19ECE211 Signal Processing

Module 1

Prepared by,
Dr.J.Aravinth

31-Jan-24 Dr.J.Aravinth 1
Outline

• Linear filtering using DFT

• Linear and Circular convolution
• Efficient computation using of DFT Algorithms

31-Jan-24 Dr.J.Aravinth 2
 TIME-FREQUENCY ANALYSIS OF FOURIER SERIES AND
FOURIER TRANSFORMS
Time Frequency
Domain Domain

Continuous-Time Continuous & Discrete &

Fourier Series (CTFS) Periodic Aperiodic
Continuous-Time Continuous &
Continuous
Fourier Transform (CTFT) Periodic
Discrete & Discrete &
Discrete Fourier Series (DFS)
Periodic Aperiodic
Discrete-Time Continuous &
Discrete
Fourier Transform (DTFT) Periodic
Discrete & Discrete &
Discrete
31-Jan-24 Fourier Transform (DFT)
Dr.J.Aravinth 3
2
Periodic Periodic
 DFT Vs DTFT
• is the FT transform of DT-signal x(n).The range of ‗w‘ is from 0
to 2π (or) - π to π.Hence it is not possible to compute on digital
computer.
• Because in expression the range of summation is from - to .
• If we make the range is finite, then it is possible to do the calculations on
computer

31-Jan-24 Dr.J.Aravinth 4
Discrete-Time Fourier Transform
• Many sequences can be expressed as a weighted sum of complex
exponentials as

( ) =  xne

j − jn
Xe (forward transform)
n = −

• Where the weighting is determined as

xn =
1 
2 − 
X (e j
)e jn
d (inverse transform)

– X (e ) is the Fourier spectrum of the sequence x[n]

j

– The phase wraps at 2 hence is not uniquely specified

– The frequency response of a LTI system is the DTFT of the impulse
response
( ) =  hk e 1 
( )

He j − jk
and hn =  H e j
e jn
d
k = − 2 − 

31-Jan-24 Dr.J.Aravinth 5
 DISCRETE FOURIER TRANSFORM (DFT)
❑ DFT-Defines a relationship between a signal in the Time-Domain and its
representation in Frequency Domain.
❑ The DFS provides a mechanism for numerically computing the DTFT. It also alerted
us to a potential problem of aliasing in the time domain.
❑ Sampling of the DTFT result in a periodic sequence x˜(n). But most of the signals in
practice are not periodic. They are likely to be of finite duration.

Possibility of numerically computable Fourier representation

for such signals?
❑ From the theoretical conclusion, define a periodic signal whose primary shape is that
of the finite-duration signal and then using the DFS on this periodic signal.
❑ In Practical, define a new transform called the discrete Fourier transform (DFT),
which is the primary period of the DFS.
❑ This DFT is the ultimate numerically computable Fourier transform for arbitrary
finite-duration
31-Jan-24 sequences. Dr.J.Aravinth 6
 DISCRETE FOURIER TRANSFORM (DFT)
❑ The discrete-time Fourier transform provided the frequency domain (ω)
representation for absolutely summable sequences. The transforms (DTFT and
ZT) have two features in common.
1. The Transforms are defined for infinite-length sequences. (ie. Numerical
computation viewpoint or from MatLab‘s viewpoint)
2. They are functions of continuous variables (ω or z).

❑ These two features are troublesome because one has to evaluate infinite sums at
uncountably infinite frequencies.

❑ To compute these transforms using software(MatLab, Scilab) the sequence have

to truncated and then evaluate the expressions at finitely many
points.(Evaluations approximated to the exact calculations).

❑ Therefore the DTFT and the ZT are not numerically computable transforms.

31-Jan-24 Dr.J.Aravinth 7
 DISCRETE FOURIER TRANSFORM (DFT)
❑ Numerically computable transform –
by sampling the DTFT in the frequency domain (or the z-transform on the unit
circle).

❑ This transform is developed by analyzing periodic sequences.

❑ In Fourier analysis, a periodic function (or sequence) represented by a linear

combination of harmonically related complex exponentials (which is a form of
sampling). This gives us the discrete Fourier series (DFS) representation.

❑ The extension of DFS to finite-duration sequences, which leads to a new

transform, called the discrete Fourier transform (DFT).

❑ The DFT avoids the TWO PROBLEMS mentioned and is a numerically

computable transform that is suitable for computer implementation.
31-Jan-24 Dr.J.Aravinth 8
 DISCRETE FOURIER TRANSFORM PAIRS

Analysis Equation

Synthesis Equation

31-Jan-24 Dr.J.Aravinth 12
 DISCRETE FOURIER TRANSFORM (DFT)

❑ The DFT provides uniformly spaced samples of the Discrete-Time

Fourier Transform (DTFT).

