IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO.

3, JUNE 2019 5387

The Realization of Chipless RFID Resonator for

Multiple Physical Parameter Sensing
Tharindu Athauda and Nemai Chandra Karmakar, Senior Member, IEEE

Abstract—We present the design of asymmetric circular split parameters in the environment. A chipless RFID sensor gen-
ring resonator as a chipless radio frequency identification (RFID) erally consists of an RFID tag and a smart material which acts
sensor tag design that can be deployed to measure multiple as the sensor element that changes its dielectric properties, for
physical parameters wirelessly in ultrawideband frequencies.
Secondarily, the proposed tag design can extend to be used as a given physical parameter in the environment. Chipless RFID
a chipless RFID tag with actual data encoded for product identi- has been deployed as individual sensors to detect humidity [3],
fication purposes. The proposed resonator design consists of four temperature [4], applied pressure [5], pH [6] as well as gas
complementary split circular rings. Each gap between the rings concentration [7]. However, the use of a single chipless RFID
acts as a capacitive sensor which is optimized to have a resonance sensor for multiple physical parameter sensing has not been
with a high-quality factor. Each resonance can be considered as
a sensor to deploy smart materials that are sensitive to differ- thoroughly investigated.
ent environmental physical parameters, which can be operated Chipless RFID sensing tags can be designed to resonate in
in wireless conditions. The fabricated chipless RFID sensor pro- a given frequency band by changing the elements of shape,
vides a higher Q factor, aligned with the optimized simulation size, and composition of the tag design [8]. The resonance
model. Chipless RFID sensors are comparatively low-cost, but frequency response is a key element of a chipless RFID sensor
they provide enormous design flexibility in diversified applica-
tions. The proposed design holds a great promise for sensing as the changes occurring in the induced electrical field can
multiple physical parameters in a remote setting. augment the response; hence, it is a valuable tool to determine
the sensor performance in microwave frequencies.
Index Terms—Chipless radio frequency identification (RFID)
sensor, chipless RFID tag, quasi-static equivalent model, remote Various attempts have been recorded, to develop an inte-
sensing, split ring resonator (SRR). grated sensing platform in the ultrawideband (UWB) domain.
Fletcher suggested a temperature sensor based on ELC
resonator [4] and a magnetic material-based chipless RFID
temperature sensor [9]. Fletcher also suggested the use of three
I. I NTRODUCTION different layers of magnetic materials which change the mag-
OOD waste is a critical concern in the food industry netic spectrum with temperature. However, the fabrication of
F where 20% of all food produced ends up in landfills due
to compromise of food freshness and supply chain vulnera-
such a multilayer magnetic material-based sensor is considered
to be challenging due to the complexity of fabrication; hence,
bilities. By incorporating technology to detect food freshness, the physical prototype has not yet been realized. Furthermore,
especially for packaged food, smart packaging can play a very a passive surface acoustic wave (SAW)-based RFID temper-
important role in reducing food waste [1]. Smart remote sens- ature and pressure sensor were also presented by Buff [10].
ing using RFID has gained a huge traction in research and In this SAW sensor, a physical or chemical influence changes
industries and the sensors are able to measure the influ- the propagation characteristics of the SAW sensor, influencing
ence of intrinsic and extrinsic factors that can harm food the frequency response of the device. SAW sensor is also con-
freshness. However, the current RFID- based food freshness sidered to be the only commercially available chipless RFID
sensing technologies are expensive and dependent on a single sensor product to date; however, the nonplanar behavior and
physical parameter which only provides a 1-D view of intrin- fabrication limitation has restrained its wide reach to the main-
sic/extrinsic changes taking place inside a food package [2]. stream market. Therefore, the realization, as well as the cost
A sensing technique which could measure several intrin- of chipless RFID multiparameter sensors, has not been com-
sic/extrinsic parameters at a single time may help improve mercially viable and also the increased complexities restrained
the food security at a reduced cost. the use of sensors in real-world applications.
Radio frequency identification (RFID) sensor technologies In this paper, the focus is given to the optimization of
are renowned for their use in detecting and measuring physical chipless RFID sensor design and the radio frequency (RF)
characterization of the tag for multiple resonators with a higher
Manuscript received October 29, 2018; revised November 28, 2018, Q factor. This paper further elaborates how each resonator
January 8, 2019, and February 4, 2019; accepted February 20, 2019. Date
of publication February 25, 2019; date of current version June 19, 2019. can be used to measure individual environmental parame-
(Corresponding author: Tharindu Athauda.) ters in a remote setting. This analysis further emphasizes the
The authors are with the Department of Electrical and Computer design criteria to select split ring resonator (SRR) as a chip-
Systems Engineering, Monash University, Clayton, VIC 3800, Australia
(e-mail: tharindu.athauda@monash.edu; nemai.karmakar@monash.edu). less RFID sensor, the lumped circuit model, and the optimum
Digital Object Identifier 10.1109/JIOT.2019.2901470 RF performance of the fabricated resonator.
2327-4662 c 2019 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.
Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 5388 IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO. 3, JUNE 2019

