CK COLLEGE OF ENGINEERING AND TECHNOLOGY

DEPARTMENT OF EEE
Fundamentals
SUBJECT CODE : EE 3024 SEM :V
SUBJECT NAME : Digital Signal processing and system design.
YEAR : III
UNIT III DISCRETE FOURIER TRANSFORM & COMPUTATION

DFT properties, magnitude and phase representation - Computation of DFT using FFT algorithm
– DIT & DIF - FFT using radix 2 – Butterfly structure.

DFT
 The drawback in DTFT is that the frequency domain representation of discrete time
signal is obtained using DTFT will be continuous function of ω.
 The DFT has been developed to convert a continuous function of ω to a discrete function
of ω.
 The DFT of a discrete time signal can be obtained by sampling the DTFT of the signal.
 The sampling of DTFT is conventionally performed at N equally spaced frequency points
in the period, 0 ≤ ω ≤ 2π.
 The sampling frequency points are denoted as ωk= for k=0, 1, 2,…., N-1.
 DFT sequence starts at k=0, corresponding to ω=0 but does not include k=N,
corresponding to ω=2π.
 The DFT defined along with number of samples and is called N point DFT.
 The number of samples N for a finite duration sequence x(n) of length L should be such
that, N ≥ L, in order to avoid aliasing of frequency spectrum.
 The X(K) is also called discrete frequency spectrum(or signal spectrum) of the discrete
time signal x(n).
 The N point DFT of x(n), where N ≥ L, is defined as X(K)= ; for
k=0,1,2,….,N-1.
DFT – Magnitude and phase Representation
 The plot of samples of magnitude sequence versus K is called Magnitude
spectrum. ;

 The plot of samples of phase sequence versus K is called phase spectrum.

 The IDFT of the sequence X (K) of length N is defined as ; for

n=0,1,2,…,N-1.
DFT Properties
 Linearity property: If DFT DFT{x1 (n)}  X 1 ( K )andDFT{x2 (n)}  X 2 ( K ) then,
DFT{a1 x1 (n)  a2 x2 (n)}  a1 X 1 ( K )  a2 X 2 ( K )
 Periodicity Property: If a sequence x(n) is periodic with periodicity of N samples then
N point DFT, X(K) is also periodic with a periodicity of N samples. ;
for all n and ; for all k
 Time shifting property: DFT{x(n-m)N} =
 Time reversal property: DFT[x(N-n)]=
 Conjugation property: DFT {x  (n)}  X  ( N  K )
j 2nm

 Multiplication by an exponential sequence: DFT {x(n)e N

}  X ( K  m) N
 Circular Convolution Property: DFT{x1(n)  x2(n)}=X1(K)X2(K).
Multiplication Property: DFT {x1 (n) x 2 (n)}  X 1 ( K ) X 2 ( K )
1

N
 j 2 kn
 The relation between Z transform and DFT is X(K)= X (Z ) Z e N .
 The DFT supports only circular convolution and so, the linear convolution using DFT has
to be computed via circular convolution.
Computation of DFT using FFT
 The FFT is a method (or algorithm) for computing the DFT with reduced number of
calculation.
 In N point DFT by radix-r FFT, the number of stages of computation will be “m” times,
where m=log2N.
 In direct computation of N-point DFT, the total number of complex additions are N(N-1)
and number of complex multiplications are N2.
 In computation of N point DFT, the total number of complex additions are Nlog2N and
total number of complex multiplications are (N/2)log2N.
 The complex valued phase factor or twiddle factor WN is defined as,
 The term W in phase factor represents a complex number 1-2π.
 The multiplication by K of the phase value -2π of W can be represented as Wk.
 The division by N of the phase value -2π of W can be represented as WN.
DIT and DIF algorithm
 In DIT the time domain sequence is decimated, whereas in DIF the frequency domain
sequence is decimated.
 In radix-2 FFT algorithm, the N point DFT can be realized from two numbers of N/2
point DFTs, the N/2 point DFT can be realized from two numbers of N/4 points DFTs,
and so on.
 In radix-2 FFT, N/2 butterflies per stage are required to represent the computational
process.
 In radix-2 DIT FFT, the input should be in bit reversed order and the output will be in
normal order.
 In radix-2 DIF FFT, the input should be in normal order and the output will be in bit
reversed order.
 In butterfly computation of DIT, the multiplication of the phase factor takes place before
the addition or subtraction operation.
 In butterfly computation of DIF, the multiplication of the phase factor takes place after
the addition or subtraction operation.
 In FFT, the phase factor for computing inverse DFT will be conjugate of phase factors for
computing DFT.
 CK COLLEGE OF ENGINEERING AND TECHNOLOGY
DEPARTMENT OF EEE
SUBJECT CODE : EE 8591 SEM : V
SUBJECT NAME : Digital Signal Processing YEAR : III
(100% Numerical)
UNIT III DISCRETE FOURIER TRANSFORM & COMPUTATION
DFT properties, magnitude and phase representation - Computation of DFT using FFT algorithm
– DIT & DIF - FFT using radix 2 – Butterfly structure.

Solved Two Marks

DFT :
1. Define DFT and IDFT of a discrete time sequence. May-06
The Discrete Fourier transform of finite duration sequence x(n)obtained by
sampling the fourier transform X(ejw) at N equally spaced points over the interval
0<ω<2π with a spacing of 2π/N.
N 1
X (k)   x(n) e  j 2 kn / N
n 0 ; for k=0,1,2,….,(N-1)

The IDFT is used to convert the N-point frequency domain sequence X(k) to an
point time domain sequence. The IDFT of the sequence X(k) of length N is defined as
2 kn
1 N 1

j
x ( n)  X ( k )e N
N n 0 ; for n=0,1,2,….,(N-1)

2. Mention the applications of DFT.

 The DFT is used for spectral analysis of signals using a digital computer.
 The DFT is used to perform filtering operations on signals using digital computer.
 The DFT is used to power spectrum estimation or calculation of harmonics.
 DFT can be computed quickly FFT algorithms.
 Linear filtering using DFT is computationally efficient because of FFT
algorithms.
3. List the limitations of using DFT for spectral analysis.
 For more accurate spectrum, proper window function and large values of N and L
are required. This increase processing time.
 Because of the windowing, the power leaks out in the entire frequency
range.Hence selection of windows depend upon type of signal to be resolved.
 Since finite length of the input sequence is taken, spectrum is not perfect.

4. Compute the DFT of the sequence whose for one period is given by x (n) = {1,1,-2,-2}.
N/D-10

Solution:
N 1
X (k)   x(n) e j 2 kn / N
n 0
k=0,1,….N-1
 3
=  x(n)e
n0
 j kn /2
k=0,1,..3
3
X (0)   x(n)  (1  1  2  2)  2
n 0
3
X(1)   x(n)e  j n /2  (1  1( j)  (2)(1)  2(j))
n 0
 (3  3 j)
3
X(2)   x(n)e j n  1  1(2)  2(1)  2(1)  0
n 0
3
X(3)   x(n) e  j 3 n /2  1  1(j)  2(1)  2( j)  3  3 j
n 0

X (k)  {2,3  3 j, 0,3  3 j}

5. Find the IDFT of Y(k) = {1,0,1,0}.

N 1
1
y (n) 
N
 Y (k)e
k 0
 j 2 kn / N
n  0,1,...N  1

1 3
  Y (k)e j kn / 2 n  0,1, 2, 3
4 k 0
1 3 1 1
y (0)  [ Y (k)]  [1  0  1  0] 
4 k 0 4 2
1 3 1
y (1)  [ Y (k) e j k / 2
]  [1  0  ( 1)  0]  0
4 k 0 4
3
1 1 1
y(2)  [ Y (k) e j k ]  [1  0  1  0] 
4 k 0 4 2
1 3 1
y (3)  [ Y (k) e j 3 k / 2
]  [1  0  1  0]  0
4 k 0 4
y(n)  [0.5, 0, 0.5, 0]
6.Distinguish between DFT and DTFT.

DFT DTFT

Obtained by performing sampling

Sampling is performed only in time
operation in both the time and
domain not in frequency domain.
frequency domains.

Discrete values of frequency k Continuous function of frequency ω.

X(k) repeats after N X(ω) repeats after N

7.What is the relationship between Fourier series coefficients of a periodic sequence and
DFT?
A periodic sequence xp(n) with fundamental period N can be represented
in a Fourier series of the form
N 1
x p (n)   c k e j 2 nk / N    n  
k 0
 1 N 1
c k 
N k 0
x p (n) e j 2 nk / N k  0,1,.... / N

Where x(n)  x p (n)0  n  N  1

0
1
Therefore, c k  X (k)  X(k)  Nc k
N
8. Find the DFT of the causal 3-sample average.
Solution:
h(n) = 1/3 for 0≤n≤2
= 0 otherwise
N 1
H(k) =  h(n) e
n 0
 j 2 kn / N

N 1
H(k) = 1/ 3e
n 0
 j 2 kn / N
1/ 3[1  e j 2 k / N  e j 4 k / N ]

1/ 3e j 2 k / N [e j 2 k / N  1  e  j 2 k / N ]
 1/ 3e  j 2 k / N [1  2cos(2 k/ N)]
1  2cos(2 k/ N)
H(k)  e j 2 k / N [ ]
3
9. Why it is requires to do zero padding in DFT analysis?
We want to find N point DFt of x(n). But the length of x(n) =M, then (N-M)
number of zeros are added to x(n). This is called zero padding
Uses:
1. Frequency spectrum is good
2.DFT is used in linear filtering because of zero padding
3. Zero padding in the time domain is used extensively in practice to compute
heavily interpolated spectra by taking the DFT of the zero-padded signal. Such spectral
interpolation is ideal when the original signal is time limited (nonzero only over some
finite duration spanned by the original samples).

10. Find the DFT of the Non-Causal 3-sample average. N/D-16

Solution:
h(n) =1/3 for -1≤ n≤ 1
= 0 otherwise
N 1
H(k) =  h(n) e
n 0
 j 2 k / N

= 1/ 3[1  e j 2 k / N  e j 2(N 1) k/ N ]

1  2 cos(2 k/ N)
=
3
11. Compute the 2N point DFT of y(n) in terms of X(k) where X(k) is the N-point DFT
 x ( n2 ) neven 
of sequence x(n), 0  n  N-1 y(n)   0 nodd 
 
Solution:
The 2N point DFT of the sequence y(n) is
 2 N 1
Y (e j 2 k /2 N )   y(n)e
n 0
 j 2 kn / N
k  0,1,...2 N  1

We know that only even value of y(n) is non-zero, where n=0,..N-1. therefore

N 1
Y (e j 2 k /2 N )   y (2 n)e  j 2 kn / N k  0,1,...2 N  1
n 0
j 2 k / N
 X (e ) k  0,1,.......2 N  1
That is the 2N-point DFT of the sequence y(n) is concatenation of 2 DFTs of x(n).

12. Find the DFT of the following signals (i) x(n) = δ(n). M/J-13
Solution:
x(n) = δ(n)
N-1
X(k) = ∑ x(n) e-j2πnk/N
n=0
N-1
= ∑ δ(n) e-j2πnk/n = 1 δ(n) = 1 for n= 0
n=0 = 0 for n ≠ 0

i.e., X(k ) = 1 for 0 ≤ k ≤ N-1

13. Find the DFT of the following signals (i) x(n) = an (ii) x(n) = δ(n-n0).
Solution:
(i)x(n) = an

N 1
X(k) = a e
n 0
n  j 2 nk / N
for 0  k  N  1

1  a N e j 2 k
=
1  ae j 2 k / N
X(k) = 1, ≤ 0≤k≤ N/2 -1

(ii) x(n) = δ(n-n0)

N 1 N 1
X (k)   x(n)e  j 2 k / N
   (n  n 0 ) e  j 2 n 0 / N
n 0 n 0
 j 2 n 0 k / N
e for 0  k  N  1

14. Consider a signal with 0.5 sec duration with a bandwidth 5 kHz. The resolution of
the spectrum is 50 Hz or less. Calculate the value of N.
Solution:
fs≥2fm = 2(5kHz)
= 10 kHz.
2f m
Where f  50 Hz 
N
 2(5000)
 50 Hz 
N
2(5000)
N  200
50
Choose the next largest power of 2, we get
N=256.
15. Find the DFT of the signal x(n)=u(n). A/M-10
Solution: Given: x(n) = u(n)
N-1
X(k) = ∑ x(n) e-j2πnk/N
n=0
N-1
= ∑ e-j2πnk/n u(n) = 1 for n≥ 0
n=0
1  e  j 2k

1  e  j 2k / N
16. Find the DFT of the sequence x(n)={1,-1,1,-1}. A/M-10
Solution:
Given: x(n)={1,-1,1,-1}
N 1
X (k)   x(n) e j 2 kn / N
n 0
k=0,1,….N-1

3
=  x(n)e
n0
 j kn /2
k=0,1,..3
3
X (0)   x(n)  {1  1  1  1}  0
n 0
3
X(1)   x(n)e j n /2  {1  1( j)  1(1)  1(j)}  0
n 0
3
X(2)   x(n)e  j n  1  1(2)  1(1)  1(1)  5
n 0
3
X(3)   x(n) e  j 3 n /2  1  1(j)  1(1)  1( j)  0
n 0

X (k)  {0, 0,5, 0}

17. Calculate the DFT of x(n)={1,1,-2,2}. N/D-10
Solution:
Given: x(n)={1,1,-2,2}
N 1
X (k)   x(n) e j 2 kn / N
n 0
k=0,1,….N-1

3
=  x(n)e
n0
 j kn / 2
k=0,1,..3
 3
X (0)   x(n)  {1  1  2  2}  2
n 0
3
X(1)   x(n)e  j n /2  {1  1( j)  2(1)  2(j)}  3  3 j
n 0
3
X(2)   x(n)e  j n  1  1(2)  2(1)  2(1)  0
n 0
3
X(3)   x(n) e  j 3 n /2  1  1(j)  2(1)  2( j)  3  3 j
n 0

X (k)  {2,3  3 j , 0,3  3 j}

18. Analog data to be spectrum analyzed are sampled at 10 kHz and the DFT of 1024
sample computed. Determine the frequency spacing between samples.
Solution:
Given spectrum analyzed = 10 kHz
Sample computed of DFT = 1024.
2f m
N= f
f  10 Hz
19. Let xu(t) be an analog signal with band width B=3KHz. we wish to use a N=2 m ;point
DFT to compute the spectrum of the signal with a resolution less than or equal to
50Hz.Determine (i) the minimum sampling rate .(ii) The minimum number of
required samples. (iii) The minimum length of the analog signal record.
Solution: Given
stB= 3KHz
 f = 50Hz
Minimum sampling rate
fs≥2(3000)≥6000 samples/sec
2 B 6000
  120
f 50
N≥  N  128
1 1
t r
 0.02sec
f 50
20. What is the relation between DFT and Z-transform? A/M-11

The z-transform of a sequence x(n) is X(z) =  x(n) z
n 
n
with a ROC that

includes the unit circle.