Discrete Fourier transform (DFT) of a discrete-time signal x[n] with finite

extent n [0, N-1]

Twiddle factor:

31-Jan-24 Dr.J.Aravinth 13
 DISCRETE FOURIER TRANSFORM (DFT)

With these definitions, the N-point DFT

may be expressed in matrix form as,

The inverse DFT

Or

I is an NxN identity matrix.

31-Jan-24 N Dr.J.Aravinth 14
 DISCRETE FOURIER TRANSFORM (DFT)

Let N-point vector x Nof the signal sequence x(n) , n = 0, 1, 2, ….(N-1) and
N-point vector XN of frequency samples and NxN matrix WNas

31-Jan-24 Dr.J.Aravinth 15
 EXAMPLE
Compute the DFT of the 4-point sequence

then

31-Jan-24 Dr.J.Aravinth 16
 EXAMPLE
1. Given x(n) = {1, 1, 0, 0}, the DFT of this 4-point sequence can be computed
using the matrix formulation as

31-Jan-24 Dr.J.Aravinth 17
 DFT-Properties

1. Periodicity:

If x(n) and X(k) are N-point DFT pairs

2. Symmetry:

X(k) = X*(-k)N = X*(N-k)

3. Linearity:

31-Jan-24 Dr.J.Aravinth 18
 DFT-Properties

4. Circular Shift of a sequence:

• The desired shift, called the circular shift, is defined using a modulo
operation:

• For n 0>0 (right circular shift), the above equation implies

31-Jan-24 Dr.J.Aravinth 19
 DFT-Properties
Concept of a circular shift

31-Jan-24 Dr.J.Aravinth 20
 DFT-Properties

5. Multiplication:

6. Parseval‘s relation:

Energy spectrum
Power spectrum

31-Jan-24 Dr.J.Aravinth 21
 DFT-Properties

7. Circular Convolution:

31-Jan-24 Dr.J.Aravinth 22
 DFT-Properties

For the two finite-duration sequences of

length N, x1(n) and x2(n)

Convolution Sum called

Circular Convolution

31-Jan-24 Dr.J.Aravinth 23
 Example:

31-Jan-24 Dr.J.Aravinth 24
 Example:
Perform the circular convolution of the following 2 sequence:

m=0 m=2

= 14
= 14
m=1 m=3

= 16
= 16

31-Jan-24 Dr.J.Aravinth 25
 Continued…

31-Jan-24 Dr.J.Aravinth 26
 Continued…

31-Jan-24 Dr.J.Aravinth 27
 Continued…

31-Jan-24 Dr.J.Aravinth 28
 Linear filtering methods based
on the DFT
➢DFT provides a discrete frequency
representation of a finite duration sequence
in the frequency domain.
➢Computational tool for linear system
analysis, especially for linear filtering.
➢DFT can be used to perform linear filtering
in the frequency domain.

31-Jan-24 Dr.J.Aravinth 29
 Linear filtering methods based on the
DFT
1.Use of the DFT in Linear Filtering
2.Filtering of long data sequence
• Overlap-save method
• Overlap-add method

31-Jan-24 Dr.J.Aravinth 30
 Use of the DFT in Linear Filtering

• Our objective is to determine the output of a

linear filter to a given input sequence.

Linear Filter
h(n) y ( n) Output or
Input x(n) Response

• Linear Convolution
• By using DFT and IDFT
• Circular Convolution
31-Jan-24 Dr.J.Aravinth 31
 Linear Convolution
• A sequence x(n) of length L filtered by an
FIR filter h(n) of length M
M −1
y( n ) =  h( n ) x( n − k )
k =0

Example x(n) =[1, 2, 2, 1] , h(n) =[1, 2, 3]

L=4,M=3
x(n)=[1 ,2,2,1] ,h(n)=[1, 2,3]

The length of sequence y(n) is N=L+M-1=6

y(n)={1,4,9,11,8,3}
31-Jan-24 Dr.J.Aravinth 32
31-Jan-24 Dr.J.Aravinth 33
 DFT and IDFT
• Therefore ,a DFT of size N  L+M-1 is required
to represent y(n) in the frequency domain.
• Using the DFT notation
Y (k ) = H (k ) X (k ), k = 0,1,.........( N − 1)

X(k) and H(k) are the DFT (with zero padding)

of x(n) and h(n), respectively.
• Performing the inverse DFT of Y(K)
y(n) = IDFT (Y (k )