II. D ESIGN C RITERIA FOR C HIPLESS RFID S ENSOR

The resonant circuit which is used in chipless RFID sens-
ing comprises of lumped discrete elements such as inductors,
resistors, and capacitors. These elements also play a key role in
chipless RFID sensing as the lump circuit elements correspond
to the surface currents on the tag, induced by the electrical
and magnetic fields by the electromagnetic energy transferred
through the reader antenna. Smart materials [11] which are
inherently sensitive to certain environmental condition (e.g.,
humidity and temperature) can be used to detect changes in
the environment. The selected smart materials trigger changes
in its dielectric properties, based on the existence or thresh-
old level of certain environmental condition, which results in Fig. 1. Asymmetric circular four split ring resonator (ACiSRR).
a change in resonance frequency response. Such a change will
reflect in the frequency response, either a change in amplitude
and/or shifting of the frequency response from its original res-
onance frequency, which can be used to determine the changed
behavior of the given physical parameter [12]. Therefore, the
selection of appropriate smart materials also plays importance
in chipless RFID sensor design.
SRRs have also been known as a metamaterial [13], which
showcases negative refractive properties [14]. SRR was also
used in sensing application [15], due to its symmetrical prop-
erties, capacitive behavior and also used in experiments related
to wave dispersion, relative permittivity [16], and quality fac-
tor (Q factor) measurements [17]. In this paper, the SRR has
been selected as the chipless RFID tag architecture, due to
SRR’s ability to provide multiple capacitive resonators. Each
gap between rings can be used as an independent capacitive
sensor. Theoretically, (n) a number of circular rings can allo-
cate (n − 1) number of capacitors and this principle was used
to develop multiple capacitors in one resonator design. In this
paper, the development of four SRRs with asymmetric splits
and the analysis of the resonator design as a multiparameteric
sensor (as shown in Fig. 1) are discussed.
The mean circumference of any ring resonator should be
equal to the integral multiple of the wavelengths. By cutting Fig. 2. (a) Chipless RFID sensor excitation using a plane wave using CST
a split in a closed ring resonator, the total wavelength that Microwave Studio 2016. (b) Measurement (S11) setup of chipless RFID tag
using patch antenna with VNA.
is used to excite the whole ring becomes half wavelengths
while creating a weak capacitance at the split [18]. In this
asymmetric circular SRR (ACiSRR), when the external elec- dipoles electrically and vice-versa; however, a magnetic field
trical/magnetic field applies, it generates an electrical field is known to be the dominant excitation mechanism [17].
around the rings which induces current on ring’s surfaces. ACiSRR has multiple transmission lines; hence, multiple
Fig. 2 elaborates the excitation mechanism of the chipless modes are generated according to the strip lengths of each
RFID sensor using a plane wave and then practical setup to micro-stripline. This phenomenon generates multiple reso-
measure return loss (S11) with the patch antenna and vector nances which are elaborated as shown in Fig. 3, using its
network analyzer (VNA). Patch antenna acts as a trans-receiver RCS value.
which sends excitation signal and receive it back then shows RCS response of ACiSRR in Fig. 3 implies that there
the S11 through the VNA. exist three distinct notches with a higher Q factor at 3.86,
The split stops the current flow across the ring; which helps 4.6, and 5.75 GHz frequencies, respectively, which occurs
to create two equal capacitors in both left and right as well due to the three gaps in between rings. This scenario can
as a weak capacitor at the split. On the other hand, the time- be showcased by the analysis of surface currents as shown
varying electrical field can also be able to excite SRR based in Fig. 4(a) below. The highest reflection occurring at the
on the electric charge distribution among half rings. frequencies where the microstrip length is complemented with
Each half ring adjacent to each other acts as two paral- the corresponding λ/2; therefore, the peaks were recorded at
lel dipoles; hence, they have the ability to induce the current 3.7, 4.35, and 5.3 GHz, respectively. Fig. 4 shows the induced
in presence of electric field. Such phenomena lead to cross surface current at 3.7 GHz when the plane wave was excited
polarization effects as the given magnetic field can polarize as shown in Fig. 2(a). For this analysis, CST Microwave

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 ATHAUDA AND KARMAKAR: REALIZATION OF CHIPLESS RFID RESONATOR FOR MULTIPLE PHYSICAL PARAMETER SENSING 5389

Fig. 5. ELC resonator and spiral resonator designs.

Fig. 3. Simulated RCS response of ACiSRR using CST.

Fig. 6. RCS response of ELC resonator versus spiral resonator versus

SRR (ACiSRR).

Fig. 4. Surface current at (a) 3.7 GHz, (b) 4.35 GHz, and (c) 5.3 GHz. TABLE I
K EY F RESHNESS I NDICATORS

Studio 2016 was used and field monitors [19] were introduced.
Field monitors help to understand the specific electrical and
magnetic field behavior on a given frequency.
As shown in Fig. 4, the resonance characteristics of the
SRR are correlated with the surface current concentrations.
The maximum surface current density is shown along the
gaps between ring resonators (denoted by the red color) and Especially when it comes to food safety, the proposed sensor
comparatively low current density recorded on top of the may have the potential to measure multiple parameters which
microstrip lines. This observation helps to interpret that the are key freshness indicators in perishable food products such
higher the reflection coefficient, the lower the concentration as humidity, temperature, and pH.
of electromagnetic power; hence, the capacitance. Therefore,
the concentration of EM power implies the lack of reflec-
tion of electromagnetic waves resulting weak attenuation at III. I NVESTIGATION OF S MART M ATERIALS AND
the receiver’s end. The analyses of the surface current of the A NALYSIS OF C HIPLESS RFID S ENSOR B EHAVIOR
ACiSRR provided a cohesive correlation between capacitive As proposed in the previous section, the suggested ACiSRR
properties and surface currents. design is capable of generating three notches where each notch
From the comparison standpoint, spiral resonator design, as acts as a capacitive sensor. In the food safety applications,
well as popular ELC resonator RF performances, were also where humidity, temperature, and pH measurements are key
analyzed. The main reason (Figs. 5 and 6) behind this com- to guarantying the food quality [20], [21] as extrinsic measures
parison is to highlight the better Q factor values produced by (Fig. 7), the following suggestions were made.
the ACiSRR in contrast to other resonators. 1) Environment Condition: As the plan was to measure
As visually explained in Fig. 6, SRR (ACiSRR) stands bet- food safety using the proposed sensor the following
ter in contrast to Q factor as well as RCS value. ELC has parameters with thresholds in Table I were established.
very low RCS which may due to the smaller size as well as As per Table I, if the sensor exposes to room tempera-
thinner widths of the microstrip lines. Each line in the spiral ture the dielectric constant of the rGO changes; hence,
resonator are continuous hence unable to make multiple reso- it resulting in a change in permittivity [16] which even-
nances; however, it shows better Q factor in contrast to ELC tually impacts the frequency response. Similar scenarios
resonator. However, SRR still has better performances when can be applied to humidity and pH measurements as
comparing with the RCS values of spiral resonator. well.
Apart from its RF performances, the proposed ACiSRR tag 2) Selection of Smart Material: Three distinct graphene
also considers the following key factors in UWB frequency compounds were selected and their respective dielectric
such as simplicity, low-cost, robustness and scalability. properties were elaborated in Table II.

