Scribd
Upload
50 views

5. FFT

Uploaded by

durgaindira.n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
50 views

5. FFT

Uploaded by

durgaindira.n
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 35

Fast Fourier Transform

(FFT)

FFT is an efficient algorithm to


compute DFT

12/19/24
12/19/24
DFT
Analysis Equation

Synthesis Equation

12/19/24
Limitations of DFT

12/19/24
Limitations of DFT

12/19/24
Comparison of Computational Complexity for the Direct
Computation of the DFT versus the FFT Algorithm:

12/19/24
Radix-2 FFT
By adopting a divide and conquer approach, a
computationally efficient algorithm for the DFT
can be developed. This approach depends on
the decomposition of an N-point DFT into
successively smaller size DFTs.

If N is factored as, N r1r2 . . . . rL

where r1 r2 . . . . rL r ,

then N r L .
12/19/24
r

Radix-2 FFT . . .
• Hence the DFT will be the size of ‘r ’, where ‘r ’
is called radix of FFT algorithm.
• Here we describe or study the radix-2 FFT
algorithms.
• The radix-2 FFT algorithm is divided into two
types: Radix-2 DIT FFT and DIF FFT
algorithms.

12/19/24
Decimation In Time (DIT) FFT algorithm
The decimation-in-time (DIT) algorithm
decomposes the DFT by sequentially splitting
input samples x[n] in the time domain into
sets of smaller and smaller sub sequences
and then forms a weighted combinations of
the DFTs of these sub-sequences.
Let us assume that x[n] represents a sequence
of ‘N’, where ‘N’ is an integer power of 2, i.e.,
.

12/19/24
N 2 L
12/19/24
12/19/24
12/19/24
For N 8
X [0] G[0]  W80 H [0]
X [1] G[1]  W81H [1]
X [2] G[2]  W82 H [2]
X [3] G[3]  W83 H [3]
X [4] G[4  82 ]  W84 H [4  82 ] G[8]  W84 H [8] G[0]  W84 H [0]
X [5] G[5  82 ]  W85 H [5  82 ] G[9]  W85 H [9] G[1]  W85 H [1]
X [6] G[6  82 ]  W86 H [6  82 ] G[10]  W86 H [10] G[0]  W86 H [0]
X [7] G[7  82 ]  W87 H [7  82 ] G[11]  W87 H [11] G[3]  W87 H [3]

12/19/24
Radix-2 FFT

12/19/24
Fig.1 shows the flow graph of the decimation-in-time
decomposition of an 8-point DFT computation into two
4-point DFT computations.

Fig.1. Flow graph of first stage Decimation-In-Time FFT


algorithm
12/19/24
for N = 8
Results of Stage-1

12/19/24
12/19/24
12/19/24
The corresponding flow graph of a 2-point DFT is
shown in Fig.2(a).

Fig.2. (a) Flow graph of a 2-point DFT (b) Modified flow


graph

12/19/24
12/19/24
The modified N-point FFT for N=8 is shown in Fig.3.

W80 1, W81 0.707  j 0.707, W82  j W83  0.707  j 0.707


12/19/24
12/19/24 Dr. M. Venu Gopala Rao
Decimation In Frequency (DIF) FFT
algorithm
The decimation-in-frequency FFT decomposes the DFT
by recursively splitting the sequence elements
X [ K ] in the frequency domain into sets of smaller
and smaller sub-sequences.

12/19/24
12/19/24
N 1
2
X [2r ]    x[ n ]  (  1)
2 
2 r x[ n  N ] W 2 rn
N
n 0
N 1
2
   x[n]  x[n  N2 ] W Nrn/ 2 , 0 r  N  1
2 (3)
n 0
N 1
2
  rn , where g[ n]  x[ n]  x[ n  N ]
g[ n] W N /2 2
n 0
N 1
2
X [2r  1]    x[n]  ( 1) 2r 1 x[n  N2 ]W N(2r 1)n
n 0
N 1
2
   x[n]  x[n  N2 ]W Nn W Nrn/ 2 , 0 r  N  1
2
(4)
n 0
N 1
2
  h[ n] W Nn W Nrn/ 2 , where h[ n]  x[ n]  x[ n  N ]
2
12/19/24 n 0
g[0]  x[0]  x[4] h[0]  x[0]  x[4]
For N = 8
g[1]  x[1]  x[5] h[1]  x[1]  x[5]
g[2]  x[2]  x[6] h[2]  x[2]  x[6]
g[3]  x[3]  x[7] h[3]  x[3]  x[7]
The flow graph of the first stage of an 8-point DIF FFT
computation scheme shown in Fig.1.

12/19/24 Dr. M. Venu Gopala Rao


N 1 N 1 N 1
2 4 2
X [2r ]   g[n] W N2rn   g[n] W N2rn   g[n] W N2rn , 0 r  N  1
2
n 0 n 0 n N
4
N 1 N 1
4 4 2r ( n  N )
  g[n] W N2rn   g[ n  N] WN 4 , 0 r  N  1
n 0 n 0
4 2
(5)
N 1 N 1
4 2r N 4
  g[n] W N2rn  W N 4
 g[ n  N ] W N2rn , 0 r  N  1
4 2
n 0 n 0

When r 2l (even)


N 1
4
X [4l ]   g[n]  g[n  N4 ]W Nln/ 4 , 0 l  N  1
4 (6)
n 0
N 1
4
  A[n] W Nln/ 4 , where A[n] g[n]  g[n  N ], 0 l  N  1
4 4
n 0
12/19/24
12/19/24
The flow chart is shown in Fig2.

Fig.2. The flow chart for computation of DIF-FFT


algorithm for N=8
12/19/24
Radix-2 D.I.F.F.F.T algorithm

8-point DFT

12/19/24
12/19/24 Dr. M. Venu Gopala Rao
Inverse FFT

12/19/24
Inverse FFT using DITFFT algorithm

12/19/24
Inverse FFT using DIFFFT algorithm

12/19/24
End

12/19/24

You might also like