IOP PUBLISHING MEASUREMENT SCIENCE AND TECHNOLOGY

Meas. Sci. Technol. 21 (2010) 035201 (9pp) doi:10.1088/0957-0233/21/3/035201

Wireless interrogation of passive antennasensorsS Deshmukh and H Huang1

Department of Mechanical and Aerospace Engineering, University of Texas at Arlington,500 W First Street, WH211, Arlington, TX 76019, USA

E-mail: huang@uta.edu

Received 21 October 2009, in final form 19 December 2009Published 25 January 2010Online at stacks.iop.org/MST/21/035201

AbstractRecently, we discovered that the resonant frequency of a microstrip patch antenna is sensitiveto mechanical strains or crack presence in the ground plane. Based on this principle, antennasensors have been demonstrated to measure strain and detect crack in metallic structures. Thispaper presents a wireless method to remotely interrogate a dual-frequency antenna sensor. Aninterrogation horn antenna was used to irradiate the antenna sensor with a linear chirpmicrowave signal. By implementing a light-activated switch at the sensor node andperforming signal processing of the backscattered signals, the resonant frequencies of theantenna sensor along both polarizations can be measured remotely. Since the antenna sensordoes not need a local power source and can be interrogated wirelessly, electric wiring can beeliminated. The sensor implementation, the signal processing and the experimental setup thatvalidate the remote interrogation of the antenna sensor are presented. A power budget modelhas also been established to estimate the maximum interrogation range.

Keywords: passive wireless sensor, antenna backscattering, sensor interrogation, antennasensor, patch antenna.

(Some figures in this article are in colour only in the electronic version)

1. Introduction

Wireless sensors have many advantages over traditional wiredsensors. Sensor installation and maintenance costs are muchlower for wireless sensors because the need for cables iseliminated. In addition, the mobile nature of wireless sensorsalso makes reconfiguration of the sensor networks easier.Using wireless communication as the sensor interrogationmethod eliminates the physical limitation on the maximumnumber of sensors that can be interfaced to a monitoringsystem. Wireless sensors can be categorized into two groups:active wireless sensors and passive wireless sensors. Activewireless sensor configuration includes a power source, a sensornode, an analog to digital converter (ADC), an embeddedmicroprocessor and a wireless data transmitting unit (Houand Lynch 2006, Kiremidjian et al 2004, Kurata et al 2005,Nagayama et al 2007). The requirement for a local powersource increases the weight and size of the sensor and limits its

1 Author to whom any correspondence should be addressed.

active life-time. One popular method to eliminate the batteryis using an energy-harvesting device (Wang and Yuan 2007,Beeby et al 2006). The energy harvester generates energy fromother energy sources available in the ambient environment,such as mechanical vibrations, temperature gradient, opticalpower, etc. The energy harvester requires special componentsto generate and store energy, and thus increases the unit cost,its physical dimensions and the overall weight. Moreover,the energy density from the ambient energy sources is usuallylow. It takes considerable amount of time for the energyharvester to gather sufficient energy for the wireless sensorto function. This means that the wireless sensors poweredby an energy harvester can only acquire and transmit thesensing information intermittently. Passive wireless sensorsdo not have a local power source at the sensor node. Themost common passive wireless sensors are based on radiofrequency identification (RFID) concept and surface acousticwave (SAW) devices (Abraham and Varadan 2006, Pernget al 2007, Wen et al 2007). This type of sensor couples aSAW device to a passive RFID tag. The RFID tag intercepts

0957-0233/10/035201+09$30.00 1 © 2010 IOP Publishing Ltd Printed in the UK

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

Figure 1. Configuration of a microstrip patch antenna.

the interrogation signal transmitted by the RFID tag readerwhile the SAW device converts the interrogation signal to aSAW packet and uses it to perform the measurement. Thereflected SAW packet is again converted back to RF signalsand wirelessly transmitted to the tag reader using the RFIDtag. Because of the double conversion of RF signals toSAWs, the insertion loss of the sensor is relatively high.Passive wireless sensors utilizing inductively coupled energyfor measurement and wireless transmission have also beendeveloped (Butler et al 2002, Gould et al 2009, Harpster et al2002). The sensor, a parallel-connected LC tank, is connectedto an inductive coil. The measurand changes the capacitanceor inductance of the LC tank, which in turn changes theresonant parameters of the sensor coil. By setting up amagnetic field near the sensor coil, a reader coil measures theresonant parameter changes of the sensor coil to decipher themeasurement information. Inductive coupled passive wirelesssensors usually have high losses. Therefore, the maximuminterrogation distance between the sensor and the reader isvery limited, usually of the order of a few centimeters.

