Constrained Pole-Zero Synthesis of Phase-Oriented RFID Sensor Antennas · 2016-04-08 · Constrained Pole-Zero Synthesis of Phase-Oriented ... The basic operation principle[8] ...

496 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 2, FEBRUARY 2016

Constrained Pole-Zero Synthesis of Phase-OrientedRFID Sensor Antennas

Stefano Caizzone, Emidio DiGiampaolo, and Gaetano Marrocco

Abstract—Passive sensing by means of radiofrequencyidentification has been extensively explored for various applica-tions, such as gas detection, temperature change, and deformation.The sensing indicator is generally based on the amplitude andphase of the backscattered field. However, a degradation ofthe communication performance must be usually accepted forachieving the sensing capability. This work introduces a designmethod suitable for phase-based RFID sensors that permitsto shape the phase response while preserving the impedancematching between the antenna and the microchip. The RFIDsensor is modeled as a two-ports scatterer comprising a lumpedsensor at one of the ports and an RFID chip at the other port.A pole-zero representation of the electromagnetic interactionbetween the reader and the RFID sensor allows to introduce aconstrained design of the antenna with a full control on the sensordynamic range and on the communication performance. Theproposed method is numerically and experimentally validatedby means of a pair of strongly coupled dipoles connected to avoltage-controlled varactor emulating a dynamic sensor response.

Index Terms—Antenna, array, Internet of Things, mutualcoupling, RFID, sensors, wireless communications.

I. INTRODUCTION

W ITH THE GROWTH of connectivity and the rapidintroduction of the next-generation Internet [the

Internet of Things (IoT)], the need and request for pervasivesensors are also increasing rapidly [1]. Radiofrequency iden-tification (RFID) technology is gaining an important role insuch a sector, thanks to its already widespread use in logistics.Mass market production guarantees in fact low costs and analready available standardization of devices and interfaces [2].Sensing and communication platforms of general purposes (likeArduino or Raspberry) have already fostered the developmentof IoT thanks to their versatility and the effort of enthusiasticusers [3]. Moreover, dedicated RFID devices and systems havebeen recently proposed to perform sensing and data loggingalso far from the reader’s proximity. A possible approach usinga wireless powered data logger is shown in [4] and [5], while

Manuscript received August 18, 2015; revised December 14, 2015; acceptedDecember 19, 2015. Date of publication December 23, 2015; date of currentversion February 01, 2016.

S. Caizzone is with the Institute of Communications and Navigation,German Aerospace Center (DLR), Wessling, Germany, and also with the DISP,University of Rome, Tor Vergata, Italy (e-mail: [email protected]).

E. DiGiampaolo is with the Dipartimento di Ingegneria Industrialee dell’Informazione e Economia, University of L’Aquila, Italy (e-mail:[email protected]).

G. Marrocco is with the DISP, University of Rome, Tor Vergata, Italy(e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TAP.2015.2511788

[6] describes a battery-assisted two-chip data logger for pack-aging surveillance applications. A generic RFID equipment tobe used as a sensing node once interfaced to an active sen-sor is presented in [7]. Although these devices show advancedperformance allowing digital recording of sensor data and thepossibility to manage a storage memory, they include either abattery or complicated energy harvesting circuits for power-ing the microcontrollers: their cost is still prohibitive for truepervasive applications.

Passive RFID sensors, instead, are in the latest years collect-ing a steady increase in interest thanks to the lower cost andto the long lifetime, which are possible because of the absenceof the battery and of complicated circuitry. The basic operationprinciple [8] requires a sensor or a sensor-like component capa-ble of modifying the antenna properties (e.g., the impedanceor the gain) of the RFID tag. Such variation can be, there-fore, remotely detected and stored by interrogating the RFIDsensor tag with a reader and by processing the backscatteredsignal. This approach has clear advantages of simplicity, whichis key for low-cost sensing; however, it exhibits the drawbackof intrinsically implying a communication degradation (due tothe varying impedance or gain) aside with the sensing pro-cess. A tradeoff analysis for the two contrasting requirementscan be found in [9], with guidelines to minimize the com-munication degradation while amplifying the power orientedsensing.

Recently, the authors have shown [10] how mutual couplingeffects between two closely spaced tags can help in compen-sating the impedance variation due to the sensing process (inthat case, a crack width enlargement). Engineered coupling can,therefore, enable a stable communication while at the sametime obtaining sensing through variations of the phase of theRFID signal. In that case, the two antennas played as trans-ducers themselves converting the change of the environmentinto a controlled modulation of the backscattered signal. Inother implementations, instead, it is preferred to include a spe-cific sensor into the antenna, for instance a lumped element,in order to provide a more specific detection of the physi-cal phenomenon. Some examples are humidity-sensor polymerdrops [11], thermistors [12], and substrates [13] comprisingvapor-sensitive polymers.

