NUS/ECE EE6832

Antenna Arrays
1 Introduction
Antenna arrays are becoming increasingly important in
wireless communications. Advantages of using antenna
arrays:
1. They can provide the capability of a steerable beam
(radiation direction change) as in smart antennas.
2. They can provide a high gain (array gain) by using
simple antenna elements.
3. They provide a diversity gain in multipath signal
reception.
4. They enable array signal processing.
Hon Tat Hui
1 Antenna Arrays
NUS/ECE EE6832

2 Far-Field Expression of An Antenna Array

Observation
r1 point P(x,y,z) at
z r2 infinity
r0
ρ2
ρ1 rN
ri ρ1, ρ2,…, ρi,…, ρN
= position vectors of the
y antenna elements
r1, r2,…, ri,…, rN
= distances of the antenna
x ρi
elements from the
ρN observation point
r0 = distance of the origin from
An arbitrary antenna array the observation point
Hon Tat Hui
2 Antenna Arrays
NUS/ECE EE6832

The sum of the far fields E radiated from the array is:
N
E = ∑ Ei (1)
i =1

where Ei is the far field of the ith antena given by:

Ei = ⎡⎣aˆ θ fθ i (θ ,φ ) + aˆ φ fφi (θ ,φ ) ⎤⎦ wi K i e − jkri (2)
fθ i (θ ,φ ) = the θ component of the radiation pattern
fφi (θ ,φ ) = the φ component of the radiation pattern
wi = the weighting factor of the excitation source
K i = a constant accounting for the path loss
Note that fθ i (θ ,φ ) and fφi (θ ,φ ) are obtained with the ith
(for all i ) antenna element located at the origin.
Hon Tat Hui
3 Antenna Arrays
NUS/ECE EE6832

For identical antenna elements, the element radiation

patterns are the same and independent of i. Hence,
N
E = ⎡⎣aˆ θ fθ (θ ,φ ) + aˆ φ fφ (θ ,φ ) ⎤⎦ ∑ wi K i e − jkri
(3)
i =1

Furthermore, we assume that Ki is approximately constant

such that
K1 = K 2 = = KN = K (4)

We can also write

ri = r0 − Δr = r0 − ρi ⋅ aˆ r (θ ,φ ) (5)

Hon Tat Hui

4 Antenna Arrays
NUS/ECE EE6832

Then
N
⎡⎣aˆ θ fθ (θ ,φ ) + aˆ φ fφ (θ ,φ ) ⎤⎦ ∑ wi e
− jkr0 jkρi ⋅aˆ r (θ ,φ )
E = Ke
i =1

= Ke − jkr0 ⎣⎡aˆ θ fθ (θ ,φ ) + aˆ φ fφ (θ ,φ ) ⎤⎦ f array (θ ,φ ) (6)

The result in (6) is known as the Principle of Pattern

Multiplication, which states that the array pattern is the
product of the antenna element pattern multiplied with an
array factor f array (θ ,φ ) .

Hereafter, we focus on the study of the array factor.

Hon Tat Hui

5 Antenna Arrays
NUS/ECE EE6832

N
f array (θ ,φ ) = ∑ wi e
jkρi ⋅aˆ r (θ ,φ )

i =1
N
= ∑ wi e jbi (7)
i =1

Putting in the matrix form, we have

f array (θ ,φ ) = w T b (8)
jkρ1⋅aˆ r (θ ,φ )
⎡ w1 ⎤ ⎡e ⎤ ⎡
jb1 e ⎤
⎢w ⎥ ⎢ jb2 ⎥ ⎢ jkρ2 ⋅aˆ r (θ ,φ ) ⎥
⎢ e ⎥ ⎢e ⎥
w=⎢ 2⎥ b= =⎢ ⎥ (9)
⎢ ⎥ ⎢ ⎥
⎢w ⎥ ⎢ jbN ⎥ ⎢ jkρ N ⋅aˆ (θ ,φ ) ⎥
⎣ N⎦ ⎣e ⎦ ⎢⎣e r ⎥⎦
Hon Tat Hui
6 Antenna Arrays
NUS/ECE EE6832