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 5390 IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO. 3, JUNE 2019

TABLE II
S MART M ATERIALS AND D IELECTRIC P ROPERTIES

a) Graphene oxide [22] for humidity sensing.

b) Reduced graphene oxide for temperature sensing. Fig. 7. Key freshness factors in perishable food items.
c) Chitosan with graphene oxide for pH sensing.
In the analysis of permittivity [16], it is important to under-
stand the measurement of permittivity, and the free space
method is considered to be more promising as a method of
measurement of permittivity in a remote sensing setting
Complex Permittivity(ε) = ε − jε
where ε relative complex permittivity; ε relative real permit-
tivity (dielectric constant); ε relative imaginary permittivity.
It is also equally important to recognize that the deriving
of the values of dielectric constants and loss tangents are sub-
jected to change based on the preparation method of the smart
material. As an example, GO can vary its dielectric constant Fig. 8. Simulated frequency domain response with sensing material
substantially based on its preparation method. The values men- (Graphene oxide) between outer and second ring.
tioned in Table II, some of which were derived from theoretical
models as well as from prior experiments conducted with spe-
cific smart materials. Thereafter, these modeled values were
implemented to the CST simulation model, where the respec-
tive smart materials were deployed into the capacitive gaps
(between rings and split) of ACiSRR and then analyzed using
their respective frequency responses.
In the initial sensitivity analysis, GO was deployed in
between the outermost ring and the second ring. Thereafter,
the sensor was excited by a plane wave [Fig. 2(a)] and then
changed the dielectric property of the GO from 12 to 16, and
loss tangent 0.25–0.4, respectively, and RCS responses were Fig. 9. Simulated frequency response of ACiSRR with and without smart
obtained as shown in Fig. 8. materials GO, rGO, and chitosan with dielectric constants 12, 6.4, and
In Fig. 8, a clear frequency shift in all resonances can be 6, respectively.
observed; however, a significant shift (from 3.86 to 3.26 GHz)
was recorded of the resonance where the smart material (GO)
was deployed. This concludes that the capacitance has been in the frequency response to the left. The most likely rea-
significantly changed by the deployment of smart material son for such a frequency shift is the increase in material
in between rings. In this analysis, the capacitance generated capacitance between rings, which has increased the dielectric
by the split gap has considered negligible over the overall constant. Such an increase in dielectric constant also increases
capacitance of the ACiSRR. the overall insulation properties of the ACiSRR.
Similarly, both rGO and chitosan materials were also along Thereafter, changes in dielectric constant and loss tan-
with GO deployed to the model to understand the baseline gent were incorporated into the simulation model and similar
of the sensor prior to exposure to the different environmen- approach was followed for all three materials. Finally, the over-
tal condition as shown in Fig. 9. In this scenario, GO was all change in the RCS response was recorded as shown in
deployed to the outermost gap and chitosan deployed to the Fig. 10. As per Fig. 10, it can be seen that the humidity sen-
most inner gap between rings. sor further reduces its amplitude and shift to the left due to the
As shown in Fig. 9, the simulation results showcase that increased dielectric properties. Due to this reason, the overall
the deployed smart material has been made a significant shift insulation property increases as well. On the other hand, both

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 ATHAUDA AND KARMAKAR: REALIZATION OF CHIPLESS RFID RESONATOR FOR MULTIPLE PHYSICAL PARAMETER SENSING 5391

Fig. 12. Theoretical LC model calculation using ADS from Keysight

technologies.

Fig. 10. Simulated frequency response of ACiSRR with smart materials in

different environmental conditions (humidity, temperature, and pH).

Fig. 11. ACiSRR lump circuit model.

Fig. 13. Theoretical S parameter plot using ADS.

rGO and chitosan enhance their conductive properties by shift-
ing to the right. Moreover, reduction of loss tangent also helps
both rGO and chitosan to increase the amplitude moderately. element as shown in Fig. 12. The optimum Q factors values
The overall analysis explained in Fig. 10, can also be used were obtained using the simulations and tunning functions
to implement individual sensors. It elaborates the possible available in the Advanced Design System (ADS) modeling
frequency response changes in all sensors at once; however, software. Considering the lump values and the Q factors, the
individual sensor modeling can be done to forecast possi- S11 parameter was obtained as shown in Fig. 13.
ble changes in the frequency response. The sensor outcome As per Fig. 13, the theoretical values along with Q factors
also depends on its sensitivity to the application environ- have shifted all three resonances to the left. However, toward
ment; hence, the control of the environment is essential to the higher frequencies, the shift has minimized. The likely
get accurate sensory responses. reason for such phenomena is the lower Q factors in inductors
and capacitors in outer rings with respect to higher Q factors
IV. T HEORETICAL M ODELING in the inner rings of the ACiSRR design.
In the development of the theoretical model, the basis of
The understanding of the LC model is important to inves-
split resonator where the split gap divides the ring capaci-
tigate the change in capacitive properties in relation to the
tance into two equal capacitors (distributed capacitance) was
permittivity of the given substrate. The equivalent LC cir-
considered [28]. The overall length of the resonator consid-
cuit for the suggested can be deduced as follows in Fig. 11.
ered for the total inductance of the LC model [29]. In general,
Usually, the LC model (the equivalent circuit model) consists
SRR when two rings are excited in their fundamental mode,
of inductance, capacitance, and resistance [27]; however, in
the distributed capacitance in one half is in series with the
this LC model for ACiSRR, only the inductance and capac-
other half based on their current and voltage distributions.
itors were considered to keep the overall model simple. The
Therefore, every two rings represented two series capacitance
inductance of the system generates as a resistant force due to
while adjacent rings create parallel capacitance. These estab-
the induced current flow in the microstrip by the transmitted
lished hypotheses were used in developing the LC model as
signal. Similarly, capacitance generated by the cavity located
elaborated in Figs. 11 and 12. The mutual inductance between
in between the microstrip ring resonators. The capacitance in
parallel stems and the mutual capacitance among nonadjacent
the half rings of the ACiSRR needs to be carefully calculated
rings were excluded in the current model to keep the LC model
as the capacitance is heavily influencing the relative permittiv-
simple.
ity and the frequency response. Apart from the tag dimensions
The model explained in Fig. 11, used several equations
the substrate permittivity, loss tangent, and the thickness of the
listed below in order to come up with respective inductance
substrate, have significant influence over the total inductance
and capacitive values for the LC model
and capacitance.    
The development of the ideal theoretical model does not μ0 lavg 0.98
Total Inductance(LT ) = 4.86 ln + 1.84ρ .
justify the real world environment; hence, further analy- 2 4 ρ
sis has carried out to implement Q factors into each lump (1)