This paper introduces a generic synthesis methodology suit-able for RFID tags with an embedded lumped sensor that relyon phase monitoring. Unlike the configuration in [10], onlya single RFID chip is required. The phase of the backscat-tered system is mathematically manipulated to make explicitthe effect of the change of the impedance of the lumped sensing

0018-926X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

CAIZZONE et al.: CONSTRAINED POLE-ZERO SYNTHESIS OF PHASE-ORIENTED RFID SENSOR ANTENNAS 497

Fig. 1. Schematic representation of the two-port sensor, with one port beingconnected to an RFID chip and the other to a variable passive sensor.

element. A suitable pole/zero representation, originating fromthe multiport tag theory [14], [15], will, therefore, permit toformalize a constrained synthesis of the RFID tag to balancethe minimum allowed read range with the dynamic range ofthe phase sensing. The technique will require to properly placezeros and poles into a complex plane nomogram. Since thelumped sensor is only represented by its impedance response tothe external agent to be monitored, the proposed design methodwill be of general extent.

The paper is organized as follows. Section II provides theformulation of the problem with the explanation of the physicalparameters and their dependence on the variable load sensor.Section III introduces an useful nomogram in the complexplane of sensor impedance while the poles/zeroes synthesis isdescribed in Section IV. Finally the method implementationis demonstrated in Section V by means of a simulated andexperimental test-bed including a varactor element.

II. MATHEMATICAL REPRESENTATION OF TWO-PORT

TAG RESPONSE

The family of devices that are here investigated can bedescribed as two-ports scatterers (Fig. 1), where one port is con-nected to the RFID microchip and a second port is connected toa variable load, i.e., the passive lumped sensor of impedanceZL. The radiating body could be implemented as either twostrongly coupled scatterers or by a unique conductor providedwith two terminals. In both cases, due to the electromagneticcoupling among the two ports, the variation of the impedanceof the load will influence the induced current over the wholesystem and thus the power collected by the microchip. As aresult, the amplitude and phase of the backscattered signals willbe accordingly modified.

A. Phase Response

The antenna mode signal due to a two-port RFID tag can beexpressed in terms of impedance matrix and chip impedance asin [10]. Let us now consider the two ports of the tag (namedn and m) as sufficiently close to be electromagnetically cou-pled. If port n is loaded with an UHF microchip (Fig. 1), whilethe port m is terminated on a load, which is variable accord-ing for instance to some chemical/mechanical receptor, then the

signal received at the reader due to the antenna mode can bewritten as

VON/OFFR,(n) (Ψ)

=Znn + Zmm − 2Zmn + ZL(Ψ) + Z

ON/OFFCn(

Znn + ZON/OFFCn

)(Zmm + ZL(Ψ))− Z2

mn

f (r) (1)

where Zmn {m,n = 1, 2} are, respectively, self (m = n)and mutual (m �= n) impedances, Z

ON/OFFCn and ZL(Ψ) =

RL(Ψ) + jXL(Ψ) are, respectively, the chip impedance in thetwo ON and OFF states and the load impedance on port m,with Ψ the analyte, i.e., the physical parameter to monitor.f (r)is a function whose phase only depends on the reader-to-tags distance r and on the effective heights of tags andreader

f(r) ∝ λ2

π2

√PinRrad

R /2GRRradn Gn(hn · hR)

2 e−j2kr

4r2(2)

where λ is the wavelength, Pin is the power supplied by thereader, Rrad

R , hR, and GR are the radiation resistance, thepolarization unitary vector and the gain of the reader antenna,respectively; Rrad

n , hn, and Gn are instead the radiation resis-tance, the polarization unitary vector and the radiation gainof the tag antenna when the port m is in open circuit condi-tion.