Recently, we developed a different class of passivewireless sensors based on microstrip patch antenna technology.We have demonstrated that a patch antenna can be usedto detect and measure different physical parameters suchas cracks (Mohammad and Huang 2009, Deshmukh et al2009) and mechanical strain (Tata et al 2009). A typicalpatch antenna, as shown in figure 1, consists of a dielectricsubstrate placed between a metallic patch and the groundplane. The factors determining the resonant frequency ofthe patch antenna are the geometric length and width ofthe antenna patch, the conductivity of the ground plane andthe dielectric properties of the dielectric substrate (Balanis2005). Physical parameters having an influence on thesefactors will cause the resonant frequencies of the patch antennato change. Therefore, a patch antenna can be exploited tomeasure the physical measurands based on its effect on theresonant frequency. For example, we have demonstrated thatthe antenna resonant frequency is very sensitive to mechanicalstrain developed in a structure subjected to external loading(Tata et al 2009). In addition, metallic structures can act asthe ground plane for the patch antenna. Cracks in the groundplane under a patch antenna increase the average path lengthof the current flow, resulting in a reduction in the resonantfrequency of the patch antenna. By measuring the change in

the antenna resonant frequencies, the crack growth under thepatch antenna can be detected and quantified.

In this paper, we present a technique to wirelesslyinterrogate the antenna sensors and thus achieve passiveoperation of the antenna sensor. By eliminating the needfor a power source at the sensor level, no electric wiring isrequired for data transmission or power supply, which couldfacilitate the implementation of large-scale sensor networks.The principle of wireless interrogation based on antennabackscattering is first discussed, followed by the discussions ofhardware implementation and software development to extractthe antenna resonant frequencies from the backscatteredsignal. Experimental results that demonstrated wirelessinterrogation of the antenna sensor are presented. Developinga power budget model to estimate the maximum interrogationdistance is also presented.

2. Principle of operation

A passive antenna sensor can be wirelessly interrogated basedon the principle of antenna backscattering. If we broadcasta wideband interrogation signal toward the antenna sensorusing an antenna reader, the portion of the interrogationsignal whose frequency matches with the resonant bandof the antenna sensor will be received by the antennasensor. If the antenna sensor is not properly terminatedwith perfectly matched impedance, the received signal willbe reflected at the termination and reradiated by the antennasensor. The reradiation of the received signal by an antennais called the antenna mode backscattering. In additionto antenna backscattering, the surrounding structures andthe antenna’s physical structure will indiscriminately scatterback the wideband interrogation signal, creating structuremode backscattering. In other words, the backscatteredsignal received by the antenna reader consists of two typesof backscattered signals: the antenna mode backscatteringand the structure mode backscattering (Penttila et al 2006,Dardari and D’Errico 2008, Knott et al 1993). The structuremode backscattering usually has the same spectrum as theinterrogation signal. The antenna mode backscattering,however, has a spectrum that matches the resonant frequencyband of the antenna. Based on this principle, we canmeasure the resonant frequency of the antenna sensorremotely by performing spectral analysis of the antenna modebackscattering.

Since the surrounding structures are much larger than theantenna aperture, the structure mode backscattering is typicallyone or two orders of magnitude greater than the antenna modebackscattering. Without proper signal manipulation and dataprocessing, we will not be able to determine the resonantfrequency of the antenna sensor from the backscattered signalsince it is dominated by the structural mode backscattering.To resolve this difficulty, a signal normalization scheme thatcan significantly reduce the structure mode backscattering isneeded. This can be achieved by switching the terminatingimpedance of the antenna sensor from open to short. Foran open termination, the received signal is reflected withoutany phase change. For a short termination, however, the

