CHAPTER 3

THE DISCRETE-TIME FOURIER ANALYSIS

() h(n))
= () ()
DISCRETE-TIME FOURIER TRANSFORM (DTFT)
If x(n) is absolutely summable
() <

then

=

DTFT:
IDTFT (Inverse Discrete-Time Fourier Transform):

1

=
1
2

Example1: Determine the DTFT of = 0.5

()
Solution:

= =

=


= 0.5

0

= 0.5

=
1
1 0.5

0.5

0

Example2: Determine the DTFT of = 1,2,3,4,5
Solution:

= =

=

= 1
(1)
+ 0
0
+ (1)
1
+ x 2
2
+(3)
3



=

+ 2 +3

+4
2
+ 5
3

Properties of DTFT:
1. Periodicity

=
+2

2. Symmetry

Even symmetry:

odd symmetry:

even symmetry:

odd symmetry:
<

=<

Some Common DTFT Pairs:
Signal Type Sequence x(n)
DTFT

,
Unit Impulse () 1
Constant 1 2
Unit Step u(n)
1
1

+
Causal Exponential

()
1
1

Complex exponential

Cosine

+ +

Sine

Double exponential

() 1
2
1 2 +
2

The Properties of the DTFT:
1. Linearity

1
+
2
() =
1
+
2

3. Frequency Shifting
( ) =

4. Conjugation

() =

5. Folding
() =

2. Time Shifting
()

=

0

6. Symmetries in real sequences
=

()
if
then

() =

0
() =

7. Convolution

1

2
=
1

2
=
1

8. Multiplication

1

2
=
1

2

1
2

1

Periodic convolution
9. Energy

= ()
2

=
1
2

0
- PARSEVALS THEOREM
ENERGY DENSITY SPECTRUM

ENERGY OF x(n) in the band

1
,
2

The Frequency Domain Representation of LTI System
1. Response to a Complex Exponential

Let =

input to LTI system

h(n)

()

Then =

= ()

0
()

= ()

= ()

0

Frequency Response
- The discrete-time Fourier transform of an impulse response is called frequency
response (or transfer function) of an LTI system and is denoted by

()

H(

) =

h(n)

Magnitude or gain response function
<

Phase response function

2. Response to Sinusoidal Sequences
Let =
0
+
0

=

0

0
+
0
+ <

0

Steady state response,

H(

0
+
0

+ <

3. Response to Arbitrary Sequences
Let

= ()
and

= ()
Then using the convolution property

H(

=
1

=
1
2

Example: Determine the frequency response

of a system characterized by
= 0.9

(). Plot the magnitude and the phase responses.
Solution:

= ()

= 0.9

0
= 0.9

=
1
1 0.9

=
1
1 0.9
2
+ 0.9
2
=
1
1.81 1.8

<

=
0.9
1 0.9

Example : Causal Exponential Pulse

<
>
=

0 t 0
0 t
) (
bt
Ae
t v
From:
f j2 b
A

e A
) (
) ( ) (
0
)t j2 (b -
0
2
0
2
t
t
t
t
+
=
}
=
}
=
}
=

dt
dt e Ae f V
dt e t v f V
f
ft j bt
ft j
AN EXAMPLE ON HOW TO COMPUTE FOR AMPLITUDE
( )
A
f b
f j b
f j b
f j b
f j b
A
f V
2
2
2
2

2
2
2
) (
t
t
t
t
t
+

=
|
.
|

\
|

+
=
( ) | |
| |
2 2
2 2
) 2 (
2
) ( Im ) (
) 2 (
Re ) (
f b
fA
f V f V
f b
bA
f V f V
o
e
t
t
t
+
= =
+
= =
Convert to polar form to get and amplitude and phase spectrum:

( ) ( )
( ) ( )
( ) ( )
( )
( )
( )
( )
2
2
2
2
2
2
2
2
2
2 2
2
2
2
2 2
2
2 2
2
2 2
2 2
2
A

2
2
2
2
2
2 bA

) 2 (
2
) 2 (
) ( ) ( ) (
f b
f b
f b A
f b
f b A
f b
fA
f b
f
f b
bA
f V f V f V
c e
t
t
t
t
t
t
t
t
t
t
+
=
+
+
=
+
+
=
+
+
=
|
.
|

\
|
+
+ |
.
|

\
|
+
= + =
Amplitude Spectrum:
Phase Spectrum:
( )
( )
b
f 2
-arctan
bA
fA 2
- arctan
2
2
2
arctan
) (
) (
arctan ) ( arg
2
2
2
2
t
t
t
t
t
=
=
+
+

=
=
f b
bA
f b
fA
f V
f V
f V
e
o
4. Frequency Response Function from Difference Equations
+

( )

=0

+1

then

()
=

=1

()

=0

=0
1 +

=1

Example: An LTI system is specified by the difference equation
= 0.8 1 + ()
a. Determine

b. Calculate and plot the steady-state response

() to
Solution:
0.8 1 = ()
a.