The phase measured by the RFID reader can hence bewritten as

ϕn[ZL(Ψ)] = arg{V OFFR,(n) [ZL(Ψ)]− V ON

R,(n)[ZL(Ψ)]} (3)

where the dependence from ZL(Ψ) has been made clear.It is assumed, for simplicity, that the radiating body has a

symmetric geometry with respect to the two ports, hence Z11 =Z22 = ZS and Z21 = ZM . Phase becomes, therefore

ϕn[ZL(Ψ)] = arg

{− 2ZS − 2ZM + ZL(Ψ) + ZOFF

Cn(ZS + ZOFF

Cn

)(ZS + ZL(Ψ))− Z2

M

+2ZS − 2ZM + ZL(Ψ) + ZON

Cn(ZS + ZON

Cn

)(ZS + ZL(Ψ))− Z2

M

}+ ϕ0

(4)

ϕ0 = arg{f(r)} can be considered as a constant withrespect to variations of ZL(Ψ), because the polarization uni-tary vector [hn in (2)] is defined when port m is in open circuitand thus it does not depend on the load of port m. Moreover,if the measurement setup is kept fixed along with the sensingprocess, ϕ0 will not vary at all (as, besides open circuit param-eters, it only depends on the distance between reader and tag).The variations of the phase will then be due only to variationsof ZL(Ψ) and hence information on such load variations willbe remotely retrievable by measuring the phase of the tag asreceived by the reader.

B. Amplitude Response

On the other hand, communication properties for down-link limited passive systems are controlled by the realized


gain, which can be conveniently expressed in the followingform [16]:

Gn[ZL(Ψ)] = Gnχnτn

= Gn4RONCn Rrad

S

∣∣∣∣∣ ZS − ZM + ZL(Ψ)(ZS + ZON

Cn

)(ZS + ZL(Ψ))− Z2

M

∣∣∣∣∣2

χn.

(5)

χn is the polarization factor and

τn = 4RONCn Rrad

S

∣∣∣∣∣ ZS − ZM + ZL(Ψ)(ZS + ZON

Cn

)(ZS + ZL(Ψ))− Z2

M

∣∣∣∣∣2

is the embedded power transfer coefficient. Equation (5) showshow also the realized gain depends on the (variable) load ofthe sensing port m and will, therefore, need to be properlyengineered to avoid substantial degradation along the sensingprocess. It is worth clarifying that τn is a mathematical func-tion, different from the conventional ratio between the powerabsorbed by the chip impedance and the power collected by theantenna.

Since the reader/tag is kept at fixed distance and the gains andeffective heights are defined in open circuit condition for portm, the realized gain will only experience changes by means ofvariations of the lumped impedance ZL(Ψ) of the sensor.

C. Pole/Zero Representation

Equations (4) and (5) are the key equations for sensingand communication properties, respectively. After simple math-ematical manipulations, they can be rewritten in the morecompact form

ϕn[ZL(Ψ)] =

arg

{d · (ZL(Ψ)− z)

2

cONcOFF (ZL(Ψ)− POFF ) (ZL(Ψ)− PON )

}+ ϕ0

(6)

τn [ZL(Ψ)] =4RON

Cn RradS

|cON |2∣∣∣∣ ZL(Ψ)− z

ZL(Ψ)− PON

∣∣∣∣2 (7)

where

z = −ZS + ZM (8)

PON =− (

bON + a)

cON(9)

POFF =− (

bOFF + a)

cOFF(10)

and

a = (ZS + ZM ) (ZS − ZM )

bOFF/ON =(ZSZ

OFF/ONCn

)cOFF/ON =

(ZS + Z

OFF/ONCn

)d =

(ZONCn − ZOFF

Cn

).

The performance features both for sensing and for com-munication can hence be expressed in terms of poles({PON , POFF }) and zeroes (z), with the purpose of allowingan easier and fully mathematical control of their trends.

It is worth noticing that the functions (6) and (7) both havetwo coinciding zeroes. Equation (6) has two different polesPON and POFF , while (7) has two coinciding poles in PON .The presence of poles in (7) may in theory allow the τn tobe arbitrary large for a suitable choice of the load impedanceZL(Ψ) (i.e., it may tend to infinite). However, this possibil-ity vanishes for physically feasible antennas and passive loads.Indeed, a feasible antenna impedance [17] requires that 1) theself resistance RS > 0 due to passivity and 2) the self andmutual impedance to be subjected to the constraints |RM | <|RS |, |XM | < |XS |. Accordingly, it has been proved numeri-cally that the poles will always lay in the negative resistancehalf-space of the complex load impedance plane. As a conse-quence of the imposed constraints, also the zero z = −ZS +ZM will have a negative real part.

In such conditions, therefore, the embedded power trans-fer coefficients will never grow indefinitely using only passivedevices, as expected from common sense: τn in fact would tendto the pole only for negative values of RL, i.e., for negative loadresistances, not admissible for passive devices.

Although poles and zeroes belong to the RL < 0 half-plane,nevertheless they will influence the phase and power transmis-sion coefficient trends on the half-plane RL > 0 as well, asshown in next Section.