2

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

Interrogating antenna

Antenna sensor

Backscattered signal

Incident chirp signal

Switch

(a)

FET device

Photocell

Antenna sensor

D

GS

(b)

Figure 2. Wireless interrogation of the antenna sensor based on antenna backscattering: (a) antenna impedance switching;(b) implementation of light-activated RF switching.

received signal is reflected with a 180◦ phase change at thetermination. Therefore, there is a 180◦ phase differencebetween the antenna mode backscattering measured whenthe patch antenna is terminated with an open circuit andthat measured when the patch antenna is terminated witha short circuit. On the other hand, the structure modebackscattering remains the same regardless of the antennatermination. Subtracting the backscattered signals obtained atthese two antenna termination states, therefore, will cancel outthe structure mode backscattering while doubling the antennamode backscattering. The normalized antenna mode can thenbe analyzed in the frequency domain to determine the resonantfrequency of the antenna sensor.

The implementation of wireless interrogation of theantenna sensor using impedance switching is illustrated infigure 2. The antenna sensor is placed in the far-field region ofan interrogating antenna. A wideband chirp signal is radiatedfrom the interrogation antenna toward the antenna sensor. Amicrowave switch can be inserted between the antenna sensorand its ground plane to switch the termination of the antennasensor from open to short, as shown in figure 2(a). In orderto make the antenna sensor completely wireless and passive,this microwave switch should be remotely controlled as well.The implementation of a light-activated microwave switchis illustrated in figure 2(b). A low-power microwave bandfield-effect transistor (FET) device controlled by a compactphotocell is employed to perform the switching action. TheFET gate terminal is biased by the gate-source voltage, whichis provided by the photocell. A light beam can be used tocontrol the photocell voltage and thus remotely control theFET switching operation. When the light beam is off, thejunction between the drain and the source terminals of the FETis saturated. Therefore, there is a perfect electrical connectionbetween these two terminals, short-terminating the antennasensor to the ground. When the light beam is turned on, thephotocell provides −0.8 V to the gate terminal of the FET,which cuts off the junction between the source and the drainterminal of the FET. As a result, the antenna sensor is openterminated. Using this light-activated microwave switchingmechanism, wireless interrogation and passive operation ofthe antenna sensor can be achieved.

3. Implementation and evaluation of the microwaveswitching circuit

A pseudomorphic high electron mobility transistor (pHEMT),ATF36077 from Avago Technologies, was selected to

VNA

pHEMT

0.8VSwitch

(a)

D

G

S

(b)

To VNA

pHEMT

SMA connector

Microstrip transmission

line

Figure 3. Experiment setup to study the switching operation of thepHEMT: (a) schematic; (b) actual implementation.

implement the switching circuit because it has goodperformance characteristics in the microwave frequency bandand it is a discrete device available in a microstrip package.In order to verify the switching mechanism over the requiredoperating frequency range, an experimental setup as shown infigure 3 was devised. The pHEMT is controlled using a voltagesource and the reflections of the microwave signal supplied tothe drain terminal at different switching states are measuredusing a vector network analyzer (VNA—Rohde & SchwarzZVA24). The pHEMT, mounted on a dielectric substrate, isconnected to the VNA through a SMA connector by connectingits drain terminal and the pin of the SMA connector usinga microstrip transmission line. The source terminal of thepHEMT is connected to the ground. The VNA was set tomeasure the reflection S-parameter S11 from 3 GHz to 10 GHzwith 4000 points. This experiment was performed for gateterminal voltages of 0 V and −0.8 V, corresponding to thecase when the pHEMT is in saturation mode and in cutoffmode, respectively. The phase responses of the measuredS11 parameters at the two pHEMT switching states weresubtracted to determine the phase difference between thesetwo reflections.