If X(z) is sampled at the N equally spaced points on the unit circle Z k= ej2πk/N,
k=0,1,2…N-1, obtain
X (k)  X(z) | z e j 2 k / N k  0,1,....N  1

  x(n) e
n 
 j 2 nk / N

If the equation x(n) has a finite duration of length N, then

 N 1
X (z)   x(n)Z  n
n 0
N 1 N 1
1
 [  X (k) e j 2 k / N
]z  n
n 0 N k 0
N 1 N 1 n
1

N
 X (k) (e j 2 k / N z 1 )
k 0 k 0
 N N 1
1 z X (k)
 
N k 0 1  e  j 2 k / N 1
z
21. What is the relation between DFT and Fourier transforms

Relationship to the Fourier transforms:

Fourier transform X(ejw) of a finite duration sequence x(n) having length N is given by
N 1
X(ejw) =  x(n)e jwn
n 0
The discrete fourier transform is given by
N 1
X(k) =  x (n) e  j 2 kn / N k=0,1,2,3,….N-1
n 0
Comparing the above two equations
X(k)=X(ejw)/w=2πk/N
DFT Properties:
22. Mention the properties of DFT.
Periodicity
If X(k) is N-point DFT of a finite duration sequence x(n), then
x(n+N) = x(n) for all n
X(k + N) = X(k) for all k
Linearity If X1(k) = DFT [x1 (n)] and
X2(k) = DFT [x2(n)],
Then DFT [a1x1(n) + a2x2(n)] = a1X1(k) + a2X2(k)
Time reversal of a sequence:
If DFT [x(n)] = X(k), then
DFT [x(-n)N] = DFT [x(N-n)] = X((-k)) N = X (N-k)
Circular time shifting of a sequence:
If DFT [x(n)] = X(k),
Then
DFT [x((n-1))N] = X(k)e-j2πkl/N
23. Define Parseval’s theorem. May-07, N/D’14
Parseval’s theorem relates energy of the duration sequence in terms of its
frequency components,
 2  2
1
E   x ( n)   X (e jw ) dw
n  2 

If the sequence x(n) is real and odd (or) imaginary and even, then X(k) is purely
imaginary.
If the sequence x(n) is real and even (or) imaginary and odd, then X(k) is
purely real.
24. State and prove time shifting property of DFT.
 Solution:
DFT[x(n)] = X(k)
e j 2 km/ N X (k)
DFT [x((n-m))N] =
Proof
N 1
DFT  x   n  m   N     x  n  m   N e  j 2 kn / N
n 0

subn  m  p
n pm
N 1
  x  p  e  j 2 k  p  m  / N
p 0
N 1
  x  p  e  j 2 kp / N e  j 2 km / N
p 0

 e  j 2 km / N X  k 
25. Define circular frequency shifting property of DFT.
Solution:
If DFT [x(n)] = X(k)
Then DFT [x(n)ej2πln/N] = X ((k  l)) N
Thus, shifting the frequency components of DFT circularly is equivalent to multiplying
the time domain sequence by ej2πln/N

26. Let X(z) denote the z-transform of the sequence

X(n) = {2, 1, 1, 0, 3, 2, 0, 3, 4, 6}
Let X1(k) represents the samples of X(z) evaluate on the unit circle at 7 equally
j 2 k
spaced points given by z  e 7 , k  0,1,......6. determine the 7-point IDFT x1(n) of
X1(k).
Solution:
The length of the sequence L = 10
The number of samples taken on the unit circle is 7, thus a 7-point DFT is
obtained. That is N = 7 since N < L time domain aliasing occurs. In that case if we take
IDFT of X1(k) the resulting sequence x1(n) is not equal to x(n). due to time domain
aliasing last N – L = 3 samples are added to first three samples. That is
x1(n) = {2+3, 1+4, 1+6, 0, 3, 2, 0}
= {5, 5, 7, 0, 3, 2, 0}

27. State and prove Conjugation property of DFT.

Solution:

Let x(n) be a complex N-point discrete sequence and x*(n) be its conjugate sequence
Now if, DFT {x(n)}=X(k), then DFT{x*(n)}=X*(N-k).
 Pr oof :
 j 2 nk  j 2 nk *
N 1  N 1 
DFT {x *(n)}   x *(n)e N
   x ( n )e N 
n 0  n 0 
j 2 nk *j 2 nk  j 2  nN *
 N 1  j 2 n   N 1 
   x ( n )e     x ( n )e  e  j 2 n  1
N

N N
e e
 n 0   n 0 
 j 2 nN  k *
 N 1 
   x(n)e N   [ X ( N  k )]*  X* ( N  k )
 n 0 

28. Compute the N-point DFT of the sequence

x(n)=cos(2  rn/N ,0  n  N – 1, and 0  r  N – 1

Proof:
x(n)=cos(2  rn/N)

j 2 nr  j 2 nr
e N
e N

2
N 1 j 2 kn

X(k)=  x ( n) e
n 0
N

j 2 nr j 2 kn j 2 nr j 2 kn
1  N 1  N 1   
   x ( n) e N
e N
  x ( n) e N
e N 
2  n 0 n 0 
 
1  N 1 
j 2 ( r  k ) n N 1 j 2 ( r  k ) n

   x ( n) e N   x ( n) e N

2  n 0 n 0 
1
  N  (k  r)  n  (k  N  r) 
2
State and prove Linearity property of DFT
29. The linearity property of DFT states that the DFT of a linear weighted
combination of two or more signals is equal to similar linear weighted combination of the
DFT of individual signals.
Let , DFT  x1 n   X 1 (k ) and DFT {x2 (n)}  X 2 (k ). Then by linearity property
DFT{a1x1 (n)  a 2 x 2 (n)}  a1X1 (k)  a 2X 2 (k), where a1 and a 2 are constants
Proof :
By definition of discrete Fourier transforms
N 1  j 2 kn
x1 (k )  DFT {x1 (n)}   x1 (n)e N
           (1)
n 0
 N 1  j 2 kn
x2 (k )  DFT {x2 (n)}   x2 ( n)e N
           (2)
n 0
N 1  j 2 kn
DFT {a1 x1 (n)  a2 x2 (n)}   [ a1 x1 (n)  a2 x2 ( n)]e N

n 0
N 1  j 2 kn
  [a1 x1 (n)  a2 x2 (n)]e N

n 0
N 1  j 2 kn N 1  j 2 kn
 a1  x1 (n)e N
 a2  x2 (n)e N

n 0 n 0

 a1 X 1 (k)  a2 X 2 (k) using equ(1) & (2)

30. Let X(k) denote the N-point DFT of an N-point sequence x(n). if the DFT of X(k) is
computed to obtain a sequence x1(n). Determine x1(n) in terms of x(n).
Solution:
N 1  j 2 kn
X (k)   x(n) e N

n 0
x1(m) = DFT of X(k)
N 1  j 2 km
  X(k) e N
m  0,1,............ N  1
k 0
N 1 N 1  j 2 km  j 2 kn
  x(n) e N
e N

n 0 k 0
N 1 N 1  j 2 k ( m  n )
  x(n) e N

n 0 k 0
 j 2 k ( m  n )
e N
 1 if n  m  0,  N ,  2 N ,......
 0 otherwise

We take the case n+m =0

n=-m
x1(m) = x(- m)N
x1(n) = Nx(- n)

31. Prove that If a sequence x (n) is periodic with periodicity of N samples then N- point
DFT, X (k) is also periodic with a periodicity of N samples.
Solution:
If x (n) and X (k) are N point DFT pair then, x (n+N) = x (n); for all n
X (k+N) = X (k); for all k
Pr oof
By definition of DFT , the (k , N )th coefficient of X (k ) is given by,
N 1  j 2 n ( k  N ) N 1  j 2 n ( k )  j 2 n N
X ( k  N )   x ( n)e N
  x ( n)e N
e N

n 0 n 0
N 1  j 2 nk N 1  j 2 nk
  x ( n )e N
e  j 2 n   x ( n ) e N
[ for integer n, e  j 2 n  1]
n 0 n 0
32. State and prove Circular time shift property of DFT.
The circular time shift property of DFT says that if a discrete time signal is circularly
shifted in time
 j 2 km
N
By M units then its DFT is multiplied by e
 j 2 km
i.e if , DFT 
x  n   X  k  , then DFT x  n  m N  X  k  e  N

Proof:
N 1  j 2 kn N 1  j 2 k ( p  m )
DFT {x((n  m)) N }   x((n  m)) N e N
  x( p )e N

n 0 p 0

Let , P  n  m, n  p  m
N 1  j 2 kp  j 2 km
  x ( p )e N
e N

p 0

 N 1  j 2 kp
  j 2 km
  x( p)e N  e N
 p 0 
 j 2 kp
 X (k )e N

33. If N-point sequence x(n) has N-point DFT X(k) then what is the DFT of the
following (i)x*(n) (ii) x*(N-n) (iii) x((N-1)) N (iv) x(n)ej2πln/N
Solution:
(i)DFT [x*(n)] = X*(N-k)
(ii)DFT [x*(N-n)] = X*(k)
(iii)DFT [x((n-1))N] = X(k) e –j2πkl/N
(iv)DFT [X(n) ej2πln/N] = X((k-l)) N
34. Find the DFT of the sequence x[n]={1,1,0,0} A/M-15
Solution:
N-1
X K =  x(n)e
n=0
-j2πkn/N
;K=0,1,2,3
We know that 3
X(K)= x(n)e-jπkn/2
n=0
3
X(0)=  x(n)e-j2π(0)n/N
n=0

X(0)=1+1+0+0=2
X(0)=2
X(1)=1e-0 +1e-jπ/2 +0e-jπ +0e-j3π/2
X(1)=1e-0 +1(cosπ/2  j sin π/2)+0(cosπ  j sin π)+0(cos3π/2  j sin 3π/2)
X(1)=1-j
X(2)=1e-0 +1e-jπ +0e-j2π +0e-j3π
X(2)=1e-0 +1(cosπ  j sin π)+0(cos2π  j sin 2π)+0(cos3π  j sin 3π)
X(2)=0
X(3)=1e-0 +1e-j3π/2 +0e-j3π +0e-j9π/2
X(3)=1e-0 +1(cos3π/2  j sin 3π/2)+0(cos3π  j sin 3π)+0(cos9π/2  j sin 9π/2)
 X(3)=1+j
X(K)={2,1-j,0,1+j}

35. The first five points of the eight point DFT of a real valued sequence are {0.25,
0.125-j0.3018, 0.125-j0.0518, 0}. Determine the remaining DFT coefficients.
N/D’13
Solution:
Given: X (K) = {0.25, 0.125-j0.3018, 0.125-j0.0518, 0} the first five points of the eight
point DFT.
We know that, DFT [x*(N-n)] = X*(k)
X(5) = X*(8-7) = X*(3) = (0)* = 0.
X(6) = X*(8-5) = X*(3) = (0.125-j0.0518)* = 0.125+j0.0518.
X(7) = X*(8-6) = X*(2) = (0.125-j0.3018)* = 0.125+j0.3018.
36. Given the sequences x1(n)={1,2,3,4};x2={1,1,2,2}. find x3(n) such that X3(k)=
X1(k)X2(k).
Solution: x3= IDFT [X3(k)]= IDFT[X1(k)X2(k)]
= x1(n) x2(n)
N

so we find x3(n) by circular convolution x1(n)and x2(n)

Given
x1(n)={1,2,3,4}
x2(n)={1,1,2,2}
representing x2(n) as NxN matrix form and x1(n) as column matrix and multiplying
 1 2 2 1  1  15
    
 1 1 2 2   2   17 
 2 1 1 2   3  15 
    
 2 2 1 1   4  13
y(n) = {15,17,15,13}
37. Define circular convolution. May-09
Let x1(n) and x2(n) are finite duration sequences both of length N with DFTs
X1(k) and X2(k).
If X3(K) = X1(k)X2(k), then the sequences x3(n) can be obtained by circular
convolution, defined as,
N 1
x 3 (n)   x 1 (m) x 2 ((n  m)) N
m0

38. Distinguish between linear and circular convolution of two sequences.

Linear Convolution Circular Convolution

If x(n) is a sequence of L Number of If x(n) is a sequence of L Number of

samples and h(n) with M number of samples and h(n) with M samples, after
samples, after convolution y(n) will contain convolution y(n) will contain N= Max(L,M)
N= L+M-1 samples. samples.
Linear convolution can be used to find the Circular convolution cannot be used to find
response of a linear filter. the response of a linear filter.
Zero padding is not necessary to find the Zero padding is necessary to find the
response of a linear filter. response of a filter.
39. Obtain the circular convolution of the following sequences x(n) = {1,2,1};
h(n) = {1,-2,2}
 Solution:
The circular convolution of the above sequence can be obtained by using matrix
approach. In this method can write h(n) as 3x3 matrix and x(n) as column matrix. The
product of these two matrices gives the output.
 h(0) h(2) h(1)  x(0)   y(0) 
    
 h(1) h(0) h(2)  x(1)    y(1) 
 h(2) h(1) h(0)  x(2)   y(2) 
    
 1 2 2  1   3 
    
 2 1 2  2    2 
 2 2 1  1   1
    
y(n) = x(n) N h(n)={3,2,-1}

40. Find linear convolution of the sequences x(n) = {1,2,3}; h(n)={-1,-2} using circular
convolution.
The linear convolution of two sequences x(n) and h(n) gives an output having
3+2-1 = 4 samples. In order to obtain the number of samples in circular convolution
equal to 4, append one zero to sequence x(n) and two zeros to sequence h(n) i.e.,
x(n) = {1,2,3,0}; h(n)= {-1,-2,0,0}
The circular convolution of x(n) and h(n) can be obtained by using matrix approach
 h(0) h(3) h(2) h(1)  x(0)   y(0) 
    
 h(1) h(0) h(3) h(2)  x(1)    y(1) 
 h(2) h(1) h(0) h(3)  x(2)   y(2) 
    
 h(3) h(2) h(1) h(0)  x(3)   y(3) 
 1 0 0 2  1   1 
    
 2 1 0 0  2    4 
 0 2 1 0  3   7 
    
 0 0 2 1  0   6 
x(n)N
h(n) = {-1,-4,-7,-6}
41. Mention the difference between overlap-save method and overlap-add method.
Overlap-save method Overlap-add method

In this method the size of the input data block In this method the size of the input data block
is N=L+M-1. is L.
Each data block consists of the last M-1 data Each data block is L points and append M-1
points of the previous data block followed by L zeros to compute N-point DFT.
new data points.
In each output block M-1 points are corrupted In this no corruption due to aliasing, as linear
due to aliasing, as circular convolution is convolution is performed using circular
employed. convolution.
To form the output sequence the first M-1 data To form the output sequence, the last M-1
points are discarded in output block and the points from each output block is added to the
remaining data are fitted first (m-1) points of the succeeding block.
Computation of DFT using FFT:
42. What is FFT?
 The fast Fourier transform (FFT) is an algorithm used to compute the DFT. It
makes use of the symmetry and periodicity properties of twiddle factor W Nk to
effectively reduce the DFT computation time. It is based on the fundamental principle of
decomposing the computation of DFT of a sequence of length N into successively
smaller discrete Fourier transforms. The FFT algorithm provide speed-increase factors,
when compared with direct computation of the DFT, pf approximately 64 and 205 for
256-point and 1024-point transforms respectively.
43. What is the need for FFT?
N 1
The direct evaluation of DFT using the formula X(k) =  x(n) e
n 0
 j 2 nk / N
requires

N2 complex multiplications and N(N-1) complex additions. Thus for reasonably large
values of N(N-1) complex additions. Thus for reasonably large values of N (in order of
1000) direct evaluation of the DFT requires an inordinate amount of computation. By
using FFT algorithms the number of computations can be reduced. For example, for an
N
N-point DFT, the number of complex multiplications required using FFT is log 2 N . If
2
N=16. The number of complex multiplications required for direct evaluation of DFT is
256 only 32 multiplications are required.
44. What is the speed improvement factor in calculating 64-point DFT of a sequence
using direct computation and FFT algorithm?
The number of complex multiplications required using direct computation is
N2 = 64 = 4096.
The number of complex multiplications required using FFT is
N 64
log 2 N  log 2 64  192. 4096
2 2 Speed improvement factor =  21.33
192
45. What is the main advantage of FFT?
FFT reduces the computation time required to compute discrete Fourier transform.
Calculate the number of multiplications needed in the calculations of DFT using FFT
algorithm with 32-point sequence.
For N-point DFT the number of complex multiplication needed using FFT
N
algorithm is log 2 N . For N=32, the number of complex multiplications is equal
2
32
to log 2 32  16 5  80.
2
0
46. The butterfly diagram of N = 64 FFT algorithm has a twiddle factor w64 for one
of the butterflies in the last stage. Is the FFT a decimation-in-time or decimation-in-
frequency algorithm.
Solution:
Decimation-in-time algorithm is used to calculate the DFT of a N-point sequence. The
0
FFT is a decimation-in-time algorithm, since the decimation-in-time algorithm has w64
term for one of the butterflies. In decimation-in-frequency algorithm all butterflies have
0
w64 term. In DIF algorithm the output sequence X(k) is divided into smaller and smaller
subsequences.
47. Compute the number of multiplications and additions for 32-point DFT &
FFT?A/M-17
 The number of multiplications required to compute N-point DFT using radix-2 FFT is
Nlog2N .
The number of additions required to compute N-point DFT using radix-2 FFT is
N
log 2 N .
2
48. What is meant by radix-2 FFT and Mention the applications of FFT.June-
2011,2014,2015
The FFT algorithm is most efficient in calculating N-point discrete fourier transform. If
the number of output points N can be expressed as a power of 2, that is, N=2 M, where M
is an integer, then this algorithm is known as radix-2 fast fourier transform algorithm.

