A Review of Passive Wireless Sensors for Structural Health Monitoring

Wireless sensors for Structural Health Monitoring (SHM) is an emerging new technology that promises to overcome many disadvantages pertinent to conventional, wired sensors. The broad field of SHM has experienced significant growth over the past two decades, with several notable developments in the area of sensors such as piezoelectric sensors and optical fibre sensors. Although significant improvements have been made on damage monitoring techniques using these smart sensors, wiring remains a significant challenge to the practical implementation of these technologies. Wireless SHM has recently attracted the attention of researchers towards un-powered and more effective passive wireless sensors. This article presents a review of some of the underlying technologies in the field of wireless sensors for SHM - with a focus on the research progress towards the development of simple, powerless, yet effective and robust wireless damage detection sensors. This review examines the development of passive wireless sensors in two different categories: (1) use of oscillating circuits with the help of inductors, capacitors and resistors for damage detection; and (2) use of antennas, Radio Frequency Identification (RFID) tags and metamaterial resonators as strain sensors for wireless damage monitoring. An assessment of these electromagnetic techniques is presented and the key issues involved in their respective design configurations are discussed


Introduction
Over the past few decades, the field of Structural Health Monitoring (SHM) has attracted considerable research.Several effective damage monitoring techniques like strain measurement, electro-mechanical impedance, scattering of guided waves, acoustic emissions, dynamic response and optical techniques have been developed.Several sensors like strain gauges, piezoelectric sensors and optical fibre sensors have been employed (Housner et al., 1997;Chang, 2002;Auweraer & Peeters, 2003;Chang, Flatau, & Liu, 2003;Wang & Rose, 2003;Wang & Chang, 2005).
The above mentioned sensors are extensively used for damage monitoring; however, they do present certain limitations.Most of the existing sensors require an input power supply.When the sensors detect any change in strain or stress, they need to transfer the information, for signal processing and analysis, to the data acquisition system which may be located at the base station far away from the structure being monitored.The necessary connection of sensors by wires for power and data transmission often renders the SHM system complex to implement and difficult to maintain.The whole structure sometimes needs to be redesigned to accommodate the connections among these sensor networks; therefore, increasing the cost of manufacture.The technical difficulties of designing sensor systems along with their connections become more pronounced when the structure under investigation contains moving parts, such as a helicopter rotor.Furthermore, wiring between sensors and base-stations increases the cost of replacing damaged or degraded sensors.
In order to tackle these challenges, researchers have started investigating options which could result in wireless SHM.By making the sensors wireless through the incorporation of energy coupling and communication functionalities, it is possible to integrate the data acquisition and signal processing system in the same sensor unit.
Where f is the natural frequency; L is the inductance; and C is the capacitance.The inductance of the solenoid can be calculated by, Where k is the form factor; µ is the permeability; N is the number of turns; A is the cross sectional area; and l 0 is the solenoid height.
By applying strain the cross sectional area of the solenoid changes and from Equation 3 its resonant frequency changes accordingly.A high frequency oscillator was used to measure the resonant frequency of the solenoid.
The dip in the RF power is measured to find the frequency of the sensor and thus the applied strain.The experimental setup of this concept is illustrated in Figures 2 and 3.
Figure 2. Schematic of guillotine compressing non-embedded sensor coil (Butler et al., 2002) Figure 3. Illustration of experimental setup (Butler et al., 2002) Chuang, Thomson, andBridges (2005) developed an embeddable wireless strain sensor which works with the same principle as the previous case.The difference is that instead of a solenoid a coaxial resonant RF cavity was used as the sensor (Figure 4).The cavity length changes under the applied load thereby changing its resonant frequency.The shift in the resonant frequency with respect to the applied strain is shown in Figure 5.This sensor was shown to be linear up to 130 με.The shift in the frequency was about 2.42 kHz per με.The relationship between the applied strain and the resonant frequency of the cavity was derived using the following equations.

