4316 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO.

10, OCTOBER 2016

A Frequency Agile Microstrip Patch Phased Array

Antenna With Polarization Reconfiguration
Behrouz Babakhani, Student Member, IEEE, Satish Kumar Sharma, Senior Member, IEEE,
and Nathan R. Labadie, Member, IEEE

Abstract— A novel multifunctional 1 × 4 phased array antenna optimization compared to a wideband antenna system. There
employing wideband frequency agile microstrip patches with is a wide range of published research on FAAs [2]. Com-
simultaneous polarization reconfiguration has been designed and mon tunable components used in these designs are varactor
experimentally verified. Each radiating element consists of a
circular microstrip patch connected to an annular microstrip diodes as variable load or liquid crystal materials, which have
ring via four varactor diodes for achieving frequency agility been reported in [3]–[6]. Polarization reconfiguration is also
between 1.5 and 2.4 GHz (frequency agility bandwidth ≈46%, desirable in communication systems as it provides filtering
S11 ≤ −10 dB). Employing two feed points per radiator enables and duplexing in addition to enhanced detection by avoiding
switching among four polarization senses (two linear and two fading loss and polarization purity impairments. Various polar-
circular polarizations) using a polarization feed network (PFN).
For realizing beam steering, the optimum amplitude and phase ization reconfigurable antennas have been reported in [7]–[12].
excitation coefficients for each radiating element were calculated Radiation pattern reconfigurable antennas provide communi-
using projection matrix method based on active element pattern, cation system with benefits such as the longer communica-
which is then applied to each corresponding radiating element tion range, and better signal to noise ratio (SNR). Single
using programmable digital phase shifters, low noise amplifiers, element antennas or antenna arrays with either frequency
and attenuators arranged in a beam forming network. Measure-
ment shows ±52° beam peak steering at 1.5 GHz and ±28° beam agility or polarization reconfiguration in combination to pat-
peak steering at 2.4 GHz based on 3 dB gain variation criteria tern reconfiguration have been reported in [13]–[24]. Two
for both the linear and circular polarizations. The simulated and PFN designs that can generate four polarizations are reported
measured results agree reasonably well. in [22] and [23]. These two circuitries are narrow band and
Index Terms— Antenna array, beam steering, beam synthesis, are not suitable for our design.
frequency agile, microstrip patch, multifunctional antenna, The authors of this paper have previously reported a
RF switches. wideband frequency tunable and polarization reconfigurable
microstrip antenna solution over 30% frequency agility range
I. I NTRODUCTION along with the polarization reconfigurable feed network [24].
In this paper, a multifunctional 1 × 4 linear phased array
M ULTIFUNCTIONAL antennas are a point of interest
in recent system design and research topics for their
ability to improve communication system performance. These
of frequency agile circular microstrip patches with polariza-
tion reconfiguration has been considered for beam steering
antennas typically incorporate dynamic design features such applications. To the best knowledge of the authors, such a
as frequency agility, polarization reconfiguration, and radiation beam steering antenna which provides wideband frequency
pattern reconfiguration. Many wireless communication devices agility with simultaneous polarization reconfiguration is not
are required to operate in two or more frequencies as there are presented earlier in the literature except in [25]–[27]. The
often different frequency allocations for different services [1]. radiating element is chosen based on [24] which offered only
Frequency agile antennas (FAAs) offer narrow instantaneous 30% frequency agility with simultaneous polarization recon-
bandwidths that can be tuned to operate over a large range of figuration. Brief detail of the antenna geometry and agility
frequencies. These antennas improve communication system range has been provided in Section II. Each radiating element
capability by reducing the need of RF filtering and dynamic has two orthogonal feed points. Therefore, by controlling the
port excitations, four different radiation polarizations can be
Manuscript received October 18, 2015; revised July 17, 2016; accepted
July 25, 2016. Date of publication August 4, 2016; date of current version generated. A novel polarization control feed network (PFN)
October 4, 2016. This work was supported by the National Science Foundation consisting of RF switches, power divider, and coupled line
through the CAREER Program under Grant ECCS-0845822. phase shifter has been proposed, which is substantially differ-
B. Babakhani and S. K. Sharma are with the Antenna and Microwave
Laboratory, Department of Electrical and Computer Engineering, San ent than our prior work presented in [24]. The PFN geometry
Diego State University, San Diego, CA 92182-1309 USA (e-mail: and its performance result is presented in Section III. The
[Link]@[Link]; ssharma@[Link]). radiating elements along with its dedicated PFNs are arranged
N. R. Labadie was with the Antenna and Microwave Laboratory, Department
of Electrical and Computer Engineering, San Diego State University, San in a linear 1 × 4 array. Beam steering is accomplished by
Diego, CA 92182-1309 USA. He is now with Cubic Defense Systems, San the beam forming network (BFN) connected to the four PFN
Diego, CA 92123 USA (e-mail: nrlabadie@[Link]). output ports. A MATLAB code based on the projection matrix
Color versions of one or more of the figures in this paper are available
online at [Link] method using the active element patterns (AEPs) [28] was
Digital Object Identifier 10.1109/TAP.2016.2598156 developed to calculate the optimum amplitude and phase
0018-926X © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See [Link] for more information.