FFT applications:
 Linear filtering
 Correlation
 Spectrum analysis
49. Calculate the number of multiplications needed in the calculation of 512 point radix
2FFT when compared to direct DFT.
Solution:
The number of complex multiplications required using direct computation is
N2 = 5122 = 262144.
N
The number of complex multiplications required using FFT is log 2 N .
2
Number of multiplication needed as 2304.
50. How can we calculate IDFT using FFT algorithm?

The inverse DFT of an N-point sequence X(k); k=0,1,….N-1 is defined as

1 N 1
x(n) =  X (k)W  nkN
N k 0
If take complex conjugate and multiply by N, we get
N 1
Nx*(n) =  X *(k)W
k 0
nk
N

The right hand side of the above equation is DFT of the sequence X*(k) and may be
computed using any FFT algorithm. The desired output sequence x(n) can then be
obtained by complex conjugating the DFT of and dividing by N to give
1 N 1
x(n)  [ X *(k) W nkN ] *
N k 0
DIT & DIF - FFT using radix 2 – Butterfly structure:
51. What is a decimation-in-time algorithm?
Decimation-in-time algorithm is used to calculate the DFT of a N-point sequence.
The idea is to break the N-point sequence into two sequences, the DFTs of which can be
combined to give the DFT of the original N-point sequence. Initially the N –point
N
sequence is divided into two -point sequences xe(n) and x0(n),which have even and
2
N
odd members of x(n) respectively. The -point DFTs of these two sequences are
2
N
evaluated and combined to give the N-point DFT. Similarly the -point DFTs can be
2
 N
expressed as a combination of -point DFTs. This process is continued until are left
4
with 2-point DFT. This algorithm is called decimation-in-time because the sequence x(n)
is often split into smaller subsequences.
52. Draw the flow graph of a 4 point radix -2 DIT – FFT butterfly structure for DFT.
M\J-16

The signal flow graph is also called butterfly diagram because it resembles a butterfly. In
radix-2 FFT, Z/2 butterflies per stage are required to represent the computational process.
The butterfly diagram used to compute the 4point DFT via radix-2 DIT FFT

53. What are the applications of FFT algorithm? M\J-16

FFT's importance derives from the fact that in signal processing and image processing it has
made working in frequency domain equally computationally feasible as working in temporal or
spatial domain. Some of the important applications of FFT includes,
 Fast large integer and polynomial multiplication
 Efficient matrix-vector multiplication for Toeplitz, circulant and other structured matrices
 Filtering algorithms
 Fast algorithms for discrete cosine or sine transforms (example, Fast DCT used for JPEG,
MP3/MPEG encoding)
 Fast Chebyshev approximation
 Fast discrete Hartley transform
 Solving difference equations
 Computation of isotopic distributions.

54. What is decimation-in-frequency algorithm?

In DIF algorithm the output sequence X(k) is divided into smaller and smaller
subsequences, that is why the name decimation in frequency. Initially te input sequence
N
x(n) is divided into two sequences x1(n) and x2(n) consisting of the first -samples of
2
N N
x(n) and the last -samples od x(n) respectively. Then, find the -point sequences
2 2
f(n) and g(n) as f(n)= x1(n) + x2(n) and g(n)=[x1(n)-x2(n)] W Nn
 N
The -point DFTs of the above two sequences give even membered and odd membered
2
output samples respectively.The above procedure can now be iterated to express each
N N N
-point DFT as a combination of two -Point DFT as a combination of two -point
2 2 2
DFTs. This process is continued until are left with 2point DFT.

55. What is phase factor or twiddle factor?A/M-17

The complex number WkN is called phase factor or twiddle factor.
The WkN represent a complex number e-j2π/n.
WkN= e-j2π/n
It also represents an Nth root of unity.
56. What are the differences and similarities between DIF and DIT algorithm? N/D-10
Differences:
1. For DFT, the input is bit reversal while the output is in natural order, whereas for
DIF, the input is in natural order while the output is bit reversal.
2. The DIF butterfly is slightly different from the DIF butterfly, the difference being
that the complex multiplication takes place after the add-subtract operation in DIF.
Similarities:
Both algorithms require same number of operations to compute the DFT. Both
algorithms can be done-in-place and both need to perform bit reversal at some place
during the computation.
57. The DFT of a sequence x(n) that has N = 2m can be calculated using two
algorithms. Algorithm A computes the DFT by direct evaluation and take N 2
seconds to run. Algorithm B implements DIT-FFT and takes 5Nlog2N seconds to
run. What is the shortest sequence N such that algorithm B runs faster than
algorithm A.
Solution:
N Algorithm A Algorithm B
(N2 seconds) (5N log2 N seconds)
2 4 10
4 16 40
8 64 120
16 256 320
32 1024 800
Thus, we see that a sequence with length N = 32 is the shortest sequence for which
algorithm B runs faster than algorithm A.

58. The butterfly shown below is a decimation-in-time FFT with N = 32. Assume that
the three stage of the butterfly diagram are indexed by m = 1,….3. what are the
possible values of k for each of the three stage.
 k
w32
The twiddle factor exponents for each stage are given by
Nt
k m ; t  0,1, 2,3,................2m 1  1
2
for stage1; m 1
Nt
k ; t  0,1, 2,3,................2m 1  1
2m
Nt
k ; t 0
2
k 0
for stage 2; m2
32t
k t  0, 1
4
k  0, 8
for stage 3; m3
32t
k t  0, 1, 2, 3
8
k  0, 4, 8, 12
59. What are the basic operations of the DIF algorithms? M/J-13
The basic operation of the DIF algorithm is the so called butterfly in which two
inputs Xm(p) and Xm(q) are combined to give the outputs Xm+1(p) and Xm+1(q) via the
operation.
Xm+1(p) = Xm(p) + Xm(q)
Xm+1(q) = [Xm(p) – Xm(q)] W Nk
Where W Nk is twiddle factor
60. Draw the basic butterfly flow graph for the computation in the DIT and DIF
algorithm. A/M”15, May-11
xm(p) p xm+1(p) = xm(p) + xm(q)

xm(q) q W Nk xm+ 1(q)=xm(p)- W Nk xm(q)

The basic butterfly diagram for DIT algorithm is

Xm(p) p W Nk Xm+1(p) = Xm(p) + Xm(q)

Xm(q) q Xm+ 1(q)=[Xm(p)- Xm(q) ] W Nk

PART B (13 Marks Solved)
DFT and its properties
State and prove the properties of DFT.
1. Refer A.Nagoor kani page no.:5.4-5.9 (13)A/M-11
1. Linearity property: (2)
 The linearity property of DFT states that the DFT of a linear weighted
combination of two or more signals is equal to similar linear weighted combination of
the DFT of individual signals.
Let , DFT  x1 n   X 1 (k ) and DFT {x2 (n)}  X 2 (k ). Then by linearity property
DFT{a1x1 (n)  a 2 x 2 (n)}  a1X1 (k)  a 2X 2 (k), where a1 and a 2 are constants
Proof :
By definition of discrete Fourier transforms
N 1  j 2 kn
x1 (k )  DFT {x1 (n)}   x1 (n)e N
           (1)
n 0
N 1  j 2 kn
x2 (k )  DFT {x2 (n)}   x2 ( n)e N
           (2)
n 0
N 1  j 2 kn
DFT {a1 x1 (n)  a2 x2 (n)}   [ a1 x1 (n)  a2 x2 ( n)]e N

n 0
N 1  j 2 kn
  [a1 x1 (n)  a2 x2 (n)]e N

n 0
N 1  j 2 kn N 1  j 2 kn
 a1  x1 (n)e N
 a2  x2 (n)e N

n 0 n 0

 a1 X 1 (k)  a2 X 2 (k) using equ(1) & (2)

2. Periodicity property: (2)
If a sequence x (n) is periodic with periodicity of N samples then N- point DFT, X
(k) is also periodic with a periodicity of N samples.
Hence, if x (n) and X (k) are N point DFT pair then, x (n+N) = x (n); for all n
X (k+N) = X (k); for all k
Pr oof
By definition of DFT , the (k , N )th coefficient of X (k ) is given by,
N 1  j 2 n ( k  N ) N 1  j 2 n ( k )  j 2 n N
X ( k  N )   x ( n)e N
  x ( n)e N
e N

n 0 n 0
N 1  j 2 nk N 1  j 2 nk
  x ( n )e N
e  j 2 n   x ( n ) e N
[ for integer n, e  j 2 n  1]
n 0 n 0
3. Circular time shift: (2)
The circular time shift property of DFT says that if a discrete time signal is circularly
shifted in time
 j 2 km
N
By M units then its DFT is multiplied by e
 j 2 km
i.e if , DFT 
x  n   X  k  , then DFT x  n  m N  X  k  e  N

Proof:
 N 1  j 2 kn N 1  j 2 k ( p  m )
DFT {x((n  m)) N }   x((n  m)) N e N
  x( p )e N

n 0 p 0

Let , P  n  m, n  p  m
N 1  j 2 kp  j 2 km
  x ( p )e N
e N

p 0

 N 1  j 2 kp
  j 2N km
  x( p)e N
e
 p 0 
 j 2 kp
 X (k )e N

4. Time reversal: (2)

The time reversal property of DFT says that reversing the N-point sequence in time is
equivalent to reversing the DFT sequence.
i.e, if DFT{x(n)=X(k), then DFT{x(N-n)}=X(N-k).
Pr oof :
N 1  j 2 kn N 1  j 2 k ( N  m )
DFT {x(N  n}   x( N  n)e N
  x(m)e N
Let , m  N  n, n  N  m
n 0 m 0
N 1  j 2 kN j 2 km N 1  j 2 km
  x(m)e N
e N
  x(m)e N  j 2 k
e sin ce k is an int eger , e 2 k  1.
m 0 m 0
N 1 j 2 km N 1  j 2 km
  x(m)e N
  x(m)e N  j 2 k
e sin ce k is an int eger , e 2 km  1.
m 0 m 0
N 1 j 2 km  j 2 mN N 1  j 2 m ( N  K )
  x(m)e N
e N
  x(m)e N

m 0 m 0

 X (N  K)

5. Conjugation property: (2)

Let x(n) be a complex N-point discrete sequence and x*(n) be its conjugate sequence
Now if, DFT {x(n)}=X(k), then DFT{x*(n)}=X*(N-k).
Pr oof :
 j 2 nk  j 2 nk *
N 1  N 1 
DFT {x *(n)}   x *(n)e N
   x ( n )e N 
n 0  n 0 
j 2 nk *j 2 nk  j 2  nN *
 N 1  j 2 n   N 1 
   x ( n )e     x ( n )e  e  j 2 n  1
N

N N
e e
 n 0   n 0 
 j 2 nN  k *
 N 1 
   x(n)e N   [ X ( N  k )]*  X* ( N  k )
 n 0 
6. Circular frequency shift property: (2)

Circular frequency shift property of DFT says that if a discrete time signal is multiplied
 j 2 mn
N
by e its DFT is circularly shifted by m units
 j 2 mn
 
i .e , if , DFT  x  n  X  k  then DFT 
x ( n ) e N
  X ((k  m)) N
 
Pr oof :
j 2 mn j 2 mn  j 2 kn
  N 1
 x ( n )e N
   x ( n )e N e N
  n 0
N 1  j 2 ( k  m ) n
  x ( n )e N

n 0

 X ((k  M )) N
7. Multiplication: (2)
The multiplication property of DFT says that the DFT of product of two direte time
sequences is equivalent to circular convolution of the DFT s of the individual sequences
by a factor 1/N.
1
i .e. if DFT { x  n   X  k , then DFT  x1 (n) x2 (n)   X1 (k ) * X 2 (k )
N
Pr oof :
N 1 j 2 kn N 1 j 2 mn
1 1
By definiton of inverse DFT , x1 (n) 
N
 X (k )e
k 0
1
N

N
 X (m)e
k 0
1
N
let k  m    (1)

By definition of DFT ,
 j 2 nk
N 1   j 2 nk  j 2 kn
N 1
1 
DFT {x1 (n) x2 (n)}   x1 ( n) x2 ( n)e    X 1 (m)e N x2 (n) e N
N
u sin g the equation(1)
N 0 m 0  N 
N 1   j 2 nk j 2 nm
1 N 1 
   X 1 (m) x2 (n)e N e N  Re arrangine order of summation.
N m 0  n 0 
 j 2 k  m ) n
1 N 1  N 1  1 N 1
  X 1 (m)   x2 (n)e N    X 1 ( m) X 2 ( k  m) N
N m 0  n 0  N m 0
1
  X 1 (k )  X 2 (k ) 
N
8. Parseval’s relation (2)
Let DFT{X1(n) =X1 (k) and DFT {x2(n)} =X2(k)
Proof:
Let x1(n) and x2(n) be N-point sequences.
N 1  j 2 nk
Now by definition of DFT , X1  k    x1 (n)e N

n 0
 j 2 nk
1 N 1

Now by definition of DFT , x 2  k  
N k 0
X 2 (k) e N

Consider the right  hand sideterm of parseval ' srelation.

  j 2 nk
1 N 1  1 N 1 N 1 
 X1 (k) X*2 (k )   1
k 0  n 0
x (k) e N
 X 2 *(k )
N k 0 N 
 j 2 nk
N 1 1 N 1 
  x1 (n)   X 2* (k) e N 
k 0  N k 0 
j 2 nk *
N 1 1 N 1 
  x1 (n)  X 2 (k) e N

k 0 N k 0 
N 1
  x1 (n) x2* (n)
n 0
2.Determine the 8 –point DFT of the sequence x(n)={1,1,1,1} (13) N/D’14
Solution:
Given: x(n)={1,1,1,1} N=8
Therefore the input x (n) = {1, 1, 1, 1, 0, 0, 0, 0}
N 1
X (k)   x(n) e  j 2 kn / N k  0,1,...N  1
n 0 (2)
For N=8
7
X (k)   x(n) e  j kn /4 k  0,1, 2....N  1
n 0
For K=0
7
X (0)   x(n)
n 0 (2)
X(0) = X(0)+X(1)+X(2)+X(3)+X(4)+X(5)+X(6)+X(7)
= 1+1+1+1+0+0+0+0
=4
For K=1
7
X (1)   x (n) e j n /4
n0

X 1 = x  0   x 1 e  j /4  x  2  e  j /2  x  3 e  j3 /4

 x  4  e j  x  5  e j5 /4  x  6  e  j3 /2  x  7  e  j7 /4
X 1  1  0.707  j0.707  j  0.707  j0.707+0+ 0  0  0
X 1  1  j2.414
For K=2
7
X (2)   x(n)e j n /2
n 0 (2)
X  2  = x  0   x 1 e  j /2
 x  2 e  j
 x  3 e  j3 /2

 x  4  e  j2  x  5  e  j5 /2  x  6  e  j3  x  7  e  j7 /2

X  2  1  j 1  j  0  0  0  0
X  2  0
For K=3 (2)
 7
X (3)   x(n)e  j 3 n /4
n 0

X  3  x  0   x 1 e  j3 /4  x  2  e  j3 /2  x  3 e  j9 /4

 x  4  e j3  x  5  e  j15 /4  x  6  e  j9 /4  x  7  e  j21 /4
X  3  1  0.707  j0.707  j  0.707  j0.707+0+0+0+0
X  3  1-j0.414
For K=4 (2)
7
X (4)   x(n) e j n
n 0

X  4   x  0   x 1 e  j  x  2  e  j 2  x  3 e  j 3
 x  4  e j 4  x  5  e  j 5  x  6  e  j 6  x  7  e  j 7
X  4  1 1  1 1  0  0  0  0
X  4 = 0
For K=5
7
X (5)   x(n)e  j 5 n /4
n 0