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 5392 IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO. 3, JUNE 2019

TABLE III
In (1), LT stands for total inductance where μ0 is the vac- C ALCULATED A MPLITUDE (dB) VALUES F ROM
uum permeability, and lavg is the average strip length which THE T HEORETICAL M ODEL
can be calculated using the following equations [30]
lavg = 4[l − (N − 1)(w + s)](2) (2)
where w is the width of the strips and s is the gap between
rings. N stands for the number of rings where N = 4 in this
particular ACiSRR design. ρ stands for the filling ratio as
elaborated in (3) [29]
(N − 1)(w + s) TABLE IV
ρ= . (3)
[l − (N − 1)(w + s)] TACONIC -TLX8 S UBSTRATE M ATERIAL P ROPERTIES

Finally, the total inductance was calculated as 4.72 nH.

Moreover, (4)–(8) were used to calculate the total capacitance
of the LC model
(C1 + Cgap1 )(C2 + Cgap2 )
CT =   (4)
C1 + Cgap1 + (C2 + Cgap2 )
where CT denotes the total capacitance of the two capacitors
in series and Cgap provides the capacitance of each split gap.
A similar approach can be used to calculate the capacitance Table III showcases the different amplitude values obtained
of other series branches for C3–C6. Each capacitance above from the notches in comparison to values in the simulation.
(except split gap) can be calculated using (5). In this calcu- These values in Table III emphasizes that the amplitude of
lation, per unit length capacitance between two parallel strips the RCS response of the simulation is low in comparison to
was considered √ values from the theoretical model in the ADS. This can be
K 1 − k2 explained that the ADS model represents the ideal scenario
Cn = ε0 (5) where ideal inductance and capacitors were used to calculate
K(k)
frequency resonances. There are more factors to be considered
where k stands for the elliptic integral of the first kind and such as resistivity of the elements, dielectric losses which were
ε0 for the relative permittivity of air. The value for k can be not represented in the theoretical model. However, resonance
found using the following equation: frequencies provide a closer approximation in both simula-
s
tion and the theoretical model. The analysis of the ACiSRR
k= 2
. (6)
w+ s
2
chipless sensor tag in CST simulations for its multiple mode
resonances and the LC model derived from theoretical calcula-
The capacitance of the split gap can be calculated as in (9),
tions have provided an established foundation of the ACiSRR
where g stands for the split gap distance. The value of g varies
tag’s expected RF performance. It can also conclude that
among each ring resonator which results in different capacitive
multiple smart materials can be deployed into the ACiSRR
values. The surface area of each ring is calculated using w
sensor design which is useful for single or multiple physical
(width of the strip) and the height of the conductive layer (h)
parameter sensing applications. These proposed sensors are
ε0 wh useful for food safety application as elaborated in Section III.
Cgap = . (7)
g The interesting alignment of such smart materials can also be
Once the all gap capacitors and the capacitance between deployed for other sensor-based applications such as gas sens-
each ring resonator is calculated, the total capacitance can be ing, air quality monitoring, precision agriculture, health care,
calculated as and well-being, which further emphasizes the usefulness of
this microwave resonator design.
Ctotal = CT1 + CT2 + CT3 . (8)
Following all above analysis, the promising performances
These CT1 –CT3 were used along with calculated inductance of the ACiSRR sensor design obtained from simulation mod-
to develop LC model in ADS software to verify the reso- els have helped to make a decision to fabricate the sensor
nance frequencies as shown in Fig. 10(a). The relationship and measure tag’s RF measurements as elaborated in the next
between LC and resonance frequencies of an LC model can section.
be calculated using the following equation:
V. FABRICATED P ERFORMANCE A NALYSIS
1 1
Resonant Frequency(f0 ) = . (9) The ACiSRR sensor tag was developed using Taconic-
2π LT CT
TLX8 [31] substrate which has low loss tangent in comparison
The derived resonance as in Fig. 12, are well aligned with to other Teflon-based dielectric substrates. The properties of
the frequencies obtained from the simulation, which concludes the substrate used are elaborated in Table IV. Low loss tan-
that the LC circuit model is aligned with the ACiSRR chipless gent (tanθ ) is useful in sensor development in order to achieve
sensor design. a high Q factor in S11 measurements. Thereafter, the ACiSRR

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 ATHAUDA AND KARMAKAR: REALIZATION OF CHIPLESS RFID RESONATOR FOR MULTIPLE PHYSICAL PARAMETER SENSING 5393

Tag was fabricated and return loss (S11) measurements were

taken as shown in Fig. 15(a) and (b), respectively. To measure
the performance of the CSRR tag, a single patch antenna was
used [as elaborated in Fig. 2(b)] in a laboratory environment
using a VNA. In the calibration of the chipless RFID tag
in the measurement environment, we consider that a single
antenna will act as both transmitter and receiver as shown in
Fig. 2(b). Moreover, as the different resonators in the ACiSRR
resonate at distinct frequencies, the transmitted interrogated
pulse from the patch antenna is immediately reflected from
the tag, another part of the substrate as well as from the other
objects in the environment. The total received signal can be Fig. 14. Measurement of ACiSRR on Taconic-TLX8 substrate and experi-
modeled by considering the following formula: mental setup.