After the switching mechanism of the pHEMT wasverified, the light-activated switching circuit was implementedusing the pHEMT and a photocell (Clare model—CPC 1832).The integration of the antenna sensor and the light-activatedswitching circuit is shown in figure 4. The patch antennawas fabricated by bonding a dielectric substrate (Kapton HN,125 μm in thickness, dielectric constant εr = 3.4) on analuminum block using super glue, followed by bonding theantenna patch on the dielectric substrate, again using superglue. The antenna patch pattern and the transmission linewere cut from a copper tape (3M 118, 68 μm in thickness,12.75 mm in width and 15 mm in length) using a sharp blade.The source terminal of the pHEMT, the anode of the photocelland the ground of the 560 k� resistor were bonded to the

3

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

Antenna patch Photocell

Kapton substrate pHEMT Resistor

D G

S

Figure 4. Implementation of the passive antenna sensor with alight-activated RF switch.

antenna ground plane using conductive epoxy. The inset-fedantenna sensor was connected to the drain terminal of thepHEMT through a microstrip transmission line. The gateterminal of the pHEMT was connected to the cathode of thephotocell. The pHEMT offers an open circuit from its drainterminal to the ground plane when its gate voltage is kept at−0.8 V and a short circuit when its gate voltage is 0 V. The580 k� resistor was employed to reduce the voltage output (atconstant current) of the photocell from an open circuit valueof 8 V to 0.8 V when it was exposed to a light beam generatedby a flashlight. To make sure that the photocell produces astable output, the light beam should be strong enough so thatthe photocell is operating in saturation. It was observed thatthe ambient light was not sufficient to cause any change in thepHEMT biasing.

4. Wireless interrogation of the passive antennasensor

The experimental setup for wireless interrogation of thepassive antenna sensor is illustrated in figure 5(a). The antennasensor was placed in the far-field region of an interrogatinghorn antenna connected to a VNA. A light source was placedpointing at the photocell that switches the termination ofthe antenna sensor. The light source and the VNA werecontrolled by a program developed using Microsoft Excel.

VNA

1.34 m

Photocell

pHEMT

Interrogating antenna

GD

S

Light source

Control program

_

+ Antenna sensor

(a)

580KΩ

(b)

Horn Antenna

Antenna sensor

VNA

Light source

Figure 5. Wireless interrogation of the passive antenna sensor: (a) diagram; (b) experimental implementation.

The implementation of the wireless interrogation systemis shown in figure 5(b). An A6100 horn antenna fromSinger was employed as the interrogation horn antenna. Theantenna sensor and the light-activated switching circuit wereimplemented on an aluminum block, which is then mounted ona large aluminum sheet that imitates a large metallic structure.The distance between the antenna sensor and the horn antennawas 1.34 m. The horn antenna was connected to port 1 ofa VNA, which was programmed to collect 4000 data pointsfrom 5 GHz to 9 GHz. The VNA was calibrated with itsreference plane at the horn antenna feeding point and theoutput power of the VNA was set to 20 dBm. A flashlightwas used as the light source. To control the flashlight usingthe Excel control program, the flashlight was connected to apower supply through a relay (OEG model OUAZ-SS-105D).The switching signal for the relay was generated using theparallel port of the computer. The VNA was connected to thecomputer using an Ethernet cable. An Excel control programwas developed to initiate the VNA at the beginning of the test,trigger the VNA to collect data at different antenna terminationstates, receive data from the VNA and store the data in the harddisk of the computer. The same program also controls the lightsource so that the switching of the antenna termination and thedata collection are synchronized. The measurement was donein two steps. The spectrum of the backscattered signal was firstmeasured when the light source was turned off. At this state,the pHEMT was in saturation mode and the antenna sensorwas short terminated. The light source was then turned on tobias the pHEMT into the cut-off mode and thus terminate theantenna sensor with open impedance. The VNA was againtriggered to collect the spectrum of the backscattered signal.The measured spectra were downloaded from the VNA to acomputer for data processing.

A rectangular patch antenna can radiate at twofundamental radiation modes: the TM01 mode is polarizedalong the length direction of the antenna patch, while theTM10 mode is polarized along the width direction of theantenna patch (Tata et al 2009). As long as the antennapatch is not square, it will display two resonant frequencies:each corresponding to a fundamental radiation mode. Thesetwo resonant frequencies, denoted as f 01 and f 10, can beinterrogated separately by aligning the polarization of the

4

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

Width

Length

Horn Antenna

PatchAntenna

Width dimension polarization (a)

Width

Length

Length dimension polarization (b)

Figure 6. Remote interrogation of the dual-frequency antennasensor: (a) width dimension frequency interrogation; (b) lengthdimension frequency interrogation.

interrogating signal to be parallel to the polarization directionof a particular radiation mode. For example, the horn antennaused in figure 5(b) has two connectors: one for verticalpolarization and the other for horizontal polarization. Asshown in figure 6(a), in order to interrogate the antennafrequency associated with the width dimension of the antennasensor, namely the f 10 frequency, the connector of the hornantenna for vertical polarization is connected to the VNA. Infigure 6(b), the VNA connection is switched to the horizontalpolarization to interrogate the f 01 frequency.

