FFT

Decimation in Frequency

Spring
2009
Ammar Abu-Hudrouss -Islamic
University Gaza

Decimation in Frequency
The discrete Fourier transform can be found using
N 1

X ( k ) x ( n )W Nkn

k 0,1, 2 ,....., N 1

n 0

Where N = 2, 4, 8, 16, and

WN e j 2 / N
X (k ) can be expressed as
N / 2 1

X (k )

N 1

x ( n )W Nkn

n 0

N / 2 1

x ( n )W Nkn

n 0

x (n N / 2 )W

N / 2 1

x ( n )W

n 0

kn
N

W NkN / 2

x ( n N / 2 )W
n 0

Digital Signal Processing

Slide

k ( n N / 2)
N

n 0

N / 2 1

X (k )

kn
N

n N /2

N / 2 1

X (k )

x ( n )W

kn
N

Decimation in Frequency
But

W N( N / 2 ) k e 2k ( N / 2 ) / N e k ( 1) k
Then
N / 2 1

X (k )

N / 2 1

x (n)WNkn (1) k

n 0

x(n N / 2)W

kn
N

n 0

If k = 2m or an even number
N / 2 1

x ( n ) x ( n N / 2 ) W

X (2m)

2 mn
N

n 0

N / 2 1

X (2m)

a(n)W

mn
N /2

n 0

Digital Signal Processing

Slide

Decimation in Frequency
Noting That

WN2 mn e j 2 mn2 / N e j 2mn /( N / 2 ) WNmn/ 2

a ( n ) x ( n ) x ( n N / 2)
Then X(2m) is N/2-point DFT for a(n)
If k = 2m+1 (odd number) and using the same method
N / 2 1

X ( 2 m 1)

x ( n ) x ( n N / 2) W

n
N

W N2 mn

n0

N / 2 1

X (2m 1)

b(n)W W
n
N

mn
N/2

n 0

b(n) x (n) x (n N / 2)
n
X(2m+1) is N/2-point DFT for b(n)WN

Digital Signal Processing

Slide

Decimation in Frequency
a(0)

x(0)
x(1)

a(3)

x(3)

b(0)

x(4)

b(1)

-1

b(3)

X(4)

X(1)

1
8

N/2-point
DFT

-1

b(2)

X(2)

X(6)
0
8

-1

x(5)

x(7)

N/2-point
DFT

a(2)

x(2)

x(6)

X(0)

a(1)

2
8

3
8

X(3)
X(5)
X(7)

-1

Digital Signal Processing

Slide

Decimation in Frequency

x(0)
x(1)
0
8

1
4

2
8

-1
W W

x(3)
W

x(4)

0
8

N/4 point
DFT

X(2)

N/4 point
DFT

X(1)

N/4 point
DFT

X(3)

-1

-1

W81

x(5)

-1
W

2
8

W
-1

0
4

0
8

W W

-1

-1

x(7)

W41 W82

3
8

-1

Digital Signal Processing

Slide

X(0)

W W

x(2)

x(6)

0
4

N/4 point
DFT

X(4)

X(6)

X(5)

X(7)

Decimation in Frequency
x(0)

X(0)
W80

x(1)

-1
0
8

x(2)

2
8

0
8

x(4)

0
8

-1

-1

1
8

0
8

-1
W

2
8

x(7)
-1

0
8

3
8

X(3)

2
8

0
8

-1

X(5)

-1

-1

-1

X(6)
X(1)

-1

x(5)
x(6)

X(2)

-1

x(3)

X(4)

X(7)

-1

Digital Signal Processing

Slide

Decimation in Frequency
Using the previous algorithm , the complex multiplications
needed is only 12. While using the normal DFT would
require 64 complex multiplications
In general
Complex multiplication of DFT is: N2
Complex multiplication of FFT is (N/2) log2(N)
If N = 1024
Complex multiplication of DFT is: 1,048,576
Complex multiplication of FFT is: 5,120

Digital Signal Processing

Slide

Decimation in Frequency
Index mapping for Fast Fourier Transform

Input Data
index
n

Index Bits

Reversal
Bits

Output data
index
k

000

000

001

100

010

010

011

110

100

001

101

101

110

011

111

111

Digital Signal Processing

Slide

Decimation in Frequency
Example Given a sequence x(n) where x(0) = 1, x(1) = 2, x(2) = 3,
x(3) = 4 and x(n) = 0 elsewhere ,find DFT for the first four points
solution
4

x(0) =1

x(1) =2

W 1

-1

-2

x(2) =3
x(3) =4

0
4

-2
-1

-1

Digital Signal Processing

Slide

W40 1

W41 j

X(0)

-2

X(2)

-2+2j

W40 1

-1

10

-2-2j

X(1)
X(3)

Inverse Fourier Transform

The inverse discrete Fourier transform can be found using
x (n)

1
N

N 1

X ( k )W

kn
N

n 0,1, 2 ,....., N 1

k 0

Which can be expressed as

x (n)

1
N

N 1

X (k )W

kn
N

n 0,1, 2 ,....., N 1

k 0

where

W Nkn W N kn

is called the twiddled factor

We can see that the difference between the inverse discrete

Fourier and forward Fourier transform is the twiddled factor
and the division by 1/N

Digital Signal Processing

Slide

Inverse Fourier Transform

1/N

X(0)
~
0
8

X(1)

-1

-1

-1

-1

-1

1/N x(1)
W80
~

-1

-1

-1
0
8

-1

-1

Slide

x(3)

~
2
8

3
8

1/N x(5)
1/N

W80

W82

-1

1/N x(6)

Digital Signal Processing

-1

W81

X(5)

X(7)

0
8

W80

X(4)

x(2)

W
~

X(6)

~
2
8

X(3)

1/N x(4)
1/N

W80

X(2)

x(0)

1/N x(7)

Inverse Fourier Transform

Example Given a sequence X(n) given in the previous example.
Find the IFFT using decimation in frequency method
solution
8

X(0) =10

-4

X(1) = -2+2j

12

X(2) =-2
X(3) = -2-2j

4
~

-1

j4

-1

0
4

W 1
0
4

W 1
~

0
4

1
4

W 1

W j

-1

Slide

x(0) = 1

12

1/4

x(2) = 3

1/4

x(1) = 2

16

1/4

x(3) = 4

-1

Digital Signal Processing

1/4