Lecture 3

Discrete-Time Fourier Analysis Discrete-Time Fourier Analysis

(chapter 3)
Review
The output y(n) of a linear system to an arbitrary input x(n)
is called impulse response (denoted by h(n,k))
For a LTI system, unit impulse response becomes
] ) ( [ ) ( ) ( ) ( )] ( [ ) (

+
=
+
=
=
(

= =
n k
k n L k x k n k x L n x L n y
)] ( [ k n L
For a LTI system, unit impulse response becomes
A LTI system can be represented in terms of its response
to the unit sample sequence.
The convolution representation is based on the fact that
any signal can be represented by a linear combination of
scaled and delayed unit samples.
) ( )] ( [ ) , ( k n h k n L k n h = =
Representation
Also can represent any arbitrary discrete signal as a linear
combination of basis signals introduced in Chap 2.
Each basis signal set provides a new signal representation.
Each representation has some advantages and
disadvantages depending upon the type of system under disadvantages depending upon the type of system under
consideration.
Consider DTFT: It is based on the complex exponential
signal set { }
n j
e

Discrete-Time Fourier Transform
If x[n] is absolutely summable, then its
DTFT is given by

+
=

= =
n
jwn jw
e n x n x F e X ) ( )] ( [ ) (

= =

dw e e X e X F n x
jwn jw jw
) (
2
1
)] ( [ ) (
1
DTFT:
IDTFT:

+

< | ) ( | n x
F[.] transforms a discrete signal x(n) into a complex-valued F[.] transforms a discrete signal x(n) into a complex-valued
continuous function X of real variable w, called a digital
frequency, which is measured in radians.
Time domain Frequency domain
Discrete Continuous
Real valued Complex-valued
Summation Integral
The range of w:
The integral range of w:
+
+
Discrete-Time Fourier Transform
If x[n] is absolutely summable, then its
DTFT is given by

+
=

= =
n
jwn jw
e n x n x F e X ) ( )] ( [ ) (

= =

dw e e X e X F n x
jwn jw jw
) (
2
1
)] ( [ ) (
1
DTFT:
IDTFT:

+

< | ) ( | n x
F[.] transforms a discrete signal x(n) into a complex-valued F[.] transforms a discrete signal x(n) into a complex-valued
continuous function X of real variable w, called a digital
frequency, which is measured in radians.
Ex.31: Determine the Discrete-Time Fourier Transform of
This seq. is ABS SUM, thus its DTFT exists
) ( ) 5 . 0 ( ) ( n u n x
n
=
Two Important Properties
Periodicity:
The DTFT is periodic in w with period 2pi
Implication: we need only one period for analysis
and not the whole domain
Symmetry:
) ( ) (
] 2 [ +
=
w j jw
e X e X
+ < <
] , [ or ], 2 , 0 [
Symmetry:
For real-valued x(n), X is conjugate symmetric.
Implication: to plot X, we now need to consider
only a half of X: [0,pi]
) ( ) (
* jw jw
e X e X =

Matlab Implementation
If x(n) is of infinite duration, then Matlab can not be used
directly to compute X from x(n). But we can use it to
evaluate the expression X over [0,pi] frequencies and then
plot its magnitude & angle (or real & imaginary parts).
Evaluate in exp.3.1 at 501 equi-spaced points
between [0,pi]
) (
jw
e X
between [0,pi]
Strongly recommend to plot freq. in the units of pi
For a finite duration, if we evaluate at equi-spaced
freq. between [0,pi], the DTFT can be implemented as a
matrix-vector multiplication operation.
) (
jw
e X
Matlab Implementation
When are arranged as column vectors x
and X respectively, we have
In Matlab
{ } { } ) ( and ) (
k
jw
l
e X n x
Wx X =
In Matlab, a dtft function was made for this operation.
This way of calculation produces N x (M+1) matrix: memory
cost DFT FFT for more efficient computation
Example 3.5: periodicity / symmetry ?
Example 3.6: periodicity / symmetry ?
The properties of the DTFT
1. Linearity:
The DTFT is a linear transformation.
2. Time shifting:
A shift in time domain corresponds to the phase shifting.
3. Frequency shifting:
Multiplication by a complex exponential corresponds to a
shift in the frequency domain.
4. Conjugation:
Conjugation in the time domain corresponds to the folding
and conjugation in the frequency domain.
The properties of the DTFT
5. Folding:
Folding in the time domain corresponds to the folding
in the frequency domain.
6. Symmetries in real sequence:
Implication: If the sequence x(n) is real and even, then
X is also real and even.
7. Convolution:
Freq. domain representation of LTI system
The Fourier transform representation is the most
useful signal representation for LTI systems.
It is due to the following result:
Response to a complex exponential e
jw0n
Response to sinusoidal sequences
Response to arbitrary sequences
Response to a complex exponential
Frequency response: the DTFT of an impulse response is
) (n h
n jw
e
0
n jw
w w
n jw
e n h F e n h
0
0
0
] | )] ( [ [ * ) (
=
=
Frequency response: the DTFT of an impulse response is
called the frequency response (or transfer function) of an
LTI system and is denoted by H().
) (
jw
e H
n jw
e
0
n jw jw
e e H
0 0
) (
The output sequence is the input exponential sequence
modified by the response of the system at frequency w
0
In general, the frequency response H is a complex
) ( ), ( n h e H
jw