=0
1 +

=1
=
1
1 0.8

= cos 0.05 ()
b. In the steady state the input is = cos 0.05 with frequency

0
= 0.05 and
0
= 0
0

therefore

= 4.0928 cos 0.05 0.5377 = 4.0928 0.05 3.42

0.05
=
1
1 0.8
0.05
= 4.0928
0.5377

Example: a 3
rd
-order LPF is described by the difference equation
= 0.081 +0.0543 1 +0.0543 2 + 0.0181 3 +
1.76 1 1.1829 2 + 0.2781 3

1.76 1 +1.1829 2 +0.2781 3 =
0.081 + 0.0543 1 +0.0543 2 +0.0181 3
Use matlab to plot
Introduction
2 mathematical tools in the analysis and
design of LTI systems:
Fourier Transform for nonperiodic signal
Fourier Series for periodic signal
Above signal representations decompose the
signals in terms of sinusoidal or complex
exponential components frequency domain

Frequency Analysis of Continuous-
Time Signals
Illustrated in prism by Isaac Newton
Joseph Fraunhofer observed that light consists
of distinct color lines
Gustav Kirchoff and Robert Bunsen found that
each chemical element, when heated to
incandescence, radiated its own distinct color
of light
Line spectrum identification of each chjemical
elements
Frequency analysis analysis of light into
colors
Spectrum frequency content of the signal
Spectrum estimation process of
determining the spectrum of a signal in
practice based on actual measurement of the
signal
Fourier Series for Continuous-Time
Periodic Signals
Developed by Jean Baptiste Joseph Fourier

period
1
T with
signal periodic ) ( if
0
p
2
0
l fundamenta
F
e c t x
k
t kF j
k
=
=

=
t
Frequency Analysis of Continuous Time Periodic Signals
Synthesis Equation


Analysis Equation

=
=
k
t kF j
k
e c t x
0
2
) (
t
}

=
p
T
t kF j
p
k
dt e t x
T
c
0
2
) (
1
t
If the periodic signal is real

k
k
j
k k
j
k k
e c c
then
e c c
u
u

=
=
and
( )
( ) ( ) | |
k k k
k k k
k
k k k
k
k k
c b
c a
c a
where
t kF b t kF a c c t x
t kF c c t x
u
u
t t
u t
sin 2
cos 2
2 sin 2 cos 2 ) (
real is when x(t) valued real is c where 2 cos 2 ) (
0 0
1
0 0 0
0
1
0 0
=
=
=
+ =
+ + =

=
Power Density Spectrum of Periodic
Signals
Periodic signals have:
Energy infinite
Average Power finite
signals power for relation ' ) (
1 2 2
s Parseval c dt t x
T
P
k
k
T
p
x
p
= =

}

=
Example: Find the power of the following signal
t kF j
k
e c t x
0
2
) (
t
=
Solution:

} }

=
= = =
k
k
T
k
p
T
p
x
c dt c
T
dt t x
T
P
p p
2 2 2 1
) (
1
Also called line spectrum because
the power of the periodic signal is
only at discrete values of frequencies
For real periodic signal, the total average power is
( )


=

=
+ + = + =
1
2 2 2
0
1
2
2
0
2
1
2
k
k k
k
k x
b a a c c P
k k k
k k k
c b
c a
c a
where
u
u
sin 2
cos 2
0 0
=
=
=
Example: Determine the Fourier series and the power density
spectrum of the rectangular pulse train signal as
illustrated below


2,... 1, k ,
sin A

2 F
A

2
1
) (
1
c
0 k
1
) (
1
c
x(t) signal the of component dc 0 k for
period T where
2
to
2
0
0
0
2 /
2 /
2 /
2 /
2
2
0
2
2
k
2 /
2 /
2 /
2 /
0
p
0 0
0
0
= =