III. THE τn-ϕn NOMOGRAM

The concurrent control of the variation of the phase ofthe backscattered signals (sensing) and of the embeddedpower transfer coefficient (communication) is here imple-mented through the representation of all the involved functionsover the complex plane of the load impedance. The response ofthe lumped sensor to the physical parameter under observationcan be hence mapped on the ZL-plane as an oriented con-tour �ZL(Ψ) = RL(Ψ) + jXL(Ψ) with Ψ1 < Ψ < Ψ2, where{Ψ1, Ψ2} are the initial and final states of the analyte as in[9]. By plotting the isolines of τn and ϕn functions over theZL-plane, a nomogram is obtained (Fig. 2). The dynamic rangeof the tag, in its behavior of phase-based sensor, and its readdistance are hence dependent on the intersection between thesensor contour �ZL(Ψ) and the τn-ϕn nomogram.

The position of poles and zeroes on the ZL-plane will affectthe shape and orientation of isolines and their intersection withthe �ZL(Ψ) profile, and ultimately the device performance.

Phase and τn isolines have approximately a symmetrical cir-cular shape, in particular, the symmetry axis of τn isolineslays on the conjunction line between PON and z, while thephase, though having a more complicated behavior due to itsdependence also from POFF , has a symmetry axis almostperpendicular to black that of τn isolines.

Since phase isolines are almost perpendicular to τn ones,phase and τn exhibit different trends along a specific path onthe ZL plane: for instance, if �ZL(Ψ) draws a path that follows


Fig. 2. Example of ϕ-τn nomogram over a complex ZL plane, when z =−10− j110 Ω and PON = −70− j100 Ω and ZON

c = 15− j135 Ω,ZOFFC = 150. Phase isolines are plotted every 25◦ while τn isolines are

expressed in dB in all plots of this paper. The thick line indicates a qualitative�ZL(Ψ) contour, while the blue line is the PON − z axis.

Fig. 3. ϕ-τn nomogram over a complex ZL plane corresponding to a differentposition of zero and poles with respect to Fig. 2 (e.g., z = −5− j200 Ω andPON = −20− j100 Ω) that produces a near 90◦ rotation of the PON − zaxis (blue line).

a single τn isoline, it will cross several phase isolines and viceversa.

A change in the position of poles and zero, which iscontrolled by the electromagnetic interaction among the twoports of the sensor (ZS , ZM ) and by the chip impedance(ZON/OFF

C ), thus provokes a rescaling and a rotation of thenomogram with respect to the contour �ZL(Ψ) of the sensorresponse. For instance, by perturbing the values of zeroes andpoles of Fig. 2 (details in the captions), the map of Fig. 3 willbe produced. Now, the symmetry axes of the nomogram arerotated with respect to the previous case, due to the rotationof the line between PON and z. The result is that the contour�ZL(Ψ) now intersects a region with much denser isolines ofphase and accordingly the ϕ-τn response to the physical param-eter to monitor will be sharper than in Fig. 2. It is, therefore,intuitive that the sensor’s response and communication featurescan be mathematically manipulated by a proper placement ofzero and poles on the complex impedance plane as described inthe next paragraph.

IV. DESIGN METHOD

The proposed pole/zero design method comprises two steps:1) a mathematical placement of poles and zeros on the complexplane to properly steer the τn-ϕn nomogram and 2) an elec-tromagnetic design to shape a two-port antenna having therequired feasible impedance matrix that has been derived fromthe first step. The overall design is constrained to conditionsover the sensing range and minimum communication distance.

A. Design Constraints

The placement of poles and zeroes in the ZL plane is subjectto desired requirements on sensing accuracy and communi-cation distance. Such constraints are expressed in terms ofminimum power transmission coefficient τn,min

τn,min =pc,n(

λ4πr0

)2

PinGnGRχn

(11)

that assures the required read range r0, and the minimum spanof the measurable phase variation �ϕn,min

Δϕn,min =δϕ

δΨ(Ψ2 −Ψ1) (12)

that permits to achieve the required resolution δΨ in the mea-surement of the analyte [8], δϕ being the phase resolution ofthe reader.

The aim of the design is, therefore, a radio-sensor satisfyingthe following conditions:{

ϕn[ZL(Ψ2)]− ϕn[ZL(Ψ1)] ≥ Δϕn,min

τn[ZL(Ψ)] ≥ τn,min

(13)

for the whole range of the sensing process, i.e., ∀Ψε[Ψ1, Ψ2].