Total received Signal (Yt ) = Yr + Ys + Ya . (10)

In this Yr , Ys , and Ya stand for the return loss profile of
the antenna, structural mode RCS, and antenna mode RCS,
respectively [32]. The Yr component will gradually become
redundant once the transmission is completed as there is no
reflection due to return loss. It is also to note that, the pres-
ence of a tag in front of the antenna has slightly changed
the original S11 measurement of the antenna as the S11 is
now influenced by the backscattered incident from the tag.
Therefore, the antenna can now be considered as loaded by the (a)
chipless RFID tag and can denote using the following formula:
Y(t) = F −1 S11loaded (f )X(f ) . (11)
And from (11), now we can derive the actual reflection by
the tag as follows:

Ys + Ya = F −1 S11loaded (f ) − S11(f ) X(f ) . (12)
This understanding is used to calibrate the tag against
the antenna or the same principle applies when considering
S11 measurements with tag and without the presence of the (b)
tag as follows:
  Fig. 15. (a) Measured reflection coefficient (S11) of Taconic-TLX8 ACiSRR
Ys + Ya = F −1 S11Lwithtag (f ) − S11Lwithouttag (f ) X(f ) . chipless RFID sensor tag using a single patch antenna. (b) Simulated RCS
response versus measured S11.
(13)
In the actual measurements, the measured antenna S11 the distance increases, unless the increase in source power.
(loaded with tag) from the setup shown in Fig. 14 was the plot When the power received by the tag reduces, it impacts the
against the RCS value obtained from CST Microwave Studio signal which is backscattered, hence the amplitude levels of
simulation model. The distance between the patch antenna the S11 measurement. In this measurement, the reading range
and the fabricated tag kept 3 cm; however, same performance can easily be increased up to 10 cm as the RCS value obtained
can be maintained even at 10–15 cm when we use compara- from the simulations accounted 10 cm as the distance between
tively larger tag size. The understanding of reading range can the tag and the antenna. However, the inherent fabrication
be further explained using the radar range equation [33] as errors and the loss factors have shifted the tag response as
follows: per Fig. 15(b) which can even shift further upon the use of
  smart materials for sensing. In such a case, there is a chance
er Dant 2 λ2
Ptag = Lrcv 1 −  × 2
σtag Psrc (14) that shifted resonances may overlap each other due to the
4π R2 4π shift which needs to be avoided in practical use. This is the
where Lrsv , Psrc , and  stands for coupler loss in the antenna, main reason to keep tag close to the patch antenna in this
transmitted power at the source, and complex reflection coef- experiment.
ficient, respectively. Additionally, σtag , Dant , er , λ, and R stand As observed in Fig. 15, the resonances appeared in the mea-
for RCS of the chipless RFID tag, antenna directivity, radia- sured frequency response are slightly shifted in comparison
tion loss of the antenna, the wavelength, and most importantly to the RCS response. However, it can also observe that the
the distance between the antenna and the tag. As it shows in Q factor has reduced in the measured response in contrast to
the equation, power received by tag reduces drastically when the simulated frequency response. A possible reason for such

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 5394 IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO. 3, JUNE 2019