Eq. (8) is an important expression which tells us that the

array radiation pattern (array factor) can be obtained from
a given array element weight vector w and vice versa. If
we consider the weight vector as an N dimensional
function w(xi,yi,zi), i =1, 2, …, N, then the array factor
and the weight function is related by a certain type of
transformation operation ℑ. That is,
f array (θ ,φ ) = ℑ ( w ( x, y, z ) ) (10)

Similarly, the weight function w(x,y,z) can be determined

from a given required array factor function.
w ( x, y, z ) = ℑ−1 ( f array (θ ,φ ) ) (11)
Hon Tat Hui
7 Antenna Arrays
NUS/ECE EE6832

3 Uniform Liner Arrays (ULAs)

Dipoles are Far field

parallel to the z observation
direction point
y
r
rN-1
φ
x
Dipole 1 Dipole N
d

An N-element uniform antenna array with an element

separation d and at the plane of θ = π/2

Hon Tat Hui

8 Antenna Arrays
NUS/ECE EE6832

Uniform linear arrays (ULAs) mean that the array

elements are same as each other and they are aligned along
a straight line with equal element separations. If the
excitation currents have the same amplitude (I = 1) but the
phase difference between adjacent elements is β (the
progressive phase difference), then the weight vector is:

⎡ w1 ⎤ ⎡ I ⎤
⎢ w ⎥ ⎢ Ie jβ ⎥
⎢ 2⎥ ⎢ j 2β
⎥
w = ⎢ w3 ⎥ = ⎢ Ie ⎥ (12)
⎢ ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ wN ⎥⎦ ⎢⎣ Ie j( N −1)β ⎥⎦
Hon Tat Hui
9 Antenna Arrays
NUS/ECE EE6832

The array factor (AF) for this array specified on the plane
θ = π/2 is:
N N
AF = f array (θ = π 2,φ ) = ∑ wi e jbi
= I ∑e j ( i −1) β
e j ( i −1) d cos φ

i =1 i =1

= 1 + e j ( kd cosφ + β ) + e j 2( kd cosφ + β ) + + e j ( N −1)( kd cosφ + β )

⎛ ψ⎞
sin ⎜ N ⎟ ψ
N
⎝ ⎠e ( − )
= ∑e 2 j N 1
j ( n −1)ψ
= 2 (13)
n =1 ⎛ ψ⎞
sin ⎜ ⎟
⎝2⎠
where ψ = kd cos φ + β and 0 ≤ φ , β ≤ 2π (14)

Hon Tat Hui

10 Antenna Arrays
NUS/ECE EE6832

The normalized array factor is:

where Γ is a constant to
⎛ ψ⎞
sin ⎜ N ⎟ make the largest value of
AFn (ψ ) =
1 ⎝ 2⎠ |AFn| equal to one. Note
(15)
that Γ is not necessarily
Γ sin ⎛ψ ⎞
⎜ ⎟ equal to N.
⎝2⎠
The relation between |AFn|, Ψ, d, and β is shown graphically
on next page. Note that |AFn| is a period function of Ψ,
which is in turn a function of φ. The angle φ is in the real
space and its range is 0 to 2π. However, Ψ is not in the real
space and its range can be greater than or smaller than 0 to
2π, leading to the problem of grating lobes or not achieving
the maximum values of the |AFn| expression.
Hon Tat Hui
11 Antenna Arrays
NUS/ECE EE6832

|AFn(Ψ )|

Ψ = kd cosφ +β

kd
φ
β
kdcosφ

The relation between AFn,Ψ, d, and β

Hon Tat Hui
12 Antenna Arrays
 NUS/ECE EE6832

3.1 Properties of the normalized array factor AFn:

1. |AFn| is a periodic function of ψ, with a period of 2π.
This is because
|AFn(ψ + 2π)| = |AFn(ψ)|. (16)
2. As cos(φ) = cos(-φ), |AFn| is symmetric about the line of
the array, i.e., φ = 0 & π. Hence it is enough to know
|AFn| for 0 ≤ φ ≤ π.
3. The maximum values of |AFn| occur when (see
Supplementary Notes):
ψ 1
= (kd cos φ + β ) = ± mπ , m = 0,1,2,
2 2
−1 ⎡ λ
⇒ φmax = main beam directions = cos ⎢ (− β ± 2mπ ) ⎤⎥ (17)
⎣ 2π d ⎦
Hon Tat Hui
13 Antenna Arrays
NUS/ECE EE6832

