are included in
A number of researchers have undertaken empirical studies the design. To achieve reliable wireless communication, a
to quantify the propagation distances of wireless signals Proxim Proxlink MSU2 wireless modem operating on the
within structures when communicating on different frequency 902–928 MHz ISM band is used. Consuming 135 mA of cur-
bands (Seidel and Rappaport, 1992; Davidson and Hill, 1997). rent when communicating, the wireless modem is ordinarily
Pei et al. (2005) have also measured the range and amount of kept in sleep mode where it consumes minimal power (1 mA
data loss of different wireless sensors operating on the unli- of current). The maximum open space range of the wireless
censed ISM bands in various structural monitoring applica- radio has been determined to be approximately 300 m out-
tions. doors, with a maximum data rate of 19.2 kbps. To attain a
The last subsystem of a wireless sensor would be the actu- high degree of reliability in the wireless channel, the Prox-
ation interface. Actuation provides a wireless sensor with the link radio encodes data using a DSSS technique. Finally, to
capability to interact directly with the physical system in which convert analog signals to digital forms, an eight-channel, 16-
it is installed. Actuators and active sensors (e.g. piezoelectric bit, 240 Hz Harris H17188IP sigma-delta ADC is used. An
elements) can both be commanded by an actuation interface. interesting feature of this ADC is its fixed sampling rate
The core element of the actuation interface is the digital-to- (240 Hz). With built-in line noise reduction and support for
analog converter (DAC) which converts digital data gener- the Motorola serial peripheral interface (SPI), the Harris

Table 1. Summary of academic wireless sensing unit prototypes (1998–2003).

Straser Lynch et al. Lynch et al.
Kottapalli
and Bennett et (2001, Mitchell et (2003a, Aoki et al. Basheer et
et al.
Kiremidjian al. (1999) 2002a, al. (2002) 2004a, (2003) al. (2003)
(2003)
(1998) 2002b) 2004e)
DATA ACQUISITION SPECIFICATIONS

A/D Channels 8 4 1 5 1 Multiple
Sample Rate 240 Hz 100 kHz 20 MHz 20 MHz 100 kHz
A/D Resolution 16-bit 16-bit 16-bit 16-bit 8-bit 16-bit 10-bit
Digital Inputs 0 2 0 2
EMBEDDED COMPUTING SPECIFICATIONS

Processor Motorola Hitachi H8/ Atmel Cygnal Microchip Atmel Renesas ARM7TDMI
68HC11 329 AVR8515 8051 PIC16F73 AT90S8515 H8/4069F
AVR /
MPC555Po
werPC
Bus Size 8-bit 8-bit 8-bit 8-bit 8-bit 8-bit/32-bit 8-bit 32-bit
Clock Speed 2.1 MHz 4.9 Hz 4 MHz 20 MHz 4 MHz / 20 20 MHz
MHz
Program Memory 16 kB 32 kB 8 kB 2 kB 4 kB 8 kB / 26 kB 128 kB
Data Memory 32 kB 32 kB 128 kB 192 kB 512 kB / 448 2 MB
kB
WIRELESS CHANNEL SPECIFICATIONS

Radio Proxim Radiome- Proxim Ericsson BlueChip Proxim Realtek Phillips
ProxLink trix RangeLan2 Bluetooth RBF915 RangeLan2 RTL- Blueberry
8019AS Bluetooth
Frequency Band 900 MHz 418 MHz 2.4 GHz 2.4 GHz 900 MHz 2.4 GHz 2.4 GHz
Wireless Stand- IEEE IEEE
ard 802.15.1 802.15.1
Spread Spectrum Yes Yes Yes Yes Yes Yes
Outdoor Range 300 m 300 m 300 m 10 m 500 m 300 m 50 m 100 m
Enclosed Range 150 m 150 m 10 m 200 m 150 m 50 m
Data Rate 19.2 kbps 40 kbps 1.6 Mbps 10 kbps 1.6 Mbps
FINAL ASSEMBLED UNIT ATTRIBUTES

Dimensions 15x13 15D 10x10 5x3.8x 10x5 12x10x2 cm 30x6x8 cm 2.5x2.5
x10 cm x 30 cm x5 cm 1.2 cm x1.5 cm x2.5 cm
Power 120 mW 100 mW
Power Source Battery Battery Battery (9V) Battery Battery Battery (9V) Battery
(9V) (6V) (9V)
ADC is well suited for the wireless sensing unit design; modate the two strain gages, Wheatstone bridge and ampli-
however, it should be noted that no anti-aliasing filter is fication circuits are designed as part of the wireless sensor’s
present. Although the wireless sensor proposed does not sensing interface. At the core of the wireless sensor is a
emphasize power minimization in its design, the prototype Hitachi H8/329 8-bit microcontroller. To provide ample mem-
represents the first major step by the structural engineering ory for the storage of embedded software that operates the
community towards decentralized data processing and wire- sensor, 32 kB of external ROM is included in the computa-
less SHM. tional core design. To communicate asphalt response data in
Bennett et al. (1999) have proposed the design of a wire- real time to a data logger, a narrow-band 418 MHz Radi-
less sensing unit intended for embedment in flexible asphalt ometrix wireless radio is included in the design of the wire-
highway surfaces. To record measurement data from two less sensor. The Radiometrix radio is capable of data rates of
thermometers and two thin-film strain gages, a four-channel 40 kbps and can communicate to ranges as high as 300 m in
sensing interface is designed. While the specific ADC is not open space. The completed wireless sensor prototype is
mentioned, the resolution of the ADC is 16 bits. To accom- packaged in a water-tight PTFE cylinder with a 15 cm diam-

Table 2. Summary of academic wireless sensing unit prototypes (2003–2005).

Wang et al. Shinozuka Farrar et
Casciati et Mastro- Wang
(2003, 2004); (2003); Ou et al. Sazanov et al. al. (2005); Pei et al.
al. (2003b, leon et al. et al.
Gu et al. Chung et (2004) (2004) Allen (2005)
2004) (2004) (2005)
(2004) al. (2004) (2005)

A/D 8 8 5 4/2 6 6 4
Channels
Sample > 50 Hz 480 Hz 200 kHz 100 kHz 100/500
Rate Hz
A/D 12-bit 12-bit 16-bit 8-bit / 12-bit 16-bit 16-bit 10/12/
Resolution 10-bit 16-bit
Digital multiple 0 2 16 0
Inputs

Processor Analog Micro- Atmel AVR Texas Intel Atmel Motorolla
Devices chip PIC- ATMega 8L Instruments Pentium / AVR 68HC11
ADuC832 micro MSP430F1611 Motorola ATMega
128
Bus Size 8-bit 16-bit / 8-bit 16-bit 16-bit 8-bit 8-bit
8-bit
Clock 120/233 8 MHz
Speed MHz
Program 62 kB 8 kB 16 MB 256 MB 128 kB 32 kB
Memory
Data 2 kB 1 kB Compact 128 kB 32 kB
Memory Flash

Radio Aurel XTR- Linx BlueChip Chipcon Chipcon Motorola
Max- Max-
915 Technolo- RFB915B CC1000 CC2420 neuRFon
stream Stream
gies 9XCite Xstream
Frequency 914.5 MHz 916 MHz 900 MHz 2.4 Ghz 433 MHz 2.4 GHz 2.4 GHz 900 MHz 900 MHz/
Band 2.4 GHz
Wireless IEEE IEEE IEEE 802.15.4 IEEE
Standard 802.15.1 802.11b 802.15.4
Spread No No Yes Yes Yes Yes Yes Yes Yes
Spectrum (Software)
Outdoor 152 m 200- 250 m 75 m 9.1 m 300 m
Range 300 m
Enclosed 61 m 9.1 m 100 m
Range
Data Rate 100 kbps 33.6 kbps 19.2 kbps 76.8 kbps 250 kbps 230 kbps 38.4
kbps

Dimen- 8x8x2 cm 6x9x 10x6x
sions 3.1 cm 4 cm
Power 75 mW 6W
Power Battery Battery + Battery Battery Battery
Source Solar (7.5V) (9V)
eter and 30 cm height. For power, four AA alkaline batteries (2001, 2002a, 2002b) have proposed a wireless sensor proto-
offering a total voltage of 6 V are included. type that emphasizes the design of a powerful computa-
Recognizing the importance of decentralized data process- tional core. Setting the goal to minimize power consumption
ing in wireless structural monitoring systems, Lynch et al. throughout the entire sensing unit design, the 8-bit Atmel

Table 3. Summary of commercial wireless sensing unit prototypes.

UC Berkeley- UC Berkeley- UC Berkeley- UC Berkeley- Microstrain, Rockwell,
Intel iMote,
Crossbow WeC Crossbow Crossbow Crossbow Galbreath et Agre et al.
Kling (2003)
(1999) Rene (2000) MICA (2002) MICA2 (2003) al. (2003) (1999)

A/D Channels 8 8 8 8 8 4
Sample Rate 1 kHz 1 kHz 1 kHz 1 kHz 1.7 kHz 400 Hz
(one chan-
nel)
A/D Resolution 10-bit 10-bit 10-bit 10-bit 12-bit 20-bit
Digital Inputs

Processor Atmel Atmel Atmel Atmel Zeevo MicroChip Intel Stron-
AT90LS8535 Atmega163L ATmega103L ATmega128L ARM7TDMI PIC16F877 gARM 1100
Bus Size 8-bit 8-bit 8-bit 8-bit 32-bit 8-bit 32-bit
Clock Speed 4 MHz 4 MHz 4 MHz 7.383 MHz 12 MHz 133 MHz
Program 8 kB 16 kB 128 kB 128 kB 64 kB 1 MB
Memory
Data Memory 32 kB 32 kB 512 kB 512 kB 512 kB 2 MB 128 kB

Radio TR1000 TR1000 TR1000 Chipcon Wireless BT RF Mono- Conexant
CC1000 Zeevo lithics DR- RDSSS9M
3000-1
Frequency 868 / 916 MHz 868 / 916 MHz 868 / 916 315, 433, or 2.4 GHz 916.5 MHz 916 MHz
Band MHz 868 / 916MHz
Wireless IEEE
Standard 802.15.1
Spread No No No Yes (Soft- Yes Yes
Spectrum ware)
Outdoor
Range
Enclosed 100 m
Range
Data Rate 10 kbps 10 kbps 40 kbps 38.4 kbps 600 kbps 75 kbps 100 kbps

Dimensions 2.5 x 2.5 7.3 x 7.3
x 1.3 cm x 8.9 cm
Power 575 mAh 2850 mAh 2850 mAh 1000 mAh
Power Coin Cell Battery (3V) Battery (3V) Coin Cell Battery Battery Battery
Source (3.6V) (two 9V)
AVR AT90S8515 enhanced RISC (reduced instruction set processing and data storage tasks. A low-noise single-chan-
computer) microcontroller is selected. Capable of eight million nel Texas Instrument 16-bit ADC is used to translate analog
instructions per second (MIPS), the microcontroller has high signals to a digital format for processing. However, after
computational throughput without consuming large amounts mounting the ADC within the tailored designed printed cir-
of power. The AVR microcontroller also has a wide variety of cuit board, the authors note that its resolution is reduced to
on-chip services such as internal oscillators, serial communi- 14 bits due to circuit noise. The high-speed parallel CMOS
cation transceivers, timers, pulse width modulators (PWMs), architecture of the ADC allows the sampling rate to reach
and four 8-bit general purpose input/output ports. The micro- 100 kHz. Similar to the unit proposed by Straser and Kiremid-
controller is able to take full advantage of its 8 kB of program- jian (1998), the Proxim ProxLink MSU2 wireless modem
mable flash memory, 512 bytes of SRAM (static random operating on the 902–928 MHz ISM radio band is integrated
access memory), and 512 bytes of electronically erasable with the wireless sensor. In comparison with the wireless
programmable read-only memory (EEPROM) to perform local sensing unit design proposed by Straser and Kiremidjian

Figure 2. Wireless network topologies for wireless sensor networks: (a) star; (b) peer-to-peer; (c) two-tier network topologies.
(1998), the wireless sensor described by Lynch et al. (2001) ers, an Ericsson Bluetooth wireless transceiver, operating on
is compact (10 × 10 × 5 cm3 in size) and relatively low power the 2.4 GHz radio band, is integrated. The communication
(250 mW when not transmitting data and 900 mW when range of the radio is roughly 10 m line of sight. Provided the
using the wireless modem). short range of the radio, multihopping of data between wire-
Mitchell et al. (2002) have proposed a two-tier SHM archi- less sensors is proposed. The Bluetooth radio consumes 35
tecture using wireless sensors (as shown in Figure 2c). Based mW of electrical power.
upon three generations of hardware and software designs, After data are collected by the wireless sensors, data can
their current wireless monitoring system emphasizes the par- then be transferred wirelessly to wireless data servers (cluster
titioning of the monitoring system functionality between wire- nodes). Each cluster node has both a short-range radio (for
less sensors and wireless data servers (called wireless cluster communication with wireless sensors in its cluster) as well
nodes). In their system, a compact (footprint size of 4 × 7.5 cm2) as a long-range radio (for communication with other remote
wireless sensor using a powerful Cygnal 8051F006 micro- cluster nodes). The central cluster server is designed to both
controller is proposed for data collection. Capable of 25 store and process the vast amounts of data collected from the
MIPS, the microcontroller only consumes 50 mW of battery cluster’s wireless sensors. The cluster node is designed using
power and provides 2 kB of RAM for data storage. For com- a single board computer (SBC) running the Microsoft Win-
munication between wireless sensors and wireless data serv- dows OS. MATLAB is installed in the node for processing

measurement data for signs of structural damage. A key ele- While Mitchell et al. (2002) and Kottapalli et al. (2003) have
ment of this two-tiered wireless SHM system architecture proposed attainment of an overall low-power wireless SHM
proposed by Mitchell et al. (2002) is its seamless interface to system by partitioning functionality upon multiple network
the Internet. Using the World Wide Web (WWW), structural tiers, Lynch et al. (2003a, 2004a, 2004e) focus upon the design
management professionals have the capability to remotely of a low-power but computationally rich wireless sensing
access structural response data, as well as analysis results unit. In their design, each component of the wireless sensor
performed by the monitoring system (Mitchell et al., 2001). is selected such that minimal power is required. Often, micro-
The wireless data cluster nodes are equipped with cellular controllers with high computational throughput consume
modems for long-range communication (on the order of miles) more energy from portable power supplies compared to sim-
to a single web server that is accessible from the WWW. pler microcontrollers. To address this limitation, Lynch et al.
Kottapalli et al. (2003) have presented a wireless sensor (2003a, 2004a, 2004e) have proposed a dual-processor com-
network architecture that is intended to overcome the major putational core design. Based on their earlier wireless sensing
challenges associated with time synchronization and limited unit design (Lynch et al., 2001), a low-power 8-bit Atmel
power availability in wireless SHM systems powered by bat- AVR AT90S8515 microcontroller is utilized for overall unit
teries. Similar to the two-tiered wireless SHM system pro- operation and real-time data acquisition. When data are
posed by Mitchell et al. (2002), Kottapalli et al. (2003) have ready for local processing, the unit turns on the second micro-
proposed a two-tiered wireless sensor network architecture controller, which is the 32-bit Motorola MPC555 PowerPC.
that entails the design of wireless sensing units and local site This microcontroller contains 448 kB of ROM, 26 kB of
masters. The role of the sensing unit is to simply collect meas- RAM, along with a floating-point arithmetic and logic unit
urement data and to wirelessly transmit the data to the desig- (ALU). At a clock rate of 20 MHz, intensive data processing
nated site master. Wireless sensing units communicate with algorithms, such as embedded damage detection routines
their corresponding site master using the BlueChip EVK915 stored in ROM, can be executed. When the two microcon-
915 MHz radio transceiver. Using Manchester encoding of trollers are turned on, the AT90S8515 consumes 40 mW of
the data for wireless transmission, the effective data rate of power and the MPC555 (at 20 MHz) consumes 330 mW. In
the radio is 10 kbps. To achieve wireless reliability, each sens- sleep mode, the two microcontrollers both consume 12 mW,
ing unit communicates directly with its site master by using respectively. During data collection, measurement data can be
FHSS encoding. The motivation for selecting this particular stored either in the internal RAM of the microcontrollers or
radio for inclusion with the wireless sensors is that it is very in external memory (512 kB Hitachi HM628512B SRAM). For
low power, consuming 36 mW when receiving and 150 mW data collection, a low-power single-channel ADC is included.
when transmitting. The embedded microcontroller of the wire- The Texas Instruments ADS7821 16-bit ADC has a maximum
less sensing unit prototype is an 8-bit Atmel AVR microcon- sample rate of 100 kHz and draws 80 mW of power. Included
troller. For data collection, a 16-bit ADC is also included in in the sensing interface are two additional channels for external
the unit design. The total power consumption for each indi- sensors with digital outputs. For wireless communications,
vidual sensing unit is, on average, 100 mW. Using alkaline the 2.4 GHz Proxim RangeLAN2 radio modem is selected.
AA batteries, this low power demand results in approxi- To supply power to the wireless sensor, a high-energy-density
mately 18 months of battery life before the units deplete the Li/FeS2 7.5 V battery pack is chosen because the estimated
portable energy supply. duty cycle usage life of the battery in the field is estimated to
A network of local site masters forms the upper tier of the be of the order of one year (Lynch, 2002).
sensor network. Their role is to aggregate the data originat- Aoki et al. (2003) have proposed a novel wireless sensing
ing from the low-tier wireless sensing units. Each local site unit prototype, which they call the Remote Intelligent Mon-
master is at the center of a star network topology where wire- itoring System (RIMS). Designed for the purpose of bridge
less sensing units communicate only with their designated and infrastructure SHM, each hardware component included
site master. As such, the hardware design of the local site mas- in their design is carefully chosen to reduce the cost and size
ter must have ample storage for measurement data and must of the prototype while achieving adequate performance stand-
be cable of high data rate communication. To accomplish these ards. At the core of the wireless sensor design is the Renesas
goals, the local site masters are equipped with two radios. H8/4069 microcontroller. The microcontroller has a high-speed
The first radio, the BlueChip EVK915, allows the master to processing core operating at 20 MHz and an internal 10-bit
communicate with wireless sensing units. A second radio is ADC. Tailored for monitoring dynamic structures, the wireless
included, the Proxim RangeLAN2, to facilitate communica- sensing unit design includes a dedicated three-axis microelec-
tion between local site masters. The RangeLAN2 operates tromechanical systems (MEMS) piezoresistive accelerome-
on the 2.4 GHz ISM band radio, has a data rate of 1.6 Mbps, ter (Microstone MA3-04). To enhance the storage capabilities
and can achieve long communication ranges (300 m in open of the wireless sensor design, an additional 2 MB externally
range and 150 m when shielded by heavy construction). The interfaced dynamic random access memory (DRAM) is
radios are selected to operate on two separate frequency included. The DRAM is employed for storage of time-his-
bands in order to minimize interference between site master tory data, as well as for performing local computations to
to site master and site master to wireless sensing unit connec- minimize the amount of data that need to be transmitted
tions. The RangeLAN2 consumes a large amount of power wirelessly. While no details are provided, the RIMS wireless
(800 mW when transmitting or receiving), but it is assumed sensor is capable of wireless communication with a remote
that the local site masters would be powered by outlet sources. data repository. The core component of the wireless commu-
At the core of the local site master is an 8-bit Microchip PIC nication link is the Realtek RTL-8019AS ethernet controller.
microcontroller that is employed for data storage and local Embedded within each wireless sensor is an HTTP manager
data processing. servelet. The embedded HTTP manager allows remote users

to interact with sensors and perform tasks remotely by exe- erned by the desire to find a processor that is low-power
cuting suitable servelet functions through the Internet. For without sacrificing computational throughput; the ARM proc-
instance, users can create documents for unit initialization, essor is capable of 100 MIPS. For wireless communication,
to set operational parameters, and to request the display of the Phillips Blueberry 2.4 GHz Bluetooth wireless radio is
time-history data, all from a web browser. A more recent selected for integration. The Bluetooth radio is both low-power
version of the RIMS wireless sensor has been proposed with and short-range but employs fast FHSS encoding, thereby
an improved computational core; the Renesas H8 microcon- enhancing its reliability in the presence of other radios oper-
troller is replaced by the Rabbit 3000 microcontroller offer- ating on the same frequency.
ing 12-bit analog-to-digital conversion resolution. Wang et al. (2003a) have proposed the design of a wire-
Casciati et al. (2003b) present the design of a wireless sens- less sensor specifically intended to report displacement and
ing unit intended for SHM of historic landmarks in which strain readings from a polyvinylidene fluoride (PVDF) thin-
wired monitoring systems would be too obtrusive. Again, a film sensor. Their wireless sensor is similar to that proposed
two-tier approach to the design of the wireless structural mon- by Casciati et al. (2003b) in that the wireless sensor design is
itoring system is proposed. The authors detail their design of based upon an Analog Devices ADuC832 microsystem. The
a low-power wireless sensing unit which is situated on the ADuC832 combines a powerful 8051 microcontroller with a
lowest tier of the two-tier monitoring system architecture. complete data acquisition system on a single integrated cir-
Intended to collect structural response measurements from cuit chip. To collect data from interfaced sensors (in this case,
accelerometers, the design of the wireless sensing units is a PVDF sensor), the ADuC832 provides eight sensing chan-
based upon the Analog Devices ADuC812 microsystem. nels serviced by a 12-bit ADC. Also included in the microsys-
The ADuC812 is a complete data acquisition system-on-a-chip tem are two separate 12-bit DACs. Once data are collected,
solution that includes an 8051 microcontroller core, 8 kB of the internal 8-bit 8052 microcontroller is responsible for
flash ROM, an eight-channel 12-bit ADC, and a two-channel management of the sensor data. To facilitate the storage and
12-bit DAC. The wireless communication subsystem of the processing of data, the ADuC832 microsystem has 62 kB of
wireless sensing unit is based upon the single-channel AUREL ROM reserved for the storage of executable programs and
XTR-915 RF transceiver operating at 914.5 MHz with a max- 256 bytes of SRAM for data storage. Integrated with the
imum data transmission rate of 100 kbps. Selection of this wireless sensor is a single-channel half-duplex wireless radio
transceiver is based upon its high transmission rate and low operating on the 916 MHz frequency band with a range of
power consumption (160 mW maximum but typically only 150 m and a data rate of 33.6 kbps (Gu et al., 2004).
120 mW). An important component of the wireless sensing unit Extending upon the design of the wireless sensing unit
design is the inclusion of a third-order low-pass anti-aliasing proposed by Kottapalli et al. (2003), Mastroleon et al. (2004)
filter whose pass band is adjustable through the ADuC812 have attained greater power efficiency by upgrading many of
microcontroller. the unit’s original hardware components. In particular, the
Upon the second tier of the hybrid wireless monitoring sys- computational core of their unit is designed around a Micro-
tem architecture proposed by Casciati et al. (2003b, 2004), chip PICmicro microcontroller. The PICmicro is selected for
are wireless computational units where data streams origi- its low power consumption and high computational perform-
nating from the lower tier wireless sensing units are aggre- ance. The microcontroller is capable of achieving real-time
gated and locally processed. Since the design of the wireless data processing and time synchronization by using multilevel
computational unit is not based upon the collection of meas- priority interrupts and phase-locked loop (PLL) synchroni-
urement data from interfaced sensors, the computational zation units. Moreover, the PICmicro dynamically switches
units can be placed anywhere, thereby allowing design limi- between six power management modes and possesses a fail-
tations to be less stringent on weight, dimensions, and power safe clock monitor to achieve ultralow power consumption.
consumption. To establish communication with the wireless In addition, the availability of self-programming flash mem-
sensing units, the wireless computational unit includes the ory allows embedded software to be upgraded in the field
AUREL XTR-915 RF transceiver. For inter-wireless com- through the wireless channel. Identical to the unit proposed
putation unit communication, a second wireless transceiver by Kottapalli et al. (2003), the wireless sensor employs the
operating on the 2.4 GHz wireless spectrum is included. The Bluechip RFB915B RF transceiver for wireless communica-
MaxStream 2.4 GHz XStream wireless radio is selected tion. For the sensing interface, the 18-bit Maxim MAX1402
because of the reliability provided by its use of FHSS tech- ADC is chosen. The MAX1402 is capable of sample rates as
niques. The XStream comsumes 750 mW when transmitting high as 480 Hz and can simultaneously sample sensor data
and 250 mW when receiving. In addition, the radio can attain from five channels. Acknowledging the strong dependence
a communication range of over 180 m. upon the ambient temperature of the structure and the accuracy
Basheer et al. (2003) have proposed the design of a wireless of current damage detection methods, the Maxim DS18S20
sensor whose hardware design has been optimized for collab- digital thermometer is also implemented within the wireless
orative data processing (such as damage detection) between sensing unit design.
wireless sensors. The wireless sensors proposed form build- Drawing from previous experiences with commercial wire-
ing blocks of a self-organizing sensor network called the less sensor platforms, Ou et al. (2004) have described the
Redundant Link Network (RLN). Basheer et al. (2003) call design of a new low-power academic wireless sensor proto-
their wireless sensor ISC-iBlue. The design of ISC-iBlue is type for structural monitoring. At the core of their sensor is
divided into four main components: communication, process- the low-power Atmel AVR ATmega8L microcontroller. This
ing, sensing, and power subsystems. The processing core of 8-bit microcontroller has 8 kB of flash memory for storing
the wireless sensor is designed around the ARM7TDMI embedded programs and 1 kB of SRAM for storing meas-
microprocessor. Selection of the ARM processor is gov- urement data. In total, eight sensing channels are provided

