Transform
( )
2 1 3
, max N N N
[ ] [ ]
3
3
1
1 1 3
0
, 0 1
N
n
n
N
k
X k x n W k N
[ ] [ ]
3
3
1
2 2 3
0
, 0 1
N
n
n
N
k
X k x n W k N
[ ] [ ] [ ] [ ] k bX k aX n bx n ax
DFT
2 1 2 1
+ +
[ ] n x
1 1
N [ ] n x
2
2
N
[ ] [ ] k X N n n x
DFT
1 0 ,
[ ] [ ] k X n x
DFS
~
~
[ ]
( )
[ ]
2
DFS
j k N m
x n m X k e
%
%
( ) ( )
, 0 1
N
x n m n N
]
]
Figure
circular
shift
Ex. Circular Shift of a Sequence
8.6.2 Circular Shift of a Sequence
[ ] [ ]
( )
[ ]
1 1
2
, 0 1
DFT
j k N m
x n n N X k X k e
[ ] [ ] k X N n n x
DFT
1 0 ,
[ ] [ ] k X n x
DFS
~
~
[ ]
( )
[ ]
2
DFS
j k N m
x n m X k e
%
%
[ ] ( )
jw
e X n x
[ ]
( )
jw jwm
x n m X e e
[ ] ( ) ( ) [ ] ( ) ( )
DFS
N N
x n x n X k X k
] ]
] ]
%
%
[ ] ( ) ( ) [ ] ( ) ( )
1 1 1 1
DFS
N N
x n x n X k X k
] ]
] ]
%
%
Circular Shift of a Sequence
[ ]
( )
[ ]
1
2 j k N m
X k X k e
( ) ( )
N
x n m
]
]
[ ] ( ) ( ) [ ] ( ) ( )
DFS
N N
x n x n X k X k
] ]
] ]
%
%
[ ] ( ) ( ) [ ] ( ) ( )
1 1 1 1
DFS
N N
x n x n X k X k
] ]
] ]
%
%
[ ]
( )
( ) ( )
1
2
N
N
j k N m
X k X k e
 `
 `
. ,
. ,
]
%
( )
( ) ( )
2
N
j k N m
X k e
]
]
( )
[ ]
2 j k N m
X k e
%
[ ]
1
x n
%
[ ]
( )
[ ]
2
DFS
j k N m
x n m X k e
%
%
8.6.2 Circular Shift of a Sequence
[ ]
( )
[ ]
1
2 j k N m
X k X k e
( ) ( )
N
x n m
]
]
[ ]
1
x n
%
[ ]
( )
[ ]
2
DFS
j k N m
x n m X k e
%
%
[ ]
[ ] ( ) ( )
1
1
, 0 1
0,
N
x n x n m n N
x n
otherwise
'
%
( ) ( ) [ ]
( )
[ ] k X e N n m n x
m N k j
DFT
N
2
1 0 ,
[ ] [ ] k X N n n x
DFT
1 0 ,
8.6.3 Duality
[ ] [ ] k X n x
DFT
[ ] ( ) ( )
, 0 1
DFT
N
X n Nx k k N
]
]
[ ] [ ] k X n x
DFS
~
~
[ ] ( ) ( ) [ ] ( ) ( )
DFS
N N
x n x n X k X k
] ]
] ]
%
%
[ ] [ ] k x N n X
DFS
~
~
[ ]
[ ] ( ) ( )
1
, 0 1
0,
N
Nx k Nx k k N
X k
otherwise
'
%
[ ] [ ]
1
X k Nx k
%
%
[ ] [ ]
1
x n X n
%
%
[ ] [ ]
1
x n X n
Ex.8.9 The Duality Relationship for the
DFT
[ ]
, 0 1
ep
x n n N
[ ] ( ) ( ) ( ) ( )
{
*
1
2
e
N N
x n x n x n
] ]
+
] ]
%
[ ]
, 0 1
op
x n n N
8.6.4 Symmetry Properties
[ ] ( ) ( ) [ ] 1 0 ,
* *
N n k X n x
N
DFT
( ) ( ) [ ] [ ] 1 0 ,
* *
N n k X n x
DFT
N
[ ] ( ) ( ) ( ) ( )
{
*
1
2
o
N N
x n x n x n
] ]
] ]
%
[ ] ( ) ( ) [ ] ( ) ( )
DFS
N N
x n x n X k X k
] ]
] ]
%
%
[ ] [ ] [ ]
{
*