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4317

Fig. 2. Measured and simulated reflection coefficient magnitude (S11 , dB)

of the antenna for different applied bias voltages. Solid lines are measured
and dashed lines are simulation results.

of the antenna geometry is given in [24]. Fig. 2 shows the

simulated and measured reflection coefficient magnitudes of
the antenna for different bias voltages applied to the varactor
Fig. 1. Geometry of the frequency agile patch antenna with simultaneous
polarization reconfiguration. (a) Top view. (b) Side view. diodes. Fig. 2 shows that the operating frequency varies from
1.5 to 2.4 GHz as the capacitance value is varied from
weights. Details about the array analysis and beam synthesis 2.67 to 0.63 pF. This equates to a frequency agility range
methods are provided in Section IV. Simulated beam steering of approximately 46%, which is wider than the 30% reported
performance of the array was experimentally verified and the in [24]. The absolute bandwidth at each of the tuned bands
results are given in Section V. Ansys High Frequency Structure is between 33 and 66 MHz although the fractional bandwidth
Simulator version 15 has been used to simulate and analyze the is nearly constant as expected for a resonant patch antenna.
proposed antenna in addition to using some custom developed Because of the rotational symmetry in the antenna structure,
MATLAB codes. capacitance variation has the same effect on the operational
frequency of the port 2 (S22 ). Although the simulated and
measured results are given only for some selected capacitance
II. S INGLE R ADIATING E LEMENT D EVELOPMENT
values, in reality, varactor diode capacitance variation is con-
A. Antenna Geometry tinuous and can offer any capacitance value between 0.63 and
Fig. 1 shows the geometry of the single radiating antenna 2.67 pF. The radiation performance of the antenna is similar
element. As mentioned earlier, this geometry is based on to [24] and is not included here.
the work reported in [24], which has been now scaled to
operate toward the higher frequency but with wider frequency III. P ROPOSED F EED N ETWORK FOR THE
agility bandwidth of 46% (1.5 to 2.4 GHz). For the sake AGILITY AND P OLARIZATION C ONTROL
of brevity, we are including only the most important details A. Novel Polarization Feed Network Geometry
here. The antenna consists of a microstrip circular patch
As shown in Fig. 1 the antenna has two feed points which
(R1 = 15.5 mm) and a concentric circular ring (R2 = 42 mm)
are 90° apart from each other. In the dominant TM11 mode
separated by a narrow gap (1.3 mm) to place the varactor
circular patch, each of these ports generates orthogonal surface
diodes. A 150 × 150 mm2 and 120 mil (3.05 mm) thick
currents and therefore orthogonal radiated far-fields. By appro-
Rogers 5880 RT/Duroid (εr = 2.2, tanδ = 0.0009) is
priately controlling the port excitations, one can generate
used as the substrate material. A Skyworks varactor diode
radiation patterns with different polarization properties. If only
(Model # SMV 1405-079LF) is chosen because of its low
one port is excited at a time, the antenna would generate linear
parasitic resistance (0.8 ), and large capacitance varia-
polarization, either polarized along the x-axis (LPX) or the
tion (0.63 to 2.67 pF) with the reverse bias voltage from
y-axis (LPY) based on which port is excited. If both ports
30 to 0 V. The varactor is modeled as a series RLC lumped
are excited with equal magnitude and ±90° phase difference,
component circuit. This model has been discussed in [24]. The
the antenna would generate right handed circular polariza-
parasitic values are extracted from the Skyworks data sheet.
tion (RHCP) or left handed circular polarization (LHCP).
This antenna was fabricated and experimentally verified.
Our PFN design consists of the following: three single pole
double through Skyworks AS186-302LF RF switches (less
B. Analysis and Measured Results than 1 dB insertion loss and isolation better than 45 dB), one
It was shown by analysis in [24] that, by varying the varactor diversity Skyworks 13355-374LF RF switch (less than 0.6 dB
capacitance, coupling between the annular ring and the circular insertion loss and isolation better than 22 dB), equal-split
patch would vary which consequently, changes the operational Wilkinson power divider (WPD) and a modified coupled-
frequency of the antenna. The analytical study for this type line Schiffman 90° phase shifter [29]. The proposed feed

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
4318 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 10, OCTOBER 2016

Fig. 5. Measured S-parameter response (dB) of the feed network for the
linear polarization.

would be two equally split signals which one has −90° phase
Fig. 3. Proposed polarization feed network (PFN) to control the ports difference comparing to the other one at the inputs of the
excitation of the antenna for achieving polarization reconfiguration.
diversity switch. If the diversity switch direct the 0° signal to
antenna port 1 and −90° signal to antenna port 2, we would
have RHCP. The reverse would generate LHCP. This feed
network is a general design and could be used with any two
port antenna configuration supporting two orthogonal linear
polarizations.

B. Simulated and Measured Results

Fig. 5 shows the simulated and measured results for lin-
ear polarization LPX (in Fig. 3 ANT port 1 receives the
power while ANT port 2 is matched terminated). In this
case, the signal travels through switches and CPW trans-
mission lines as shown in Fig. 4(b). The input port (S11 )
is matched for all the frequencies while one of the out-
put ports (S12 ) receives all the power and the other output
port (S13 ) receives no power. The difference between the
measured and simulated results comes from the different
sources of error. For example, the data used for model-
ing the switches and other components in simulation are
taken from datasheets where the measurement conditions
Fig. 4. Magnitude of the surface current distribution for the different
polarizations of (a) LPY, (b) LPX, (c) RHCP, and (d) LHCP at 2.0 GHz.
are typically not described in sufficient detail and possibly
The state of the switches has been defined with arrows. quite different from the end user application. It has been
mentioned in the data sheets that the values may have
5%–10% error. Another source of error is the soldering quality
network was laid out in coplanar waveguide (CPW) struc- and fabrication imperfections. In general, the simulated and
ture and simulated using the AWR’s Microwave Office on measured data are in good and acceptable agreement. The
a 62 mil (1.58 mm) thick FR4 substrate (150 × 150 mm2 , other linear polarization, LPY, is generated by switching the
εr = 4.4, tanδ = 0.02). Fig. 3 shows the geometry of the outputs using the diversity switch as shown in Fig. 4(a). Only
proposed PFN. This feed network was fabricated and measured results for the LPX case are plotted for brevity and due to the
for verification of the simulated results. fundamental similarity to the LPY case.
To illustrate how the PFN is working, the magnitude of sur- To generate circular polarization over the wide frequency
face current density for the feed network is plotted in Fig. 4 for range, a WPD combined with a coupled-line Schiffman
the different polarizations. The state of the switches has been 90° phase shifter is used. The WPD has a very wideband
defined with arrows for more clear understanding. Comparing matching bandwidth with a difference in output magnitude
surface current for the linear polarizations, LPX and LPY, of less than 0.5 dB in magnitude over our frequency range of
it is evident that the signal is traveling through a simple interest. Schiffman phase shifters can also generate a nearly
transmission line up to the diversity switch. It is the state constant phase shift over a wide range of frequencies [29].
of diversity switch that defines which port receives the power The measured and simulated results of the proposed circuit
and which port is matched terminated. Similarly, for the RHCP including the effects of both the WPD and Schiffman phase
and LHCP cases, the signal is going through the power divider shifter to generate CP are shown in Figs. 6 and 7. Fig. 6 plots
and then the 90° coupled line phase shifter. Therefore, there the S-parameter response for the CP case. It is evident that

