This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI

Impact Analysis and Calibration

Methods of Excitation Errors for Phased
Array Antennas
GUOLONG HE1 , XIN GAO1 , AND RENTIAN ZHANG1
1
Beijing Institute of Tracking and Telecommunication Technology, Beijing 100094, China
Corresponding author: Guolong He (e-mail: [email protected]).

ABSTRACT Phased array antennas have played a very important role in many different areas and
applications. It requires precise excitation of each antenna element by various synthesis techniques to obtain
desired array pattern features. However, due to reasons such as manufacturing imperfections, component
aging, and temperature variation, the realistic antenna element excitation inevitably differs from their
expected values in practice. This paper presents a tutorial-like review to deal with excitation errors for
phased array antennas. Two kinds of analysis methods, probabilistic methods and interval arithmetic (IA)
based methods, are presented to evaluate the effects of excitation errors for phased array antennas. State-of-
the-art calibration methods along with various signal processing techniques are reviewed, their advantages
and challenges are discussed in a comparative manner. Some other common errors and open research
directions are also presented.

INDEX TERMS Phased array antenna, excitation error, interval arithmetic, array calibration.

I. INTRODUCTION phased array antennas have gained a much wider range of

The first phased array antenna was invented around the turn improvements in both capability and performance. With the
of the 20th century with mechanical rotation to steer the important breakthrough with silicon(Si)-based phased array
beam [1]. It was not until World War II that remarkable technology, it has been driven to a point affordable for a huge
progress such as controlled phase shifters were made, which number of civil and commercial applications, including not
began to replace the original mechanical antenna steering. only 5G communication and wireless localization, but also
With the military need for large missile warning and space medical imaging, autonomous driving, and Internet of Things
surveillance radars during the cold war, phased array antenna (IOT) that enrich our daily lives [7-10]. Nowadays, Si-based
technologies were fast developed and quickly played an phased array technologies have been reported covering in
important role in defense areas like radar, sonar, navigation, a very wide frequency range from millimeter-wave (mm-
and wireless communications [2-3]. At that time, the phased wave) to sub-millimeter-wave band, and even in terahertz
array antennas utilized passive architecture with a centralized (THz) frequency band [11]. These modern applications with
high-power transmit amplifier and receiver. However, the increasing working frequency impose much more stringent
phased array antennas have undergone a steady evolution requirements on phased array antennas, adding significant
over the years. Thanks to advancements in GaAs microwave constraints to the size, cost, complexity, and weight, and most
circuits during the 1980s, active electronically scanned ar- importantly, performance of the array.
rays with analog beamforming architecture became very By amplitude tapering and phase control of antenna ele-
popular, where each element has its own full independent ments, phased array antennas can agilely steer the beam, con-
transmit/receive (T/R) module. By the end of the 1990s, ex- trol the sidelobe levels (SLLs), and present the nulls on the
periments and implements with hybrid analog-digital beam- interfere directions conveniently [12-19]. This requires pre-
forming architecture were also extensively investigated and cise excitation of each antenna element by various synthesis
gradually became mature [4-6]. techniques to obtain desired array pattern features, e.g., high
After entering into the 21st century, with element level dig- directivity, low SLL, and null control. However, because of
itization and fully digital beamforming (DBF) architecture, mechanical manufacturing imperfections and circuit differ-

VOLUME 4, 2016 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

FIGURE 1. The main topics and general flowchart of this paper, including analysis methods, calibration methods, and signal processing techniques of excitation
errors for phased array antennas.

ences, phased array antennas show element to element vari- element excitation including both amplitude and phase, and
ation. Moreover, due to component aging and temperature predict the array performance by its upper and lower bounds
variation, the realistic antenna element excitations inevitably [30-40]. Thanks to the inclusion property of IA to deal
differ from their expected values in practice. These errors with uncertainties, the determined bounds of antenna array
would cause array pattern distortion, gain degradation, worse pattern are finite and inclusive thus reliable. Moreover, with
SLL, and null drifts [20-23]. Therefore, knowledge about respect to standard time-consuming Monte Carlo (MC) based
the impacts of the excitation errors and the corresponding strategies, IA based methods allow analyzing the effects of
calibration methods are of high importance for phased array array excitation errors by means of analytic expressions in a
antennas. short time.

This paper aims to present a tutorial-like review of impact To compensate for the excitation errors as low as possible
analysis and state-of-the-art calibration methods to overcome and maintain acceptable array performance, phased array an-
excitation errors for phased array antennas. Many methods tennas are required to be carefully calibrated before their de-
have been proposed to analyze the impacts of excitation ployments and during their work. These calibration methods
errors on array performance. These analysis methods fall vary from different systems and applications, either suitable
into mainly two different types, probabilistic methods and for factory test or for in-field test [41]. Some methods such
interval arithmetic (IA) based methods. The probabilistic as the near-field scanning probe method and the peripheral
methods assume that the excitation errors follow some kind fixed probe method require additional equipments [42-60],
of probability distribution function (PDF) and treat this while the mutual coupling method takes advantage of the
problem via statistical derivation [24-29]. They can provide inherent property of the phased array without the need for
simple closed-form expressions of the features of the array extra hardware [61-71]. For Si-based integrated phased array
pattern by exploiting the central limit theorem. Although well antenna, it’s very common to design a dedicated coupling
established, statistical methods have not always guaranteed network that using transmission lines connected to each
reliable confidence bounds, some extreme cases can still antenna element for periodic in-field calibration [72-87].
fall outside the bounds. On the contrary, IA based methods Moreover, different trade-off strategies can be made between
introduce intervals to represent the possible values of the the complexity of measurement setup and signal process-
2 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

TABLE 1. The Example of 10-element Taylor Tapered Linear Antenna Array

for Probabilistic Analysis.

Quantity Value
Element number N = 10
Inter-element distance d = λ/2
Amplitude Tapering Taylor tapered with
SLLref = -20 dB
and n̄ = 2
Excitation amplitudes a1 = a10 = 0.542
a2 = a9 = 0.629
a3 = a8 = 0.771
a4 = a7 = 0.913
a5 = a6 = 1.000
Locations of the first nulls θ = ±17 deg
Locations of the first sidelobes θ = ±19 deg
Standard deviation of the am- σa = 0.05
plitude error
Standard deviation of the phase σp = 5 deg
error

FIGURE 2. The referenced array pattern of the 10-element Taylor tapered

linear antenna array, with Monte Carlo simulations are presented with standard
deviations of the amplitude error σa = 0.05 and phase error σp = 5 deg.

ing requirements. Some calibration techniques characterize

the antenna elements one by one, while some others can
N
handle several calibrations simultaneously. Some calibration f (θ, φ) =
P
fn (θ, φ)
techniques require the calibration source and sink must be n=1
coherent, while some others require no such synchronization N →
−
an exp[j(2π/λ k (θ, φ) · →
−
P
= rn + pn )] (1)
and can be done by amplitude or power only measurements. n=1
PN
Figure 1 illustrates the main topics and general flowchart = an exp(jbn )
n=1
of this paper, including analysis methods, calibration meth- →
−
ods, and signal processing techniques of excitation errors for where λ is the wavelength, k (θ, φ) is the unit vector of the
phased array antennas. The rest of this paper is organized observation direction (θ, φ) and →−
rn is the position of the n-th
as follows. First in Section 2, the mathematical model of antenna element, fn (θ, φ) is the array factor contributed by
excitation errors is introduced along with the probabilistic the n-th antenna element and bn is its phase. Note that {bn }
analysis methods. Then in Section 3, according to the ge- is also a function of the observation angle (θ, φ), however, to
ometry of the excitation error representation, three kinds simplify the notation this angle dependence is not shown in
of IA based methods are presented. Next, state-of-the-art the following discussion. Without considering position errors
→
−
calibration methods for phased array antennas are reviewed of the antenna elements, the term k (θ, φ) · →−
rn is constant at
in Section 4, their advantages and challenges are discussed any given direction, then the two phases {bn } and {pn } have
in a comparative manner. Some signal processing techniques the same statistics and can be treated interchangeably.
for phased array calibration are also reviewed in Section 5. There are generally two types of excitation errors, namely
Two other common errors, position errors and quantization correlated errors and random errors. Usually, it is the in-
errors, are also discussed in Section 6. Then in Section 7, two tent of the designer to ensure that the correlated errors are
open research directions are presented to encourage future appropriately removed once identified, so that all remained
research about phased array calibration. Finally, this paper is residuals are uncorrelated and random, which are difficult to
concluded in Section 8. compensate due to their randomness [24-27]. Hence, only
random excitation errors are considered in the following
discussion. With excitation errors, the real antenna element
excitation can be described according to its referenced value,
II. PROBABILISTIC METHODS FOR IMPACT ANALYSIS
areal
n = an (1 + γn )
(2)
A. THE MATHEMATICAL MODEL preal
n = p n + δn
and without position errors,
Considering a phased array antenna with N isotropic ele-
ments, each element has the excitation amplitude {an } and
breal
n = bn + δn (3)
phase {pn }, n = 1,..., N . The array radiation pattern or
the array factor f (θ, φ) can be mathematically expressed as where areal
nand preal
are the real excitation amplitude and
n
follows, phase of the n-th antenna element, and the random variables
VOLUME 4, 2016 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

2
γn and δn represent its relative amplitude error and phase And the variances of these two parts,σR (θ, φ) and σI2 (θ, φ)
error in percentage and degree, respectively. , have the following expressions,
Without loss of generality, the variances in the amplitude
and phase error are assumed to be identical among antenna N
(θ, φ) = 12 (1 + σa2 ) 1 − exp(−2σp2 )
2
 P
elements. According to the central limit theorem, both of σR a2n
these random variables are Gaussian distributed and have n=1
N
zero mean,
 P
+ (1 + σa2 ) exp(−2σp2 ) − exp(−σp2 ) a2n cos2 bn
n=1
N
γn ≈ N (0, σa2 ) 1
 P
σI2 (θ, φ) = 2 (1 + σa2 ) 1 − exp(−2σp2 ) a2n
(4)
δn ≈ N (0, σp2 ) N
n=1

a2n sin2 bn
 P
+ (1 + σa2 ) exp(−2σp2 ) − exp(−σp2 )
where σa and σp are the standard deviations of the excitation n=1
amplitude and phase error, respectively. (8)
Figure 2 illustrates an example of a half-wavelength e- Figure 3 illustrates the expectations and variances of the
qually spaced (d = λ/2) 10-element linear antenna array, real and imaginary parts of the array pattern. As can be
which is used as the benchmark architecture in the following seen, the expectation for the real part is symmetric while
discussion. The referenced array pattern is Taylor tapered the imaginary part is anti-symmetric. Their variances closely
with SLLref = -20 dB and n̄ = 2, and the corresponding follow each other at most of the angles except at the boresight
parameters and excitation amplitudes are summarized in and the grating lobe. This is because the first term of their
Table 1. It has the main lobe with the first nulls at θ = ±17 variances are the same in (8), while the latter term differs in
deg, and the highest SLLs (the first sidelobe at θ = ±19 deg) their values especially in these two regions.
are approximately equal to -20 dB as designed. Some Monte From (7) and (8), the expectation of the array power
Carlo simulations are also presented with σa = 0.05 and σp pattern P (θ, φ) can be derived as,
= 5 deg, whose results show some performance degradation,
i.e., higher SLL and broader main beam.



