Digital Signal Processing and its Applications Week 1

1. The system defined by y[n] = n(x[n] − x[n − 1]) is a linear shift-invariant

system.
(a) True
(b) False

Solution: B

2. Let x[n] be a discrete-time signal, and let y[n] = x[2n]. Now, if y[n] is
periodic, then x[n] is periodic.
(a) True
(b) False

Solution: B

3. Which of the following is/are the advantage(s) of working with digital sys-
tems as compared with an analog system?
(a) More memory efficient
(b) Essential for compatibility with modern computing devices
(c) Allow for a more generalised framework for modifying the data
(d) Easier to design

Solution: B, C

4. What is the minimum number of samples required to represent a perfectly

sinusoidal signal without losing any information?
(a) 1
(b) 2
(c) 3
(d) Frequency dependent
(e) Infinite

1
 Solution: C

5. Consider x(t) = sin(6πt) + sin(10πt), what is the (minimum) Nyquist sam-

pling rate in Hz for the signal y(t) = x(3t + 7)?

(a) 10/3
(b) 10
(c) 16
(d) 30

Solution: D

6. Which of the following is true about the Fourier Transform of g ′ (t), the
periodically repeated version of a time-limited function g(t) with period To .
[Note, g(t) equals g ′ (t) for only one period and let its Fourier transform be
G(f ).]

(a) It depends on To
(b) It includes the Dirac delta
(c) It is the periodically repeated version of G(f ) in the frequency domain
(d) It is the sampled version of G(f ) in the frequency domain

Solution: A, B, D

7. Which of the following is the correct order of components for a receiver

system?

(a) Antenna - DSP Unit - Anti-aliasing Filter - Sampler - Quantiser

(b) Antenna - Anti-aliasing Filter - Sampler - DSP Unit - Quantiser
(c) Antenna - Sampler - Anti-aliasing Filter - Quantiser - DSP Unit
(d) Antenna - Anti-aliasing Filter - Sampler - Quantiser - DSP Unit

2
 Solution: D

8. If a shift-invariant system satisfies the relation, y[n] = y1 [n] + y2 [n] for all
x[n] = x1 [n] + x2 [n] where y1 [n], y2 [n] and y[n] are the output sequences
when the inputs are x1 [n], x2 [n] and x[n] respectively, then it can be said to
be LSI.
(a) True
(b) False

Solution: B

9. Adding more and more aliased versions of a signal onto itself causes:
(a) Constructive interference at t = n · Ts
(b) Destructive interference at t ̸= n · Ts
(c) Constructive interference at t ̸= n · Ts
(d) Destructive interference at t = n · Ts

Solution: A, B

10. Any arbitrary signal x[n] satisfies the condition,

∞
X ∞
X ∞
X
2
x [n] = x2e [n] + x2o [n]
n=−∞ n=−∞ n=−∞

(a) True
(b) False

Solution: A

11. Which of the following statements are true?

(a) Given a sinusoidal signal of the form Acos(ωt + ϕ), only three distinct
valued samples are needed to determine the signal uniquely (to deter-
mine the parameters A, ω and ϕ).

3
(b) At high speeds, often a wheel having spokes appears to be rotating
backwards due to aliasing and this phenomenon can be explained using
the sampling theorem.
(c) A wheel with 30 spokes, captured using a 30 fps (frames per second)
camera appears to be stationary if it rotates at an angular velocity,
ω = 30π rad/s.
(d) The Dirichlet signal cannot be reconstructed from uniform samples ir-
respective of sampling frequency

Solution: B, C, D

4
Digital Signal Processing and its Applications Week 2

1. Let an LSI system’s impulse response be defined as

h[n] = α|n| n ∈ Z, 0 < α < 1

Which of the following are true?

(a) The output of the system for input x[n] = ejwn , will have same phase as
x[n]
(b) If the input to the system is odd ( x[−n] = −x[n]), then the output at
n = 0, will be 0
 
α+α3 −1
(c) The output of the system for the input, x[n] = u[n] at n = 2 is (1−α)
 
α−α3 +1
(d) The output of the system for the input, x[n] = u[n] at n = 2 is (1−α)

Solution: A,B,D
The system is given to be LSI, therefore output of the system will be x[n] ∗
h[n]. If the input of the system is x[n] = ejwn , then

y[n] = ejwn∗ α|n|

k=+∞
X
= ejw(n−k) · α|k|
k=−∞
k=+∞
X
−jwk
=e jwn
e · α|k|
k=−∞
" k=−1 k=∞
#
X X
y[n] = ejwn e−jwk · α−k + 1 + e−jwk · αk
k=−∞ k=1
jw −jw
 
e α e α
∴ y[n] = ejwn + 1 +
(1 − ejw α) (1 − e−jw α)
1 − α2 jwn


∴ y[n] = · e
(1 + α2 )
The output y[n] is a real positive factor times x[n], therefore it is in phase
with x[n].
(Hence, A is true)

1
 If input x[n] to the system is odd,
k=+∞
X
y[n] = x[n − k] ∗ h[k]
k=−∞

The output at n = 0,
k=+∞
X
y[0] = x[−k] ∗ h[k]
k=−∞
" k=−1 k=∞
#
X X
y[0] = x[−k] · α−k + x[0] + x[−k].αk
k=−∞ k=1

Since the input is odd, therefore two summations will cancel each other and
the only term remaining, is x[0] which is zero, therefore the output y[0] is
equal to zero.
(Hence, B is true)
If the input of the system is x[n] = u[n]
k=+∞
X
y[n] = u[k] ∗ h[n − k]
k=−∞
k=+∞
X
= h[n − k] ( since u[k] = 1 for k ≥ 0)
k=0
k=+∞
X
= α|n−k|
k=0

Substituting n = 2
k=+∞
X
2 1 0
y[2] = α + α + α + αk−2
k=3
α
y[2] = α2 + α1 + 1 +
(1 − α)
α − α3 + 1
y[2] =
(1 − α)
(Hence, D is true)
2. Output of an LTI system is
8 sin ((2ω1 + ω2 ) n) cos ((ω1 − 2ω2 ) n)

2
 for ω1 ̸= ω2 . What are the possible inputs to the system?
(a) cos ((3ω1 − ω2 ) n) + sin ((ω1 + 3ω2 ) n)
(b) cos ((3ω1 − ω2 ) n) + 2 cos ((ω1 + 3ω2 ) n)
(c) 3 cos ((2ω1 + ω2 ) n) + 5 sin ((ω1 − 2ω2 ) n)
(d) 4 sin ((3ω1 − ω2 ) n + ϕ1 ) + sin ((ω1 − 3ω2 ) n + ϕ2 )

Solution: B,D
On simplification we can write the output as
4 (sin ((3ω1 − ω2 ) n) + sin ((ω1 − 3ω2 ) n)
So this means our output is a linear combination of two sinusoids with
angular frequency (3ω1 − ω2 ) and (ω1 − 3ω2 ).
Since it is an LTI system the input would have been a linear combination
of sinusoids with the same angular frequencies. Thus the most appropriate
options are B and D.
3. Which of the following functions are analytic over R?
(a) 3z 2 + 2z + 1
2
(b) ez
(c) sin(z)
(d) |z|, (|z| denotes absolute value of z)

Solution: A,B,C
Analytic Function: A function f (z) is said to be analytic in a region Π
of the complex plane if f (z) has a derivative at each point of Π and if f (z)
is single valued at that point. Clearly, A,B, and C are analytic over R and
|z| is not differentiable at 0 so it is not analytic.
4. Consider the following system, y[n] with input x[n] defined as :
y[n] = (x[n2 ])2
The system is
(a) Time invariant
(b) Linear
(c) Time varying
(d) Linear and Time invariant

3
 Solution: C
Clearly, the system is non-linear due to squaring of the input.
For time invariance, we apply the delayed input x2 [n] = x[n−k], which gives
y2 [n] = (x[n2 − k])2 ̸= x[(n − k)2 ]2 . So, the system is time-varying.
5. Let y[n] be an LTI system with input x[n]. Which of the following is/are
true?