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4319

TABLE I
M EASURED L OSS FOR THE PFN FOR THE L INEAR
AND C IRCULAR P OLARIZATION

Fig. 6. Measured and simulated S-parameter response (dB) of the feed

network for the CP case.

Fig. 7. Simulated and measured magnitude difference (dB) and phase

difference (°) between the output ports for the CP case.
Fig. 8. Geometry of the proposed frequency agile 1 × 4 linear array antenna
with its polarization feed network (PFN) of (a) antenna array and (b) PFNs
exciting the radiating elements.
the output magnitudes (|S21 | and |S13 |) are almost equal while
the input (|S11 |) is sufficiently matched considering −10 dB
matching criteria. Fig. 7 plots the simulated and measured To compensate this loss without degrading the SNR, low
difference between the magnitude and phase of the PFN noise amplifiers (LNAs) should be placed preferably right after
outputs. Simulation shows a maximum magnitude difference the antenna ports which would require eight LNAs. However,
of 0.8 dB, whereas the measured results show a maximum in this design to reduce number of LNAs, we decided to
difference less than 0.9 dB around 2 GHz. Ideally, a constant include it in the BFN board, which will be discussed in the
90° phase difference between the ports is desired to generate coming section.
CP with adequate axial ratio (AR). Fig. 7 shows a simulated
phase variation of 8° (from 88° to 96°) compared to a
IV. A RRAY A NALYSIS AND B EAM S YNTHESIS
measured phase difference variation of 10° (from 88° and 98°).
Overall, the magnitude and phase variation of the proposed A. Array Geometry
circuit are expected to have some impact on the quality A 1 × 4 linear array antenna implementation for beam
of the CP, but an AR of less than 3 dB is achievable. steering applications (Section II) is discussed in this section.
Again only one CP result was presented in the interest Fig. 8(a) shows the geometry of the proposed linear array
of brevity and because the two cases are fundamentally aperture. The spacing between the elements is 85 mm, which
similar. is λ/2 at midway between the edges of the frequency range
From Figs. 5 and 6 it is evident that the PFN loss is around of interest at 1.765 GHz. The array was fabricated using
3 dB range. This loss is due to the lossy FR-4 substrate Rogers 5880 RT/Duroid substrate of thickness 3.2 mm. Each
material (tanδ = 0.02), RF switches loss (0.7 dB which is of the array elements has its own dedicated polarization feed
nominal value for each switch), biasing components (dc block- network (PFN). These four PFN circuits are shown in Fig. 8(b)
ing capacitor and RF choke inductors) and quality of the with marked feed points (F1 to F8) where the output ports
soldering. Loss values have been summarized for selected of the radiating elements are connected. All the RF switches
frequencies in Table I. One should notice that for linear on the combined PFN board are controlled using a dual in-
polarization signal would travel though three RF switches line package (DIP) switch. Therefore, once the polarization is
while for CP the signal should go through four RF, which set by the DIP switch, all four PFNs would have the same
generates more losses. state ensuring that all four radiating elements would produce

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
4320 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 10, OCTOBER 2016

Fig. 9. Simulated reflection coefficient magnitude of the array for

three different capacitance values.

Fig. 11. Simulated active S-parameters at 1.5 GHz for the LHCP case where
all the eight antenna array ports excited simultaneously.

