BERL 3114: Technology System

Optimization II
Chapter 4:
Fourier Transform (FT)
and
Discrete Time Fourier Transform
(DTFT)
MUHAMMAD FAREQ BIN IBRAHIM
FTKEK
PART 1 – FOURIER
TRANSFORM
Introduction
 oo

V (ω ) = ∫ v ( t ) e - jωt

dt
Fourier
- oo

Transform
V
v(t
(ω )
)
oo
1
v(t )= ∫ V (ω )e j ω t d ω
- oo
2π
Inverse Fourier
Transform
 Why do we need
Fourier
Transform?
 Basically, a signal has three information Amplitude (A), Frequency
(F) and phase (Ф).

Single frequency signal

 In DSP, filtering is based on frequency of the signal, where the

desired frequency component will be preserved while the unwanted
frequency component will be removed from the signal.
 In this case, FT is used to sort the signal based on its frequency.
When the signal has been sorted based on its frequency, the process
of preserving and removing frequency component in the signal will
be possible.

 In C-T signal, frequency is labeled with f and the unit is Hz.

 In D-T signal, frequency is labeled with f where,

Frequency of continuous signal – Hz

Sampling frequency – Hz
 ReVisit: Complex Exponential &
Sinusoids
• CT complex exponential can be derived from its sinusoidal
function as follows:
x(t ) = A cos(ω
0 t + ϕ ) e
A +ϕ)
[
j (ωo t
+ e- j (ωot +ϕ
)
=
2 ]
This comes from Euler’s Formula:

Another way to write Euler’s Formula:

e jθ = cos(θ) + j e - jθ = cos(θ) - j
sin(θ) sin(θ)
Fourier Transform
Pairs
 Properties of Fourier
Transform
• Linearity or Superposition
• Time Shifting
• Time Scaling
• Duality
• Frequency Shifting (Modulation)
• Convolution
• Differentiation
(1) Linearity or
Superposition
f1 (t ) ←⎯ → F1 (ω)
F

f 2 (t ) ←⎯ → F2 (ω)
F

af1 (t )+ bf 2 (t ) ←⎯ → aF1 (ω)+ bF2

F

(ω)
 (2) Time Shifting

f (t - t0 ) ←⎯F → F( jω )e 0

- j ωt

(3) Time Scaling

1 F ⎛⎜ j ω
f (at) ←⎯F →
⎝ ⎞
⎟
|a| ⎠ a
 (4) Duality

f (t) ←⎯F →2πf ( -

jω )
(5) Frequency Shifting (Modulation)

f (t)e jω ←⎯F →F [ j(ω - ω

0

)]
0
 (6) Convolution

f1 (t)* f 2 (t) ←⎯F → F1 ( jω)F2

( jω)
f1 (t) = e - at u(t) a>0

f 2 (t) = e - bt u(t) b>0

f (t) = f1(t)
F*f(2ω
(t)) = F (ω)F (ω) =⎜
1 2
1 ⎟⎞⎜⎛ 1
⎟⎝ ⎞ ⎠⎝ ⎠
⎛
a + jω b+
1 jω
=
(a + jω)(b +
jω )
 (7) Differentiation

v(t) ←⎯F → V
( ω)

d v(t) F (jω)
n n

dt ←⎯ →
n
V(ω)
Fourier Transform
Properties
 Continuous to
Sampling Theorem
x(t) Discrete x(n)
x(t) Continuous Discrete
x(n)

t n
Laplace Transform z Transform
x(t) X(s) x(n)
oo
n=o X(z)
X (s) = ∫ x(t)e
-
o

st
dt
X (z) = ∑ x(n)z -

n
n = - oo
- oo

Fourier Transform Discrete Fourier Transform

x(t) oo X(jω x(n) X(k)
N-1
X ( ω) = ∫ x(t)e -
) 2 πk
∑ x(n)e
- j
N n
X (k)
jωt
dt = n=0
- oo
Fourier Transformation
Formula
PART 2 –
DISCRETER TIME
FOURIER
TRANSFORM
What is DTFT
representation?
Brief Introducti on

Discrete Time Fourier Transform (DTFT)

representati on is a branch of Fourier Analysis
approaches. This Fourier representati on is very
useful to convert an APERIODIC (NON-
PERIODIC) discrete time signal (DT signal) from
the ti me-domain into its frequency domain.
DTFT Concept
What type of signals can use DTFT?
DTFT can be applied to discrete time (DT) signals ONLY. This Fourier representati on can not be
applied to conti nous time (CT) signals. DTFT can be applied to APERIODIC or NON-PERIODIC
signals.

Time-domain vs Frequency-domain
The relationship of ti me-domain DT signal and frequency domain DT signal is shown by the
following equation. represents the DT signal/sequence in time-domain.
represents the DT signal in frequency domain. and can be interchanged by using
DTFT representation. is also written as .
 DTFT Concept
What i s t h e K E Y POINT o f D T F T ?