1
2
e
x n x n x n +
% % %
[ ] [ ] [ ]
{
*
1
2
o
x n x n x n
% % %
8.6.4 Symmetry Properties
[ ] ( ) ( ) [ ] ( ) ( ) [ ] { 1 0 ,
2
1
*
+ N n n x n x n x
N N ep
[ ] ( ) ( ) [ ] ( ) ( ) [ ] { 1 0 ,
2
1
*
N n n x n x n x
N N op
0 1, for n N
( ) ( ) ( ) ( )
,
N N
n N n n n
[ ] [ ] [ ]
{
*
1
, 0 1
2
ep
x n x n x N n n N +
[ ] [ ] [ ]
{
*
1
, 0 1
2
op
x n x n x N n n N
[ ] [ ] [ ]
{
[ ] [ ]
{
[ ] {
* *
1 1
0 0 0 0 Re 0
2 2
ep
x x x N x x x + +
[ ] [ ] [ ]
{
[ ] {
*
0
1
0 0 0 Im 0
2
p
x x x j x
8.6.4 Symmetry Properties
[ ] [ ] [ ] {
, 0 1
ep e e
x n x n x n N n N +
[ ] [ ] [ ] {
, 0 1
op o o
x n x n x n N n N +
[ ] [ ] [ ]
{
*
1
, 0 1
2
ep
x n x n x N n n N +
[ ] [ ] [ ]
{
*
1
, 0 1
2
op
x n x n x N n n N
[ ] [ ] [ ]
{
*
1
2
e
x n x n x n +
[ ] [ ] [ ]
{
*
1
2
o
x n x n x n
8.6.4 Symmetry Properties
[ ] { [ ] k X n x
ep
DFT
Re [ ] { [ ] k X n x j
op
DFT
Im
[ ] [ ] { k X n x
DFT
ep
Re
[ ] [ ] { k X j n x
DFT
op
Im
[ ] [ ] [ ]
e o
x n x n x n +
% % %
[ ] [ ] [ ] [ ]
, 0 1
e o
x n x n x n x n n N +
% % %
[ ] [ ]
, 0 1
ep e
x n x n n N
%
[ ] [ ]
, 0 1
op o
x n x n n N
%
[ ] [ ] [ ]
ep op
x n x n x n +
Symmetry Properties
8.6.5 Circular Convolution
% %
[ ] ( ) ( ) ( ) ( )
1
3 1 2
0
, 0 1
N
N N
m
x n x m x n m k N
] ]
] ]
[ ] ( ) ( )
1
1 2
0
, 0 1
N
N
m
x m x n m k N
]
]
( ) ( )
, 0 1
N
since m m for k N
[ ] [ ]
1 2
X k X k
% %
[ ] [ ]
1
1 2
0
N
m
x m x n m
% %
DFS
]
]
[ ] [ ] [ ] [ ] 1 0 ,
2 1 3 3
N k k X k X k X n x
DFT
[ ] [ ]
2 1
x n N x n
[ ] ( ) ( )
1
2 1
0
N
N
m
x m x n m
]
]
1
0
2 1 3
1
N
l
N
l k X l X
N
k X
[ ] [ ] [ ] [ ] k X N k X
N
n x n x
DFT
2 1 2 1
1
'
<
[ ] [ ]
1 0
x n n n
2
2 0
[ ] [ 1]
[(( 1)) ], 1
N
x n N n
x n n n N
<
[ ] [ ] ( ) ( )
1
3 1 2
0
N
N
m
x n x m x n m
]
]
'
<
[ ] [ ]
1 0
x n n n
0
[ ] [ 1]
[(( 1)) ], 1
N
x n N n
x n n n N
<
[ ]
0
1
kn
N
W k X
[ ] [ ] [ ]
[ ] k X W
k X k X k X
kn
N 2
2 1 3
0
'
otherwise
L n
n x n x
, 0
1 0 , 1
2 1
6 L N
[ ] [ ]
'
otherwise
k N
W k X k X
N
n
kn
N
, 0
0 ,
1
0
2 1
[ ] [ ] [ ]
'
otherwise
k N
k X k X k X
, 0
0 ,
2
2 1 3
[ ]
1
3
0
1
N
kn
N
k
X k W
N
, 0 1
0,
N n L
otherwise
'
[ ] [ ] ( ) ( )
1
3 1 2
0
N
N
m
x n x m x n m
]
]
[ ] [ ] [ ]
2
2 1 3
1
1
,
`
.