31-Jan-24 Dr.J.Aravinth 34
• L=4,M=3,N=L+M-1=6
• Eight point DFT of x(n) is
• X(K)= {6,1.707 - 4.121j, -1 – j,0.2929 - j0.121, 0,0.292 + j0.121, -1+j,1.707 + j4.121}
• Eight point DFT of h(n) is
• H(K)={6,2.414 - j4.414, -2-j2,-0.4142+ j1.585,2,-0.414 -j 1.585,-2+j2,2.414 + j4.4142}
• Y(k)=X(k).H(K)={36,-14.0-j17.48,j4,0.07+j0.515,0,0.07-j0.515,-j4,-14.07+j17.48}
• Eight point IDFT of Y(K) is

31-Jan-24 Dr.J.Aravinth 35
 Circular Convolution
• Example x(n) =[1, 2, 2, 1] , h(n) =[1, 2, 3]
• L=4,M=3
• The length of sequence y(n) is N=L+M-1=6
• Example x(n)=[1 ,2,2,1,0 ,0] ,h(n)=[1, 2,3,0,0 0]

• y(n)={1,4,9,11,8,3}

31-Jan-24 Dr.J.Aravinth 36
 Filtering of Long Data Sequences
• In practical applications involving linear
filtering of signals, the input sequence x(n) is
often a very long sequence.
• Real-time signal processing applications
concerned with signal monitoring and
analysis.
• Overlap-save method
• Overlap-add method

31-Jan-24 Dr.J.Aravinth 37
 Filtering of Long Data Sequences
• When the DFT is used to implement linear
filtering, a signal is processed in blocks. Due
to the real-time requirement (low delay) and
the limitation of physical memory, the size of
the block can not be arbitrarily large.
• The length of the FIR filter is M and the length
of on block of data is L (L>M)
• Each time a block of data of length L+M-1 is
filtered by using the DFT method.
31-Jan-24 Dr.J.Aravinth 38
31-Jan-24 Dr.J.Aravinth 39
31-Jan-24 Dr.J.Aravinth 40
 Overlap-Add Method

31-Jan-24 Dr.J.Aravinth 41
31-Jan-24 Dr.J.Aravinth 42
31-Jan-24 Dr.J.Aravinth 43
 x(n)={3,-1,0,1,3,2,0,1,2,1},h(n)={1,1,1}
• Overlap-Save Method: N −1

• Let L=5,M=3,N=L+M-1=7 x3 (m) = x1 (n)  x 2 (n) =  n =0

x1 (n)x 2 ((n − m)) N

x1(n)={0,0,3,-1,0,1}, x3(n)={0,1,2,1,0,0}
x2(n)={0,1,3,2,0,1}
N −1
Perform y1(n)=x1(n) N h(n) y1 (m) = x1 (n)  h(n) =  x1(n)h((n − m))N
n =0
N −1
y2(n)=x2(n) N h(n) y2 (m) = x 2 (n)  h(n) =  x (n)h((n − m))
n =0
2 N

N −1
y3(n)=x3(n) N h(n) y3 (m) = x3 (n)  h(n) =  x3 (n)h((n − m)) N
31-Jan-24 Dr.J.Aravinth 44
n =0
 x(n)={3,-1,0,1,3,2,0,1,2,1},h(n)={1,1,1}
• Overlap-Add Method: N −1

• Let L=5,M=3,N=L+M-1=7 x3 (m) = x1 (n)  x 2 (n) =  n =0

x1 (n)x 2 ((n − m)) N

x1(n)={3,-1,0,1,0,0}, x3(n)={2,1,0,0,0,0}
x2(n)={3,2,0,1,0,0}
N −1
Perform y1(n)=x1(n) N h(n) y1 (m) = x1 (n)  h(n) =  x1(n)h((n − m))N
n =0
N −1
y2(n)=x2(n) N h(n) y2 (m) = x 2 (n)  h(n) =  x (n)h((n − m))
n =0
2 N

N −1
y3(n)=x3(n) N h(n) y3 (m) = x3 (n)  h(n) =  x3 (n)h((n − m)) N
31-Jan-24 Dr.J.Aravinth 45
n =0
• x(n) ⎯⎯
DFT
→ X(k) ⎯⎯
IDFT
→ x(n)
• Properties of DFT
Circular Convolution N −1
x3 (m) = x1 (n)  x 2 (n) = 
n =0
x1 (n)x 2 ((n − m)) N ⎯⎯
DFT
→ X3 (K) = X1 (K)X 2 (K)
• Linear Filtering Methods Based on the DFT
1.Use of the DFT in Linear Filtering
• By using DFT and IDFT
x(n) ⎯⎯
→ X(k) → H(K)
H(K) → Y(K) = X(K).H(K) ⎯⎯→ y(n
DFT IDFT