for the interface of sensors. Six of the channels support the Instead of a low-power microcontroller, the wireless sensor
conversion from analog sensor outputs into digital formats is designed using a standard PC-104 SBC with a 133 MHz
with resolutions of 8 and 10 bits. The last two channels are Pentium processor, 256 MB of RAM, and a 512 MB Compact
for measuring the output of digital sensors such as the Ana- Flash (CF) card serving as a hard drive. Other features included
log Devices ADXL202E MEMS accelerometer. To provide on the SBC are serial, Ethernet, and USB interfaces for com-
wireless communication between wireless sensors, Ou et al. munication with peripherals. To provide the wireless sensor
(2004) integrate the Chipcon CC1000 wireless transceiver. with the capability to interface with sensors, a separate sens-
This radio operates on the 433 MHz radio band and can coming board is designed. The sensing board houses a Motorola
municate at a data rate of 76.8 kbps. DSP56858 digital signal processor (DSP) that is used to sample
Shinozuka (2003) and Chung et al. (2004a) have described data from six single-channel Maxim ADCs. The maximum
the design of a wireless sensor called DuraNode. Different rate for simultaneously sampling the six ADCs is 200 Hz. After
from the previous wireless sensors that had sensor transparent data are collected by the sensing board, they can be forwarded
interfaces, the wireless sensor proposed is designed around to the SBC through the serial port. Finally, a Motorola neu-
two types of MEMS-based accelerometers: Analog Devices RFon transmission board utilizing the IEEE802.15.4 wireless
ADXL202 and Silicon Design SD1221. While the specific sensor communication standard is selected. The IEEE802.15.4
hardware components are not described, the wireless sensor transceiver operates on the 2.4 GHz ISM radio band with a
employs a 2.4 GHz 802.11b wireless network interface card data rate of 230 kbps and an indoor range of 10 m. When fully
as its wireless radio and is powered on lithium–polymer thin- packaged, the total unit volume is 1750 cm3 and consumes 6 W
film battery technology. Recognizing the limitations of battery of power. The wireless sensor platform proposed by Allen
power, they have also integrated a solar panel with DuraN- (2004) and Farrar et al. (2005) is called Husky.
ode to recharge the lithium–polymer battery. The completed Using the latest commercially available embedded system
DuraNode unit has dimensions of 6 × 9 × 3.1 cm3. components, Wang et al. (2005) have proposed a wireless sens-
In recent years, a new wireless communication standard, ing unit with multitasking capabilities. In particular, a low-
IEEE802.15.4, has been developed explicitly for wireless sen- power wireless sensor that can sample measurement data
sor networks (Institute of Electrical and Electronics Engineers, simultaneous to wirelessly transmitting data with other wire-
2003). This wireless standard is intended for use in energy- less devices is proposed. For the sensing interface, a four-
constrained wireless sensor networks because of its extreme channel Texas Instrument ADS8341 16-bit ADC is selected
power efficiency. Another important aspect of IEEE802.15.4 to convert analog sensor signals to digital formats for use by
is that it offers a standardized wireless interface for wireless the microcontroller. This ADC is selected for its low power
sensor networks, thereby ensuring compatibility between consumption and high sample rates (100 kHz maximum).
wireless sensor platforms with different designs and func- For the computational core, the low-power 8-bit Atmel
tionalities. Sazonov et al. (2004) have proposed the design of ATmega128 AVR microcontroller is selected. The microcon-
a low-power wireless sensor around the IEEE802.15.4 wire- troller has 128 kB of ROM, which is sufficient for storing dam-
less standard. For wireless communication, their unit employs age detection software. In addition to ROM, 4 kB of SRAM
the Chipcon CC2420 wireless transceiver. IEEE802.15.4- is integrated with the microcontroller; however, this amount
compliant, the radio operates on the 2.4 GHz radio spectrum of SRAM is insufficient to store all the collected data. An addi-
with a data rate of 250 kbps. The radio has a range of 10–75 tional 128 kB of SRAM (Cypress CY62128B) is interfaced
m, yet it only consumes 60 mW when receiving and 52 mW with the microcontroller for the storage of measurement
when transmitting. To design the remainder of the wireless data. The most attractive feature of the wireless sensing unit
sensor hardware to be as low power as possible, the 16-bit design is its wireless radio. With the wireless radio identified
Texas Instruments MSP430 microcontroller is selected for as one of the most power hungry elements of a wireless sen-
the computational core. The MSP430 provides the wire- sor design, Wang et al. (2005) have proposed the integration
less sensing unit with a six-channel 12-bit ADC and a two- of the MaxStream 9XCite wireless modem. This radio oper-
channel 12-bit DAC. With 2 MB of non-volatile EEPROM, ates on the 900 MHz radio band and is capable of data rates
the MSP430 is capable of storing sophisticated data interro- as high as 38.4 kbps. The communication range of the radio is
gation algorithms. When fully assembled, the proposed low- 300 m line-of-sight yet the radio only consumes 250 mW
power wireless sensor is intended to serve as the building when transmitting, 150 mW when receiving, and less than
block of a wireless intelligent sensor and actuator network 5 mW when idle. With efforts to further reduce the size of
(WISAN). the wireless sensor, the electrical circuit is printed on a com-
The previously described wireless sensor designs seek to pact two-layer circuit board (9.7 × 6 cm2). When fully assem-
minimize power consumption simultaneous to maximizing bled, the wireless sensor is 10 × 6.5 × 4 cm3 and is powered
functionality. Allen (2004) and Farrar et al. (2005) have pro- by five AA batteries.
posed a different design strategy; the emphasis of their wire- Undertaking a much broader study, Pei et al. (2005) have
less sensor design is on providing ample computational power rigorously evaluated the impact different hardware components
to perform a broad array of damage detection algorithms have on the quality of data collected by wireless sensors. To
within a wireless SHM system. In close collaboration with facilitate such an evaluation, a highly modular wireless sen-
Motorola Labs, Farrar et al. (2005) have described the design sor architecture, in which different hardware components
of a wireless sensor designed to have seamless interaction can be readily interchanged, is proposed. Some of the hard-
with DIAMOND II, an existing damage detection package ware components that can be interchanged include the wireless
written in Java. As such, the overall design of the wireless sensor’s ADC, interfaced sensors, and wireless transceivers.
sensor is based on the powerful computational core needed The common element to all of the hardware permutations is
to execute DIAMOND II-based damage detection routines. the computational core. The wireless sensor architecture pro-

posed is based upon the Motorola 68HC11 microcontroller, sented in Figure 4. The commercial platforms to be described
which is a popular microcontroller with 32 kB of SRAM and in this section are summarized in Table 3.
32 kB of ROM. The first hardware element evaluated is the The Berkeley Mote platform has been under development
ADC. In total, three different ADCs with varying resolutions since the late 1990s with the first prototype, called WeC,
(10-, 12-, and 16-bits) are interfaced with the wireless sensor produced in 1999 and commercialized as the Rene Mote by
design. To facilitate the change of the ADC, all three are Crossbow. The WeC hardware is based upon the 8-bit Atmel
selected to have the same interface with the microcontroller. AT90LS8535 AVR microcontroller for its computational core.
Included with each ADC is a four-pole Butterworth low-pass The internal eight-channel, 10-bit ADC of the microcontrol-
anti-aliasing filter (LPF) with a cutoff frequency of 35 Hz. ler serves as the primary sensing interface capable of sampling
Also of interest in their study is the impact of the wireless rates as high as 1 kHz. With only 8 kB of ROM and 512 Bytes
transceiver carrier frequency on both the range and the relia- of RAM included in the microcontroller, an additional 32 kB
bility of the wireless communication channel. The MaxStream of external RAM is included with the WeC platform. To estab-
XStream wireless transceiver is selected for integration within lish wireless communication with other wireless sensors, the
the modular wireless sensor architecture. The XStream is a RF Monolithics TR1000 wireless radio is integrated. This sin-
FHSS radio that has impressive range. The authors evaluate gle-channel TR1000 transceiver operates on the 916 MHz
two variations of the XStream radio: one operating at 900 MHz frequency, employs amplitude modulation (AM), and commu-
and another at 2.4 GHz. When operated at 900 MHz, the radio nicates with a data rate of 10 kbps (Maurer, 2003). Hill and
is capable of communication ranges of up to 450 m, while at Culler (2002) report the motivation for selecting the TR1000
2.4 GHz, its range is 180 m. As part of the study, the ranges is due to it consuming only 15 mW of battery energy with a
of the radios, as well as the number of data packets lost, are maximum communication range of 60 m. In 2001, the WeC
quantified when the sensors are installed at a variety of loca- wireless sensor was then modified to produce the Rene2
tions in typical structural environments. platform. The Rene2 Mote has an identical design to the WeC
Figure 3 presents many of the academic and commer- except that the original microcontroller is replaced with the
cial wireless sensors described in this summary review. The Atmel ATmega163L (Maurer, 2003). The ATmega163L has
academic prototypes presented include the wireless sensor larger internal memory banks including 16 kB of ROM and
prototypes proposed by Straser and Kiremidjian (1998), 1 kB of RAM.
Lynch (2002), Aoki et al. (2003), Allen (2004), and Wang Tanner et al. (2002, 2003) have presented the adoption of
et al. (2005). the Crossbow Rene2 Mote in a SHM system. During this study,
the authors report their experience of interfacing two types
2.2. Commercial Wireless Sensor Platforms of MEMS accelerometers with the Mote: the Analog Devices
ADXL202 and Silicon Devices SD-1221. While interfac-
A number of commercial wireless sensor platforms have ing the accelerometers to the microcontroller’s 10-bit ADC,
emerged in recent years that are well suited for use in SHM it is discovered that two sensing channels cannot be sam-
applications. The advantages associated with employing a pled simultaneously, resulting in a relative offset of 30 µs
commercial wireless sensor system include immediate out- between samples. This offset negatively impacts the accu-
of-the-box operation, availability of technical support from racy of embedded software used to calculate cross-correlation
the platform manufacturer, and low unit costs. For this reason, coefficients for sensor signals with high-frequency content.
many academic and industrial research teams have begun to The small amount of on-board RAM does not permit large
explore these generic wireless sensors for use within SHM buffers of sensor data to be stored. As a result, only on-the-
systems. In particular, the structural engineering community fly type embedded data interrogation algorithms have been
has focused their attention on the Mote wireless sensor platform successfully embedded in the Mote’s computational core for
initially developed at the University of California-Berkeley local data processing. A useful feature of the Mote is its three-
and subsequently commercialized by Crossbow (http://www. color light emitting diode (LED) display. The authors report
xbow.com/) (Zhao and Guibas, 2004). A major reason for the the use of the three-color LED as an indicator of the degree
Motes’ popularity is that it is an open source wireless sensor of calculated damage based on embedded damage detection
platform with both its hardware and software (TinyOS) design algorithms: red corresponds to severe damage, yellow corre-
available to the public. Since their introduction, Motes have sponds to the onset of damage, and green corresponds to the
been deployed in a number of large-scale monitoring appli- structure being undamaged.
cations. For example, over 150 Motes have been deployed to Glaser (2004) has evaluated the suitability of the hardware
monitor the weather and nesting conditions of birds on Great elements of the Crossbow Rene Mote during monitoring
Duck Island, Maine (Kumagai, 2004). Recently, Intel has pro- studies performed in the laboratory and field. After using the
duced its own version of the Mote called iMote (Kling, 2003). Rene Mote in their studies, some issues were identified with its
Well over 70 iMotes have been deployed by Intel to monitor hardware design. In particular, problems were reported with
the performance and health of pumps and motors in one of their the reliability of the single-channel RF Monolithics TR1000
microchip factories (Culler and Mulder, 2004). A number of wireless radio. During testing, the radios experience signifi-
other commercial wireless sensor platforms have been used cant communication interference, resulting in the loss of
for structural monitoring in addition to the Motes, including sensor data wirelessly communicated. The reliability of the
platforms from Ember (http://www.ember.com/), Microstrain radio is further reduced in the presence of other electronic
(http://www.microstrain.com/), and Sensametrics (http://www. equipment including cameras, cell phones, and radios. Short
sensametrics.com/). In contrast to the Motes, these wireless of these limitations, the conclusion of the study is that the
sensor platforms are proprietary and not open source. The concept of affordable wireless monitoring systems is suc-
Crossbow MICA2 and Intel iMote wireless sensors are pre- cessfully established.

Figure 3. Academic wireless sensor prototypes: (a) WiMMS wireless sensor (Straser and Kiremidjian, 1998); (b) dual-core
prototype by Lynch (2002); (c) RIMS wireless sensor based on Aoki et al. (2003); (d) Husky wireless sensor (Allen, 2004; cour-
tesy of Motorola Labs); (e) wireless sensor prototype by Wang et al. (2005).

Figure 4. Commercial wireless sensors: (a) Crossbow MICA2 Mote; (b) Intel iMote.
To provide more program and data storage and to improve accelerometer. In the frequency domain, the SD1221 accel-
the flexibility of the wireless communication channel, erometer is consistent with the PCB393B04 accelerometer,
Crossbow released the MICA Mote wireless sensor in early especially below 1.5 Hz. Based on extensive experience using
2002 as the successor to the Rene2. The computational core the MICA and MICA2 platforms, Spencer (2003) has iden-
of the MICA is based on the 8-bit Atmel ATmega103L micro- tified critical hardware issues that must be addressed before
controller (Maurer 2003). The ATmega103L is selected for the the MICA Motes can be used for SHM. In order to achieve
MICA core because of its considerable internal flash ROM sufficient measurement fidelity when using wireless sensors,
(128 kB) and RAM (4 kB) banks that facilitate the storage of the 10-bit ADC resolution must be improved. Also, time syn-
an embedded OS called TinyOS. Again, the internal eight- chronization across a large number of MICA Motes has been
channel 10-bit ADC of the microcontroller is utilized as the found to be challenging with synchronization errors of 7 ms
primary sensing interface for the MICA Mote. This ADC is encountered.
capable of sample rates up to 1 kHz. To provide additional In 2003, the MICA was modified to improve the reliabil-
memory for the microcontroller, 512 kB of non-volatile ity of the communication channel. With the original TR1000
memory is included off-chip in the MICA hardware design. single-channel radio susceptible to interference and data loss,
Similar to the WeC and Rene platforms, the MICA utilizes the MICA2 was introduced with a new radio offering greater
the single-channel amplitude modulation TR1000 wireless reliability. The Chipcon CC1000 wireless transceiver operates
transceiver. To conserve power for long-term field deploy- on the 900 MHz radio band and is a frequency modulation
ment, the MICA Motes utilize three different power modes: (FM) radio with excellent noise immunity. The carrier fre-
idle, power down, and power save. In total, the sensing unit quency of the CC1000 can be changed in software, allowing
can operate for approximately 30 h on two AA batteries. FHSS encoding techniques to be employed with the radio.
Ruiz-Sandoval et al. (2003) have reported their experiences The data rate of the CC1000 is reported as 38.4 kbps (Maurer,
using the MICA Mote wireless sensing platform for structural 2003). Like the radio, the ATmega103L microcontroller
monitoring. Their study utilizes the Crossbow MTS310CA is replaced with the Atmel ATmega128L. The ATmega128L
sensor board, which includes light, temperature, acoustic, and has the same amount of on-chip memory (128 kB ROM and
magnetic sensors along with an Analog Devices ADXL202E 4 kB RAM). Recently, the MICA2 has been upgraded with a
accelerometer. The MTS310CA sensor board plugs directly 2.4 GHz IEEE802.15.4 compliant wireless transceiver and is
to a multipin header situated on the MICA printed circuit called the MICAz (Crossbow, 2004). Finally, the most signif-
board. The performance of the ADXL202E accelerometer in icant change in the new MICA2 and MICAz designs is the
tracking the motion of a shaking table is compared to that of size reduction of the processor boards. For the MICA2 and
a PCB393B04 accelerometer attached to a tethered labora- MICAz, the total unit size is approximately 6 × 3 × 1 cm3.
tory data acquisition system. While the time histories pro- A number of researchers adopt the improved MICA2 Mote
vided by both accelerometers look identical, transformation in their research. Kurata et al. (2003a, 2003b) have reported
to the frequency domain reveals an excessive noise floor of on their use of the MICA2 to monitor the response of a lab-
the ADXL202E, hampering the accuracy of the sensor for oratory structure excited by a shaking table. The Tadeo sensor
signals below 1.5 Hz. To address these limitations, Ruiz- board initially proposed by Ruiz-Sandoval (2004) is inter-
Sandoval (2004) has proposed a new sensor board to replace faced to the MICA2 to measure the acceleration response of
the MTS310CA. Called the Tadeo sensor board, the board is the structure. While the MICA2 has an improved radio with
designed with the low-noise Silicon Devices SD1221 MEMS frequency hopping spread spectrum (FHSS) encoding, some

data loss is still experienced during testing. Ou and Li (2003) Motes. Coupled within the microcontroller is 64 kB of RAM
report similar results having used MICA2 Motes on various intended for data storage and 512 kB of ROM for running
laboratory structures. the embedded OS, TinyOS. On the wireless communication
Since the MICA2 Mote is unable to measure structural end, the 2.4 GHz Zeevo Bluetooth radio is integrated with
strain, Nagayama et al. (2004) implement a new integrated the ARM7TDMI microcontroller on a single integrated circuit
strain sensor board for the MICA2 Mote that accommodates chip. Selection of Bluetooth for wireless communication
strain gages. To be useful for structural monitoring applica- between iMotes is motivated by its high data rate (720 kbps)
tions, a sensor board capable of measuring strains spanning and high reliability (FHSS). Moreover, the Bluetooth media
from 1 to 2000 microstrains is designed and validated. At the access control (MAC) protocol allows the iMotes wireless
center of the sensor board is a standard Wheatstone bridge sensor network to be both scalable and reliable. The iMote is
circuit tailored for high resistance strain gages. The decision very compact with dimensions of 3.5 × 3.5 × 2.5 cm3 and is
to design the sensor board for a 4.5 kΩ strain gage is to limit powered by two Panasonic Lithium CR2 3V batteries. Spen-
the power consumed from the MICA2 batteries during oper- cer et al. (2004) have reported on the availability of the Intel
ation of the strain gage circuit. To ensure low levels of strain iMote platform which will potentially serve as a powerful
are measurable by the strain sensor board, a four-pole But- tool for future wireless SHM systems.
terworth low-pass filter with a high signal-to-noise ratio is Aside from the open-source efforts by researchers at the
designed to remove high-frequency noise. The output of University of California-Berkeley, Crossbow, and Intel, other
the Wheatstone bridge is amplified using an Analog Devices commercially available wireless sensor platforms have been
AD623 low-noise amplifier. The role of the amplifier is to adopted for SHM. For example, researchers at the Rockwell
overcome the low 10-bit resolution of the MICA2’s ADC. Science Center propose the design of a wireless sensing unit
Pakzad and Fenves (2004) describe a study where a novel designed for military applications which could potentially
prototype accelerometer sensor board is integrated with a include structural monitoring. The defining feature of the wire-
MICA2. With the standard Mote sensor board (MTS310CA) less sensor platform proposed by Agre et al. (1999) is its abil-
poorly suited for structural monitoring, the sensor board pro- ity to self-organize when deployed in the field. The wireless
posed by Pakzad and Fenves (2004) is intended for use in sensor prototype, called AWAIRS, adopts the powerful 32-
SHM applications. Upon the sensor board are four acceler- bit Intel StrongARM 1110 microcontroller for its computa-
ometer channels: two orthogonal channels are provided by tional core. This microcontroller includes 128 kB of SRAM
a single Analog Devices ADXL202 MEMS accelerometer and over 1 MB in flash ROM for embedded software storage.
while two Silicon Design SD1221 single-axis accelerometers The typical power consumption of the StrongARM 1110 is
are oriented parallel to the two-axes of the ADXL202. The approximately 200 mW; however, when placed in sleep mode,
Silicon Design SD1221 accelerometers have noise floors of the microcontroller only consumes 0.8 mW. To collect data
30 µg which allow them to measure small amplitude struc- from a variety of sensors, including geophones, acoustic sen-
tural vibrations. Static and dynamic laboratory testing are sors, magnetometers, and accelerometers, a 20-bit Analog
performed using the sensor board in order to assess its noise Devices AD7714 ADC is adopted. Using a standard serial
floor and frequency performance. First, the accelerometers peripheral interface, the StrongARM microcontroller is capa-
are tested in a seismically isolated vault in a static condition ble of commanding the ADC to collect measurement data
to confirm the accelerometer noise floor. While the SD1221 at sample rates as high as 400 Hz. To render networks of
is able to achieve its specified noise floor (the measured noise AWAIRS wireless sensors self-organizing, Agre et al. (1999)
floor was 32 µg), a slow varying drift was reported in the sen- have been careful in selecting a suitable wireless radio for
sor output observed over 30 min. The source of the drift is their prototype. The Conexant RDSSS9M wireless cordless
identified as temperature-dependent, thereby suggesting tem- telephone radio is selected for integration with AWAIRS.
perature compensation is needed for the accelerometer. The The 900 MHz radio employs spread spectrum encoding with
accelerometers are also tested using a low-noise vertical shak- data rates as high as 100 kbps. The communication range of
ing table. As expected, the SD1221 accelerometer outperforms the radio is well over 100 m. When fully assembled, AWAIRS
the ADXL202 during these dynamic tests. In particular, the is only 7.3 × 7.3 × 8.9 cm3 in dimension and is powered by
high noise floor of the ADXL202 results in loss of measure- two 9 V alkaline batteries.
ment accuracy at low frequencies (0–0.3 Hz) as compared to A wireless structural monitoring system proposed by
the SD1221. researchers at MicroStrain is assembled from off-the-shelf
Close research collaboration between the University of electrical components resulting in a functionally rich plat-
California-Berkeley and the Intel Research Berkeley Labora- form (Townsend et al., 2002). At the core of the wireless
tory has resulted in a next-generation Mote platform called sensor node is the 8-bit Microchip PIC16C microcontroller
iMote. As developed by Kling (2003), the hardware design of where embedded software is stored in the microcontroller’s
the iMote is different from those of the MICA, MICA2, and internal electrically erasable, programmable, read-only mem-
MICAz Motes. Recognizing that the sensing application drives ory (EEPROM). To allow for the interfacing of various sen-
the choice for the appropriate sensing interface, the iMote is sors to the node, the Analog Devices AD7714 16-bit ADC is
designed with only a computational core and wireless trans- included in the node design. An attractive feature of the
ceiver. iMotes employ a highly modular construction allow- AD7714 ADC is a programmable voltage gain on the sensor
ing sensing interfaces fabricated as separate boards to be inputs ranging from 1 to 128. To achieve wireless communi-
snapped onto the iMote circuit board. At the core of the cation back to a remote data repository, a surface acoustic
iMote is the 32-bit ARM7TDMI microcontroller operating at wave (SAW) radio operating on the 418 MHz frequency is
12 MHz. This processor selection provides four times greater selected. To modulate digital data upon the selected carrier
computational power than the previously mentioned MICA frequency, the frequency shift keyed (FSK) pulse code mod-