(a) x[n] = ejωn is an eigenfunction of y[n]

(b) x[n] = e−jωn is an eigenfunction of y[n]
(c) x[n] = e−jωn is not an eigenfunction of y[n]
(d) x[n] = ejωn is not an eigenfunction of y[n]

Solution: A,B

Let x[n] = esn , s ∈ C and h[n] be the impulse response of the system.
∞
X
y[n] = h[k]es(n−k)
k=−∞
X∞
= h[k]esn e−sk
k=−∞
∞
X
=e sn
h[k]e−sk (1)
k=−∞

. So, any exponential sequence of the form x[n] = esn , s ∈ C is an eigen-

function of y[n].
6. Suppose that we have a black box which has a circuit made of various elec-
trical components. In which of the following cases the system is LTI?

(a) The circuit is made using resistors

(b) The circuit is made using inductors and capacitors (without any resis-
tance)
(c) The circuit is made using resistors, inductors and capacitors.
(d) None of the above

4
 Solution: A,B,C
A circuit made using linear and passive components like resistors, inductors
and capacitors is LTI.
7. What is the advantage of working with Phasors over sines and cosines?
(a) Exponential functions take up less space
(b) Change of phase has a much simpler representation in the case of Phasors
(c) Summing up sines and cosines directly is not straightforward
(d) Complex exponentials are easier to visualise

Solution: B,C

8. Which of the following quantities of a sinusoid is/are preserved when input

to an LSI system?
(a) Phase
(b) Frequency
(c) Magnitude
(d) Shape

Solution: B,D
Frequency is preserved when a sinusoidal input is given to an LSI system.
Also, an LSI system cannot give rise to other frequency components apart
from the ones present in the input so the shape is preserved.
9. Additivity and homogeneity are sufficient conditions for linearity, individu-
ally. The given statement is?
(a) True
(b) False

Solution: False

10. Which of the following statements is/are true regarding the convolution of
the sequences, x1 [n] = {1, 2, 3, 4} and x2 [n] = {4, 3, 2, 1}.

5
 (a) The maximum value of the convolved sequence is 30
(b) The minimum value of the convolved sequence is 8
(c) The maximum value of the convolved sequence is 20
(d) The minimum value of the convolved sequence is 4

Solution: A,D
The convovled sequence is y[n] = {4, 11, 20, 30, 20, 11, 4}.
2πp
11. The sequence, x[n] = cos ω0 n is periodic if and only if ω = q , where p
and q are positive integers. The above statement is?
(a) True
(b) False

Solution: True

For ω0 = 2πp/q,
     
2πp(n + q) 2πpn 2πpn
cos = cos + 2πp = cos .
q q q
Therefore, the signal is periodic with period N = q.
Now we assume that there exists an integer k such that
cos (ω0 n) = cos (ω0 n + ω0 k)
= cos (ω0 n) cos (ω0 k) − sin (ω0 n) sin (ω0 k) , ∀n
=⇒ ω0 k is an integer multiple of 2π =⇒ ω0 is a rational multiple of 2π.
12. Let x[n] shown below be decomposed in terms of unit step signals.
x[n]
5

1
n
−2 −1 1 2 3 4 5

6
 Figure 1: Input Sequence x1 [n] Figure 2: Output Sequence y1 [n]

(a) The coefficient of u[n] is 2

(b) The coefficient of u[n] is 3
(c) The coefficient of u[n − 3] is -1
(d) The coefficient of u[n + 1] is -1

Solution: B,C
The sequence is x[n] = u[n + 1] + 3u[n] − 2u[n − 1] − u[n − 2] − u[n − 3]
13. Let x1 [n] shown below be the input to an LTI system and y1 [n] be the
corresponding output. The output, y2 [n] for the input sequence, x2 [n] shown
below is?

Figure 3: Input Sequence x2 [n]

(a) y2 [n] = {2, 4, 6, 6, −4, −6, −8}

(b) y2 [n] = {2, 4, 6, 6, −4, −6, 8}
(c) y2 [n] = {2, −4, −6, 6, −4, −6, 8}
(d) y2 [n] = {2, −4, −6, −6, 4, 6, 8}

7
 Solution: A
Observe that,
x2 [n] = 2x1 [n + 1] − 2x1 [n]
Since the system is LTI, the response, y2 [n] would be,
y2 [n] = 2y1 [n + 1] − 2y1 [n], y1 [n] = {0, 1, 3, 6, 9, 7, 4}
Which gives,
y2 [n] = {0, 2, 6, 12, 18, 14, 8, 0} − {0, 0, 2, 6, 12, 18, 14, 8}
y2 [n] = {2, 4, 6, 6, −4, −6, −8}
.
14. A student conducting an experiment using a natural sampler for sampling
a continuous real time domain video signal m(t). The natural sampler es-
sentially picks the signal for a small finite time window in every sampling
period. Its time domain representation can be given by
∞
X
S(t) = P∆ (t)δ (t − nTs )
n=−∞

Where, (
1
∆, 0 ≤ |t| ≤ ∆2
P△ (t) = ∆
0, 2 ≤ |t| ≤ ∞
∆ 1
The sampler is designed such that Ts = 100 follows. The sampled signal in
time domain can be given by
k(t) = m(t)S(t)
Keeping in mind that he was aware of the Nyquist theorem, and made
sure that the sampling frequency is 20 times the Nyquist frequency. To
reconstruct the signal he has a set of ideal lowpass and bandpass filters
which he can use appropriately to obtain the original time domain signal
back.
Which of the following statements  are correct in the regard of above exper-
sin(πx)
iment? (Note: sinc(x) = πx

(a) The original signal m(t) can be recovered by a using a ideal low pass
filter by extracting the spectrum of the time domain signal.

8
(b) The natural sampling technique introduces distortion in the magnitude
spectrum but not the phase spectrum which would have been obtained
from ideal sampling, because natural sampling essentially multiplies
replica with sinc function samples which is a real value.
(c) The original signal can be recovered by using the ideal Band-pass filter
centered at 200fs .
(d) In general, the reconstruction of the unique baseband signal from the
sampled signal is always possible in every sampling conditions and hence
validating Nyquist theorem is vague.

Solution: A

The sampled signal is

∞
X
k(t) = m(t) P△ (t)δ (t − nTs )
n=−∞

In frequency (f) domain, by using Fourier series coefficients of the periodic

sampler signal, and Poisson summation formula, the representation of the
sampler is given by
 
X sinc n∆
∞
Ts

n

S(f ) = δ f−
n=−∞
Ts Ts

Using convolution-Multiplication theorem, the frequency domain represen-

tation of k(t) can be obtained as

K(f ) = M (f ) ∗ S(f )

Hence, the Output of sampler in frequency domain is

 
∞ sin c n∆  
X Ts n
K(f ) = M f−
n=−∞
Ts Ts

and for the given design condition of the sampler of ratio of ∆ and Ts ,
∞ n

 
X sinc 100 n
K(f ) = M f−
n=−∞
T s Ts

9
A is correct because the spectrum at f = 0, is the spectrum of the original
time domain signal, and by using a Ideal lowpass filter, this spectrum is
extracted, in time domain, it implies the reconstruction of the original signal.
B is incorrect because the sampled sinc will have both positive and negative
values, and being multiplied by a negative number, the phase will be affected
by 180 degrees.
C is incorrect because when n = 200, the sinc(2) will be zero. Hence, the
output of the bandpass filter to extract just that part of periodic spectrum
contains no knowledge of the shape of spectrum of the original signal.
D is incorrect the reconstruction is not possible if the sampling frequency
was too low and aliasing of spectrum may take place.

10
 DSP-Assignment-Week3

July 10, 2022

Question 1
Consider the following response for discrete-time LTI system,
1
h[n] = ( )n u[n − 2]
k
Which of the following statements is true for k > 1?