B. Positioning of Poles and Zeros

In general, a fully numerical pole/zero search procedure [18],aiming at shifting and rotating the nomograms with respect tothe sensor response �ZL(Ψ) can be applied to find the opti-mal parameters {z, PON/OFF } that satisfy the requirementsin (13). At each step of the search, self and mutual impedancerequired for the two-port tag are derived from (8) to (10) byusing the following formulas:

ZS,w =PONZON

C − z2

2z − PON − ZONC

(14)

ZM,w = ZS + z (15)

and conditions have to be enforced so that mutual impedancesare physically meaningful, as discussed in Section II.

A starting trial for the synthesis procedure may initiate fromthe following heuristic observations. In order to obtain reliablecommunication (i.e., high and stable τn) and fine sensing (i.e.,strongly variable ϕn) and thus satisfy the requirements in (13),it is beneficial that the path of �ZL(Ψ) is far away from the


Fig. 4. Examples of load paths �ZL(Ψ) over a complex ZL plane, respectively,for (a) resistive type sensor and (b) reactive sensor.

position of the zero where very low τn values are expected andwhere the communication should be accordingly very poor. Onthe other hand, having both τn and ϕn a common pole (PON ),a path of �ZL(Ψ) in its vicinity would guarantee strong phasevariation (as phase isolines are particularly dense in the vicinityof the pole) and high τn values.

The two extreme cases of resistive and reactive behaviorof the sensor can be considered as reference (Fig. 4), i.e., byinvestigating the configuration where the change of the analytemostly produces a variation of the resistance or of the reactanceof the sensor, respectively.

1) Resistive Sensor: as in case of thermistors [12], thesensor impedance will mostly move along a horizontal line�Z(Ψ) = RL(Ψ) + jXL,0 in the ZL plane along the sensingprocess [Fig. 4(a)]. It would be hence beneficial if PON andz had the same real part, i.e., they laid on the same verti-cal line in the ZL plane, while the imaginary parts are suchthat Im(PON )+Im(z)

2 ∼ XL,0. By doing so, the PON − τn axiswould be substantially vertical. Accordingly, there exists aregion of the plane where the τn isolines run parallel to �Z(Ψ)path so that τn barely varies, while phase exhibits strongchanges. This configuration will be thus advantageous both forsensing and communication.

2) Reactive Sensor: as in the case of permittivity sensors,the profile �Z(Ψ) = RL,0 + jXL(Ψ) will now mostly movealong a vertical line in the complex ZL plane.

Fig. 5. (a) Dipole antennas with T-match circuit used for the demonstration ofdesign method. (b) Manufactured dipoles with the varactor (connected to thevoltage generator through the cables visible on the port on the left) as variableload.

A favorable pole/zero placement is such that the pole reac-tance is equal to the barycenter of �ZL(Ψ), e.g., |XPON | ∼|(XL(Ψ1)−XL(Ψ2)| /2, and pole and zeroes have similar realpart, namely Re(z) ∼ Re(PON ). With this arrangement, the�ZL(Ψ) trajectory will fall in a region with high embeddedpower transfer coefficient and with strong phase variations[Fig. 4(b)], thus providing the wished behavior both for sensingand communication.

C. Geometrical Design of the Two-Port Antenna

Having optimally placed poles and zeroes, an electromag-netic optimization based on numerical simulation is finallyperformed to synthesize an actual antenna geometry that is ableto provide the wished ZS,w and ZM,w. Denoting with α ={α1 . . . αM} the set of geometrical parameters of the antennathat have to be optimized, the electromagnetic design will haveto minimize the difference between the simulated impedancesand the wished ones around the working frequency fRFID, byminimizing the following goal function:

F (α) = w1 |ZS − ZS,w|+ w2 |ZM − ZM,w| (16)

where w1 and w2 are weight coefficients.An example of the design of such wireless sensing system is

given in the next section.

V. NUMERICAL AND EXPERIMENTAL EXAMPLE

The proposed method is here demonstrated by using as two-port device a couplet of dipole-like tags, such as in Fig. 5(a).

The first dipole is connected to the NXP G2XM chip withTSSOP8 packaging (ZON

C = 16− j156 Ω, ZOFFC = 150 Ω at

fRFID = 870 MHz). The variable load ZL is emulated by a


Fig. 6. Reactance of the varactor versus voltage supplied by the externalvoltage generator.

voltage (Vg) controlled varactor that is connected to the termi-nals of the second dipole, in series with a low-value resistor(with RL = 5 Ω to take into account small parasitic resis-tances). A fully steerable reactance (Fig. 6) is hence obtainedso that the sensor impedance is ZL = 5 + jXL(Vg) Ω withXL ∈ [−300; −100] Ω (Fig. 6). The optimization of the sen-sor is restricted to the range 0 < Vg < 4 V so that XL ∈[−200;−100] Ω, thus exhibiting a sharper slope.