Note that there may be more than one angles φmax

corresponding to the same value of m because cos-1(x) is
a multi-value function. If there are more than one
maximum angles φmax, the second and the subsequent
maximum angles give rise to the phenomenon of grating
lobes. The condition for grating lobes to occur is that d ≥
λ (disregarding the value of β) as shown below:
2nd grating lobe 1st grating lobe Main lobe 1st grating lobe 2nd grating lobe

Hon Tat Hui

14 Antenna Arrays
 NUS/ECE EE6832

2nd grating lobe 1st grating lobe Main lobe 1st grating lobe 2nd grating lobe
|AFn(Ψ )|

Visible region
(1) When d ≤ 0.5λ, no General conditions to
grating lobes can be kd avoid grating lobes
formed for whatever with β [0,2π] and d
value of β. (2) When d φ
≥ λ, grating lobe(s) is [0.5λ,λ]:
(are) formed for ψ=kdcosφ 1.For 0 ≤ β < π, the
whatever value of β. requirement is:
(3) When 0.5λ <d< λ, kd + β ≤ 2π
formation of grating 2. For π ≤ β < 2π, the
lobes depends on β. requirement is:
kd - β ≤ 0
Hon Tat Hui
15 Antenna Arrays
NUS/ECE EE6832

4. There are other angles corresponding to the maximum

values for the minor lobes (minor beams) but these angles
cannot be found from the formula in no. 3 above.
5. When β and d are fixed, it is possible that ψ can never be
equal to 2mπ. In that case, the maximum values of |AFn|
cannot be determined by the formula in no. 3.
6. The main beam directions φmax are not related to N. They
are functions of β and d only.
7. The nulls of |AFn| occur when:
ψ nπ ⎧n = 1,2,3,
=± , ⎨
2 N ⎩n ≠ N ,2 N ,3 N ,
−1 ⎡ λ 2n ⎤
⇒ φnull = null directions = cos ⎢ (− β ± π ) ⎥ (18)
⎣ 2π d N ⎦
Hon Tat Hui
16 Antenna Arrays
NUS/ECE EE6832

Note that there may be more than one angles φnull

corresponding to a single value of n because cos-1(x) is a
multi-value function.
8. The null directions φnull are dependent on N.
9. The larger the number N, the closer is the first null (n = 1)
to the first maximum (m = 0). This means a narrower
main beam and an increase in the directivity or gain of the
array.
10.The angle for the main beam direction (m = 0) can be
controlled by varying β or d.

Hon Tat Hui

17 Antenna Arrays
NUS/ECE EE6832

Example 1
A uniform linear array consists of 10 half-wave dipoles with
an inter-element separation d = λ/4 and equal current
amplitude. Find the excitation current phase difference β
such that the main beam direction is at 60° (φmax = 60°).
Solutions d = λ/4, θmax = 60°, N = 10
⎡ λ ⎤
main beam dirction = φmax −1
= 60° = cos ⎢ − ( β ± m 2π )⎥
⎣ 2π d ⎦
2
⇒ − ( β ± m2π ) = cos ( 60° ) = 0.5
π
π
⇒ β = − ∓ m2π = −45° + 360° = 315°, when m = 1
4
Hon Tat Hui
18 Antenna Arrays
 NUS/ECE EE6832

Other values of β corresponding to other values of m are

outside the range of 0 ≤ β ≤ 2π and are not included.