k
n jw
k
k
e A

k
n jw jw
k
k k
e e H A ) (
Response to a complex exponential
In general, the frequency response H is a complex
function of w.
The magnitude |H| is called the magnitude (gain)
response function, and
the the angle is called the phase response function.
Response to sinusoidal sequences
)) ( cos( | ) ( | ) (
) cos( ) (
0 0
0 0
0 0
+ + =
+ =
jw jw
e H n w e H A n y
n w A n x

) ) ( cos( | ) ( | ) (
) cos( ) (
0 0

+ + =
+ =
k
jw
k k
jw
k
k
k k k
e H n w e H A n y
n w A n x

Steady-state response
Response to arbitrary sequences
) (n h
) (n x
) ( * ) ( ) ( n x n h n y =
) ( ), ( n h e H
jw
) (
jw
e X
) ( ) ( ) (
jw jw jw
e X e H e Y =
Condition:
1. Absolutely summable sequence
2. LTI system
Frequency response function from
difference equations
When an LTI system is represented by the difference
equation,

= =
= +
N
l
M
m
m l
m n x b l n y a n y
1 0
) ( ) ( ) (
then to evaluate its frequency response , we would need
the impulse response h(n).
We know that when , then y(n) must be
jwn
e n x = ) (
jwn jw
e e H ) (

+
=
N
l
jwl
l
M
m
jwm
m jw
e a
e b
e H
0
0
1
) (
Implementation in Matlab
a a a a
b b b b
w jl a w jm b e H
M
T T jw
] , , [
] , , [
) exp( / ). exp( ) (
1 0
=
=
=
L
L
K k pi w
K k
N l
M m
a a a a
N
/ *
] , , 2 , 1 , 0 [
] , , 2 , 1 , 0 [
] , , 2 , 1 , 0 [
] , , [
1 0
=
=
=
=
=
L
L
L
L
Sampling & reconstruction of analog signals
Analogy signals can be converted into discrete signals
using sampling and quantization operations: analogy-to-
digital conversion, or ADC
Digital signals can be converted into analog signals using
a reconstruction operation: digital-to-analogy conversion, a reconstruction operation: digital-to-analogy conversion,
or DAC
Using Fourier analysis, we can describe the sampling
operation from the frequency-domain view-point, analyze
its effects and then address the reconstruction operation.
We will also assume that the number of quantization
levels is sufficiently large that the effect of quantization on
discrete signals is negligible.
Sampling
Continuous-time Fourier transform and inverse CTFT

+


= dt e t x j X
t j
a a
1
) ( ) (

= d e j X t x
t j
a a
) (
2
1
) (

1. Absolutely integrable
2. Omega is an analogy frequency in radians/sec
Sample x
a
(t) at sampling interval T
s
sec apart to obtain the
discrete-time signal x(n)
(
| |
=
s a
nT x n x