not be neglected as it reflects the coupling variation as the

beam scans [30]. Fig. 11 shows the array’s simulated ASP
at 1.5 GHz for the LHCP case where we have all the eight
antenna ports excited simultaneously. To test the LHCP mode,
a 90° phase difference was applied between even and odd
mode ports. At this frequency, we have the worst coupling
between the elements and the largest scan range. This would
create the worst case ASP variation. This figure shows that
Fig. 10. Simulated mutual coupling between the array ports for three different
capacitance values.
the active S-parameter is varying as we vary the progressive
phase shift between the elements, which in fact, steers the
beam peak. As we steer the beam, the induced field from the
the same polarization. The radiating element output ports and
adjacent elements causes the input impedance of one element
PFN input ports have Subminiature Adapter (SMA) female
to vary as captured in this figure.
connectors. Thus, male to male SMA adaptors (barrels) are
used to connect the PFN board to the array aperture board
in a back-to-back configuration. In principle, both the array C. Beam Synthesis Method
aperture and PFN circuits could be fabricated in one multilayer To maximize the performance of an antenna array, many
printed circuit board stackup. However, this is a more costly synthesis methods based on the array element excitation con-
approach and not necessary to demonstrate the multifunctional trol have been proposed. In this paper, the beam synthesis has
antenna concept presented in this paper. been done based on the projection matrix method described
in [28] but the AEP is used as the input data, so the mutual
B. Array Analysis coupling and replacement from the phase center for each
element has been taken into account. The AEP is defined by
Fig. 9 shows the array’s simulated reflection coefficient exciting single element in the array when all other elements are
magnitude for the three different capacitance values of the terminated in matched loads. This method is briefly described
varactor diodes. Fig. 9 implies that the operational frequency in the Appendix.
of all four radiating elements in the array is varying together
as the capacitance value of the varactor is changing while
V. E XPERIMENTAL V ERIFICATION
maintaining a good impedance matching level.
Mutual coupling between the elements in the array has A. Fabricated Array and Feed Networks
a significant impact on the radiating element’s impedance A functional block diagram of the receiver only array and
performance, which is shown in Fig. 10. Only the simulated supporting feed network blocks are shown in Fig. 12. The
coupling between the adjacent ports are plotted as these are output ports of each of the radiating elements are connected
the worst values. As expected, the lowest frequency (1.5 GHz) to the input ports of an active PFN using male SMA to male
has the highest coupling level due to the minimum element SMA barrels. Flexible coaxial cables are used to connect the
spacing in terms of wavelength. Even in this case, the coupling output ports of the PFN circuits to the BFN input ports.
between the elements at this frequency is better than −12 dB. The BFN consists of four LNAs (RAMP-33LN LNA from
The minimum coupling of better than −27 dB happens at the Mini-Circuits with an average gain of 15 dB), four 6 bit digital
highest frequency (2.4 GHz) where we have maximum element phase shifters (DPSs with 360° variation in 5.6° steps), and
spacing in terms of wavelength. four 6 bit digital step attenuators (DATs with full attenuation
Another important factor in the array analysis is the active of 31.5 dB in 0.5 dB steps) each having a unique hardware
scattering parameter (ASP). The importance of ASP should address. Due to the LNA positive gain, this board would

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4321

Fig. 12. Functional block diagram of the fully combined array with
BFN + PFN boards.

Fig. 14. Fully assembled antenna array with the polarization feed net-
works (PFNs), and BFN boards as mounted in the anechoic chamber for
the radiation pattern measurement.

Fig. 13. Top and bottom views of the active BFN board used to combine Fig. 15. Normalized measured AEP for each of the radiating elements in
the outputs from each of the subarrays for the beam synthesis. the array at 2.0 GHz and LPY.

provide BFN gain around 8 and 4.5 dB at 1.5 and 2.4 GHz,
and 4) recover the array pattern gain by de-embedding
respectively. Programmable components on the BFN are con-
the active feed network gain and associated mismatch
trolled from a laptop hosting a custom Graphical user interface
through a microcontroller (MCU). Other supporting logic and losses.
To measure AEP for each of the array elements, the signal
power circuitry is present on the BFN board. The combined
RF output of each RF channel is combined in a final Wilkinson attenuation for the element under test was set to 0 dB while all
power combiner where it can be routed to the vector network the other elements receive maximum attenuation of 31.5 dB.
The phase shift was set to 0° for all the elements. In the
analyzer receiver for the measurement. Fig. 13 shows the two
sides of the fabricated BFN. described experimental setup, the complex radiated fields were
measured for a spherical sampling of azimuthal and elevation
angles and for the different polarizations and frequencies.
B. Array Calibration Fig. 15 shows the φ = 90° cut of the measured copolarization
The antenna array was calibrated and characterized in the AEP for each of the array elements at 2.0 GHz with LPY
anechoic chamber at San Diego State University’s Antenna and polarization. To steer the beam toward the desired angle,
Microwave Laboratory (AML) for the radiation pattern perfor- the measured radiation field data was processed based on the
mance. The Orbit/FR measurement facility has the capability described projection matrix method using a MATLAB code to
to perform both far field and spherical near field radiation find the optimum excitation (amplitude and phase) for each of
pattern measurements. The BFN, PFN board, array aperture, the array elements as described in Section III. Fig. 16 shows
and supporting circuitry were integrated into the test setup the measured normalized radiation pattern of the array after
mounted in the AML’s anechoic chamber as shown in Fig. 14. processing AEP in MATLAB to steer the synthesized beam
The calibration process consists of the following four steps: toward 20°. The calculated attenuation and phase shift for
1) measure each of the AEPs; 2) compute the required complex each of the elements is also shown in Fig. 16. The expected
weighting coefficients needed to synthesize a desired beam angular error is ±1° but because of the limited accuracy of
pattern using the MATLAB code; 3) apply the computed the DATs and DPSs. Measured results show an error of ±3°
complex weights and characterize the resulting array patterns; in the worst case. To remove the effects of the feed network

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
4322 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 10, OCTOBER 2016

Fig. 17. SFG model of a single receive channel.

Fig. 16. Measured normalized radiation pattern of the array after processing
AEP using a MATLAB code to steer the beam to 20°. The required attenuation
and phase shift is calculated and shown in the figure for each of the radiating
elements.