1. System is neither causal nor BIBO stable.

2. System is BIBO stable but not causal.
3. System is causal but not stable.
4. System is causal and BIBO stable.
Solution (4)
h[n] = 0 for n¡2 since u[n] = 0 for n¡2
Thus h[n] is causal.

Consider,
∞ ∞
X 1n X 1
| u[n − 2]| = n
−∞
k 2
k
1
For k¿1, k < 1 Thus,
∞
X 1 1 1 1
n
= 2 1 =
2
k k 1− k
k(k − 1)

Since h[n] is absolutely summable, the system is BIBO stable.

Question 2
In the above question, what will happen for k < 1 ?

1. System is neither causal nor BIBO stable.

2. System is BIBO stable but not causal.
3. System is causal but not stable.

1
 4. System is causal and BIBO stable.

Solution : (3)
h[n] = 0 for n¡2 since u[n] = 0 for n¡2
Thus h[n] is causal.

1
For k¡1, k > 1 Thus,
∞
X 1
n
→∞
2
k
Since h[n] is not absolutely summable, the system is not BIBO stable.
Question 3
A discrete-time LTI system has the impulse system:
sin( πn
2 )
h[n] =
πn
Is this system BIBO stable?
1. Yes
2. No
3. Can’t Say
Solution : (2)

Consider the absolute sum.

∞ ∞
X X sin(nπ/2) 1 1 1 1 1 1 1 1 1
|h[n]| = | | = 0.5+ {.....+ + + + +1+1+ + + + ......}
−∞ −∞
nπ π 9 7 5 3 3 5 7 9

2 1 1 1 1
= 0.5 + {1 + + + + .......}
π 3 5 7 9
2 1 1 1 1 1
≤ 0.5 + { + + + + .......}
π 2 4 6 8 10
1 1 1 1 1
= 0.5 + (1 + + + + .......)
π 2 3 4 5
∞
1X1
= 0.5 +
π n=1 n
Since the above sum does not converge the system is not BIBO stable.

Question 4 Find the impulse response of the following system :

2
 1. 5u[n+1]-12u[n]+20u[n-1]-8u[n-2] + 3u[n-3]
2. 5u[n-3]-12u[n-2]+20u[n-1]-8u[n]+3u[n+1]
3. 5u[n+1]-4u[n]+6u[n]+8u[n+1]
4. 5u[n+1]-4u[n]+12u[n]+8u[n+1]

Solution : (1)
Let the two systems be

h1 [n] = u[n] − 2u[n − 1] + 3u[n − 2]

h2 [n] = 5u[n + 1] − 2u[n] + u[n − 1]
Since the two systems are cascaded the impulse response of the system will be
∞
X
h[n] = h1 [n] ∗ h2 [n] = h1 [k]h2 [n − k]
k=−∞

= 5u[n + 1] − 12u[n] + 20u[n − 1] − 8u[n − 2] + 3u[n − 3]

Question 5
Consider two systems with impulse response,

h1 [n] = u[n] + 2u[n − 1] + u[n − 2] + 0.5u[n − 3]

h2 [n] = 2u[n] − u[n − 1] + 0.5u[n − 2] + u[n − 3]

connected in parallel. Find the output of the combined system when the input
to the system is

x[n] = 2u[n] − 2u[n − 1] + u[n − 2] − 0.5u[n − 3]

1. 6u[n]−4u[n−1]+4u[n−2]−0.5u[n−3]−2u[n−4]+0.75u[n−5]−0.75u[n−6]
2. 6u[n]+4u[n−1]−4u[n−2]+0.5u[n−3]−2u[n−4]−0.75u[n−5]+0.75u[n−6]

3. 3u[n] − 2u[n − 1] + 4u[n − 2] − u[n − 3] − 2u[n − 4] + u[n − 5] − 0.75u[n − 6]

4. 6u[n]−2u[n−1]−8u[n−2]−0.5u[n−3]+2u[n−4]+0.75u[n−5]−0.75u[n−6]

Solution : (1)
Since the two systems are in parallel, the impulse response of the combined
system is given by,

h[n] = h1 [n] + h2 [n] = 3u[n] + u[n − 1] + 1.5u[n − 2] + 1.5u[n − 3]

The input to the system is ,

x[n] = 2u[n] − 2u[n − 1] + u[n − 2] − 0.5u[n − 3]

3
Thus, the output y[n] is given as,
∞
X
y[n] = h[n]∗x[n] = h[k]x[n−k] = 6u[n]−4u[n−1]+4u[n−2]−0.5u[n−3]−2u[n−4]+0.75u[n−5]−0.75u[n−6
k=−∞

Question 6
Consider the signals x[n] = 2n−1 u[−n + 2] and y[n] = 2−n+2 u[n + 1],where u[n]
is the unit step sequence.Let X(ejw ) and Y (ejw ) be the discrete time Fourier
transform of x[n] and y[n],respectively.What is the value of the integral
Z 2π
1
X(eiw )Y (e−jw ) dw
2π 0

1. 2

2. 8
3. 4
4. 1

Solution : (2)
Using the below two results of orthogonality and unit magnitude property of
eigen vectors jwn :
Z 2π
ejw(k2 −k1 ) dw = 0 f or k1 ̸= k2
0
Z 2π
ejw(k2 −k1 ) dw = 2π f or k1 = k2
0
Z 2π
1
X(ejw )Y (e−jw )dw
2π 0
Z 2π kX
1 =∞ kX
2 =∞
1
= x[k1 ]e−jwk1 y[k2 ]ejwk2
2π 0 k1 =−∞ k2 =−∞

Using orthogonality we can write,

Z 2π k=∞
1 X
= x[k]e−jwk y[k]ejwk
2π 0 k=−∞

Using the unit magnitude property,

k=∞
X
= x[k]e−jwk y[k]ejwk
k=−∞

4
 k=∞
X
= x[k]y[k]
k=−∞

= x[−1]y[−1] + x[0]y[0] + x[1]y[1]

=8
Multiple correct questions

Question 7
Which of the following systems are causal?

1. x[n] + x[n − 5] + x[n − 100]

2. nx[n] + n2 x[n − 2]
3. 2n x[n − 2] + x[n]
4. 2−n x[n + 1] + x[n]

Solution : (1) (2) (3)

For any system y[n],
If y[n] = 0 for n¡0 then the system is causal.
In (4) y[n] ̸= 0 for n=-1, y[−1] = 2. Thus system is not causal. Question 8
Which of the following can be orthogonal basis for 4-D space?

1,0,0,0 , [0,1,0,0] , [0,0,0,1] , [0,0,1,0]

1,2,3,0 , [0,-3,2,4] , [6,0,-2,1] , [1,-0.5 ,0,-6]

1,1,-1,-1 , [-1,1,1,-1] , [-1,1,1,-1] , [0,1,0,1]
0,0,0,1 , [1,2,3,0], [1,2,-3,0], [-2,1,0,0]

For the vector to form orthogonal basis, they should all be perpendicular to
each other.
We can check that all pairs of vectors for (1) (3) and (4) are perpendicular.
Also since they are all perpendicular to each other their linear combination
cannot be zero. Hence they span the 4-D space. Question 9
Consider 2 orthogonal vectors [x,5,7] , [2, 3x-4, -2]. Find the value of x
1. 1 2. 2 3. 6/17 4. 36/17 Solution : (2)
Since the two vectors are orthogonal, their dor product will be 0.
Hence,
x(2) + 5(3x − 4) + 7(−2) = 17x − 34 = 0
x=2
Question 10
What is true about the frequency response H(ejw ) of the sequence h[n]?