• Linear Convolution
• Circular Convolution
2.Filtering of Long Data Sequence
• Overlap-save method
• Overlap-add method
31-Jan-24 Dr.J.Aravinth 46
 COMPUTATIONAL COMPLEXITY

• How ever its implementation in so easy for

An implementation of DFT involves TWO basic functional blocks,

1. ADDITION
2. MULTIPLICATION
DFT-Complex Function DFT-Real Function
Length of the No.of Real MULs
No.of Complex No.of Complex No.of Real ADDs
Sequence (N) 4(N*N)
MULs (N*N) ADDs (N*N-1) 4(N*N-1)

2 4 2 16 8
3 9 6 36 24
4 16 12 64 48
: : : : :
8 64 56 256 224
: : : : :
16 256 240 1024 960
31-Jan-24 Dr.J.Aravinth 47
: : : : :
 LIMITATIONS OF DFT

❑ When DFTs are used to process a CT-Signal by sampling, several

potential error sources may be important

– Aliasing, Spectral Leakage

– Under suitable restrictions, the DFT closely approximates the

spectrum of CT-signal at a Discrete set of frequencies.

31-Jan-24 Dr.J.Aravinth 48
 Why do we need FFT?

❑ DFT properties and its use in system analysis in the numerical computation
of long sequences is prohibitively time-consuming.

❑ The importance of DFT and IDFT in practical applications is due to a large

extent on the existences of computationally efficient algorithms,
collectively known as FFT algorithms.

❑ Therefore several algorithms have been developed to efficiently compute

the DFT. These are collectively called Fast Fourier transform (or FFT)
algorithms.

31-Jan-24 Dr.J.Aravinth 49
 FAST FOURIER TRANSFORM (FFT)

❑ FFT is introduced by Cooley-Tukey, (almost a half century ago) is playing

historically sustained significant role in the development of DSP.

❑ It is widely used ―fast algorithm‖ to solve many engineering challenges,

designing filters, performing spectral analysis, estimation, noise cancellation
and benchmark testing devices and systems, etc.

❑ It is also very readily useable for computing the inverse transforms.

31-Jan-24 Dr.J.Aravinth 50
 COMPARISON OF DFT WITH FFT
DFT FFT

No.of Complex No.of Complex

Length of the
Sequence (N) No.of Complex No.of Complex MULs ADDs
MULs (N*N) ADDs (N*N-1)

2 4 2 1 2

4 16 12 4 8

8 64 56 12 24

16 256 240 32 64

: : : : :

128 16384 16256 448 896

: : : : :

31-Jan-24 Dr.J.Aravinth 51
1024 1048576 1047552 5120 10240
 FFT PERFORMANCE

• I. Speed calculation of FFT:

• For N=8;

FFT is 5.33 times faster than direct DFT

31-Jan-24 Dr.J.Aravinth 52
 FFT PERFORMANCE
• I. Percentage of Computation saved in FFT:

• For N=8;

• For N=16;

31-Jan-24 Dr.J.Aravinth 53
 Fast Fourier Transform
• A large amount of work has been devoted to
reducing the computation time of a DFT.
• This has led to efficient algorithms which are
known as the Fast Fourier Transform (FFT)
algorithms.
• Decimation In Time(DIT) Radix-2 Algorithm
• Decimation In Frequency(DIF) Radix-2 Algorithm

31-Jan-24 Dr.J.Aravinth 54
Fast Fourier Transform(FFT) X (k ) = N −1
 
N −1 2  x n W nk
N

X (k ) =  xne
−j nk n =0
N
N −1

−j
2
n =0
x(n ) =
1
 X k 
W − nk
N
e N
= WN (TwiddleFactor (or ) PhaseFactor ) N n =0
N 2 2 N 2 2
k+ −j k −j −j k −j k
− j
WN 2
=e N
e N 2
=e N
e = −e N
= −WNk , Symmetry Pr operty
N 2 2 N 2
k+ −j k −j −j k
WN 2
=e N 2
e N 22
=e N 2
= W Nk , Periodicit y Pr operty
2 2
N-Point DFT DFT FFT
N(N-1) N2 complex ‘×’ N/2 log2(N) N log2(N)
complex ‘+’ complex ‘×’. complex ‘+’.