B. THE EXPECTATION OF THE ARRAY PATTERN E (P (θ, φ)) = E (fRreal (θ, φ))2 + E (fIreal (θ, φ))2
= u2R (θ, φ) +"σR
2
(θ, φ) + u2I (θ, φ) + σI2 (θ, φ)
The expected array pattern aims to evaluate the average pat- N 2  N 2 #
tern performance and provide approximate formulas for the = exp(−σp )2
P
an cos bn +
P
an sin bn
confidence boundaries. From (1)-(3), the real array pattern n=1 n=1
f real (θ, φ) is rewritten as,  N
P
+(1 + σa2 ) 1 − exp(−2σp2 ) a2n
n=1
 N
P
N
X + (1 + σa2 ) exp(−2σp2 ) − exp(−σp2 ) a2n
f real (θ, φ) = an (1 + γn ) exp [j(bn + δn )] (5) N
n=1
 P
n=1 = exp(−σp2 )P ref (θ, φ) + 1 + σa2 − exp(−σp2 ) a2n
n=1
Generally, the array pattern is a complex value and can be (9)
splitted into the real and imaginary parts, where P ref (θ, φ) is the normal/ideal array power pattern
without excitation errors,
N
X
fRreal (θ, φ) = an (1 + γn ) cos(bn + δn )
n=1
N
(6)  2
N
X ref
P 2 
fIreal (θ, φ) = an (1 + γn ) sin(bn + δn ) P (θ, φ) = |f (θ, φ)| = 
 an exp(jbn )
n=1
"
N
2  n=1N
2 # (10)
P P
= an cos bn + an sin bn
From the Biggelaar’s work [27], the expectations of these n=1 n=1
two parts, uR (θ, φ) and uI (θ, φ), can be derived as follows,

This expression in (9) shows that the effect of random

N
X excitation errors produces the real array power pattern con-
uR (θ, φ) = E fRreal (θ, φ) = exp(−σp2 /2)


an cos bn
sisting of the ideal pattern reduced by a factor that accounts
n=1
N
for phase error, plus another term that is a constant with
uI (θ, φ) = E fIreal (θ, φ) = exp(−σp2 /2)

X
an sin bn no angular dependence [13]. It is convenient to normalize
n=1
the latter term to the peak of the resulting pattern, and the
(7) normalized sidelobe level is given by,
4 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

N
[1+σa2 −exp(−σp2 )] a2n
P
n=1
SLL = N
P 2
exp(−σp2 )( an )
n=1
2
,

N
P N
2
exp(σp2 ) − 1 a2n
P
= 1 + σa ( an ) (11)
n=1
, n=1 2


PN N
≈ 1 + σa2 1 + σp2 − 1 a2n (
P
an )
n=1 n=1


≈ σa2 + σp2 N ηA
where ηA is known as the array taper efficiency [28],

N 2, N
X X
ηA = ( an ) N a2n (12)
n=1 n=1

The symmetrical form of the amplitude and phase errors in

(11) suggests the convenience of converting them from one to
the other. This helps in the trade-off to determine how much
of the error variance to allot between the two [25]. Figure 4
presents several curves of average sidelobe level due to the
phase and amplitude errors for the linear array as presented
in Table 1. Due to their interchangeability, each sidelobe level
curve is like an arc.

C. THE PDF OF THE ARRAY PATTERN

As can been seen in Fig. 3, for most regions except at
the bore-sight, grating lobe, and other selective regions, it’s
reasonable to assume that the real and imaginary parts of the
array factor have identical mean value and share the same
variance. Hence, as sum of these two Gaussian distributed
variances, the array pattern at these regions follows Rician
distribution with the PDF defined as, FIGURE 3. The expectations and variances of the real and imaginary parts of
the array pattern for the 10-element Taylor tapered linear antenna array.
prob(r(θ, φ))
 =
r 2 (θ,φ)+u2r (θ,φ) (13)
  
r(θ,φ) rur (θ,φ)
σr2 (θ,φ) exp − σr2 (θ,φ) · I0 σr2 (θ,φ)

where I0 () is the modified Bessel function of the first kind

with order zero, and
p
ur (θ, φ) = u2R (θ, φ) + u2I (θ, φ)

(14)
σr2 (θ, φ) = 21 σR
2
(θ, φ) + σI2 (θ, φ)
Generally, the same variance assumption is good enough
for most analyses. However, it has been pointed out that this
assumption is not mandatory and a Beckman distribution can
be derived [27]. For other special regions that have uniden-
tical variances for the real ad imaginary parts, such as the
boresight (uI = 0) and null locations (uR = uI = 0), detailed
mathematical derivation can be found in Bhattacharyya’s
excellent book [29].

III. IA BASED METHODS FOR ARRAY PATTERN BOUND

ESTIMATION
For practical antenna array engineering, it’s of great interest
to evaluate the worst performance of the array pattern along FIGURE 4. Array average sidelobe level due to phase and amplitude errors
for the 10-element Taylor tapered linear antenna array.
with statistical characteristics. Recently, the IA technique
VOLUME 4, 2016 5

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

was introduced by P. Rocca et al. [30-36] to evaluate the other. If an interval variable occurs several times in the
effects of excitation errors and the corresponding upper and expression, and each occurrence is taken independently, then
lower bounds of the array patterns. Thanks to the intrinsic of this so-called dependency problem will cause expansion of
IA to deal with uncertainties, it’s very efficient to determine the resulting intervals. In the above derivation, the array
the bounds of array patterns in a robust and reliable way. It excitation error occurs in both the real and imaginary parts,
needs no detailed information unless the knowledge of the this will cause an unwanted overestimation when adding
intervals about the excitation errors, which are the only terms them together [37-38].
involved in the IA based mathematical operations. As shown
in Figure 5, according to the geometry of the excitation error B. THE CIRCULAR IA METHOD
representation, three IA based methods have been proposed, For the circular IA method, as shown in Fig. 5(b), the ge-
i.e., the Cartesian IA method, the Circular IA method, and ometry of the excitation error is represented into the circular
the Polygonal IA method. form. It’s denoted as < f (θ, φ), ρ(θ, φ) >, where f (θ, φ)
is the center and ρ(θ, φ) is the radius covering all possible
A. THE CARTESIAN IA METHOD excitation error values[35]. The advantage of the circular IA
As shown in Fig. 5(a), a specified excitation error has a fan- method is the simplicity of IA calculation, the radius of the
like geometry defined by the boundaries of amplitude interval array pattern is the sum of the radius of the interval of each
[aL , aU ] and phase interval [bL , bU ]. Please keep in mind that element, respectively,
bL and bU are also dependent on the observation direction
N
(θ, φ). For the Cartesian IA method, the error geometry is X
ρ(θ, φ) = ρn (θ, φ) (21)
expanded into the rectangular form for convenience [30-31].
n=1
Again, the complex array pattern of (1) is rewritten with the
real part fR (θ, φ) and the imaginary part fI (θ, φ), where ρn (θ, φ) is the radius covering the excitation error of
the n-th element,
f (θ, φ) = fR (θ, φ) + j · fI (θ, φ)
N N (15)
 
=
P
an cos bn + j ·
P
an sin bn ρn (θ, φ) = max{an exp(jbn ) − aLn exp(jb

L 
n) ,
n=1 n=1 an exp(jbn ) − aL U 
n exp(jbn ) ,

an exp(jbn ) − aU L  (22)
The array power pattern P (θ, φ) can be obtained by first  n exp(jbn )  ,
an exp(jbn ) − aU U 
n exp(jbn ) }
squiring these two parts then adding them together,
P (θ, φ) = PR (θ, φ) + PI (θ, φ) The bounds of the array power pattern are obtained by first
2 2 (16) calculating the bounds of the array pattern and then squaring
= |fR (θ, φ)| + |fI (θ, φ)|
it,
Since an > 0, the bound of the real part of the array pattern 2
P U (θ, φ) = 
(|f (θ, φ)| + ρ(θ, φ))
[fRL (θ, φ), fRU (θ, φ)] can be derived as 2
 (|f (θ, φ)| − ρ(θ, φ))
N
L P L (θ, φ) = if |f (θ, φ)| > ρ(θ, φ)
fRL (θ, φ) = aL
P
n (cos bn ) 0 otherwise

n=1 (17)
N (23)
U
fRU (θ, φ) aU
P
= n (cos bn ) In this way, the excitation of the antenna element as an
n=1
interval variable appears only once in (21). However, the
where radius is enlarged due to the expression of the circular IA
form.
−1  if (2k + 1)π ∈ [bL , bU ]

L
(cos b) =
min cos(bL ), cos(bU ) otherwise C. THE POLYGONAL IA METHOD
(18) As shown in Fig. 5(c), a more compact geometry is to expand
the excitation error region as a convex polygon. The outer
∈ [bL , b U ] arc of the possible region is surrounded by several tangent

1 if 2kπ
(cos b)U = lines, while the inner arc is surrounded by a line between the
max cos(b ), cos(bU )
L
otherwise
(19) two vertices. In such a case, the computation procedure of the
A similar expression can be obtained for the imaginary sum of two convex polygons is known as the Minkowski sum.
part, and the interval of the array power pattern is as, Let V and W be two convex polygons representing possible
regions of the excitation errors of two antenna elements with
P L (θ, φ) = PRL (θ, φ) + PIL (θ, φ)
(20) K vertices {v1 , ..., vK } and M vertices {w1 , ..., wM }, re-
P U (θ, φ) = PRU (θ, φ) + PIU (θ, φ)
spectively. Their Minkowski sum can be computed as follows
It should be mentioned that, due to the intrinsic of the [39],
complex notation, the real and imaginary parts of the array Step 1: initial k =1, m =1;
pattern in (15) are dependent and highly correlated to each Step 2: add vk + wm as the vertex of the sum.
6 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

FIGURE 5. Illustration of three IA based methods, (a) the Cartesian IA method, (b) the Circular IA method, and (c) the Polygonal IA method. The boundary of each
element is defined by the amplitude interval [aL , aU ] and phase interval [θ L , θ U ]. [aL U L U
R , aR ] and [aI , aI ] represent the interval of the real and imaginary parts in
the Cartesian IA method, respectively. The boundary for Circular IA method is represented with < f, ρ > denoting the center f and the radius ρ, while the
Polygonal IA method uses a convex polygon with K vertexes {vn }.