5
 1. H(ejw ) is the inner product between sequences h[n] and ejwn
P∞
2. H(ejw ) = −∞ h[n]e−jwn
3. H(ejw ) is the convolution between sequences h[n] and ejwn
P∞
4. H(ejw ) = −∞ h[n]ejwn
Solution : (1) (2)
H(ejw ) is the inner product between sequences h[n] and ejwn
∞
X
H(ejw ) = h[n]e−jwn
−∞

Question 11
Consider [x,y,z] to be a vector of magnitude 5. Find the upperbound on the
value of 5x+3y+8z.
1. 40
√
2. 35 2
3. 80/3
4. 65/2
Solution : (2)
By cauchy schwartz inequality we know that
u.v ≤ |u||v|
Therefore
p p √
5.x + 3.y + 8.z ≤ 52 + 32 + 82 x2 + y 2 + z 2 = 35 2
Question 12
Consider a sequence x[n] with frequency response X(w). Consider another se-
quence y[n] followed by sampling x[n] at intervals of k. Find Y(w) in terms of
X(w)
1. Y(w) = X(w)
2. Y(w) = X(kw)
3.
4.
Solution : (4)

y[n] = x[kn]
∞
X X∞ ∞
X
Y (w) = y[n]e−jwn = x[kn]e−jwn = x[kn]e−j(w/k)kn
n=−∞ n=−∞ n=−∞

Question 13 Find the component of a u[n] in direction of ejwn

n

6
 1 jwn
1. 1−ae−jw e
1
2. 1−ae−jw

3. an u[n]ejwn
4. aw
Solution : (2)
Consider x[n] with frequency response as X(w).
We know,
∞
X
X[w] = x[k]e−jwk
k=−∞

Therefore, if we consider x[k] and ejwk as two sequences then X[w] is the dot
product between the two.
X[W] is the component of x(k) in direction of ejwk
Question 14
Find the output of an LTI system where h[n] = (n + a)an u[n] and input is
cos(wn).

ae−jw ae−jw
1. 1−ae−jw (1 + 1−ae−jw )

ae−jw ae−jw
2. 1−ae−jw (1 + 1−ae−jw )cos(wn)

ae−jw ae−jw jwn aejw aejw −jwn

3. 1−ae−jw (1 + 1−ae−jw )e + 1−aejw (1 + 1−aejw )e
−jw
ae ae−jw jwn aejw aejw −jwn
4. 0.5( 1−ae−jw (1 + 1−ae−jw )e + 1−aejw (1 + 1−aejw )e )
Solution : (3)

∞
X ∞
X
h[k]cos(w(n − k)) = 0.5h[k](ejw(n−k) + e−jw(n−k) )
k=−∞ k=−∞
∞
X ∞
X
jwn −jwk −jwn
= 0.5(e h[k]e +e h[k]ejwk )
k=−∞ k=−∞

= 0.5(H(w)e jwn
+ H(−w)e−jwn )
where,
∞
X ae−jw ae−jw
H(w) = h[k]e−jwk = (1 + )
1 − ae−jw 1 − ae−jw
k=−∞

Question 15
Consider 2 stable LTI systems an u[n] and b−n u[n] . Find the frequency response
of the two systems. Consider a system formed by cascading the two systems.
Find the frequency response of the cascaded system.

7
 bejw
1. (1−ae−jw )(bejw −1)

1
2. (1−ae−jw )(1−be−jw

bejw
3. bejw −a

aejw
4. aejw −b

Solution : (1)

x1 [n] = an u[n]
x2 [n] = b−n u[n]
∞ ∞
X X 1
X1 (w) = x1 [n]e−jwn = an e−jwn =
n=−∞ n=0
1 − ae−jw
∞ ∞
X X bejw
X2 (w) = x2 [n]e−jwn = b−n e−jwn =
n=−∞ n=0
bejw − 1

Consider the cascaded system

∞
X
x[n] = x1 [n] ∗ x2 [n] = x1 [k]x2 [n − k]
k=−∞

∞
X
= ak u[k]b−(n−k) u[n − k]
k=−∞
n
X an+1 b − b−n
= ak b−(n−k) =
ab − 1
k=0
∞
X
X(w) = x[n]e−jwn
n=−∞
∞
1 X
= (an+1 b − b−n )e−jwn
ab − 1 n=−∞

1 ab bejw
= ( − )
ab − 1 1 − ae−jw bejw − 1
bejw
=
(1 − ae−jw )(bejw − 1)
Can you observe any relationship among the frequency responses.Try to prove
the relation for a general case.

8
Digital Signal Processing and its Applications Week 4
1. Given a sequence x[n] with DTFT X(ω), find the DTFT of x[2n].
1
(a) · (X(ω/2) + X(ω/2 − π))
2
(b) X(ω/2) + X(ω/2 − π)
1
(c) · (X(2ω) + X(2ω − π))
2
(d) X(2ω) + X(2ω − π)

Solution: A

2. Find the DTFT of the sequence x[n] given by:

(
1 , −3 ≤ n ≤ 3
x[n] =
0 , else
sin (3ω/2)
(a)
sin (ω/2)
sin (7ω/2)
(b)
sin (ω/2)
sin (9ω/2)
(c)
sin (ω/2)
sin (5ω/2)
(d)
sin (ω/2)

Solution: B

3. Determine the value of the given integral using Parseval’s theorem:

Z ∞
sinc2 (kx − c) dx
−∞
where c is a constant, here sinc(x) = sin(x)/x
(a) ∞
(b) π
(c) kπ
(d) π/k

1
 Solution: D

4. Which of the following relations and/or statements is incorrect in the context

of the sequence x[n] and its DTFT?

(a) The DTFT of the signal an x[n] is given by X(aω), and this property is
called the frequency scaling property.
(b) The existence of the DTFT of x[n] is enough to prove that the sequence
x[n] is absolutely summable.
(c) There cannot exist a sequence x[n] which is absolutely summable but
does not have finite energy or square summability (as defined by Parse-
val’s relation).
(d) The Inverse-DTFT of the sequence Y (ejω ) = X(ejω )ejω/3 is the sequence
which is identically zero.

Solution: A, B, D

5. The input-output relation of a discrete-time system is defined as:

y[n] = x[n] + 5x[n − 1]

The input to this system is the output of a ideal sampler block. Consider
the input applied to the sampler is a(t) = sin(ωt). The ideal sampler is
designed to sample the input signal at the sampling frequency which is equal
to 4 times the Nyquist frequency of the input signal.
Which of the following statements hold true regarding the above system?

(a) The system will have the same steady state output irrespective of the
frequency of the applied input
(b) The steady state output of the overall system for the given input is
5.75 · sin (πn/4 − 37.94 deg)
(c) The system is linear shift-invariant discrete-time system
(d) The output of the sampler for the specified input signal is sin(πn/8)

2
 Solution: A, B, C

6. A causal LTI system is described by the difference equation:

y[n] − a2 y[n − 1] = b2 x[n] + x[n − 1]
where a is real and is less than 1 in magnitude. Find a relationship between
a and b such that |H(ω)| = 1 for all values of ω.
(a) b = ja
(b) b2 + a2 = 0
(c) a=b
(d) None of the above

Solution: A, B

7. Let the DTFT of a discrete-time signal x[n] be X(ω). Which of the following
is true?
(a) X(ω) = X(ω + π)
(b) X(ω) = X(ω − π)
(c) X(ω − π) = X(ω + π)
(d) X(ω) = X(ω − 3π)

Solution: C

8. Evaluate the following integral using Parseval’s theorem:

Z π
1
dx
−π 65 − 16 cos x
1
(a)
63
2π
(b)
63
π
(c)
63
2
(d)
63
3
 Solution: B

9. A discrete time signal x[n] is defined as:

x[n] = [4α, 4, α2 , 3α]

where the first element is at 0 and x[n] is 0 elsewhere. Let the DTFT of the
sequence be X(ω) and it is known that the the sum of the terms of IDTFT
of X(ω–π) is 2, then α can take the values:

(a) 2
(b) -2
(c) 3
(d) -3

Solution: A, D

10. DTFT of a real signal has Hermitian symmetry, ie. X(ω) = X ∗ (−ω))

(a) True
(b) False

Solution: A

4
Digital Signal Processing and its Applications Week 5

1. Let the inverse z-transform of Xm (z) be xm [n], where Xm (z) is given by:
1
Xm (z) =
(1 + z −1 )m
Xm (z) ⇐⇒ xm [n] = f (n, m) × g[n]
Where xm [n] can be expressed as f (n, m) × g[n] and the ROC is |z| > 1.
Which of the following options is/are correct?