ulation method is employed. This pulse code modulation 3.1. Embedded Operating Systems
technique permits the wireless nodes to detect errors in the
wireless communication channel, thereby increasing the relia- TinyOS, developed by researchers at the University of
bility of the radio. When fully assembled, the wireless node California-Berkeley in collaboration with the Intel Research
is 9 × 6.5 × 2.5 cm3 and is powered by two 1.5 V (3 V total) Berkeley Laboratory, is one of the most widely utilized OSs
lithium–ion batteries. for deeply embedded wireless sensor networks. While TinyOS
Arms et al. (2004) have reported a more recent improve- is not dedicated to any one wireless sensor platform, it is the
ment on the original wireless sensor proposed by Townsend default OS embedded in the computing cores of the various
et al. (2002). The SAW wireless radio originally integrated Crossbow and Intel Motes. A distinct advantage of TinyOS
with the wireless sensor represents a poor utilization of the is that it is an open-source OS readily available to the public
wireless channel since only one sensor node can communicate for free use and modification. Researchers such as Tanner et al.
at any one point in time. Instead, Arms et al. (2004) propose (2003), Glaser (2004), and Kurata et al. (2004) have all uti-
the integration of the Chipcon CC1021 wireless transceiver lized the functionality of TinyOS to perform their laboratory-
with the wireless node. Operating on the 900 MHz radio band, based structural monitoring experiments using Motes.
multiple nodes can utilize the same wireless bandwidth through As described by Hill (2000), TinyOS is intended to maxi-
frequency division multiple access (FDMA) methods. In stark mize the potential of the limited resources available on wire-
contrast to the original time division multiple access (TDMA) less sensors to achieve a functional wireless sensor network.
methods, the new radio allows 26 wireless sensors to com- Specifically, TinyOS is intended to render wireless sensors
municate simultaneously on the 26 individual frequencies defined by short communication ranges both scalable and
equally distributed from 902 to 928 MHz. In addition to energy-efficient. In addition, the limited amount of program
improving the radio, the new wireless sensor proposed by memory often found on wireless sensors necessitates the OS
Arms et al. (2004) also includes 2 MB of on-board RAM for to be as small as possible; TinyOS is designed to occupy only
data storage. 256 bytes of RAM and 4 kB of ROM. TinyOS is written in
the high-level programming language NesC, which is based
3. Embedded Software for Wireless Sensors upon the C programming language. Included in TinyOS are
basic services such as data collection from interfaced sen-
The integration of mobile computing with wireless sensors, processing of sensor signals, and utilization of the radio
sors represents a major paradigm shift in the design and for communication with other wireless sensors. To extend the
use of structural monitoring systems. To take full advan- service life of wireless sensors powered by portable batteries,
tage of the computing power, integrated, embedded soft- numerous low-power modes of operation are included in the
ware is needed to automate wireless sensor operations and OS (see http://www.tinyos.net/).
to process structural response data. Embedded software for TinyOS distinguishes itself from other OSs by its explicit
wireless sensors is often structured as hierarchal layers support of ad hoc networking and multihop data transmis-
(Hayes, 2001; Morton, 2001). At the lowest layer is the OS sion (see http://www.tinyos.net/). When the communication
whose role is to hide implementation details of the underly- range of two wireless sensors is shorter than their physical
ing wireless sensor hardware from upper software layers. separation, connectivity can still be established by multihop-
Above the OS are layers of software dedicated to operating ping. Multihopping is defined as the retransmission of data
the wireless sensor (e.g. collect data, store data, communi- by intermediate wireless sensors so that data arrive at the
cate with other sensors) and for the execution of embedded final intended wireless sensor (or base station). Even though
data interrogation methods (e.g. damage detection algo- many wireless sensors are required for the transmission of
rithms). data in the wireless network, multihopping is more energy-
As wireless sensors continue to shrink in size and cost, the efficient than having longer-range radios which offer direct
demand for dense networks of wireless sensors in SHM connectivity between all sensors (Zhao and Guibas, 2004).
applications will continue to grow. A significant amount of TinyOS offers multihopping connectivity by autonomously
engineering has gone into the design of wireless sensors to routing all communication packets through the wireless sen-
ensure they can be deployed in large numbers. For example, sor network based upon ad hoc peer-to-peer connections.
high data rate radios employing FHSS encoding allow mul- When a wireless sensor is ready to send data, TinyOS calls
tiple wireless sensors to simultaneously share a common the RouteSelector function to recommend an optimal
wireless channel. Furthermore, the absence of cables enforc- multihop route for the packet through the network. To find the
ing a static network topology allows wireless sensor net- optimal path, the RouteSelector function calls estimator
works to form ad hoc topologies built upon simple peer-to- subfunctions that determine the link quality between wireless
peer connectivity. The embedded OS is responsible for man- sensors, geographical position estimates, and power estimates
aging the operation of hardware and to form effective net- to recommend a multihop data route with the highest link
work topologies. quality and lowest power consumption.
With a robust OS in place, engineers can focus their With every wireless sensor playing an important role in the
energies upon the coding of interrogation algorithms. To communication network, the embedded OS must be capable
date, engineering algorithms explored for embedment in of concurrent operation. Concurrency allows wireless sensors
wireless sensors have been from the system identification to perform local tasks such as data processing, while simul-
and damage detection fields. In this section we outline some taneously participating in network communication tasks. To
of the field’s major accomplishments in writing embedded allow TinyOS to handle communication tasks quickly and
interrogation software for wireless sensors deployed for asynchronously, a scheduler is included in the OS to priori-
SHM. tize the servicing of requested services. To ensure no wire-

less sensor reduces the performance of the global wireless sensors generally consume the greatest amount of energy from
sensor network, high priority tasks (e.g. multihop communi- portable batteries. As a means of preserving battery life, local
cations) interrupt lower priority tasks (e.g. data processing). data processing is significantly more energy efficient than
While TinyOS has gained popularity due to its embedment transmitting raw time-history data via the wireless commu-
in the Mote platform, researchers have developed OSs for nication channel.
their own academic prototype wireless sensors. Lynch et al. Straser and Kiremidjian (1998) were the first to describe
(2004a) have described a two-layer embedded software archi- algorithms for determining the health of a civil structure using
tecture intended for local data interrogation by wireless wireless sensors. Their method is to detect the general struc-
sensors including damage detection. The first layer of the tural state immediately following a seismic event. At the core
embedded software acts as a single-thread OS that hides the of their embedded analysis is the use of the normalized Arias
details of the underlying hardware from the upper software intensity (Arias, 1970). As an indirect measure of a structure’s
layer. The OS is written in a modular fashion to encapsulate kinetic energy, the method can be used to detect when energy
the functionality of various subsystems of the wireless sen- is dissipated by the structure during the formation of damage.
sor including the sensing interface and the wireless commu- Essentially summing the square of the structural acceleration
nication channel. Working closely at the hardware level, there measured over the earthquake duration, this decentralized
are six software modules written in C and embedded in the damage detection method is easy to calculate within a wire-
computational core to control the key hardware components less sensor’s microcontroller.
of the wireless sensing unit hardware. The first module oper- The Cooley–Tukey implementation of the fast Fourier
ates the serial port (UART) of the microcontroller for opera- transform (FFT) is successfully embedded in the computa-
tion of the wireless radio. The second and third modules tional core of a wireless sensing unit by Lynch et al. (2003a).
operate the RangeLAN2 wireless modem by implementing the The FFT embedded in the wireless sensing unit is utilized
modem’s communication protocols. To translate analog sen- during field deployments of the wireless monitoring system
sor signals into digital formats upon a precisely timed sched- to provide the frequency response functions (FRFs) of instru-
ule, a fourth module operates the wireless sensor ADC in real mented structures. The accuracy of the complex-valued Fou-
time. The fifth software module performs efficient storage of rier amplitude spectra calculated by the wireless sensor is
sensor data within on-board memory. The module organizes compared to that determined by MATLAB using the same
memory as two data stacks; two stacks permit data to be time-history data. The wireless sensor is shown to provide
buffered in one stack while the wireless radio transmits sen- identical results compared to those generated by MATLAB
sor data from the second stack. The final module is to take (Lynch, 2002).
measurements from digital sensors interfaced to the wireless While many researchers have proposed the use of modal
sensor’s two digital sensor channels. frequencies as a primary damage indicator, the method lacks
Most recently, Wang et al. (2005) have undertaken a rede- sensitivity in structures where environmental factors also
sign of the original OS proposed by Lynch et al. (2004a). contribute to modal frequency shifts (Doebling et al., 1996).
Their new design allows for multiple threads to be executed To fully account for the environmental and operational vari-
at the same time. This permits wireless sensors to simultane- ability of structures, a damage detection methodology based
ously collect measurement data and wirelessly transmit the upon a pattern recognition framework is proposed by Sohn
data, all in real time. In addition, features are added to the and Farrar (2001). Their method begins with the stationary
OS including high-precision clock synchronization across response time history of the structure at a single measurement
the network. location. Using these data, an autoregressive (AR) time series
model is fit to the data. The residual error of the fitted AR
3.2. Embedded Engineering Analyses for time series model and the structural output are then used to fit
Structural Health Monitoring a second autoregressive with exogenous input (ARX) time
series model. The final residual error of the ARX model is
What distinguishes wireless sensor networks from tradi- identified as the damage sensitive feature of the proposed
tional tethered structural monitoring systems is the colloca- method. To accommodate for environmental variability, AR–
tion of computing power with the sensor. This embedded ARX time series models are determined for the structure in
computing power can be utilized by the wireless sensor to its undamaged state when exposed to different operational
self-interrogate structural response measurements it has col- conditions. These AR–ARX time series models form a library
lected. In contrast to a tethered structural monitoring system, of baseline models describing the structure in its undamaged
the wireless sensing infrastructure can be utilized for in-net- state. When the structural response is taken from the structure
work processing of response data for detection of structural in an unknown state (damage or undamaged), an AR–ARX
damage. Autonomous execution of damage detection algo- time series pair is fitted to the data. The coefficients of this AR–
rithms by the wireless sensor represents an important step ARX model pair are then compared to the library of AR–
towards automated SHM. However, allowing sensors to inter- ARX coefficients corresponding to the undamaged structure.
rogate their own data has many other benefits. For example, The undamaged AR–ARX model pair closest (based on the
as the cost and size of wireless sensors drastically reduce, Euclidian distance of the AR coefficients) to that of the AR–
hundreds of sensors could be installed in a single structure. ARX coefficients of the unknown structure is selected from
Such a scenario would result in a centralized data repository the library. If the structure in the unknown state is not dam-
being inundated with measurement data it must collect. There- aged, then the AR–ARX model pair corresponding to the
fore, in-network processing of measurement data can be seen undamaged structure will fit the response data of the unknown
as a means of minimizing data glut in the monitoring system. structure well. If the AR–ARX model pair selected from the
Furthermore, the wireless radios integrated with the wireless database does not fit the data well, then the structure is iden-

tified as damaged. The metric for determining the quality of events. The damage index method, modified from a damage
the model fit is the standard deviation of the ARX model resid- index model initially proposed by Kratzig et al. (1989), consid-
ual error; damage is concluded when the ARX error standard ers the peak response of a structural element in the time domain.
deviation is above an established threshold. Using only the response peaks, the damage index model cal-
The proposed AR–ARX time series method of damage culates an index that falls between 0 and 1. When the index
detection has two attractive features: it is inherently decen- is 0, the element is in its uncracked virgin state. In contrast,
tralized by determining damage in the vicinity of a single when the index equals 1, the element has experienced severe
measurement location, and it does not consume significant damage localization leading to cracking failure of the com-
computational resources. For these two reasons, Lynch et al. ponent. The damage index method is intended to serve as a
(2004d) propose embedding the complete AR–ARX time quick estimate of the health of instrumented cementitious
series method of damage detection in a wireless sensing unit components to allow facility owners to prioritize inspections
for autonomous execution. The methodology allows sensor after an earthquake. The method is both computationally sim-
nodes to be computationally independent so that they can col- ple to implement and sufficiently accurate.
lect raw time-history records and generate results without node- To provide evidence of the energy efficiencies gained by
to-node interaction or the exchange of time-history data. As utilizing wireless sensors for local data processing, Lynch et
described by Lynch et al. (2003d), the coefficients of the AR al. (2003e, 2004d) and Lynch (2004) present a validation study
model are calculated by a wireless sensing unit using Burg’s that compares the differences in battery energy consumed by
method to solving the Yule–Walker equations. Burg’s method transmitting the entire time-history data versus locally inter-
offers many advantages over the least-squares solution to the rogating the data and communicating analysis results only.
Yule–Walker equations, including stability and accuracy. The To measure the energy consumed by the wireless sensing
implemented method uses the wireless sensors to determine unit, the electrical power of each hardware element is meas-
the AR coefficients while AR coefficients are wirelessly trans- ured in the laboratory. After measuring the power, each soft-
mitted to a centralized data repository where the AR–ARX ware operation is precisely timed using the internal timer of
model pairs for the undamaged structure are stored. Once the the wireless sensor’s microcontroller. For example, the amount
closest AR–ARX model is selected from the undamaged of power consumed by the wireless sensor to calculate pri-
library, the coefficients of the undamaged AR–ARX time mary modal frequencies using an embedded 4096-point FFT
series models are transmitted back to the wireless sensor where and communicating peak frequencies is only 2% of the energy
the residual error of the ARX model is found using the orig- needed to transmit the original time-history record. A simi-
inal response time history. The standard deviation of the lar study is performed with the wireless sensing unit deter-
residual error is then checked by the wireless sensor to see if mining the coefficients of AR models. Depending on the size
it has exceeded an established threshold that defines the struc- of the AR model (number of coefficients) and the number of
ture as damaged in the vicinity of the measurement point. data points in the time-history record, the battery energy
Caffrey et al. (2004) propose a decentralized method of consumed by calculating the AR coefficients is between 30%
detecting damage using a network of Motes installed in a and 50% of the energy needed to transmit the original time-
structural system. Recognizing the transmission of sensor history record.
time histories as a wasteful use of the Motes’ batteries, they When the need to transfer time-history records between
propose each wireless sensor to calculate the Fourier spectra wireless sensing units arises, Lynch et al. (2003e) have pro-
of structural acceleration time histories so that modal frequen- posed the use of lossless data compression. Lossless com-
cies and the signal energy contained in each corresponding pression is chosen over lossy compression to preserve the
mode can be determined. After the modal frequencies and integrity of time-history data at all times. Lossless Huffman
modal signal energy contributions are calculated at each sen- coding is encoded in the computational core of wireless
sor location, they are wirelessly transferred to a centralized sensing units (Salomon, 2004). Before encoding the data, the
data repository. The data repository is then given the task of time-history data are decorrelated using a Daubechies-4 dis-
assessing changes in modal frequencies and signal energy crete wavelet transform. Using typical time-history response
contributions to diagnose damage. To validate the proposed data collected in the laboratory, compression rates of 60–
damage detection methodology, a laboratory-based study is 80% are measured when the wireless sensors utilize lossless
performed with the IASC–ASCE Structural Health Monitor- Huffman coding prior to wirelessly communicating data.
ing Benchmark Structure (Johnson et al., 2004). Different com- Alternatively, Caffrey et al. (2004) also advocate compressing
binations of sensors and actuators are employed to actuate and raw time-history data as a means of saving battery power.
to gather the corresponding structural response. By using a However, in contrast to Lynch et al. (2003e), Caffrey et al.
total of 360 different damage patterns, the proposed algorithm (2004) propose the use of lossy compression techniques using
is adequate for detecting damage, especially when a dense sen- wavelet transforms.
sor network with multiple actuators to induce forced excita- Although significant power savings are achieved by Lynch
tions on the structure is present. et al. (2003b), Sazonov et al. (2004) disagree with the notion
To illustrate the ability of wireless sensors to monitor the of leveraging local data interrogation to achieve energy effi-
health of structures based upon the response of individual ciency for their prototype wireless sensor. Instead, they pro-
structural elements (e.g. columns, beams, joints), Lynch et al. pose on-the-fly compression of acquired data to minimize
(2004e) have presented the embedment of a local-based dam- bandwidth consumption during wireless communications.
age detection method based upon damage index models. The Their wireless sensor network system also assumes data are
damage index method proposed is intended for use in struc- aggregated at multiple local data repositories where a con-
tures constructed of cementitious materials that are exposed stant energy source, such as grid power or solar cells, is read-
to cyclic loadings, such as those encountered during seismic ily available. At each local data repository, Sazonov et al.

(2004) propose a damage detection methodology based on AR-ARX model pair using measurement data, AR-ARX_
changes in the strain energy of modes estimated from meas- Database to match time series coefficients to a database of
urement data. coefficients corresponding to the undamaged structure, and
Damage_Index, which determines the damage index at the
3.3. Agent-based Structural Health Monitoring node. To determine the coefficients, the AR-ARX_Damage
Systems service solves the Yule–Walker equations to calculate the
coefficients of the AR–ARX models. If the damage index
Recognizing the potential for future wireless SHM systems determined by the NodeManager agent exceeds an established
to be defined by hundreds of sensor nodes, Ruiz-Sandoval threshold, the acquaintance model permits the agent to com-
(2004) has proposed the development of embedded software municate this information to the BaseStationManager
for wireless sensors based upon an agent framework. An agent, agent, which in turn communicates the presence of possible
as defined by Jennings et al. (1998), is a computer system damage to the end user through the UserHandler agent.
capable of autonomous operation to meet an established To validate the proposed MAS-based wireless SHM system
behavioral objective. When such agents are assembled as an design, Ruiz-Sandoval (2004) implements a 10-story shear
interactive societal group, multi-agent systems (MAS) emerge. building in Simulink. Each agent and corresponding service
The interest of Ruiz-Sandoval (2004) in a MAS framework and acquaintance model are defined using the StateFlow tool-
for automated wireless SHM systems is a result of recog- box within Simulink. Damage is introduced in the structure
nized MAS advantages: (1) high computational throughput by reducing the stiffness of the fourth story by 20%. The
due to concurrent data processing; (2) reduced demand for complete MAS architecture is proven very effective in iden-
communication because of embedded data processing; (3) tifying the presence of this story stiffness reduction using the
excellent scalability due to role decomposition and function- embedded processing of the NodeManager agent.
ality encapsulation. Specifically, for large-scale wireless sen-
sor networks where sensing and processing are collocated at 3.4. Data Repository Driven Software Architectures
each sensing node, MAS frameworks can render resulting
wireless SHM systems more scalable and reliable than non- In contrast to those writing static software to be embedded
MAS implementations. in the computational cores of wireless sensors prior to instal-
Ruiz-Sandoval (2004) selects the Gaia MAS design proc- lation in a structure, Allen (2004) has presented a radically
ess for writing embedded software for wireless sensors. The different architecture for the design of software for wireless
Gaia methodology, proposed by Wooldridge et al. (2000), is SHM systems. In his design, Allen (2004) describes a dynamic
a software abstraction that implements various agent behavio- software architecture that empowers the data repository to
ral models along with rules for inter-agent interaction. Within manage the processing of response data collected by remote
the Gaia framework, three organizational models can be wireless sensors. The software architecture is based upon the
defined: the agent, services, and acquaintance models. The graphical linking and assembly of syntax structure (GLASS)
agent model defines the primary role of each agent in the concept. The GLASS software architecture for a wireless SHM
system. For example, Ruiz-Sandoval (2004) identifies three system begins with the data repository. The data repository
distinct agent models within his wireless SHM system. First, is programmed to contain an extensive library of damage
each wireless sensor is an instantiation of a “node” agent; detection algorithms written for MATLAB. Known as DIA-
secondly, the system data repository is modeled as a “base MOND-II, this library is modular, thereby allowing a user to
station” agent; finally, the human user interacting with the assemble a selected set of algorithms into a single analysis
system is defined as the “user” agent. Once the agents are procedure. To assist the user in designing a damage detec-
defined, the service model defines the roles and functions that tion methodology based upon damage detection algorithms
are to be associated with each agent. For example, considering available in DIAMOND-II, a graphical user interface written
the wireless SHM system, the service model for the “node” in Java is provided to offer “click, drag, and drop” features
agent would include capabilities to collect, process, and com- for algorithm selection and integration. User designed proc-
municate measurement data. With each node’s role defined, esses can then be saved for execution by the data repository
the last step in the Gaia methodology is to define the acquaint- using structural response data. An advantage of Java is that
ance model. The acquaintance model defines the rules for the virtual machine paradigm offered by this high-level pro-
establishing communication links between the agents. gramming language renders the GLASS-based software port-
In order to evaluate the ability of the MAS-based wireless able to any hardware platform.
SHM system to carry out decentralized data processing, Ruiz- Another advantage associated with the modularity of the
Sandoval (2004) implements the multitiered time series GLASS software architecture is that it permits the user to
damage detection algorithm proposed by Sohn and Farrar dynamically assign the responsibility of software execution
(2001). The AR–ARX damage detection method is imple- to various elements in the wireless SHM system. In particular,
mented primarily within the NodeManager agent using the elements of the damage detection process can be dynamically
NodeStructuralHealthMonitoring agent service model. uploaded to wireless sensors via the wireless link for auton-
Unlike the implementation presented by Lynch et al. (2003d) omous execution. This allows the system to take full advan-
where the undamaged structure’s AR–ARX model pairs are tage of computational resources coupled with each wireless
stored in the base station, Ruiz-Sandoval (2004) has proposed sensor to achieve concurrency for software execution.
storing the database of undamaged model pairs within each To validate the proposed software architecture, Allen (2004)
wireless sensor node. Associated with the NodeStructural- has described a laboratory-based experiment using the Husky
HealthMonitoring service model are three primary serv- wireless sensor node to monitor the health of a bolted frame.
ices: AR-ARX_Damage, for determining the coefficients of an Damage is introduced into the frame by loosening preload on