A. Design

The constraints for both minimum embedded power transfercoefficient and minimum phase variation as in (13) are fixed toτn,min = −5 dB and Δϕn,min = 50◦. By assuming a readerresolution of δϕ = 3◦, these requirements translate in a systemresolution δVg = δϕΔVg/Δϕn,min 0.25 V .

1) Pole/Zero Placement: Following the heuristic guidelinesin the previous section, optimal results satisfying the require-ments specified above were obtained for z = −40− j240 Ωand PON

w = −40− j150 Ω (POFFw = −22− j180 Ω), so that

Re(z) ∼ Re(PON ) and |XPON | ∼ |(XL(Ψ1)−XL(Ψ2)| /2,as suggested in Section IV-B1. Accordingly, the mutual andself impedance of the two dipoles were ZS,w = 26 + j192 Ωand ZM,w = −13− j48 Ω. Both ZS,w and ZM,w are compliantwith the electrical feasibility specified in Section II and hencecan be used in the goal function (16) to be minimized by thee.m. optimizer.

The corresponding ϕ-τn nomogram is shown inFig. 4(b) together with the trajectory of the varactor response�ZL(V ). Optimal performance is attained, as expected, in therange XL ∈ [−200;−100] Ω, i.e., in the proximity of the pole.The variation of phase and power transfer coefficient in suchzone (with RL = 5 Ω) is plotted in Fig. 7.

In the whole impedance range XL ∈ [−300;−100] Ω, i.e.,the full varactor range, two different regions can be observed:for XL < −200 Ω, the influence of the zero is strong and, there-fore, although a good phase sensitivity is recorded, low valuesof power transmission coefficient are produced. On the otherhand, for XL > −200 Ω, the impact of the pole becomes dom-inant, allowing for good matching (i.e., good values of τn), aswell as good variation for the phase (�ϕ/ΔXL ∼ 60◦/100 Ω).In such zone, therefore, the RFID couplet is best suited for

Fig. 7. Simulated ϕ and τn behavior of the optimized layout.

Fig. 8. Fitness function chart for the couplet of Fig. 5. The cross marks theposition of the optimal result.

optimal communication and sensing capabilities, as expectedfrom the theoretical results of Section II.

2) Electromagnetic Design of the Couplet Geometry: Theshape of the two dipoles was then designed in order to synthe-size the impedance matrix of the two-port network previouslyderived.

By fixing the mutual spacing among dipoles to d = 4 mmand considering the free geometrical parameters α = {b/a, l},the fitness function F (b/a, l) in (16) was evaluated (Fig. 8) forw1 = w2 = 0.5.

A useful result is achieved for b = 10 mm, a = 18 mm, l=122 mm, where ZS = 35 + j208 Ω and ZM = −6− j56 Ω atfRFID = 867MHz, thus pretty close to the required values(ZS,w = 26 + j192 Ω and ZM,w = −13− j48 Ω).

B. Prototype and Measurement

The manufactured RFID tags are shown in Fig. 5(b) togetherwith the cables to supply the varactor. The tags were placedover a wooden desk and interrogated through a Thingmagic M6reader [19], connected to a 6 dBi circularly polarized antennathat was placed at a distance of 90 cm above the desk. Phaseand turn-on power (useful to calculate the measured realizedgain as specified in [15]) were collected from the RFID reader.In order to calculate an average value of the phase signal, themeasurement was repeated 30 times per each voltage step atthe varactors terminals. Error bars in Fig. 9 represent, therefore,


Fig. 9. Measured ϕ and ˜Gn versus XL for the manufactured dipole antennacouplet (at fRFID = 867 MHz) for the varactor case.

the measured standard deviation of phase along each set of 30measurements.

Also voltage values provided by the voltage generator wererecorded, as they determine the response XL(Vg) of the varac-tor. The relationship between V and XL was shown in Fig. 6and is a fixed property of the specific varactor used. The mea-sured realized gain Gn = Gτn and phase response ϕn areshown in Fig. 9 versus the load reactance, together with thesimulated results for comparison purpose.

The measured values are in good agreement with simula-tion. In the useful shadowed area, the phase variation is morethan 50◦ that, assuming a reader resolution of 3◦, permits torecognize at most 15 levels of the phenomenon (here the var-actor voltage Vg). At the same time, the gain is rather stableand more than −5 dBi that means a read distance (in broadsideobservation) exceeding 4.5 m.