(a) f(n, m) is the number of ways to select a team of n students from a class
of n + m −1 students.

n
(b) f (n, m) =
m
(c) g[n] = u[n + m − 1], where u[n] is the unit step function.
(d) If P (z) = (1+az1−1 )m , then the ratio between the IDTFTs of P (z) and
X(z) is an+m−1

Solution: A,C
1
Using the ROC of |z| > a, If X(z) = 1+az −1 , then its IDTFT x[n] would be
n
a u[n] Now, as we know by the properties of z-transform,
dX
nx[n] ⇐⇒ −z
dz
which then gives us:
dX
nx[n] ⇐⇒ −z
dz
az −2
∴ nan u[n] ⇐⇒ −z
(1 + az −1 )2
1
∴ (n + 1)an u[n + 1] ⇐⇒
(1 + az −1 )2
Here, we can use Mathematical Induction to obtain the following relation:
az −2 d
 
1 1
=
(1 + az −1 )m m − 1 dz (1 + az −1 )m−1
az −2 d
 −2  
az d 1
=
m − 1 dz m − 2 dz (1 + az −1 )m−2

1
 Continuing this series in a similar manner, we use the first equation to find
the discrete IDTFT of X(z)
1 n (n + 1)(n + 2)(n + 3) . . . (n + m − 1)
−1 m ⇐⇒ a u[n + m − 1]
(1 + az ) (m − 1)!
 
n n+m−1
=a u[n + m − 1]
n
Thus, if we put a = 1, then:
 
n+m−1
f (n, m) =
n
g[n] = u[n + m − 1]
Plus, if we take X1 (z) as mentioned above, then the ratio would simply be
an . Hence, A and C are correct.
2. Let h[n] be the impulse response of the LTI system y[n] = x[n] − y[n − 1].
Assuming the system to be causal, a student says that the system is BIBO
stable. The student’s claim is?

(a) True
(b) False

Solution: False

Since we know that the impulse response exists and that the system is LTI,
we use x = δ[n], to find h[n]. Notice that since we know the system is causal
h[−1] = 0.
h[0] = δ[0] − h[−1] = 1
h[1] = δ[1] − h[0] = −1
h[2] = δ[2] − h[1] = 1 . . . and so on
We have,
h[n] = (−1)n u[n]
Its clear from the above that the system is not BIBO stable since h[n] is
NOT absolutely summable!

2
3. Let h[n] be the impulse response of the causal LTI system y[n] = x[n] −
y[n − 1]. Now, we wish to improve the stability of the system, by making
the following modification:

y[n] = x[n] − αy[n − 1]

The range of values (multiple possible) of α for which this system is BIBO
stable is?

(a) 13 , 54


(b) 18 , 34


(c) 12 , 1


(d) − 23 , 67


Solution: B,D

Since we know the system is causal h[−1] = 0.

h[0] = δ[0] − h[−1] = α
h[1] = δ[1] − h[0] = −α
h[2] = δ[2] − h[1] = α . . . and so on
We have,
h[n] = (−α)n u[n]
Thus, the above is absolutely summable if and only if |α| < 1. Thus, the
system is BIBO stable only when |α| < 1.
4. State True or False: For a causal LTI system, the ROC is outside the
outermost pole.
(a) True
(b) False

Solution: True

5. Consider a stable and causal system with impulse response h[n] and rational
system H(z). Suppose it is known that H(z) contains a pole at z = 1/2, and

3
 zero somewhere on the unit circle. The precise number and locations of all
of the other poles and zeros are unknown. Which of the following statements
is/are true?


(a) H ejw = 0 for some ω
(b) h[n] has finite duration
(c) h[n] has infinite duration
(d) g[n] = n[h[n] ∗ h[n]] is the impulse response of a stable system

Solution: A,C,D

A is true, since there is a zero on the unit circle. For any sequence to be a
finite-duration sequence, it must have ROC that includes the entire z-plane,
except possibly at z = 0/∞. Since we have a pole at z = 1/2, the sequence
is not a finite duration sequence. Hence, B is false and C is true.
Since the system is casual h[n] = 0, for n < 0, and h[n] ∗ h[n] = 0, for
n < 0, the system with h[n] ∗ h[n] as it’s impulse response is causal. Hence
the system g[n] is also casual. The z-transform for the system h[n]]∗ h[n] =
H(z)2 . By differentiation property, the system function corresponding to
g[n] is  
d 2 d
G(z) = −z H (z) = −2zH(z) H(z)
dz dz
From the above equation, we can say that the poles of G(z) are at the same
locations as those of H(z), with the possible exception of the origin. Hence
since H(z) has all its poles inside the unit circle, so must G(z). It follows
that g[n] is the impulse response of a causal and stable system. Hence, D is
also true.
6. Consider a series cascade system having impulse responses, h1 [n] = {1, 2}
and h2 [n] = {1, −2}, where ’, denotes n = 0. Find the input sequence x[n]

for which this system produces an output sequence,

y[n] = {1, −1, −2, 3, −8, 4}

4
 (a) x[n] = {1, 1, 2, −1}
(b) x[n] = {1, −1, 2, −1}
(c) x[n] = {1, −1, −2, −1}
(d) x[n] = {1, −1, 2, 1}

Solution: B

We know that,
x[n] ∗ h1 [n] ∗ h2 [n] = y[n]
Taking the z-transform on both sides,
X(z)H1 (z)H2 (z) = Y (z)
X(z) = Y (z)/ (H1 (z)H2 (z))
1 − z −1 − 2z −2 + 3z −3 − 8z −4 + 4z −5


X(z) =
(1 + 2z −1 ) (1 − 2z −1 )
X(z) = 1 − z −1 + 2z −2 − z −3
Hence, we obtain x[n] by taking inverse z-transform as,
x[n] = {1, −1, 2, −1}
Hence, B is the correct choice.
7. State True or False: The function, X(z) = z can be the z-transform of a
sequence x[n]. (z denotes the complex conjugate of z)
(a) True
(b) False

Solution: False
The function, X(z) = z is non-analytic using Cauchy-Riemann conditions.
Hence, it cannot be the z-transform of any sequence x[n] as z-transform
must be analytic.
8. State True or False: Region of convergence of derivative of z-transform
is the same as Region of convergence of the z-transform of original sequence.

(a) True
(b) False

5
 Solution: False
The claim is not true in general. Refer to https://doi.org/10.1049/el.
2016.0189 for a detailed treatment of this matter.
sin nπ/2
9. Conisder the sequence x[n] = sinc(n/2) = nπ/2 . Which of the following
is/are true?
(a) z-transform of x[n] does not exist ∀z
(b) z-transform of x[n] exists on the unit circle
(c) z-transform of x[n] exists with ROC |z| ≥ 1
(d) z-transform of x[n] exists with ROC |z| < 1

Solution: A


1,
 n=0
sin(nπ/2) 
sinc(n/2) = = 0, n = 2m, m ̸= 0
nπ/2  (−1)m

0.5π(2m−1) , n = 2m − 1



For |z| = 1,
∞ ∞
X 4X 1
| sinc(n/2)| = 1 + = ∞.
n=−∞
π m=1
2m − 1
Since,
∞ ∞
1 X 1 X
3m > 2m − 1, ∀ m > 1 =⇒ > =∞
m=1
2m − 1 m=1
3m

For |z| =
̸ 1,
∞
X −1
X ∞
X
−n −n
z | sinc(n/2)| = z | sinc(n/2)| + z −n | sinc(n/2)|
n=−∞ n=−∞ n=0
P∞ −n
Now, when
P−1 |z| > 1 =⇒ n=0 z | sinc(n/2)| diverges and when |z| <
−n
1 =⇒ n=−∞ z | sinc(n/2)| diverges. So, the z-transform does not exist.