5. Digital signal processing for antenna modenormalization and time gating

As discussed in section 4, two sets of backscattering signalspresented in frequency domain were acquired using the VNAat the two termination states of the antenna sensor. These twosets of data should be subtracted to isolate the antenna modebackscattering from the structural mode backscattering. Inaddition, the VNA measures the frequency components of thebackscattered signal over a long time span. The arrival of theantenna mode backscattering, on the other hand, is determinedby the distance between the interrogating antenna and theantenna sensor. Therefore, the signal-to-noise ratio (SNR)of the antenna mode signal can be improved by time gatingthe normalized backscattered signal. In order to performtime gating, the frequency-domain signal measured by theVNA must be converted to time domain first. After timegating, the time-domain antenna mode signal again needs tobe converted back to frequency domain to extract the resonantfrequency of the antenna sensor. A digital signal processingprogram, therefore, is needed to perform the time gating,

VNA data-antenna open terminated

VNA data -antenna short

terminated

IFFT

IFFT

Timegate

FFTSTFT SpectrogramAntenna Resonant Spectrum

Figure 7. Signal processing of measurement data.

the frequency-time domain conversions and the normalizationscheme. A flow diagram of the signal processing algorithmis shown in figure 7. The two measurement data files areimported into the program first. Subsequently, the frequency-domain signals are converted to time-domain signals usinginverse fast Fourier transformation (IFFT). The time-domainsignals are then subtracted to obtain a normalized signal thatsupposedly contains the antenna mode backscattering only.Next, a short-time Fourier transformation (STFT) is performedto display the time–frequency components of the normalizedsignal, based on which the starting and ending time of theantenna mode backscattering can be selected. Finally, thenormalized antenna mode signal is time gated and is convertedto frequency domain using fast Fourier transformation (FFT).Based on the frequency spectrum of the time-gated antennamode backscattering, the resonant frequency of the antennasensor corresponding to the polarization of the interrogatingsignal can be determined.

6. Power budget model

The power received by the antenna sensor Prs, placed ata distance R from the interrogating horn antenna, can becalculated from the Friis transmission equation (Balanis 2005)as

Prs = Pt

(λ

4πR

)2

GhGs, (1)

where Gh and Gs are the gains of interrogating horn antennaand the antenna sensor, respectively. λ is the wavelengthof the interrogation signal and Pt is the power of theinterrogation signal. When the pHEMT is in saturation mode,the pHEMT introduces an insertion loss due to the impedancemismatch between the microwave switching circuit and thepatch antenna. Therefore, the power of the signal that passesthrough the pHEMT and reaches the ground is given by

Pts = Dds × Prs, (2)

where Dds is the insertion loss of the pHEMT for signalstransmitting from the drain terminal to the source terminal.Reflected by the ground, the signal has to go through thepHEMT again to reach the patch antenna. As a result, thepower of the signal that is reflected back to the patch antennais

Pbs = Dsd × Pts = DdsDsdPrs. (3)

Again, Dsd is the insertion loss experienced by the signaltransmitted from the source terminal to the drain terminal ofthe pHEMT. The reflected power Pbs is transmitted by the

5

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

patch antenna and received by the interrogating horn antenna.The received power at the interrogating horn antenna is againcalculated from the Friis transmission equation, i.e.