() 2(
) 2 1 Figure 4. Electromagnetic coaxial cavity sensor.The dominant TEM001 resonant mode is shown (Chuang et al., 2005) www.ccsen   10.The biasing and sensing elements of the sensor are separated by a flexible layer to provide proper strain for a given compressive force (Tan et al., 2008) In this section, several wireless strain sensing techniques utilising LC or RC oscillating circuits are discussed.The summary of the most important techniques are presented in Table 1.For these circuits to resonate at a specific frequency, power has to be supplied to the circuits.Some of these techniques employed external frequency oscillators and hand interrogators to provide power to the LC/RC circuits wirelessly.Thus, it is important to ensure an efficient energy coupling between the transmitter and the receiver.In SHM in order to determine and predict the crack propagation, it is essential to measure the strain in the structural member and to determine its spatial distribution.The work reviewed in this section does not refer primarily to the strain spatial distribution which is a major concern for researchers.Although these techniques are shown to be linear, it is important to ensure an acceptable reliability and repeatability for practical applications.These limitations provided the direction for further research, leading to the development of techniques discussed in the next section.
Table 1.Various techniques employed using oscillator circuits for wireless strain measurement Authors Technique Butler et al., 2002 The strain changes the dimensions of the inductor and hence the inductance in the LC circuit, thus changing the resonant frequency.

Chuang et al., 2005
The strain changes the cavity length of the coaxial RF cavity thereby changing the resonant frequency.

Umbrecht et al., 2005
The strain moves the incompressible liquid through the capillary which is wirelessly read using ultrasound imaging technique.

Matsuzaki & Todoroki, 2005
The strain changes the capacitance of the RC oscillating circuit and hence changes the resonant frequency.

Jia et al., 2006a
The strain changes the capacitance of the interdigital capacitor coupled with the spiral inductor thereby changing the frequency of the LC circuit.

Tan et al., 2008
The strain deforms the flexible layer between the magnetically soft material and the permanent magnet hence changing the harmonic spectrum.