10. Conisder the sequence x[n] = sinc2 (n/2). Which of the following is/are true?
(a) z-transform of x[n] does not exist ∀z

6
 (b) z-transform of x[n] exists on the unit circle
(c) z-transform of x[n] exists with ROC |z| ≥ 1
(d) z-transform of x[n] exists with ROC |z| < 1

Solution: B


1, n=0


2
sinc (n/2) = 0, n = 2m, m ̸= 0
 1
n = 2m − 1


0.25π 2 (2m−1)2 ,

For |z| = 1,
∞ ∞
X 8 X 1
| sinc(n/2)| = 1 + 2 2
< ∞.
n=−∞
π m=1
(2m − 1)

Therefore, the z-transform exists on |z| = 1. Since,

∞
X 1
2
= π 2 /8
m=1
(2m − 1)

For |z| =
̸ 1,
∞
X −1
X ∞
X
−n −n
z 2
| sinc (n/2)| = z 2
| sinc (n/2)| + z −n | sinc2 (n/2)|
n=−∞ n=−∞ n=0
P∞ −n 2
Now, when
P−1 |z| > 1 =⇒ n=0 z | sinc (n/2)| diverges and when |z| <
−n 2
1 =⇒ n=−∞ z | sinc (n/2)| diverges. So, the z-transform does not exist.

11. State True or False: If x[n] is a finite duration two sided sequence, then
the ROC is entire z-plane excluding at z = 0 and z = ∞.
(a) True
(b) False

Solution: True
The expansion will contain terms of the form z k for k > 0 which would
exclude z = ∞ and z −k for k > 0 which would exclude z = 0 from ROC.

7
12. Which of the following is/are true related to ROC of z-transform of sequences
x[n] and y[n] with ROC Rx and Ry respectively.?
(a) ROC of the linear combination ax[n] + by[n] is atleast Rx ∪ Ry
(b) ROC of the linear combination ax[n] + by[n] is atleast Rx ∩ Ry
(c) ROC of the convolution x[n] ∗ y[n] is atleast Rx ∪ Ry
(d) ROC of the convolution x[n] ∗ y[n] is atleast Rx ∩ Ry

Solution: B,D

13. Find x[n] from X(z) below, where x[n] is known to be causal, i.e., x[n] = 0
for n < 0.
3 + 2z −1
X(z) =
2 + 3z −1 + z −2

n
(a) 21 − 12 u[n] + (−1)n u[n]

n
(b) − 12 u[n] + (−1)n u[n]

n
(c) − 12 u[n] + (−1)n+1 u[n]

n
(d) − 14 u[n] + (−1)n u[n]

Solution: A

3 + 2z −1 3 + 2z −1 1 1
= = +
2 + 3z −1 + z −2 (2 + z −1 ) (1 + z −1 ) 2 + z −1 1 + z −1
1
ROC for the 2 terms are |z| > 2 and |z| > 1 respectively for a causal system.
1
2 1
= + As x[n] is
1 + 21 z −1 1 + z −1
 n
1 1
= − u[n] + (−1)n u[n]
2 2

8
 DSP-Assignment-Week6

July 10, 2022

Question 1 Consider a signal x[n] = (0.5)n u[n] . Find the z transform and
the region of convergence of x[n-k]

1. |z| < 2
2. |z| > 2
3. |z| < 0.5
4. |z| > 0.5
Solution : (4)
The z-transform of x[n] is
∞
X
X(z) = x[n]z −n
−∞

Let z-transform of x[n-k] be X’(z)

∞
X
X ′ (z) = x[n − k]z −n
−∞
∞
X
X ′ (z) = 0.5n−k u[n − k]z −n
−∞
∞
X
′ −k
X (z) = 0.5 0.5n u[n − k]z −n
−∞
∞
X
X ′ (z) = 0.5−k 0.5n z −n
k
−k
z
X ′ (z) = 0.5−k 0.5k ; ROC|z| > 0.5
1 − 0.5z −1
z −k
X ′ (z) = ; ROC|z| > 0.5
1 − 0.5z −1
Question 2 If X[z] is the z-transform of signal x[n]. What is the z-transform
of signal (n + 1)2 x[n]

1
 1. z 2 X ′′ (z)
2. X ′′ (z) + 2X ′ (z) + X(z)
3. z 2 X ′′ (z) + zX ′ (z) + X(z)

4. z 2 X(z) + zX ′ (z) + X ′′ (z)

Solution : (3)
Let y[n] = x[n](n + 1)2 = n2 x[n] + 2nx[n] + x[n] Let the z-tranform of n2 x[n]
be X2 (z)
∞
X
X2 (z) = n2 x[n]z −n
n=−∞

We know,
∞
X
X(z) = x[n]z −n (1)
n=−∞

Taking derivative wrt to z on both sides we get

∞
X
X ′ (z) = − nx[n]z −n−1 (2)
n=−∞

Taking derivative wrt to z on both sides we get

∞
X
X ′′ (z) = n(n + 1)x[n]z −n−2 (3)
n=−∞

∞
X
X ′′ (z) = n(n + 1)x[n]z −n−2 (4)
n=−∞
∞
X
X ′′ (z) = z −2 n(n + 1)x[n]z −n (5)
n=−∞
∞
X
′′
2
z X (z) = n(n + 1)x[n]z −n (6)
n=−∞

Adding z*eqn (2) + eqn(1) we get

∞
X
(n + 1)2 x[n]z ( − n) = z 2 X ′′ (z) + zX ′ (z) + X(z)
n=−∞

Question 3
Consider the causal digital filter
For what values of k is the system stable?

1. k ≤ 1

2
 2. k ≥ 1.5
3. k ≤ 1.5
4. k ≥ 1
Solution : (3)
k
y[n] = p[n] − p[n − 1]
4
2k
p[n] = x[n] − p[n − 1]
3
2k −1
=⇒ P (z)(1 + z = X(z)
3
k
Y (z) = P (z)(1 − z −1
4
1 − k4 z −1
=⇒ H(z) = 2k −1
1+ 3 z
−2k
Pole = 3
k
Zero = 4 For system to be causal and stable, all poles have to lie inside unit
circle,
2|k|
≤ 1 =⇒ |k| ≤ 1.5
3
Question 4
In the above question, find y[n] if k=1 and x[n] = ( 13 )n

1 1 n
1. 12 ( 3 )
1 1 n
2. 4(3)

3. ( 31 )n
1 1 n
4. 6(3)

Solution : (1)
1
x[n] = ( )n
3
We have,
2 x[n − 1]
y[n] + y[n − 1] = x[n] −
3 4
2 1 3 1 1
y[n] + y[n − 1] = ( )n (1 − ) = ( )n
3 3 4 4 3
Let y[n] = α( 13 )n
1 2 1 1 1
α( )n + α( )n−1 = ( )n
3 3 3 4 3

3
 1
α=
12
Hence,
1 1 n
y[n] = ( )
12 3
Question 5
Consider the system defined by the following LCCDE

y[n] = 5.25y[n − 1] + 1.25y[n − 2] + x[n]

Find the impulse response of the system given its is stable

20 1 n
1. 19 ( 4 ) u[n]
1 n
2. 19 5 u[−n − 1]
20 1 n 1 n
3. 19 ( 4 ) u[n] + 19 5 u[−n − 1]
1 1 n 20 n
4. 19 ( 4 ) u[n] + 19 5 u[−n − 1]
Solution : (3)
Let Y(z) and X(z) be the z-transforms of y[n] and x[n] respectively. Then from
the given equation we get,

Y (z) = 5.25z −1 Y (z) + 1.25z −2 Y (z) + X(z)

Y (z) Y (z) Y (z)
= 5.25z −1 + 1.25z −2 +1
X(z) X(z) X(z)
H(z) = 5.25z −1 H(z) + 1.25z −2 H(z) + 1
1
H(z) =
1 − 5.25z −1 + 1.25z −2
This is the system function H(z)
1
H(z) =
(1 − 5z −1 )(1 − 0.25z −1 )
Thus the system has poles at z=5 and z = 0.25. The complex plane can thus
be divided into 3 regions.
By partial fraction,
1 20 1
H(z) = = −
(1 − 5z −1 )(1 − 0.25z −1 ) 19(1 − 5z −1 ) 19(1 − 0.25z −1 )
For the system to be stable, we need the unit circle to lie within ROC. Hence
the only possible ROC is 0.25 ≤ |z| ≤ 5. Under this ROC, the impulse response
h[n] is;
20 1 n 1
h[n] = ( ) u[n] + 5n u[−n − 1]
19 4 19
Question 6
In the above question, Find the impulse response of the system given its is causal