from the measurements and recover the array pattern gain,

G θ (φ = 90°), a de-embedding process is required. We quan-
tify the gain and efficiency of the array aperture sepa-
rately from the feed network. In our particular implemen-
tation, there are no LNAs at the input to the PFN, which
will result in a noise figure increase roughly equal to
the PFN losses. Had this been done, the receiver sensitiv-
ity would have been determined primarily by the aperture
inefficiencies alone. Furthermore, it is a relatively simple
design change to move the existing LNAs from the BFN
inputs to the PFN inputs rendering the impact of the feed Fig. 18. Boresight gain variation for the linear polarization over frequency.
Solid lines are measured and dashed lines are simulated.
network losses negligible with respect to maintaining an
adequate SNR at the receiver. The de-embedding proce-
dure used in this paper is the same as the method used Measured results show that the array has 13.29 dBi gain
in [21]. The array pattern gain can be computed by normal- at 2.4 GHz. This gain decreases gradually to 9.66 dBi at
izing the measured gain of the full array, G θM (φ = 90°), 1.5 GHz. This gain variation over frequency is to be expected
n
to the active feed network channel gains, S21 . The active feed from the corresponding variation of the array aperture elec-
network channel scattering parameters are measured in 50  trical size. For example, a quick analytical calculation of the
source system and load impedances. In the fully assembled directivity from the approximate aperture area (85×340 mm2 )
array, the inputs to the active feed network are loaded by the yields an expected 9.6 dBi at 1.5 GHz and 13.7 dBi at
patch antenna impedances. Therefore, the actual array pattern 2.4 GHz. This calculation does not account for fringing
gain will depend on the output reflection coefficient of the fields that may increase the expected directivity by some
patch antennas,  A , and the input reflection coefficient of the amount. However, the approximately 4 dB gain variation
n
active feed network, S11 . The antenna and feed network are from analysis over frequency is consistent with measurement.
also connected by a male to male SMA adapter. This is treated The difference between the measured (13.29 dBi at 2.4 GHz
as a perfectly matched 50  transmission line with a phase and 9.66 dBi at 1.5 GHz) and simulated gain (14.0 dBi at
shift of exp(β j ). Fig. 17 shows the equivalent signal flow 2.4 GHz and 10.32 dBi at 1.5 GHz) values can be compensated
graph (SFG) for a simplified model of a single receive channel by considering higher parasitic resistance for the varactor
on the array. Using this model, we derive the array pattern gain diodes. This suggests that, the varactor diode has higher value
as [21] of resistance than what was mentioned in the data sheet.
G θM Table II summarizes the simulated and measured results for
Gθ =      . the boresight gain of the linear polarization.
N  n 2 /|1 −  n S n exp (− j 2β)|2
n=1 |S21 | 1 − S11
n 2
A 11 Similarly, the LHCP case has been simulated and measured.
(1) Fig. 19 shows the simulated and measured boresight CP gain
for six different frequency agile bands. Measured data shows
gain variation from 12.34 dBic at 2.4 GHz to 8.45 dBic at
C. Measured Radiation Patterns 1.5 GHz, which is again consistent with change in aperture
1) Frequency Agility Performance: Fig. 18 shows the mea- electrical size over frequency.
sured and simulated boresight linear polarization gain pat- The gain variation in simulation is from 13.8 dBic at
terns of the array for six operational frequency agile bands. 2.4 GHz to 9.81 dBic at 1.5 GHz. Compared to the linear

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4323

TABLE II TABLE III

S IMULATED AND M EASURED G AIN D ATA FOR L INEAR P OLARIZATION S IMULATED AND M EASURED G AIN D ATA FOR LHCP

TABLE IV
M EASURED AND S IMULATED R ADIATION P ERFORMANCE PARAMETERS
FOR D IFFERENT P OLARIZATIONS AT 2.0 GHz

Fig. 19. Boresight gain variation for the circular polarization over frequency.
Solid lines are measured and dashed lines are simulated data.
Table III summarizes the simulated and measured results for
the boresight gain of the LHCP case.
2) Polarization Reconfiguration: Fig. 21 shows the mea-
sured and simulated radiation patterns for four polariza-
tions (LPX, LPY, RHCP, and LHCP) at 2.0 GHz. Again,
the simulated results show higher gain compared to measure-
ment. As was mentioned before, this difference comes from
the higher varactor parasitic resistance compared to the one
considered in the simulations. For linear polarization, the sim-
ulated cross polarizations are lower compared to the measured
ones because the feed network has better port isolation in the
simulation than the measurement. Additional degradation to
polarization purity may be attributed to scattering from cables
and the large wooden sheet used for mounting the antenna.
Fig. 20. Boresight AR (θ = 0° and ϕ = 90°) variation over frequency. Table IV summarizes the measured and simulated radiation
Solid lines are measured and dashed lines are simulated. performance for the four different polarizations.
3) Beam Steering Performance: Fig. 22 shows beam steer-
ing performance of the array at 2.0 GHz for one linear and
polarization, the CP case shows lower gain values which is one circular polarization. At 2.0 GHz interelement spacing
attributed to the resistance in the varactors. For CP case, all (d = 85 mm) is more than half wavelength, which introduces
four varactor diodes are seen by the surface currents whereas in a grating lobe beyond a maximum scan angle. Theoretically,
the linear polarization, the surface currents pass through only the maximum scan angle before grating lobe occurs can be
two of the varactor diodes. Fig. 20 shows the measured and calculated using the equation
simulated boresight AR of the array for six tuned frequency
bands. Simulation and measured AR values are in agreement λ
−1.
cos(θgr ) = (2)
except at 2.2 and 2.4 GHz. The reason for this discrepancy d
is the difference in the measured and simulated performance At 2 GHz, with d = 85 mm interelement spacing, beam
of the PFN, as shown in Fig. 6 and discussed in Section III. scanning up to 40° can be achieved without grating lobe

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
4324 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 10, OCTOBER 2016

Fig. 22. Main lobe gain (dBi/dBic) and sidelobe gain (dBi/dBic) at 2.0 GHz
as the array scans from θ = +50° to θ = −50°.