one of the frame’s bolted connections. A small electrodynamic interface to extend the permissible voltage range of the single-
shaker is used to excite the frame while accelerometers are channel interface from –5 to +5 V. Similar to previous wire-
used to record the frame response. Allen (2004) has designed less sensing unit designs proposed by Lynch (2004), the 32-bit
within the GLASS-framework a damage detection method Motorola MPC555 PowerPC microcontroller is selected for
using autoregressive with moving average (ARMA) time series the computational core because of its tremendous computa-
models. The ARMA damage detection method is uploaded tional capabilities. The 10-bit ADC included in the MPC555
from the data repository to the Husky wireless sensor where serves as a 32-channel high-speed sensing interface for the
it is commanded to autonomously execute. Damage (loss wireless active sensing unit design. Both the actuation and
of bolt pre-load) is successfully identified in the study. sensing interface are capable of high-speed data acquisition;
Wong et al. (2005) have presented another approach to the both interfaces have maximum sample rates of 40 kHz. To
software design of a data repository driven structural moni- achieve wireless communication between wireless active sens-
toring system. Their work examines the integration of wireless ing units, the Proxim ProxLink wireless modem operating
sensor networks with a large data network utilized in struc- upon the 900 MHz ISM radio band is selected.
tural testing laboratories. Called the Network for Earthquake To illustrate the full operational capabilities of the proposed
Engineering Simulation (NEES), this network is intended to wireless active sensing unit, Lynch et al. (2004c) have per-
offer global researchers an entire information technology infra- formed a laboratory test using a cantilevered aluminum plate
structure for testing large-scale structural specimens. As part with crack damage introduced. To actuate and sense the alu-
of the network, cable-based sensors installed upon test spec- minum plate, two small lead–zirconate–titanate (PZT) piezoe-
imens are utilized to pass data into a nationally centralized lectric pads are mounted upon the plate surface. The prototype
data repository remotely located on the Internet. In particular, unit employs the actuation interface to command the PZT pad
the system of Wong et al. (2005) utilizes the MICA2 Motes to emit dynamic stationary white noise input signals with zero
to continuously stream measurement data via a wireless com- mean (equal energy across the full frequency spectrum). The
munication link to a local wireless access point. Data trans- second PZT pad is connected to the sensing interface to meas-
mitted by the wireless sensors use specific headers that ure the response of the plate to the input acoustic waves. In the
allow metadata to be created for the wireless sensor data described experiment, actuation and sensing are performed
prior to its entry into the national data repository. The NEES at the maximum sampling rate of 40 kHz in order to capture the
metadata standard is intended to allow the data repository to high-frequency dynamics of the PZT–aluminum plate setup.
be fully curated for easy use by other researchers. Once the data have been collected, the computational core of
the sensing unit locally executes an ARX time series model to
4. Emerging Wireless Sensor Concepts for determine the characterizing system transfer function using
Structural Health Monitoring the input–output response of the plate stored in memory.
Grisso et al. (2005) have described the design of a wire-
4.1. Wireless Active Sensors less active sensing system for SHM. Their design includes a
sensing interface, actuation interface, local data processor, and
In recent years, a number of researchers have proposed power-harvesting energy source. The prototype wireless active
enhancing the functionality of wireless sensors by including sensing unit is tailored to use piezoelectric elements for sens-
capabilities to command actuators. Actuation provides wire- ing, actuation, and power-harvesting. The prototype unit is
less sensors with the capability to be interactive with the sys- called the MEMS-Augmented Structural Sensor (MASSpatch)
tem in which they are installed. Actuation is an important and is to be used for performing impedance-based SHM. At
development in the evolution of wireless sensor technology the center of MASSpatch design is the Diamond System’s
because it closes the gap between traditional structural mon- Prometheus PC104 board which houses a microprocessor
itoring and NDE methods. For example, wireless sensors (ZFx86 at 100 MHz), 16-bit eight-channel ADC, and 32 MB
that actuate can be used to command NDE technologies such flash disk. The ADC contained upon the Prometheus PC104
as active piezoelectric sensors, ultrasonic transducers, among board is configured to operate at a maximum sample rate of
many others. As such, wireless active sensors enhance the 50 kHz, thereby allowing the sensing unit to collect data on
ability of the structural monitoring system to be able to mon- the low range of useful impedance frequencies. To provide
itor structures with a component-level (local) perspective. A the unit with wireless connectivity, two wireless radios, the
summary of wireless active sensing units described herein is Radiometrix RX2M-485-5 and TX2M-458-5, are used for
presented in Table 4. receiving and transmitting, respectively. The radios are tuned
Lynch et al. (2003f, 2004b, 2004c) have proposed the devel- to two distinct frequencies (458.5–491 MHz and 433.05–
opment of a novel prototype wireless “active” sensing unit 434.79 MHz), have communication speeds as high as 5 kbps,
that includes a sensor interface, actuation interface, compu- and can achieve communication ranges of up to 1 km. A novel
tational core, and a wireless communication channel. The feature of the wireless active sensing unit design is the inclu-
prototype design emphasizes functionality, such as simulta- sion of a power-harvesting component. Specifically, MASS-
neous high-speed actuation and sensing, without optimizing patch utilizes piezoelectric elements to transform mechanical
the power efficiency of the hardware design. The design of the energy from ambient motion into electrical charge that can
wireless active sensing unit is not specified for any one type be stored in the sensor’s battery.
of actuator; rather, a transparent actuation interface is sought. To validate the capabilities of the wireless active sensor
The actuation interface is constructed around a Texas Instru- prototype, Grisso et al. (2005) have proposed experiments in
ment DAC7624 digital-to-analog converter that converts which the prototype is used to monitor a damaged structural
digital signals from the microcontroller into an analog com- system. The structural system in question is a bolted mechan-
mand signal. Additional circuitry is included in the actuation ical joint in which the bolt pre-load can be varied to simulate

Table 4. Summary of academic active wireless sensing unit prototypes.

PERFORMANCE Mitchell et al. Lynch et al. Grisso et al. Allen (2004) Liu et al.
ATTRIBUTE (1999, 2000) (2003f, 2004c) (2005) (2005)

A/D Channels 8 32 8 6 1
Sample Rate 30 kHz 40 kHz ≥ 50 kHz 200 kHz 10 MHz
A/D Resolution 8-bit 10-bit 16-bit 16-bit 10-bit
Digital Inputs 0 0 1 0
ACTUATION SPECIFICATIONS
D/A Channels 1
Sample Rate 40 kHz
D/A Resolution 12-bit
Voltage Outputs –5 to +5 V

Processor Phytec miniModul- Motorola Diamond Sys- Pentium Atmel AT94K10AL
535/515 CAN MPC555 Pow- tems PC 104 (MCU/FPGA)
erPC
Bus Size 32-bit 32-bit 8-bit
Clock Speed 10 MHz 40 MHz 100 MHz 133 MHz
Program Memory 32 kB 448 kB 64 kB 256 MB 32 kB
Data Memory 128 kB 512 kB 32 MB Compact Flash

Radio Link Technologies Proxim ProxLink Radiometrix Motorola neuR- SmartRF
MDEV-900-HP Fon AT86RF211 from
Atmel
Frequency Band 900 MHz 902–928 MHz 480 MHz 2.4 GHz
Wireless Standard IEEE 802.15.4
Spread Spectrum Yes Yes Yes Yes (Software)
Outdoor Range 400 m 300 m 1,000 m 9.1 m 305 m
Enclosed Range 150 m 9.1 m 30.5 m
Data Rate 50 kbps 5 kbps 230 kbps 14.4 kbps

Dimensions 7 x 7 x 2.5 cm
Power 6W
Power Source Battery (2.6–16V) Battery (9V) Battery (9V)
damage. A complete impedance-based damage detection undamaged) response. When damage is identified in the struc-
procedure is programmed for autonomous execution by ture, the wireless active sensor sends a wireless signal to the
the wireless active sensing unit. First, a function generator is base station, resulting in the base station turning on a blink-
triggered by the PC 104 board to actuate the structure by ing LED light to signify when damage has been detected.
sending an analog sinusoidal signal to a piezoelectric mac- Recognizing the need to offer even greater actuation
rofiber composite (MFC) actuator mounted to the structure. and sampling frequencies, Liu et al. (2005) propose the
The corresponding response voltage of the same MFC pad is design of an ultrahigh-speed wireless active sensing unit. To
recorded by the wireless active sensing unit. The wireless achieve sample rates (both actuation and sensing) as high as
active sensor generates impedance curves by calculating the 1 MHz, they pursue the adoption of a unique processor that
FRF of the input and sensed MFC voltage signals. The labo- combines an 8-bit microcontroller and field programmable
ratory experiments demonstrate that MASSpatch can clearly gate array (FPGA) within a single integrated circuit. The Atmel
identify damage based on observable shifts in the peaks of AT94K10AL provides both an 8-bit reduced instruction set
the real component of the impedance signature between the computer (RISC) and FPGA to offer a flexible and powerful
baseline (undamaged) and unknown structure (damage versus computational core. In particular, the high-speed FPGA core

of the processor is used to control and operate the sensing including linear controllers and fuzzy logic controllers. To
and actuation circuit peripherals. To record data from inter- validate the hardware and software designs of the prototype
faced sensors, an Analog Devices AD9200 ADC is adopted. wireless controller units, a simple laboratory experiment
This ADC has a 10-bit resolution and can sample as high as involving a three-story shear structure is devised. The struc-
20 MHz. The FPGA can also be utilized to command inter- ture is mounted to a shaking table for application of seismic
faced DACs at similar sample rates. The final component base motions. Using an active mass damper actuator mounted
of the proposed wireless active sensing unit is an Atmel to the topmost story of the structure, the wireless controller
AT86RF211 wireless transceiver. This wireless radio oper- unit is used to successfully control the structure using a fuzzy
ates on a single radio frequency selectable by the user from logic controller.
400 to 900 MHz. Seth et al. (2004, 2005) have presented a feasibility study
The previous wireless active sensing unit designs are pri- of using wireless sensors to execute the control commands
marily utilized to actuate and sense piezoelectric elements needed in a structural control system. Using the wireless active
for acoustic and ultrasonic inspection of structural elements. sensing unit initially proposed by Lynch et al. (2004c), the
However, the wireless active sensing units can also assume properties of the wireless communication channel are assessed.
other powerful roles within a SHM system. For example, Hou In particular, the global performance of the control system is
and Lynch (2005) have explored the use of wireless active quantified in the face of delays in the delivery of state data
sensors to emit electrical signals into cementitious structural between wireless sensors, including potential data loss. Seth
elements. Cementitious composites, such as concrete and fiber- et al. (2005) have found a direct relationship between the total
reinforced cementitious composites (FRCCs), have electrical system demand for the wireless bandwidth and the wireless
resistances that vary in linear proportion to strain. Using the channel’s quality of service; higher demand results in dimin-
prototype wireless active sensors developed by Lynch et al. ished communication reliability. Since poor reliability in the
(2004c), the wireless active sensor injects desired electrical communication channel undermines the real-time requirements
signals into structural elements constructed from high-per- of the control system, novel decentralized wireless control
formance FRCCs (HPFRCCs). In various laboratory tests, an system architectures are presented. The control architectures
MTS 810 load frame is employed to induce monotonic ten- proposed leverage the computing power of each wireless
sile loading on HPFRCC plate specimens while the resistance active sensing unit to execute a single state estimator model
of the specimen is measured using wireless active sensors. common to the system. If the accuracy of the estimated state
The wireless unit collects all voltage measurements, stores is poor, only then does a wireless active sensing unit commu-
them in memory, and calculates the resistance of the material nicate its true state measurement at a given time-step. In
monitored. The tests reveal the ability of the wireless sensors this way, the wireless active sensing units only transmit state
to correlate changes in resistance to both strain and damage data when needed, resulting in drastically reduced demands
introduced in the HPFRCC plates. placed upon the wireless communication channel. Perform-
Wireless active sensing can also play a very important role ing simulation studies in MATLAB with the wireless channel
in future generations of structural control systems. Casciati realistically modeled, the proposed wireless control system
and Rossi (2003) have presented the design of a wireless con- architecture is shown to perform well.
troller unit whose design has been optimized for use in a struc-
tural control system. At the core of their hardware design is the 4.2. Radio Frequency Identification Based Sensors
Cygnal C8051F007 microcontroller, which is easy to program
using high-level programming languages, such as C. To pro- A challenge associated with current wireless sensor tech-
vide a means of collecting structural response data from mul- nology is the finite life expectancy of portable power sources.
tiple channels, a four-channel 12-bit ADC is integrated with Current battery technologies only offer operational lives of the
the wireless controller. To issue real-time commands to actu- order of one day to one year. While promising technologies
ators, a single-channel 12-bit DAC is implemented. This DAC are being researched (e.g. power-harvesting devices), many of
is capable of outputting command signals spanning from 0 to these technologies are still in their infancy. As a result, new
10 V or from –5 to +5 V. Provided the stringent real-time design concepts that address the issue of power are needed
requirement of a controller, Casciati and Rossi (2003) identify for wireless sensors deployed in structures. One approach,
the wireless radio as one of the most important components proposed by many researchers in the structural monitoring
to judiciously select. In particular, a wireless radio offering field, is to employ radio frequency identification (RFID)
low-latency protocols and ensuring the greatest probability wireless technologies for both the delivery of power to wire-
of packet delivery is sought. They select the MaxStream less sensors as well as for data communication. Much simpler
XStream wireless radio for integration with their wireless in design than ordinary wireless sensors, RFID-based wire-
controller design. The XStream is capable of data rates as high less sensors take advantage of remote interrogators (readers)
as 19.2 kbps and communication ranges of over 1 km. When for the transmission of operational power through near-field
fully assembled, the wireless controller is powered by two inductive coupling. The RFID-based wireless sensor cap-
batteries: one +15 V and another –15V. If the wireless con- tures the delivered energy and stores it in temporary capaci-
troller is to be used strictly for data acquisition, a smaller tive storage elements in order to operate an ultralow-power
power source can be used such as a portable 9 V battery. sensor circuit (Finkenzeller, 1999). Once data are collected
After the hardware design of the wireless controller has by the sensor, the stored power is used to modulate the sen-
been selected, Casciati and Rossi describe an impressive array sor data on a radio frequency signal readable by the remote
of software tools written to operate the wireless controller interrogator. The advantage of these battery-free wireless
for structural control applications. In particular, a number of sensors is that they can operate indefinitely in the field.
different controller architectures are offered to the end user, However, passive RFID-based technologies only offer short

communication ranges (< 5 m) between the reader and the sor is designed using a straight thin wire held by two blocks.
device. This range can be drastically reduced when the sensor One block is held fixed while the other is permitted to glide
is embedded in structural materials. In this section we sum- along a linear track. When a structural element undergoes
marize some of the key RFID sensor prototypes designed some displacement, the force acting on the gliding block is
explicitly for structural monitoring applications. greater than the static friction between the wire and block,
Das et al. (1998) have proposed an RFID-based wireless resulting in the wire pulling out a small amount from the fixed
sensor and actuator integrated as a single device using a mul- block. Once the wire pulls out, structural displacements in the
tilayer piezoelectric–dielectric-strip grating design. The pro- direction opposite to what induced the wire pull out cause
posed Simultaneous Sensing and Actuating Smart Antenna the wire to elastically buckle, thus memorizing the experi-
Element (SSASAE) contains a single microstrip antenna pat- enced peak displacement. The wire used in the prototype
terned upon a piezoelectric substrate along with digital and presented by Mita and Takahira is made from fluorocarbon
analog circuitry. The piezoelectric layer is a vital component with a diameter of 0.219 mm. A number of methods can be
in the wireless prototype design since it can be used for adopted to measure the maximum displacement of the wire
both sensing and actuating a dynamic structure to which the relative to the gliding block. One method entails measuring
SSASAE is mounted. A unique aspect of the proposed sen- the electrical resistance of the thin wire. A more elegant rea-
sor is the use of patterned narrow metal strips printed upon a dout mechanism is the use of a capacitor in series with an
buried layer of the device that selectively passes electromag- inductor. A capacitor and inductor, when combined in series,
netic waves in one polarization orientation. This allows the can be inductively coupled with an RFID reader. The tuned
device to be remotely queried using a single wireless signal wireless frequency of the sensor is a function of the capaci-
with different information modulated upon the two orthogo- tance and inductance magnitudes. One of the blocks of the
nal polarization orientations of the signal. As a result, one peak displacement sensor proposed by Mita and Takahira
polarization orientation can be dedicated to querying the (2002) is constructed as two concentric aluminum pipes sep-
SSASAE device for sensor measurements while the other arated by a dielectric material. The concentric pipes act as
orthogonal polarization orientation is dedicated to command- a capacitor whose capacitance changes proportional to the
ing the device as an actuator. Irrespective of what role the change in the peak displacement of the sensor. This results
SSASAE device is assigned (sensing versus actuation), in a change in the frequency of the inductor–capacitor (LC)
power is delivered to the microstrip antenna by a remote circuit. To deliver energy to the sensor (especially when bur-
reader. When the SSASAE device is assigned the role of ied within a structure) and to measure the tuned frequency of
sensor, the device modulates the voltage of the piezoelectric the LC circuit, an external wireless dip meter is used as a
substrate upon the carrier frequency of the remote reader in remote sensor reader. Any shift in frequency of the LC cir-
order to transmit the reading to a data repository. Proof-of- cuit can be correlated to the peak strain or peak displacement
concept devices are built and tested. However, Das et al. recorded by the sensors.
note that the sensors only function well when interrogated at Mita and Takahira (2003) have performed laboratory-
close distances. based experiments to verify the operational principle of the
Jung et al. (1999) have developed a passive RFID-based peak displacement sensor. In the experiment, the resonant fre-
sensor for monitoring the behavior and health of composite quency of the wireless LC circuit is initially tuned to 2.16 MHz.
structures. Consistent with other RFID technologies, the sen- This wireless frequency is set by designing the sensor capac-
sor is powered using a remote reader. The same radio link is itor to have an initial capacitance of 217 pF. The peak dis-
used to transmit data between the wireless sensor and the placement sensor is mounted to a test apparatus that allows
reader. The embeddable RFID wireless sensor uses a flat the sensor to undergo cyclic displacement. The resonant fre-
loop antenna patterned on a conformable polymer substrate quency of the wirelessly interrogated sensor is read using a
for inductive coupling with the reader. When the sensor is dip meter as the apparatus induces displacements. For base-
queried by the reader, the embedded sensor coil antenna gen- line comparisons, the displacement of the apparatus is also
erates an electrical current that is temporarily stored as power measured using a high-precision laser displacement trans-
to operate the sensor circuitry. The sensor circuit consists of ducer. The stated goal of the test is to determine whether the
both a low-power microcontroller and an ADC. The ADC wireless peak displacement sensor can memorize the dis-
included in the embedded sensor design can accommodate placement peak of the test apparatus. From the experimental
multiple sensing channels. While the specific circuit compo- results, and comparing them with those obtained from the laser
nents are not discussed, their total power consumption is less displacement sensor, it can be seen that the sensors perform
than 5 mW. After each sensor transducer output is read and as expected. During the unloading stage, the sensor manages
converted to a digital format, the data are modulated upon to memorize the peak strain by indicating a constant displaced
the same carrier frequency of the remote reader using ampli- state over the unloading time period. Furthermore, when sub-
tude modulation. The embedded wireless sensor has an area jected to greater applied strains, the sensor functions linearly
of 2.5 × 2.5 cm2 while it is only 3 mm thick, rendering it well and increases until it memorizes the next peak strain.
suited for installation within carbon composite plates. Testing To validate the use of the peak displacement sensor for
is performed to assess whether the sensor can be damaged by monitoring the displacement of isolator pads, the wireless
the high-pressure, high-temperature environment of carbon sensor is installed at the base of a seven-story base-isolated
composite manufacturing; tests suggest the sensor experiences building on the campus of Keio University (Mita and Takahira,
no adverse effects. 2004). Again, a laser displacement transducer is used to simul-
Mita and Takahira (2002) have proposed an RFID-based taneously measure the displacement of the isolator pad. Based
sensor that is designed to memorize the peak strain or peak on the experiment using isolator pads, the prototype peak
displacement of an instrumented structural element. This sen- displacement sensor performs almost identically to the laser

displacement transducer, except the wireless sensor success- is capable of being queried by a dip meter to distances of up
fully retains the peak strain during unloading. to 10 cm.
Novak et al. (2003) have proposed the design of a novel Other research teams have also proposed wireless sensors
wireless sensor for SHM. The sensor proposed utilizes an based on RFID technology for monitoring the amount of chlo-
inductor–capacitor (LC) circuit to allow a remote wireless ride ingress in concrete bridge decks. Watters et al. (2003) have
reader to power the sensor and to receive measurement data. proposed the Smart Pebble wireless chloride sensor under
Called a state sensor, the concept of the sensor is similar to development at SRI International. The name is derived from
that proposed by Mita and Takahira (2002); the tuned char- the small size of the final sensor prototype. The final size is
acteristic frequency of the LC circuit serves as the primary rea- equivalent to a moderately sized aggregate typically used in
dout mechanism. The wireless sensor proposed is designed to concrete bridge decks. A direct result of the sensor’s small
identify two possible damage states in civil structures: crack- size is its ability to be fully embedded in a bridge deck dur-
ing in welded steel connections and corrosion in reinforced ing construction. At the core of the hardware design of the
concrete elements. To identify the damage states, the design Smart Pebble is the Microchip MCRF202 RFID device, which
of the wireless state sensor is divided into two parts. The first provides a complete wireless interface for remote readers to
part is the LC resonant circuit that allows the wireless sensor query threshold data from a sensor attached to the MCRF202.
to be read by electromagnetic coupling with a wireless reader. An inductive antenna, constructed of 300 turns of copper wire
The second part is a switch designed to trigger when one of spun on a 2.5 cm diameter spool, is attached to the MCRF202
the damage states occurs in the structure. A core element of to allow the device to electromagnetically couple with an
the switch is a capacitor of a fixed value initially placed in external reader for power and communication. The resonant
parallel with the capacitor of the LC resonant circuit. When frequency of the antenna is designed to be 125 kHz, which is
the switch is initially closed, the resonant frequency of the the radio frequency of the MCRF202 device.
wireless sensor is fixed at one value. When damage or corro- To sense the amount of chloride ingress in concrete bridge
sion in the structure occurs, the switch is opened, thereby decks, Watters et al. (2003) have proposed the use of an
removing the second capacitor from the LC circuit; as a electrolytic cell to correlate the concentration of chloride ions
result, the characteristic frequency of the sensor changes. to a readable voltage signal. In the two-electrode electrolytic
The switch of the wireless state sensor proposed by cell, one electrode is sensitive to chloride ions, and the second
Novak et al. (2003) is modified to allow it to identify one electrode serves as a reference. The electrolyte cell based
of the two damage states sought. To be used as a crack chloride sensor exhibits a strong dependence on temperature.
sensor, the switch consists of a copper foil tape that can be In order to minimize this dependence, the Smart Pebble is
applied to a weld. At one end of the copper tape, a 200 pF designed with a temperature compensation circuit. The com-
capacitor is placed in parallel to the LC circuit whose capac- pleted wireless sensor prototype is packaged in a durable
itance and inductance are 100 pF and 4 µH, respectively. polyurethane rubber enclosure with a cementitious plug on
Together, the total capacitance of the LC circuit and closed one end of the device to allow for the migration of chloride
switch is 300 pF, resulting in a wireless characteristic fre- ions into the sensor cell. A variety of laboratory experiments
quency of roughly 8 MHz. When a crack occurs in an instru- are conducted with the Smart Pebble immersed in saturated
mented weld, the crack causes the copper tape to tear, thereby solutions with high concentrations of chloride ions. Test
isolating the 200 pF capacitor from the circuit. This results in results reveal the sensor to be fully capable of detecting var-
a significant change in the resonant frequency to 4.8 Mhz. ious thresholds of chloride ions.
Although the sensitivity of the proposed crack sensor is the- Carkhuff and Cain (2003) have proposed a passive RFID-
orized to be 0.05 inch, experimental evidence suggests the based wireless sensor to monitor concrete bridge decks for
adhesives used for mounting these sensors strongly affect corrosion. Their prototype wireless sensor, called Smart Aggre-
their performance and sensitivity. Initial tests show the sen- gate (SA), employs short-range RFID telemetry to both power
sors respond to cracks 0.25 inch or larger as a result of the the sensor and to read measurements. Unlike many of the
weak copper tape adhesive. previous RFID-based wireless sensors discussed, the proto-
To functionalize the state sensor for corrosion monitor- type presented by Carkhuff and Cain (2003) adopts two induc-
ing, Novak et al. (2003) replaced the copper tape switch tive coils. One coil, tuned to 1 MHz, is to pick up power from
with a thin steel wire. The steel wire is selected to ensure it a remote interrogator, while a second coil, tuned to 10.5 MHz,
corrodes at the same rate as the reinforcement. When fully is for radio communication. When power is inductively cap-
corroded, the steel wire breaks, which induces a measurable tured by the wireless sensor using the first coil, the alternating
change in the sensor characteristic frequency. Prior to the current (AC) signal is converted to a steady direct current (DC)
steel wire breaking, the characteristic frequency of the sen- signal using a voltage doubler. To ensure a constant voltage
sor is tuned to 8.6 MHz. When the steel wire breaks, this is applied to the sensor circuitry, a Maxim MAX1726 regu-
frequency reduces to 6 MHz. It should be noted that depend- lator is adopted to provide a 3.3 V reference signal. At the
ing on the thickness of the steel wire, the damage threshold core of the SA wireless sensor is the MicroChip 8-bit PIC
can be selected for the sensor. Simonen et al. (2004) have 12LC672 microcontroller. The microprocessor consists of
described an extensive set of successful laboratory tests where an 8-bit ADC used to take sensor measurements and various
the proposed wireless corrosion state sensor is embedded in input–output ports used for communications. To wirelessly
large-scale concrete slabs exposed to various levels of chlo- communicate with the remote reader, a Linear Technology
ride ingress. Andringa et al. (2004) have further presented LTC6900 radio frequency (RF) transceiver is incorporated
a series of additional laboratory tests that allow them to into the SA prototype design. The LTC6900 RF transceiver
refine the hardware design of the wireless corrosion state is capable of communicating on various carrier frequencies
sensor. After optimizing the design of the sensor, the sensor spanning from 1 kHz to 20 MHz using the second inductor