VI. CONCLUSION

This work has introduced a constrained synthesis methodol-ogy for phase-based RFID wireless sensing enabling to convert,in a controlled way, a variation of a load impedance con-nected at an RFID antenna into a wireless readable phasevariation. It has been shown, both numerically and experimen-tally, that it is possible to optimize a two-port RFID tag sothat both strong phase variation (i.e., superior sensing perfor-mance) and near-constant communication properties (i.e., verylimited degradation in read range) can be obtained for the givenimpedance range of the sensor. This method is of general appli-cation but requires the radiofrequency response of the sensor,namely the �ZL(Ψ) diagram, to be known. Since most of com-mercial off-the-shelf lumped sensors are fully characterizedonly in dc, the experimental determination of the �ZL(Ψ) curvebecomes the first step of the design.

It is moreover worth recalling that the proposed wirelesssensing modality, using the phase of the signal backscatteredby RFID tags, needs a fixed measurement setup (i.e., fixeddistance and orientation between reader and RFID couplet):this is tolerable in many application cases, such as labora-tory tests or with readers integrated into an infrastructure. Theextension to fully random reading procedures requires insteadadditional research that is being conducted currently by theauthors.

REFERENCES

[1] C. Perera, A. Zaslavsky, P. Christen, and D. Georgakopoulos, “Sensing asa service model for smart cities supported by Internet of Things,” Trans.Emerging Telecommun. Technol., vol. 25, no. 1, pp. 81–93, Jan. 2014.

[2] D. M. Dobkin, The RF in RFID—Passive UHF RFID in Practice.Waltham, MA, USA: Newnes, 2012.

[3] K. A. Hribernik, Z. Ghrairi, C. Hans, and K.-D. Thoben, “Co-creating theInternet of Things—First experiences in the participatory design of intel-ligent products with arduino,” in Proc. Int. Conf. Concurr. Enterprising,Aachen, Germany, Jun. 2011, pp. 1–9.

[4] D. J. Yeager, P. S. Powledge, R. Prasad, D. Wetherall, and J. R. Smith,“Wirelessly-charged UHF tags for sensor data collection,” in Proc. IEEEInt. Conf. RFID, Las Vegas, NV, USA, Apr. 2008, pp. 320–327.

[5] J. R. Smith, Wirelessly Powered Sensor Networks and ComputationalRFID. Berlin, Germany: Springer-Verlag, 2013.

[6] T. Unander, J. Siden, and H. E. Nilsson, “Designing of RFID-based sensorsolution for packaging surveillance applications,” IEEE Sens. J., vol. 11,no. 11, pp. 3009–3018, Nov. 2011.

[7] M. S. Khan, M. S. Islam, and H. Deng, “Design of a reconfigurableRFID sensing tag as a generic sensing platform toward the future Internetof Things,” IEEE Internet Things J., vol. 1, no. 4, pp. 300–310, Aug.2014.

[8] C. Occhiuzzi, S. Caizzone, and G. Marrocco, “Passive UHF RFID anten-nas for sensing applications: Principles, methods and classifications,”IEEE Antennas Propag. Mag., vol. 55, no. 6, pp. 14–34, Dec. 2013.

[9] C. Occhiuzzi and G. Marrocco, “Constrained design of passive UHFRFID sensor antennas,” IEEE Trans. Antennas Propag., vol. 61, no. 6,pp. 2972–2980, Jun. 2013.

[10] S. Caizzone, E. DiGiampaolo, and G. Marrocco, “Wireless crack moni-toring by stationary phase measurements from coupled RFID tags,” IEEETrans. Antennas Propag., vol. 62, no. 12, pp. 6420–6412, Dec. 2014.

[11] S. Manzari, C. Occhiuzzi, S. Nawale, A. Catini, C. Di Natale, andG. Marrocco, “Humidity sensing by polymer-loaded UHF RFID anten-nas,” IEEE Sens. J., vol. 12, no. 9, pp. 2851–2858, Jun. 2012.

[12] S. Manzari, S. Caizzone, C. Rubini, and G. Marrocco, “Feasibility ofwireless temperature sensing by passive UHF-RFID tags in ground satel-lite test beds,” in Proc. IEEE Conf. Wireless Space Extreme Environ.(WiSEE), Oct. 2014, pp. 1–6.

[13] R. A. Potyrailo et al., “Selective quantitation of vapors and their mixturesusing individual passive multivariate RFID Sensors,” in Proc. IEEE Int.Conf. RFID, Orlando, FL, USA, Apr. 2010, pp. 22–28.