Fig. 23. Simulation and measured LHCP copolarized gain patterns steered
to two different angles at 2.0 GHz. (a) θ = −40°. (b) θ = +40°.

the radiation patterns for the LHCP at 2.0 GHz as the beam
steers to −40° and +40°. Results show good copolarization
and cross-polarization discrimination. There is a slight differ-
ence between simulation and measurement but the results are
Fig. 21. Measured and simulated radiation patterns at 2.0 GHz for four overall in good agreement. Some of this discrepancy can be
different polarizations. (a) LPX. (b) LPY. (c) RHCP. (d) LHCP.
attributed to imperfections of the measurement setup such as
alignment issues and scattering that has not been accounted
for in simulation. Table V summarizes the beam steering
coming into picture. Fig. 22 shows the gain variation as we performance for the LHCP case and at different frequencies.
steer the beam along the array axes from −50° to +50° and for In this Table V, θ3 dB is the angle at which the array pattern
both the LPY and LHCP polarizations. It also shows that the gain has dropped by 3 dB from boresight while θmax represents
main lobe gain is almost constant from −40° to +40° for both the maximum scan angle before the main lobe and sidelobe
the linear and circular polarizations. Looking at the AEP as gain level become equal.
shown in Fig. 15, it is evident that the single element radiation The antenna array was also tested using the spherical near
patterns get flattened because of coupling to the neighborhood field system to capture 3-D radiation patterns. Fig. 24 shows
radiating elements. Therefore, the array is expected to have the normalized simulated and measured radiation patterns at
almost constant gain as beam steers in this limited range. The 2.0 GHz for the LHCP case when the beam peak is steered
gain drops sharply at scan angles beyond ±40° due to the to 0°, 20°, and 40°. Measured results show higher sidelobes
sharp roll off of the element pattern amplitudes. Fig. 23 shows but overall it is in good agreement with the simulation results.

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4325

TABLE V
LHCP B EAM S TEERING R ESULT AT D IFFERENT F REQUENCIES

Fig. 25. Simulated and measured AR in the form of intensity plot at 2.0 GHz
for the LHCP case at different beam peak scan angles. (a) Simulated at θ = 0°.
(b) Measured at θ = 0°. (c) Simulated at θ = 20°. (d) Measured at θ = 20°.
(e) Simulated at θ = 40°. (f) Measured at θ = 40°.

difference between measured and simulated PFN performance

discussed earlier.
VI. C ONCLUSIONS
A multifunctional antenna consisting of a linear array of
wideband frequency agile circular microstrip patches with
simultaneous polarization reconfiguration is simulated, fab-
ricated and experimentally verified for the beam steering
application. The operational frequency band is from 1.5 to
2.4 GHz (46% frequency agility) and can be controlled
using varactor diodes, which is significantly wider than our
previously reported result. Radiation pattern polarization is
Fig. 24. Normalized simulated and measured radiation patterns at 2.0 GHz
for the LHCP case at different beam peak scan angles. (a) Simulated at θ = 0°. reconfigured by controlling port excitations which can gen-
(b) Measured at θ = 0°. (c) Simulated at θ = 20°. (d) Measured at θ = 20°. erate LPX, LPY, LHCP, and RHCP. A novel polarization
(e) Simulated at θ = 40°. (f) Measured at θ = 40°. control feed network (PFN) consisting of RF switches, WPD
and coupled lined Schiffman phase shifter has been used for
realizing the polarization reconfiguration. To steer the beam
Fig. 25 shows the simulated and measured AR intensity plot toward a desired scan angle, a BFN consisting of program-
for LHCP case and beam peak at 0°, 20°, and 40° at 2 GHz. mable DPSs and variable gain attenuators fully control by a
Measured AR results shows higher values and narrower usable MCU has been used. A MATLAB code based on measured
AR beamwidth in both the azimuthal and elevation planes AEP and projection matrix method is used to find optimum
compared to the simulation results. This is attributed to the complex coefficients to steer the beam. The proposed array

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
4326 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 64, NO. 10, OCTOBER 2016

