Digital Signal Processing Pre-Reqs PDF

1
1
Copyright 2005, S. K. Mitra
Discrete Fourier Transform Discrete Fourier Transform
Definition- The simplest relation between a
length-N sequence x[n], defined for
, and its DTFT is
obtained by uniformly sampling on
the -axis between at ,
From the definition of the DTFT we thus have
1 0 N n
1 0 N k
) (
j
e X
) (
j
e X
< 2 0 N k
k
/ 2 =
, ] [ ) ( ] [
1
0
/ 2
/ 2

=

=
= =
N
n
N n k j
N k
j
e n x e X k X
1 0 N k 2
Note: X[k] is also a length-N sequence in
the frequency domain
The sequenceX[k] is called the discrete
Fourier transform(DFT) of the sequence
x[n]
Using the notation the DFT
is usually expressed as:
N j
N
e W
/ 2
=
1 0 , ] [ ] [
1
0
=

=
N k W n x k X
N
n
n k
N
3
The inverse discrete Fourier transform
(IDFT) is given by
To verify the above expression we multiply
both sides of the above equation by
and sum the result from n =0to 1 = N n
n
N
W
l
1 0 , ] [
1
] [
1
0
=

=

N n W k X
N
n x
N
k
kn
N
4
resulting in

=
=
1
0
1
0
1
0
1
N
n
n
N
N
k
kn
N
N
n
n
N
W W k X
N
W n x
l l
] [ ] [

=

=
1
0
1
0
1
N
n
N
k
n k
N
W k X
N
) (
] [
l

=

=
1
0
1
0
1
N
k
N
n
n k
N
W k X
N
) (
] [
l
5
Making use of the identity
we observe that the RHS of the last
equation is equal to
Hence
=

1
0
N
n
n k
N
W
) ( l
,
,
0
N , rN k = l for
otherwise
r an integer
] [l X
] [ ] [ l
l
X W n x
N
n
n
N
=
=
1
0
6
Example- Consider the length-N sequence
Its N-point DFT is given by
1 1 0
0 1

=
N n
n
,
,
= ] [n x
1 0
0
1
0
= = =

=
N
N
n
kn
N
W x W n x k X ] [ ] [ ] [
1 0 N k
2
7
Its N-point DFT is given by
= ] [n y
1 1 1 0 0
1
+
=
N n m m n
m n
, ,
,
km
N
km
N
N
n
kn
N
W W m y W n y k Y = = =

=
] [ ] [ ] [
1
0
1 0 N k
8
defined for
Using a trigonometric identity we can write
1 0 ), / 2 cos( ] [ = N r N rn n g
( )
N rn j N rn j
e e n g
/ 2 / 2
2
1
] [

+ =
( )
rn
N
rn
N
W W + =

2
1
1 0 N n
9
TheN-point DFT of g[n] is thus given by
=
=
1
0
N
n
kn
N
W n g k G ] [ ] [
,
2
1
1
0
) (
1
0
) (
+ =

=
+
=

N
n
n k r
N
N
n
n k r
N
W W
1 0 N k
10
we get
=
=
=
otherwise 0
for 2
for 2
,
, /
, /
] [ r N k N
r k N
k G
=

1
0
N
n
n k
N
W
) ( l
,
,
0
N , rN k = l for
otherwise
r an integer
1 0 N k
11
Matrix Relations Matrix Relations
The DFT samples defined by
can be expressed in matrix form as
where
1 0 , ] [ ] [
1
0
=

=
N k W n x k X
N
n
kn
N
x D X
N
=
[ ]
T
N X X X ] [
.....
] [ ] [ 1 1 0 = X
[ ]
T
N x x x ] [
.....
] [ ] [ 1 1 0 = x
12
and is the DFT matrix given by
N
D N N
2
1 1 2 1
1 2 4 2
1 2 1
1
1
1
1 1 1 1
) ( ) ( ) (
) (
) (
N
N
N
N
N
N
N
N N N
N
N N N
N
W W W
W W W
W W W
L
M O M M M
L
L
L
D
3
13
Likewise, the IDFT relation given by
can be expressed in matrix form as
where is the IDFT matrix
1 0 , ] [ ] [
1
0
1
=