Pr,st = Pbs

(λ

4πR

)2

GhGs. (4)

Substituting equations (1) and (3) into equation (4), we have

Pr,st = Pt

(λ

4πR

)4

G2hG

2sDdsDsd. (5)

When the pHEMT is in cut-off mode, the patch antenna isopen circuit terminated. Similarly, the power of the signalreceived by the horn antenna when the antenna sensor is openterminated is calculated as

Pr,op = Pt

(λ

4πR

)4

G2hG

2s S11, (6)

where S11 is the S-parameter measuring the reflection from thepHEMT. The power of the normalized antenna mode is thengiven by

Pr = Pr,st + Pr,op = Pt

(λ

4πR

)4

G2hG

2s (DdsDsd + S11) . (7)

In order to interrogate the antenna sensor with a required SNR,the power of the normalized antenna mode should satisfy

Pr � NF × SNR, (8)

where NF is the noise figure of the measurement system.The maximum distance Rmax at which the sensor can besuccessfully interrogated, therefore, can be derived fromequations (7) and (8) as

Rmax = c

4πf

[(Dds × Dsd + S11) × Pt × G2

hG2s

NF × SNR

]1/4

. (9)

7. Results and discussions

Results from two experiments will be discussed in thissection. The first experiment was conducted to study theswitching operation of the pHEMT. The second experimentwas conducted to verify the wireless interrogation of anantenna sensor. In addition, the insertion losses of thepHEMT and the noise figure of the measurement system weremeasured, based on which the maximum interrogation distancecan be estimated.

To verify the switching operation of the pHEMT, theS11 parameters of the pHEMT-based switching circuit weremeasured using the experimental setup shown in figure 3.The S11 curves obtained when the pHEMT was in cut-off andsaturation states are shown in figure 8(a), which correspond toan open-terminated and a short-terminated transmission line,respectively. It can be observed from figure 8(a) that the returnloss of the pHEMT is flat from 4.2 GHz to 7.5 GHz for bothstates. The return loss of the pHEMT in saturation has a valleyat around 4.1 GHz due to the loop resonance between the SMAconnector and the ground plane. This resonance effect will notbe present in the antenna sensor since the SMA connector willbe replaced by the patch antenna (see figure 3). The time-domain response of the pHEMT at these two states is shown

(a)

(b)

(c)

Time gate

Figure 8. Characterization of the pHEMT-based switching circuit:(a) S11 measurement; (b) reflected signals in time domain; (c) phasedifference of the reflected signals.

in figure 8(b). Since the VNA is calibrated to the plane ofthe SMA connector feed, the response from 0 ns to 0.2 ns isdue to the transmission line. The pHEMT response covers thetime span from 0.2 ns to 0.37 ns. The large pHEMT responseamplitude is due to the open and short termination of thepHEMT. A phase difference of approximately 180◦ betweenthe responses of the two pHEMT modes is clearly visible. Ifwe time gate the pHEMT response and process them usingFFT, the phase difference between the pHEMT responses forthe saturation mode and the cutoff mode can be obtained, asshown in figure 8(c). The frequency-domain phase differenceis pretty constant for the entire frequency range of operationbut is slightly lower than 180◦. This is due to the additionaltransmission line needed to connect the pHEMT to the ground.Nevertheless, the phase difference is sufficient to achieve thenormalization scheme.

Wireless interrogations of the antenna sensor wereperformed using the experimental setup shown in figure 5. Theresonant frequencies of the antenna sensor were first simulatedusing an EM simulation tool, Sonnet Pro. As shown in figure 9,the S11 curve of the antenna sensor displayed two resonant

6

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

f10f01

Figure 9. Simulated S11 curve of the antenna sensor.

Antenna mode

Antenna mode signal

(a) (b)

Figure 10. Normalized antenna mode signal: (a) time domain; (b) spectrogram.

f01 f10

(a) (b)

Figure 11. Spectra of the normalized antenna mode signal after time gating: (a) length dimension; (b) width dimension.

frequencies, i.e. f 01 = 6.3 GHz and f 10 = 7.4 GHz. Thepolarization of the interrogation horn antenna was configuredto interrogate the f 01 frequency first. The normalizedbackscattered signal is shown in figure 10(a). Obviously, thenormalization process did not cancel out the structural modebackscattering completely. Nevertheless, the spectrogram ofthe normalized signal shown in figure 10(b) displayed a strongfrequency component at 6.5 GHz between 10 ns and 11.5 ns.Based on the spectrogram, the wave packet from 9.8 ns to