along with the PFNs and BFN is fabricated and measured for [4] A. Khidre, F. Yang, and A. Z. Elsherbeni, “A patch antenna
verification of the simulated results. Measurement shows ±52° with a varactor-loaded slot for reconfigurable dual-band opera-
tion,” IEEE Trans. Antennas Propag., vol. 63, no. 2, pp. 755–760,
beam peak steering toward the lower frequency end (1.5 GHz) Feb. 2015.
and ±28° beam peak steering toward the higher frequency [5] B. Avser and G. M. Rebeiz, “Tunable dual-band antennas for
end (2.4 GHz) based on 3 dB gain variation criteria. Simulated 0.7–1.1-GHz and 1.7–2.3-GHz carrier aggregation systems,” IEEE
Trans. Antennas Propag., vol. 63, no. 4, pp. 1498–1504, Apr. 2015.
and measured results are in agreement for most of the part. [6] A. Dey and G. Mumcu, “Microfluidically controlled frequency-tunable
Such an array antenna is an example of a multifunctional monopole antenna for high-power applications,” IEEE Antennas Wireless
antenna which can be use in several applications requiring fre- Propag. Lett., vol. 15, pp. 226–229, 2016.
quency agility with simultaneous polarization reconfiguration [7] J. M. Kovitz, H. Rajagopalan, and Y. Rahmat-Samii, “Design and imple-
mentation of broadband MEMS RHCP/LHCP reconfigurable arrays
and beam steering. using rotated E-shaped patch elements,” IEEE Trans. Antennas Propag.,
vol. 63, no. 6, pp. 2497–2507, Jun. 2015.
[8] J.-S. Row and J.-F. Wu, “Aperture-coupled microstrip antennas with
A PPENDIX switchable polarization,” IEEE Trans. Antennas Propag., vol. 54, no. 9,
To maximize the performance of an antenna array, many pp. 2686–2691, Sep. 2006.
synthesis methods based on the array element excitation con- [9] Z.-X. Yang, H.-C. Yang, J.-S. Hong, and Y. Li, “Bandwidth enhancement
of a polarization-reconfigurable patch antenna with stair-slots on the
trol have been proposed. In this paper, the beam synthesis has ground,” IEEE Antennas Wireless Propag. Lett., vol. 13, pp. 579–582,
been done based on the projection matrix method described 2014.
in [28] but the AEP is used as the input data, so the mutual [10] D. Seo and Y. Sung, “Reconfigurable square ring antenna for switchable
circular polarisation,” Electron. Lett., vol. 51, no. 6, pp. 438–440,
coupling and replacement from the phase center for each Mar. 2015.
element has been taken into account. This method can be [11] J.-B. Yan, R. D. Hale, A. Mahmood, F. Rodriguez-Morales,
described as follows. C. J. Leuschen, and S. Gogineni, “A polarization reconfigurable low-
profile ultrawideband VHF/UHF airborne array for fine-resolution
Consider a general linear array in φ = φ0 plane. The sounding of polar ice sheets,” IEEE Trans. Antennas Propag., vol. 63,
elements could have different patterns and/or may be nonuni- no. 10, pp. 4334–4341, Oct. 2015.
formly spaced. The vector [ A PD ]m×1 defines the desired array [12] K. M.-J. Ho and G. M. Rebeiz, “A 0.9–1.5 GHz microstrip antenna with
full polarization diversity and frequency agility,” IEEE Trans. Antennas
pattern with m samples in the θ direction. Therefore, we have Propag., vol. 62, no. 5, pp. 2398–2406, May 2014.
[13] A. Khidre, F. Yang, and A. Z. Elsherbeni, “Circularly polarized beam-
[ A PD ]m×1 = [P1 P2 ··· Pn ] ∗ [α1 α2 · · · αn ]T scanning microstrip antenna using a reconfigurable parasitic patch of
T
= [P]m×n ∗ [α]1×n (3) tunable electrical size,” IEEE Trans. Antennas Propag., vol. 63, no. 7,
pp. 2858–2866, Jul. 2015.
where [P]m×k ; 1 ≤ k ≤ n is the AEP of the kth radiating [14] C. H. Wu and T. G. Ma, “Pattern-reconfigurable self-oscillating active
integrated antenna with frequency agility,” IEEE Trans. Antennas
element in the array. [α]1×n
T defines the required excitation Propag., vol. 62, no. 12, pp. 5992–5999, Dec. 2014.
for each element in the array to form the desired pattern. [15] Y. Cai, Y. J. Guo, and P. Qin, “Frequency switchable quasi-Yagi dipole
Each term of the vector [α] is a complex number and can array for base station antennas,” in Proc. IEEE Int. Symp. Antennas
be calculated as Propag. (APSURSI), Jul. 2011, pp. 738–741.
[16] Y. Cai and Y. J. Guo, “A frequency-agile compact array with a recon-
[α] = pinv([P]) ∗ [ A PD ]. (4) figurable decoupling and matching network,” IEEE Antennas Wireless
Propag. Lett., vol. 10, pp. 1031–1034, 2011.
Multiplying pseudoinverse matrix of [P]m×n with [17] S.-C. Yen and T.-H. Chu, “A beam-scanning and polarization-agile
antenna array using mutually coupled oscillating doublers,” IEEE Trans.
[ A PD ]m×1 yields the optimal [α]1×n in the sense that the Antennas Propag., vol. 53, no. 12, pp. 4051–4057, Dec. 2005.
mean square error between the desired pattern ([ A PD ]) and [18] H. Liu, S. Gao, and T. H. Loh, “Compact MIMO antenna with fre-
the actual pattern is minimized [28]. A pulse function has quency reconfigurability and adaptive radiation patterns,” IEEE Antennas
Wireless Propag. Lett., vol. 12, pp. 269–272, 2013.
been used as a desired pattern with the peak at θ0 (desired [19] R. J. Weber and Y. Huang, “A frequency agile beamforming array,”
angle) in order to maximize the gain of the array. Other in Proc. IEEE Int. Symp. Antennas Propag. (APSURSI), Jul. 2011,
desired array patterns can be substituted in order to trade off pp. 2035–2038.
gain for other radiation pattern parameters such as sidelobe [20] J. R. De Luis and F. De Flaviis, “Frequency agile switched beam
antenna array system,” IEEE Trans. Antennas Propag., vol. 58, no. 10,
level or cross-polarization level. pp. 3196–3204, Oct. 2010.
[21] N. R. Labadie, S. K. Sharma, and G. M. Rebeiz, “A novel approach
to beam steering using arrays composed of multiple unique radiating
ACKNOWLEDGMENT modes,” IEEE Trans. Antennas Propag., vol. 63, no. 7, pp. 2932–2945,
The authors would like to thank M. Bruno, laboratory tech- Jul. 2015.
nician, for his help with the some of the circuit components [22] Y.-F. Wu, C.-H. Wu, D.-Y. Lai, and F.-C. Chen, “A reconfig-
urable quadri-polarization diversity aperture-coupled patch antenna,”
population on the boards. IEEE Trans. Antennas Propag., vol. 55, no. 3, pp. 1009–1012,
Mar. 2007.
[23] Y. Wu and Z. Nie, “A novel reconfigurable microstrip antenna and
R EFERENCES its performance analysis in diversity reception system,” Microw. Opt.
[1] G. H. Huff and J. T. Bernhard, “Reconfigurable antennas,” in Modern Technol. Lett., vol. 49, no. 10, pp. 2516–2519, 2007.
Antenna Handbook, C. A. Balanis, Ed. Hoboken, NJ, USA: Wiley, 2014. [24] B. Babakhani and S. Sharma, “Wideband frequency tunable concentric
[2] A. Petosa, Frequency-Agile Antennas for Wireless Communications. circular microstrip patch antenna with simultaneous polarization recon-
Boston, MA, USA: Artech House, 2014. figuration,” IEEE Antennas Propag. Mag., vol. 57, no. 2, pp. 203–216,
[3] M.-C. Tang, R. W. Ziolkowski, S. Xiao, M. Li, and J. Zhang, Apr. 2015.
“Frequency-agile, efficient, near-field resonant parasitic monopole [25] B. Babakhani and S. K. Sharma, “Frequency tunable microstrip array
antenna,” IEEE Trans. Antennas Propag., vol. 62, no. 3, pp. 1479–1483, antenna with polarization reconfiguration,” in Proc. IEEE Antennas
Mar. 2014. Propag. Soc. Int. Symp. (APSURSI), Jul. 2014, pp. 997–998.