) ( ) (
Sampling

+
=
(

|
|

\
|
=
l
s s
a
s
jw
l
T T
w
j X
T
e X
2 1
) (
1. X is a countable sum of amplitude-scaled, frequency-
scaled, and translated version of X
a
2. The above relation is known as the aliasing formula
The analog and digital frequencies
s
s
s
T
F
T w
1
=
=
Fs: the sampling frequency, sam/sec
s
T
Amplitude scaled factor: 1/T
s
;
Frequency-scaled factor: =T
s
(=0~2)
Frequency-translated factor: 2k/T
s
;

0
T
S
-
0
T
S

0
T
S
-
0
T
S
Sampling Interpretation
Suppose signal band is limited to
0
,
If T
s
is small,
0
T
s
<, or
F
0
=
0
/2 < Fs/2=1/2T
s
F
0
=
0
/2 < Fs/2=1/2T
s
Then the freq. resp. of x(t) is an infinite replica series of
its analog signal x
a
(t),
If T
s
is large,
0
T
s
>, or
F
0
=
0
/2 > Fs/2=1/2T
s
Then the freq. resp. of x(t) is a overlaped replica of its
analog signal x
a
(t), so cannot be reconstructed
Band-limited signal
A signal is band-limited if there exists a finite
radians frequency
0
such that X
a
(j ) is zero for
| |>
0
. | |>
0
.
The frequency F
0
=
0
/2pi is called the signal
bandwidth in Hz
Referring to Fig.3.10, if pi>
0
T
s
, then
s s s s
a
s
jw
T T
w
T T
w
j X
T
e X

<
|
|

\
|
= ;
1
) (
Sampling Principle
A band-limited signal x
a
(t) with bandwidth F
0
can
be reconstructed from its sample values
x(n)=x
a
(nT
s
) if the sampling frequency F
s
=1/T
s
is
greater than twice the bandwidth F of x (t) , F greater than twice the bandwidth F
0
of x
a
(t) , F
s
>2 F
0
.
Otherwise aliasing would result in x(n). The
sampling rate of 2 F
0
for an analog band-limited
signal is called the Nyquist rate.
Reconstruction
Impulse train
conversion
Ideal lowpass filter
) (n x
) (t x
a
L L + + + + + =

+
) ( ) 1 ( ) ( ) 0 ( ) ( ) 1 ( ) ( ) ( T t x t x T t x nT t n x L L + + + + + =

=
) ( ) 1 ( ) ( ) 0 ( ) ( ) 1 ( ) ( ) (
s s
n
s
T t x t x T t x nT t n x

+
=
=
n
s s a
nT t F n x t x )] ( [ sinc ) ( ) (
1. Lowpass filter band-limited to the [-Fs/2,Fs/2] band
2. The ideal interpolation is not practically feasible because
the entire system is noncausal and hence not realizable.
Interpolating formula
Reconstruction
Practical D/A converters
Zero-order-hold (ZOH) interpolation:
In this interpolation a given sample value is held for the
sample interval until the next sample is received.
It can be obtained by filtering the impulse train through It can be obtained by filtering the impulse train through
an interpolating filter of the form

=
+ =
otherwise 0
0 1
) (
) 1 ( ), ( ) (

0
s
s s a
T t
t h
T n n nT n x t x
Zero-order-hold (ZOH) interpolation
The resulting signal is a piecewise-constant (staircase)
waveform which requires an appropriately designed
analog post-filter for accurate waveform reconstruction.
ZOH Post-Filter
) (n x
) (t x
a
) (

t x
a
First-order-hold (FOH) interpolation
In this case the adjacent samples are joined by straight
lines.

+ 0 1
s
T t
T
t

+
=
otherwise 0
2 1
0 1
) (
1 s s
s
s
s
T t T
T
t
T t
T
t h
Cubic-order-hold (COH) interpolation
This approach uses spline interpolation for a smoother,
but not necessarily more accurate, estimate of the analog
signal between samples.
Hence this interpolation does not require an analog post- Hence this interpolation does not require an analog post-
filter
The smoother reconstruction is obtained by using a set of
piecewise continuous third-order polynomials called cubic
splines
s s
s s s a
T n t nT
nT t n nT t n nT t n n t x
) 1 (
, ) )( ( ) )( ( ) )( ( ) ( ) (
3
3
2
2 1 0
+
+ + + =