8 56 64 12 24
16 240 256 32 64
32 992 1024 80 160
64 4032 4096 192 384
31-Jan-24 Dr.J.Aravinth 55
128 16256 16384 448 896
 DIT Radix-2 FFT Algorithm
• Let us consider the computation of the N-Point DFT
N = r m,r-Radix, m- No.of Stages
First Step: x[n] = x[0], x[1], …, x[N-1]
Split the N-point data sequences into two N/2-point
data sequences f1(n) and f2(n)
Lets divide the sequence x[n] into even and odd
Sequences


N 
f1(n) =x[2n] = x[0], x[2], …, x[N-2] ; n = 0,1,  − 1 
2  − − −(1)
f2(n) =x[2n+1] = x[1], x[3], …, x[N-1] ; n = 0,1, N − 1
2 
31-Jan-24 Dr.J.Aravinth 56
 N −1
X (k ) =  xnWNnk ; 0  k  N − 1 N-point DFT
n =0
N N
−1 −1

X (k ) =    
2 2
 x
n =0
2 n W N
2 nk
+  x 2 n + 1 WN
( 2 n +1) k
n =0
2 2
−j 2 nk −j nk
WN2 nk = e N
=e N 2
= W Nnk
2

WN(2 n +1)k = WNk  W Nnk

2
N N
−1 −1

X (k ) =  x2nWNnk + WNk  x2n + 1WNnk

2 2

n =0 2 n =0 2
N N
−1 −1
2 2
=  f1 (n)WNnk + WNk  f 2 (n)WNnk
n =0 2 n =0 2

X (k ) = F1 (k ) + WNk F2 (k ); k = 0,( N − 1) − − − − − − − −(2)

31-Jan-24 Dr.J.Aravinth 57
• Where F1(k) and F2(k) N/2-point DFTs
N
−1

F1 (k ) =
2

n =0
f1 ( n)WNnk
2
N
−1

F2 (k ) =
2

n =0
f 2 ( n)WNnk
2
k+N / 2
• Since F1(k+N/2)= F1(k) and F2(k+N/2)= F2(k), WN = −WNk

N 
X (k ) = F1 (k ) + W F2 (k ); k = 0, − 1 − − − − − (3.1)
k

2 
N

 N N 
X  k +  = F1 (k ) − WN F2 (k ); k = 0, − 1 − − − − − (3.2)
k

 2 2 

31-Jan-24 Dr.J.Aravinth 58
Second Step: Split the N/2-point data sequences into
two N/4-point sequences
 N 
 v11(n) =f1(2n) ; n = 0,1,  
 4

− 1



 v12(n) =f1(2n+1)--------------- (4.1)

 N 
 v21(n) =f2(2n) ; n = 0,1,  − 1
 4 

 v22(n) =f2(2n+1)----------------(4.2)
N
−1

F1 (k ) =  f1 (n)WNnk
2

n =0 2
N N
−1 −1
4 4
=  f1 (2n)WN2 nk +  f1 (2n + 1)WN( 2 n+1) k
n =0 2 n =0 2
N N
−1 −1

( ) ( ) − (5)
4 4

31-Jan-24
=  11
v
n =0
( n )W nk
N + W N  Dr.J.Aravinth
k
v 12 ( n ) W nk
N = V11 k + W N V12 k − − 59
k

4 2 n =0 4 2
 F 1 (k ) = V11 (k ) + WNkV12 (k ) ; k = 0,1,..........., ( N − 1)
2
4

 N
F k +  = V11 (k ) − WN V12 (k ) − − − (6.1)
k

1 4  2

F 2 (k ) = V21 (k ) + WN V22 (k )
N
k ; k = 0,1,..........., ( − 1)
4
2

 N
F k +  = V (k ) − W ( )
N V22 k − − − (6.2)
k

2 4 
21
2

• Where V11(k),V12(k),V21(k) and V22(k)

• N/4 point DFT of the sequences v11(n),v12(n),v21(n) and v22(n)
N N
−1 −1

V11 (k ) =  v11 (n)WN ;V12 (k ) =  v12 (n)WN

4 4
nk nk

n =0 4 n =0 4
N N
−1 −1

V21 (k ) =  v21 (n)WN ;V22 (k ) =  v22 (n)WN

4 4
nk nk

n =0 4 n =0 4
31-Jan-24 Dr.J.Aravinth 60
• N=8, N =2 - 3
Three Stages
• Four 2-point(N/4) DFT’s --->Two 4-point(N/2) DFT’s --->One 8-point (N)DFT
v11(n) =f1(2n)
 v11(0) = f1(0) = x(0) v12(n) =f1(2n+1)
 v11(1) = f1(2) = x(4) v21(n) =f2(2n)
v22(n) =f2(2n+1) ; n=0,1,…(N/4-1)
 v12(0) = f1(1) = x(2) f1(n) =x[2n]

 v12(1) = f1(3) = x(6) f2(n) =x[2n+1] ; n=0,1,…(N/2-1)