⎡ ⎛π ⎞ ⎤
sin ⎢5 ⎜ cos φ + β ⎟ ⎥
AFn =
1 ⎣ ⎝ 2 ⎠⎦
Γ ⎡1 ⎛ π ⎞ ⎤
sin ⎢ ⎜ cos φ + β ⎟ ⎥
⎣2⎝ 2 ⎠⎦
Γ = 10

Hon Tat Hui

19 Antenna Arrays
NUS/ECE EE6832

3.2 Phased (Scanning) Arrays

It was mentioned earlier that by controlling the values of d
or β, the maximum radiation direction of an array can be
arbitrarily pointed to any direction. In practice, the
element separation d is usually fixed while the excitation
current phase β between elements is controlled
electronically. The current amplitudes of the all the
elements are assumed to be the same. This kind of
steerable direction arrays is called uniform phased
scanning arrays. To accomplish this, the excitation
current phase β must be adjusted so that:
ψ = kd cos φ + β φ =φ = 0,
0
⇒ β = − kd cos φ0 (19)
Hon Tat Hui
20 Antenna Arrays
NUS/ECE EE6832

For a phased scanning array, the length L = (N-1)d of the

array (where N is the number of elements) can be
determined from the graph on next page. The graph
shows the relation between the array length, the half-
power beamwidth, and the maximum radiation direction.
The half-power beamwidth is an alternative way to
specify the gain of the array. Once the half-power
beamwidth and the maximum radiation direction are
specified, the number
y of elements required to design such
an array can be calculated.
L

Ant. 1 Ant. N
d
Hon Tat Hui
21 Antenna Arrays
NUS/ECE EE6832

Maximum radiation direction

Hon Tat Hui
22 Antenna Arrays
NUS/ECE EE6832

Example 2
Design a uniform linear phased scanning array whose
maximum radiation direction is in 30° (θ = 30°). The
desired half-power beamwidth is 2° while the element
separation is d = λ/4. Determine the excitation current phase
β, the length of the array L, and the number of elements N in
the array.
Solutions
Since the array is uniform, the current amplitude is same for
all elements. The excitation current phase β is found from:
2π λ
β = − kd cosθ 0 = − cos30 = −1.36 rad = −77.94
o o

λ 4
Hon Tat Hui
23 Antenna Arrays
NUS/ECE EE6832

To find the length L of the array, we use the graph on page

24. From that graph with HPBW=2° and maximum
radiation direction θ =30°:
(L+d)/λ=50

Therefore with d = λ/4,

L = 49.75λ

The number of elements is then:

L 49.75
N = +1 = + 1 = 200
d 0.25
Hon Tat Hui
24 Antenna Arrays
NUS/ECE EE6832

4 Circular Arrays

Δrn = a cosψ n
= a sin θ cos (φ − φn )

Hon Tat Hui

25 Antenna Arrays
NUS/ECE EE6832

4.1 Advantages of Circular Arrays

1. Unlike linear arrays, circular arrays can provide a 2D

angular scan, both horizontal φ and vertical θ scans.
2. Unlike 2D planar arrays, circular arrays are basically
1D linear arrays but in a circular form.
3. Unlike linear arrays, a circular array can scan
horizontally for 360° with no distortions near the
end-fire directions.
4. Unlike linear arrays, distortions in the array pattern
of a circular array due to mutual coupling effect are
same for each element and this makes it easier to
deal with the mutual coupling effect.
Hon Tat Hui
26 Antenna Arrays
NUS/ECE EE6832

4.2 Array Factor

For a uniform circular array with N elements and an
equal excitation current amplitude I0 and a current
phase of βn (reference to the central point of the array)
for the nth element (and φn=2πn/N), the array factor is:

j[ ka sin θ cos(φ −φ1 )+ β1 ] j[ ka sin θ cos(φ −φ2 )+ β 2 ]

AF = e +e
j[ ka sin θ cos(φ −φN ) + β N ]
+ +e
N
= ∑e
j[ ka sin θ cos(φ −φn )+ β n ]
(20)
n =1

(For a detailed derivation of the AF above, see ref. [1])