=
(

=
=
=
=
= = =
=

}
}
} }
t t
t t t
t
t
t
t t t t
t
t
t
t
t
t
t
t
kF
kF
T
j
e e
kT
kF j
e
T
A
dt Ae
T
dt t x
T
for
T
A
Adt
T
dt t x
T
l fundamenta
T T
p
kF j kF j
p
T
T
kt F j
p
t kF j
p
p
T
T
p p p
p p
p
p
p
p
Solution:
since the signal is even, then integration interval is from
t t | |
t t
t t
0
0
0
kF where c sin
kF
kF sin
= =
| c sin =
varied with if t fixed T
p

t t >
p p
T T fixed when varied with if
Fourier Transform for Continuous-Time
Aperiodic Signals
If the periodic signal with
p
T
Then we can consider the signal as aperiodic/nonperiodic signal
}
}

=

=
s s =
=
= =
=
2 /
2 /
2
k
p p
p
2 /
2 /
2
k
o
2
p
0
0
) (
1
c
2
T
2
T -
for ) ( ) ( x sin
) (
1
c
1
F , ) ( x
) ( lim ) (
p
p
p
p
o
p
T
T
t kF j
p
T
T
t kF j
p
p
p
k
t kF j
k
p
T
dt e t x
T
then
t t x t ce
dt e t x
T
where
T
e c t from
t x t x
t
t
t
}

-
Ft j2 -
x(t)e X(F)
x(t) of Transform Fourier X(F)
dt
let
t

T
1
F
1
) ( x
) ( T
) (
1
c
) (
1
c from
p
2 2
p
0 p
0 k
2 /
2 /
2
k
0 0
0
= A
|
|
.
|

\
|
= =
|
|
.
|

\
|
= =
=
=

}

k
t kF j
p p
k
t kF j
k
p
k
p
T
T
t kF j
p
e
T
k
X
T
e c t from
T
k
X kF X c
or
kF X
T
dt e t x
T
p
p
t t
t
( )
( )
}

=
A

=
=
A A = =

A A
A A
=
-
2
2
T
p p
2
p
) ( ) x(t
lim ) ( ) ( lim
and
F F k and dF F ) ( ) ( x , T
) ( x
0
p
0
dF e F X
F e F k X t x t x so
t x t as
F e F k X
t
Ft j
k
t kF j
F
p
k
t kF j
t
t
t
FREQUENCY ANALYSIS OF CONTINUOUS TIME
APERIODIC SIGNALS
Synthesis Equation (inverse transform
Analysis equation (direct transform)
}

=
-
Ft j2 -
x(t)e X(F) dt
t
}

=
-
2
) ( x(t) dF e F X
Ft j t
IN RADIAN FORM
( )
( )
}
}

O
= O
O O =
dt e t x X
d e X t x
t j
t j
) (
2
1
) (
t
The Fourier Transform exists if it follows the Dirichlet Conditions:
1. The signal x(t) has a finite number of finite discontinuities.
2. The signal x(t) has a finite number of maxima and minima.
3. The signal x(t) is absolutely integrable, that is

}

< dt t x ) (
ENERGY DENSITY SPECTRUM OF APERIODIC SIGNALS
} }
}
} }
} } }
}

= =
=
(

- =
(

- = - =
=
-
2 2
x
-
2
- -
2
2
x
2
x
RELATION ' X(F) ) ( E
X(F)
) ( (F)dF X
) ( ) ( ) ( ) ( E
) ( E
signal energy finite - x(t)
S PARSEVAL dF dt t x
therefore
dF
dt e t x
dF e F X dt t x dt t x t x
dt t x
Let
Ft j
Ft j
t
t
( )
spectrum density energy symmetry even has signal real ) ( ) ( S
signal the of Energy ) (
x(t) of SPECTRUM DENSITY ENERGY
frequency of function a as
signal in the energy of ) ( ) ( S and
spectrum phase X(F) (F)
spectrum X(F)
X(F) X(F)
form polar in
xx
F
2
xx
1
1
=

=
Z = O

=
}
A +
O
F S F
and
Total dF F S
on distributi F X F
magnitude where
e
xx
F F
xx
F j

>
s
=
2
t , 0
2
t A,
x(t)
: following the of
spectrum density energy the and Transform Fourier the Determine :
t
t
Example
Ft
Ft
A F X
dt Ae dt e t x F X
Ft j Ft j
t
t
t
t
t t
sin
) (
) ( ) (
2
2
2 2
=
= =
} }

p
p p p
k
T
k
T
k
X
T
kF X
T
c
=

|
|
.
|

\
|
= =
o
0
kF
s frequencie at X(f) of samples
1
) (
1
The zero-crossings occur at
multiples of
t
1
Main lobes width contains
most of the signal
=
t
2
( )
2
2
xx
sin
) ( S
Spectrum Density
|
.
|