4
 1 1 n 20 n
1. 19 ( 4 ) u[n] + 19 5 u[−n − 1]
1 1 n 20 n
2. 19 ( 4 ) u[n] + 19 5 u[n]
20 1 n 1 n
3. 19 ( 4 ) u[n] + 19 5 u[−n − 1]
20 1 n 1 n
4. 19 ( 4 ) u[n] + 19 5 u[n]

Solution : (4)
As calculated above,
20 1
H(z) = −
19(1 − 5z −1 ) 19(1 − 0.25z −1 )

For the system to be causal, we need |z| → ∞ to lie within the ROC. Hence the
only possible ROC is |z| ≥ 5. Under this ROC, the impulse response h[n] is;
20 1 n 1
h[n] = ( ) u[n] + 5n u[n]
19 4 19
Question 7
Consider the transfer function of a discrete LTI system,

2 − 65 z −1
H(z) =
1 − 65 z −1 + 16 z −2

Choose the correct options

1. System is stable and causal for —z—¿0.5
2. System is stable and causal for —z—¿1 3Systemisstableandcausalf or 1 3<|z|< 1
2
z −1
is cascaded with H1 (z) = (1 − 2 )
−1
1
4.
3. System is stable and causal for 3 < |z| is cascaded with H1 (z) = (1 − z 2 )
Solution : (1), (4)
For a system to be stable the unit circle must lie in the ROC.
For a system to be causal, the ROC must extend to ∞
Poles of the system : 12 , 13

Question 8
A discrete time signal x[n] is defined as

u[n]
x[n] =
(n + 2)(n + 3)

Find out the Z- transform of x[n]

1. X(z) = z 2 − z
2 + (z − 1)log(1 − z −1 )
2. X(z) = z 2 − z
2 + z 2 (z − 1)log(1 − z −1 )

5
 3. X(z) = z 2 (z − 1)log(1 − z −1 )
4. X(z) = z 2 (z − 1)
Solution : (2)
Let us start with the summation,
∞
X z −n z −1 z −2 z −3
= + + .... = −log(1 − z −1 )
1
n 1 2 3

We will use this identity later. By partial fractions,

u[n] u[n]
x[n] = −
(n + 2) (n + 3)

Taking z- transform, evaluating each term separately,

∞ ∞
X z −n X z −n z −0 z −1 z −2
u[n] = = + + ....
−∞
n+2 0
n+2 2 3 4

z −2 z −3 z −4
= z2( + + ....)
2 3 4
= −z 2 (log(1 − z −1 ) − z −1 )
Similarly the 2nd term
∞
X z −n z −2
u[n] = −z 3 (log(1 − z −1 ) − z −1 − )
−∞
n+3 2

So, z-transform of x[n] is

z
X(z) = z 2 − + z 2 (z − 1)log(1 − z −1 )
2
Question 9
Given the LCCDE:

y[n] − 3y[n − 1] + 2y[n − 2] = x[n]

and x[n] = 2n u[n] also given that the system is causal find y[n].
1. y[n] = (1 + 0.5n)2n u[n] + 3n 2u[n]
2. y[n] = (1 − 0.5n)2n u[n] + 3n 2u[n]
3. y[n] = (1 + 0.5n)2n u[n] − 3n 2u[n]

4. y[n] = (1 + 0.5n)2n u[n] + 3n 2u[n]

6
Solution : (4)
First we find the system function H(z) from the given LCCDE

y[n]5y[n1] + 6y[n2] = x[n] =⇒ Y (z)(15z 1 + 6z 2 ) = X(z)

Thus,
Y (z) 1
H(z) = =
X(z) 1 − 5z + 6z −2
−1

Thus H(z) has poles at z = 2, 3 and for the given input X(z) has poles at z =
2 thus Y (z) will have one pole of multiplicity 2 at z = 2 and another pole of
multiplicity 1 at z = 3. Thus,

y[n] = (an + b)2n u[n] + c3n u[n]

Since the system is causal y[n] = 0 for all n ≤ 0 thus y[0] = x[0] = 1, y[1] =
x[1] + 3y[0] = 2 + 3 = 5 and y[2] = x[2] + 3y[1] 2y[0] = 4 + 15 1 = 18. Now,
we have three equations and three variables to solve for

b+c=1

2a + 2b + 3c = 5
8a + 4b + 9c = 18
This gives a = 0.5, b = -1 and c = 2 thus we get the output y[n] as

y[n] = (1 + 0.5n)2n u[n] + 3n 2u[n]

Question 10
If the input to an LSI system is x[n] = ( 31 )n u[n] + 2n u[−n − 1] and its corre-
sponding output is y[n] = 5( 13 )n u[n] − 5 23 )n u[n] Which of the following is true
for the system?
1. Stable as well as causal
2. Stable but not causal
3. Causal but not stable
4. Neither causal nor stable
Solution : (1)
1 1 1
X(z) = 1 −1 − ; ≤ |z| ≤ 2
1− 3z
1 − 2z −1 3
n z 1
(Using a u[n] →
− 1−az −1 ; |z| < |a|
z 1
−an u[−n − 1] →
− 1−az −1 ; |z| > |a|)

5 5 2
Y (z) = − ; |z| >
1 − 13 z −1 1 − 32 z −1 3

7
 Y (z) 1 − 2z −1 2
H(z) = = ; |z| >
X(z) 1 − 23 z −1 3
Stable : The system is stable as ROC includes the unit circle
Causal : The system is causal because |z| → ∞ is included in the ROC.

Question 11
Consider a causal LSI system given by the transfer function
z
H1 (z) =
(z − 0.8)(z − 0.2)

and another system with transfer function H2 (z) such that its ROC is given by

Com-
ment on the stability of the system constructed by H1 and H2
1. BIBO Stable

2. unstable
3. can’t say
Solution :
From the graph, we can see that ROC is between 0.8 and 1.2 ant the system is

8
stable. For H1 (z) the the ROC is ≥ 0.8, the system is stable. For the cascaded
system,
RH1 ∩ RH2 ⊆ RH
Thus, RH contains the unit circle. Therefore the system is BIBO stable.

Question 12
The discrete time system’s transfer function is given by the following
1
H(z) =
1 − 4Z −1
hen which of the options is/are possible LCCDE relating the input x[n] and
output y[n]?
1. y[n] = 4y[n − 1] − x[n]
2. y[n] = 4y[n − 1] + x[n]
3. y[n] = 6y[n − 1] − 8y[n − 2] + x[n] − 2x[n − 1]

4. y[n] = 8y[n − 1] − 16y[n − 2] + x[n] − 4x[n − 1]

Solution : (2), (3)
Using the time shift property of z-transform
z
→ Y (z); y[n − k]xrightarrow[]zz −k Y (z)
y[n] −

Question 13
A series LC network forms a
1. High Pass filter

2. Low pass Filter

3. Band Pass filter
4. Bandstop Filter
Solution :
A series LC network forms a Band Pass filter
s2 LC + 1
H(s) =
sC
Question 14
The frequency response of a filter h[n] with real impulse response has an even
phase response
1. True
2. False

9
Solution : (2)
∞
X ∞
X ∞
X
−jwn jwn ∗
H(w) = h[n]e = h[n](e ) = (h[n]ejwn )∗ = H(w)∗
−∞ −∞ −∞
∞
X
H(−w) = h[n]ejwn
−∞

H(−w) = H(w)∗
H(w) is a Hermitian function. Hence, phase response is odd.