Hon Tat Hui
27 Antenna Arrays
NUS/ECE EE6832

Note that in the above expression, AF is a sum of N

complex exponentials. The magnitude of each complex
exponential is 1. Hence the maximum value of the
magnitude of |AF| is the addition of magnitudes of the
complex exponentials, i.e., N. The maximum radiation
direction (θmax, φmax) is therefore achieved when:

⎧ka sin θ max cos (φmax − 2π N ) + β1 = ±2qπ

⎪ka sin θ cos (φ − 4π N ) + β = ±2qπ
⎪ max max 2
⎨ (21)
⎪
⎪⎩ka sin θ max cos (φmax − 2π ) + β N = ±2qπ
q = 0,1,2,
Hon Tat Hui
28 Antenna Arrays
 NUS/ECE EE6832

One excitation method to achieve the above maximum

radiation direction is:

β n = ±2qπ − ka sin θ max cos (φmax − 2π n N ) , n = 1,2, , N (22)

q is choosen to make β n ∈ [ 0,2π )

Hence in a circular array, we first choose a desired

maximum radiation direction (θmax, φmax). Then the
excitation phase βn for each element is determined
according to the above formula. The βn so determined
may not be equally increasing from one element to the
next. This is different from the case of a linear array.
Hon Tat Hui
29 Antenna Arrays
NUS/ECE EE6832

Example 3
A uniform circular array with a radius a = 0.5λ and the
number of elements N = 8. The maximum radiation
direction of the array factor AF is at (60°, 30°). What should
be the excitation phases βn for the elements?
Solutions
Using the formula for βn , we have:
2π λ π ⎛ π 2π ⎞
β1 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 2π − 2.63
= 3.66
Hon Tat Hui
30 Antenna Arrays
NUS/ECE EE6832

2π λ π ⎛ π 4π ⎞
β 2 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 2π − 1.36
= 4.92
2π λ π ⎛ π 6π ⎞
β 3 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 0 + 0.70
= 0.70
2π λ π ⎛ π 8π ⎞
β 4 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 0 + 2.36
= 2.36
Hon Tat Hui
31 Antenna Arrays
NUS/ECE EE6832

2π λ π ⎛ π 10π ⎞
β 5 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 0 + 2.63
= 2.63
2π λ π ⎛ π 12π ⎞
β 6 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 0 + 1.36
= 1.36
2π λ π ⎛ π 14π ⎞
β 7 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 2π − 0.70
= 5.58
Hon Tat Hui
32 Antenna Arrays
NUS/ECE EE6832

2π λ π ⎛ π 16π ⎞
β8 = ±2qπ − sin cos ⎜ − ⎟
λ 2 3 ⎝6 8 ⎠
= 2π − 2.36
= 3.92

Azimuth pattern (θ = 60°) Vertical pattern (φ = 30°)

Hon Tat Hui
33 Antenna Arrays
NUS/ECE EE6832

Note that in the above discussion on circular arrays, we

have only derived the array factor AF. The array pattern
of any circular array with practical antenna elements
must be obtained by multiplying the array factor with the
actually element radiation pattern. For example, if the
elements are half-wave dipole antennas, then the array
pattern F(θ) is:
cos [(π 2 ) cosθ ]
F (θ ) = AF
sin θ
cos [(π 2 ) cosθ ] N j[ ka sin θ cos(φ −φn )+ βn ]
=
sin θ
∑
n =1
e (23)

Hon Tat Hui

34 Antenna Arrays
NUS/ECE EE6832

References:
1. C. A. Balanis, Antenna Theory, Analysis and Design, John Wiley
& Sons, Inc., New Jersey, 2005.
2. W. L. Stutzman and G. A. Thiele, Antenna Theory and Design,
Wiley, New York, 1998.
3. David K. Cheng, Field and Wave Electromagnetic, Addison-
Wesley Pub. Co., New York, 1989.
4. John D. Kraus, Antennas, McGraw-Hill, New York, 1988.
5. Fawwaz T. Ulaby, Applied Electromagnetics, Prentice-Hall, Inc.,
New Jersey, 2007.

Hon Tat Hui

35 Antenna Arrays