an outcome can be the fabrication errors and minor unevenness

in the Taconic substrate itself. It has also been seen that the
amplitude level has reduced in the S11 measurement. However,
the resonances are clearly visible; hence, the outcome can
be measured without deploying advanced signal processing
techniques.
There may be an argument can be made, the key simi-
larity of RCS and S11 (reflection coefficients) measurements
according to Fig. 15. It can be considered as similar when
it comes to the chipless RFID backscattering principle. The
RCS was calculated using the amount of reflection that the
Fig. 16. S11 measurements in various read range.
conductive elements of the chipless RFID can reflect back to
the source using a plane wave excitation mechanism. In such
a scenario, the total incident power transmitted is calculated efficiency of the power transmitted to the tag antenna which
against the power scattered through the unit angel reflection may also increase the ability to get a higher reflection back
as the RCS. This phenomenon is prominent when it comes to to the receiver. The noise level of the measurement setup was
the far field as the self-interference from the electric field has −60 dBm. For example, use of a horn antenna increases the
a minimum impact over the transmitted/reflected electric field. reading range almost twice as the current reading distance;
This understanding has helped to elaborate reflection coeffi- however, the main expectation of this experiment is to use
cient in the actual measurement and its correlation RCS. The practical experimentation model that supports the truly wire-
definition of return loss is hypothetically equal to the reflec- less nature of the tag when it is used as a chipless RFID
tion coefficient [34]. The return loss is the reflected scattering sensor in the real world. Furthermore, RCS (σ ) value of the
power which is similar to RCS when it comes to reflection tag is 625 mm2 as the main intention of the tag development
per given unit length/direction. Therefore, it can be reasonably is to keep the tag size as small as possible to lower the cost
assumed that RCS and S11 are same when it comes to remote of the tag as possible. The increased tag size will certainly
measurements given the consideration that both recognized as increase the conducting surface areas which help to achieve
far-field measurements. higher reading range. Path loss (4π ) is also considered in this
In order to get rid of noise clutter, a further assumption calculation. By inputting all the above values, it was found that
is taken that in both simulation and S11 measurements, clut- the maximum reading range of the antenna is approximately
ter remains as statistically independent. The simulation model 8 cm. The frequency sweep is used as the interrogation signal
itself model the noise level in the environment given the throughout the experiment. There are some other approaches
dielectric values and loss factors of the environment and tag is also available to provide different approaches to find read
materials which are actually used in the measurements in real range when using multiple antennas and [37], [38]; however,
life scenario. The reading distance which is also proportional the current use of Friss free space equation provides a more
to the noise (far the distance, the lesser the reflection comes suitable theoretical approximation to the measurement setup.
back to the source) is also considered in the simulation model The few more measurements were taken to validate the read-
as if in the real measurements. This is one of the reasons ing range and following S11 responses were obtained as in
the RCS value comes very close to S11 measurements; how- Fig. 16. As it can be seen that even at the 10-cm distance the
ever, there is always an error factor which can cause certain three resonances are clearly visible irrespective of the changes
deviations. This error may happen due to the undesirable envi- in the amplitude levels and frequency shifts.
ronmental reflections beyond the simulation model and also The increasing distance also increases the background
happen due to the fabrication errors. reflection. In an ideal event only the amplitude should get
Moreover, the analytical calculations are carried out to sup- decrease with the increased distance; however, in practical
port the limitation exist in the chipless RFID reading system measurement notches can also shift due to the changing dielec-
elaborating the key parameters used in this measurement tric properties of the environment as this is not a closed system
according to the Friss free space equation [35] as follows: during measurements and not additional steps taken (e.g., foam
 1/4 to absorb noise) to reduce noise as this should mimic the real
GT GR λ2 PT
Reading range(R) = σ (15) world measurements. Moreover, the near field and far filed
(4π )3 PR distances are also calculated [39] using the antenna dimen-
where the GT and GR stands for the gain of the trans- sions (3-cm diameter) and it is safe to assume that these
mitting and receiver antenna and it is 6 dBi as the same measurements are considered as far-field measurements. Far-
patch antenna is used as a trans-receiver. The wavelength field measurement comes to 3 cm which is the exact distance
considered for the frequency of 5 GHz and PT the transmit- that is used for the measurements of the chipless RFID tag.
ted power of the antenna is 0 dBm according to the FCC Furthermore, as shown in Fig. 17, another simulation was
regulations [36]. PR value is considered as −50 dBm based on carried out to see the performances of the chipless RFID tag
the sensitivity of the receiver antenna which can certainly be design, when one more ring added to the design. The reason
improved by using a horn antenna which has a higher sensitiv- behind such model development is to understand the system
ity. Horn antenna usually has a higher gain, thus increases the capacitance when more and more rings added to the design.

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 ATHAUDA AND KARMAKAR: REALIZATION OF CHIPLESS RFID RESONATOR FOR MULTIPLE PHYSICAL PARAMETER SENSING 5395

TABLE VI
M EASUREMENT V ERSUS S IMULATION E RROR

Fig. 17. Simulated RCS response of five ring versus ACiSRR chipless RFID
sensor tag (Taconic-TLX8). toward higher frequencies, the difference of Q factor has been
reduced significantly. Moreover, the frequencies also shifted
TABLE V
C ALCULATED Q FACTOR VALUES by approximately 100–150 MHz which may happen due to
the fabrication errors as well as form the noise from the envi-
ronment during the backscattering process. The use of a single
patch antenna where the innermost ring provides most of the
reflection can be a reason for obtaining similar Q factor in
contrast to simulation Q factor values. This can be further
explained by considering the beam pattern of the single patch
antenna where directivity of the main lobe more aligns with
the innermost ring. Nevertheless, measured S11 provides the
evidence of higher Q factor of resonances. The overall error
As per Fig. 17, it is clearly visible that the amplitude has
between simulation and measured results can be summarized
reduced in the RCS response has reduced when more rings
in Table VI. This table further emphasizes that variation of
added and alternatively, four resonances (instead of three)
Q factor as explained above; however, in terms of amplitude
occurred in the same bandwidth. Having resonances closer to
variation resonance at 4.6 GHz is closer to simulation RCS
each other may develop some sensitivity concerns when this
value in comparison to other two resonances.
five-ring design is used as a sensor, as the resonances may
By all these analyses from the surface currents, RCS value, Q
move and overlap with the deployment of smart materials.
factor (resolution), simulation models as well as the fabricated
But with additional microstrip ring added, more capacitance
performances, it can be reasonably concluded as follows.
is added to the system as a parallel capacitor. This scenario
1) Circular SRRs are suitable to be used as an integrated
enables, the use five ring (or more) SRR as a potential chipless
sensing platform due to the following.
RFID sensor with more sensing elements. It is also recom-
a) Its higher RCS.
mended to investigate other low-cost substrates and fabrication
b) Higher Q factor.
techniques such as screen printing [40] in order to understand
2) The increase of a number of rings will add more capaci-
the commercial potential. Taconic-TLX8 is one of the most
tance to the sensor system. However, careful consideration
expensive printed circuit board (PCB) substrates in the mar-
is necessary to maintain a higher RCS and higher Q value
ket. Screen printing is one of the known techniques that are
to be used as a chipless RFID sensor platform.
currently in use in developing low-cost chipless RFID tags.
In printing, the performance of tag is heavily dependent on VI. C ONCLUSION
the conductivity of the printing substrate as well as the sub- In this paper, we have presented the design of chipless RFID
strate used as the carrier. The printed tag provides not only resonator tag design suitable for employing multiple smart
a cheaper alternative but also more flexibility which is helpful materials as sensing media. Particularly, we have detailed out
in the diversified application where the bendability/pliability the design rationale, a quasi-static model of the suggested
is demanded, which is important in food safety applications. design, capacitive properties, analysis of surface currents, tag
The performance of printed chipless RFID sensor can also resolution, theoretical modeling, and more importantly com-
be affected by the fabrication errors, impurities in ink mix- parison of fabricated performances using both PCB and screen
ture, and measurement errors (effect of the environment, e.g., printed techniques. Finally, the results of this analysis can be
higher humidity). used to design a tag with higher data bits, which is useful
Finally, the resolution of tag response has been taken as part in multiphysical parameter sensing or as a tag ID for product
of the overall performance analysis. Usually, the resolution of identification purposes.
a tag can be denoted by its Q factor. As per the simulation
of the RCS value (Fig. 3) and measured S11(Fig. 13), the R EFERENCES
Q factors [41] of three notches can be calculated as below in [1] D. Gunders. (Feb. 22, 2017). Wasted: How America Is Losing Up to 40
Table V Percent of Its Food From Farm to Fork to Landfill. [Online]. Available:
Notch Frequency https://www.nrdc.org/sites/default/files/wasted-food-IP.pdf
Q= . (16) [2] averydennison.com/RFID. RFID: Tracking a Tangible Approach to
f1 − f2 Gain Efficiencies in Food Retail. Accessed: Mar. 3, 2018. [Online].
Available: http://rfid.averydennison.com/content/dam/averydennison/
It can be clearly observed that the Q factor has changed sig- rbis/global/en/documents/White%20Pages/Avery-Dennison-RFID-Food-
nificantly from simulation RCS to the measured S11. However, White-Paper-03-16.pdf