Using DTFT representation, we are actually fi nding the magnitude,

and phase, of the signal. Therefore, it is convenient to write
in the POLAR form. The following equation shows the relationship
between magnitude and phase of the signal in the POLAR form.
 Periodici
ty
Linearity

Properties of DFT Time Shifting

Modulation

Symmetry

Circular Convolution
 01 Periodicity
Let the DFT of a DT signal, x[n]
given as:
𝐷𝐹𝑇
𝑥 𝑛 ;𝑁
𝑋[𝑘]
X[k] is N periodic;

𝑋 𝑘+𝑁 =
Properties of DFT 𝑋[𝑘]
This property follows periodicity of DTFT,

x[n] is als𝑋o

p(e𝑒ri𝑗oΩd)ic;

𝑥 𝑛 =
+𝑁 𝑥[𝑛]
 02 Linearity
If x1[n] and x2[n] are two DT
sequences with the following N-
point DFT pairs:

Properties of DFT
then the linearity property states that:

for any arbitrary constants a and b, which

may be complex-valued.
 03 Time
Shifting
The shifting of sequence in time
domain by M samples equals to
multiplying the sequence in frequency
domain by the following term:

Properties of DFT
Therefore, the relationship of time shifting
in time-domain and frequency domain is:
 04 Modulati
on
If a DT signal, x[n] is modulated
(multiplied) by a complex sinusoid with
2𝜋
the frequency of
𝑚;
�
�

Properties of DFT

2 𝐷𝐹𝑇;
𝑥 𝑛 𝑒 𝜋
𝑗 𝑁 𝑋[𝑘
− 𝑀]
𝑁 𝑀𝑛
 05 Symmetry
Property
If x[n] is real, DFT can be found using
symmetry property.
In general, the symmetry property can
be written as:

Properties of DFT 𝑋∗ 𝑘 = 𝑋 −𝑘
=𝑋 𝑁−𝑘

𝑘 = 0,1, … (𝑁 − 1)
Thus, to obtain X[k] for a symmetry
sequence, ONLY values for need
to be computed, which is up to:

𝑘
�
=
2
�
 06 Circular
Convolution
If x1[n] and x2[n] are two DT
sequences with the following N-
point DFT pairs:

Properties of DFT
The circular convolution of the x1[n] and
x2[n] in time-domain equals multiplication
of their DFT coefficients in frequency
domain:
 06 Circular
Convolution
To apply the circular convolution
property in the frequency domain,
the following condition must be
fulfilled to ensure the output

Properties of DFT samples are not overlapped. For

instance;

The number of samples N to be

computed

𝑁 =
is:

max(𝑁1, 𝑁2)
Example of DTFS
From given Time-domain signal, x[n], find the Discrete Time Fourier Series (DTFS), X[k].
Solution
Step 1: Find number of periodic (N)

N=5
We know that: o=2/N.
Therefor, o=2/5
Solution
N1
1
Step 2: Insert the value in equation  X k    xne  jk o n

N n 0

1
2 − 𝑗𝑘 ( 2𝜋
)𝑛
𝑋 [ 𝑘 ]=
5 ∑ 𝑥[𝑛]𝑒
5

𝑛 =− 2

Step 3: Insert the amplitude of x(n) in equation for all “n”

[
1 − 𝑗𝑘( ) ( )
]
2𝜋 2𝜋

( )
( − 1) − 𝑗𝑘 (1 )
1 1
𝑋 [ 𝑘 ]=
5 5
0+ 𝑒 +1+ − 𝑒 +0
5 2 2

n = -2 n = -1 n=0 n=2
n=1
 Solution
Euler′s Formula
𝑒 𝑗 𝑥 = cos𝑥 +
Step 4: Simplify the equation (mostly refer to Euler’s Formula) if possible

𝑗𝑠𝑖𝑛 𝑥
[
1 − 𝑗𝑘( ) ( )
] cos 𝑥
2𝜋 2𝜋

( ) 1 𝑒 𝑗𝑥 +
( − 1) − 𝑗𝑘 (1 )
1 1
2
𝑋 [ 𝑘 ]=
5 5
0+ 𝑒 +1+ − 𝑒 +0
=
5 2 2
sin 𝑥 1 𝑒 −𝑗𝑥
𝑋 [ 𝑘 ]=
1 1 −
5 2
𝑒[ 𝑗𝑘
( 25𝜋 ) ( −1) +1+ ( − ) 𝑒
1
2
− 𝑗𝑘
( 25𝜋 ) (1 )
] =
2𝑗
𝑗𝑥
−𝑗𝑥𝑒 − 𝑒

[ (𝑒 ( ) − 𝑒 − 𝑗𝑘 ( ))
]
2𝜋 2𝜋
𝑗𝑘
5 5
1
𝑋 [ 𝑘 ]= 1+
5 2

𝑋 [ 𝑘 ]=
1
5 [1 + 𝑗 sin
( ( 𝑘
2𝜋
5 ))]