Step 3: if angle(vk vk+1 ) ≤ angle(wm wm+1 ), then k = k+1; and applications, either suitable for factory test or for in-
else if angle(vk vk+1 ) ≥ angle(wm wm+1 ), then m = field test. According to Seker’s work [41], these calibration
m+1; methods can be generally divided into four main categories,
else k = k+1, m = m+1. i.e., the near-field scanning probe method, the peripheral
Step 4: repeat step 2 and step 3 until k = K+1 and m = fixed probe method, the mutual coupling method, and the
M +1. built-in network method. The advantages and challenges of
The above procedure runs in linear time because at each these calibration methods are shortly summarized in Table 2
execution of the repeat loop either k or m is incremented. and presented in the following subsections.
Moreover, the Minkowski sum is also a convex polygon with
at most K + M edges [39]. Hence, By sequentially adding A. THE NEAR-FIELD SCANNING PROBE METHOD
each element with N -1 times of Minkowski sum calcula- The traditional near-field scanning probe method, also known
tion, the interval of the array pattern can be obtained with as "park and probe", is one of the most widely used methods
improved bounds thus more reliable results. An extensive in the industry for phased array antenna calibration [42-
comparison of these three IA based methods with repre- 46]. It’s considered to be very reliable and accurate with
sentative numerical results can be found in [32]. Generally element-level pattern accuracy as high as 0.1 dB and 0.1 deg,
speaking, the Polygonal IA method performs the best for respectively [42]. The procedure is straightforward, the tested
most scenarios, while the Circular IA method performs the element is excited to its default state, a robotic manipulator
worst. An analytic method is also proposed to analyze and places a near-field scanning probe antenna at the boresight
further dig the information from the shape of the convex of the tested element, to directly measure the relative phase
polygon of the array pattern, and map it into the probability and amplitude by Vector Network Analyzer (VNA). Then
distribution of the bound interval [40]. it moves to test the next element under the same testing
condition, and repeats this procedure until all elements of the
IV. CALIBRATION METHODS antenna array are tested. As depicted in Fig. 6(a), supposing
The objective of phased array antenna calibration is to sample gpt is the transmit gain of the calibration probe, Cpn is the
each antenna element in the array, and compare the obtained coupling coefficient between the probe antenna and the n-th
results to identify the differences among the elements. Phased antenna element under test, and gnr is the receive gain of the
array calibration for both transmit and receive are similar, n-th element to be calibrated. Then the signal received by the
the primary differences lie in the test-signal distribution at n-th antenna element Rpn can be formulated as,
the input and the combination at the output. For receive
calibration, a calibration source and/or a distribution network Rpn = gpt Cpn gnr (24)
is required to inject the test signal into the input of each
Analogously, the signal received by the m-th antenna
element, the array beamformer can be used as a test signal
element Rpm can be formulated as,
combiner. For transmit calibration, it requires a calibration
source signal into each element and additional equipment
Rpm = gpt Cpm gm
r
(25)
or beam combiner for combined signal monitoring. Only
receive calibration is presented here to keep the discussion Since the technique scans and tests all the antenna ele-
focused, the extension for transmit calibration is straightfor- ments under the same conditions, it assumes that the coupling
ward. The calibration methods vary from different systems coefficient between the probe and the tested element is the
VOLUME 4, 2016 7

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

FIGURE 6. Simplified diagram of the phased array calibration methods.

TABLE 2. The advantages and Challenges of Phased Array Calibration Methods.

Method Advantages Challenges

The near-field scanning probe method Reliable and accurate. Requiring a precise automated mechanical system.
Time-consuming.
Only suitable for in-factory test.
The peripheral fixed probe method Widely used in-field test. Requiring additional hardware.
Calibration can be done in parallel. Involving complicated signal processing.
Calibration can be done in noncoherent.
The mutual coupling method No external equipment required. Would encounter unwanted edge effect.
Calibration can be done in parallel.
The built-in network method High accuracy and efficiency. Requiring dedicated coupling network.
Calibration can be done in parallel.
Calibration can be done in noncoherent.

8 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

same for all array elements, i.e., Cpn = Cpm . Hence, the
relative excitation error between these two elements can be
obtained as,

gnr Rpn
r
= (26)
gm Rpm
Usually, probe position error will result in some of the
measurement error in near-field measurement [47]. However,
for phased array near-field scanning probe calibration, it is
not necessary to keep the probe antenna precisely at the
boresight of each antenna element under test. It is sufficient
to just keep the relative position and angle between the
probe antenna and the tested element unchanged. In that
case, even if the probe is somewhat offset from its desired
position for every antenna element, the amount of offsets
and therefore couplings will be still the same for all tested
FIGURE 7. Coupling schemes for rectangular phased array calibration by
antenna elements, making this technique still useful. performing the mutual coupling method [41, 70].
The near-field scanning probe method usually requires
a precise automated mechanical system including elements
such as the axes controllers, actuators, motors for accurate if the coupling amounts were previously measured or cali-
probe movement. This makes the setup suitable to work only brated and moved out. This can be done by using a near-
in the near-field environment, but too complicated to arrange field scanning probe in factory test, then compared with the
it in the field test. Moreover, due to the mechanical movement results from peripheral fixed probes made in-field to calculate
of the probe antenna, the measurement process is very time- the couplings. For example, the SAMPSON Multi-Function
consuming, especially for large-scale phased array antennas. Radar has four fixed open waveguide auxiliary radiators used
For example, for the THAAD (Theater High Altitude Area for calibration [60]. The signal received from an individual
Defense) radar, each of the 25,344 T/R modules had to be active channel is compared with a stored reference level ob-
near-field scanned individually, result in a very long and tained during the factory test of the phased array antenna. In
even unacceptable test time [48, 49]. Therefore, the near-field this way, a replacement module or one whose characteristics
scanning probe method is most suitable for initial factory test have simply drifted with time may be reset to the original
rather than periodic in-field calibration. RF performance, thus restoring the original factory-fresh
condition of the antenna. With the peripheral fixed probe
B. THE PERIPHERAL FIXED PROBE METHOD method, some advanced signal processing techniques can be
In some large phased array antennas, unlike the near-field adopted to reduce the complexity of measurement hardware
scanning method with a moveable probe antenna, one or requirements, which is presented in detail in Section 5.
several fixed probe antennas are placed at the periphery of
the array [50-54], or moving platform such as a balloon or C. THE MUTUAL COUPLING METHOD
unmanned aerial vehicle (UAV) [55, 56]. In [57], calibration The mutual coupling method was first proposed by H.M.
is made by external fixed probe antennas at certain locations Aumann et al. [61] on a linear antenna array, based on
near the array. A circle around the array center is chosen for the idea that the inherent mutual coupling among the array
symmetry reasons. It’s proposed that one probe antenna is in- elements can be used by transmitting from an element and re-
tegrated at the side for a spaceborne phased array calibration ceiving from another. The measured signals between all pairs
[58]. High calibration accuracy can be achieved for actual of elements in the array allow a complete characterization
satellite systems because this allows easy realization with of the relative amplitude and phase of each element in the
high signal-to-noise ratio (SNR). The Space Fence, used to array. Compared to the near-field scanning probe method and
detect, track, and catalog small objects in Space Situational the peripheral fixed probe method, this method utilizes the
Awareness for the U.S. Air Force Space Surveillance Net- inherent property of mutual coupling and requires no external
work, both the transmit and receive arrays are calibrated with hardware resources, resulting in a much faster calibration
horn antennas mounted on calibration towers in the near-field process and more suitable for in-field calibration [62-68].
[59]. The mutual coupling method was extended to two-
If the fixed probe is at the far-field of the array under dimensional (2D) antenna array and became more practical-
test, then the effects of each antenna element can be viewed ly useful [63]. The detail of the mutual coupling method
as the same. Otherwise, the coupling between the probe depends on the specific array geometry. Taking the most
and each antenna element is going to be different. In this common rectangular array as an example, as shown in Fig.
case, the peripheral fixed probe method can be applied only 7, there are four elements under test and two parameters
VOLUME 4, 2016 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

t
need to be calibrated for each element. Let gm denotes the are built using the same material and have the same radius
r
transmit gain of the m-th element and gn denotes the receive and length, their effects would be very similar. For example,
gain of the n-th element to be calibrated. A mutual coupling the ELTA/IAI Systems used microstrip line couplers near
measurement consisting of a signal transmitted from the m- parallel plate waveguide radiators to individually probe T/R
th element and received by the n-th element. When the #2 modules through the analog beamformers to supplement ini-
and #3 elements as receive and the #1 and #4 elements as tial near-field calibration. The TerraSAR-X imaging satellite
transmit, two pairs of mutual coupling measurements are was one of the first SAR systems using embedded coupling
formulated as, to supplement the initial calibration once it is launched into
space [84,85]. For the S-band digital array testbed at Lincoln
R12 = g1t C12 g2r , R13 = g1t C13 g3r Laboratories, a coupler was used for injecting in front of the
(27)
R42 = g4t C42 g2r , R43 = g4t C43 g3r LNAs for receive-only calibration monitoring [86]. In [87], a
coupled line network that is weakly coupled to the antenna
From these four measurements, these elements can be array is designed to provide a calibration signal path without
calibrated relative to each other as, adding any switch after the antenna port.
1/2 1/2 V. SIGNAL PROCESSING TECHNIQUES FOR PHASED
g2r g1t
 