[14] G. Marrocco, “RFID grids—Part I: Electromagnetic theory, IEEE Trans.Antennas Propag., vol. 59, no. 3, pp. 1019–1026, Mar. 2011.

[15] S. Caizzone and G. Marrocco, “RFID grids—Part II: Experimentations,”IEEE Trans. Antennas Propag., vol. 59, no. 8, pp. 2896–2904, Aug.2011.

[16] S. Caizzone, E. DiGiampaolo, and G. Marrocco, “Passive RFID cou-plets as wireless interface for sensor applications,” in Proc. Eur. Conf.Antennas Propag. (EUCAP), Apr. 2015, pp. 1–4.

[17] C. Balanis, Antenna Theory: Analysis and Design. Hoboken, NJ, USA:Wiley.

[18] M. H. Tu and C.-M. Lin, “Synthesis of pole/zero assignment control lawwith minimum control input,” IEE Proc. Control Theory Appl., vol. 139,no. 3, pp. 291–295, May 1992.

[19] [Online]. Available: http://www.thingmagic.com/index.php/embedded-rfid-readers/mercury6e

Stefano Caizzone received the M.Sc. degree intelecommunications engineering and the Ph.D.degree in geoinformation from the University ofRome “Tor Vergata,” Rome, Italy, in 2009 and 2015,respectively.

Since 2010, he is with the Antenna Group ofthe Institute of Communications and Navigation ofthe German Aerospace Center (DLR), Wessling,Germany, where he is responsible for the devel-opment of innovative miniaturized antennas. Hisresearch interests concern small antennas for RFIDs

and satellite navigation, antenna arrays and grids with enhanced sensing capa-bilities, and controlled radiation pattern antennas for robust satellite navigation.


Emidio DiGiampaolo received the Laurea degreein electronic engineering and the Ph.D. degree inapplied electromagnetics from the University ofL’Aquila, L’Aquila, Italy, in 1994 and 1998, respec-tively.

From 1998 to 2004, he was a PostdoctoralResearcher with the University of L’Aquila. In thespring of 2000, he was Visiting Researcher at theEuropean Space Research and Technology Centre(ESTEC), Noordwijk, The Netherlands. From 2005to 2009, he was a Researcher with the University of

Rome Tor Vergata, Rome, Italy. Since 2010, he has been with the Universityof L’Aquila, as an Assistant Professor. His research interests include numer-ical methods for modeling radio-wave propagation in complex environments,antennas, RFID, and radio localization.

Gaetano Marrocco received the Laurea degreein electronic engineering and the Ph.D. degree inapplied electromagnetics from the University ofL’Aquila, L’Aquila, Italy, in 1994 and 1998, respec-tively.

In 1994, he was with the University of Illinoisat Urbana-Champain, Champaign, IL, USA, asa Postgraduate Student. In 1997, he joined theUniversity of Roma Tor Vergata, Rome, Italy, as aResearcher. In 1999, he was a Visiting Researcherat the Imperial College, London, U.K. Currently he

serves as Associate Professor of Electromagnetics with the University ofRoma Tor Vergata, Rome, Italy, and Chairs the Pervasive ElectromagneticsLaboratory. In 2014, he received Full Professor qualification. In 2015, he wasa Guest Professor with the University of Paris-est Marne la Vallée, Marnela Vallée, France. He was active in the development of FDTD methods formicrowave hyperthermia and in the modeling and design of pulsed arrays. Hisresearch interests include application of radiofrequency identification (RFID) tomedical and industrial diagnostics as well as to the modeling and design ofdistributed and miniaturized conformal antenna clusters over ships, micro andnanosatellites and aircrafts within the framework of European Space Agency,NATO, Italian Space Agency, and the Italian Navy research projects.

Dr. Marrocco serves as an Associate Editor of the IEEE Antennas andWireless Propagation Letters, as Associate Editor of the IEEE RFID VIRTUAL

JOURNAL and as a Vice-Chair of the Italian delegation URSI Commission D:Electronics and Photonics. He was the Chair of Local Committee of EUCAP-2011 in Roma and TPC Chair of the 2012 IEEE-RFID TA in Nice, France.He is the Co-Founder and President of the University spin-off RADIO6ENSEthat is active in the short-range electromagnetic sensing for Industry, Internetof Things and Smart Cities.

Constrained Pole-Zero Synthesis of Phase-Oriented RFID Sensor Antennas · 2016-04-08 · Constrained Pole-Zero Synthesis of Phase-Oriented ... The basic operation principle[8] ...

Documents