coil included in the SA device. The final prototype is pack- system (Hewlett Packard 35655A) ordinarily interfaced to a
aged in a durable ceramic housing to ensure survivability of personal computer via a data cable, is modified to employ
the sensor during the harsh conditions of pouring wet con- wireless modems. An Aerotelemetry Gina wireless modem,
crete in the field. operating on the 900 MHz ISM radio band, is used to estab-
To validate the performance of the SA wireless sensor, lish two-way communication between the personal computer
Carkhuff and Cain (2003) interface two sensing transducers and the data acquisition system. The modems employ spread
to monitor the health of concrete bridge decks: a thermome- spectrum encoding to minimize interference and are capable
ter and resistance meter. The thermometer is used to monitor of data rates of 38.4 kbps. A plate specimen, densely instru-
the curing process of the concrete deck immediately after pour- mented with strain gages, is monitored by the proposed struc-
ing, while the resistance meter is intended to monitor deck tural monitoring system. During testing, no data are lost and
resistivity as a function of chloride ingress over the deck life- communication ranges of up to 1.5 km (line of sight) are val-
time. For validation, 15 SA prototype sensors are encased in idated.
a 5 cm thick concrete slab while bulk conductivity measure- To quantify the precision of the wireless sensor proposed
ments are made. In addition, 10 SA devices are embedded in by Straser and Kiremidjian (1998), the wireless sensor is
a real bridge deck where the temperature and bulk resistance mounted to the surface of a 1.5 × 1.5 m2 unidirectional shak-
are successfully measured. ing table. A 4 Hz sinusoidal input acceleration is applied to
Saafi and Romine (2004) have proposed a novel design of the table while the prototype wireless sensing unit is mounted
a passive RFID-based corrosion sensor designed to be upon the table surface. To measure the table acceleration, the
embedded into concrete during construction. An innovative MEMS-based EG&IC 3145 accelerometer is also mounted
element of the proposed sensor is the use of MEMS fabrica- to the table surface and interfaced to the wireless sensing unit’s
tion processes to create high-sensitivity sensing transducers 16-bit ADC. Straser and Kiremidjian (1998) have reported
integrated within the RFID-based corrosion sensor platform. that their wireless sensor is able to successfully capture the
In particular, to sense the complete environmental parame- input excitation with high precision. The recorded time history
ters within concrete bridge decks, a MEMS sensor capable of the shaking table is converted to the frequency domain
of monitoring the pH, relative humidity, and the concentra- where little energy is observed outside the 4 Hz frequency. It
tion of chloride ions and CO2, is designed and fabricated. should be noted that this laboratory-controlled experiment
The design of the MEMS sensor is based upon cantilever ele- focuses only on low-frequency excitations. Nevertheless, the
ments micromachined into a silicon substrate upon which experiment successfully verifies the reliability of using spread
polymer films are deposited. Each polymer film is chemically spectrum wireless transmission with no data loss experienced
sensitized to the environmental parameter measured (e.g. pH, between the wireless sensor and a remote base station.
CO2, Cl–, and relative humidity). As the environmental param- To validate the performance of the wireless sensing unit
eter changes, the polymer thin film contracts and expands, proposed by Lynch et al. (2002b), the prototype unit is used to
resulting in the resonant frequencies of the cantilevers to measure the acceleration response of a five-story aluminum
change. To provide the MEMS sensors with RFID-based wire- test structure mounted to a shaking table. On the fifth story
less connectivity, a proprietary wireless system is supplied of the structure, three MEMS accelerometers are installed,
by Microstrain. The Microstrain interrogator consists of an including the Analog Devices ADXL210, Bosch SMB110,
oscillator, demodulator/level shifter, data logger, and antenna and a high-performance planar accelerometer designed and
which is capable of querying the sensor from distances of fabricated by Partridge et al. (2000). The accelerometers are
1.5 cm above the concrete bridge deck with sensors embed- interfaced to a single wireless sensing unit prototype for data
ded 2.5 cm below the deck surface. logging and processing. The five-story shear structure is lat-
erally excited by numerous excitations applied to the shaking
5. Performance Validation of Wireless Sensors table. An interesting feature of their study is the use of the
wireless sensing unit’s computational core to execute an
5.1. Laboratory-based Validation Studies embedded FFT algorithm. The resulting FRFs are wire-
lessly communicated to a laptop computer and are found
As a first step towards deployment of wireless sensor to be nearly identical to the theoretical FRFs derived from
networks in real structures, researchers often validate the a numerical model of the structure. Similar to Straser and
performance of wireless sensors within the well-controlled Kiremidjian (1998), Lynch et al. (2002b) have observed high
environment offered by the laboratory. For the majority of precision data collection while experiencing no data loss
the studies reported in the literature, partial- and full-scale during testing.
structural models have been instrumented using many of the Tanner et al. (2002, 2003) have undertaken a validation
academic and commercial wireless sensor platforms previ- study of the Crossbow Mote wireless sensing platform using
ously summarized in this review paper. For many of these a small structural frame tested in the laboratory. Their study
laboratory-based studies, the primary goal has been to assess assesses the capabilities of the Cossbow Mote for locally
the accuracy and reliability of the wireless sensors installed. processing structural response data using statistical process
However, more recent studies have begun to explore the com- control methods based on cross-correlation coefficients. The
putational potential of wireless sensor networks as a tool for Analog Devices ADXL202 accelerometer included in the
automated interrogation of structural response data for sys- Mote wireless sensor is used to measure the acceleration of
tem identification and damage prognosis applications. the sensor when firmly mounted to the aluminum bolted frame
A conceptual framework for a wireless SHM system has structure. Damage is introduced into the frame by changing
been proposed by Pines and Lovell (1998). To validate the the pre-load of one of the frame’s bolted joints. The structure
feasibility of their framework, a laboratory data acquisition is excited horizontally near the frame base with a 100 Hz

sinusoidal excitation while the Mote sensors measure the floor while 33 Motes are mounted to a single glue laminated
acceleration of the frame in the vicinity of the bolted joint. beam. The intention of this instrumentation strategy is to high-
Tanner et al. note some performance limitations encountered light the capability of a dense wireless sensor network to
using the Crossbow Motes to wirelessly communicate the monitor both global and local structural responses. A key
frame acceleration to a central data repository. In particular, finding reported by Glaser (2004) is the poor performance of
a limited communication range, low analog-to-digital reso- the wireless communication channel when data are sampled
lution, and the inability to simultaneously sample multiple at 100 Hz. Peer-to-peer channel failure is found to be highly
channels are identified. To test the computational capabili- dependent upon the interference encountered from cell phones,
ties of the Motes, a simple damage detection method based radios, TV cameras, and other electronics in the testing area.
upon statistical process control methods is embedded. Spe- To account for lock-ups in the communication channel, the
cifically, the cross-correlation coefficient between the accel- structural response data are alternatively downloaded from
erations measured by two Mote wireless sensors is calculated each problematic Mote from a laptop base station placed in
by the wireless sensors. The wireless sensors are also utilized close proximity. After the data are collected from the wire-
to determine the mean and variance of the cross-correlation less sensor network by the central data repository, various
coefficients which serve as damage sensitive features in the post-collection analyses are performed to successfully iden-
damage detection method employed. tify the presence of both global and local damage.
To assess the performance of the Motes to monitor the Kurata et al. (2004) have devised a laboratory experiment
acceleration response of large-scale civil structure, Casciati to assess the performance of the MICA and MICA2 wireless
et al. (2003a) have described a series of laboratory tests Mote sensors for SHM. A small two-story frame structure is
performed at the ELSA Laboratory located at the European constructed and mounted to the surface of a unidirectional
Joint Research Centre. In their tests, two types of structures shaking table under base excitation. During the first set of
are instrumented with Motes to monitor their acceleration tests, MICA wireless sensors are mounted upon the shaking
response. The first structure instrumented is a three-story frame table surface and each floor of the test structure. Using the
structure which is monitored during free vibration experi- MTS310 sensor board attached to each MICA Mote, the accel-
mentation. The second structure is a portion of an Austrian eration response of the structure is measured during applied
steel railway bridge which is excited by an electromagnetic base motion. Compared to a reference accelerometer mounted
shaker. During testing of these structures, the researchers to the structure, the MICA Mote is capable of accurate response
report significant amounts of data lost during wireless com- measurement. However, the wireless communication chan-
munication to a central data server. For the excitation tests nel of the MICA is cited as unreliable with data loss as high
conducted using the Motes installed upon the Austrian steel as 30%. In the second set of tests, MICA2 are used in lieu of
bridge section, data losses as high as 40% are reported (Cas- the MICA Motes. With application software developed by
ciati et al., 2003c). The data loss problem is so severe that it the Open Systems Laboratory at the University of Illinois at
prevents the researchers from obtaining FRFs for the two Urbana-Champaign, the wireless communication channel of
structures. To improve the reliability of the wireless moni- the improved MICA2 proves significantly more robust. Dur-
toring system, the researchers attempt to reduce the sample ing testing, three MICA2 Motes could be simultaneously
rates. As a result, the communication reliability is improved, queried by a data repository for response data with only 0.5%
but aliasing is subsequently encountered. data loss encountered. Furthermore, the acceleration meas-
To record the dynamic response of full-scale residential urements of the MICA2, when compared to reference accel-
timber buildings, Arici and Mosalam (2003) have presented erometers attached to a cable-based laboratory data acquisition
their work using a dense wireless sensor network for moni- system, are accurate.
toring. In total, 56 wireless Motes are installed upon the first Extending upon the original research conducted by Kurata
floor of a three-story timber structure that is excited at its base et al. (2003a, 2004), Kurata et al. (2003b) have presented
by a shaking table applying real seismic ground motions (e.g. research aimed at assessing the feasibility of using wireless
the 1999 Izmit, Turkey record). Analog Devices ADXL202 Motes to detect structural damage in the two-story test struc-
accelerometers coupled with each Mote are used to record ture. In their tests, the two-story test structure is augmented
the acceleration response of the structure. After the response with added mass (3.3 kg) at each floor to induce P–∆ collapse
is collected, the data are wirelessly communicated to a cen- during lateral excitation. Using a single MICA Mote installed
tral data repository where system identification interrogation upon the topmost story of the structure, the progressive col-
occurs. In parallel to the wireless monitoring system is a dense lapse of the test structure under the JMA-Kobe (NS) ground
array of traditional piezoresistive accelerometers whose out- motion record can be observed. Based upon the acceleration
puts are recorded by a tethered data acquisition system. A response of the instrumented structure, four phases of pro-
comparison of the acceleration time-history records show gressive global structural failure are identified by the MICA
that the structural accelerations recorded by the Motes are Mote using embedded software written to monitor the health
comparable to those recorded by the conventional piezore- of the structure.
sistive accelerometers. Although some Motes exhibit commu- Hou et al. (2005) have utilized the four-channel wireless
nication errors including loss of data, the test successfully sensing units developed by Wang et al. (2005) to monitor the
illustrates the potential for the installation of dense arrays of behavior of a cyclically loaded bridge pier specimen in the
wireless sensors for structural monitoring applications. laboratory. The bridge pier specimen, with a circular diameter
Similar to the study reported by Arici and Mosalam (2003), of 0.4 m and a height of 1.2 m, is constructed from a new
Glaser (2004) has described the installation of a wireless civil engineering material called high-performance fiber
sensor network upon the same full-scale wood-frame build- reinforced cementitious composite (HPFRCC). The labora-
ing. During the study, 25 Motes are installed upon the first tory tests performed on the bridge pier are intended to illus-

trate the potential for wireless sensors to monitor the Alamosa Canyon Bridge (as shown in Figure 5a) to validate
health of the HPFRCC bridge pier under earthquake load- its performance. The bridge has seven independent sections
ing. To monitor the response of the pier under quasi-static spanning 15 m. The construction of each span consists of
lateral load reversals, two linear voltage displacement trans- seven deep steel girders supporting an 18 cm concrete deck.
ducers (LVDTs), a strain gage (mounted to a steel reinforce- An attractive feature of this bridge is that it has been previ-
ment bar), and an accelerometer are installed upon the test ously instrumented as part of a system identification study.
specimen and are read using a wireless sensing unit. Two of As such, the modal properties of the structure are well docu-
the wireless sensing unit channels are used to measure flex- mented. Using the northernmost span of the bridge, five wire-
ural deformations on the tensile and compressive faces of the less sensing units are installed along one of the span’s steel
column base using the LVDTs. The intention of using LVDTs girders. Each wireless sensor records the output of an EG&IC
in this region is to assess if macroscopic-sized cracks form in 3145 MEMS accelerometer mounted to the web of the girder.
the pier base. In addition, the MEMS Crossbow CXL02LF1 Researchers from Los Alamos National Laboratory have
accelerometer is mounted to the top of the pier to measure installed, in parallel to the wireless sensors, a traditional teth-
the tilt angle of the pier under lateral loading. The final chan- ered structural monitoring system to serve as a performance
nel of the wireless sensing unit is used to record the strain baseline. One observation reported by Straser and Kiremid-
in a reinforcement bar within the pier specimen. A special jian (1998) is the time to install both structural monitoring
Wheatstone bridge circuit with amplification, as presented systems. The installation of the wireless monitoring system
by Lynch (2002), is used to amplify and condition the meas- takes 30 min, which is roughly five times faster than the time
urement signal of the 120 Ω strain gage. An identical set of needed to install the cable-based monitoring system. After
instrumentation is installed upon the pier and monitored using the wireless monitoring system has been installed, the relia-
a baseline cable-based data acquisition system. Comparing bility of the communication between a central data repository
the response records obtained from the wireless monitoring (which is a laptop) and the wireless sensing units is assessed.
system with those obtained using the cable-based monitor- Having the data repository issue a data acquisition command
ing system, the two data sets are in complete agreement. Fur- to the wireless sensors, the wireless sensors communicate
thermore, the reliable wireless communication software the ambient response data of the bridge without error. Dur-
experiences no data loss during multiple days of testing. ing the second set of tests, a modal hammer is used to induce
forced vibrations into the instrumented bridge span. Com-
5.2. Field Deployment in Civil Infrastructure paring the acceleration response of the bridge measured by
Systems the wireless sensors and the tethered monitoring system, the
time-history response records are in strong agreement. After
The deployment of wireless sensors and sensor networks the bridge response is collected by the data repository, the
in actual civil structures is perhaps the best approach to assess- frequency domain transfer function of the bridge is calcu-
ing the merits and limitations of this nascent technology. In lated by the data repository. The modal frequencies of the
particular, bridges and buildings provide complex environ- bridge are identical when comparing the transfer function
ments in which wireless sensors can be thoroughly tested. calculated by each monitoring system.
The transition of wireless monitoring systems from the labo- Assessing the performance of the wireless sensor devel-
ratory to the field has been demonstrated by a number of oped for monitoring asphalt pavements, Bennett et al. (1999)
research studies. In all of these studies, the goal of the research- have described a series of field experiments of their device
ers has been to assess the performance of a variety of wireless embedded in an actual asphalt highway surface. Interfaced
sensor platforms for the accurate measurement of structural to the wireless sensor are two strain gages to measure the
acceleration and strain responses. Common to most of the tensile strain of the asphalt lower surface, as well as two ther-
studies reported, the sensitivity and accuracy of the wireless mometers to measure the asphalt temperature. To ensure no
monitoring systems are compared to that of traditional cable- data loss between the embedded wireless sensor and a laptop
based monitoring systems which have been installed along- acting as the data repository, the laptop is positioned roughly
side their wireless counterparts. 4 m from the buried sensor. Before regular traffic is permit-
Perhaps the earliest field validation of wireless teleme- ted on the highway, the reliability of the data channel is
try for monitoring the performance of highway bridges was tested by having the wireless sensor send empty data packets
described by Maser et al. (1996). Their wireless monitoring to the reader. Over a 20 min time period, 100% of the data
system, called the Wireless Global Bridge Evaluation and packets sent by the buried sensor are received. Once the wire-
Monitoring System (WGBEMS), consists of two levels of less channel is tested, the highway is opened to traffic while
wireless communication. The first level includes a wireless the asphalt temperature and strain are continuously recorded.
transceiver coupled with a traditional sensing transducer (e.g. Bennett et al. find that the system records the asphalt tem-
strain gage or accelerometer). This wireless connectivity is perature with an accuracy of 0.2°C and strains with resolu-
intended to be short range for transfer of measurement data tions of 5–10 µε.
from the transducer to an on-site data repository. On the sec- Using the same bridge as Straser and Kiremidjian (1998),
ond level of wireless communication, cellular telephony tech- the performance of the wireless sensing prototype developed
nology is utilized to transfer the aggregated bridge response by Lynch et al. (2003a) is validated in the field. Seven wire-
data to transportation officials situated far from the instru- less sensing units are installed upon an interior span of the
mented bridge site. The total cost of the system is roughly Alamosa Canyon Bridge to measure the bridge response to
$1000 per sensor node and $2000 for the data repository. forced excitations induced by modal hammer blows and truck
After completing the design of their academic wireless traffic. Crossbow CXL01LF1 MEMS accelerometers are
sensor prototype, Straser and Kiremidjian (1998) utilized the epoxy mounted to the web of the span girders and interfaced

Figure 5. Various bridges upon which wireless monitoring systems have been validated: (a) Alamosa Canyon Bridge, NM,
USA; (b) Di Wang Tower, Guangdong, China; (c) Geumdang Bridge, Icheon, Korea.
with the wireless sensing units mounted to the girder flanges. the tethered monitoring system. In contrast to Straser and
MEMS accelerometers are interfaced with the wireless sen- Kiremidjian (1998), the purpose of the field validation study
sors because of their relative low cost compared to more is to determine the modal frequencies of the instrumented
accurate accelerometer types. In parallel to the wireless mon- span using the processing capabilities of the wireless sensors.
itoring system, a traditional cable-based data acquisition sys- The FRF is calculated by the wireless sensing units using an
tem is installed to serve as a performance baseline. The sensor embedded FFT algorithm. When comparing the FRF calcu-
selected for the cable-based monitoring system is the Piezo- lated by the wireless sensors with those calculated by the
tronics PCB336 piezoelectric accelerometer; PCB336 accel- tethered monitoring system, the locations of the primary modal
erometers are mounted adjacent to the Crossbow MEMS frequency peaks and antiresonance valleys are identical. How-
accelerometers. The wireless monitoring system is reported ever, the relatively high noise floor of the MEMS accelerom-
to take half the time to install compared to the cable-based eter results in a lack of agreement between the FRFs below
monitoring system. In comparing the collected time-history 2 Hz. In addition to impulse loadings introduced by modal
acceleration records of the instrumented span during modal hammer blows, the Alamosa Canyon Bridge is dynamically
hammer impact blows, there is strong agreement between excited by a speeding truck. Due to the lack of knowledge of
those recorded by the wireless sensing unit prototypes and the input excitation delivered by the truck crossing the bridge,

the Fourier amplitude spectra calculated by the wireless sens- wireless sensors calculate the level crossing histograms,
ing units are not used as part of a rigorous system identifica- they can be wirelessly downloaded upon demand by the data
tion study. Although calculations of modal frequencies are repository or even to a bridge inspector’s personal digital
possible using the wireless sensing unit prototypes, the absence assistant (PDA).
of an accurate method of time synchronization between the Chung et al. (2004a, 2004b) have described a detailed study
wireless sensing units impedes accurate calculation of sys- taken to validate the performance of their DuraNode wire-
tem mode shapes. less sensing unit prototype. Using two different MEMS accel-
Galbreath et al. (2003) demonstrate the use of a wireless erometers (Analog Devices ADXL210 and Silicon Design
sensor network to monitor the performance of a steel girder SD1221) interfaced to the wireless sensing unit, the ambient
composite deck highway bridge spanning the LaPlatte River and forced response of a 30 m long steel truss bridge is
in Shelburne, Vermont. They select the Microstrain SG-Link recorded. The truss bridge, located upon the campus of the
wireless sensor platform to measure flexural stain on the bot- University of California-Irvine, is for pedestrian traffic only.
tom surface of the bridge girders. To accurately measure strain, The two accelerometers interfaced to the wireless sensor
high-resolution differential variable reluctance transducers are instrumented in the middle of the bridge span. During
(DVRTs), also known as half-bridge linear displacement volt- vibration testing of the instrumented bridge, the accelera-
age transducers, are magnetically mounted to the lower flange tion response of the bridge is collected by the wireless sen-
surfaces of the steel girders and interfaced with the SG-Link sor protytpe and wirelessly communicated to a laptop situated
wireless sensor. The DVRT selected (Microstrain nano-DVRT) 150 m away. To compare the accuracy of the wireless mon-
has a linear displacement resolution of 10 nm and a gage itoring system, a traditional cable-based monitoring system
length of 100 mm. The motivation for installing the SG-Link is also installed; the cable-based system uses piezoelectric
wireless sensors upon the LaPlatte River Bridge is to vali- PCB 393C accelerometers as its primary sensing transducer.
date the continuous real-time streaming performance of the Results from the field study show very strong agreement in
wireless sensor platform. A second goal of the field test is to the acceleration time histories recorded by both the wireless
assess the capability of the sensors to simultaneously record and cable-based monitoring systems. After the completion
2 MB of sensor data in the wireless sensor data bank. During of the field test, the time-history response data recorded by
testing, a data repository responsible for coordination of the the three accelerometers (ADXL210, SD1221, PCB393C) are
wireless sensors, including data collection, is placed 35 m loaded on a personal computer where the FRFs from the
away from the wireless sensors mounted to the bridge. Com- bridge impulse response are calculated for each acceleration
munication between the sensor and the repository is bidirec- record. The first three modes of the bridge can be identified
tional, thereby allowing users to configure the nodes wirelessly from the three FRFs. Subsequently, a theoretical computer
or to trigger the network to collect bridge response data at model is created using SAP 2000; the theoretical modal fre-
any time. Strain within the instrumented bridge girders is meas- quencies predicted by SAP 2000 are compared to those
ured while the bridge remains open to traffic. The study finds obtained from the actual bridge response data. The theoreti-
that the effective resolution of the DVRT strain sensor, when cal model predicts only two modal frequencies in the fre-
interfaced to the wireless sensors, is approximately 1.5 µε. quency region of interest whereas the actual response data
When sampled at 2 kHz, the resolution of the DVRT sensors illuminate three modal frequencies. The two modal frequen-
is sufficient to identify the passing of trucks over the bridge cies predicted by the theoretical model are in agreement with
when viewing the strain time-history records collected. the second and third modal frequencies revealed by the bridge
Aoki et al. (2003) have outlined the validation of their response data. The first mode observed in the experimental
Remote Intelligent Monitoring System (RIMS) wireless sen- response data that is missing from the theoretical model is
sor platform. To test the accuracy of their wireless monitoring attributed to the potential existence of a lateral-torsion deflec-
system, field tests are performed using a flexible light pole tion mode.
mounted to the surface of the Tokyo Rainbow Bridge, Japan. Binns (2004) has presented a wireless sensor system devel-
With fatigue failure common in light poles subjected to fre- oped by researchers at the University of Dayton, Ohio for
quent excitation, the study is intended to illustrate the poten- bridge monitoring. The wireless monitoring system, called
tial of the RIMS wireless monitoring system to monitor the WISE (Wireless InfraStructure Evaluation System), can per-
long-term health of non-structural components on bridges. The form wireless monitoring of bridge structures using any type
selected light pole is instrumented with a tri-axial Micros- of analog sensor. Once installed upon a bridge, communica-
tone MA-3 MEMS accelerometer in order to measure the tion with the WISE system can be established with a laptop
acceleration of the pole top in three orthogonal directions. computer or an inspector’s PDA. The advantage of WISE,
The RIMS wireless sensing unit records the response of the besides the compatibility with any off-the-shelf sensors, is its
pole using the MA-3 accelerometer and wirelessly commu- ability to incorporate an unlimited number of sensor chan-
nicates that measured response to a laptop computer using nels in the global monitoring system (Farhey, 2003). During
wireless local area network (WLAN) communication proto- a field validation study on a highway bridge in Ohio, a WISE
cols. With the data repository situated 50 m away from the system consisting of 16 wireless sensors is installed in 30
wireless sensor prototype, Aoki et al. report no data loss in min. For that bridge study, LVDTs are mounted to the bot-
the wireless communication channel. Their study also tem- tom flanges of steel girders to measure flexural strain result-
porarily stores the three acceleration channel time histories ing from bridge traffic. Time-history records reveal the
in on-board RAM for further local data processing. For exam- ability of the WISE system to accurately detect truck-induced
ple, the computing capabilities of the prototype wireless sens- vibrations of the bridge.
ing unit are employed to calculate a histogram of the level Ou et al. (2005) have described a series of field experiments
crossings of the acceleration time-history records. Once the using MICA Motes installed in a large building. The Di Wang