Bit Re versal − Shuffling ofData
 v21(0) = f2(0) = x(1) x0 = x000  x000 = x0
 v21(1) = f2(2) = x(5) x1 = x001  x100 = x4
x2 = x010  x010 = x2
 v22(0) = f2(1) = x(3) x3 = x011  x110 = x6
x4 = x100  x001 = x1
 v22(1) = f2(3) = x(7)
x5 = x101  x101 = x3
x6 = x110  x011 = x5
31-Jan-24 x7 = x111  x111 = x7
Dr.J.Aravinth 61
• FirstNStage
−1
-Four 82-point(N/4)
−1
DFT’s
1
V11 (k ) =
4 4
 11
v ( n )W nk
N =  11
v ( n )W 8
nk
=  11
v ( n )W2
nk

n =0 4 n =0 4 n =0 V11 (0)
= v11 (0)W2( 0 ) k + v11 (1)W2(1) k V11 (1)
k = 0;V11 (0 ) = v11 (0)W2( 0 ) 0 + v11 (1)W2(1) 0
= v11 (0) + W20 v11 (1) = x(0) + W80 x( 4) V12 (0)

k = 1;V11 (1) = v11 (0)W2( 0 )1 + v11 (1)W2(1)1 V12 (1)

= v11 (0) + v11 (1)W21 = x(0) − W80 x(4)

V12 (0 ) = x( 2) + W80 x(6) V21 (0)

V12 (1) = x( 2) − W80 x(6) V21 (1)

V21 (0 ) = x(1) + W80 x(5)

V22 (0)
V21 (1) = x(1) − W80 x(5)
V12 (1)
V22 (0 ) = x(3) + W x(7)
8
0
2 ( 0 )
V22 (131-Jan-24
) = x(3) − W x(7) −j
W80 = e = 162
0 Dr.J.Aravinth 8
8
 Second Stage-Two 4-point(N/2) DFT’s
F1 (0)

F 1 (k ) = V11 (k ) + WNkV12 (k ) = V11 (k ) + W8kV12 (k ) F1 (1)

2 2

k = 0;F 1 (0) = V11 (0) + W40V12 (0) = V11 (0) + W80V12 (0) F1 (2)
k = 1;F 1 (1) = V11 (1) + W41V12 (1) = V11 (1) + W82V12 (1)
F1 (3)
 N  8
F  k +  = V11 (k ) − WN V12 (k ) → F  k +  = V11 (k ) − W8kV12 (k )
k

1 4 2 1 4 2
F2 (0)
k = 0;F 1 (2) = V11 (0) − W40V12 (k ) = V11 (0) − W80V12 (k )
k = 1;F 1 (3) = V11 (1) − W41V12 (1) = V11 (0) − W82V12 (k ) F2 (1)

k = 0;F 2 (0) = V21 (0) + W40V22 (0) = V21 (0) + W80V22 (0)
F2 (2)
k = 1;F 2 (1) = V21 (1) + W V (1) = V21 (1) + W V (1)
1
4 22
2
8 22

k = 0;F 2 (2) = V21 (0) − W40V22 (0) = V21 (0) − W80V22 (0) F2 (3)

k = 1;F 2 (3) = V21 (1) − W41V22 (1) = V21 (1) − W82V22 (1) −j
2 ( 0 )

W8 = e
0 8
=1
2 ( 2 ) 
−j −j
W8 = e =e =−j
31-Jan-24 Dr.J.Aravinth 2 8
63
2
 Third Stage-One 8-point(N) DFT
X (k ) = F1 (k ) + WNk F2 (k )
k = 0; X (0 ) = F1 (0 ) + W80 F2 (0 )
k = 1; X (1) = F1 (1) + W81 F2 (1)
k = 2; X (2 ) = F1 (2 ) + W82 F2 (2 )
k = 3; X (3) = F1 (3) + W83 F2 (3)
 N
X k +  = F (k ) − W k
F2 (k )
 2
1 N

k = 0; X (4 ) = F1 (0 ) − W80 F2 (0 )
k = 1; X (5) = F1 (1) − W81 F2 (1) 2 ( 0 )
−j

k = 2; X (6 ) = F1 (2 ) + W82 F2 (2 ) W =e =1
0 8
8
2  (1 ) 
−j −j

k = 3; X (7 ) = F1 (3) + W8 F2 (3)
3 W8 = e
1 8
=e 4
= 0707 − j 0.707
2 ( 3 ) 3
−j −j
W8 = − j;W8 = e
2 3 8
=e 4
= −0707 − j 0.707
31-Jan-24 Dr.J.Aravinth 64
 1st stage 2nd stage 3rd stage
x(0) X(0)