X  5   x(0)  x(1)e  j5 /4  x(2)e  j5 /2  x(3)e  j5 n /4

 x(4)e  j5 +x(5)e  j25 /4 +x(6)e  j15 /2 +x(7)e  j35 /4
X  5   1  0.707  j0.707  j  0.707  j0.707+0+0+0+0
X  5   1  j0.414
For K=6 (2)
7
X (6)   x(n)e  j 3 n /2
n 0

X  6  = x  0  +x 1 e-j3π /2 +x  2  e-j3π +x  3 e-j9π /2

+x  4  e-j6π +x  5  e-j15π +x  6  e-j9π +x  7  e -j21π /2
X  6  = 1+j-1-j+0+0+0+0
X 6 = 0
For K=7
7
X (7)   x(n)e j 7 n /4 (2)
n 0

X  7   x  0  x 1 e j7 /4  x  2 e j7 /2  x 3 e j21 /4

 x  4  e  j7  x  5  e  j35 /4  x  6  e  j21 /2  x  7  e  j49 /4
X  7   1  0.707  j0.707  j  0.707  j0.707+0+0+0+0
X  7   1  j2.414
X(K)={4, 1-j2.414, 0, 1-j0.414, 0, 1+j0.414, 0, 1+j2.414} (2)

3.(i) Determine the 8-point DFT of the sequence x(n)={1,1,1,1,1,1,0,0} (9)M/J-14,A/M-17

 N 1
X (k)   x(n) e  j 2 kn / N k  0,1,...N  1
n 0 (2)
For N=8
7
X (k)   x(n) e  j kn /4 k  0,1, 2....N  1
n 0
For K=0
7
X (0)   x(n)
n 0 (2)
X(0)=X(0)+X(1)+X(2)+X(3)+X(4)+X(5)+X(6)+X(7)
= 1+1+1+1+1+1+0+0
=6
For K=1
7
X (1)   x(n) e j n /4
n 0
X(1) = x(0) +x(1) e-jπ/4+x(2) e-jπ/2+x(3) e-j3π/4+x(4) e-jπ+x(5) e-j5π/4+x(6) e-j3π/2+x(7) e-j7π/4
= 1+0.707-j0.707-j-0.707-j0.707-1-0.707+j0.707
= -0.707-j1.707
For K=2
7
X (2)   x(n)e  j n /2
n 0
X(2) = x(0)+x(1) e-jπ/2+x(2) e-jπ+x(3) e-j3π/2+x(4) e-j2π+x(5) e-j5π/2+x(6) e-j3π+x(7) e-j7π/2
= 1-j-1+j+1-j
= 1-j
For K=3 (2)
7
X (3)   x(n)e  j 3 n /4
n 0
X(3) = x(0)+x(1) e-j3π/4+x(2) e-j3π/2+x(3) e-j9π/4+x(4) e-j3π+x(5) e-j15π/4+x(6) e-j9π/4 +x(7) e-
j21π/4

= 1-0.707-j0.707+j+0.707-j0.707-1+0.707+j0.707
= 0.707 + j0.293
For K=4 (2)
7
X (4)   x(n) e  j n
n 0
X (4) = x(0)+x(1) e-jπ +x(2) e-jπ2 +x(3) e-jπ3 +x(4) e-jπ4 +x(5) e-jπ5 +x(6) e-jπ6 +x(7) e-jπ7
= 1-1+1-1+1-1 = 0
For K=5
7
X (5)   x(n)e  j 5 n /4
n 0
X (5) = x(0)+x(1) e-j5π/4+x(2) e-j5π/2+x(3) e-j5πn/4+x(4) e-j5π+x(5) e-j25π/4+x(6) e-j15π/2
+x(7) e-j35π/4
= 1-0.707+j0.707-j+0.707+j0.707-1+0.707-j0.707
= 0.707-j0.293
For K=6 (2)
7
X (6)   x(n)e  j 3 n /2
n 0
 X(6) = x(0)+x(1) e-j3π/2+x(2) e-j3π+x(3) e-j9π/2+x(4) e-j6π+x(5) e-j15π+x(6) e-j9π+x(7) e-j21π/2
= 1+j-1-j+1+j
= 1+j
For K=7
7
X (7)   x(n)e j 7 n /4
n 0
X(7) = x(0) +x(1) e-j7π/4+x(2) e-j7π/2+x(3) e-j21π/4+x(4) e-j7π+x(5) e-j35π/4+x(6) e-j21π/2+x(7) e-
j49π/4

=1+0.707+j0.707+j-0.707+j0.707-1-0.707-j0.707
=-0.707+j1.707
(2)
X(K)={6,0.707-j1.707,1-j,0.707+j0.293,0,0.707-j0.293,1+j,-0.707+j1.707}
(ii) Find the circular convolution of the sequence using concentric circle method.
x1={1,1,2,1} and x2={1,2,3,4} (4) M/J’14
Solution: Given : x1={1,1,2,1} and x2={1,2,3,4}
y (n) =x1(n) N x2(n)

1 4 3 2  1  1  4  6  2  13
2 1 4 3  1   2  1  8  3  14 
  
3 2 1 4   2  3  2  2  4  11
      
4 3 2 1  1   4  3  4  1  12 

y(n) = {13,14,11,12}

y (0)   (11)  (1 2)  (2  3)  (1 4) 

x1(0)=1
y (0)  (1  2  6  4)
y (0)  13
1
x1(1)=1 4 2 x1(3)=1
3

x1(0)=1

x1(2)=2

2
x1(1)=1 3 x1(3)=1
 y (1)   (1 2)  (11)  (2  4)  (1 3) 
y (1)  (2  1  8  3)
y (1)  14

x1(0)=1

y (2)   (1 3)  (1 2)  (2 1)  (1 4) 

3
x1(1)=1
2 4 x1(3)=1 y (2)  (3  2  2  4)
1
y (2)  11

x1(2)=2

x1(0)=1

y (3)   (1 4)  (1 3)  (2  2)  (11) 

4 y (3)  (4  3  4  1)
x1(1)=1 1 x1(3)=1
3
2 y (3)  12

x1(2)=2 y(n) = {13,14,11,12}

4.(i) Find IDFT of the sequence X (K) = (5,0,1-j,0,1,0,1+j,0) (5)

Solution:
We have
1 N 1
x(n)=  X (k)e j 2 kn / N n  0,1,...., N  1
N k 0 (2)
For N=8
1 N 2

x(n)= 8 k 0
X (k) e j kn /4 n  0,1,....7
 7
1
For n=0; x(0)=  X (k) = [5+0+1-j+0+1+j+0]=1 (2)
k 0 8
1 7 1
For n=1; x(1)=  X (k) e j k /4 = [5+-(1-j)(j)+1(-1)+(1+j)(-j)]= 6/8= 0.75
8 k 0 8
1 7 1
For n=2; x(2)=  X (k) e j k /2 = [5+(1-j)(-1)+1(1)+(1+j)(-1)]=4/8 =0.5
8 k 0 8
7
1
For n=3; x(3)=  X (k) e j 3 k /4 = [5+(1-j)(-j)+1(-1)+ (1+j)(j)]= 2/8 = 0.25
8 k 0
1 7 1
For n=4; x(4)=  X (k) e j 5 k /4 = [5+(1-j)(1)+1(1)+(1+j)(1)]=8/8=1
8 k 0 8
7
1 1
For n=5; x(5)=  X (k) e j 5 k /4 = [5+(1-j)(j)+(1)(-1)+(1+j)(-j)]=6/8=0.75
8 k 0 8
1 7 1
For n=6; x(6)=  X (k) e j 3 k /2 = [5+(1-j)(-1)+1(1)+(1+j)(-j)]=4/8=0.5
8 k 0 8
7
1
For n=7;x(7)=  X (k) e j 7 k /2 =[5+(1-j)(-j)+1(1)+(1+j)(j)]=2/8=0.25
8 k 0
x(n) = {1,0.75,0.5,0.25,1,0.75,0.5,0.25}
(4)
ii) Use 4-point inverse FFT for the DFT result 6,  2  j 2,  2,  2  j 2 and
determine the input sequence. A/M”15 (5M)
Solution:
Given
4-point IDFT 6,  2  j 2,  2,  2  j 2
The decimation of given sequence is.
{6, -2} {-2+2j, -2-2j}
Take complex conjugate of above decimated sequence.
{6, -2} {-2-2j, -2+2j}

x(n)={o, 3, 2, 1} (4)

(iii) Give the relationship between DFT and other transforms. (3)
Refer A.Nagoor kani page no.:5.9-5.10
Solution:
Relationship to the Fourier transforms: (2)
 Fourier transform X(ejw) of a finite duration sequence x(n) having length N is given by
N 1
X(ejw) =  x(n)e jwn
n 0
The discrete fourier transform is given by
N 1
X(k) =  x (n) e  j 2 kn / N k=0,1,2,3,….N-1
n 0
Comparing the above two equations
X(k)=X(ejw)/w=2πk/N

Relationship to the z- transform: (2)

Let us consider a sequence x(n) of finite duration N with z-transform.
N 1
X(z)=  x(n)z
n 0
n

1 N 1
We have x(n) = 
N k 0
X (k)e j 2 kn / N

Substitute x(n) in X(z)

n 1
1 N 1
 
n 0 N k 0
X (k )e j 2 kn / N z  n
X(z)=
1 N 1 N 1
  X (k) e j 2 k / N ( z 1 ) n
N k 0 n 0

1 zN N 1
X (k)
=
N

k 0
j 2 k
z 1
1 e N
5.(i) Perform circular convolution of the following sequences x(n)= (1,1,2,1) and h(n)
= (1,2,3,4) using DFT & IDFT method. (7) M/J-14
Solution:
We know that X3(k) = X1(k)X2(k)
N 1
X1(k) =  x 1 (n)e  j 2 kn / N k=0,1,2,…N-1 (2)
n 0
Given x1(n) = {1,1,2,1} and N=4
2
X1(0) =  x 1 (n) = 1+1+2+1=5
n 0
3
X1(1) =  x 1 (n) e j n /2 = 1-j-2+j=-1
n 0
3
X1(2) =  x 1 (n)e  j n = 1-1+2-1=1
n 0
3
X1(3) =  x 1 (n)e  j 3 n /2 = 1+j-2-j=-1
n 0
X1(k)= (5,-1,1,-1) (2)
N 1 3
X2(k) = x
n 0
2 (n)e j 2 kn / N k  0,1, 2....N  1  x 2 (n) k=0,1,2,…N-1
n 0
X2(0) ==1+2+3+4 =10
 3
X2(1) = x
n 0
2 (n) e  j n /2 =1+2(-j)+3(-1)+4(j) = -2+2j
3
X2(2) =  x 2 (n) e  j n =1+2(-1)+3(1)+4(-1) = -2
n 0
3
X2(3) = x
n 0
2 (n) e  j 3 n /2 =1+2(j)+3(-1)+4(-j) =-2-2j

X2(k) = {10, -2+2j,-2, -2-2j}

X3(k)= X1(k) X2(k) = {50, 2-2j,-2, 2+2j} (2)
N 1
1
x3(n) =  X 3 (k)e j 2 kn / N n  0,1, 2....N  1
N k 0
1 3 1
x3(0) =  X 3 (k) = (50+2-j2-2+2+2j) = 13
4 k 0 4
4
1
x3(1) =  X 3 (k) e j k /2 =1/4[50+(2-j2)j+(-2)(-1)+(2+j2)(-j)] = 14
4 k 0
1 4
x3(2) =  X 3 (k)e j k =1/4[50+(2-j2)(-1)+(-2)(1)+(2+j2)(-1)] = 11
4 k 0
1 4
x3(3) =  X 3 (k) e j 3 k /2 =1/4[50+(2-j2)(-j)+(-2)(-1)+(2+j2)(j)] = 12 (2)
4 k 0
x3(n) = {13, 14, 11, 12}
(ii) Using linear convolution find y(n)=x(n)*h(n) for the sequences x(n)=(1,2,-1,2,3,-2,-3,-
1,1,1,2,-1) and h(n)=(1,2).Find the result by solving overlap-save method. (6)
Solution:
Linear convolution:

1 2 -1 2 3 -2 -3 -1 1 1 2 -1

1 1 2 -1 2 3 -2 -3 -1 1 1 2 -1

2 2 4 -2 4 6 -4 -6 -2 2 2 4 -2

y(n)= (1,4,3,0,7,4,-7,-7,-1,3,4,3,-2)
Overlap-save method
The input sequence can be divided in to blocks of data as follows.
M-1 zeros appended
x1(n)= (0, 1, 2, -1 )
3 data’s
x2(n) = (-1, 2, 3, -2)
3 data’s
M-1= 1 data from previous block
x3(n) = (-2,-3,-1, 1); x4(n)=(1,1,2,-1); x5(n)=(-1,0,0,0)
Given h (n) = (1, 2).Appending two zeros to the sequence we obtain
h (n) = (1, 2, 0, 0)
y1 (n) =x1(n) N h (n) (4)

1 0 0 2   0   2 
2 1 0 0   1   1 
 
0 2 1 0  2   4 
    
0 0 2 1   1  3 

y2(n)=x2(n) N h(n)

1 0 0 2   1  5
2 1 0 0   2   0 
 
0 2 1 0  3   7 
    
0 0 2 1   2   4 

y3(n)=x3(n) N h(n)

1 0 0 2   2   0 
2 1 0 0   3  7 
 
0 2 1 0   1  7 
    
0 0 2 1   1   1

y4(n)=x4(n) N h(n)

1 0 0 2   1   1
2 1 0 0   1   3 
 
0 2 1 0  2   4 
    
0 0 2 1   1  3 

y5(n)=x5(n) N h(n)

1 0 0 2   1  1
2 1 0 0   0   2 
 
0 2 1 0  0   0 
    
0 0 2 1  0   0 

y1(n)= -2 1 4 3
y2(n)= -5 0 7 4
y3(n)= 0 -7 -7 -1
y4(n)= -1 3 4 3

y5(n)= -1 -2 0 0
y(n)= 1 4 3 0 7 4 -7 -7 -1 3 4 3 -2 (4)

y(n)= {1,4,3,0,7,4,-7,-7,-1,3,4,3,-2}
6. (i) Compute the circular convolution of the sequences. x1(n) = {1,2,0,1} and
x2(n)={2,2,1,1} using DFT approach. (7)
Solution:
N=4
From circular convolution we have
DFT [ x1(n)©x2(n) ] = X1(K)X2(K) (4)
[ x1(n)©x2(n) ] = IDFT [X1(K)X2(K)]
Given x1(n) = {1,2,0,1}; N=4
N 1
X1(k) =  x(n) e
n 0
 j 2 kn / N
k  0,1,.....N  1
3
  x 1 (n) e j kn /2
n 0
X1(0) = 4
3
X1(1) =  x(n) e
n 0
 j n / 2

X1(1) = x(0)e0+x(1)e-jπ/2+x(2)e-j2π/2+x(3)e-j3π/2
= (1x1)+(2xe-jπ/2)+(0)+(1xe-j3π/2)
= (1-j1)
3
X2(2) =  x (n) e
n 0
1
 j n /2

= x(0)e0+x(1)e-jπ/2+x(2)e-j2π/2+x(3)e-j3π/2
= (1x1)+(2xe-jπ/2)+(0)+(1xe-j3π/2)
= -2
3
X1(3) =  x 1 (n) e j 3 n /2
n 0
= x(0)e0+x(1)e-j3π/2+x(2)e-j6π/2+x(3)e-j9π/2
= (1x1)+(2xe-j3π/2)+(0)+(1xe-j9π/2)
= 1+j1
N 1
X (k)   x(n) e  j 2 kn / N k  0,1,.....N  1
n 0
3
X2(0) =  x 2 (n) e  j kn /2
n 0
= x(0)e0+x(1)e0+x(2)e0+x(3)e0
=2+2+1+1
X2(0) = 6
3
X2(1) =  x 2 (n) e j n /2
n 0
=x(0)e0+x(1)e-jπ/2+x(2)e-j2π/2+x(3)e-j3π/2
=(2x1)+(2xe- jπ/2)+(1x e-j2π/2)+(1x e-j3π/2)
= 1-j1
3
X2(2) = x
n 0
2 (n) e  jk n /2