Tower, located in Guangdong, China, is selected for the instal- viewed as being partially wireless. For the interested reader,
lation of a wireless structural monitoring system comprised a representative list of researchers who have experimented
of eight MICA Motes. The Di Wang Tower, shown in Figure with wireless communications between instrumented struc-
5(b), is 79 stories tall and is constructed as a hybrid structures and the remotely located structural manager/owner is
tural system using steel and reinforced concrete. Potentially as follows: Oshima et al. (2000); Mufti (2003); Sereci et al.
susceptible to vibrations during typhoons, the building is (2003); Elgamal et al. (2003); Karbhari et al. (2003); Wang
instrumented to better understand its wind response behavior. et al. (2004); Tan et al. (2004).
The wireless sensors, using ADXL202 accelerometers, meas-
ure the acceleration of the Di Wang Tower’s 69th floor. The 5.3. Offshore Structures and Naval Vessels
sensors are configured to sample data at 100 Hz and to trans-
mit their measurements to a central data repository. Acceler- Aside from SHM of land-based structures, a number of
ation response data collected by the wireless monitoring research teams have explored the use of wireless sensors on
system are nearly identical to those recorded by a cable-based sea-based structures. Even as naval architects progress toward
monitoring system. sophisticated ship design concepts, the extensive lengths of
Lynch et al. (2005) have installed 14 wireless sensing unit cables that are often needed to connect sensors and actuators
prototypes to monitor the forced vibration response of the with control units distributed throughout a vessel remain a
Geumdang Bridge in Korea. The Geumdang Bridge, presented technological challenge. In addition to the high cost of rout-
in Figure 5(c), is a newly constructed concrete box girder ing wires during construction, naval vessels represent a com-
bridge continuously spanning 122 m. The vertical acceleration plex and harsh environment in which extensive lengths of
of the bridge is measured by the wireless sensing units using wires are vulnerable to detriments such as heat, moisture, and
PCB 3801 capacitive accelerometers mounted on the interior toxic agents (MacGillivray and Goddard, 1997). With wires
spaces of the box girder. In tandem with the wireless monitor- vulnerable to failure when exposed to these harsh conditions,
ing system is a cable-based monitoring system with PCB reduction or outright elimination of wire-based communi-
393C piezoelectric accelerometers mounted adjacent to the cation would greatly enhance the reliability of on-board
wireless sensing unit accelerometers. The stated goals of the engineering control systems while reducing installation and
field validation study are to assess the measurement accuracy maintenance costs.
of the wireless sensing units, to determine the ability of a cen- Ships and offshore structures pose a challenging setting
tral data repository to time synchronize the wireless sensor for the propagation of wireless signals. Estes et al. (2001) have
network, and to use the wireless sensors to calculate the Fou- explored the feasibility of wireless radios for both intra- and
rier amplitude spectra from the recorded acceleration records. inter-compartment shipboard communications within various
Since the bridge was closed to ordinary traffic, Lynch et al. naval vessels (ex-USS America, USS Ross, and USS Carr).
(2005) utilized trucks of known weights (15, 30, and 40 tons) Their study considers radio frequencies between 800 MHz
crossing the bridge at fixed speeds (40, 60, and 80 km h–1) to and 2.5 GHz, which are typical radio frequencies for com-
introduce vibrations. Comparing the recorded time histories mercial off-the-shelf (COTS) radios. Based on the high RF
of the bridge using both monitoring systems (wireless and reflectivity of steel, multipath effects are discovered to dom-
cable-based), the accuracy of the wireless sensing units is inate received radio signals during inter-compartment wire-
confirmed. In addition, the time synchronization procedure less communication. To overcome multipath influences, only
implemented by Wang et al. (2005) is shown to be perfect FHSS wireless radios are found to work. When assessing the
for almost all of the wireless sensing units. Only two of the feasibility of inter-compartment communication, ship bulk-
wireless sensing units had synchronization errors greater than heads severely attenuate wireless signals (of the order of mag-
one time-step (0.014 s) when the monitoring system is sam- nitude of 20–30 dB) but communication through two or three
pled at 70 Hz. Simultaneous to recording the bridge acceler- bulkheads is still found to be possible. Steel is a near-perfect
ation, the wireless sensing units are utilized to execute FFT conductor that reflects electromagnetic waves, thereby limit-
algorithms to locally process the bridge Fourier amplitude ing radio signal penetration. However, on modern ships, a
spectra at each sensor location. The primary modal frequen- number of non-steel elements are present in the bulkheads
cies of the bridge, as measured by the wireless sensors, are (e.g. hatch seals, ducts, cable transits) that allow wireless sig-
shown to be consistent with modal frequencies obtained using nals to penetrate. Mokole et al. (2000) have undertaken a sim-
the tethered monitoring system. After the Fourier amplitude ilar wireless communication feasibility study using COTS
spectra are wirelessly communicated to the centralized data wireless modems that communicate on the 800 MHz to 3 GHz
repository, the operational deflection shapes are calculated. radio frequencies. Their study has found that radio commu-
Since the input to the bridge is not directly monitored, the nication is possible for inter-compartment communication
modes of the bridge are not calculated. If the input to the using commercial wireless radios, even when bulkhead clo-
bridge is a white noise excitation and the modal frequencies sures are securely fastened.
are well spaced, then the operational deflection shapes are Gause et al. (1999) have presented a prototype wireless
equivalent to the bridge mode shapes. sensor for monitoring strain within the composite decks of
It is important to note that the researchers mentioned above naval vessels with a high electromagnetic interference (EMI)
have dedicated their efforts to evaluate the performance of environment. Called the remotely queried embedded micro-
wireless sensors to monitor large-scale civil structures. A sensor (RQEM), the wireless sensor is designed to perform
number of other research teams have explored the use of wire- in the face of common naval EMI conditions. The wireless
less communications to transfer data from a traditional cable- sensors are intended to be impregnated within composite
based structural monitoring system to data repositories located elements; with no connection to other sensors, the sensors
far from the structure’s location. Such systems could also be must be wirelessly queried using RFID readers situated near

the composite structure. To assess the accuracy of the strain posed wireless monitoring system upon an offshore structure
sensor and the reliability of RFID-based communication in in China.
the face of EMI, two RQEM transponders are installed on
the Advanced Technology Demonstration (AEM/S) System 5.4. Aircrafts and Aircraft Components
on board the USS Arthur W. Radford. The two RQEM
sensors are installed between the fiberglass/epoxy layers Gause et al. (1999) have proposed the use of RQEM wire-
of the ship’s composite mast structure. The selection of less sensors on military aircraft so that SHM can be performed
this particular location is due to the EMI resulting from in a cost-effective manner. In their study, two RQEM devices
ultrahigh frequency (UHF) based air search radars mounted similar to those used in the USS Radford are installed in the
to the mast. Having monitored the sensors with the ship at carbon epoxy composite access panel and aft metal door
sea in constant operation and going through hurricane seas, panel of a Boeing AV-8B Harrier plane. Interrogation of the
the study deems the sensors fully operational, accurate, and embedded sensors is accomplished from the aircraft exterior
reliable. Thus, this study reveals that despite high EMI envi- using an RFID reader. Initial studies show that interrogation
ronments, RFID RQEM sensor devices can withstand the can be accomplished through the access panels and aircraft
high level of radar activity for potential monitoring of naval skin to allow for such sensor installments. In addition to testing
structures. the ability to query these sensors on an aircraft, the sensors are
Schwartz (2002) has presented a novel ship design con- tested to ensure they can withstand high temperatures, high
cept using wireless sensors embedded in various ship systems acceleration vibrations, and acoustic waves of 170 dB or higher.
to reduce manning requirements. Wireless communication Using a commercial Trovan RFID reader, measurements are
between embedded sensors and existing shipboard local area taken by the RQEM sensors during short 2 hour flights to
networks (LANs) are also aimed to reduce ship construction ensure the sensors can survive typical flight environments.
and maintenance costs. Wireless sensors are proposed to After over 50 hour of total flight time, these sensors appear
monitor the environmental parameters of ship spaces, the to have no degradation despite the harsh flight conditions.
structural integrity of the hull, and the operational health of Varadan and Varadan (2000) have presented an integrated
critical ship machinery. The framework proposed has wire- wireless sensor package of MEMS sensors, inter-digital trans-
less sensors interfacing to a ship LAN through 802.11 wire- ducers (IDTs), and conformable antennas. At the core of their
less access points (WAPs). The system has been validated proposed wireless system is the use of an inter-digital trans-
successfully on numerous naval vessels including the USS ducer micromachined upon silicon substrates. The IDT devices
Monterey and the ex-USS Shadwell. are capable of actuating and sensing SAWs introduced in the
Ploeger et al. (2003) have described a cost-effective wireless substrate. Changes in the propagation properties of the SAW
monitoring system that monitors the operational health of in the substrate can be correlated to various response parame-
a shipboard ventilation system. The wireless system is con- ters of a structure, thereby serving as a sensor. A core element
structed from wireless data acquisition nodes, called the Intel- of the proposed platform is the integration of an inductive
ligent Component Health Monitor (ICHM), that are capable coil antenna to allow for power delivery by a remote reader
of collecting sensor data from analog sensors and commu- in addition to a two-way communication channel between
nicating that data via Bluetooth wireless radios to a central- the device and the reader. Embracing RFID-based wireless
ized data repository, called the Compartment Health Monitor telemetry, IDT wireless sensors can be mounted to the surface
(CHM). The system described is well suited for intra-com- or impregnated in between the layers of composite structures.
partment communication because of the short 20–30 m com- Most recently, Kim et al. (2002) have presented the fabri-
munication range associated with Bluetooth radios. Simple cation of the IDT sensor upon a piezoelectric substrate. The
data processing of sensor data, including threshold detection piezoelectric substrate is used to introduce Lamb or
and spectrum analysis, is performed at the compartment’s Rayleigh waves into structural elements to detect crack dam-
CHM server. As validation, ventilation fans upon an opera- age. In particular, they discuss the potential impact wireless
tional aircraft carrier are monitored for overall health using IDT sensors can have on the health monitoring of aircraft. In
the described system. particular, the sensors are proposed for integration in aircraft
Most recently, Takahashi (2004) has reported on the use skins as well as in turbine engines. To illustrate the function-
of wireless sensors for wireless monitoring of oil tankers. ality of the MEMS–IDT sensors, a set is mounted on the sur-
Wireless sensors manufactured by Dust Networks are being face of an existing airfoil. Since these sensors are surface
installed throughout various oil tankers, especially in criti- mounted and must be protected from harsh flight environ-
cal regions where structural or mechanical problems could ments, a thin coating of UV curable multifunctional polymer
potentially occur. For example, wireless sensors are installed (100 µm thick) is applied above the MEMS–IDT sensor. Ini-
upon oil tanker motors to detect out-of-ordinary vibrations tial studies on the cracked airfoil indicate that the MEMS–
so that mechanics can be alerted. Currently, British Petroleum IDT sensor can successfully detect cracks in aircraft struc-
(BP) has also installed wireless sensors on offshore platforms tures. Furthermore, the reflected Lamb wave signals from
as well as on tanker trucks to assess their utility. the MEMS–IDT sensors indicate that the signal response
Li et al. (2003) and Ou and Li (2003) have looked into the changes in concert with the detected crack size.
feasibility of using wireless sensors for monitoring the health Ihler et al. (2000) have proposed a unique approach to
of offshore oil platforms. In short, the wireless sensor net- wirelessly monitoring corrosion cracking in aircraft skins.
work (WSN) consists of multiple sensor nodes wirelessly con- Their wireless system is similar in operation to the RFID-
nected to a server base station with the base station gathering based wireless sensor proposed by Novak et al. (2003) for
all the data for processing and distribution through a LAN or monitoring corrosion of steel in reinforced concrete. The
the Internet. The authors have plans to instrument the pro- device proposed by Ihler et al. (2000) is a wireless crack-

wire sensor sufficiently resilient to withstand the extreme again based on detectable changes in the characteristic fre-
temperature cycles encountered during fabrication of carbon quency of the wireless sensor’s LC circuitry. In order to suc-
fiber composite laminates. The design of the wireless sensor cessfully embed this sensing device within structural elements,
begins with four conductive wires mounted to the surface of a polymer thick film (PTF) ink is used to screen print an
an epoxy-based substrate, which in turn is mounted to the sur- inductive coil antenna onto Mylar. Since screen printing
face of an aircraft. When cracking occurs in the surface, the allows different patterns to be formed, this fabrication meth-
wires begin to break, resulting in a change in the circuitry odology provides the designer with the flexibility to achieve
and a corresponding change of the resonant frequency of the desired antenna properties. Moreover, various substrates such
sensor antenna. The magnitude of the crack is tracked by the as ceramic, printed wiring boards, and polyimide film are also
number of wires broken and the changes in the sensor fre- compatible with PTF ink screen printing. In total, the thick-
quency resulting from those wire failures. Power is delivered ness of the printed PTF board is less than 25 µm. To validate
to the wireless sensor through a reader located in the vicinity the performance of the RFID prototype, a parallel resistor
of the impregnated sensor. To ensure the wireless signal can LC (RLC) system is printed. The applied stress of an instru-
propagate through the aircraft frame, the wireless sensor mented structural element changes the resistance of the cir-
prototype is tuned to 24 GHz. cuitry and causes a change in both the characteristic frequency
In anticipation of the development of next-generation and quality factor of the RLC circuit. The laboratory-based
reusable space vehicles, Milos et al. (2001) have proposed validation study indicates that the reader, a grid dip meter
two RFID-based launch-vehicle health monitoring prototypes (GDM), can successfully detect the resonant frequency change
(SensorTag) to monitor the temperature of a space vehicle’s and observe the frequency shift from an undamaged to a
thermal protection system (TPS). Due to tens of thousands of damaged state.
individual parts susceptible to damage on a reusable launch
vehicle TPS, a high density of sensors would be needed. While 5.5. Railroad Vehicles and Railway Structures
a large number of sensors are needed, vehicle design require-
ments restrict the total mass of the vehicle. In order to have a Nejikovsky and Keller (2000) present a comprehensive
large number of wireless sensors taking up negligible weight, monitoring system to monitor the dynamic behavior and struc-
passive low-frequency (125 kHz) RFID tags are selected for tural condition of railway cars. An onboard monitor (OM)
the prototype SensorTag units. These tags are small in size, system is proposed to measure various railway car operational
require no power, and can be embedded within the TPS. The conditions including acceleration, speed, temperature, etc.
design of the circuitry of the prototype SensorTag utilizes a The OM offers eight analog sensing channels to which a vari-
fuse/switch within an RFID circuit. As soon as the tempera- ety of sensors can be interfaced. In addition to these eight
ture exceeds a predetermined temperature correlated to high channels, a GPS receiver is integrated in the design of the OM
occurrences of damage, the fuse is blown to cause an open to provide a real-time clock and location position capability.
circuit in part of the RFID circuitry. This causes a shift in To communicate railcar responses to railroad officials in real
the characteristic frequency of the RFID tag from 125 kHz time, cellular and satellite wireless radios are integrated in the
to approximately 156 kHz. Since each RFID SensorTag is OM design. The entire OM wireless sensing unit is housed in
embedded with a unique identification code, the acquired a rugged NEMA-4 enclosure and is powered by an unregu-
data correspond to the location of the sensor. An initial set of lated AC (110V) or DC (74 or 36V) power source.
prototype RFID sensors is embedded within the TPS of an Nejikovsky and Keller (2000) have installed OM devices
X-34 aircraft. Laboratory-based validation studies at SRI on railcars to create a comprehensive Remote Rail Monitoring
International and arc-jet heating tests at the National Aeronau- System (RRMS) for railroad companies. The first field appli-
tics and Space Administration (NASA) Ames verify the abil- cation of this monitoring system is installed for CONRAIL’s
ity to read data from a large number of embedded SensorTags SD-50 locomotives to monitor various engine performance
using a standard RFID reader. While the reader is found parameters including water and oil temperatures, engine
reliable in reading the characteristic frequency of each Sen- temperature switch statuses, fans statuses, throttle positions,
sorTag, SensorTags are found to sustain temperatures only rotations per minute, and traction motor currents. Using a two-
below 200°C. To rectify this shortcoming, a second set of pro- way satellite communication messaging service, detailed
totypes utilizes a ferrite rod, insulated copper wire, capacitor, information is remotely displayed at the railroad command
microfuse, and a Microchip Technologies MCRF202 micro- center in real time. A second field application of the RRMS
controller. This second generation design allows for higher system is the installation of the system on the Pacific North-
operation temperatures of up to 400°C. In addition, instead west rail passenger cars to monitor their structural conditions.
of identifying frequency shifts, the embedded microfuse and In particular, each OM is intended to measure the lateral accel-
microcontroller allow the RFID reader to pick up a bit-inverted erations of railcars during curves to ensure they do not exceed
code when the microfuse exceeds the operational tempera- government cant safety levels. During four months of field
ture of the vehicle. testing on various locomotive trains, the system identifies nine
A research team led by Walsh et al. (2001) has designed a banked track locations where lateral accelerations exceeded
prototype RFID-based sensor for rotary wing flight systems. the federal safety limits.
Although the design of their sensor is similar to other RFID
wireless sensors, the research team has successfully integrated 6. Conclusions
the sensor into the vehicular structure without disrupting the
performance of the structure. To accomplish this task, a pas- In this paper we have explored the historical development
sive RFID-based system is selected for embedment in key of wireless sensors and sensor networks intended for SHM.
structural elements. The damage detection methodology is Since the mid-1990s, a number of research teams in both

academia and industry have proposed an impressive array devices (e.g. cell phones, PDAs) in everyday use, battery
of wireless sensor prototypes featuring a wide offering of technology might begin to mature at an increasing rate. In
functionalities. As can be observed in more recent proto- the meantime, other solutions are needed to address the strong
types, embedded system technology continues to mature at dependence wireless sensors currently have on finite battery
ever faster rates with components offering greater function- sources. One immediate approach to maximizing the life of
ality at lower costs. These trends are critically important to existing battery sources is to devise an optimal usage strat-
ensure wireless SHM is adopted by business and govern- egy for wireless sensors. Such strategies include maximizing
ment entities seeking to take full advantage of the favorable the time sensors are placed in sleep mode, as well as pursuit
cost–benefit trade-off offered by the technology. of duty cycle usage schemes. In addition, minimizing the
The hardware architecture is a critical element in the design need to transfer long time histories of structural response
of wireless sensors optimized for monitoring the perform- data, by programming sensors to locally interrogate their data
ance and health of structures. Equally important is the design first, seems appropriate in most applications. Energy can also
of embedded software that operates each wireless sensor in be conserved by designing the wireless sensor hardware to be
the field. In this paper, a review of embedded software is as low power as possible. Another direction is to eliminate
provided, including OSs and embedded engineering analy- the power source all together, as has been done in the area of
ses. The true power of wireless sensors is the collocation of RFID-based wireless sensors. Innovation in the area of RFID-
mobile computing power with the sensor transducer. There- based sensors will continue, especially for applications such
fore, to truly harness the potential of a wireless sensor net- as monitoring the health of layered composite structures includ-
work, well designed software is necessary for the wireless ing military aircraft and ships.
sensors. One justification for emphasizing local processing The research community has proposed another approach
by wireless sensors is to prolong the life expectancy of bat- to addressing the limitation of current battery technologies:
tery power sources. For example, in many of the wireless power harvesting. Power harvesting entails the use of trans-
sensor prototypes discussed in this paper, the wireless radio ducers that convert ambient energy sources (e.g. solar power,
consumes more electrical energy than the computational thermal, wind, vibrations) into usable and storable electrical
core; therefore, energy can be saved by minimizing the use energy. A number of power-harvesting devices have been
of the wireless communication channel. Local data processing proposed to take advantage of structural vibrations as the
is also a powerful tool for ensuring the scalability of wireless energy source to harvest. A number of innovative vibration-
sensor networks since it is a primary means of minimizing based power-harvesting technologies have been proposed
data glut in the network. (Meninger et al., 1999, 2001; Elvin et al., 2001; Casciati et al.,
Wireless sensor networks are sufficiently mature that many 2003d; Churchill et al., 2003; Sodano et al., 2003, 2004; Wang
field validation studies have been undertaken. A wide assort- et al., 2003b). The validity of using solar power for power-
ment of structures, ranging from aircrafts to bridges, has been ing wireless sensors installed on bridges has also been pro-
utilized to showcase the merits of wireless structural moni- posed by Chung et al. (2004a). While many of the vibration-
toring. In recent years, networks of ever greater numbers of based power-harvesting technologies show tremendous prom-
wireless sensors have been installed on structures with great ise, more research is needed to render power harvesting ready
success. In many of the field studies, researchers have been for widespread commercial adoption.
able to utilize the in-network data processing capabilities of A significant advantage of wireless sensor networks over
wireless sensors to interrogate measurement data in near real traditional cable-based monitoring systems is the colloca-
time to derive performance parameters such as modal fre- tion of computational power with the sensing transducer. In
quencies and damage indices. essence, this feature transforms the wireless monitoring
In many respects, wireless sensor networks are in their system into a genuine SHM system where damage detection
infancy. The majority of wireless sensor prototypes described is fully automated. To date, many engineering algorithms,
in this review are passive devices that only record the including Fourier transforms, wavelet transforms, and sys-
response of the structure. In future years, wireless sensors tem identification models, have been embedded. However,
with actuation interfaces will prove to be even more power- wireless sensor networks should be viewed as a decen-
ful for monitoring structures for damage. Already, a number tralized architecture offering parallel processing of meas-
of research teams have showcased the capabilities of wire- urement data. More research is needed to arrive at truly
less active sensing units for ultrasonic NDE analyses. Active distributed data interrogation schemes designed explicitly
sensing is not the only application for wireless sensors with for the parallelism and decentralization offered by wireless
actuation capabilities. For example, structural control is another sensor networks.
ripe application for a sensor platform capable of interacting As the field of wireless sensors and sensor networks matures,
with the structure. A more futuristic use of actuation might the technology must continuously be installed in real struc-
even include the use of mobile wireless sensors. One such tures to fully validate performance in the complex field envi-
system, proposed and illustrated by Huston et al. (2003), ronment. In the future, researchers will attempt to install
includes the design of robotic or mobile wireless sensors that ever greater numbers of wireless sensors in actual structures.
render a SHM system easier to install and adaptive to chang- Large-scale deployments, defined by higher nodal densities,
ing structural conditions. will continue to illustrate the scalability of wireless sensor
A remaining limitation of current wireless sensors is the networks for SHM. To date, the majority of wireless systems
finite energy sources used to power devices in the field. Bat- have been left within a structure for the duration of testing.
tery technology has only progressed incrementally; this is in In the future, field tests will be devised to test wireless sen-
stark contrast to the Moore laws encountered in the micro- sors in longer-term deployments. Tests like these could offer
processor and wireless fields. Given the ubiquity of mobile opportunities to refine duty cycle usage strategies, to assess