11.5 ns of the normalized signal was identified as the antennamode backscattering, as highlighted in figure 10(a). Thearrival time of the antenna mode backscattering, at around9.8 ns, is the time taken for the interrogation signal to propagatefrom the horn antenna feed to the antenna sensor and back. Theantenna mode signal was time gated and processed using FFTto obtain the frequency spectrum of the antenna mode signal, asshown in figure 11(a). The f 01 frequency of the antenna sensorwas determined from the frequency at which the amplitude of

7

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

antenna mode is the largest. The f 01 frequency measured fromfigure 11(a) is 6.35 GHz, which matched with the simulatedf 01 frequency very well. In addition, repeated experimentsyielded consistent results, which verified the stability of themeasurements over time. The f 10 frequency of the antennasensor was measured following the same procedure, exceptchanging the polarization of the interrogation signal. Thefrequency spectrum of the f 10 antenna mode signal is shown infigure 11(b). The f 10 frequency measured from figure 11(b)is 7.4 GHz, which again matched with the simulatedf 10 frequency very well. The SNR of the time-gatednormalized antenna mode at f 10 frequency was estimatedto be approximately 1.16 (0.64 dB). Averaging the sevenmeasurements improved the SNR slightly (see dashed greencurves in figures 11(a) and (b)). This indicates that the ‘noisefloor’ is probably contributed by other deterministic scatteringsources due to the imperfect normalization.

To measure the insertion losses of the pHEMT and thenoise figure of the measurement system, the transmissionS-parameters from the horn antenna to the patch antennawere first measured by connecting the horn antenna to port1 of the VNA and the patch antenna to port 2 of theVNA. The S12 and S21 parameters were measured to bearound −34 dB. Assuming the interrogation frequency is f =7.5 GHz, the gain of the horn antenna is Gh = 12 dBi =15.84, the gain of the antenna sensor is Gs = 6 dBi = 3.98and the distance R = 1.34 m, the calculated horn antenna topatch antenna transmission loss is 34.5 dB, which matchedwith the measurements very well. Next, the switching circuitwas added between the patch antenna and the port 2 ofthe VNA. When the pHEMT is in saturation, i.e. the patchantenna is short-connected to the SMA connector, the S12 andS21 parameters were measured to be 43.4 dB. Therefore, theinsertion losses of the switching circuit, Dds and Dsd, can becalculated from the S12 and S21 parameters measured beforeand after the switching circuit was inserted, i.e. Dds = Dsd =43.4–34.5 = 8.9 dB. The S11 parameter of the pHEMT in cut-off mode can be measured from figure 8(a), which is around−1 dB. Substituting these parameters into equation (7), thepower of the normalized antenna mode signal is calculated tobe −49.91 dBm, assuming the interrogation power is 20 dBm.The noise figure of the system can be then estimated fromthe calculated receiving power and the measured SNR to be−50.55 dBm, which is 12.55 dB higher than the noise figure ofthe VNA calculated from the specifications of the VNA. Theadditional noises are likely contributed by the surroundingmedium and the structural mode backscattering that was notcompletely cancelled out by the normalization process. Toincrease the interrogation distance and the SNR, the powerof the interrogation signal and the gain of the interrogationantenna should be increased. For example, increasing thepower of the interrogation signal to 30 dBm and the gain ofthe interrogation antenna to 20 dB, the interrogation distancecan be increased to 3.5 m with a SNR of 10 dB, assuming theother parameters remain the same (Dds = Dsd = 8.9 dB andNF = −50.55 dBm). Reducing the insertion losses Dds andDsd of the switching circuit will also increase the interrogationrange. However, their influences on the interrogation rangeare not very significant, based on equation (9).

The wireless interrogation scheme using a VNA requiresmultiple transformations from frequency domain to timedomain and vice versa. These multiple transformations canbe avoided if a time-domain measurement technique such astime-domain reflectometry (TDR) is used. However, TDRusually has a smaller dynamic range and higher noise floorthan a VNA. Therefore, a lock-in amplifier maybe needed toimprove the TDR SNR. In addition, the wireless interrogationsystem can be improved by replacing the light-activated switchwith a RF-activated switch that does not require direct line-of-sight propagation.