R12 R42 R12 R13
= , = (28) ARRAY CALIBRATION
g3r R13 R43 g4t R42 R43
For the peripheral fixed probe method and the built-in net-
From the other three topographies with six more pairs work method for phased array calibration, it’s very con-
of measurements, the relative transmit and receive gains of venient to apply advanced signal processing techniques to
all four elements can be solved. Moreover, a mathematical save hardware or time requirements. As shown in Fig. 8,
framework by the least square method is proposed that using these techniques can be distinguished as serial or parallel
measurement results as many as possible [69]. by the number of elements to be calibrated each time. In
There is also something to be mentioned for the mutual serial calibration techniques, the antenna elements are char-
coupling method. It relies on the assumption that coupling acterized one by one, and the characterizations of different
between an antenna element and its neighbors are the same array elements are independent. Some techniques save time
for all elements in the array. Nonetheless, it was revealed by by handling several calibrations simultaneously, thus can
using a prototype system, it does not hold for small arrays due be viewed as more efficient. These techniques can also be
to unwanted edge effects [70-71]. A before/after approach distinguished as coherent or noncoherent by the measure-
is introduced to avoid errors introduced by edge effects, the ment requirements between the calibration source and the
ratio between the after and before status will quantify the element under calibration. For coherent measurement, it’s
changes suffered by the elements. required that the calibration source and sink must be coherent
to obtain the complex value containing both amplitude and
D. THE BUILT-IN NETWORK METHOD phase information. On the contrary, noncoherent measure-
It’s very common to use a built-in network connected to each ment requires no such synchronization, which can be done by
antenna element for periodic in-field calibration of integrat- amplitude or power only measurements. All these four kinds
ed phased array antennas, especially for Si-based RF and of techniques can be conducted with both the peripheral fixed
mmWave-band integrated phased array [72-83]. As shown probe method and the built-in network method, only slight
in Fig. 6(d), this technique employs microstrip transmission differences in the measurement setup are required depending
line as test signal injector or weak signal coupler embedded on practical applications.
under the element for calibration, with the coupling ratio
between the transmission line and the radiating element A. SERIAL AND COHERENT TECHNIQUES
usually several tens of dB. These calibration lines sample The serial and coherent techniques simply calibrate each el-
the signals received or transmitted by the antenna elements. ement serially and require a coherent measurement to gather
The measured signals are then used to calculate the amplitude the element excitation information. The near-field scanning
and phase differences among the elements. An early example probe method can also be conducted as one of such kind of
of built-in performance monitoring system was developed techniques. For some smart antenna applications [88] and
and tested in [78]. Test distribution networks can also be mobile satellite communication applications [89,90], these
integrated with advanced signal processing techniques on- simple and direct calibration techniques are very common,
chip as described in Section 5. especially with the element-level digital phased array. One
To successfully calibrate a phased array using this method, of the main advantages of digital beamforming over every
phase shifts and amplitude losses caused by the transmission element is the ease that amplitude and phase errors among
lines and the couplers that connect them should be equal elements can be calculated in the digital domain.
or already known. A high-quality, embedded calibration net- There are also some measurement setups involving phase
work near the array face could source or receive nearly iden- toggling. In [78], the phase of the element under calibration
tical signal levels at each element. If the transmission lines is switched between two states of 0 and π, while the other
10 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

FIGURE 8. Signal processing techniques for phased array calibration.

elements stay unchanged. By comparing the combined array (power) measurements to determine the complex excitation
signal of these two measurements, the element excitation of one array element.
can be easily determined, while the other stationary element In [98-99], multiple noncoherent power detectors are dis-
outputs can be canceled out. tributed across the array for amplitude and phase extrac-
tion. Power detectors can also be placed between adjacent
B. SERIAL AND NONCOHERENT TECHNIQUES elements, allowing the measurement of relative phase in
Compared to coherent measurement methods, noncoherent addition to amplitude [100].
techniques which involve amplitude or power only mea-
surements are considered to be more convenient. In such C. PARALLEL AND COHERENT TECHNIQUES
a case, coherent measurements are no longer required with Compared to serial techniques, parallel techniques that si-
precise synchronization between the calibration source and multaneously calibrate a number of array elements are more
the element under calibration. However, this is usually done efficient, which is especially suitable for large phased arrays
with some additional costs. e.g., more power detectors or with small calibration time slots. Some parallel techniques
more times of measurements. The rotating element electric measure the array combined signals with several measure-
field vector (REV) method is the most well-known power ment probes at multiple positions [101]. However, this re-
only measurement method with only one power detector. quires some hardware burden. A more popular alternative
It measures the amplitude of the array combined signal by technique allows simultaneously measuring excitations of
shifting the element phase from 0 to 2π continuously [91- multiple elements relying on code modulation [102-106].
94]. The element excitation can be determined from statistics A method that controls the element phases based on time
of three parameters, the maximum and minimum power of multiplexed orthogonal codes was proposed in [102], thus
the array signal, and the element rotation phase to maximum the individual element excitation can be derived from the
power. combined array signal. The pseudo-random (PN) code for
In [95], it was pointed out that measurement results at phased array calibration has been successfully verified in a
four orthogonal phase shifts are sufficient enough to obtain space born environment by TerraSAR-X [107]. In addition, a
a maximum likelihood estimation for each array element. similar technique with Walsh codes was used for Sentinel-
An even simpler expression was further derived in [96] 1[108]. A recursive matrix-forming method for Hadamard
that requires only two phase shifts of π/2 and π to yield matrix construction is presented in [109] for phased array
the element complex excitation information. Another novel calibration.
power only measurement method is derived to calibrate the Besides these aforementioned methods, the spectrum anal-
element excitation [97]. For each array element, two power ysis technique can also be adopted for phased array parallel
measurements are used to determine the element complex calibration. In [110,111], a multi-element phase toggling
excitation, one by shifting the element’s phase of π/2 and the method with Fast Fourier Transform (FFT) based spectrum
other by turning the element under test off. From the math- analysis is proposed for parallel calibration. Each element
ematical point of view, it’s required at least two amplitude under calibration is toggled with a particular odd time of
VOLUME 4, 2016 11

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

fundamental step frequency. By applying FFT analysis of the representing element position errors is proposed in [120],
combined complex signal, both the amplitude and phase of which considers both the relevance and the randomness of
each element can be found in the corresponding spectrum. the position errors of adjacent planar array elements. The
IA based method is also applied to evaluate the effects of
D. PARALLEL AND NONCOHERENT TECHNIQUES element position errors on the statistic performance [121].
The parallel and noncoherent techniques, with multi-element
element calibration each time and no coherent measuremen- B. QUANTIZATION ERRORS
t required, should be the most preferred technique at the With the advance of modern RF technologies, digital am-
expense of more complex signal processing. In [112], the plitude attenuators and digital phase shifters are nowadays
phases of multiple antenna elements are successively shifted extensively used than analog ones. Thus, the optimized
with the specified phase intervals. The measured array power continuous amplitude and phase must be rounded off to
variation is expanded into Fourier series and the terms are the closest quantized value for implementation in practice
rearranged to put them into the form of the conventional REV [122, 123]. Since the errors of these approximate phase and
method. In [113], another spectrum analysis based method amplitude are highly correlated rather than random, this often
is proposed by periodically modulating the element phase yields large grating lobes and degrade the array sidelobe
sequentially. The phase of the antenna element is shifted to performance. Detailed derivations about array performance
π on one time slot while staying unchanged for the other degradation and grating lobe levels due to quantization can
time slots. By applying FFT analysis of the combined array be found in [13]. To reduce the grating lobe, it is common
signal, the element complex excitation is obtained from the engineering practice to randomize the amplitude and phase
corresponding modulation frequency harmonics. quantization errors thus break up the periodicity. A detailed
In [114, 115], a novel calibration technique that employs comparison of several methods can be found in [124] that are
code-modulated interferometry is proposed for parallel mea- used to reduce the peak sidelobe level.
surements. Moreover, it requires only a simple power squar-
ing detector without the need for a coherent receiver. First, VII. OPEN RESEARCH DIRECTIONS
the common test signal in the form of a single-frequency tone Currently, phased array calibration is still one of the hottest
injected into each element, both the in- and quadrature-phase research topics for phased array technology. Many aspects
components of each element are encoding, then combined are expected to be improved with higher efficiency and better
and squared using an on-chip power detector. Benefiting performance. Two open research directions, from the au-
those orthogonal code products (OCPs) where the product thors’ personal point of view, are presented here to encourage
of any two codes is another unique code, the squaring op- future researches about phased array calibration.
eration downconverts the combined signal to baseband and
creates an interference pattern between all of the individual A. WIDEBAND PHASED ARRAY CALIBRATION
elemental responses. This pattern contains complex cross- The higher data rate requirements for future wireless com-
correlations that are each modulated according to the OCP. munications have significantly driven the need for wideband
Using each OCP, the correlation of interest can be demod- phased array antennas. Future phased array antennas must
ulated, and then the full set of correlations can be used to address very wideband signals as well as nonlinear behaviors,
extract amplitude and phase information for all elements. adaptivity, and reconfigurability. Hence, array calibration
techniques must have to be developed to meet such chal-
VI. SOME OTHER COMMON ERRORS lenges as spurious signals and pulse dispersion for wideband
Although the excitation errors are analyzed in this paper, antenna arrays [125,126]. The phase-shift approximation
the real phased array antenna will suffer some other errors used in the narrowband beamformer implementation is not
in practice. Here two other common errors of phased array valid for processing wideband signals. True time delay (TTD)
antenna, position errors and quantization errors, are shortly units rather than phase-shifters are required for wideband
discussed to inspire the readers’ broader envision for phased beamforming. Several methods have been proposed for wide-
array antenna calibration. band phased array calibration. For very long baseline inter-
ferometry (VLBI) in radio astronomy, comb tones with a step
A. POSITION ERRORS frequency of several MHz are used to provide wideband and
As described in Section II-A, the position errors of array high-stability calibration. The comb tones are injected into
elements will also influence the performance of the array. the receiver front end to calibrate the phase delay of receive
A large number of studies have been focusing on the po- channel over wide band [127-129]. Another mainstream idea
sition tolerance of antenna elements [116-119]. A coupled is to compensate for the channel imbalances corresponding
structural-electromagnetic model is developed to evaluate to the channel response. In [130], a digital wideband phased
the effects of random position errors on the performance of array working on 1-8 GHz is first characterized using impulse
phased antenna arrays [117,118]. The relationship between response measurement. A chirp signal is injected as the prob-
the position tolerance of the antenna element and the gain ing signal to measure all 16 channels simultaneously. Based
loss of the array is derived in [119]. A conical scheme for on the spectrum-modified time reversal technique, the im-
12 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