system performance versus environmental factors, and to Caffrey, J., Govindan, R., Johnson, E., Krishnamachari, B., Masri, S.,
test the long-term reliability of wireless sensors. Sukhatme, G., Chintalapudi, K., Dantu, K., Rangwala, S., Sridharan,
A., Xu, N., and Zuniga, M., 2004, “Networked Sensing for Structural
Prior to the efforts of the open-source community to Health Monitoring,” in Proceedings of the 4th International Work-
embrace a common homogeneous wireless sensor platform shop on Structural Control, New York, NY, June 10–11, 57–66.
(e.g. the Mote platform), a number of academic research teams Carkhuff, B. and Cain, R., 2003, “Corrosion Sensors for Concrete Bridges,”
had undertaken the design of proprietary wireless sensors. IEEE Instrumentation and Measurement Magazine, Vol. 6, No. 2,
19–24.
The benefit gained by a researcher in designing a wireless Casciati, F. and Rossi, R., 2003, “Fuzzy Chip Controllers and Wireless
sensor platform is the ability to attain the specific perform- Links in Smart Structures,” in Proceedings of the AMAS/ECCO-
ance features desired. However, a drawback of having a het- MAS/STC Workshop on Smart Materials and Structures (SMART’03),
erogeneous set of proprietary wireless sensors is their inability Warsaw, Poland, September 2–5.
Casciati, F., Faravelli, L., Borghetti, F., and Fornasari, A., 2003a, “Future
to easily communicate with one another. In recent years, a Trends in Infrastructure Monitoring,” in Proceedings of the Interna-
new wireless communication protocol has emerged that has tional Conference on Structural Health Monitoring and Intelligent
been explicitly designed for wireless sensor networks. Called Infrastructure, Tokyo, Japan, November 13–15, Vol. 2, 997–1002.
IEEE 802.15.4, this wireless personal area network (WPAN) Casciati, F., Faravelli, L., Borghetti, F., and Fornasari, A., 2003b, “Tuning
the Frequency Band of a Wireless Sensor Network,” in Proceedings
standard provides mobile battery-dependent devices a wire- of the 4th International Workshop on Structural Health Monitoring,
less media access protocol of low complexity (IEEE 2003). Stanford, CA, September 15–17, 1185–1192.
In addition, the physical layer design of 802.15.4 is intended Casciati, F., Faravelli, L., and Borghetti, F., 2003c, “Wireless Links
to provide the most energy-efficient wireless communica- Between Sensor-device Control Stations in Long Span Bridges,” in
Smart Structures and Materials: Smart Systems and Nondestructive
tion protocol available. While an immediate benefit of IEEE Evaluation, San Diego, CA, March 3–6, Proceedings of the SPIE,
802.15.4 is its low power consumption, a potentially greater Vol. 5057, 1–7.
benefit is that it offers a common protocol for wireless sen- Casciati, F., Civera, P., and Rossi, R., 2003d, “Power Harvesting for Wire-
sor networks. As such, wireless sensors of different designs less Smart Sensors,” in Proceedings of the International Workshop
on Advanced Sensors, Structural Health Monitoring, and Smart
and with different performance features can interoperate. It Structures, Tokyo, Japan, November 10–11.
should be pointed out that other communication protocols, Casciati, F., Casciati, S., Faravelli, L., and Rossi, R., 2004, “Hybrid Wire-
such as Bluetooth, could also offer the same interoperability less Sensor Network,” in Smart Structures and Materials: Sensors
within heterogeneous wireless sensor networks. and Smart Structures Technologies for Civil, Mechanical, and Aero-
space Systems, San Diego, CA, March 15–18, Proceedings of the
SPIE, Vol. 5391, 308–313.
References Celebi M., 2002, “Seismic Instrumentation of Buildings (With Emphasis
on Federal Buildings),” Technical Report No. 0-7460-68170, United
Agre, J. R., Clare, L. P., Pottie, G. J., and Romanov, N. P., 1999, “Develop- States Geological Survey, Menlo Park, CA.
ment Platform for Self-organizing Wireless Sensor Networks,” in Chung, H-C., Enomoto, T., Shinozuka, M., Chou, P., Park, C., Yokoi, I.,
Proceedings of the Conference on Unattended Ground Sensor Tech- and Morishita, S., 2004a, “Real-time Visualization of Structural
nologies and Applications, Orlando, FL, April 8–9, Proceedings of Response with Wireless MEMS Sensors,” in Proceedings of the 13th
the SPIE, Vol. 3713, 257–268. World Conference on Earthquake Engineering, Vancouver, BC,
Allen, D. W., 2004, “Software for Manipulating and Embedding Data Canada, August 2–6.
Interrogation Algorithms into Integrated Systems,” M.S. Thesis, Chung, H-C., Enotomo, T., Loh, K., and Shinozuka, M., 2004b, “Real-time
Department of Mechanical Engineering, Virginia Polytechnic Insti- Visualization of Bridge Structural Response Through Wireless
tute and State University, Blacksburg, VA. MEMS Sensors,” in Testing, Reliability, and Application of Micro-
Andringa, M. M., Neikirk, D. P., and Wood, S. L., 2004, “Unpowered and Nano-Material Systems II, San Diego, CA, March 15–17, Pro-
Wireless Analog Resistance Sensor,” in Smart Structures and Mate- ceedings of SPIE, Vol. 5392, 239–246.
rials: Sensors and Smart Structures Technologies for Civil, Mechani- Churchill, D. L., Hamel, M. J., Townsend, C. P., and Arms, S. W., 2003,
cal, and Aerospace Systems, San Diego, CA, March 15–18, “Strain Energy Harvesting for Wireless Sensor Networks,” in Smart
Proceedings of the SPIE, Vol. 5391, 356–367. Structures and Materials: Smart Electronics, MEMS, BioMEMS,
Aoki, S., Fujino, Y., and Abe, M., 2003, “Intelligent Bridge Maintenance and Nanotechnology, San Diego, CA, March 3–6, Proceedings of
System Using MEMS and Network Technology,” in Smart Systems SPIE, Vol. 5055, 319–327.
and NDE for Civil Infrastructures, San Diego, CA, March 3–6, Pro- Crossbow, 2004, MICAz MPR2400CA Data Sheet, Crossbow, Sunnyvale,
ceedings of the SPIE, Vol. 5057, 37–42. CA.
Arias, A., 1970, “A Measure of Earthquake Intensity,” in Seismic Design Culler, D. E. and Mulder, H., 2004, “Smart Sensors to Network the
for Nuclear Power Plants, R. Hanson, ed., MIT Press, Cambridge, MA. World,” Scientific American, Vol. 290, No. 6, 84–91.
Arici, Y. and Mosalam, K. M., 2003, “Modal Analysis of a Densely Instru- Das, N. K., Khorrami, F., and Nourbakhsh, S., 1998, “A New Integrated
mented Building Using Strong Motion Data,” in Proceedings of the Piezoelectric–Dielectric Microstrip Antenna for Dual Wireless
International Conference on Applications of Statistics and Probabil- Actuation and Sensing Functions,” in Smart Electronics and MEMS,
ity in Civil Engineering, San Francisco, CA, June 6–9, 419–426. San Diego, CA, March 2–4, Proceedings of the SPIE, Vol. 3328,
Arms, S. P., Townsend, C. P., Churchill, D. L., Hamel, M. J., Galbreath, J. 133–146.
H., and Mundell, S. W., 2004, “Frequency Agile Wireless Sensor Davidson, A. and Hill, C., 1997, “Measurement of Building Penetration
Networks,” in Smart Electronics, MEMS, BioMEMS, and Nanote- into Medium Buildings at 900 and 1500 MHz,” IEEE Transactions
chnology, San Diego, CA, March 15–18, Proceedings of the SPIE, on Vehicular Technology, Vol. 46, No. 1, 161–168.
Vol. 5389, 468–475. Doebling, S. W., Farrar, C. R., Prime, M. B., and Shevitz, D. W., 1996,
Basheer, M. R., Rao, V., and Derriso, M., 2003, “Self-organizing Wireless “Damage Identification and Health Monitoring of Structural and
Sensor Networks for Structural Health Monitoring,” in Proceedings Mechanical Systems from Changes in their Vibration Characteris-
of the 4th International Workshop on Structural Health Monitoring, tics: A Literature Review,” Report No. LA-13070-MS, Los Alamos
Stanford, CA, September 15–17, 1193–1206. National Laboratory, Los Alamos, NM.
Bennett, R., Hayes-Gill, B., Crowe, J. A., Armitage, R., Rodgers, D., and Doebling, S. W., Farrar, C. R., and Prime, M. B., 1998, “A Summary
Hendroff, A., 1999, “Wireless Monitoring of Highways,” in Smart Review of Vibration-based Damage Identification Methods,” Shock
Systems for Bridges, Structures, and Highways, Newport Beach, and Vibration Digest, Vol. 30, No. 2, 91–105.
CA, March 1–2, Proceedings of the SPIE, Vol. 3671, 173–182. Elgamal, A., Conte, J. P., Masri, S., Fraser, M., Fountain, T., Gupta, A.,
Bensky, A., 2004, Short-Range Wireless Communication, 2nd Ed., Else- Trivedi, M., and El Zarki, M., 2003, “Health Monitoring Framework
vier, Amsterdam. for Bridges and Civil Infrastructure,” in Proceedings of the 4th Inter-
Binns, J., 2004, “Bridge Sensor System Delivers Results Quickly,” Civil national Workshop on Structural Health Monitoring, Stanford, CA,
Engineering, Vol. 74, No. 9, 30–31. September 15–17, 123–130.

Elvin, N. G., Elvin, A. A., and Spector, M., 2001, “A Self-powered International Conference of Building Officials (ICBO), 2002, “2001 Cali-
Mechanical Strain Energy Sensor,” Smart Materials and Structures, fornia Building Code – California Code of Regulations,” Title 24,
Vol. 10, No. 2, 293–299. Part 2, Volume 2, ICBO, Whittier, CA.
Estes, D. R. J., Welch, T. B., Sarkady, A. A. and Whitesel, H., 2001, “Ship- Jennings, N. R., Sycara, K., and Wooldridge, M., 1998, “A Roadmap of
board Radio Frequency Propagation Measurements for Wireless Agent Research and Development,” Autonomous Agents and Multi-
Networks,” in Proceedings of the IEEE Military Communications Agent Systems, Vol. 1, 7–38.
Conference (MILCOM), McLean, VA, Vol. 1, 247–251. Johnson, E. A., Lam, H. F., Katafygiotis, L. S., and Beck, J. L., 2004,
Farhey, D. N., 2003, “Rapid-response Diagnostics and Smart Structure “Phase 1 IASC–ASCE Structural Health Monitoring Benchmark
Monitoring in Bridge Testing,” in Proceedings of the County Engi- Problem Using Simulated Data,” Journal of Engineering Mechanics,
neers Association of Ohio Bridge Conference, Columbus, OH. Vol. 130, No. 1, 3–15.
Farrar, C. R., 2001, “Historical Overview of Structural Health Monitor- Jung, K., Bredow, J. W., and Joshi, S. P., 1999, “Electromagnetically Cou-
ing,” Lecture Notes on Structural Health Monitoring Using Statis- pled Embedded Sensors,” in Industrial and Commercial Applications
tical Pattern Recognition, Los Alamos Dynamics, Los Alamos, of Smart Structures Technologies, Newport Beach, CA, March 2–4,
NM. Proceedings of the SPIE, Vol. 3674, 148–156.
Farrar, C. R., Allen, D. W., Ball, S., Masquelier, M. P., and Park, G., 2005, Karbhari, V. M., Guan, H., and Sikorsky, C., 2003, “Web-based Structural
“Coupling Sensing Hardware with Data Interrogation Software for Health Monitoring of a FRP Composite Bridge,” in Proceedings of
Structural Health Monitoring,” in Proceedings of the 6th Interna- the International Conference on Structural Health Monitoring and
tional Symposium on Dynamic Problems of Mechanics (DINAME), Intelligent Infrastructure, Tokyo, Japan, November 13–15, Vol. 1,
Ouro Preto, Brazil, February 29–March 4. 217–225.
Finkenzeller, K., 1999, RFID Handbook: Radio Frequency Identification Kim, J. S., Vinoy, K. J., and Varadan, V. K., 2002, “Wireless Health Mon-
Fundamentals and Applications, John Wiley and Sons, Chichester, itoring of Cracks in Structures with MEMS–IDT Sensors,” in Smart
UK. Structures and Materials, San Diego, CA, March 18, Proceedings of
Gadre, D. V., 2001, Programming and Customizing the AVR Microcontrol- the SPIE, Vol. 4700, 342–353.
ler, McGraw-Hill, New York. Kling, R. M., 2003, “Intel Mote: An Enhanced Sensor Network Node,” in
Galbreath, J. H., Townsend, C. P., Mundell, S. W., Hamel, M. J., Esser, B., Proceedings of the International Workshop on Advanced Sensors,
Huston, D., and Arms, S. W., 2003, “Civil Structure Strain Monitor- Structural Health Monitoring and Smart Structures, Tokyo, Japan,
ing with Power-efficient, High-speed Wireless Sensor Networks,” in November 10–11.
Proceedings of the 4th International Workshop on Structural Health Kottapalli, V. A., Kiremidjian, A. S., Lynch, J. P., Carryer, E., Kenny, T.
Monitoring, Stanford, CA, September 15–17, 1215–1222. W., Law, K. H., and Lei, Y., 2003, “Two-tiered Wireless Sensor
Gause, L. W., Krantz, D. G., Biermann, P. J., and Belk, J. H., 1999, “Early Network Architecture for Structural Health Monitoring,” in Smart
Demonstration of Remotely Queried Microsensors,” in Smart Elec- Structures and Materials, San Diego, CA, March 3–6, Proceedings
tronics and MEMS, Newport Beach, CA, March 1–3, Proceedings of of the SPIE, Vol. 5057, 8–19.
the SPIE, Vol. 3673, 190–194. Kratzig, W. B., Meyer, I. F., and Meskouris, K., 1989. “Damage Evolution
Glaser, S. D., 2004, “Some Real-world Applications of Wireless Sensor in Reinforced Concrete Members Under Cyclic Loading.” in Pro-
Nodes,” in Smart Structures and Materials: Sensors and Smart Struc- ceedings of the 5th International Conference on Structural Safety
tures Technologies for Civil, Mechanical, and Aerospace Systems, and Reliability, San Francisco, CA, 795–802.
San Diego, CA, March 15–18, Proceedings of the SPIE, Vol. 5391, Kumagai, J., 2004, “Life of Birds – Wireless Sensor Network for Bird
344–355. Study,” IEEE Spectrum, Vol. 41, No. 4, 30–37.
Grisso, B. L., Martin, L. A., and Inman, D. J., 2005, “A Wireless Active Kurata, N., Spencer, B. F. Jr., and Ruiz-Sandoval, M., 2003a, “Risk Moni-
Sensing System for Impedance-based Structural Health Monitoring,” toring of Buildings Using Wireless Sensor Network,” in Proceedings
in Proceedings of the 23rd International Modal Analysis Conference, of the International Workshop on Advanced Sensors, Structural
Orlando, FL, January 31–February 3. Health Monitoring, and Smart Structures, Tokyo, Japan, November
Gu, H., Peng, J., Zhao, Y., Lloyd, G. M., and Wang, M. L., 2004, “Design 10–11.
and Experimental Validation of a Wireless PVDF Displacement Kurata, N., Spencer, B. F. Jr., Ruiz-Sandoval, M., Miyamoto, Y., and Sako,
Sensor for Structure Monitoring,” in Non-destructive Detection and Y., 2003b, “A Study on Building Risk Monitoring Using Wireless
Measurement for Homeland Security II, San Diego, CA, March 16– Sensor Network MICA-Mote,” in Proceedings of the International
17, Proceedings of the SPIE, Vol. 5395, 91–99. Conference on Structural Health Monitoring and Intelligent Infra-
Hamburger, R. O., 2000, “A Policy Guide to Steel Moment-frame Con- structure, Tokyo, Japan, November 13–15, Vol. 1, 353–357.
struction,” Technical Report No. 354, Federal Emergency Manage- Kurata, N., Spencer, B. F. Jr., and Ruiz-Sandoval, M., 2004, “Building
ment Agency, Washington DC. Risk Monitoring Using Wireless Sensor Network,” in Proceedings of
Hayes, J. R., 2001. “Modular Programming in C,” Embedded Systems Pro- the 13th World Conference on Earthquake Engineering, August 2–6,
gramming, Vol. 14, No. 13, 18–24. Vancouver, BC, Canada.
Hill, J., 2000, “A System Architecture Supporting Networked Sensors,” Li, H. W., Zhou, Z., and Ou, J. P. 2003, “Offshore Platform Health Moni-
Master’s Thesis, University of California-Berkeley, Berkeley, CA. toring Applications of Wireless Sensor,” in Proceedings of the 21st
Hill, J. L. and Culler, D. E., 2002, “MICA: A Wireless Platform for Deeply International Modal Analysis Conference (IMAC XXI), Orlando,
Embedded Networks,” IEEE Micro, Vol. 22, No. 6, 12–24. FL, February 3–6.
Hou, T-C. and Lynch, J. P., 2005, “Monitoring Strain in Engineered Liu, S. C. and Tomizuka, M., 2003a, “Strategic Research for Sensors and
Cementitious Composites Using Wireless Sensors,” in Proceedings Smart Structures Technology,” in Proceedings of the International
of the International Conference on Fracture (ICF XI), Turin, Italy, Conference on Structural Health Monitoring and Intelligent Infra-
March 20–25. structure, Tokyo, Japan, November 13–15, Vol. 1, 113–117.
Hou, T-C., Lynch, J. P., and Parra-Montesinos, G., 2005, “In Situ Wireless Liu, S. C. and Tomizuka, M., 2003b, “Vision and Strategy for Sensors and
Monitoring of Fiber Reinforced Cementitious Composite Bridge Smart Structures Technology Research,” in Proceedings of the 4th
Piers,” in Proceedings of the 23rd International Modal Analysis Con- International Workshop on Structural Health Monitoring, Stanford,
ference, Orlando, FL, January 31–February 3. CA, September 15–17, 42–52.
Huston, D., Esser, B., Miller, J., and Wang, X., 2003, “Robotic and Mobile Liu, L., Yuan, F. G., and Zhang, F., 2005, “Development of Wireless Smart
Sensor Systems for Structural Health Monitoring,” in Proceedings of Sensor for Structural Health Monitoring,” in Smart Structures and
the 4th International Workshop on Structural Health Monitoring, Materials, San Diego, CA, March 6–10, Proceedings of the SPIE,
Stanford, CA, September 15–17, 1249–1256. Vol. 5765, No. 1, 176–186.
Ihler, E., Zaglauer, H. W., Herold-Schmidt, U., Dittrich, K. W., and Wies- Lynch, J. P., 2002, “Decentralization of Wireless Monitoring and Control
beck, W., 2000, “Integrated Wireless Piezoelectric Sensors,” in Technologies for Smart Civil Structures,” Ph.D. Thesis, Department
Smart Structures and Materials, Newport Beach, CA, March 7, Pro- of Civil and Environmental Engineering, Stanford University, Stan-
ceedings of the SPIE, Vol. 3991, 44–51. ford, CA.
Institute of Electrical and Electronics Engineers (IEEE), 2003, Part 15.4: Lynch, J. P., 2004, “Overview of Wireless Sensors for Real-time Health
Wireless Medium Access Control (MAC) and Physical Layer (PHY) Monitoring of Civil Structures,” in Proceedings of the 4th Interna-
Specifications for Low-Rate Wireless Personal Area Networks (LR- tional Workshop on Structural Control, New York, NY, June 10–11,
WPANs), IEEE, New York, NY. 189–194.