=

N n W k X n x
N
k
n k
N
N
N N
1
N
D
X D x
1
=
N
14
where
Note:
2
1 1 2 1
1 2 4 2
1 2 1
1
1
1
1
1 1 1 1
) ( ) ( ) (
) (
) (
N
N
N
N
N
N
N
N N N
N
N N N
N
W W W
W W W
W W W
L
M O M M M
L
L
L
D
*
N N
N
D D
1
1
=
15
DFT Computation Using DFT Computation Using
MATLAB MATLAB
The functions to compute the DFT and the
IDFT arefft andifft
These functions make use of FFT
algorithms which are computationally
highly efficient compared to the direct
computation
Programs 5_1.mand 5_2.millustrate the
use of these functions
16
DFT Computation Using DFT Computation Using
MATLAB MATLAB
Example- Program 5_3.mcan be used to
compute the DFT and the DTFT of the
sequence
as shown below
15 0 ), 16 / 6 cos( ] [ = n n n x
0 0.2 0.4 0.6 0.8 1
0
2
4
6
8
10
Normalized angular frequency
M
a
g
n
itu
d
eindicates DFT samples
17
DTFT from DFT by DTFT from DFT by
Interpolation Interpolation
TheN-point DFT X[k] of a length-N
sequencex[n] is simply the frequency
samples of its DTFT evaluated at N
uniformly spaced frequency points
Given theN-point DFT X[k] of a length-N
sequence x[n], its DTFT can be
uniquely determined fromX[k]
) (
j
e X
) (
j
e X
1 0 , / 2 = = N k N k
k
18
Thus
=

=

1
0
] [ ) (
N
n
n j j
e n x e X
n j
N
n
N
k
n k
N
e W k X
N

=
1
0
1
0
] [
1
=

=
=

1
0
1
0
) / 2 (
] [
1
N
k
N
n
n N k j
e k X
N
S
4
19
To develop a compact expression for the
sum S, let
Then
From the above
) / 2 ( N k j
e r

=
=

=
1
0
S
N
n
n
r
1 1 S
1
1 1
+ + = =

= =
N N
n
n N
n
n
r r r r
1 1
1
0
+ = + =

=
N N N
n
n
r S r r
20
Or, equivalently,
Hence
N
r r r = = 1 S ) 1 ( S S
)] / 2 ( [
) 2 (
1
1
1
1
S
N k j
k N j N
e
e
r
r

=
] 2 / ) 1 )][( / 2 [(
2
2
sin
2
2
sin

=
N N k j
e
N
k N
k N
21
Therefore

=

=

1
0
] 2 / ) 1 )][( / 2 [(
2
2
sin
2
2
sin
] [
1
N
k
N N k j
e
N
k N
k N
k X
N
) (
j
e X
22
Sampling the DTFT Sampling the DTFT
Consider a sequencex[n] with a DTFT
We sample at N equally spaced points
, developing theN
frequency samples
TheseN frequency samples can be
considered as anN-point DFT Y[k] whoseN-
point IDFT is a length-N sequencey[n]
) (
j
e X
) (
j
e X
N k
k
/ 2 = 1 0 N k
)} ( {
k
j
e X

23
Now
Thus
An IDFT of Y[k] yields
=

=

l
l
l
j j
e x e X ] [ ) (
) ( ) ( ] [
/ 2 N k j j
e X e X k Y
k

= =
= =

=
=

l
l
l
l
l l
k
N
N k j
W x e x ] [ ] [
/ 2
=

=

1
0
] [
1
] [
N
k
n k
N
W k Y
N
n y
24
i.e.

=
=
1
0
1
N
k
n k
N
k
N
W W x
N
n y
l
l
l] [ ] [
=

=

=
1
0
1
N
k
n k
N
W
N
x
) (
] [
l
l
l
=

1
0
1
N
n
r n k
N
W
N
) (
,
,
0
1 mN n r + = for
otherwise
5
25
we arrive at the desired relation
Thusy[n] is obtained fromx[n] by adding
an infinite number of shifted replicas of
x[n], with each replica shifted by an integer
multiple of N sampling instants, and
observing the sum only for the interval
1 0 + =

=
N n mN n x n y
m
], [ ] [
1 0 N n
26
To apply
to finite-length sequences, we assume that
the samples outside the specified range are
zeros
Thus if x[n] is a length-M sequence with
, theny[n] =x[n] for
1 0 + =

=
N n mN n x n y
m
], [ ] [
N M 1 0 N n
27
If M >N, there is a time-domain aliasing of
samples of x[n] in generatingy[n], andx[n]
cannot be recovered fromy[n]
Example- Let
By sampling its DTFT at ,
and then applying a 4-point IDFT to
these samples, we arrive at the sequence y[n]
given by
} { ]} [ { 5 4 3 2 1 0 = n x
) (
j
e X 4 / 2 k
k
=
3 0 k
28
,
i.e.
{x[n]} cannot be recovered from{y[n]}
] [ ] [ ] [ ] [ 4 4 + + + = n x n x n x n y 3 0 n
} { ]} [ { 3 2 6 4 = n y
29
Numerical Computation of the Numerical Computation of the
DTFT Using the DFT DTFT Using the DFT
A practical approach to the numerical
computation of the DTFT of a finite-length
sequence
Let be the DTFT of a length-N
sequencex[n]
We wish to evaluate at a dense grid
of frequencies , ,
where M >>N:
) (
j
e X
) (
j
e X
M k
k
/ 2 = 1 0 M k
30
Define a new sequence
Then
= =