= x(0)e0+x(1)e-jπk/2+x(2)e-jπ2k/2+x(3)e-j3πk/2
= (2 x 1)+(2 x e-jπk/2) + (1 x e-jπ2k/2) + (1 x e-j3πk/2)
 =0
3
X2(3) =  x 2 (n) e j 3 kn /2
n 0
= x(0)e0+x(1)e-j3πk/2+x(2)e-j6πk/2+x(3)e-j9πk/2
= (2 x 1) + (2 x e-j3πk/2) + (1 x e-j6πk/2 ) + (1 x e-j9πk/2)
= 1+j1
X(k) = x1(k)x2(k)
X(0)=24; X(1)=-2j; X(3)=2j
1 N 1
 X (k) e j 2 kn/ N
N k 0
X(n)=
1 3
  X (k)e jk n /2
4 k 0
x(0) = 6; x(1)=7;x(2)=6;x(3)=5 (4)
x(n) = { 6,7,6,5}
(ii) If x (n) =cos (nπ/2) and n=0,1,2,3 find DFT. (6)
Solution:
x (n) =cos (nπ/2) and n=0,1,2,3
x(n) = {1,0,-1,0}
N 1
X(k) =  x(n) e
n 0
 j 2 kn / N
k  0,1, 2,3
3
  x(n) e  j kn /2 (1)
n 0
N 1
X (0)   x(n) e  j 2 (0) n / N
n 0

X (0)  1  0  1  0  0 (2)
X (0)  0
N 1 N 1
X (1)   x(n) e  j 2 (1) n /4   x(n) e  j n /2
n 0 n 0
 j (0)/2  j (1)/2
X (1)  1e  0e  1e  j (2)/2  0 e  j (3)/2
X (1)  1e 0  0 e  j /2  1e  j  0 e  j 3 /2 (2)
X (1)  1  1  2
X (1)  2
N 1 N 1
X (2)   x(n) e  j 2 (2) n /4   x(n) e  j n
n 0 n 0
 j (0)  j (1)  j (2)
X (2)  1e  0e  1e  0e j (3)
X (2)  1e0  0e j  1e j 2  0e j 3
X (2)  1  1  0
X (2)  0
 N 1 N 1
X (3)   x(n) e  j 2 (3) n /4   x(n) e  j 3 n /2
n 0 n 0
 j 3 (0)/2  j 3 (1)/2
X (3)  1e  0e  1e  j 3 (2)/2  0 e  j 3 (3)/2
X (3)  1e 0  0 e  j 3 /2  1e  j 3  0 e  j 9 /2 (2)
X (3)  1  1  2
X (3)  2
X(K)={0,2,0,2} (1)
7.(i) Determine the N-point DFT of the following sequences. x(n)=δ(n) and
x(n)= δ(n-n0) (3)
Solution:
i. X(n)= δ(n) (2)
N 1
X ( K )   x(n)e  j 2 kn / N k  0,1, 2,...N  1
n 0
N 1
X ( K )    (n) e  j 2 kn / N
n 0

= 1 i.e. X(k)=1 for 0≤k≤N-1

ii. x(n)= δ(n-n0) (2)
N 1
X ( K )   x(n)e  j 2 kn / N k  0,1, 2....N  1
n 0
N 1
   (n  n 0 )e  j 2 kn 0 / N
n 0 for 0≤k≤N-1
 j 2 kn 0 / N
e
(ii) Suppose you have a number of eight-point FFT chips. Suggest a scheme to
interconnect four such chips to compute a 32-point DFT. (4)
Solution:
Let x(n) has 32 samples for n=0….31. Divide the sequence x(n) in to four subsequences.
x1(n)=x(4n) n=0,1,2,….7
x2(n)=x(4n+1) n=0,1,2,….7
x3(n)=x(4n+2) n=0,1,2,….7
x4(n)=x(4n+3) n=0,1,2,….7 (2)
7 7 7 7
X (k )   x(4n) w 4Nnk   x(4n  1) w (4N n 1) k   x(4n  2) w (4N n  2) k   x(4n  3) w (4N n 3) k
n 0 n 0 n 0 n 0
7 7 7 7
  x1 (n) w nk
N   x2 ( n) w N w N   x3 ( n) w N w N   x4 ( n) w N w N
nk k 2k k 3k k

n 0 4 n 0 4 n 0 4 n 0 4

 X1 (k )  w kN w 3k
N (2)
2

Where X1 , X 2 , X 3 and X 4 are 8  po int DFTs of the subsequences x1 , x2 , x3 , and x4 respectively.

(iii) Consider the finite length sequence x(n)= {1,2,2,1,0}. The five point of DFT of x(n)
4 k
is denoted by X(k). Plot the sequence whose DFT is Y (k)  e 5
X (k ) . (6)

Solution:
 Using the circular time shifting property of DFT, we have
 j 2 k (2)
DFT [ x((n  2))5 ]  e 5
X (k )
Hence
y  n   x   n  2  5
y  0   x   0  2  5  x(3)  1
y 1  x  1  2  5  x((1))5  x(4)  0
y  2   x   2  2  5  x((0))5  x(0)  1
y  3  x   3  2  5  x((1))5  x(1)  2
y  4   x   4  2  5  x((2))5  x(2)  2
y(n)={1, 0,1, 2,2}
8.(i) Determine the DFT of the sequence x(n)=(1,1,0,0) and find the IDFT of
Y(k) = (1,0,1,0). (9)

Solution:
Let us assume N=L=4.
N-1
X K =  x(n)e
n=0
-j2πkn/N
;K=0,1,2,3
We have 3
X(K)= x(n)e-jπkn/2
n=0
3 3
X(0)=  x(n)e-j2π(0)n/N =  x(n)
n=0 n=0

X(0)=1+1+0+0=2
X(0)=2
3 3
X(1)=  x(n)e-j2π(1)n/4 =  x(n)e-jπn/2
n=0 n=0
-jπ(1)/2
X(1)=1e -jπ(0)/2
+1e +0e-jπ(2)/2 +0e-jπ(3)/2
X(1)=1e-0 +1e-jπ/2 +0e-jπ +0e-j3π/2
X(1)=1e-0 +1(cosπ/2  j sin π/2)+0(cosπ  j sin π)+0(cos3π/2  j sin 3π/2)
X(1)=1-j+0+0
X(1)=1-j
 3 3
X(2)=  x(n)e-j2π(2)n/4 =  x(n)e-jπn
n=0 n=0
-jπ(1)
X(2)=1e -jπ(0)
+1e +0e-jπ(2) +0e-jπ(3)
X(2)=1e-0 +1e-jπ +0e-j2π +0e-j3π
X(2)=1e-0 +1(cosπ  j sin π)+0(cos2π  j sin 2π)+0(cos3π  j sin 3π)
X(2)=1-1+0+0
X(2)=0
3 3
X(3)=  x(n)e-j2π(3)n/4 =  x(n)e-j3πn/2
n=0 n=0
-j3π(1)/2
X(3)=1e -j3π(0)/2
+1e +0e-j3π(2)/2 +0e-j3π(3)/2
X(3)=1e-0 +1e-j3π/2 +0e-j3π +0e-j9π/2 (4)
X(3)=1e-0 +1(cos3π/2  j sin 3π/2)+0(cos3π  j sin 3π)+0(cos9π/2  j sin 9π/2)
X(3)=1+j+0+0
X(3)=1+j
X(K)={2, 1-j, 0, 1+j}
Y(K) = {1,0,1,0}
1 N-1
yn = 
N K=0
Y(K)e j2πkn/N ;n=0,1,2,3
We know that
1 3 1 3
y  n  =  Y(K)e j2πkn/4   Y(K)e jπkn/2
4 K=0 4 K=0
3 3
1 1
y  0  =  Y(K)e j2πk(0)/4 =  Y(K)
4 K=0 4 K=0
1
y (0)  [1  0  1  0]
4
1
y (0)   0.5
2
1 3 1 3
y 1 =  Y(K)e j2πk(1)/4 =  Y(K)e jπk(1)/2
4 K=0 4 K=0
3
1
y 1 =  Y(K)e jπk/2
4 K=0
1
y(1)= (1e jπ(0)/2 +0e jπ(1)/2 +1e jπ(2)/2 +0e jπ(3)/2 ) (4)
4
1
y(1)= (1+0+1(cosπ+jsinπ)+0)
4
1
y(1)= (1+0-1+0)
4
y(1)=0
 1 3 1 3
y  2 = 
4 K=0
Y(K)e j2πk(2)/4   Y(K)e jπk
4 K=0
3
1
y  2  =  Y(K)e jπk
4 K=0
1
y(2)  (1e jπ(0)  0e jπ(1)  1e jπ(2)  0e jπ(3) )
4
1
y(2)  (1  0  1(cos 2  j sin 2 )  0)
4
1
y(2)  (1  0  1  0)
4
1
y(2)  (2)  0.5
4
1 3 1 3
y  3 =  Y(K)e j2πk(3)/4   Y(K)e jπk(3)/2
4 K=0 4 K=0
3
1
y  3 =  Y(K)e j3πk/2
4 K=0
1
y(3)  (1e j3π(0)/2  0e j3π(1)/2  1e j3π(2)/2  0e j3π(3)/2 ) (4)
4
1
y(3)  (1  0  1(cos 3  j sin 3 )  0)
4
1
y(3)  (1  0  1  0)
4
y(3)  0
y(n)={0.5, 0, 0.5, 0}
(ii) State and proof convolution property of DFT. (4)
Solution:
Let x1(n) and x2(n) are finite duration sequences both of length N with DFTs
X1(k) and X2(k).Now we find a sequence x3(n) for which the DFT is X3(k)
Where
X3(k) = X1(k) X2(k) - eqn (1)
N 1
X 3 p (n)   x1 p (m) x2 p ( n  m)
m0
or
N 1
x3 ((n)) N   x1 ((m)) N x2 ((n  m)) N
m0 (2)
For 0  n  N  1; x3 ((n)) N  x3 (n).similarly x1 ((m)) N  x1 (m)
N 1
x 3 (n)   x1 (m) x 2 (( n  m)) N equ.(2)
m 0

Eq(1) represents the circular convolution of x1 (n) andx 2 (n) represented as

x 3 (n)  x1 (n) N x 2 (n)
From eq. (1) and eq. (2) we find that (2)
DFT [(x1(n) N x2(n)] = X1(k) X2(k)
9.Consider the x1(n)={0,1,2,3,4}, x2(n)={0,1,0,0,0} and s(n)={1,0,0,0,0}
(i) Determine a sequence y(n) so that Y(K) = X1(K)X2(K).
(ii) Is there a sequence x3(n) such that S(K) = X1(K)X3(K)? (13) N/D ‘13
Solution:
Given : x1(n)={0,1,2,3,4}, x2(n)={0,1,0,0,0} and s(n)={1,0,0,0,0}
y (n) =x1(n) N x2(n)
using circular convolution Y(K)= X1(K)X2(K).

0 0 0 0 1   0  0  0  0  0  4   4 
1 0 0 0 0  1  0  0  0  0  0  0 

y ( n)   0 1 0 0 0   2    0  1  0  0  0   1 
      
0 0 1 0 0  3   0  0  2  0  0   2  (4)
0 0 0 1 0   4  0  0  0  3  0  3 
y(n)={4,0,1,2,3}
N-1
X K =  x(n)e
n=0
-j2πkn/N
;K=0,1,2,3,4
4
X1 (K)=  x(n)e-j2πkn/5
n=0
4 4
X1 (0)= x(n)e-j2π(0)n/5 = x(n)
n=0 n=0

X1 (0)=0+1+2+3+4=10
X1 (0)=10
4 3
X1 (1)=  x(n)e-j2π(1)n/5 =  x(n)e-j2πn/5
n=0 n=0

X1 (1)=0e -j2π(0)/5
+1e -j2π(1)/5
+2e-j2π(2)/5 +3e-j2π(3)/5  4e-j2π(4)/5
X1 (1)=0e0 +1e-j2π/5 +2e-j4π/5 +3e-j6π/5  4e-j8π/5
X1 (1)=0e-0 +1(cos2π/5  j sin 2π/5)+2(cos4π/5  j sin 4 π/5)
+3(cos6π/5  j sin 6π/5)  4(cos8π/5  j sin 8π/5)
X1 (1)=-0.309-j0.951-1.618-j1.176-20427+j1.763+1.236+j3.804
X1 (1)=-2.5+3.44j
4 3
X1 (2)=  x(n)e-j2π(2)n/5 =  x(n)e-j4πn/5
n=0 n=0

X1 (2)=0e -j4π(0)/5
+1e -j4π(1)/5
+2e -j4π(2)/5 +3e -j4π(3)/5  4e -j4π(4)/5
X1 (2)=0e0 +1e-j4π/5 +2e -j8π/5 +3e -j12π/5  4e -j16π/5
X1 (2)=0e-0 +1(cos4π/5  j sin 4π/5)+2(cos8π/5  j sin 8π/5)
+3(cos12π/5  j sin12π/5)  4(cos16π/5  j sin16π/5)
X1 (2)=-0.809-j0.588+0.618+j1.902+0.927-j2.853-3.236+j2.351
X1 (2)=-2.5+0.812j
4 3
X1 (3)= x(n)e-j2π(3)n/5 =  x(n)e-j6πn/5
n=0 n=0

X1 (3)=0e -j6π(0)/5
+1e -j6π(1)/5
+2e-j6π(2)/5 +3e-j6π(3)/5  4e-j6π(4)/5
X1 (3)=0e0 +1e-j6π/5 +2e-j12π/5 +3e-j18π/5  4e-j24π/5
X1 (3)=0e-0 +1(cos6π/5  j sin 6π/5)+2(cos12π/5  j sin12π/5)
+3(cos18π/5  j sin18π/5)  4(cos24π/5  j sin 24π/5)
X1 (3)=-0.809+j0.588+0.618-j1.902+0.927+j2.853-3.236-j2.351
X1 (3)=-2.5-0.812j
4 3
X1 (4)=  x(n)e-j2π(4)n/5 =  x(n)e-j8πn/5
n=0 n=0

X1 (4)=0e -j8π(0)/5
+1e -j8π(1)/5
+2e-j8π(2)/5 +3e-j8π(3)/5  4e-j8π(4)/5
X1 (4)=0e0 +1e-j8π/5 +2e-j16π/5 +3e-j24π/5  4e-j32π/5
X1 (4)=0e-0 +1(cos8π/5  j sin 8π/5)+2(cos16π/5  j sin16π/5)
+3(cos24π/5  j sin 24π/5)  4(cos32π/5  j sin 32π/5)
X1 (4)=-0.309+j0.951-1.618+j1.176-20427-j1.763+1.236-j3.804
(4)
X1 (4)=-2.5-3.44j
X1(K)={10,-2.5+3.44j,-2.5+0.812j,-2.5-0.812j,-2.5-3.44j}
s(n)={1,0,0,0,0}
N-1
S K  =  x(n)e
n=0
-j2πkn/N
;K=0,1,2,3,4
4
S(K)=  x(n)e-j2πkn/5
n=0
4 4
S(0)=  x(n)e-j2π(0)n/5 =  s(n)
n=0 n=0

S (0)=1+0+0+0+0=1
S (0)=1
4 3
S (1)=  x(n)e-j2π(1)n/5 =  x(n)e-j2πn/5
n=0 n=0

S (1)=1e -j2π(0)/5
+0e -j2π(1)/5
+0e-j2π(2)/5 +0e-j2π(3)/5  0e-j2π(4)/5
S (1)=1e0 +0e-j2π/5 +0e-j4π/5 +0e-j6π/5  0e-j8π/5
S (1)=1+0+0+0+0
S (1)=1
4 3
S (2)=  x(n)e -j2π(2)n/5
=  x(n)e-j4πn/5
n=0 n=0