Lynch, J. P., 2005, “Design of a Wireless Active Sensing Unit for Local- Maser, K., Egri, R., Lichtenstein, A., and Chase, S., 1996, “Field Evaluation
ized Structural Health Monitoring,” Journal of Structural Control of a Wireless Global Bridge Evaluation and Monitoring System,” in
and Health Monitoring, Vol. 12, No. 3–4, 405–423. Proceedings of the 11th Conference on Engineering Mechanics, Fort
Lynch, J. P., Law, K. H., Kiremidjian, A. S., Kenny, T. W., Carryer, E., Lauderdale, FL, May 19–22, Vol. 2, 955–958.
and Partridge, A., 2001, “The Design of a Wireless Sensing Unit for Mastroleon, L., Kiremidjian, A. S., Carryer, E., and Law, K. H., 2004,
Structural Health Monitoring,” in Proceedings of the 3rd Interna- “Design of a New Power-efficient Wireless Sensor System for Struc-
tional Workshop on Structural Health Monitoring, Stanford, CA, tural Health Monitoring,” in Non-destructive Detection and Meas-
September 12–14. urement for Homeland Security II, San Diego, CA, March 16–17,
Lynch, J. P., Law, K. H., Kiremidjian, A. S., Kenny, T. W., and Carryer, Proceedings of the SPIE, Vol. 5395, 51–60.
E., 2002a, “A Wireless Modular Monitoring System for Civil Struc- Maurer, W., 2003, The Scientist and Engineer's Guide to Smart Dust and
tures,” in Proceedings of the 20th International Modal Analysis Con- TinyOS Programming, DSPLab Press, Livermore, CA.
ference (IMAC XX), Los Angeles, CA, February 4–7, 1–6. Meninger, S., Mur-Miranda, J. O., Amirtharajah, R., Chandrakasan, A. P.,
Lynch, J. P., Law, K. H., Kiremidjian, A. S., Carryer, E., Kenny, T. W., and Lang, J. H., 1999, “Vibration-to-electric Energy Conversion,” in
Partridge, A., and Sundararajan, A., 2002b, “Validation of a Wire- Proceedings of the International Symposium on Low Power Electron-
less Modular Monitoring System for Structures,” in Smart Structures ics and Design, Digest of Technical Papers, San Diego, CA, 48–53.
and Materials: Smart Systems for Bridges, Structures, and High- Meninger, S., Mur-Miranda, J. O., Amirtharajah, R., Chandrakasan, A. P.,
ways, San Diego, CA, March 17–21, Proceedings of the SPIE, Vol. and Lang, J. H., 2001, “Vibration-to-electric Energy Conversion,”
4696, No. 2, 17–21. IEEE Transactions on Very Large Scale Integration (VLSI) Systems,
Lynch, J. P., Sundararajan, A., Law, K. H., Kiremidjian, A. S., Carryer, E., Vol. 9, No. 1, 64–76.
Sohn, H., and Farrar, C. R., 2003a, “Field Validation of a Wireless Milos, F. S., Watters, D. G., Pallix, J. B., Bahr, A. J., and Huestis, D. L.,
Structural Health Monitoring System on the Alamosa Canyon 2001, “Wireless Subsurface Microsensors for Health Monitoring of
Bridge,” in Smart Structures and Materials: Smart Systems and Non- Thermal Protection Systems on Hypersonic Vehicles,” in Advanced
destructive Evaluation for Civil Infrastructures, San Diego, CA, Non-destructive Evaluation for Structural and Biological Health
March 3–6, Proceedings of the SPIE, Vol. 5057, 267–278. Monitoring, Newport Beach, CA, March 6, Proceedings of the SPIE,
Lynch, J. P., Sundararajan, A., Law, K. H., Kiremidjian, A. S., and Car- Vol. 4335, 74–82.
ryer, E., 2003b, “Power-efficient Wireless Structural Monitoring Mita, A. and Takahira, S., 2002, “Health Monitoring of Smart Structures
with Local Data Processing,” in Proceedings of the International Using Damage Index Sensors,” in Smart Structures and Materials –
Conference on Structural Health Monitoring and Intelligent Infra- Smart Systems for Bridges, Structures, and Highways, San Diego,
structure, Tokyo, Japan, November 13–15, Vol. 1, 331–338. CA, March 18–20, Proceedings of the SPIE, Vol. 4696, 92–99.
Lynch, J. P., Sundararajan, A., Law, K. H., Kiremidjian, A. S., Kenny, T., Mita, A. and Takahira, S., 2003, “A Smart Sensor Using a Mechanical
and Carryer, E., 2003d, “Embedment of Structural Monitoring Algo- Memory for Structural Health Monitoring of a Damage-controlled
rithms in a Wireless Sensing Unit,” Journal of Structural Engineer- Building,” Smart Materials and Structures, Vol. 12, No. 2, 204–209.
ing and Mechanics, Vol. 15, No. 3, 285–297. Mita, A. and Takahira, S., 2004, “Damage Index Sensor for Smart Struc-
Lynch, J. P., Sundararajan, A., Law, K. H., Kiremidjian, A. S. and Carryer, tures,” Structural Engineering and Mechanics, Vol. 17, No. 3–4,
E., 2003e, “Power-efficient Data Management for a Wireless Struc- 331–346.
tural Monitoring System,” in Proceedings of the 4th International Mitchell, K, Dang, N., Liu, P., Rao, V. S., and Pottinger, H. J., 2001,
Workshop on Structural Health Monitoring, Stanford, CA, Septem- “Web-controlled Wireless Network Sensors for Structural Health
ber 15–17, 1177–1184. Monitoring,” in Smart Structures and Materials – Smart Electronics
Lynch, J. P., Sundararajan, A., Law, K. H., Sohn, H., and Farrar, C. R, and MEMS, Newport Beach, CA, March 5–7, Proceedings of the
2003f, “Design and Performance Validation of a Wireless Active SPIE, Vol. 4334, 234–243.
Sensing Unit,” in Proceedings of the International Workshop on Mitchell, K., Rao, V. S., and Pottinger, H. J., 2002, “Lessons Learned
Advanced Sensors, Structural Health Monitoring, and Smart Struc- About Wireless Technologies for Data Acquisition,” in Smart Struc-
tures, Tokyo, Japan, November 10–11. tures and Materials 2002: Smart Electronics, MEMS, and Nanotech-
Lynch, J. P., Sundararajan, A., Law, K. H., Carryer, E., Farrar, C. R., Sohn, nology, San Diego, CA, March 18–21, Proceedings of the SPIE,
H., Allen, D. W., Nadler, B., and Wait, J. R., 2004a, “Design and Vol. 4700, 331–341.
Performance Validation of a Wireless Sensing Unit for Structural Mittag, L., 2001, “Magic in the Air,” Embedded Systems Programming,
Health Monitoring Applications,” Structural Engineering and Vol. 14, No. 10, 49–60.
Mechanics, Vol. 17, No. 3, 393–408. Mokole, E. L., Parent, M., Samaddar, S. N., Tomas, E. and Gold, B. T.,
Lynch, J. P., Sundararajan, A., Sohn, H., Park, G., Farrar, C., and Law, K., 2000, “Radio-frequency Propagation Measurements in Confined
2004b, “Embedding Actuation Functionalities in a Wireless Struc- Ship Spaces Aboard the ex-USS Shadwell,” Report Number
tural Health Monitoring System,” in Proceedings of the International A987183, Naval Research Laboratory, Washington DC.
Workshop on Smart Materials and Structures Technology, Honolulu, Morton, T. D., 2001, Embedded Microcontrollers, Prentice-Hall, Upper
HI, January 12–14. Saddle River, NJ.
Lynch, J. P., Sundararajan, A., Law, K. H., Sohn, H., and Farrar, C. R., Mufti, A. A., 2003, “Integration of Sensing in Civil Structures: Devel-
2004c, “Design of a Wireless Active Sensing Unit for Structural opment of the New Discipline of Civionics,” in Proceedings of
Health Monitoring,” in Proceedings of the 11th Annual International the International Conference on Structural Health Monitoring and
Symposium on Smart Structures and Materials, San Diego, CA, Intelligent Infrastructure, Tokyo, Japan, November 13–15, Vol. 1,
March 14–18, Proceedings of the SPIE, Vol. 5394, 157–169. 119–129.
Lynch, J. P., Sundararajan, A., Law, K. H., Kiremidjian, A. S., and Car- Nagayama, T., Ruiz-Sandoval, M., Spencer, B. F. Jr., Mechitov, K. A., and
ryer, E., 2004d, “Embedding Damage Detection Algorithms in a Agha, G., 2004, “Wireless Strain Sensor Development for Civil
Wireless Sensing Unit for Operational Power Efficiency,” Smart Infrastructure,” in Proceedings of the 1st International Workshop on
Materials and Structures, Vol. 13, No. 4, 800–810. Networked Sensing Systems, Tokyo, Japan, June 22–23.
Lynch, J. P., Parra-Montesinos, G., Canbolat, B. A., and Hou, T-C., 2004e, Nejikovsky, B. and Keller, E., 2000, “Wireless Communications Based
“Real-time Damage Prognosis of High-performance Fiber Rein- System to Monitor Performance of Rail Vehicles,” in Proceedings of
forced Cementitious Composite Structures,” in Proceedings of the IEEE/ASME Joint Railroad Conference, Newark, NJ, April 4–6,
Advances in Structural Engineering and Mechanics (ASEM’04), 111–124.
Seoul, Korea, September 2–4. Novak, L. J., Grizzle, K. M., Wood, S. L., and Neikirk, D. P., 2003,
Lynch, J. P., Wang, Y., Law, K. H., Yi, J-H., Lee, C. G., and Yun, C. B., “Development of State Sensors for Civil Engineering Structures,” in
2005, “Validation of a Large-scale Wireless Structural Monitoring Smart Structures and Materials: Smart Systems and Non-destructive
System on the Geumdang Bridge,” in Proceedings of the Interna- Evaluation for Civil Infrastructures, San Diego, CA, March 3–6,
tional Conference on Safety and Structural Reliability (ICOSSAR), Proceedings of the SPIE, Vol. 5057, 358–363.
Rome, Italy, June 19–23. Oshima, T., Rahman, M. S., Mikami, S., Yamazaki, T., Takada, N., Lesko,
MacGillivray, P. and Goddard, K., 1997, “Advanced Sensor Technology J. J., and Kriz, R. D., 2000, “Application of Smart Materials and Sys-
for Marine Propulsion Control Systems,” in Proceedings of the 11th tems to Long-term Bridge Health Monitoring,” in Non-destructive
Ship Control Systems Symposium, Southampton, UK, April, Vol. 2, Evaluation of Highways, Utilities, and Pipelines, Newport Beach,
245–257. CA, March 7–9, Proceedings of the SPIE, Vol. 3995, 253–263.

Ou, J-P. and Li, H-W., 2003, “Wireless Sensors Information Fusion for Seth, S., Lynch, J. P., and Tillbury, D. M., 2005, “Wirelessly Networked
Structural Health Monitoring,” in Multisensor, Multisource Informa- Distributed Controllers for Real-time Control of Civil Structures,” in
tion Fusion: Architectures, Algorithms, and Applications, Orlando, Proceedings of the American Controls Conference (ACC05), Port-
FL, April 23–25, Proceedings of the SPIE, Vol. 5099, 356–362. land, OR, June 8–10.
Ou, J., Li, H., and Yu, 2004, “Development and Performance of Wireless Shinozuka, M., 2003, “Homeland Security and Safety,” in Proceedings of
Sensor Network for Structural Health Monitoring,” in Smart Struc- the International Conference on Structural Health Monitoring and
tures and Materials, San Diego, CA, March 15–18, Proceedings of Intelligent Infrastructure, Tokyo, Japan, November 13–15, Vol. 2,
the SPIE, Vol. 5391, 765–773. 1139–1145.
Ou, J. P., Li, H. W., Xiao, Y. Q., and Li, Q. S., 2005, “Health Dynamic Simonen, J. T., Andringa, M. M., Grizzle, K. M., Wood, S. L., and Neikirk,
Measurement of Tall Building Using Wireless Sensor Network,” in D. P., 2004, “Wireless Sensors for Monitoring Corrosion in Rein-
Smart Structures and Materials, San Diego, CA, March 6–10, Pro- forced Concrete Members,” in Smart Structures and Materials: Sen-
ceedings of the SPIE, Vol. 5765, No. 1, 205–215. sors and Smart Structures Technologies for Civil, Mechanical, and
Pakzad, S. N. and Fenves, G. L., 2004, “Structural Health Monitoring Aerospace Systems, San Diego, CA, March 15–18, Proceedings of
Applications Using MEMS Sensor Networks,” in Proceedings of the SPIE, Vol. 5391, 587–596.
4th International Workshop on Structural Control, New York, NY, Sohn, H. and Farrar, C., 2001, “Damage Diagnosis Using Time-series
June 10–11, 47–56. Analysis of Vibrating Signals,” Smart Materials and Structures, Vol.
Park, G., Cudney, H. H., and Inman, D. J., 2000, “Impedance-based Health 10, No. 3, 446–451.
Monitoring of Civil Structural Components,” Journal of Infrastruc- Sohn, H., Farrar, C. R., Hemez, F. M., Czarnecki, J. J., Shunk, D. D., Stin-
ture Systems, Vol. 6, No. 4, 153–160. emates, D. W., and Nadler, B. R., 2004, “A Review of Structural
Park, G., Sohn, H., Farrar, C. R., and Inman, D. J., 2003, “Overview of Pie- Health Monitoring Literature: 1996–2001,” Report Number LA-
zoelectric Impedance-based Health Monitoring and Path Forward,” 13976-MS, Los Alamos National Laboratory, Los Alamos, NM.
Shock and Vibration Digest, Vol. 35, No. 6, 451–463. Sodano, H. A., Park, G., Leo, D. J., and Inman, D. J., 2003, “Use of Piezo-
Partridge, A., Reynolds, J. K., Chui, B. W., Chow, E. M., Fitzgerald, A. electric Energy Harvesting Devices for Charging Batteries,” in
M., Zhang, L., Maluf, N. I., and Kenny, T. W., 2000, “A High-per- Smart Structures and Materials: Smart Sensor Technology and
formance Planar Piezoresistive Accelerometer,” Journal of Microe- Measurement Systems, San Diego, CA, March 3–6, Proceedings of
lectromechanical Systems, Vol. 9, No. 1, 58–66. the SPIE, Vol. 5050, 101–108.
Pei, J-S., Kapoor, C., Graves-Abe, T. L., Sugeng, Y., and Lynch, J. P., Sodano, H. A., Inman, D. J., and Park, G., 2004, “A Review of Power Har-
2005, “Critical Design Parameters and Operating Conditions of vesting from Vibration using Piezoelectric Materials,” Shock and
Wireless Sensor Units for Structural Health Monitoring,” in Pro- Vibration Digest, Vol. 36, No. 3, 197–205.
ceedings of the 23rd International Modal Analysis Conference Spencer, B. F. Jr., 2003, “Opportunities and Challenges for Smart Sensing
(IMAC XXIII), Orlando, FL, January 31–February 3. Technology,” in Proceedings of the International Conference on
Pines, D. J. and Lovell, P. A., 1998, “Conceptual Framework of a Remote Structural Health Monitoring and Intelligent Infrastructure, Tokyo,
Wireless Health Monitoring System for Large Civil Structures,” Japan, November 13–15, Vol. 1, 65–71.
Smart Materials and Structures, Vol. 7, No. 5, 627–636. Spencer, B. F. Jr., Ruiz-Sandoval, M. E., and Kurata, N., 2004, “Smart
Ploeger, R., Newton, W. W., Rabiner, A., and Lally, R., 2003, “Wireless Sensing Technology: Opportunities and Challenges,” Journal of
e-diagnostics Reduces Workload and Improves Shipboard Quality Structural Control and Health Monitoring, Vol. 11, No. 4, 349–
of Life,” White Paper, Oceana Sensor Technologies, Virginia Beach, 368.
VA. Straser, E. G. and Kiremidjian, A. S., 1998, “A Modular, Wireless Damage
Rappaport, T. S., 2002, Wireless Communications: Principles and Prac- Monitoring System for Structures,” Technical Report 128, John A.
tice, Prentice-Hall, Englewood Cliffs, NJ. Blume Earthquake Engineering Center, Stanford University, Stan-
Ruiz-Sandoval, M. E., 2004, “ ‘Smart’ Sensors for Civil Infrastructure ford, CA.
Systems,” Doctor of Philosophy Thesis, Department of Civil Engi- Takahashi, D., 2004, “Dusting for Data,” TheMercuryNews.com. http://
neering and Geological Sciences, University of Notre Dame, Notre www.mercurynews.com/mld/mercurynews/business/technology/
Dame, IN. 9770734.htm
Ruiz-Sandoval, M., Spencer, B. F. Jr., and Kurata, N., 2003, “Development Tan, G. H., Ng, T. G., and Brownjohn, J. M. W., 2004, “Real-time
of a High Sensitivity Accelerometer for the Mica Platform,” in Pro- Monitoring and Alert Systems for Civil Engineering Applications
ceedings of the 4th International Workshop on Structural Health Using Machine-to-machine Technologies,” in Proceedings of the
Monitoring, Stanford, CA, September 15–17, 1027–1034. International Conference on Structural and Foundation Failures, Sin-
Saafi, M. and Romine, P., 2004, “Embedded MEMS for Health Monitoring gapore.
and Management of Civil Infrastructure,” in Smart Structures and Tanner, N. A., Farrar, C. R., and Sohn, H., 2002, “Structural Health Moni-
Materials: Sensors and Smart Structures Technologies for Civil, toring Using Wireless Sensing Systems with Embedded Processing,”
Mechanical, and Aerospace Systems, San Diego, CA, March 15–18, in Non-destructive Evaluation and Health Monitoring of Aerospace
Proceedings of the SPIE, Vol. 5391, 331–343. Materials and Civil Infrastructure, Newport Beach, CA, March 18–
Salomon, D., 2004, Data Compression: The Complete Reference, Springer, 19, Proceedings of the SPIE, Vol. 4704, 215–224.
New York. Tanner, N. A., Wait, J. R., Farrar, C. R., and Sohn, H., 2003, “Structural
Sazonov, E., Janoyan, K., and Jha, R., 2004, “Wireless Intelligent Sensor Health Monitoring Using Modular Wireless Sensors,” Journal of
Network for Autonomous Structural Health Monitoring,” in Smart Intelligent Material Systems and Structures, Vol. 14, No. 1, 43–56.
Structures and Materials: Smart Sensor Technology and Measure- Townsend, C. P., Hamel, M. J., Sonntag, P, Trutor, B., and Arms, S. W.,
ment Systems, San Diego, CA, March 15–17, Proceedings of the 2002, “Scaleable Wireless Web-enabled Sensor Networks,” in Smart
SPIE, Vol. 5384, 305–314. Systems for Bridges, Structures, and Highways, San Diego, CA,
Schwartz, G., 2002, “Reliability and Survivability in the Reduced Ship’s March 18–20, Proceedings of the SPIE, Vol. 4696, 1–9.
Crew by Virtual Presence System,” in Proceedings of the IEEE Varadan, V. K. and Varadan, V. V., 2000, “Microsensors, Microelectro-
International Conference on Dependable Systems and Networks mechanical Systems (MEMS), and Electronics for Smart Structures
(DSN’02), Washington, DC, June 23–26, 199–204. and Systems,” Smart Materials and Structures, Vol. 9, No. 6, 953–
Seidel, S. Y. and Rappaport, T. S., 1992, “914 MHz Path Loss Prediction 972.
Models for Indoor Wireless Communications in Multifloored Build- Walsh, S. M., Butler, J. C., Belk, J. H., and Lawler, R. A., 2001, “Develop-
ings,” IEEE Transactions on Antennas and Propagation, Vol. 40, ment of a Structurally Compatible Sensor Element,” in Advanced
No. 2, 207–217. Non-destructive Evaluation for Structural and Biological Health
Sereci, A. M., Radulescu, D., and Radulescu, C., 2003, “Recent Installa- Monitoring, Newport Beach, CA, March 6–8, Proceedings of the
tions of Real-time Structural Monitoring Systems,” in Proceedings SPIE, Vol. 4335, 63–73.
of the 4th International Workshop on Structural Health Monitoring, Wang, M. L., Gu, H., Lloyd, G. M., and Zhao, Y., 2003a, “A Multichannel
Stanford, CA, September 15–17, 223–230. Wireless PVDF Displacement Sensor for Structural Monitoring,” in
Seth, S., Lynch, J. P., and Tilbury, D. M., 2004, “Feasibility of Real- Proceedings of the International Workshop on Advanced Sensors,
time Distributed Structural Control upon a Wireless Sensor Net- Structural Health Monitoring and Smart Structures, Tokyo, Japan,
work,” in Proceedings of the 42nd Annual Allerton Conference on November 10–11.
Communication, Control and Computing, Allerton, IL, September Wang, M., Chang, P. C., and Newcomb, R., 2003b, “Power Scavenging
29–October 1. from Highway Bridge Vibration,” in Proceedings of the International

Conference on Structural Health Monitoring and Intelligent Infra- Structures and Materials, San Diego, CA, March 3–6, Proceedings
structure, Tokyo, Japan, November 13–15, Vol. 2, 1117–1126. of the SPIE, Vol. 5057, 20–28.
Wang, X., Wang, M. L., Zhao, Y., Chen, H., and Zhou, L. L., 2004, “Smart Wong, J-M., Goethals, J., and Stojadinovic, B., 2005, “Wireless Sensor
Health Monitoring System for a Prestressed Concrete Bridge,” in Seismic Response Monitoring System Implemented on Top of
Smart Structures and Materials: Sensors and Smart Structures NEESgrid,” in Health Monitoring and Smart Non-destructive Evalu-
Technologies for Civil, Mechanical, and Aerospace System, San ation of Structural and Biological Systems IV, San Diego, CA,
Diego, CA, March 15–18, Proceedings of the SPIE, Vol. 5391, March 6–10, Proceedings of the SPIE, Vol. 5768, 74–84.
597–608. Wooldridge, M., Jennings, N. R., and Kinny, D., 2000, “The Gaia Method-
Wang, Y., Lynch, J. P., and Law, K. H., 2005, “Wireless Structural Sen- ology for Agent-oriented Analysis and Design,” Journal for Autono-
sors Using Reliable Communication Protocols for Data Acquisition mous Agents and Multi-Agent Systems, Vol. 3, No. 3, 285–312.
and Interrogation,” in Proceedings of the 23rd International Modal Wu, F. and Chang, F-K., 2001, “A Built-in Active Sensing Diagnostic Sys-
Analysis Conference (IMAC XXIII), Orlando, FL, January 31– tem for Civil Infrastructure Systems,” in Smart Structures and Mate-
February 3. rials, San Diego, CA, March 5–7, Proceedings of the SPIE, Vol.
Watters, D. G., Jayaweera, P., Bahr, A. J., Huestis, D. L., Priyantha, N., 4330, 27–35.
Meline, R., Reis, R., and Parks, D., 2003, “Smart Pebble: Wireless Zhao, F. and Guibas, L., 2004, Wireless Sensor Networks: An Information
Sensors for Structural Health Monitoring of Bridge Decks,” in Smart Processing Approach, Morgan Kaufman, San Francisco, CA.


10 1 1 126 6482 PDF

Uploaded by

Copyright:

Available Formats

10 1 1 126 6482 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

10 1 1 126 6482 PDF

Uploaded by

Copyright:

Available Formats

The Shock and Vibration Digest

The online version of this article can be found at:

Email Alerts: http://svd.sagepub.com/cgi/alerts

Citations (this article cites 65 articles hosted on the

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

A Summary Review of Wireless Sensors and Sensor

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Figure 1. Functional elements of a wireless sensor for structural monitoring applications.

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Table 1. Summary of academic wireless sensing unit prototypes (1998–2003).

DATA ACQUISITION SPECIFICATIONS

EMBEDDED COMPUTING SPECIFICATIONS

WIRELESS CHANNEL SPECIFICATIONS

FINAL ASSEMBLED UNIT ATTRIBUTES

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Table 2. Summary of academic wireless sensing unit prototypes (2003–2005).

DATA ACQUISITION SPECIFICATIONS

EMBEDDED COMPUTING SPECIFICATIONS

WIRELESS CHANNEL SPECIFICATIONS

FINAL ASSEMBLED UNIT ATTRIBUTES

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Table 3. Summary of commercial wireless sensing unit prototypes.

DATA ACQUISITION SPECIFICATIONS

EMBEDDED COMPUTING SPECIFICATIONS

WIRELESS CHANNEL SPECIFICATIONS

FINAL ASSEMBLED UNIT ATTRIBUTES

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Table 4. Summary of academic active wireless sensing unit prototypes.

DATA ACQUISITION SPECIFICATIONS

EMBEDDED COMPUTING SPECIFICATIONS

WIRELESS CHANNEL SPECIFICATIONS

FINAL ASSEMBLED UNIT ATTRIBUTES

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008

Downloaded from http://svd.sagepub.com at PENNSYLVANIA STATE UNIV on April 17, 2008