8. Conclusions

This paper presented a technique to wirelessly interrogatepassive antenna sensors. Experimental results demonstratedthat an antenna sensor with dual polarizations can beinterrogated wirelessly by implementing a light-activated RFswitch. This capability enables the antenna sensors to transmitmeasurement information along both its length and widthdirection. A power budget model was established, which canbe used to estimate the maximum interrogation distance withdifferent measurement configurations. We are evaluating thewireless interrogation scheme to measure the crack-induced orstrain-induced resonant frequency shifts of the passive antennasensors. The results of the experiment will be a subject ofsubsequent publications.

Acknowledgments

This project is supported by the Air Force Office of ScientificResearch under contract no FA9550-07-1-0465. The supportand encouragement of the program managers, Drs DavidStargel and Victor Giurgiutiu, are greatly appreciated.

References

Abraham J K and Varadan V 2006 Modeling of passive wirelesssensors for RFID tag applications Proc. SPIE 6172 617204

Balanis C 2005 Antenna Theory Analysis and Design 3rd edn(New York: Wiley-Interscience)

Beeby S P, Tudor M J and White N M 2006 Energy harvestingvibration sources for microsystems applications Meas. Sci.Technol. 17 R175–95

Butler J C, Vigliotti A J, Verdi F W and Walsh S M 2002 Wireless,passive, resonant-circuit, inductively coupled, inductive strainsensor Sensors Actuators A 102 61–6

Dardari D and D’Errico R 2008 Passive ultrawide bandwidth RFIDGlobal Telecommunication Conf. IEEE Globecom 1–6doi:10.1109/GLOCOM.2008.ECP.757

Deshmukh S, Mohammad I, Wu T, Tentzeris M and Huang H 2009Crack detection and monitoring using passive wireless antennasensor ASME Conf. on Smart Materials Adaptive Structuresand Intelligent Systems Oxford CA SMASIS2009-1326

Gould D, Meiners M, Benecke W and Lang W 2009 Condensationdetection using a wirelessly powered RF temperature sensorIEEE Trans. Veh. Technol. 58 1667–72

Harpster T J, Stark B and Najafi K 2002 A passive wirelessintegrated humidity sensor Sensors Actuators A 95 100–7

Hou T and Lynch J P 2006 Rapid-to-deploy wireless monitoringsystems for static and dynamic load testing of bridges:validation on the grove street bridge Proc. SPIE 6178 105–16

8

Meas. Sci. Technol. 21 (2010) 035201 S Deshmukh and H Huang

Kiremidjian G K, Kiremidjian A and Lynch J P 2004 Wirelessstructural monitoring for homeland security applications Proc.SPIE 5395 82–90

Knott E, Shaeffer J and Tuley M 1993 Radar Cross Section 2nd edn(Boston, MA: Artech House)

Kurata N, Spencer B F and Ruiz-Sandoval M 2005 Risk monitoringof buildings with wireless sensor networks Struct. ControlHealth Monit. 12 315–27

Mohammad I and Huang H 2009 Passive wireless patch antennasensor for crack detection The 8th Int. Workshop on StructuralHealth Monitoring (Stanford University)

Nagayama T, Spencer B F and Rice J A 2007 Structural healthmonitoring utilizing Intel’s Imote2 wireless sensor platformProc. SPIE 6529 652943

Penttila K, Keskilammi M, Sydanheimo L and Kivikoski M 2006Radar cross-section analysis for passive RFID systems IEEProc. Microw. Antennas Propag. 153 103–9

Perng J K, Hunt W D and Edmonson P J 2007 Development of ashear horizontal SAW RFID biosensor IEEE Sensors Conf.pp 691–4

Tata U, Huang H, Chiao J C and Carter R 2009 Exploiting patchantenna for strain measurement Meas. Sci. Technol. 20 015201

Wang L and Yuan F G 2007 Energy harvesting by magnetostrictivematerial (MsM) for powering wireless sensors in SHM Proc.SPIE 6529 652941

Wen W, Lee K, Kim T and Yang S 2007 Development of wirelessMEMS sensor for RFID tag and temperature/pressuremonitoring Proc. SPIE 6463 64630Q

9