pulse response is compensated and both phase and amplitude of IA to deal with uncertainties, the determined bounds of
are realigned for all elements. In the future, more efficient antenna array pattern are finite and inclusive thus reliable.
calibration methods are expected to reduce the complexity of To meet the requirements and maintain acceptable perfor-
measurement setup and calibration signal processing. mance, phased array antennas must be carefully calibrated to
compensate the errors as low as possible. These calibration
B. MACHINE LEARNING BASED CALIBRATION methods vary from different systems and applications, either
TECHNIQUES suitable for factory test or for in-field test. The near-field
scanning probe method that employs a robotic scanner is
The rise and proliferation of artificial intelligence (AI) has
well-known and widely used as the standard for factory test.
found numerous applications including natural language pro-
The peripheral fixed probe method requires external probes
cessing, remote sensing, image recognition, and fraud de-
with calibration signal processing. In contrast, the mutual
tection. Machine learning (ML) approaches are scientific
coupling method takes advantage of the inherent property
disciplines that build a mathematical model based on training
of mutual coupling among elements of the array, and uses
data, to enable AI by improving an outcome through the ex-
it to avoid the employment of external equipment. Some
perience without being explicitly programmed. Some popular
dedicated coupling network that using transmission lines
ML techniques are radial basis functions (RBFs), support
connected to each antenna element for the periodic in-field
vector machines(SVMs), artificial neural networks (ANN),
calibration of integrated phased array antenna, especially
and deep neural networks (DNNs), just to name a few [131,
for RF and mmWave-band integrated phased array. It’s very
132]. Recently, it has also been gaining increasing popularity
convenient to apply advanced signal processing techniques to
in the antenna and propagation community to solve complex
save hardware or time requirements. These techniques can be
electromagnetic problems [133-135]. For phased array cali-
distinguished as serial or parallel by the number of elements
bration, a graph coloring theory based method is proposed
to be calibrated each time, or distinguished as coherent or
in [136], which transforms the calibration problem into a
noncoherent by the measurement requirements between the
coloring problem that aims at minimizing the number of used
calibration source and the element under calibration.
colors. In [137], the number of calibration measurements is
minimized by using a compressed sensing (CS) approach.
Sparsity is introduced into the combined signals by assigning REFERENCES
binary delay vectors, which allows to recovery excitation [1] R. L. Haupt and Y. Rahmat-Samii, “Antenna Array Developments: A
Perspective on the Past, Present and Future,” IEEE Antennas Propag.
of individual array elements in a computationally effective Mag., vol. 57, no. 1, pp.86-96, Feb. 2015.
way. In [138], the ANN approach is applied to calibrate [2] C. A. Fowler, “Old radar types never die, they just phased array or 55
both transmit and receive channels for phased array antenna, years of trying to avoid mechanical scan,” IEEE Aerosp. Electron. Syst.
which allows simple hardware implementation and requires Mag., vol. 13, no. 9, pp. 24A-24L, Sep. 1998.
[3] A. J. Fenn, D. H. Temme, W. P. Delaney, and W. E. Courtney, “The
little computational power. ML techniques are undoubtedly Development of Phased Array Radar Technology,” Lincoln Laboratory
great analysis tools, particularly for problems with large and Journal, vol. 12, no. 2. pp. 321-340, 2000.
complex data sets. As ML techniques grow in sophistication [4] D. Parker and D. C. Zimmermann, “Phased Arrays, Part I: Theory and
Architectures,” IEEE Trans. Microw. Theory Techn., vol.53, no.2, pp. 678-
and are still in boosting, more powerful ML based calibration 687, Mar. 2002.
techniques and practical applications are expected for phased [5] D. Parker and D. C. Zimmermann, “Phased Arrays, Part II: implementa-
array calibration and status monitoring. tions, applications, and future trends,” IEEE Trans. Microw. Theory Techn.,
vol.53, no.2, pp. 688-698, Mar. 2002.
[6] A. S. Y. Poon and M. Taghivand, “Supporting and enabling circuits for
VIII. CONCLUSION antenna arrays in wireless communications,” Proc. IEEE, vol. 100, no. 7,
pp. 2207-2218, Jul. 2012.
Phased array antennas require precise excitation of each [7] E. Brookner, “Developments and breakthroughs in radars and phased-
antenna element by various synthesis techniques to obtain de- arrays,” in 2016 IEEE Radar Conf., Philadelphia, PA, USA, 2016, pp. 1-6,
sired array pattern features. However, because of mechanical doi: 10.1109/RADAR.2016.7485273.
[8] D. L. Jones, “Technology challenges for the Square Kilometer Array,”
manufacturing imperfections, component aging, temperature IEEE Aero. Electronic Systems Mag., vol. 28, no. 2, pp. 18-23, Feb. 2013.
variation, and circuit differences, the realistic antenna ele- [9] P. F. Li, S. W. Qiu, and S. W. Yang, “Two-Dimensional Imaging Based
ment excitation inevitably differs from their expected values on Near-Field Focused Array Antenna,” IEEE Antennas Wireless Propag.
in practice. These errors would cause array pattern distortion, Lett., vol. 18, no. 2, pp. 274-278, Feb. 2019.
[10] F. Roos, J. Bechter, C. Knill, B. Schweizer, and C. Waldschmidt, “Radar
gain degradation, worse SLL, and null drifts. This paper Sensors for Autonomous Driving: Modulation Schemes and Interference
aims to present a tutorial-like review of impact analysis and Mitigation,” IEEE Microw. Mag., vol. 20, no. 9, pp. 58-72, Sep. 2019.
state-of-the-art calibration methods to overcome excitation [11] B. Sadhu, X. Gu, and A. Valdes-Garcia, “The More (Antennas), the Mer-
rier: A Survey of Silicon-Based mm-Wave Phased Arrays Using Multi-IC
errors for phased array antenna. The probabilistic methods Scaling,” IEEE Microw. Mag., vol. 20, no. 12, pp. 32-50, Dec. 2019.
provide simple closed-form expressions of the features of the [12] R. S. Elliott, Antenna Theory and Design Hoboken, NJ: Wiley-
array pattern by exploiting the central limit theorem. The IA Interscience, IEEE, 2003.
based methods introduce intervals to represent the element [13] R. J. Mailloux, Phased Array Antenna Handbook, 3rd edition. Boston,
MA: Artech House, 2017.
excitation errors and predict the array performance with its [14] R. L. Haupt, Antenna Arrays: A Computational Approach, New York:
upper and lower bounds. Thanks to the inclusion property Wiley, 2010.