S (2)=1e -j4π(0)/5
+0e +0e-j4π(2)/5 +0e-j4π(3)/5  0e-j4π(4)/5
-j4π(1)/5

S (2)=1e0 +0e-j4π/5 +0e-j8π/5 +0e-j12π/5  0e-j16π/5 =1+0+0+0+0

S (2)=1
4 3
S (3)= x(n)e-j2π(3)n/5 = x(n)e-j6πn/5
n=0 n=0

S (3)=1e -j6π(0)/5
+0e -j6π(1)/5
+0e-j6π(2)/5 +0e-j6π(3)/5  0e-j6π(4)/5
S (3)=1e0 +0e-j6π/5 +0e-j12π/5 +0e-j18π/5  0e-j24π/5
S (3)=1+0+0+0+0
 S (3)=1
4 3
S (4)=  x(n)e-j2π(4)n/5 =  x(n)e-j8πn/5
n=0 n=0

S (4)=1e -j8π(0)/5
+0e -j8π(1)/5
+0e-j8π(2)/5 +0e-j8π(3)/5  0e-j8π(4)/5
S (4)=1e0 +0e-j8π/5 +0e-j16π/5 +0e-j24π/5  0e-j32π/5 (2)
S (4)=1+0+0+0+0
S (4)=1
S(K)={1,1,1,1,1}
S(K)=X1(K)X3(K)
X3(K)=S(K)/X1(K)

1 1 1 1 1 
X3 (K )   , , , ,  (2)
10 2.5  3.44 j 2.5  0.812 j 2.5  0.812 j 2.5  3.44 j 
X3 ( K )  0.1, 0.138  0.19 j, 0.362  0.118 j, 0.362  0.118 j, 0.138  0.19 j
1 N-1
x3  n  =
N K=0
 X3 (K)e j2πkn/N ;n=0,1,2,3,4

1 4 1 4
x3  n  =  X3 (K)e j2πkn/5   X3 (K)e j2πkn/5
5 K=0 5 K=0
4
1 1 4
x3  0  =  X 3 (K)e j2πk(0)/5   X3 (K)
5 K=0 5 K=0
1
x3  0  =  0.1  0.138  0.19 j  0.362  0.118 j  0.362  0.118 j  0.138  0.19 j 
5
1
x3  0  =  0.9   0.18
5
x3  0  =  0.18
1 4 1 4
x3 1 =  3
5 K=0
X (K)e j2πk(1)/5
  X3 (K)e j2πk/5
5 K=0
1 0.1   0.138  0.19 j  e   0.362  0.118 j  e j4π/5  
j2π/5

x3 1 =  
5  0.362  0.118 j  e j6π/5   0.138  0.19 j  e j8π/5 
1 0.1   0.138  0.19 j  cos 2π/5  j sin 2π/5   0.362  0.118 j  (cos 4π/5  j sin 4π/5) 
x3 1 =  
5    0.362  0.118 j  cos 6π/5+jsin6π/5   0.138  0.19 j  cos8π/5+jsin8π/5 
1 0.1   0.138  0.19 j  0.309  0.951 j    0.362  0.118 j  ( 0.809  0.558 j ) 
x3 1 =  
5    0.362  0.118 j  0.809  0.588 j    0.138  0.19 j  0.309  0.951 j  
1
x3 1 =  0.1  0.138  0.189 j  0.224  0.308 j  0.224  0.308 j  0.138  0.189 j 
5
1
x3 1 =  0.882  0.16
5
x3 1 =0.16
 1 4 1 4
x3  2  = 
5 K=0
X3 (K)e j2πk(2)/5   X3 (K)e j4 k/5
5 K=0
1 0.1   0.138  0.19 j  e   0.362  0.118 j  e j8π/5  
j4π/5

x3  2  =  
5  0.362  0.118 j  e j12π/5   0.138  0.19 j  e j16π/5 
1 0.1   0.138  0.19 j  cos 4π/5  j sin 4π/5   0.362  0.118 j  (cos8π/5  j sin 8π/5) 
x3  2  =  
5    0.362  0.118 j  cos12π/5+jsin12π/5   0.138  0.19 j  cos16π/5+jsin16π/5 

1 0.1   0.138  0.19 j  0.809  0.558 j    0.362  0.118 j  (0.309  0.951 j ) 

x3  2  =  
5    0.362  0.118 j  0.309  0.951 j    0.138  0.19 j  0.809  0.588 j  
1
x3  2  =  0.1  0.223  0.073 j  0.00036  0.38 j  0.00036  0.38 j  0.223  0.073 j 
5
1
x3  2  =  0.55  0.11
5
x3  2  =0.11
1 4 1 4
x 3  3 = 
5 K=0
X 3 (K)e j2πk(3)/5
= 
5 K=0
X 3 (K)e j6πk/5

1 0.1   0.138  0.19 j  e   0.362  0.118 j  e j12π/5  

j6π/5

x3  3 =  
5  0.362  0.118 j  e j18π/5   0.138  0.19 j  e j24π/5 
1 0.1   0.138  0.19 j  cos 6π/5  j sin 6π/5    0.362  0.118 j  (cos12π/5  j si12π/5) 
x3  3 =  
5    0.362  0.118 j  cos18π/5+jsin18π/5    0.138  0.19 j  cos 24π/5+jsin24π/5  

1 0.1   0.138  0.19 j  0.809  0.558 j    0.362  0.118 j  (0.309  0.951 j ) 

x3  3 =  
5    0.362  0.118 j  0.309  0.951 j    0.138  0.19 j   0.809  0.588 j  
1
x3  3 =  0.1  0.000034  0.235 j  0.224  0.308 j  0.224  0.308 j  0.000034  0.235 j 
5
1
x3  3 =  0.348  0.06
5
x3  3 =  0.06
1 4 1 4
x3  4 = 
5 K=0
X 3 (K)e j2πk(4)/5 =  X 3 (K)e j8πk/5
5 K=0
1 0.1   0.138  0.19 j  e   0.362  0.118 j  e j16π/5  
j8π/5

x3  4  =  
5  0.362  0.118 j  e j24π/5   0.138  0.19 j  e j32π/5 
1 0.1   0.138  0.19 j  cos8π/5  j sin 8π/5    0.362  0.118 j  (cos16π/5  j sin16π/5) 
x3  4  =  
5    0.362  0.118 j   cos 24π/5+jsin24π/5    0.138  0.19 j  cos 32π/5+jsin32π/5 
1 0.1   0.138  0.19 j  0.309  0.951 j    0.362  0.118 j  (0.809  0.558 j ) 
x3  4  =  
5    0.362  0.118 j  0.809  0.588 j    0.138  0.19 j  0.309  0.951 j  
 1
x3  4  = 0.1  0.223  0.07 j  0.223  0.308 j  0.223  0.308 j  0.223  0.07 j 
5
1
x3  4  =  0.1  0.02
5
x3  4  =0.02
x3(n) = {-0.18, 0.16, 0.11, -0.06, 0.02} (4)
10.Compute the DFT of the 3-point sequence x(n)={2,1,2} using the same sequence compute
the 6 point DFT. (13)
Solution:
N=3, x(n) = {2,1,2}
N 1
x(k) =  x(n)e j 2 kn / N k  0,1, 2....N  1
n 0 (2)
2
=  x(n)e
n 0
 j 2 kn /3

x(0) = 5
2
x(1) =  x(n) e  j 2 n /3 =0.5+j0.866 (2)
n 0
2
x(2) =  x(n)e  j 4 n /3 =0.5-j0.866
n 0
X(k)={ 5, 0.5+j0.866, 0.5-j0.866}
For n=6
x(n) = {2,1,2,0,0,0} (2)
N 1
x(k) =  x(n)e
n 0
 j 2 kn / N

5
=  x(n)e
n 0
 j 2 kn /6

5
=  x(n)e
n 0
 j kn /3

= x(0)e0
x(0)=5
5
x(1) =  x(n) e
n 0
 j n /3

x(1)= x(0) e0  x(1) e j /3  x(2) e j 2 /3  x(3) e j 3 /3  x(4) e j 4 /3  x(5) e j 5 /3

= (2 x 1) +1 e-jπ3 +2e-j2π/3
= 1.5-j2.6 (4)
5
x(2) =  x(n)e  j 2 n /3
n 0
x(2)=x(0)e0+x(1) e-j2π/3+x(2) e-j4π/3+x(3)e-j6π/3+x(4)e-j8π/3+x(5)e-j10π/3
x(2)=(2x1) + (1xe-j2π/3) + (2x e-j4π/3) +0+0+0
= 0.5+j0.866
5
x(3) =  x(n) e
n 0
 j n

(2)
0 -jπ -j2π -j3π -j4π -j5π
x(3)=x(0)e + x(1)e +x(2)e +x(3)e +x(4)e +x(5)e
 x(3)=(2x1)+(1xe-jπ)+(2xe-j2π)+0+0+0
=3
5
x(4) =  x(n) e
n 0
 j 4 n /3

x(4)=x(0)e0+x(1)e-j4π/3+x(2)e-j8π/3+x(3)e-j12π/3+x(4)e-j16π/3+x(5)e-j20π/3
x(4)=(2x1)+(1xe-j4π/3)+(2xe-j8π/3)+0+0+0 = 0.5-j0.866
5
x(5) =  x(n) e
n 0
 j 5 n /3

(4)
0 -j5π/3 -j10π/3 -j15π/3 -j20π/3 -j25π/3
x(5)= x(0)e +x(1)e +x(2)e +x(3)e +x(4)e +x(5)e
x(5)= (2x1)+(1xe-j5π/3)+(2xe-j10π/3)+0+0+0
= 1.5+j2.6
X(k)={ 5, 1.5-j2.6, 0.5+j0.866,3, 0.5-j0.866, 1.5+j2.6}
11.(i) Find the Inverse DFT of X(k)={1,2,3,4} Dec-2009 (7)
The inverse DFT is defined as
N 1
1 j 2 nk / N , n0,1,2,3,...N 1
x ( n) 
N
 X ( k )e
k 0
(2)

1 3
Given N  4, x(n)  
4 k 0
X (k )e j 2 nk / N , n  0,1, 2,3

when n  0
1 3
x(0)  
4 k 0
X (k )e j (0) k /2

1
 (1  2  3  4)
4
5
x(0) 
2
when n  1
1 3 1
x(1)  
4 k 0
X (k )e j (1) k /2  (1  2e j /2  3e j  4e j 3 /2 )
4
1 1
 (1  2( j )  3(1)  4( j ))  (2  2 j )
4 4
1 1
x(1)    j
2 2
when n  2
1 3 1 (2)
x(2)  
4 k 0
X (k )e j k  (1  2e j  3e j 2  4e j 3 )
4
1 1
 (1  2(1)  3(1)  4(1))  (2)
4 4
1
x(2)  
2
 when n  3
1 3 1
x(3)  
4 k 0
X (k )e j 3 k /2  (1  2e j 3 /2  3e j 3  4e j 9 /2 )
4
(2)

1
 (1  2( j )  3(1)  4 j )
4
1 1 1
 (2  2 j )   j
4 2 2
1 1
x ( n)    j
2 2

5 1 1 1 1 1
x ( n)   ,   j ,  ,   j  (2)
2 2 2 2 2 2

(ii) Find the 4 point DFT of the two sequences x (n) and y (n) using a single 4-point
DFT. X(n)={1,-1,1,-1}and y(n)={0,2,4,6} (6)
Solution:
To find DFT of x(n) and y(n) using a single point DFT we construct a complex
sequence z(n)= x(n)+jy(n). That is x(n)=Re[z(n)] and y(n) = Im[z(n)]. Now we find
Z(k) = DFT [x(n)+jy(n)] = X(k)+j Y(k) (2)
From z(n) = x(n) + j y(n) we can find that
1
x (n)  [z(n)  z* (n)] and
2
1
y(n)  [z(n)  z* (n)]
2j
1 1
X(k)  [z(k)  z* (N  k)]  [z(k)  z* (( k)) N
2 2
1 1
Y(k)  [z(k)  z* (N  k)]  [z(k)  z* (( k)) N ]
2j 2j (2)
That is if we find Z(k) and use above relation we can obtain X(k) and Y(k). Let us form a
sequence
Z(n)=x(n) + j y(n)= {1,-1+j2,1+j4,-1+j6} (2)
3

Z(k) =  z (n) e
k 0
 j kn / 2

Z(0)=1+(-1+j2)+(1+j4)+(-1+j6) = j12
Z(1)= 1+(-1+j2)(-j)+(1+j4)(-1)+(-1+j6)(-1)= -4-j4
Z(2)= 1+(-1+j2)(-1)+(1+j4)(1)+(-1+j6)(-1)=4-j4
Z(3)= 1+(-1+j2)(j)+(1+j4)(-1)+(-1+j6)(-j)=4-j4
Z(k) = {j12, -4-j4, 4-j4, 4-j4}
z * (N  k) = {-j12, 4+j4,4+j4,-4+j4}
1
X(k)  [z(k)  z* (N  k)] = {0, 0, 4, 0}
2
X(k)= {0, 0, 4, 0}
 1 1 4  j 4 4  j4
Y(k)  [j24, 8  j8,  j8,8  j8] = { [12, , 4, ]}
2j 2j j j (2)
Y (k) = {12,-4+j4, -4, -4-j4}
12.(i) Compute the DFT of the four-point sequence x[n]={1,1,0,0} (4)A/M-11, 15
Solution:
N-1
X K =  x(n)e
n=0
-j2πkn/N
;K=0,1,2,3
We know that 3
X(K)= x(n)e-jπkn/2
n=0
3 3
X(0)=  x(n)e-j2π(0)n/N =  x(n)
n=0 n=0

X(0)=1+1+0+0=2
X(0)=2
3 3
X(1)=  x(n)e-j2π(1)n/4 =  x(n)e -jπn/2
n=0 n=0
-jπ(1)/2
X(1)=1e -jπ(0)/2
+1e +0e -jπ(2)/2 +0e -jπ(3)/2
X(1)=1e-0 +1e-jπ/2 +0e-jπ +0e-j3π/2
X(1)=1e-0 +1(cosπ/2  j sin π/2)+0(cosπ  j sin π)+0(cos3π/2  j sin 3π/2)
X(1)=1-j+0+0
X(1)=1-j
3 3
X(2)=  x(n)e -j2π(2)n/4
=  x(n)e-jπn
n=0 n=0
-jπ(1)
X(2)=1e -jπ(0)
+1e +0e-jπ(2) +0e-jπ(3)
X(2)=1e-0 +1e-jπ +0e-j2π +0e-j3π
X(2)=1e-0 +1(cosπ  j sin π)+0(cos2π  j sin 2π)+0(cos3π  j sin 3π)
X(2)=1-1+0+0
X(2)=0
3 3
X(3)=  x(n)e-j2π(3)n/4 =  x(n)e-j3πn/2
n=0 n=0
-j3π(1)/2
X(3)=1e -j3π(0)/2
+1e +0e-j3π(2)/2 +0e-j3π(3)/2
X(3)=1e-0 +1e-j3π/2 +0e-j3π +0e-j9π/2
X(3)=1e-0 +1(cos3π/2  j sin 3π/2)+0(cos3π  j sin 3π)+0(cos9π/2  j sin 9π/2) (4)
X(3)=1+j+0+0
X(3)=1+j
X(K)={2,1-j,0,1+j}

(ii) Summarize the difference between overlap save and add method. (6)
Overlap add Overlap save
 The overlap-add procedure cuts the signal The overlap-save procedure cuts the signal up
up into equal length segments with no into equal length segments with some overlap.
overlap.