=

=

1
0
/ 2
1
0
] [ ] [ ) (
N
n
M kn j
N
n
n j j
e n x e n x e X
k k

=
1 , 0
1 0 ], [
] [
M n N
N n n x
n x
e
=

=
1
0
/ 2
] [ ) (
M
n
M kn j
e
j
e n x e X
k
6
31
Thus is essentially anM-point DFT
of the length-M sequence
The DFT can be computed very
efficiently using the FFT algorithm if M is
an integer power of 2
The functionfreqz employs this approach
to evaluate the frequency response at a
prescribed set of frequencies of a DTFT
expressed as a rational function in
) (
k
j
e X

] [k X
e
] [n x
e
] [k X
e
j
e
32
DFT Properties DFT Properties
Like the DTFT, the DFT also satisfies a
number of properties that are useful in
signal processing applications
Some of these properties are essentially
identical to those of the DTFT, while some
others are somewhat different
A summary of the DFT properties are given
in tables in the following slides
33
Table 5.1: Table 5.1: DFT Properties: DFT Properties:
Symmetry Relations Symmetry Relations
x[n] is acomplex sequence
34
Table 5.2: Table 5.2: DFT Properties: DFT Properties:
Symmetry Relations Symmetry Relations
x[n] is areal sequence
35
Table 5.3: Table 5.3: General Properties General Properties
of DFT of DFT
36
Circular Shift of a Sequence Circular Shift of a Sequence
This property is analogous to the time-
shifting property of the DTFT as given in
Table 3.4, but with a subtle difference
Consider length-N sequences defined for
Sample values of such sequences are equal
to zero for values of n <0 and N n
1 0 N n
7
37
If x[n] is such a sequence, then for any
arbitrary integer , the shifted sequence
is no longer defined for the range
We thus need to define another type of a
shift that will always keep the shifted
sequence in the range
] [ ] [
o
n n x n x =
1
o
n
1 0 N n
1 0 N n
38
The desired shift, called the circular shift,
is defined using a modulo operation:
For (right circular shift), the above
equation implies
] [ ] [
N o c
n n x n x =
0 >
o
n
=
], [
], [
] [
n n N x
n n x
n x
o
o
c
o
o
n n
N n n
<

0 for
1 for
39
Illustration of the concept of a circular shift
] [n x ] 1 [
6
n x
] 5 [
6
+ = n x ] 2 [
6
+ = n x
] 4 [
6
n x
40
As can be seen from the previous figure, a
right circular shift by is equivalent to a
left circular shift by sample periods
A circular shift by an integer number
greater thanN is equivalent to a circular
shift by
o
n
o
n N
o
n
N o
n
41
Circular Convolution Circular Convolution
This operation is analogous to linear
convolution, but with a subtle difference
Consider two length-N sequences, g[n] and
h[n], respectively
Their linear convolution results in a length-
sequence given by ) ( 1 2 N
2 2 0
1
0
=

=
N n m n h m g n y
N
m
L
], [ ] [ ] [
] [n y
L
42
In computing we have assumed that
both length-N sequences have been zero-
padded to extend their lengths to
The longer form of results from the
time-reversal of the sequenceh[n] and its
linear shift to the right
The first nonzero value of is
, and the last nonzero value
is
1 2 N
] [n y
L
] [n y
L
] [n y
L
] [ ] [ ] [ 0 0 0 h g y
L
=
] [ ] [ ] [ 1 1 2 2 = N h N g N y
L
8
43
To develop a convolution-like operation
resulting in a length-N sequence , we
need to define a circular time-reversal, and
then apply a circular time-shift
Resulting operation, called a circular
convolution, is defined by
1 0 ], [ ] [ ] [
1
0
=