VOLUME 4, 2016 13

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

[15] C. Fulton, M. Yeary, D. Thompson, J. Lake, and A. Mitchell, “Digital [37] G. He, X. Gao, H. Zhou, and H. Zhu, “Comparison of three interval arith-
Phased Arrays: Challenges and Opportunities,” Proc. IEEE, vol. 104, no. metic based algorithms for antenna array pattern upper bound estimation,”
3, pp. 487-503, Mar. 2016. Electron. Lett., vol. 55, no. 14, pp. 775-776, Jul. 2019.
[16] J. S. Herd and M. D. Conway, “The Evolution to Modern Phased Array [38] G. He, X. Gao, and H. Zhou, “Matrix-based Interval Arithmetic for Linear
Architectures,” Proc. IEEE, vol. 104, no. 3, pp. 519-529, Mar. 2016. Array Tolerance Analysis With Excitation Amplitude Errors,” IEEE Trans.
[17] S. H. Talisa, K. W. O’Haver, T. M. Comberiate, M. D. Sharp, and O. F. Antennas Propag., vol.67, no.5, pp.3516-3520, May 2019.
Somerlock, “Benefits of Digital Phased Array Radars,” Proc. IEEE, vol. [39] M. de Berg, O. Cheong, M. van Kreveld, and M. Overmars, Computa-
104, no. 3, pp. 530-543, Mar. 2016. tional Geometry Algorithms and Applications, 3rd ed. Berlin, Germany:
[18] C. Fulton, R. Palmer, M. Yeary, J. Salazar, H. Sigmarsson, M. Weber, Springer-Verlag, 2008, ch. 13, p. 297.
and A. Hedden, “Horus: A testbed for fully digital phased array radars,” [40] P. Rocca, N. Anselmi, A. Benoni, and A. Massa, “Probabilistic Interval
Microw. J., vol. 63, no. 1, pp. 20-36, Jan. 2020. Analysis for the Analytic Prediction of the Pattern Tolerance Distribution
[19] M. Yeary, R. Palmer, C. Fulton, J. Salazar, and H. Sigmarsson, “Recent in Linear Phased Arrays With Random Excitation Errors,” IEEE Trans.
advances on an S-band all-digital mobile phased array radar,” in 2019 Antennas Propag., vol.68, no.12, pp.7866-7878, Dec. 2020.
IEEE Int. Sym. Phased Array Syst.Techn., Waltham, MA, USA, 2019, pp. [41] I. Seker, “Calibration methods for phased array radars,” Proc. SPIE, vol.
1-5, doi: 10.1109/PAST43306.2019.9020988. 8714, pp. 1-15, May 2013.
[20] M. Yeary, D. Conway, J. Herd, M. Fosberry, M. Harger, and K. Hondl, [42] C. Fulton, “Digital array radar calibration and performance monitoring
“A least mean squares approach of iterative array calibration for scalable techniques, including direct conversion and dual polarization architec-
digital phased array radar panels,” in 2013 IEEE Int. Sym. Phased Array tures,” Ph.D. dissertation, Purdue Univ., West Lafayette, IN, USA, 2011.
Syst. Techn., Waltham, MA, USA, 2013, pp. 276-278, doi: 10.1109/AR- [43] D. Hess, A. Fenn, D. Garneski, and C. Kryzak, “Phased array antenna
RAY.2013.6731841. measurements tutorial session,” Literature for tutorial session at 2010 the
[21] R. Lebron, J. Diaz, J. Salazar-Cerreno, “A Procedure to Characterize IEEE Int. Symp. on Phased Array Antennas Techn., , Waltham, MA, USA,
and Predict Active Phased Array Antenna Radiation Patterns from Planar Oct. 2010.
Near-Field Measurements,” in AMTA 2018 Proceedings, Williamsburg, [44] W. Patton and L. Yorinks, “Near-field alignment of phased-array anten-
VA, USA, 2018, pp. 1-4. nas,” IEEE Trans. Antennas Propag., vol. 47, no. 3, pp. 584-591, Mar.
[22] A. Dreher, N. Niklasch, F. Klefenz, and A. Schroth, “Antenna and receiver 1999.
system with digital beamforming for satellite navigation and communica- [45] K. Hassett,“Phased array antenna calibration measurement techniques and
tions,” IEEE Trans. Microw. Theory Tech., vol. 51, no. 7, pp. 1815-1821, methods,” in Proc. Eur. Conf. Antennas Propag., 2016, pp. 1-4.
Jul. 2003. [46] A. Stark, B. Rohrdantz, U. Johannsen, and A. F. Jacob, “In-situ probes
[23] M. Joler, “How FPGAs Can Help Create Self-Recoverable Antenna Ar- for patch antenna array calibration,” Int. J. Microw. Wireless Tech., vol. 3,
rays,” Int. J. Antennas Propag., vol. 2012, Article ID 196925, 10 pages, no.3, pp. 273-280, 2011.
2012. https://doi.org/10.1155/2012/196925 [47] Y. V. Korotetskiy, A. M. Shitikov, and V. V. Denisenko, “Phased Array An-
[24] J. Ruze, “The effect of aperture errors on the antenna radiation pattern,” tenna Calibration with Probe Positioning Errors,” IEEE Antennas Propag.
Nuovo Cimento, vol. 9, no. 3, pp. 364-380, 1952. Mag., vol. 58, no.3, pp. 65-80, Jun. 2016.
[48] J. Mulcahey and M. Sarcione, “Calibration and diagnostics of the THAAD
[25] J. K. Hsiao, “Design of error tolerance of a phased array,” Electron. Lett.,
solid state phased array in a planar near field facility,” in 1996 IEEE Int.
vol. 21, no. 19, pp. 834-836, Sep. 1985.
Symp. Phased Array Syst. Technol., Boston, MA, USA, 1996, pp. 322-326,
[26] J. Lee, Y. Lee, and H. Kim, “Decision of error tolerance in array element
doi: 10.1109/PAST.1996.566108.
by the Monte Carlo algorithm,” IEEE Trans. Antennas Propag., vol. 53,
[49] M. Sarcione, J. Mulcahey, D. Schmidt, K. Chang, M. Russell, P. Enzmann,
no. 4, pp. 1325-1331, Apr. 2005.
P. Rawlinson, W. Guzak, R. Howard, and M. Mitchell, “The design,
[27] A. J. van den Biggelaar, “Improved Statistical Model on the Effect of
development and testing of the THAAD (Theater High Altitude Area
Random Errors in the Phase and Amplitude of Element Excitations on
Defense) solid state phased array (formerly Ground Based Radar),” in
the Array Radiation Pattern,” IEEE Trans. Antennas Propag., vol. 66, no.
IEEE Int. Symp. Phased Array Syst.Technol., Boston, MA, USA, 1996,
5, pp. 2309-2317, May 2018.
pp. 260-265, doi: 10.1109/PAST.1996.566096.
[28] E. L. Holzman, “A different perspective on taper efficiency for array [50] A. Agrawal and A. Jablon, “A calibration technique for active phased array
antennas,” IEEE Trans. Antennas Propag., vol. 51, no. 10, pp. 2963-2967, antennas,” in Proc. IEEE Int. Symp. Phased Array Syst. Technol., Boston,
Oct. 2003. MA, USA, 2003, pp. 223-228, doi: 10.1109/PAST.2003.1256985.
[29] Arun K. Bhattacharyya, Phased Array Antennas, Floquet Analysis, Synthe- [51] J. Nicolas, “In situ array antenna diagnosis using microwave cir-
sis, BFNs and Active Array Systems, John Wiley and Sons, Inc., Hoboken, cular holography,” in Proc. IEEE Int. Symp. Phased Array Syst.
NJ, 2006. Technol., Waltham, MA, USA, 2010, pp. 298-305, doi: 10.1109/AR-
[30] P. Rocca, L. Manica, N. Anselmi, and A. Massa, “Analysis of the pattern RAY.2010.5613353.
tolerances in linear arrays with arbitrary amplitude errors,” IEEE Antennas [52] H. Pawlak, A. Charaspreedalarp, and A. F. Jacob, “Experimental investi-
Wireless Propag. Lett., vol. 12, pp. 639-642, Dec. 2013. gation of an external calibration scheme for 30 GHz circularly polarized
[31] N. Anselmi, L. Manica, P. Rocca, and A. Massa, “Tolerance analysis DBF transmit antenna arrays,” in Euro. Microw. Conf., Manchester, UK,
of antenna arrays through interval arithmetic,” IEEE Trans. Antennas 2006, pp.764-767, doi: 10.1109/EUMC.2006.281031.
Propag., vol. 61, no. 11, pp. 5496-5507, Nov. 2013. [53] C. Fulton, J. L. Salazar, Y. Zhang, et. al., “Cylindrical Polarimetric Phased
[32] N. Anselmi, P. Rocca, M. Salucci, and A. Massa, “Optimization of exci- Array Radar: Beamforming and Calibration for Weather Applications,”
tation tolerances for robust beam-forming in linear arrays,” IET Microw. IEEE Trans. Geoscience and Remote Sensing, vol. 55, no. 5, pp. 2827-
Antennas Propag., vol. 10, no. 2, pp. 208-214, Feb. 2016. 2841, May 2017.
[33] N. Anselmi, P. Rocca, M. Salucci, and A. Massa, “Power pattern sensitivity [54] B. Rousta, M. Naser-Moghadasi, B. A. Arand and R. A. Sadeghzadeh, “An
to calibration errors and mutual coupling in linear arrays through circular Efficient Scheduling for Fast Calibration of Digital Beamforming Active
interval arithmetics,” Sensors, vol. 16, no. 6, pp. 1-14, 2016. Phased Array Radar,” IETE J. Res., pp. 1-7, 2019.
[34] A. Massa, P. Rocca, E. Giaccari, and A. Farina, “Tolerance Anal- [55] M. Fernandez, Y. Lopez, and L. Fernando, “Dual-Probe Near-Field Phase-
ysis of Antenna Arrays Through Interval Analysis,” in 2014 IEEE less Antenna Measurement System on Board a UAV,” Sensors, vol. 19, no.
Radar Conf., Cincinnati, OH, USA, 2014, pp. 0896-0899, doi: 21, pp.1-18, 2019.
10.1109/RADAR.2014.6875718. [56] M. Fernandez, Y. Lopez, and L. Fernando, “Unmanned aerial system for
[35] N. Anselmi, M. Hannan, and P. Rocca, “Optimal Synthesis of Maximally antenna measurement and diagnosis: evaluation and testing,” IET Microw.
Robust Antenna Arrays by Means of Circular Interval Arithmetics,” in Antennas Propag., vol. 13 no. 13, pp. 2224-2231, 2019.
2019 IEEE Int. Symp. Antennas Propa. and USNC-URSI Radio Science [57] H. Pawlak and A. F. Jacob, “An external calibration scheme for DBF
Meeting, Atlanta, GA, USA, 2019, pp. 749-750, doi: 10.1109/APUS- antenna arrays,” IEEE Trans. Antennas Propag., vol. 58, no. 1, pp. 59-67,
NCURSINRSM.2019.8888575. Jan. 2010.
[36] L. Tenuti, N. Anselmi, P. Rocca, M. Salucci, and A. Massa, “Minkowski [58] T. Takahashi, N. Nakamoto, M. Ohtsuka, M. Ohtsuka, T. Aoki, Y. Konishi,
sum algorithm for planar arrays sensitivity analysis with uncertain-but- I. Chiba, M. Yajima, “On-Board Calibration Methods for Mechanical
bounded excitation tolerances,” IEEE Trans. Antennas Propag., vol. 65, Distortions of Satellite Phased Array Antennas,” IEEE Trans. Antennas
no. 1, pp. 167-177, Jan. 2017. Propag., vol . 60, no. 3, pp. 1362-1372, Mar. 2012.