k
N
Lk
N
W
W
k X k X k X
[ ] [ ] ( ) ( )
1
3 1 2
0
N
N
m
x n x m x n m
]
]
procedure
Compute the Npoint DFT x1[k] and x2[k] of the two sequence x1(n) and x2(n)
Compute x3[k]=x1[k]*x2[k] for
8.7 Linear Convolution using the
Discrete Fourier Transform
m
m n x m x n x
2 1 3
[ ]
3
0, x n
2 0 + P L n
for
is maximum length of
( ) 1 + p L
[ ] n x
3
L
[ ] n x
2
[ ] n x
1
length
L P
8.7.2 Circular Convolution as
Linear Convolution with Aliasing
and linear convolution
[ ] [ ]
1 2
* x n x n
circular convolution corresponding to DFTs:
[ ] [ ]
1 2
X k X k
[ ] [ ]
1 2
x n N x n
, Whether they are same?
depends on the length of the DFT in relation to the
length of and
[ ]
2
x n
[ ]
1
x n
8.7 Linear Convolution using the
Discrete Fourier Transform
1. Compute the Npoint DFT and of the
two sequence and
[ ] k X
1
[ ] k X
2
[ ] n x
1
[ ] n x
2
2. Compute for
[ ] [ ] [ ] k X k X k X
2 1 3
1 0 N k
3. Compute as the inverse DFT of
[ ] k X
3
[ ] [ ] [ ] n x N n x n x
2 1 3
Implement a convolution of two sequences by the
following procedure:
Review
8.7.2 Circular Convolution as
Linear Convolution with Aliasing
[ ]
[ ]
, 0 1
0,
p
x n n N
x n
otherwise
'
( )
( )
( )
2
, 0 1
0,
j k N
X e k N
X k
otherwise
'
( )
( )
( )
2 j k N
X k X e
%
[ ] [ ]
r
x n x n rN
%
DFS
m
m n x m x n x
2 1 3
Linear convolution:
Define a DFT
( ) ( )
1 2
X k X k
0 1 k N
'
otherwise
N n rN n x
n x
r
p
, 0
1 0 ,
3
3
[ ] [ ] [ ] n x N n x n x
p 2 1 3
8.7.2 Circular Convolution as
Linear Convolution with Aliasing
( ) ( ) ( ) 1 0
2 1 3
N k k X k X k X
[ ] [ ] [ ] n x N n x n x
p 2 1 3
[ ]
[ ]
'
otherwise
N n rN n x
n x
r
p
, 0
1 0 ,
3
3
And
From
m
m n x m x n x
2 1 3
Linear convolution:
8.7.2 Circular Convolution as
Linear Convolution with Aliasing
m
m n x m x n x
2 1 3
D
F
T
D
T
F
T
8.7.2 Circular Convolution as
Linear Convolution with Aliasing
m
m n x m x n x
2 1 3
D
F
T
D
T
F
T
linear convolution
6 points shift right of
the linear convolution
6 points shift left of
the linear convolution
6 points circular convolution=
linear convolution with aliasing
12 points circular convolution
= linear convolution
Ex. 8.12 Circular
Convolution as Linear
Convolution with Aliasing.
[ ] [ ]
1 2
x n x n
[ ] [ ]
3 3 p
r
x n x n rN
N=6
N=12
[ ]
3
x n N +
[ ]
3
x n N
[ ]
3
x n
Which points of
Circular Convolution
equal that of Linear
Convolution when
Aliasing?
[ ] [ ]
3 3 p
r
x n x n rN
0 1 n N
[ ] n x
1
[ ] n x
2
Consider of
length L and of
length P, where P < L
1 L P +
N L
Linear
Convolution
Circular Convolution
1 L P + 1 P
Fig.8.19
Fig.8.20
4 , 8 P L
8 N L
1 11 N L P +
8.7.3 Implementing Linear Time
Invariant Systems Using the
DFT
[ ]
[ ]
, 0 1
0,
r
x n rL n L
x n
otherwise
+
'
overlapadd method
(1) segment into sections of length
L;
) (n x
(2) fill 0 into and some section of
, then do L+P1 points FFT ;
) (n h
) (n x
2 ,... 0 )} ( ) ( { ) ( + P L n k X k H IFFT n y
) (n y
(3) calculate
[ ] [ ] [ ] [ ]
0
r
r
y n x n h n y n rL
P1 points
(4)add the points n=0P2 in
y[n] to the last P1 points in the
former section y[n] the output
for this section is the points
n=0L1
( ) { ( ) ( )} y n IFFT H k X k
) (n y
(3) calculate
(1) segment into sections of length
L;
) (n x
(2) fill 0 into and some section of
, then do L+P1 points FFT
) (n h
) (n x
0,..., 2 P n L +
[ ] [ ]
0
r
r
x n x n rL
overlapadd method
[ ] [ ] [ ]
r r
where y n x n h n
L=16
4 , 8 P L
8 N L
8.7.2 Circular Convolution as Linear
Convolution with Aliasing
overlapsave method
P1 points
(4) the output for this section is
LP+1 points of y[n]
n=P1,L1
(1) segment into sections of length
L, overlap P1 points;
) (n x
(2) fill 0 into and some section of
, then do L points FFT
) (n h
) (n x
( ) { ( ) ( )} y n IFFT H k X k
) (n y
(3) calculate
0,.. 1 ., n L
L=25