=
N n m n h m g n y
N
m
N C
] [n y
C
44
Since the operation defined involves two
length-N sequences, it is often referred to as
an N-point circular convolution, denoted as
y[n] =g[n] h[n]
The circular convolution is commutative,
i.e.
g[n] h[n] =h[n] g[n]
N
N N
45
Example- Determine the 4-point circular
convolution of the two length-4sequences:
as sketched below
}, { ]} [ { 1 0 2 1 = n g } { ]} [ { 1 1 2 2 = n h

n
3 2 1 0
1
2
] [n g
n
3 2 1 0
1
2
] [n h
46
The result is a length-4sequence
given by
From the above we observe
] [n y
C
, ] [ ] [ ] [ ] [ ] [
3
0
4
= =
= m
C
m n h m g n h n g n y 4
3 0 n
=
=
3
0
4
] [ ] [ ] 0 [
m
C
m h m g y
] 1 [ ] 3 [ ] 2 [ ] 2 [ ] 3 [ ] 1 [ ] 0 [ ] 0 [ h g h g h g h g + + + =
6 2 1 1 0 1 2 2 1 = + + + = ) ( ) ( ) ( ) (
47
Likewise =
=
3
0
4
] 1 [ ] [ ] 1 [
m
C
m h m g y
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 2 3 3 2 0 1 1 0 h g h g h g h g + + + =
7 1 1 1 0 2 2 2 1 = + + + = ) ( ) ( ) ( ) (
=
=
3
0
4
] 2 [ ] [ ] 2 [
m
C
m h m g y
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 3 3 0 2 1 1 2 0 h g h g h g h g + + + =
6 1 1 2 0 2 2 1 1 = + + + = ) ( ) ( ) ( ) (
48
and
The circular convolution can also be
computed using a DFT-based approach as
indicated in Table 5.3
5 2 1 2 0 1 2 1 1 = + + + = ) ( ) ( ) ( ) (
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 0 3 1 2 2 1 3 0 h g h g h g h g + + + =
=
=
3
0
4
] 3 [ ] [ ] 3 [
m
C
m h m g y
] [n y
C
9
49
Example- Consider the two length-4
sequences repeated below for convenience:
The4-point DFT G[k] of g[n] is given by
n
3 2 1 0
1
2
] [n g
n
3 2 1 0
1
2
] [n h
4 2
1 0
/
] [ ] [ ] [
k j
e g g k G

+ =
4 6 4 4
3 2
/ /
] [ ] [
k j k j
e g e g

+ +
3 0 2 1
2 3 2
+ + =

k e e
k j k j
,
/ /
50
Therefore
Likewise,
, ] [ 2 1 2 1 2 = = G
, ] [ j j j G = + = 1 2 1 1
j j j G + = + = 1 2 1 3] [
, ] [ 4 1 2 1 0 = + + = G
4 2
1 0
/
] [ ] [ ] [
k j
e h h k H

+ =
4 6 4 4
3 2
/ /
] [ ] [
k j k j
e h e h

+ +
3 0 2 2
2 3 2
+ + + =

k e e e
k j k j k j
,
/ /
51
Hence,
The two4-point DFTscan also be
computed using the matrix relation given
earlier
, ] [ 6 1 1 2 2 0 = + + + = H
, ] [ 0 1 1 2 2 2 = + = H
, ] [ j j j H = + = 1 1 2 2 1
j j j H + = + = 1 1 2 2 3] [
52
j
j
j j
j j
g
g
g
g
G
G
G
G
1
2
1
4
1
0
2
1
1 1
1 1 1 1
1 1
1 1 1 1
3
2
1
0
3
2
1
0
4
] [
] [
] [
] [
] [
] [
] [
] [
D
j
j
j j
j j
h
h
h
h
H
H
H
H
1
0
1
6
1
1
2
2
1 1
1 1 1 1
1 1
1 1 1 1
3
2
1
0
3
2
1
0
4
] [
] [
] [
] [
] [
] [
] [
] [
D
is the 4-point DFT matrix
4
D
53
If denotes the4-point DFT of
then from Table 3.5 we observe
Thus
3 0 = k k H k G k Y
C
], [ ] [ ] [
] [k Y
C
] [n y
C
2
0
2
24
3 3
2 2
1 1
0 0
3
2
1
0
j
j
H G
H G
H G
H G
Y
Y
Y
Y
C
C
C
C
] [ ] [
] [ ] [
] [ ] [
] [ ] [
] [
] [
] [
] [
54
A 4-point IDFT of yields ] [k Y
C
] [
] [
] [
] [
*
] [
] [
] [
] [
3
2
1
0
4
1
3
2
1
0
4
C
C
C
C
C
C
C
C
Y
Y
Y
Y
y
y
y
y
D

=
5
6
7
6
2
0
2
24
1 1
1 1 1 1
1 1
1 1 1 1
4
1
j
j
j j
j j
10
55
Example- Now let us extended the two
length-4sequences to length 7by
appending each with three zero-valued
samples, i.e.