14 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

[59] J. A. Haimerl, B. Hudson, G. P. Fonder, and D. K. Lee, “Overview of [79] A. Valdes-Garcia et al., “A Fully Integrated 16-Element Phased-Array
the large digital arrays of the space fence radar,” in IEEE Int. Symp. Transmitter in SiGe BiCMOS for 60-GHz Communications,” IEEE J.
Phased Array Syst. Technol., Waltham, MA, USA, 2016, pp. 1-8, doi: Solid-State Circuits, vol. 45, no. 12, pp. 2757-2773, Dec. 2010.
10.1109/ARRAY.2016.7832538. [80] A. Natarajan et al., “A Fully-Integrated 16-Element Phased-Array Receiv-
[60] M. Scott, “Sampson MFR active phased array antenna,” in IEEE Int. Symp. er in SiGe BiCMOS for 60-GHz Communications,” IEEE J. Solid-State
Phased Array Syst. Technol., Boston, MA, USA, 2003, pp. 119-123, doi: Circuits, vol. 46, no. 5, pp. 1059-1075, May 2011.
10.1109/PAST.2003.1256967. [81] O. Inac, D. Shin, and G. M. Rebeiz, “A phased array RFIC with built-in
[61] H. M. Aumann, A. J. Fenn, and F. G. Willwerth, “Phased array antenna self-test capabilities,” IEEE Trans. Microw. Theory Techn., vol. 60, no. 1,
calibration and pattern prediction using mutual coupling measurements,” pp. 139-148, Jan. 2012.
IEEE Trans. Antennas Propag., vol. 37, no. 7, pp. 844-850, Jul. 1989. [82] O. Inac, F. Golcuk, T. Kanar, and G. Rebeiz, “A 90-100 GHz phased-array
[62] G. F. Lewis and E. Boe, “Self-phase up of array antennas with non-uniform transmit/receive silicon RFIC module with built-in self-test,” IEEE Trans.
element mutual coupling and arbitrary lattice orientation,” US Patent US Microw. Theory Techn., vol. 61, no. 10, pp. 3774-3782, Oct. 2013.
5657023, Aug. 12, 1997. [83] S. Y. Kim, O. Inac, C. Y. Kim, D. Shin, and G. M. Rebeiz, “A 76-84
[63] C. Shipley and D. Woods, “Mutual coupling-based calibration of GHz 16-element phased-array receiver with a chip-level built-in self-test
phased array antennas,” in Proc. IEEE Int. Conf. Phased Array system,” IEEE Trans. Microw. Theory Techn., vol. 61, no. 8, pp. 3083-
Syst. Technol., Dana Point, CA, USA, 2000, pp. 529-532, doi: 3098, Aug. 2013.
10.1109/PAST.2000.859012. [84] B. Brautigam, M. Schwerdt, M. Bachmann, and M. Stangl, “Individual
[64] T. Gao, J. Wang, Y. Guo, and X. Chen, “Large phased array antenna T/R module characterisation of the terraSAR-X active phased array anten-
calibration using mutual coupling method,” in 2001 CIE Int. Conf. on na by calibration pulse sequences with orthogonal codes,” in Proc. IEEE
Radar, Beijing, China, 2001, pp. 223-226, doi: 10.1109/ICR.2001.984660. Int. Conf. Geosci. Remote Sens. Symp., Barcelona, Spain, 2007, pp. 5202-
[65] M. A. Salas Natera, R. Martinez Rodriguez-Osorio, L. de Haro Ariet and 5205, doi: 10.1109/IGARSS.2007.4424034.
M. Sierra Perez, “Calibration Proposal for New Antenna Array Archi- [85] M. Schwerdt, M. Bachmann, D. Schrank, B. Doring, B. Brautigam, J. H.
tectures and Technologies for Space Communications,” IEEE Antennas Gonzalez, and C. Schulz, “Precise calibration techniques for complex SAR
Wireless Propag. Lett., vol. 11, pp. 1129-1132, 2012, doi: 10.1109/LAW- systems based on active phased array antennas,” in Proc. IEEE Int. Symp.
P.2012.2215952. Phased Array Syst. Technol., Waltham, MA, USA, 2010, pp. 695-699, doi:
10.1109/ARRAY.2010.5613290.
[66] Y. Neidman, R. Shavit, and A. Bronshtein, “Diagnostic of phased arrays
with faulty elements using the mutual coupling method,” IET Microw. [86] D. J. Rabideau, R. J. Galejs, F. G. Willwerth, and D. S. McQueen,
Antennas Propag., vol. 3, no. 2, pp. 235-241, 2009. “An S-band digital array radar testbed,” in Proc. IEEE Int. Symp.
Phased Array Syst. Technol., Boston, MA, USA, 2003, pp. 113-118, doi:
[67] C. Fulton and W. Chappell, “Calibration techniques for digital phased ar-
10.1109/PAST.2003.1256966.
rays,” in IEEE Inter. Conf. Microw., Commu., Antennas and Electron. Syst.,
[87] S. C. Chae, H. W. Jo, J. I. Oh, G. Kim, and J. W. Yu, “Coupler Integrated
Tel Aviv, Israel, 2009, pp. 1-10, doi: 10.1109/COMCAS.2009.5385979.
Microstrip Patch Linear Phased Array for Self-Calibration,” IEEE Anten-
[68] F. S. Gerino and P. M. Belcaguy, “Mutual coupling internal calibration nas Wireless Propag. Lett., vol. 19, no.9, pp. 1615-1619, Sep. 2019.
method applied to a polarimetric phased array antenna model,” in 2016
[88] T. W. Nuteson, J. E. Stocker, J. S. Clark, D. S. Haque, and G. S. Mitchell,
IEEE Global Electromag. Compatibility Conf., Mar del Plata, Argentina,
“Performance characterization of FPGA techniques for calibration and
2016, pp. 1-5, doi: 10.1109/GEMCCON.2016.7797310.
beamforming in smart antenna applications,” IEEE Trans. Microw. Theory
[69] D. Bekers, R. van Dijk, and F. van Vliet, “Mutual-coupling based phased- Tech., vol. 50, no. 12, pp. 3043-3051, Dec. 2002.
array calibration: A robust and versatile approach,” in Proc. IEEE Int. [89] G. Liang, W. Gong, H. Liu, and J. Yu, “61-Channel Digital Beamforming
Symp. Phased Array Syst. Technol., Waltham, MA, USA, 2013, pp. 630- (DBF) Transmitter Array,” in IEEE 2009 Asia Pacific Microw. Conf.,
637, doi: 10.1109/ARRAY.2013.6731903. Singapore, Nov. 2009, pp. 739-743.
[70] R. M. Lebron, P. S. Tsai, J. M. Emmett, C. Fulton, J. L. Salazar-Cerreno, [90] G. Liang, W. Gong, H. Liu, and J. Yu, “Development of 61-Channel Digital
“Validation and Testing of Initial and In-situ Mutual Coupling-Based Beamforming (DBF) Transmitter Array for Mobile Satellite Communica-
Calibration of a Dual-polarized Active Phased Array Antenna,” IEEE tion,” Prog. Electromagn. Res., vol. 97, pp. 177-195, 2009.
Access, vol.8, pp. 78315-78329, 2020. [91] A. M. Shitikov and A. V. Bondarik, “Multi-element PAA calibration with
[71] J. Lujan, C. J. Fulton, M. Yeary, E. A. Langley, S. McCormick, and A. REV method,” in IEEE Int. Conf. Antenna Theory Tech., Sevastopol,
Hedden, “Phased array radar initial alignment algorithm using mutual Ukraine, Sep. 9-12, 2003, pp. 761-764.
coupling: an iterative approach,” in Proc. SPIE 11408, Radar Sensor [92] S. Mano and T. Katagi, “A method for measuring amplitude and phase
Technology XXIV, no. 1140811, Apr. 2020. of each radiating element of a phased array antenna,” Electron. Commun.
[72] J. Herd, “Experimental results from a self-calibrating digital beamforming Jpn., vol. 65, no. 5, pp. 58-64, 1982.
array,” in Proc. Antennas Propag. Soc. Int. Symp. Dig., Dallas, TX, USA, [93] T. Takahashi, H. Miyashita, Y. Konishi, and S. Makino, “Theoretical study
1990, pp. 384-387 vol.1, doi: 10.1109/APS.1990.115127. on measurement accuracy of rotating element electric field,” Electron.
[73] A. Dreher, N. Niklasch, F. Klefenz, and A. Schroth, “Antenna and receiver Commun. Jpn., vol. 90, no. 1, pp. 22-33, 2006.
system with digital beamforming for satellite navigation and communica- [94] T. Takahashi, Y. Konishi, and I. Chiba, “A Novel Amplitude-Only Mea-
tions,” IEEE Trans. Microw. Theory Tech., vol. 51, no. 7, pp. 1815-1821, surement Method to Determine Element Fields in Phased Arrays,” IEEE
Jul. 2003. Trans. Antennas Propag., vol. 60, no. 7, pp. 3222-3230, Jul. 2012.
[74] O. Inac, S. Kim, D. Shin, C. Kim, and G. Rebeiz, “Built-in self [95] R. Sorace, “Phased array calibration,” IEEE Trans. Antennas Propag., vol.
test systems for silicon-based phased arrays,” in IEEE MTT-S In- 49, no. 4, pp. 517-525, Apr. 2001.
t. Microw. Symp. Dig., Montreal, QC, Canada, 2012, pp. 1-3, doi: [96] R. Long, J. Ouyang, F. Yang, W. Han, and L. Zhou, “Fast Amplitude-
10.1109/MWSYM.2012.6259455. Only Measurement Method for Phased Array Calibration,” IEEE Trans.
[75] D. Kissinger, B. Laemmle, L. Maurer, and R. Weigel, “Integrated test for Antennas Propag., vol. 65, no.4 , pp. 1815-1822, Apr. 2017.
silicon front ends,” IEEE Microw. Mag., vol. 11, no. 3, pp. 87-94, May [97] G. He, X. Gao, and H. Zhou, “Fast Phased Array Calibration by Power-
2010. Only Measurements Twice for Each Antenna Element,” Int. J. Antennas
[76] D. Kissinger, R. Agethen, and R. Weigel, “A versatile built-in test archi- Propag., vol. 2019, Article ID 6432149, 10 pages, 2019.
tecture for integrated millimeter-wave radar receiver front-ends,” in Proc. [98] S. B. Sleiman, A. Akour, W. Khalil, and M. Ismail, “Millimeter-wave BiST
IEEE Int. Instrum. Meas. Technol. Conf., Graz, Austria, 2012, pp. 254-258, and BiSC using a high-definition sub-ranged detector in 90 nm CMOS,”
doi: 10.1109/I2MTC.2012.6229244. in IEEE Int. Midwest Symp. Circuits Syst., Seattle, WA, USA, 2010, pp.
[77] X. Gu, A. Valdes-Garcia, A. Natarajan, B. Sadhu, D. Liu and S. K. 477-480, doi: 10.1109/MWSCAS.2010.5548737.
Reynolds, “W-band scalable phased arrays for imaging and communica- [99] Y. Takeda, T. Fujibayashi, Y. S. Yeh, W. Wang, and B. Floyd, “A 76- to 81-
tions,” IEEE Communications Mag., vol. 53, no. 4, pp. 196-204, Apr. 2015. GHz transceiver chipset for long-range and short-range automotive radar,”
[78] K.M. Lee, R.S. Chu, and S.C. Liu, “A built-in performance monitor- in Proc. IEEE MTT-S Int. Microw. Symp., Tampa, FL, USA, 2014, pp. 1-3,
ing/fault isolation and correction (PM/FIC) system for active phased-array doi: 10.1109/MWSYM.2014.6848490.
antennas,” IEEE Trans. Antennas Propag., vol. 41, no. 11, pp. 1530-1540, [100] E. Cohen, M. Ruberto, M. Cohen, O. Degani, S. Ravid, and D. Ritter, “A
Nov. 1993. CMOS bidirectional 32-element phased-array transceiver at 60 GHz with