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.
BABAKHANI et al.: FREQUENCY AGILE MICROSTRIP PATCH PHASED ARRAY ANTENNA 4327

[26] B. Babakhani and S. K. Sharma, “A beam steering linear antenna array Satish Kumar Sharma (SM’04) received the
with novel simultaneous frequency agility and polarization reconfig- [Link]. degree in electronics engineering from the
urablity,” in Proc. USNC-URSI Radio Sci. Meeting (Joint AP-S Symp.), Kamla Nehru Institute of Technology, Sultanpur,
Jul. 2015, p. 185. India, and the Ph.D. degree in electronics engineer-
[27] B. Babakhani and S. K. Sharma, “A multifuctional phased array antenna ing from IIT Varanasi, Varanasi, in 1991 and 1997,
with frequency agility and polarization reconfigurablity,” in Proc. IEEE respectively.
Antennas Propag. Soc. Int. Symp. (APSURSI), to be published. He was a Post-Doctoral Fellow in the Department
[28] A. K. Bhattacharyya, “Projection matrix method for shaped beam syn- of Electrical and Computer Engineering, Univer-
thesis in phased arrays and reflectors,” IEEE Trans. Antennas Propag., sity of Manitoba, Winnipeg, MB, Canada, from
vol. 55, no. 3, pp. 675–683, Mar. 2007. 1999 to 2001. He was a Senior Antenna Engi-
[29] J. L. R. Quirarte and J. P. Starski, “Novel Schiffman phase shifters,” neer with InfoMagnetics Technologies Corporation,
IEEE Trans. Microw. Theory Techn., vol. 41, no. 1, pp. 9–14, Jan. 1993. Winnipeg, from 2001 to 2006. He was a Research Associate with the
[30] W. Kahn, “Active reflection coefficient and element efficiency in arbi- University of Manitoba from 2001 to 2006. In 2006, he joined San Diego State
trary antenna arrays,” IEEE Trans. Antennas Propag., vol. 17, no. 5, University, San Diego, CA, USA, as an Assistant Professor in the Department
pp. 653–654, Sep. 1969. of Electrical and Computer Engineering, where he has developed an Antenna
Laboratory, teaches courses in Applied Electromagnetics, and advises several
M.S. and Ph.D. graduate students. He is currently a Professor and Director of
the Antenna and Microwave Laboratory, San Diego State University. He has
authored or co-authored over 150 research papers in referenced international
journals and conferences. He holds one U.S. and one Canadian patent.
Recently, he co-edited three volumes of “Handbook of Reflector Antennas
and Feed Systems, Volume 1: Theory and Design of Reflectors, Volume II:
Feed Systems, and Volume III: Applications of Reflectors,” (Artech House,
USA), which also has several co-authored chapter contributions by him. His
current research interests include microstrip antennas, ultrawide, wideband,
multiband, and broadband antennas, reconfigurable, tunable, and frequency
agile antennas, feeds for reflector antennas, waveguide horns and polarizers,
electrically small antennas, MIMO antennas, phased array antennas, wire
antennas, active antennas, and microwave passive components.
Dr. Sharma is a Full Member of the USNC/URSI, Commission B,
Fields and Waves. He received the IEEE AP-S Harold A. Wheeler Prize
Paper Award in 2015, the National Science Foundation’s prestigious faculty
early development CAREER Award in 2009, the Young Scientist Award
of URSI Commission B, Field and Waves, during the URSI Triennial
International Symposium on Electromagnetic Theory, Pisa, Italy, in 2004.
He was recognized as an Outstanding Associate Editor (AE) of the IEEE
T RANSACTIONS ON A NTENNAS AND P ROPAGATION (TAP) journal in 2014.
He has been serving as an AE of TAP since 2010. He was the Chair/
Co-Chair of several Student Paper Contests in different conferences and
symposia and served on the sub-committee of the Education Committee for
the IEEE Antennas and Propagation Society for the organization of the Student
Paper Contests.

Nathan R. Labadie (M’16) received the B.S.

degree in physics from the University of Cali-
fornia at Davis, Davis, CA, USA, in 2005, the
M.S.E.E. degree in antenna and electromagnetics
Behrouz Babakhani (S’13) was born in Iran in research from San Diego State University (SDSU),
1985. He received the [Link]. degree in electrical engi- San Diego, CA, USA, in 2009, and the Ph.D. degree
neering from the Ferdowsi University of Mashhad, in electrical and computer engineering through a
Mashhad, Iran, in 2009, and the [Link]. degree from joint program between the University of California
the Sharif University of Technology, Tehran, Iran, at San Diego, San Diego, CA, USA, and SDSU, in
in 2011. He is currently pursuing the Ph.D. degree 2015.
with San Diego State University, San Diego, CA, He is currently a Principal Antenna Engineer at
USA. Cubic Defense Systems, San Diego, CA, USA, where he develops antenna
His current research interests include tunable and systems for military data link applications. His current research interests
reconfigurable antennas and arrays. include wideband and wide scanning phased arrays, reconfigurable antenna
elements, and multibeam digital beamforming communications systems.

Authorized licensed use limited to: National University of Singapore. Downloaded on January 19,2026 at 10:58:48 UTC from IEEE Xplore. Restrictions apply.