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.
 5396 IEEE INTERNET OF THINGS JOURNAL, VOL. 6, NO. 3, JUNE 2019

[3] Y. Feng, L. Xie, Q. Chen, and L.-R. Zheng, “Low-cost printed chipless [28] J. D. Baena et al., “Equivalent-circuit models for split-ring resonators
RFID humidity sensor tag for intelligent packaging,” IEEE Sensors J., and complementary split-ring resonators coupled to planar transmis-
vol. 15, no. 6, pp. 3201–3208, Jun. 2015. sion lines,” IEEE Trans. Microw. Theory Techn., vol. 53, no. 4,
[4] K. N. Chandra, A. M. Emran, and S. J. Kumar, “Chipless RFID temper- pp. 1451–1461, Apr. 2005.
ature memory and multiparameter sensor,” in Chipless RFID Sensors. [29] F. Bilotti, A. Toscano, and L. Vegni, “Design of spiral and multiple
Hoboken, NJ, USA: Wiley, 2016, p. 272. split-ring resonators for the realization of miniaturized metamaterial
[5] A. Faul and J. Naber, “Design and test of a 915MHz, RFID-based pres- samples,” IEEE Trans. Antennas Propag., vol. 55, no. 8, pp. 2258–2267,
sure sensor for Glaucoma,” in Proc. IEEE 5th Latin Amer. Symp. Circuits Aug. 2007.
Syst., 2014, pp. 1–4. [30] F. Bilotti et al., “Equivalent-circuit models for the design of metamateri-
[6] R. A. Potyrailo, C. Surman, T. Sivavec, and T. Wortley, “Passive als based on artificial magnetic inclusions,” IEEE Trans. Microw. Theory
multivariable RFID pH sensors,” in Proc. IEEE Int. Conf. RFID Technol. Techn., vol. 55, no. 12, pp. 2865–2873, Dec. 2007.
Appl., 2011, pp. 533–536. [31] (Jun. 4, 2017). Microwave & RF Laminates Products. [Online].
[7] M. R. Das. (Sep. 2006). Chipless RFID—The End Game. Available: Available: http://www.4taconic.com/page/microwave-rf-laminates-
https://www.idtechex.com/research/articles/chipless_rfid_the_end_game 66.html
_00000435.asp [32] P. Kalansuriya, N. C. Karmakar, and E. Viterbo, “On the detec-
[8] L. Wang, T. Liu, J. Sidén, and G. Wang, “Design of chipless RFID tag tion of frequency-spectra-based chipless RFID using UWB impulsed
by using miniaturized open-loop resonators,” IEEE Antennas Propag., interrogation,” IEEE Trans. Microw. Theory Techn., vol. 60, no. 12,
vol. 66, no. 2, pp. 618–626, Feb. 2018. pp. 4187–4197, Dec. 2012.
[33] M. I. Skolnik, Radar Handbook. New York, NY, USA: McGraw-Hill,
[9] N. G. R. Fletcher, “Wireless monitoring of temperature,” U.S. Patent
1970.
6 208 253 B1, 2000.
[34] Y. Liu, K. Zhao, M. G. B. Drew, and Y. Liu, “A theoretical and practical
[10] W. Buff, S. Klett, M. Rusko, J. Ehrenpfordt, and M. Goroli, “Passive clarification on the calculation of reflection loss for microwave absorbing
remote sensing for temperature and pressure using SAW resonator materials,” AIP Adv., vol. 8, no. 1, 2018, Art. no. 015223.
devices,” IEEE Trans. Ultrason., Ferroelect., Freq. Control, vol. 45, [35] M. Khaliel, A. El-Awamry, A. Fawky, and T. Kaiser, “Long reading
no. 5, pp. 1388–1392, Sep. 1998. range for the frequency coded chipless RFID system based on reflec-
[11] N. C. Karmakar, E. M. Amin, and J. K. Saha, “Smart materials for tarray antennas,” Int. J. Microw. Wireless Technol., vol. 10, no. 2,
chipless RFID sensors,” in Chipless RFID Sensors. Hoboken, NJ, USA: pp. 187–195, 2018.
Wiley, 2016, p. 272. [36] D. Dardari, R. D. Errico, C. Roblin, A. Sibille, and M. Z. Win,
[12] N. C. Karmakar, E. M. Amin, and J. K. Saha, “Chipless RFID sensor for “Ultrawide bandwidth RFID: The next generation?” Proc. IEEE, vol. 98,
real‐time environment monitoring,” in Chipless RFID Sensors. no. 9, pp. 1570–1582, 2010.
Hoboken, NJ, USA: Wiley, 2016, p. 272. [37] G. Khadka and S.-S. Hwang, “Tag-to-tag interference suppression tech-
[13] R. Marqués, J. Martel, F. Mesa, and F. Medina, “Left-handed-media nique based on time division for RFID,” Sensors, vol. 17, no. 1, p. 78,
simulation and transmission of EM waves in subwavelength split-ring- 2017.
resonator-loaded metallic waveguides,” Phys. Rev. Lett., vol. 89, no. 18, [38] J. D. Griffin and G. D. Durgin, “Gains for RF tags using multiple
2002, Art. no. 183901. antennas,” IEEE Trans. Antennas Propag., vol. 56, no. 2, pp. 563–570,
[14] R. Marques, F. Mesa, J. Martel, and F. Medina, “Comparative analysis 2008.
of edge- and broadside- coupled split ring resonators for metamate- [39] S. Huang, O. P. Gan, Z. Q. Hwang, H. Song, and L. Xie, “Passive UHF
rial design—Theory and experiments,” IEEE Trans. Antennas Propag., far-field RFID based localization in smart rack,” in Proc. IECON 41st