Use of Resonators and Antennas as Strain Sensors
The works discussed in the previous section are fairly simple and effective; however, researchers started to develop sensors which could directly convert strain into frequency shifts that could be read wirelessly.Antennas and electromagnetic resonators are passive devices which could be illuminated by incident electromagnetic waves and the backscattered signals could be received wirelessly using other antennas.Researchers have further tried to determine the direction of the strain induced in the structure.However, at present, it seems the work in this field is just starting.This section presents the recent techniques employed for wireless strain and damage monitoring which utilise resonators or antennas as strain sensors.
Das, Khorrami, and Nourbakhsh (1998) designed a novel sensor/actuator system which utilizes a patch antenna with a multilayer substrate (Figure 11).The multilayer consists of a dielectric layer and a piezoelectric layer.The piezoelectric layer is the sensing unit which converts the measured strain/vibration into voltage.The antenna receives wireless EM signal from the base station and generates a voltage which gets added up to the piezoelectric voltage and this modulated signal is transmitted back to the base station.This antenna can also be used for actuating the piezoelectric layer by supplying the required voltage to the piezoelectric by receiving wireless electromagnetic power.Das et al. (1998) developed a dielectric-piezoelectric grating technique to distinguish sensing and actuating activities.Due to this grating technique, the sensing and actuating functions are activated separately using orthogonal polarization orientation techniques.It is also feasible to stack such microstrip patch antennas with dimensions to operate at different frequencies.This sensor integrates wireless power reception, sensing and data communication in one simple unit.However, this sensor could function well only when it is interrogated from a very close distance.(SWNT) films.These films could act as a strain or a pH sensor because their capacitance or resistance changes accordingly.When these films are integrated with a coil antenna, they could be inductively coupled using a RFID reader and thus rendering the sensor completely wireless.Because these sensors act as a RLC oscillating circuit, the resonant frequency changes with the change in mechanical behaviour of the structure.Use of conducting carbon nanotube-gold nanocomposites as an inductor for wireless coupling was also investigated.However, the inductance was shown to be low, thereby limiting its wireless range to a very small distance.The size of the film sensor is 2.5 cm × 2.5 cm and is stated to be sensitive and linear.Although this technique might be useful, it is believed that the manufacturing of such film nanocomposites could be expensive.
Matsuzaki, Melnykowycz, and Todoroki (2009) developed a very innovative technique for wireless detection of damage in CFRP.The CFRP structure (e.g. the wing structure) can be modeled as a half-wavelength dipole antenna (Figures 12 and 13).The resonant frequency of the antenna is dependent on the length of the structure.
When there is a crack perpendicular to the fibre direction, the dipole length decreases and hence the resonant frequency increases.Therefore, by measuring the frequency, the length of the dipole could be back calculated.
With the length value, the crack location could be precisely identified.This method can only be used in structures with a specific geometry and it can only detect the crack when the crack reaches its critical length.
Figure 12.Simulation model of a rectangular dipole antenna (Matsuzaki et al., 2009) Figure 13.Schematic of the wireless crack detection mechanism (Matsuzaki et al., 2009) Bhattacharyya, Floerkemeier, and Sarma (2009) investigated a RFID tag antenna sensor for displacement measurement (Figure 14).A simple RFID tag was kept at a very close distance to a metal surface which was attached to the structure.As the structure deforms, the metal surface comes closer to the RFID tag which affects the antenna's impedance and hence changes the backscattered power.It also affects the threshold power required to turn the RFID tag 'on'.This RFID tag can be queried wirelessly from a convenient location using an RFID tag reader/transmitter.By processing the backscattering from the RFID tag, the displacement of the structure could be evaluated.Although this sensor is very cheap and simple to design, there are certain challenges associated with this design.Obtaining the displacement data from backscattering becomes difficult if there are other metallic elements in the host structure.Due to the randomly moving metallic components, the sensor might give false positive results.Moreover, the sensor is sensitive to the displacements of the structure only in one direction.However, it is mentioned that this sensing technique could be optimized and utilized for an effective passive wireless displacement sensing system.This RFID tag requires an IC chip which increases the complexity of the structure.The shift in the power level as a result of the applied strain cannot be distinguished from the shift caused by other parameters that may influence the power transmitted (i.e.propagation path-loss, reflection, diffraction etc.).Caizzone and Marrocco (2012) further studied the application of this sensor in a RFID network grid to monitor the deformation of the structure.Their study shows that increasing the number of RFID tags does not further improve the sensitivity of the grid when the spacing between the sensors becomes lower than an optimum threshold.
Mandel, Schussler, and Jakoby (2011) proposed another concept for wireless passive strain measurement based on RFID tags principles.The proposed structure is composed of two layers of metal divided by a dielectric layer.
The two metal layers are connected through the dielectric using an interconnecting via.This "mushroom structure" can be considered as a special case of a short-circuited microstrip patch antenna.One sensor structure comprised different elements, which were separated by gaps in the top metal layer and the substrate.The resonance frequency of each element is determined by the gap capacitance and the via inductance.Different fixed resonant frequencies can be used for identification purposes.The performance of this sensor was investigated using numerical simulations and experimental measurements.However, the linearity of sensor with applied strain is not discussed and further studies are required to quantify its performance.
Another recent study on RFID tags for passive wireless strain measurement was presented by Yi et al. (2011a).