=
6 4 0
3 0
n
n n g
n g
e
,
], [
] [

=
6 4 0
3 0
n
n n h
n h
e
,
], [
] [
56
We next determine the7-point circular
convolution of and :
From the above
6 0 , ] [ ] [ ] [
6
0
7
=
=
n m n h m g n y
m
e e
] [n g
e
] [n h
e
] [ ] [ ] [ ] [ ] [ 6 1 0 0 0
e e e e
h g h g y + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 1 6 2 5 3 4 4 3
e e e e e e e e
h g h g h g h g + + + +
2 2 1 0 0 = = = ] [ ] [ h g
57
Continuing the process we arrive at
, ) ( ) ( ] [ ] [ ] [ ] [ ] [ 6 2 2 2 1 0 1 1 0 1 = + = + = h g h g y
] [ ] [ ] [ ] [ ] [ ] [ ] [ 0 2 1 1 2 0 2 h g h g h g y + + =
, ) ( ) ( ) ( 5 2 0 2 2 1 1 = + + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 0 3 1 2 2 1 3 0 3 h g h g h g h g y + + + =
, ) ( ) ( ) ( ) ( 5 2 1 2 0 1 2 1 1 = + + + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ 1 3 2 2 3 1 4 h g h g h g y + + =
, ) ( ) ( ) ( 4 2 1 1 0 1 2 = + + =
58
As can be seen from the above that y[n] is
precisely the sequence obtained by a
linear convolution of g[n] andh[n]
, ) ( ) ( ] [ ] [ ] [ ] [ ] [ 1 1 1 1 0 2 3 3 2 5 = + = + = h g h g y
1 1 1 3 3 6 = = = ) ( ] [ ] [ ] [ h g y
] [n y
L
] [n y
L
59
TheN-point circular convolution can be
written in matrix form as
Note: The elements of each diagonal of the
matrix are equal
Such a matrix is called a circulant matrix
N N
] 1 [
] 2 [
] 1 [
] 0 [
] 0 [ ] 3 [ ] 2 [ ] 1 [
] 3 [ ] 0 [ ] 1 [ ] 2 [
] 2 [ ] 1 [ ] 0 [ ] 1 [
] 1 [ ] 2 [ ] 1 [ ] 0 [
] 1 [
] 2 [
] 1 [
] 0 [
N g
g
g
g
h N h N h N h
h h h h
h N h h h
h N h N h h
N y
y
y
y
C
C
C
C
M
L
M O M M M
L
L
L
M
60
Tabular Method
We illustrate the method by an example
Consider the evaluation of
where {g[n]}and {h[n]}are length-4
sequences
First, the samples of the two sequences are
multiplied using the conventional
multiplication method as shown on the next
slide
] [ O ] [ ] [ * n g n h n y =
11
61
] [ ] [
] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ : ] [
:
3 0
2 1 2 0
1 2 1 1 1 0
0 3 0 2 0 1 0 0
3 2 1 0
3 2 1 0
3 2 1 0
h g
h g h g
h g h g h g
h g h g h g h g
h h h h n h
g g g g n g
n
] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [
] [ ] [
3 3 3 2 3 1
2 3 2 2
1 3
h g h g h g
h g h g
h g
4
4
4
5
4
6
The partial products generated in the 2
nd
, 3
rd
, and 4
th
rows
are circularly shifted to the left as indicated above
62
The modified table after circular shifting is
shown below
The samples of the sequence are
obtained by adding the 4partial products in
the column above of each sample
] [ ] [ ] [ ] [ : ] [
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [
] [ ] [ ] [ ] [ : ] [
] [ ] [ ] [ ] [ : ] [
:
3 2 1 0
3 0 3 3 3 2 3 1
2 1 2 0 2 3 2 2
1 2 1 1 1 0 1 3
0 3 0 2 0 1 0 0
3 2 1 0
3 2 1 0
3 2 1 0
c
y
c
y
c
y
c
y n
c
y
h g h g h g h g
h g h g h g h g
h g h g h g h g
h g h g h g h g
h h h h n h
g g g g n g
n
]} [ { n y
c
63
Thus
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 3 1 2 2 1 3 0 0 0 h g h g h g h g y
c
+ + + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 3 2 2 3 1 0 0 1 1 h g h g h g h g y
c
+ + + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 3 3 2 0 1 1 0 2 2 h g h g h g h g y
c
+ + + =
] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ 3 0 2 1 1 2 0 3 3 h g h g h g h g y
c
+ + + =

Digital Signal Processing Pre-Reqs PDF

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Digital Signal Processing Pre-Reqs PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

1

Anda mungkin juga menyukai