vol. 51, no. 10, pp. 2572–2581, Oct. 2003. Annu. Conf. IEEE Ind. Electron. Soc., 2015, pp. 26–31.
[15] G. Galindo-Romera, F. J. Herraiz-Martínez, M. Gil, [40] T. Blecha, “Screen printed chipless RFID tags,” in Proc. 37th Int. Spring
J. J. Martínez-Martínez, and D. Segovia-Vargas, “Split-ring resonator- Seminar Electron. Technol., 2014, pp. 143–146.
based sensor for thin-film sensing applications,” in Proc. 10th Eur. [41] R. Want, “An introduction to RFID technology,” IEEE Pervasive
Conf. Antennas Propag. (EuCAP), 2016, pp. 1–5. Comput., vol. 5, no. 1, pp. 25–33, Jan./Mar. 2006.
[16] (Oct. 2016). Dielectric Constant—Permittivity. [Online]. Available:
http://maxwells-equations.com/materials/permittivity.php
Tharindu Athauda received the bachelor of
[17] K. Chang and L. H. Hsieh, Microwave Ring Circuits and Related information technology degree from the School
Structures. New York, NY, USA: Wiley, 2004.
of Computing, University of Colombo, Colombo,
[18] A. Vitas, V. V. George, E. Chatzarakis, and L. Ekonomou, “Review of Sri Lanka, in 2009, and the B.Sc. engineering
different ring resonator coupling methods,” in Proc. 9th WSEAS Int. degree (Hons.) in textile process engineering from
Conf. Telecommun. Informat., 2010, pp. 227–231. the University of Moratuwa, Moratuwa, Sri Lanka,
[19] CS Technology. (Apr. 5, 2018). CST Field Monitors. [Online]. Available: in 2010. He is currently pursuing the Ph.D. degree
https://www.cst.com/products/cstps/solvers/wakefieldsolver in electrical engineering at Monash University,
[20] F. Toldra, Lawrie’s Meat Science, 8th ed. Duxford, U.K.: Woodhead, Clayton, VIC, Australia.
2017. He has comprehensive research and innovation
[21] Safe Food Australia—A Guide to the Food Safety Standards, Food Stand. experience in diversified interest areas in wearable
Australia New Zealand, 2016. Accessed: Feb. 21, 2018. technology and smart materials platforms over seven years in Sri Lanka
[22] T. Daio et al., “In-Situ ESEM and EELS observation of water uptake and and Europe. His current research interests include wearable antennas for
ice formation in multilayer graphene oxide,” Sci. Rep., vol. 5, Jul. 2015, portable devices, super wideband antennas, low-cost chipped and chipless
Art. no. 11807. radio frequency identification (RFID) sensors, and chipless RFID printed
[23] F. C. Salomão, E. M. Lanzoni, C. A. Costa, C. Deneke, and E. B. Barros, electronics.
“Determination of high-frequency dielectric constant and surface poten-
tial of graphene oxide and influence of humidity by kelvin probe force Nemai Chandra Karmakar (S’91–M’91–SM’99)
microscopy,” Langmuir, vol. 31, no. 41, pp. 11339–11343, Oct. 2015. received the Ph.D. degree in information technol-
[24] T. L. X. Huang. Graphene Oxide Dielectric Permittivity at GHz and Its ogy and electrical engineering from the University
Applications for Wireless Humidity Sensing. Accessed: Mar. 30, 2018. of Queensland, St. Lucia, QLD, Australia, in 1999.
[Online]. Available: https://arxiv.org/ftp/arxiv/papers/1711/1711.03435. He is currently an Associate Professor with the
pdf Department of Electrical and Computer Systems
[25] V. K. Singh et al., “Microwave absorbing properties of a thermally Engineering, Monash University, Clayton, VIC,
reduced graphene oxide/nitrile butadiene rubber composite,” Carbon, Australia. He has 20 years of teaching, design, and
vol. 50, no. 6, pp. 2202–2208, May 2012. research experience in smart antennas, microwave
[26] S. Xu and S. Matysiak, “Effect of pH on chitosan hydrogel polymer active and passive circuits, and chipless radio
network structure,” Chemcomm, vol. 53, pp. 7373–7376, Jun. 2017. frequency identifications (RFIDs) in both industry
[27] J. Voldman, Course Materials for 6.777J/2.372J Design and and academia in Australia, Canada, Singapore, and Bangladesh. He has
Fabrication of Microelectromechanical Devices, Spring 2007. MIT authored or co-authored over 230 journal and conference papers, 24 book
OpenCourseWare, Massachusetts Inst. Technol., Cambridge, MA, USA. chapters, and 3 edited and 1 co-authored books in the field of RFID. He
Accessed: Apr. 30, 2018. [Online]. Available: http://ocw.mit.edu/ holds two patent applications for chipless RFIDs.

Authorized licensed use limited to: UNIVERSITAT POLITECNICA DE CATALUNYA. Downloaded on February 16,2024 at 10:24:13 UTC from IEEE Xplore. Restrictions apply.