\
|
=
t t
t t
t
F
F
A F
Energy
Assignment
1. Find the Fourier Transform and energy density spectrum
of
) (
0
t t x
Sampling and Reconstruction of Analog Signals
Sampling
Let

() be an analog (absolutely integrable) signal

Continuous-Time Fourier Transform (CTFT)

()

()

Where analog frequency (rad/sec)

=
1
2

()

=
1

ALIASING FORMULA
Digital And Analog frequencies are related through
=

with

, /
With =

Definition:
A signal is bandlimited if there exists a finite radian frequency
0
such that

= 0 for >
0
. The frequency
0
=

0
2
is called the signal bandwidth in Hz.
Sampling Principle Theorem:
A bandlimited signal

() with a bandwidth
0
can be reconstructed from its sample
Values =

if the sampling frequency

=
1

is greater than twice the bandwidth

0
of

()

> 2

Otherwise aliasing would result in x(n). The sampling rate of 2
0
for an analog bandlimited
Signal is called the Nyquist rate.
Example: The analog signal

= 4 +2 150 +

3
+4 350 is sampled
at

= 200/ to obtain the discrete-time signal x(n). Determine x(n)

and its corresponding DTFT

.
Solution:
Highest frequency of

=175

= 200/
The sampled signal is then
=

= 4 +2
150
200
+

3
+4
350
200

=

= 4 + 2 0.75 +

3
+4 1.75
Since 1.75 Is outside the primary interval
Then aliasing occurs
The alias of the frequency 1.75 is

=

= 4 + 2 0.75 +

3
+4 1.75 2
=

= 4 + 2 0.75 +

3
4 0.25
Using Eulers identity
= 4 +
/3

0.75
+
/3

0.75
+ 2
0.25
2
0.25

From the DTFT table, the DTFT of x(n) is

= 8 + 2
/3
0.75 +2
/3
+ 0.75
+4 0.25 4 + 0.25

Reconstruction
Condition for reconstruction:
1. Signal is bandlimited
2. Sampling above Nyquist sampling rate
To reconstruct:
1. Samples are converted into a weighted impulse train

= + 1 +

+ 0 + 1

=

2. Impulse train is filtered through an ideal analog LPF (low pass filter)
bandlimited to the

band
x(n)
Impulse train
conversion
Ideal LPF

()
Can be described mathematically as

=

where
=

Example: Consider =

= 4 + 2 0.75 +

3
+4 1.75
With

= 200/
Obtain the reconstructed signal

(), and determine whether the sampling/

reconstruction operation resulted in any aliasing. Also, plot the Fourier
Transform

and

Solution:
With
=

= 4 + 2 0.75 +

3
+4 1.75

= |
=

= |
=200

= 4 + 2 0.75200 +

3
4 0.25200

() = 4 + 2 150 +

3
4 50
Using Eulers identity, the CTFT

is given by

= 8 +2
/3
150 + 2
/3
+ 150
+4 350 4 +350
With = 2

2 = 8 2 + 2
/3
2 150 + 2
/3
2 + 150
+4 2 350 4 2 +350

2 = 4 +
/3
75 +
/3
+75
+2 175 2 + 175
Using = 2 =
1
2

and

2 = 4 +
/3
75 +
/3
+75
+2 25 2 + 25
from

() = 4 + 2 150 +

3
4 50

() = 4 + 2 150 +

3
4 50
From the DTFT table, the DTFT of x(n) is

= 8 + 2
/3
0.75 +2
/3
+ 0.75
+4 0.25 4 + 0.25

Practical D/A Converters
1. Zero-order-Hold (ZOH) interpolation:

= ,

< +1

The interpolating filter is

=
1, 0

0,

A rectangular pulse
x(n) ZOH

Postfilter

()
Postfilter used for accurate reconstruction
2. 1
st
-order-hold (FOH) interpolation:

1
=
1 +

, 0

0,

3. Cubic spline interpolation - does not need an analog postfilter

=
0
+
1

+
2

2
+
3

3
,

+1

, 0 3 are the polynomial coefficients which are determined

by using least-squares analysis on the sample values.