Then it zero-pads the segments and takes Then it takes the DFT of the segments and saves
the DFT of the segments. Part of the the parts of the convolution that correspond to the
convolution result corresponds to the circular convolution
circular convolution

Results in the aliasing that occurs in No lost information in throwing away parts of the
circular convolution. linear convolution.

 x[n  rL] 0  n  L  1 xr [n]  x[n  r ( L  P  1)  P  1], 0  n  L 1

xr [n]  
 0 otherwise

 
y[n]  x[n]  h[n]   yr [n  rL] y[n]   yr [n  r ( L  P  1)  P  1]
m  r 0

Result of circularly convolving each section with

Fir filtering by using the overlapping-add h[n]. The portions of each filter section to be
method discarded in forming the linear convolution are
indicated
(iii) Compute the DFT of the four-point sequence x[n]={0, 1, 2, 3} (3)A/M-11,N/D-16
 ComputeThe DFT of the four  po int sequencesx (n)  {0,1, 2,3}.
Givenx(n)  {0,1, 2,3}
  j kn
X (k)   x(n)e N
k  0,1, 2,3
n 0

X (0)  0  1  2  3  6
3  j kn
X (1)   x(n)e 2

n0

 0  1( j )  2(1)  3( j )  2  j2
3
X (2)   x(n)e  j kn
n0

 0  1(1)  2(1)  3(1)  2

3  j kn
X (3)   x(n)e 2

n0

 0(1)  1( j )  2(1)  3(  j )  2  j 2
X (k )  {6, 2  j 2, 2, 2  j 2} (4)

DFT Using FFT

13.(i) Summarize the steps of radix – 2 DIT – FFT algorithm. (7)
1. In each computation two complex numbers”a” and “b” are considered. (4)
2. The complex number “b” is multiplied by a phase factor “WNkn”
3. The product “ bWNkn” is added to complex number “a” to form new complex number “A”
4. The product “ bWNkn” is subtracted from comblex number “a” to form new complex
number “B”.
The above basic computation can be expressed by a signal flow graph (2)

The signal flow graph is also called butterfly diagram since it resembles a butterfly. In
radix-2 FFT, Z/2 butterflies per stage are required to represent the computational process.
The butterfly diagram used to compute the 8point DFT via radix-2 DIT FFT . (2)

The sequence x(n)is arranged in

rsed order and then decimated into two sample sequences .

x(0) x(2) x(1) x(3)

x(4) x(6) x(5) x(7) (2)

Fist stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)

Second stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)

Third stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)
Combined stage for computation:

(ii)Compute 4-point DFT of the sequence x(n)={0,2,4,6} (6)

Solution:
 Let us assume N  L  4
N 1
DFT X (k )   x(n)e  j 2 kn / N (2)
n 0
3
X (0)   x(n)
n 0

 x(0)  x(1)  x(2)  x(3)

 0246
X (0)  12 (1)
3
X (1)   x(n)e  j n /2
n 0

 x(0)  x(1)e  j /2  x(2)e  j  x(3)e  j 3/2

 0  2( j )  4(1)  6( j )
X (1)  4  4 j (1)
3
X (2)   x(n)e  j n
n 0

 x(0)  x(1)e  j  x(2)e  j 2  x(3)e  j 3

 0  2(1)  4(1)  6(1)
X (2)   4 (1)
3
X (3)   x(n)e  j 3 n /2
n 0

 x(0)  x(1)e  j 3 /2  x(2)e  j 6 /2  x(3)e  j 9 /2

 0  2( j )  4(1)  6( j )
X (3)  4  4 j
X (k )  {12, 4  4 j, 4, 4  4 j} (1)

DIT & DIF - FFT using radix 2 – Butterfly structure:

Draw and explain the basic butterfly diagram or flow graph of DIT radx-2 FFT.
14. Refer A.Nagoor Kani page no.: 5.27-5.29 (13M)
1. In each computation two complex numbers”a” and “b” are considered. (4)
2. The complex number “b” is multiplied by a phase factor “WNkn”
3. The product “ bWNkn” is added to complex number “a” to form new complex number “A”
4. The product “ bWNkn” is subtracted from comblex number “a” to form new complex
number “B”.
The above basic computation can be expressed by a signal flow graph (2)
The signal flow graph is also called butterfly diagram since it resembles a butterfly. In
radix-2 FFT, Z/2 butterflies per stage are required to represent the computational process.
The butterfly diagram used to compute the 8point DFT via radix-2 DIT FFT . (2)

The sequence x(n)is arranged in bit reversed order and then decimated into two sample
sequences .

x(0) x(2) x(1) x(3)

x(4) x(6) x(5) x(7) (2)

Fist stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)

Second stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)
Third stage of flow graph for 8-point DFT via radix-2 DIT FFT (2)

Combined stage for computation:

 (i) Find the DFT of the sequence x(n) = {1,2,3,4,4,3,2,1} using radix 2 decimation in
15. time FFT algorithm. (7)M/J-13
Solution:
Given {1, 2,3, 4, 4,3, 2,1}
Decimation in radix 2
 {1,3, 4, 2}{2, 4,3,1}

 {1, 4}{3, 2}{2,3}{4,1}

 2 
 j k
We knowthat W N  e
k  N 

Given N  8
 2 
 j 0
Hence,W 8  e 1
1  8 

 2 
 j 1  
W 8  e  cos  j sin  0.707  j 0.707
1  8 
4 4
 2 
 j 2  
W 8  e  cos  j sin j
2  8 
2 2 (2)
 2 
 j 3 3 3
W 8  e  cos  j sin  0.707  j 0.707
3  8 
4 4
 (4)

COMPUTATION

Input Output of stage 1 Output of stage 2 Output of stage 3

1 1+4=5 5+5=10 10+10=20

4 1-4=-3 -3+(-j)1=-3-j -3-j+(0.707-j0.707)(-1-3j)
=-5.828-j2.414
3 3+2=5 5-5=0 0
2 3-2=1 -3-(-j)1=-3+j (-3+j)+(-0.707-j0.707)(-1+3j)
=-0.707-j0.414
2 2+3=5 5+5=10 10-10=0
3 3-2=-1 -1+(-j)3=-1-3j -3-j-(0.707-j0.707)(-1-3j)
=-0.172+j0.414
4 4+1=5 5-5=0 0
1 4-1=3 -1-(-j)3=-1+3j (-3+j)-(-0.707-j0.707)(-1+3j)
=-5.828+j2.414

X (k )  {20, 5.828  j 2.414, 0, 0.172  j 0.414, 0, 0.172  j 0.414, 0  5.828  j 2.414} (2)

(ii) Given x(n)=n+1, and N=8, find X(K) using DIT, FFT algorithm. A/M”15 (6M)
Solution:
Given
x(n)=n+1, N=8
To find X(K)
x(n) = {1,2,3,4,5,6,7,8}
By decimating the above sequence, we get
x(n) = {1,3,5,7}{2,4,6,8} again decimating the sequence
x(n) = {1,5}{3,7}{2,6}{4,8}
 2 
 j k
We knowthat W N  e
k  N 

Given N  8
 2 
 j 0
Hence,W 8  e 1
1  8 

 2 
 j 1   (2)
W 8  e  cos  j sin  0.707  j 0.707
1  8 
4 4
 2 
 j 2  
W 8  e  cos  j sin j
2  8 
2 2
 2 
 j 33 3
W 8  e  cos
 j sin  0.707  j 0.707
3  8 
4 4
First stage computation (2)

Output sequence of first stage is {6, -4, 10, -4, 8, -4, 12, -4}

Second stage computation (2)

Output sequence of second stage is {16, -4+4j, -4, -4-4j, 20, -4+4j, -4, -4-4j}
Third stage computation (2)

Output sequence of third stage is

X(k) = {36, -4+9.656j, -4+4j, -4+1.656j, -4, -4-1.656j, -4-4j, -4-9.656j }

16.(i)Describe the need for Bit reversal and the Butterfly structure. For a sequence
x(n) = {4, 3, 2, 1, -1, 2, 3, 4}. Obtain the 8pt FFT computation using DIT method.
(7) N/D-16
Given
x(n) = {4, 3, 2, 1, -1, 2, 3, 4}
By decimating the above sequence, we get
x(n) = {4,2,-1,3}{3,1,2,4} again decimating the sequence
x(n) = {4,-1}{2,3}{3,2}{1,4}
 2 
 j k
We knowthat W N  e
k  N 

Given N  8
 2 
 j 0
Hence,W 8  e 1
1  8 

 2 
 j 1   (2)
W 8  e  cos  j sin  0.707  j 0.707
1  8 
4 4
 2 
 j 2  
W 8  e  cos  j sin j
2  8 
2 2
 2 
 j 3 3 3
W 8  e  cos  j sin  0.707  j 0.707
3  8 
4 4

(ii)Find the 8-point DFT of a sequence using radix-2 DIT Algorithm (8)
 1; 0n2
x n   A/M’17
0; otherwise

17.(i) Compute IDFT of the sequence X (k) = {7,-0.707-j0.707,-j, 0.707-j0.707, 1,

0.707+j0.707, j,-0.707+j0.707} using DIT algorithm. (7)
Solution: X (k) = {7,-0.707-j0.707,-j, 0.707-j0.707, 1, 0.707+j0.707, j,-0.707+j0.707}
 2 
 j k
We knowthat W N  e Given N  8
k  N 

 2 
 j 0
Hence,W 8  e 1
0  8 

 2 
 j 1  
e  cos  j sin  0.707  j 0.707
1  8 
W 8
4 4

 2 
 j 2  
W8e  cos  j sin j
2  8 
2 2 (2)
 2 
 j 3 3 3
e  cos  j sin  0.707  j 0.707
3  8 
W 8
4 4

(4)

Input Output of stage 1 Output of stage 2 Output of stage 3

7 7+1=8 8+0=8 8+0=8
1 7-1=6 6+(-j)2j=8 8+0=8
J j-j=0 8-0=8 8+0=8
-j j-(-j)=2j 6-(-j)(2j)=4 4+(-0.707-j0.707) -
(-2.828+j2.828)= 8
-0.707+j0.707 (0.707+j0.707) + 8-0=8
(0.707-j0.707) 0
=0
0.707-j0.707 (-0.707+j0.707) - (-1.414+j1.414) + (-j) 8-0=8
(0.707-j0.707) (1.414+j1.414)
=-1.414+j1.414 =0
0.707+j0.707 (0.707+j0.707) + 8-0=8
(-0.707-j0.707)=0 0
-0.707-j0.707 (0.707+j0.707) - (-1.414+j1.414) - (-j) 4 - (-0.707-j0.707) -
(-0.707-j0.707) (1.414+j1.414) (-2.828+j2.828) = 0
=1.414+j1.414 =2.828+j2.828

Nx*(n)={8,8,8,8,8,8,8,0} (2)
x(n)={1,1,1,1,1,1,1,0}
(ii) Compute the 8-point DFT of the sequence x (n) = (0.5, 0.5, 0.5, 0.5, 0, 0, 0, 0) using the
radix-2 DIT algorithm. June-2011,June-2013 (6)

Solution:
 2 
 j k
We knowthat W N  e
k  N 

 2 
 j 0
when N  2; W2e 1
0  2 

when N  4,
 2   2 
 j 0  j 1  
e  1; e  cos  j sin j
0  4  1  4 
W 4 W 4
2 2
whenN  8
 2 
 j 0
e 1
0  8 
W 8
 2 
 j 1  
e  cos  j sin  0.707  j 0.707
1  8 
W 8
4 4
 2 
 j 2  
W8e  cos  j sin j
2  8 
2 2
 2 
 j 3 3 3
W8e  cos  j sin  0.707  j 0.707
3  8 
(2)
4 4
 (4)
INPUT OUTPUT AT OUTPUT AT OUTPUT AT
STAGE 1 STAGE 2 STAGE 3
x0  0.5 0.5+0=0.5 0.5+0.5=1 1+1=2
x4  0 0.5-0=0.5 0.5-0.5j 0.5-0.5j+(0.5-0.5j)(0.707-j0.707)
=0.5-1.207j
x2  0.5 0.5+0=0.5 0.5-0.5=0 0+0=0
x6  0 0.5-0=0.5 0.5+0.5j 0.5+0.5j+(0.5-0.5j)(-0.707-j0.707)
=0.5-0.207j
x1  0.5 0.5+0=0.5 0.5+0.5=1 1-1=0
x5  0 0.5-0=0.5 0.5-0.5j 0.5-0.5j-(0.5-0.5j)(0.707-j0.707)
=0.5+0.207j
x3  0.5 0.5+0=0.5 0.5-0.5=0 0-0=0
x7  0 0.5-0=0.5 0.5+0.5j 0.5+0.5j-(0.5-0.5j)(-0.707-j0.707)
=0.5+1.207j

X(k)={2,0.5-j1.207,0,0.5-j0.207,0,0.5-j0.207,0,0.5-j1.207} (2)

18.(i) Find the DFT for 1,1, 2,0,1, 2,0,1 using FFT DIT butterfly Algorithm and plot the
spectrum. (7)Dec-2013
  2   2 
 j k  j 0
We knowthat W N  e ; when N  2; W2e 1
k  N  0  2 

when N  4,
 2   2 
 j 0  j 1  
e  1; e  cos  j sin j
0  4  1  4 
W 4 W 4
2 2
whenN  8
 2   2 
 j 0  j 1  
W8e  1; W8e  cos  j sin  0.707  j 0.707
0  8  1  8 
4 4
 2 
 j 2  
W8e  cos  j sin j
2  8 
2 2
 2 
 j 3 3 3
W8e  cos  j sin  0.707  j 0.707
3  8 
4 4 (2)

(4)

INPUT OUTPUT AT OUTPUT AT OUTPUT AT

STAGE 1 STAGE 2 STAGE 3
x0  1 1+1=2 2+2=4 4+4=8
x4  1 1-1=0 0+2(-j)=-2j -2j+(-1+j)(0.707-j0.707)
=-0.586j
x2  2 2+0=2 2-2=0 0+(2)(-j)=-2j
x6  0 2-0=2 0-2(-j)=2j =2j+(-1-j)(-0.707-j0.707)
=3.414j
x1  1 1+2=3 3+1=4 4-4=0
x5  2 1-2=-1 -1+(-1)(-j)=-1+j -2j-(-1+j)(0.707-j0.707)
=-3.414j
x3  0 0+1=1 3-1=2 0-(2)(-j)=2j
x7  1 0-1=-1 -1-(-1)(-j)=-1-j =2j-(-1-j)(-0.707-j0.707)
=0.586j
(2)
X  k   8, 0.586 j, 2 j,3.414 j,0, 3.414 j, 2 j,0.586 j
ii) Find the 4-point inverse DFT of X(K)={10,-2+2j,-2,-2-2j}. (6) N/D’14
Solution:
Given: X (K) = {10,-2+2j,-2,-2-2j}
1 N-1
x n =
N K=0
 X(K)e j2πkn/N ;n=0,1,2,3
1 3 1 3
x  n  =  X(K)e j2πkn/4   X(K)e jπkn/2
4 K=0 4 K=0
3
1 1 3
x  0  =  X(K)e jπk(0)/2   X(K)
4 K=0 4 K=0
1
x  0  = 10  2  2 j  2  2  2 j 
4
1
x  0  = [4]  1
4
x  0  =1

1 3 1 3
x 1 = 
4 K=0
X(K)e jπk(1)/2
  X(K)e jπk/2
4 K=0
1
x 1 = 10  (2  2 j )e jπ/2  (2)e jπ  (2  2 j )e j3π/2 
4
1
x 1 = 10  (2  2 j )(cos π/2+jsinπ/2)  (2)(cos π+jsinπ)+(2  2 j )(cos 3π/2  j sin 3π/2)
4
1
x 1 = [10  (2  2 j )( j )  (2)(1)  (2  2 j )( j )]
4
1 1
x 1 = [10  2 j  2  2  2 j  2]  [8]
4 4
x 1 =2 (4)
1 3 1 3
x  2 = 
4 K=0
X(K)e jπk(2)/2
 