The sensor consists of a folded rectangular microstrip patch antenna with an IC chip.This passive wireless strain sensor operates based on a change in the impedance of the patch antenna as a result of the applied strain, which introduces a mismatch between the antenna and the IC package.When the EM power is sent wirelessly from a remote interrogator, the patch antenna receives the power and transfers it to the chip.This transfer from the tag to the chip is maxima when the interrogation frequency matches with the resonance frequency of the patch antenna.The effect of changes in the impedance of the microstrip line between the rectangular patch and the IC was not considered in the model.The strain measurement was based on the change in the transmitted power which could also be affected by other factors.Due to the lack of sharpness in the transmitted power trace the exact resonant frequency of the RFID tag was difficult to ascertain.Hence, the resonant frequency of the tag was extracted using curve fitting techniques for the wireless strain measurements.Effect of antenna substrate thickness on the interrogation distance and strain transfer rate is further studied by Yi et al. (2011b).
Further investigations by Yi et al. (2012a) show that the shift in the resonant frequency of this sensor can be identified without the need for curve fitting for higher strain values (>4997 με).Also, this sensor can be used for monitoring crack growth when the crack propagates in one direction through the sensor.Yi et al. (2012b) implemented this sensor in an array to measure strain in different locations wirelessly.It is shown that, using the RFID tag protocols, different sensors in the array can be activated individually with minimum interference with other neighbouring sensors.
Figure 15.The RFID folded patch antenna strain sensor (Yi et al., 2012a) Melik, Pergoz, Unal, Puttlitz, and Demir ( 2008) developed a passive on-chip RF-MEMS strain sensor for bio medical applications.As the material is stressed, the area of the sensor (spiral resonator) decreases thus its resonant frequency shifts.To make it completely wireless, two antennas of the same configuration were used for receiving and transmitting the signals.The micrograph of the fabricated sensor system is shown in Figure 16.This system is very small in size and has a very high quality factor.(Melik et al., 2008) because they have more gaps between the rings.These gaps reduce during compression and increase during tension both changing the capacitance and thus shifting the resonant frequency.This sensor was designed to be used for monitoring fracture healing and other biomedical applications.Figure 17 shows the shift in the resonant frequency of the SRR sensor under compressive strain.) 4 Where c is the speed of light; f is the frequency; D sd and D ds are insertion losses due to the switch; |S 11 |is the scattering parameter; P t is the transmitted power; NF is the noise factor; SNR is the signal to noise ratio; and G h and G s are gains of the horn and antenna sensor, respectively.This formula gives a good idea about the factors that affect the wireless interrogation range.The wireless range could be increased by increasing the transmitted power, antennas' gain, and reducing the SNR.It is mentioned that by increasing the interrogation power to 30 dBm and gain of the horn antenna to 20 dB, the interrogation distance could be up to 3.5 m.Deshmukh et al. (2009), Erdmann, Deshmukh, andHuang (2010) and Mohammad and Huang (2010) tried to employ the rectangular patch antenna for sensing of fatigue crack growth.The patch antenna was placed on the crack in a lap joint structure and the shift in the resonant frequency with respect to crack growth was investigated (Figure 24).The effect of the resonant frequency due to the plate on top of the antenna sensor was also studied.
Figure 24.Shift of |S 11 | curves with crack growth (Erdmann et al., 2010) The concept of crack detection using microstrip patch antennas was followed by Mohammad, Gowda, Zhai, and Huang (2012).Numerical simulations and experimental measurements were employed to detect the direction of the crack in the structure in addition of its presence.It was shown that a normalized frequency ratio (ratio of    None of the above work on microstrip patch antenna strain sensors provided the answer for a passive wireless strain sensor which does not require additional integrated circuit units for wireless interrogation of the sensor.In the work by Deshmukh et al. (2009) and Tata et al. (2009a) the wireless interrogation of the rectangular patch antenna was achieved using an additional switch which requires wires or batteries.In the study of Deshmukh (2010) this switch was activated using a photo cell which further increases the complexity of the sensor unit and also requires a direct light source to activate the switch.In these works the resonant frequency of the patch antenna was read wirelessly; however, the strain was not measured.In other works on microstrip patch antennas a SMA connector was used to read the resonant frequency of the antenna sensor.Daliri et al. (2012b) developed a method for wireless interrogation of CMPA strain sensors without the need for additional circuit elements.The CMPA was excited using a linearly polarised double ridged horn antenna to read its resonant frequency.This concept was studied using computational simulations and experimental measurements.The strain in aluminium and CFRP panels was measured wirelessly using this technique.However, the interrogation distance was limited to 5 cm.This technique also enables measuring strain in any desired direction because the linear horn antenna excites the CMPA in the direction of its polarization.By rotating the horn antenna the strain can be measured in the corresponding direction.Daliri et al. (2012c) further increased the interrogation distance of the CMPA sensor up to 20 cm by using a high quality factor CMPA.The high quality factor CMPA was developed using a substrate with low loss and high permittivity.
Regardless of the novelty and promising future of the discussed antenna sensor structures, there are certain key issues which have to be addressed to make these aforementioned works suitable for practical applications.Metamaterial sensors are very small and sensitive; however, they have not been employed to detect cracks till date and have practical limitations for monitoring strain in metallic structures.Moreover, there is not much information about the distance up to which these sensors could be wirelessly read.The work of Matsuzaki et al. (2009) involves the use of CFRP due to its good electrical conductivity.This technique cannot be extended to GFRP and other non-conducting structures.
Microstrip patch antenna sensors have been very useful in predicting the strain along with its direction.However, it is essential to design the antenna to be very sensitive to strain such that the shift in the frequency is more detectable.The wireless interrogation distance of these sensors needs to be improved significantly in order to make these sensors suitable for practical applications.Once these issues are addressed, these sensors could be installed in an array on the host structure for wireless SHM.Table 2 summarizes the important techniques that are provided in this section.Use of split rings, spirals and metamaterial based strain sensors whose resonant frequency changes due to the change in their dimensions.Matsuzaki et al., 2009 Use of carbon fibres in the structure as a dipole antenna.The crack reduces the length of the dipole thereby shifting its frequency.