W80
x(4) X(1)
-1
W80
x(2) X(2)
-1
W80 W81
x(6) X(3)
-1 -1
W80
x(1) X(4)
-1
W80=1 W80 W81
x(5) X(5)
-1 -1
W80 W82
x(3) X(6)
-1 -1
W80 W81 W83
x(7) X(7)
-1 -1 -1

31-Jan-24 Dr.J.Aravinth 65
31-Jan-24 Dr.J.Aravinth 66
Decimation In Frequency(DIF) Radix-2 Algorithm
• Frequency Domain sequence is decimated
• Butterfly Diagram

31-Jan-24 Dr.J.Aravinth 67
Decimation In Frequency(DIF) Radix-2 Algorithm

31-Jan-24 Dr.J.Aravinth 68
31-Jan-24 Dr.J.Aravinth 69
31-Jan-24 Dr.J.Aravinth 70
31-Jan-24 Dr.J.Aravinth 71
31-Jan-24 Dr.J.Aravinth 72
31-Jan-24 Dr.J.Aravinth 73
 DIT ! DIF

• Time Domain Sequence Frequency Domain

decimated Sequence decimated
• Input Sequence is bit Normal order
Reversal order
• Output sequence is Bit-reversal order
Normal order
• Phase factor is Multiplied After subtraction
Before addition & subtraction
31-Jan-24 Dr.J.Aravinth 74
 Problems
• x(n)={1,-1,-1,-1,1,1,1,-1}

31-Jan-24 Dr.J.Aravinth 75
• First Stage

V11 (0)

V11 (0) = x(0) + W80 x(4) = 1 + 1.1 = 2 V11 (1)

V11 (1) == x(0) − W80 x(4) = 1 − 1.1 = 0
V12 (0 ) = x(2) + W80 x(6) = −1 + 1.1 = 0 V12 (0)

V12 (1) = x(2) − W80 x(6) = −1 − 1.1 = −2 V12 (1)

V21 (0 ) = x(1) + W80 x(5) = −1 + 1.1 = 0
V21 (1) = x(1) − W80 x(5) = −1 − 1.1 = −2 V21 (0)
V22 (0 ) = x(3) + W80 x(7) = −1 + 1.( −1) = −2
V21 (1)
V22 (1) = x(3) − W
8
0
x(7) = −1 − 1.( −1) = 0
V22 (0)

V12 (1)
2 ( 0 )
−j
31-Jan-24 Dr.J.Aravinth
W80 = e 8
= 176
 Second Stage
F1 (0)

F1 (1)
F 1 (0) = V11 (0) + W80V12 (0) = 2 + 1.0 = 2
F1 (2)
F 1 (1) = V11 (1) + W8 V12 (1) = 0 + (− j )(−2) = 2 j
2

F 1 (2) = V11 (0) − W80V12 (k ) = 2 − 1.0 = 2 F1 (3)

F 1 (3) = V11 (0) − W82V12 (k ) = 0 − (− j )(−2) = −2 j

F2 (0)
F 2 (0) = V21 (0) + W8 V22 (0) = 0 + 1.( −2) = −2
0

F 2 (1) = V21 (1) + W82V22 (1) = −2 + (− j ).0 = −2 F2 (1)

F 2 (2) = V21 (0) − W80V22 (0) = 0 − 1(−2) = 2 F2 (2)

F 2 (3) = V21 (1) − W82V22 (1) = −2 − (− j ).0 = −2 F2 (3)

2 ( 0 )
−j
W8 = e
0 8
=1
2 ( 2 ) 
−j −j
W8 = e =e =−j
31-Jan-24 Dr.J.Aravinth 2 8
77
2
 Third Stage