4 K=0
X(K)e jπk

1
x  2  = 10  (2  2 j )e jπ  (2)e j2π  (2  2 j )e j3π 
4
1
x  2  = 10  (2  2 j )(cos π+jsinπ)  (2)(cos 2π+jsin2π)+(2  2 j )(cos 3π  j sin 3π) 
4
1
x  2  = [10  (2  2 j )( 1)  ( 2)(1)  ( 2  2 j )( 1)]
4
 1 1
x  2  = [10  2  2 j  2  2  2 j ]  [12]
4 4
x  2  =3
1 3 1 3
x  3 = 
4 K=0
X(K)e jπk(3)/2   X(K)e j3πk/2
4 K=0
1
x  3 = 10  (2  2 j )e j3π/2  (2)e j3π  (2  2 j )e j9π/2 
4
1
x  3 = 10  (2  2 j )(cos 3π/2+jsin3π/2)  (2)(cos 3π+jsin3π)+(2  2 j )(cos 9π/2  j sin 9π/2) 
4
1
x  3 = [10  (2  2 j )( j )  (2)(1)  (2  2 j )( j )]
4
1 1
x  3 = [10  2 j  2  2  2 j  2]  [16]
4 4 (4)
x  3 =4
x(n)={1,2,3,4}
19.(i) Find the 8-point DFT of a sequence using radix-2 DIT Algorithm (7)
1; 0  n  2
x n   May-2015
0; otherwise
 2   2 
 j k  j 0
We knowthat W N  e ; when N  2; W2e 1
k  N  0  2 

when N  4,
 2   2 
 j 0  j 1  
W4e  1; W4e  cos  j sin j
0  4  1  4 
2 2
whenN  8
 2   2 
 j 0  j 1  
e  1; e  cos  j sin  0.707  j 0.707
0  8  1  8 
W 8 W 8
4 4
 2 
 j 2   (2)
W8e  cos  j sin j
2  8 
2 2
 2 
 j 3 3 3
e  cos  j sin  0.707  j 0.707
3  8 
W 8
4 4
 (4)
INPUT OUTPUT AT OUTPUT AT OUTPUT AT
STAGE 1 STAGE 2 STAGE 3
x0  1 1+0=1 1+1=2 2+1=3
x4  0 1-0=1 1+1(-j)=1-j 1-j+(1)(0.707-j0.707)
=1.707-1.707j
x2  1 1+0=1 1-1=0 0+1(-j)=-j
x6  0 1-0=1 1=1(-j)=1+j 1+j+(1)(-0.707-0.707j)
=0.293+0.293j
x1  1 1+0=1 1+0=1 2-1=1
x5  0 1-0=1 1+0(-j)=1 1-j-(1)(0.707-j0.707)
=0.293-0.293j
x3  0 0+0=0 1-0=1 0-1(-j)=j
x7  0 0-0=0 1-0(-j)=1 1+j-(1)(-0.707-0.707j)
=1.707+1.707j
X  k   3,1.707  1.707 j,  j,0.293  0.293 j ,1,0.293  0.293 j , j ,1.707  1.707 j (2)
(ii) An 8-point sequence x(n)={1, 1, 1, 1, 1, 1, 1, 1}. Compute 8-point DFT of x(n) by
radix 2 DIT FFT. (6)
A/M-11
Solution:
  2 
 j k
We knowthat W N  e
k  N 

Given N  8
x(n)  {1,1,1,1,1,1,1,1}
 2 
 j 0
Hence,W 8  e 1
1  8 
(2)
 2 
 j 1  
W8e  cos  j sin  0.707  j 0.707
1  8 
4 4
 2 
 j 2  
W8e  cos  j sin j
2  8 
2 2
 2 
 j 3 3 3
W8e  cos  j sin  0.707  j 0.707
3  8 
4 4

x(0)=1 2 4
8
1
0 0
x( 4)=1
W80 -1 0

0
x(2)=1 2
1 W80
-1 0
0
0 0
x(6)=1 -1
W8 0 -1 W82

2 4
x(1)=1 0
1 W80 -1
0 0
x(5)=1
W8 0 -1 W81
0
-1
2 0
x(3)=1 1 -1 0
W80 W82 -1
0 0
x(7)=1 W80
0
-1 W82 -1 W83 -1

(4)

Input Output of stage 1 Output of stage 2 Output of stage 3

1 1+1=2 2+2=4 4+4=8

1 1+1=2 2+2=4 4-4=0

1 1+1=2 2-2=0 0

1 1+1=2 2-2=0 0
1 1-1=0 0 0

1 1-1=0 0 0

1 1-1=0 0 0

1 1-1=0 0 0

X(K)={8,0,0,0,0,0,0,0} (2)

20.Find the IDFT of the sequence

X(K) = {4, 1-j2.414, 0, 1-j0.414, 0, 1+j0.414, 0, 1+j2.414 } using DIF algorithm

(13)(M\J-16)

Solution:

Given Output at stage 1 Output at stage 2 Output at stage 3

4 4 4 8

1-j2.414 2+j2 4 0

0 0 4 8

1-j0.414 2-j2 4 0

0 4 4 8

1+j0.414 2+j2 4 0

0 0 4 8

1+j2.414 2-j2 4 0
The output 8x*(n) is in bit reversal
Therefore
X(n)={1,1,1,1,0,0,0,0}
21.(i) Find the 8-point DFT of the sequence x(n)={1, -1, 1, -1, 0, 0, 0, 0} using radix 2
DIT algorithm. (7)N/D-13
 2 
 j k
We knowthat W N  e
k  N 

Given N  8
x(n)  {1, 1,1, 1, 0, 0, 0, 0}
 2 
 j 0
Hence,W 8  e 1
0  8 

 2 
 j 1  
e  cos  j sin  0.707  j 0.707
1  8 
W 8
4 4
 2 
 j 2  
e  cos  j sin j
2  8 
W 8
2 2
(2)
 2 
 j 3 3 3
e  cos  j sin  0.707  j 0.707
3  8 
W 8
4 4
INPUT OUTPUT AT OUTPUT AT OUTPUT AT
STAGE 1 STAGE 2 STAGE 3
x0  1 1+0=1 1+1=2 2-2=0
x4  0 1-0=1 1+1(-j)=1-j 1-j+(-1+j)(0.707-j0.707)
=1+j0.414
x2  1 1+0=1 1-1=0 0+0=0
x6  0 1-0=1 1=1(-j)=1+j 1+j+(-1-j)(-0.707-0.707j)
=1+j2.414
x1  1 -1+0=-1 -1(-1)=-2 2-(-2)=2+2=4
x5  0 -1-0=-1 -1+(-1)(-j)=-1+j 1-j-(-1+j)(0.707-j0.707)
=1-j2.414
x3  1 -1+0=-1 -1-(-1)=-1+1=0 0-0=0
x7  0 -1-0=-1 -1-(-1)(-j)=-1-j 1+j-(-1-j)(-0.707-0.707j)
=1-j0.414

(4)
X(K)={0,1+j0.414,0,1+j2.414,4,1-j2.414,0,1-j0.414} (2)
ii) Determine the response of LTI system when the input sequence is x(n)={-1,1,2,1,-1}
using radix 2 DIF FFT. The impulse response is h(n)={-1,1,-1,1}. (6) N/D-10.
Solution:
Given:
x(n)={-1,1,2,1,-1} and h(n)={-1,1,-1,1}
x(n)  {1,1, 2,1, 1, 0, 0, 0}
 2 
 j 0
Hence,W 8  e 1
0  8 

 2 
 j 1  
e  cos  j sin  0.707  j 0.707
1  8 
W 8
4 4
(2)
 2 
 j 2  
e  cos  j sin j
2  8 
W 8
2 2
 2 
 j 3 3 3
e  cos  j sin  0.707  j 0.707
3  8 
W 8
4 4
 X(0)=-1 0 2 4
1 2 0
X(1)=1 W 8 0
2 -2 0
X(2)=2 1 0
W 8 -2
-2
X(3)=1 -2 -2 -2-2j W
2
W
0

1 8 8

0
-2-j3.414
X(4)=-1 0.707-j0.707 -j1.414 W
1 8
-2-j0.586
X(5)=0 2 -2j -2+2j 0
W 8
-2+j0.586
X(6)=0 1 -0.707-j0.707 1.414 2 -j1.414 0
W 8 W 8 -2.+j3.414
X(7)=0
X(K)={4,-2-j3.414,-2,-2+j0.586,0,-2-j0.586,-2,-2+j3.414}
h(n)={-1,1,-1,1,0,0,0,0} -2
X(0)=-1 -1

1 2 0
0
X(1)=1 W 8 -4
-1 0
0
X(2)=-1 1 0
W 8
j
0
0
X(3)=1 W
2
W
0

1 -1 -2 -2+j 8 8

0
-2-j0.414
X(4)=0 0.707-j0.707 -j1.414 W
1 8
-2+j2.414
X(5)=0 -1 j -2-j
0
W 8
-2-j2.414
X(6)=0 1 -0.707-j0.707 1.414 2 -j1.414 0
W 8 W 8 -2+j0.414
X(7)=0
{0,-2-j0.414,0,-2-j2.414,-4,-2+j2.414,0,-2+j0.414}
X(K)={4,-2-j3.414,-2,-2+j0.586,0,-2-j0.586,-2,-2+j3.414}
Y(K)=H(K)X(K)

Y(K)={0,2.587+j7.656,0,5.414+j3.656 ,0,5.414-j3.656 ,0, 2.587-j7.656}

0
0 0
16
8+j4 16 0
2.587+j7.656 W 8 -16
0 0
0

8-j4 j8 W 8
8
8
5.414+j3.656 -8
2 0
0 -2 0 W 8 W 8

12
0 -2.827+j11.312 6+j10 12 W
0
8
5.414-j3.656 -12
0
0 j 0 W 8
-20
0 2.827+j11.312 6-j10 j20
 2 0
2.587-j7.656 W 8 W 8 20
-20
y(n)=1/8{16,12,8,-20,-16,-12,-8,20} (2)

y(n)={2,1.5,1,-2.5,-2,-1.5,-1,2.5}

PART- C (15 Marks)

1. Compute the N-point DFTs of the signals
i. x(n)   (n)
ii. x(n)   (n  n 0 ), 0  n0  N
iii. x(n)  a n , 0 n  N 1
2
j( ) k0
iv. x(n)  e N
, 0 n  N 1
2
v. x(n)  cos k0 n, 0  n  N 1
N
2
vi. x(n)  sin k0 n, 0  n  N 1
N
Solution:
N 1 2
j
( a ) X ( k )    ( n) e
kn
N

n 0

 1 0  k  N-1
N 1 2
j
(b) X (k )    (n  n0 )e
kn
N

n 0
2
j kn
e N
0  k  N-1
N 1 2
j
(c ) X ( k )   a e
kn
n N

n 0
n
  j 2 k 
N 1
X (k )    a n N 
n 0  
2
j kn
e N
0  k  N-1
1 aN
= 2
j kn
1  ae N

N 1  j 2 kn
(d ) X (k )   e N

n 0
2 N
j k
1 e N 2
 2
j kn
1 e N
1  (1) k
 2
j kn
1 e N
 N 1 2 2
k0 n  j
( e) X ( k )   e
j kn
N N
e
n 0
N 1 2
( k  k0 ) n
 e
j
N

n 0

 N  ( k  k0 )

1 j 2 kn 1  j 2 kn
( f ) x ( n)  e N  e N
2 2
Hence from (e)
We obtain
N
X(k)=  (k  k0 )   (k  N  k0 ) 
2

2. Consider the finite duration signal x(n)  {1, 2,3,1}

i. Compute its four point DFT by solving explicitly the 4-by-4 system of linear
equations defined by the inverse DFT formula.
ii. Check the answer in part (a) by computing the four-point DFT, using its definition.

Solution:
N 1 2
1
 X(k)e
j kn
(a ) x(n)  N
N k 0
N 1 2

 X(k)e
j kn
N
 Nx(n)
k 0

X (0)  X (1)  X (2)  X (3)  4

 
j j
X (0)  X (1) e 2  X (2) e j  X (3) e 2
8
j j 2 j 3
X (0)  X (1) e  X (2) e  X (3) e  12
3 9
j j
X (0)  X (1) e 2
 X (2) e j 3  X (3) e 2
4
1 1 1 1 X (0) 4
1 j -1 -j X (1) 8

1 -1 1 -j X (2) 12
1  j 1 j X (3) 4
X (0) 7
X (1) 2  j

X (2) 1
X (3) 2  j
 3 2
j
(b) X (k )   x( n)e
kn
4

n 0
3
X (0)   x(n)
n 0

7
3 
j n
X (1)   x(n)e 2

n 0

 2  j
3
X (2)   x(n)e  j n
n 0

1
3 3
j
X (3)   x(n)e
n
2

n 0

 2  j

3. (a)The z-transform of the sequence x(n) = u(n) – u(n-7) is sampled at five points on the
unit circle as follows: x(k)  X(z) z  e j 2 k 5 , k  0,1, 2,3, 4
Determine the inverse DFT x’(n) of X(k). compare it with x(n) and explain the results.
(b)Consider a finite duration sequence x(n),0≤n≤7, with Z transform X(z). We wish to
compute X(Z) at the following set of values;
Z k  0.8e j ( 2k / 8) (  / 8)] for 0≤k≤7.Sketch the points in Z planand also determine the
sequence.
Solution:
X(z)  1  Z 1,  ...........  Z 6
X (k )  X ( z )
2
j
where Z  e 5

2 4 6 12
j j j j
 1 e 5
e 5
e 5
 ....  e 5

2 4 6 8
j j j j
 2  2e 5  e 5  e 5
 ....  e 5

x '(n)  {2, 2,1,1,1}

x '(n)   x(n  7m)
m

n  0,1....4

Temporal aliasing occurs first two points of x’(n0 because X(z) is not sampled at
sufficiently small spacing on the unit circle.
(b) Z k  0.8e j ( 2k / 8) / 8)]
 X ( K )  X (Z ) z  zk
n
7   
 2k  
j
  x(n) 0.8e  8 8  
n 0  

j n
x(n)  x(n)0.8e 8

4. The basic butterfly in the radix-2 decimation – in – time FFT algorithm is also
X n1 (k)  X n (k)  WNm X n (l)
X n1 (k)  X n (k)  WNm X n (l)
1 1
i. If we require that X n (k)  and X n (l)  , show that
2 2
Re  X n1 (k)  1, Re  X n1 (l)  1
Im  X n1 (k)  1, Im  X n1 (l)  1
Thus overflow does not occur .
ii. Prove that
max  X n1 (k) , X n1 (l)   max  X n (k) , X n (l) 
max  X n1 (k) , X n1 (l)   2 max  X n (k) , X n (l) 
Solution:
x(n) = {1,1,1,1,1,1,1,1}
 2 
 j k
We knowthat W N  e Given N  8
k  N 

 2 
 j 0
Hence,W 8  e 1
0  8 

 2 
 j 1  
W8e  cos  j sin  0.707  j 0.707
1  8 
4 4
 2 
 j 2  
e  cos  j sin j
2  8 
W 8
2 2 (2)
 2 
 j 3 3 3
e  cos  j sin  0.707  j 0.707
3  8 
W 8
4 4
Input Output of stage 1 Output of stage 2 Output of stage 3

1 1+1=2 2+2=4 4+4=8

1 1+1=2 2+2=4 4-4=0

1 1+1=2 2-2=0 0

1 1+1=2 2-2=0 0

1 1-1=0 0 0

1 1-1=0 0 0

1 1-1=0 0 0

1 1-1=0 0 0

X(k)={8,0,0,0,0,0,0,0}

5. a) Determine the fourier transform X(ω) of the signal

x(n)  {1, 2,3, 2,1, 0}

b) compute the six point DFT V(k) of the signal
v(n) = {3, 2, 1, 0, 1, 2}
c) is there any relation between X(ω) and V(k)? Explain.
d)Compute the DFT of the sequence x(n)=(-1)n for the period N=16.

Solution:
 
(a) X ( )   x ( n)e
n 
 j n

j 2 j
 e  2e  3  2e j  e j 2
 3  2cos(2 )  4cos(4 )
5
(b) V(k)   v(n)e jkn
n 0
5 2
j
  v(n)e
kn
6

n 0
5 
 j kn
  v(n)e 3

n 0
   
j k  j 2k  j 4k  j 5k
 3  2e 3
0ee e
3 3 3

 2
 3  4 cos( k )  24 cos( k )
3 3
(c) V(k)  X ( w)
where w= 2 k
6
This is apparent from the fact that v(n) is one period ( 0  n  7 of a periodic sequence is
obtained by repeating x(n)
d) The 16 point DFT of x9n) is given by
2nk nk n
15 j 15 j 15   j nk 
X ( k )   x ( n )e 16
  (1) e n 8
    e 8 
n 0 n 0 n 0  
16
  j nk8 
1    e 
 j 2k
    1 e
  j nk8  j
nk

1    e  1 e 8

 
1  (cos 2k  j sin 2k ) 1  cos 2k
 
j
nk
 j nk
j
nk
 j
nk
k
e 16  e 16  e 816  e 16 * 2 cos
  16
1  cos 2k j 16
nk

 e
k
2 cos
16