Ability to detect strain direction
Could be designed to detect strain in any direction.
Generally unidirectional.Could be designed to detect strain in any direction.

Wireless range
Can transmit signals over long distances.
Can transmit signals and receive power effectively only over few centimetres.

Design complexity
Highly complex due to integration of various components.
Complex due to presence of inductor and capacitor integration.Simple design due to the presence of only the antenna.

Size and weight
Large in size and heavy due to presence of battery and antenna.
Moderately large in size and less heavy compared to MEMS.
Smaller in size and much lighter compared to other sensors.

Conclusion
Structural health monitoring has been an important research area for the last few decades.The cumbersome use of wires to connect sensors with base stations has prompted researchers to investigate the feasibility of wireless SHM by incorporating various electromagnetic theories to overcome the limitations of the wired sensors.
Significant amount of research has been done in the field of MEMS systems and their application to SHM.Several different types of architectures with different specifications of microprocessors have been investigated.In order to address the power requirements of these sensing systems, few power harvesting techniques have also been investigated for effective functioning of these sensor systems.However, the design of such sensor units are relatively more complicated due to the incorporation of the sensing unit, signal conditioning unit, antennas and power harvesting devices.Thus, these sensors become expensive and the performance of these sensors has to be monitored regularly.In recent years, researchers have started using resonators and antennas as strain sensors where the resonant frequency of the resonator shifts due to the change in their dimensions during a structural deformation.These devices could be made small and simple which becomes elegant to install and read wirelessly.Moreover, the use of metamaterial based resonators could make these sensors very small thereby providing the feasibility for an array of sensors for damage monitoring.
In the use of patch antennas, along with the strain, the spatial orientation of the strain could also be wirelessly estimated which becomes crucial in terms of crack direction and growth prediction.However, the work on the use of resonators and antennas as sensors in SHM is presently at the embryonic stage.More effort towards these sensors could prove fruitful, especially work directed towards effective wireless SHM with good damage discrimination, sensitivity and linearity.It is envisaged that once these resonators or antennas can be designed to exhibit linear frequency shifts with strain, its utility can be extended to assess damage in structures, irrespective of the dielectric properties of the host structure.Given the comparative parameter-based assessment of the technologies discussed in this paper (in Table 3), it might be that the way ahead in wireless SHM should consist of a combination of the technologies discussed in this paper, since no single technique had all the desirable properties. Fig
microstrip patch antennas whose resonant frequency changes with change in its dimension due to strain.Yi et al., 2011 a & b; 2012 a & bUse of rectangular microstrip patch antennas in combination with IC chips to form a RFID strain sensor whose resonant frequency changes with change in its dimension due to strain. ).

Table 2 .
Table 3 compares the various techniques covered in this article in general and in terms of important design parameters based on the authors' understanding of the sensor types.It is clearly evident from this table that each sensor type has its own advantages and limitations and no single technique has all the desirable properties.Various techniques investigated for use of resonators as strain sensors

Table 3 .
Comparison of various types of sensors based on design parameters