X (0) = F1 (0) + W80 F2 (0) = 2 + 1.( −2) = 0

X (1) = F1 (1) + W81 F2 (1) = 2 j + (0.707 − j 0.707 )( −2)
= −1.414 + j 3.414
X (2) = F1 (2) + W82 F2 (2) = 2 − j 2
X (3) = F1 (3) + W83 F2 (3) = 1.414 − j 0.585
X (4) = F1 (0) − W80 F2 (0) = 4
X (5) = F1 (1) − W81 F2 (1) = 1.414 + j 0.585
X (6) = F1 (2) + W82 F2 (2) = 2 + j 2
X (7 ) = F1 (3) + W83 F2 (3) = −1.4142 − j 3.414
2 ( 0 )
−j
W8 = e
0 8
=1
2  (1 ) 
−j −j
W8 = e
1 8
=e 4
= 0707 − j 0.707
2 ( 3 ) 3
−j −j
W8 = − j;W8 = e
2 3 8
=e 4
= −0707 − j 0.707
31-Jan-24 Dr.J.Aravinth 78
 Application of FFT Algorithm
• Linear Filtering
• Correlation
• Spectrum Analysis

31-Jan-24 Dr.J.Aravinth 79
 Use Of FFT in Linear Filtering
• Let h(n).0nM-1,be the sample response of the FIR
filter
• x(n)-Input data Sequence
Linear Filter
h(n) y ( n) Output or
Input x(n) Response

• The block size of the FFT Algorithm is N

• N=L+M-1 and L is the number of new data samples
being processed by the filter.
• N is a power of 2 . N = 23
31-Jan-24 Dr.J.Aravinth 80
1. The N-point DFT of h(n),which is padded by L-1
zeros, is denoted as H(K) using either DIT or DIF
algorithm
2. The N-point DFT of x(n), is denoted as X(K) using
either DIT or DIF algorithm.
3. Multiply X(k) and H(K), Y(k)=X(k).H(K)
4. The inverse DFT can be computed by use of an FFT
algorithm *
1 N −1
nk 
• Step 1: Conjugate of Y(K) y(n ) =   Y (k )WN 
*

N  n =0 
• Step 2: DFT of Y*(k) using either DIT or DIF radix-2
Algorithm
• Step 3:conjugate of result of step 2
• Step 4: The result of Step 3 Divided by N, we get y(n).
31-Jan-24 Dr.J.Aravinth 81
 Determine the response of LTI system when the input sequence
x(n)={-1,1,2,1,-1} by radix-2 DIT FFT. The impulse response of the
systems is h(n)={-1,1,1,-1}
• L=5,M=4,N=L+M-1=8
• x(n)= {-1,1,2,1,-1,0,0,0}
• h(n)={-1,1,1,-1,0,0,0,0}
• Determine X(K) using DIT or DIF algorithm
• X(K)={ 2,- 3.414j,-4,0.585j,-2,-0.5858j,-4,3.414j
• Determine H(K) using DIT or DIF algorithm
• H(K)={ 0,-1- 0.414j,0,-1-2.414j,-4,-1+2.414j,0,-1+0.414j}
• Y(k)={0,-1.413+3.414j,0,1.412-j0.585,8,1.12+0.585j,0,-1.413-3.414j}
• Y*(k)={0,-1.413-3.414j,0,1.412+j0.585,8,1.12-0.585j,0,-
1.413+3.414j}
• DFT of Y*(k) using Either DIT or DIF
• The result of previous step divide by N=8
• y(n)={1,-2,0,-1,1,0,2,-1}
31-Jan-24 Dr.J.Aravinth 82
 APPLICATIONS OF FOURIER TRANSFORM

• Image Processing and filters

• Transformation, representation, and encoding
• Smoothing and sharpening
• Restoration, blur removal, and Wiener filter
• Data Processing and Analysis
• Seismic arrays and streamers
• Multibeam echo sounder and side scan sonar
• Synthetic Aperture Radar (SAR) and Interferometric SAR (In-SAR)
• High-pass, low-pass, and band-pass filters
• Cross correlation — transfer functions — Coherence
• Signal and noise estimation — encoding time series.

31-Jan-24 Dr.J.Aravinth 83
 REFERENCES
1. John G. Proakis and Dimitris G. Manolakis, ―Digital Signal Processing:
Principles,Algorithms, and Applications,‖ 4th edition, 2007.

2. http://pws.npru.ac.th/sartthong/data/files/Digital%20Signal%20Processing%20-
%20Computer%20Based%20Approach%20-%20Sanjit%20K.%20Mitra.pdf.

3. https://www.scribd.com/doc/180108714/Digital-signal-processing-by-sk-mitra-
4th-edition-pdf.

4. ―Digital Signal Processing with Matlab Programs‖- Dr.S.Sanjay Sharma,

5. http://downloadingstarted.com/digital-signal-processing-by-sanjay-sharma-pdf-
free-download.html

6. John G. Proakis and Dimitris G. Manolakis, Digital Signal Processing:

Principles,Algorithms, and Applications, 4th edition, 2007.

31-Jan-24 Dr.J.Aravinth 84