VOLUME 4, 2016 15

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

LTCC antenna,” IEEE Trans. Microw. Theory Techn., vol. 61, no. 3, pp. [122] T. H. Ismail and Z. M. Hamici, “Array pattern synthesis using digital
1359-1375, Mar. 2013. phase control by quantized particle swarm optimization,” IEEE Trans.
[101] R. Long, J. Ouyang, F. Yang, Y. Li, K. Zhang, and L. Zhou, “Calibration Antennas Propag., vol. 58, no. 6, pp. 2142-2145, Jun. 2010.
method of phased array based on near-field measurement system,” in Proc. [123] W. P. M. N. Keizer, “Low sidelobe phased array pattern synthesis with
IEEE Antennas Propag. Soc. Int. Symp. (APSURSI), Memphis, TN, USA, compensation for errors due to quantized tapering,” IEEE Trans. Antennas
2014, pp. 1161-1162, doi: 10.1109/APS.2014.6904907. Propag., vol. 59, no. 12, pp. 4520-4524, Dec. 2011.
[102] S. D. Silverstein, “Application of orthogonal codes to the calibration of [124] M. S. Smith and Y. C. Guo, “A Comparison of Methods for Randomizing
active phased array antennas for communication satellites,” IEEE Trans. Phase Quantization Errors in Phased Arrays,” IEEE Trans. Antennas
Signal Process., vol. 45, no. 1, pp. 206-218, Jan. 1997. Propag., vol. 31, no. 6, pp. 821-827, Nov. 1983.
[103] E. Lier, M. Zemlyansky, D. Purdy, and D. Farina, “Phased array cal- [125] R. L. Haupt, “Antenna Arrays in the Time Domain,” IEEE An-
ibration and characterization based on orthogonal coding: Theory and tennas Propag. Mag., vol. 59, no. 3, pp. 33-41, June 2017. DOI:
experimental validation,” in Proc. IEEE Int. Symp. Phased Array Syst. 10.1109/MAP.2017.2686082.
Technol., Waltham, MA, USA, 2010, pp. 271-278, doi: 10.1109/AR- [126] R. L. Haupt and P. Nayeri, “Pulse Dispersion in Phased Arrays,” Int.
RAY.2010.5613357. J. Antennas Propag., Volume 2017, Article ID 5717641, 8 pages, 2017.
[104] Y. Lin, X. Bu, W. Zhao, and S. Wang, “Parallel Calibration Method for https://doi.org/10.1155/2017/5717641
Phased Array Antennas with Orthogonal and Nonorthogonal Codes,” in [127] J. B. Thomas, “The Tone Generator and Phase Calibration in VLBI
IEEE Int. Conf. Communications and Networking in China, Shanghai, Measurements,” The Deep Space Network Progress Report 42-44, Jet
China, 2015, pp. 883-886, doi: 10.1109/CHINACOM.2015.7498062. Propulsion Laboratory, Pasadena, California, pp. 63-74, April 15, 1978.
[105] Y. Lin, S. Wang, X. Bu, C. Xing, and J, An, “NOMA-Based Calibration [128] C. S. Jacobs, “Phase Calibration Tone Processing With the Block II VLBI
for Large-Scale Spaceborne Antenna Arrays,” IEEE Trans. Vehicular Correlator,” TMO Progress Report 42-134, Jet Propulsion Laboratory,
Tech., vol. 67, no.3, pp. 2231-2242, Mar. 2018. Pasadena, California, pp. 1-22, August 15, 1998.
[106] G. H. C. van Werkhoven and A. K. Golshayan, “Calibration aspects of the [129] R. Hamell, B. Tucker, and M. Calhoun, “Phase Calibration Generator,”
APAR antenna unit,” in IEEE Int. Conf. Phased Array Syst. Technol., Dana IPN Progress Report 42-154, Jet Propulsion Laboratory, Pasadena, Cali-
Point, CA, USA, 2000, pp. 425-428, doi: 10.1109/PAST.2000.858989. fornia, pp. 1-14, August 15, 2003.
[107] M. Bachmann, M. Schwerdt, and B. Brautigam, “TerraSAR-X Antenna [130] L. L. Liou, D. M. Lin, M. Longbrake, P. Buxa, J. McCann, T. Dalrymple,
Calibration and Monitoring Based on a Precise Antenna Model,” IEEE J. B. Tsui, R. Qiu, Z. Hu and N. Guo, “Digital Wideband Phased Array
Trans. Geoscience and Remote Sens., vol. 48, no. 2, pp. 690-701, Feb. Calibration and Beamforming Using Time Reversal Technique,” in 2010
2010. IEEE Int. Sym. Phased Array Syst. Techn., Waltham, MA, USA, 2010, pp.
[108] M. Schwerdt, K. Schmidt, N. T. Ramon, P. Klenk, N. Yague-Martinez, P. 261-266, doi: 10.1109/ARRAY.2010.5613359.
Prats-Iraola, M. Zink, and D. Geudtner “Independent System Calibration [131] A. Massa, G. Oliveri, M. Salucci, N. Anselmi, and P. Rocca, “Learning-
of Sentinel-1B,” Remote Sensing, vol. 9, no. 511, pp. 1-34, 2017. by-examples techniques as applied to electromagnetics,” J. Electro-
[109] R. Long, J. Ouyang, F. Yang, W. Han, and L. Zhou, “Multi-Element mag. Waves and Appl., vol. 32, no. 4, pp. 516-541, 2017. DOI:
Phased Array Calibration Method by Solving Linear Equations,” IEEE 10.1080/09205071.2017.1402713
Trans. Antennas Propag., vol. 65, no.4, pp. 2931-2939, Jun. 2017. [132] L. Li, L. Wang, F. L. Teixeira, C. Liu, A. Nehorai, and T. Cui, “DeepNIS:
[110] G. A. Hampson and A. B. Smolders, “A fast and accurate scheme for Deep Neural Network for Nonlinear Electromagnetic Inverse Scattering,”
calibration of active phased-array antennas,” in Proc. IEEE Antennas IEEE Trans. Antennas Propag., vol. 67, no. 3, pp. 1819-1825, Mar. 2019.
Propag. Soc. Int. Symp., Orlando, FL, USA, 1999, pp. 1040-1043, doi: [133] A. Massa, D. Marcantonio, X. Chen, M. Li, and M. Salucci, “DNNs
10.1109/APS.1999.789490. as Applied to Electromagnetics, Antennas, and Propagation-A Review,”
[111] A. B. Smolders and G. A. Hampson, “Deterministic RF nulling in phased IEEE Antennas Wireless Propag. Lett., vol. 18, no. 11, pp. 2225-2229,
arrays for the next generation of radio telescopes,” IEEE Antennas Propag. Nov. 2019.
Mag., vol. 44, no. 4, pp. 13-22, Aug. 2002. [134] S. D. Campbell, R. P. Jenkins, P. J. O’Connor, and D. H. Werner, “The
[112] T. Takahashi, Y. Konishi, S. Makino, H. Ohmine, and H. Nakaguro, Explosion of Artificial Intelligence in Antennas and Propagation,” IEEE
“Fast measurement technique for phased array calibration,” IEEE Trans. Antennas Propag. Mag., to appear. DOI: 10.1109/MAP.2020.3021433.
Antennas Propag., vol. 56, no. 7, pp. 1888-1899, Jul. 2008. [135] F. Zardi, P. Nayeri, P. Rocca, and R. L. Haupt, “Artificial Intelligence for
[113] C. He, X. Liang, J. Geng, and R. Jin, “Parallel calibration method for Adaptive and Reconfigurable Antenna Arrays,” IEEE Antennas Propag.
phased array with harmonic characteristic analysis,” IEEE Trans. Antennas Mag., to appear. DOI: 10.1109/MAP.2020.3036097.
Propag., vol. 62, no. 10, pp. 5029-5036, Oct. 2014. [136] L. Yang, R. Dang, M. Li, K. Zhao, C. Song, and Z. Xu, “A Fast Cali-
[114] K. Greene, V. Chauhan, and B. Floyd, “Built-In Test of Phased Arrays bration Method for Phased Arrays by Using the Graph Coloring Theory,”
Using Code-Modulated Interferometry,” IEEE Trans. Microw. Theory Sensors, vol. 18, no. 12, pp. 1-19, 2018. https://doi.org/10.3390/s18124315
Techn., vol.66, no.5, pp. 2463-2479, May 2018. [137] G. Baburs, D. Caratelli, and A. Mirmanov, “Phased Array Calibration
[115] K. Greene, V. Chauhan, and B. Floyd, “Code-modulated embedded test by Binary Compressed Sensing,” Prog. Electromagn. Res. M, vol. 73, pp.
for phased arrays,” in 2016 IEEE 34th VLSI Test Symposium (VTS), Las 61-70, 2018.
Vegas, NV, 2016, pp. 1-4. [138] C. Shan, X. Chen, H. Yin, W. Wang, G. Wei, and Y. Zhang, “Diagnosis
[116] H. S. C. Wang, “Performance of Phased-Array Antennas with Mechani- of calibration state for massive antenna array via deep learning,” IEEE
cal Errors,” IEEE Trans. Aero. Electronic Syst., vol. 28, no. 2, pp. 535-545, Wireless Commun. Lett., vol. 8, no. 5, pp. 1431-1434, Oct. 2019. doi:
1992. 10.1109/LWC.2019.2920818.
[117] C. Wang, M. Kang, W. Wang, J. Zhong, Y. Zhang, C. Jiang, and B. Duan,
“Electromechanical coupling based performance evaluation of distorted
phased array antennas with random position errors,” Int. J. Applied Elec-
tromag. Mechanics, vol. 51, no. 3, pp. 285-295, 2016.
[118] C. Wang, M. Kang, W. Wang, B. Duan, L. Lin, and L. Ping, “On the per-
formance of array antennas with mechanical distortion errors considering
element numbers,” Int. J. Electronics, vol. 104, no. 3, pp. 462-484, 2017.
[119] C. Wang, Y. Wang, X. Yang, W. Gao, C. Jiang, L. Wang, Y. Zhang, and
M. Wang, “Effect of Randomness in Element Position on Performance of
Communication Array Antennas in Internet of Things,” Wirel. Commu.
Mobile Comp.,Volume 2018, Article ID 6492143, 8 pages.
[120] F. Guo, Z. Liu, G. Sa, and J. Tan, “A Position Error Representation
Method for Planar Arrays,” IEEE Antennas Wireless Propag. Lett., vol.
19, no. 1, pp. 109-113, Jan. 2020.
[121] C. Wang, S. Yuan, W. Gao, C. Jiang, C. Zhu, P. Li, Z. Wang, X. Peng,
and Y. Shi, “A Taylor-Surrogate-Model-Based Method for the Electrical
Performance of Array Antennas Under Interval Position Errors,” IEEE
Antennas Wireless Propag. Lett., vol. 19, no. 7, pp. 1221-1225, Jul. 2020.

16 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2021.3073222, IEEE Access

He et al.: Impact Analysis and Calibration Methods of Excitation Errors for Phased Array Antennas

GUOLONG HE received the B.S. degree in elec-

tronic engineering from Shanghai Jiao Tong Uni-
versity, Shanghai, China, in 2008, and the Ph.D.
degree in aerospace engineering from Tsinghua
University, Beijing, China, in 2016. Since 2008,
he has been with the Beijing Institute of Tracking
and Telecommunication Technology, Beijing, Chi-
na, and involved in the design, construction, and
maintenance of the Chinese s. He is in charge of
several projects with both parabolic antennas and
phased array antennas. Currently, he is the Project Manager of 4×35-m
deep space antenna array in the Kashi Deep Space Station. His research
interests include design of satellite communication and navigation systems,
and phased array signal processing.

XIN GAO received the B.S. degree in electron-

ic engineering from Northwest University, Xi’an,
Shaanxi, China, in 1995, the M.S. degree in op-
tics engineering from Northwestern Polytechnical
University, Xi’an, Shaanxi, China, in 1998, and
the Ph.D. degree in optics engineering from X-
i’an Institute of Optics and Precision Mechanics,
Chinese Academy of Sciences, Xi’an, Shaanxi,
China, in 2007. Now he is a Full Professor in Bei-
jing Institute of Tracking and Telecommunication
Technology, Beijing, China. His research interests include design of satellite
communication and navigation systems, and phased array signal processing.

RENTIAN ZHANG received the B.S. degree in

electronic engineering from Huazhong University
of Science and Technology, Wuhan, China, in
2009, and the M.S. degree in electronic engi-
neering from Beihang University, Beijing, China,
in 2011. Now he is an Assistant Professor in
Beijing Institute of Tracking and Telecommuni-
cation Technology, Beijing, China. His research
interests include design of satellite communication
and navigation systems, and phased array signal
processing.

VOLUME 4, 2016 17

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/