Question 15
The frequency response of a filter with real impulse response has an even mag-
nitude response
1. True
2. False

Solution : (1)
∞
X ∞
X ∞
X
H(w) = h[n]e−jwn = h[n](ejwn )∗ = (h[n]ejwn )∗ = H(w)∗
−∞ −∞ −∞
∞
X
H(−w) = h[n]ejwn
−∞

H(−w) = H(w)∗
H(w) is a Hermitian function. Hence, magnitude response is even.

10
Digital Signal Processing and its Applications Week 7

1. For which of the following, will the filter resource requirements increase?

(a) Large stop-band tolerance

(b) Small pass-band tolerance
(c) Sharp transition band
(d) (B) and (C)

Solution: D

2. The bilinear transformation is a mapping for obtaining a rational z-transform

Hd (Z) from a rational Laplace transform Hc (s). Consider an all pass filter
a−s
Hc (s) = a+s . Which of the following is true for a being real and positive?

(a) Pole of Hd (z) lies inside unit circle

(b) Zero of Hd (z) lies inside unit circle
(c) |Hc (jω)| = 1
(d) Zero of Hd (z) lies outside the unit circle

Solution: A, C, D

3. An analog filter is always:

(a) Rational and unstable

(b) Irrational and causal
(c) Rational, stable and causal
(d) Unstable

Solution: C

4. Following up on the discussion on bilinear transformations, a student claims

that, “all bilinear transformations from s to z will result in a satisfactory
transformation (as per the definition in lecture 20C)”. However, you don’t

1
 agree with this claim and are trying to disprove it with a counter-example.
Let’s consider that your initial guess for the counter-example was the fol-
lowing bilinear transformation:
2+s
z=
2−s
For this counter-example, select which property is not satisfied, thus, dis-
proving the claim

(a) The mapping is not strictly monotonic between Ω and ω

(b) A rational system in s may be mapped to an irrational system in z
(c) The mapping does not map imaginary axis to unit circle
(d) The mapping does not map LHP to the inside of unit circle
(e) None of the above and thus, the student’s claim is actually true
(f) None of the above, but the claim is false and we can give a better counter-
example to disprove it

Solution: F

5. Which of the following statements are true?

(a) Any nth order butterworth filter satisfying just the condition
D2 ΩS
log ≤ 2N log
D1 ΩP
meets all the design criteria
(b) All the poles of a normalized butterworth filter lie on the unit circle
(c) The second order butterworth filter is given by √1 and the third
s2 + 2s+1
order butterworth filter is given by s3 +2s21+2s+1
(d) The high frequency roll-off of an nth order normalized butterworth filter
(Ωc = 1) is -n

Solution: B, C, D

6. Phase response has the following characteristics:

2
 (a) Phase response can be specified along with magnitude response
(b) If magnitude response is specified, phase response cannot be insisted
upon
(c) Phase response must be satisfied to ensure causality
(d) A system can be causal with 0 phase response

Solution: B, C

7. Study the Butterworth filter graph, and select the correct statements:

(a) N is increasing from blue to red i.e lower filter order sharpens the graph.
(b) N is decreasing from blue to red i.e. low filter order smoothens the graph.
(c) To get a smaller transition band, we have to reduce the order of the
filter.
(d) The magnitude of ωc at the point at which these meet is 0.707 (approx-
imately).

Solution: B, D

8. Why do we need a tangent function of ω from s-plane to z-plane mapping?

(a) This gives a one-to-one mapping between Ω to ω

(b) The mapping is invertible and monotonically increasing
(c) Tangent function can be represented as a rational in the z-plane
(d) All of the above

Solution: D

9. Which statement is correct for the following transform?

s = log(z)
Assume that the imaginary part of log(z) ∈ (−π, π].

(a) Mapping is one-to-one from the z-domain to the s-domain

3
 (b) Each circle in the z-plane is mapped to an infinite line in the s-plane
(c) The interior of the |z| = 1 is mapped to a subset of the region Re(s) < 0
(d) The entire s-plane is covered by the transform

Solution: A, C

10. Which of the following is true about the FIR realization of a filter?

(a) Require less resources than an IIR filter

(b) Always causal
(c) Always stable
(d) Can be irrational

Solution: C

4
Figure 1: Butterworth filter graph

5
Digital Signal Processing and its Applications Week 8
1. Which of the following options is not true regarding the nature of poles of
function, Hanalog (s) of Low-pass Butterworth Filter?
(a) The poles occur on a circle with radius Ωc
(b) The non-real poles occur in complex conjugate pairs.
(c) The non-real poles appear as mirror image about the imaginary axis.
(d) The poles are at an angle π6 to each other for N = 3.

Solution: C,D

A, B follow directly from definition of Hanalog (s).

C is false as non-real poles occur in complex conjugate pairs.
D is not true as poles appear at angles of π3 for N=3.
2. How does odd or even N impact Butterworth and Chebyshev Filter?
(a) The magnitude response of Chebyshev filter is not affected by N odd/even.
(b) The magnitude response of Butterworth filter is not affected by N odd/even.
(c) The magnitude response of Chebyshev filter for even N has a factor of
√1
1+ϵ
(d) The magnitude response of Chebyshev filter for odd N has a factor of
√1
1+ϵ

Solution: B,C
Follows from the formula for the magnitude response of the filters discussed
in the class.
3. For a discrete Butterworth filter to have a passband edge frequency of 10Hz,
a stopband edge frequency of 90Hz, δ1 is at 0.05 and δ2 is at 0.1. The
sampling frequency is at 200 Hz, the Normalized Digital Frequencies, ωp
and ωs are:
(a) ωp = 0.314 rad/s and ωs = 1.885 rad/s
(b) ωp = 0.157 rad/s and ωs = 1.885 rad/s
(c) ωp = 0.157 rad/s and ωs = 0.9425 rad/s
(d) ωp = 0.314 rad/s and ωs = 0.9425 rad/s

1
 Solution: A
Normalised frequency ω for a frequency f is given by:
2πf
ω=
fs
4. For the Butterworth filter in the previous question, the corresponding fre-
quencies of the analog LPF Ωp and Ωs are:
(a) Ωp = 0.0765 rad/s and Ωs = 0.688 rad/s
(b) Ωp = 0.0765 rad/s and Ωs = 1.376 rad/s
(c) Ωp = 0.153 rad/s and Ωs = 1.376 rad/s
(d) Ωp = 0.314 rad/s and Ωs = 0.688 rad/s

Solution: C
The corresponding frequencies of the analog LPF Ωp and Ωs are given by:
ω
Ω = tan
2
5. For the Butterworth filter in the previous question, the order of the low pass
filter N can be:
(a) 1
(b) 2
(c) 3
(d) 4

Solution: B,C,D
1 1
D1 = (1−δ1 )
2 − 1 = 0.11 D1 =− 1 = 99
δ2 2
 q  
D2
 log q D1  
N≥ 
 log ΩS 
 Ωp 

6. For the Butterworth filter in the previous question, the cut-off frequency
(Ωc ) of the low pass filter is for minimum order of the filter:
(a) 0.2
(b) 0.4
(c) 0.6
(d) 0.8

2
 Solution: B
ΩP Ωs
1 ≤ Ωc ≤ 1
D12N
D22N
0.265 ≤ Ωc ≤ 0.436
7. For the Butterworth filter in the previous question, the normalized transfer
function of second order analog function (H(s)) for minimum order of the
filter is:



(a) H(s) = 0.2305/ s2 + 0.87s + 0.26



(b) H(s) = 0.1904/ s2 + 0.61s + 0.19



(c) H(s) = 0.1723/ s2 + 0.23s + 0.87



(d) H(s) = 0.2367/ s2 + 0.24s + 0.49

Solution: B

Ωc 2
H(s) = √

s2 + Ωc 2s + Ωc 2
8. For the Butterworth filter in the previous question, the z-transform of the
normalized transfer function (H(z)) for minimum order of the filter is:
(a) H(z) = 0.105 + 0.210z −1 + 0.105z −2 / 1 − 0.896z −1 + 0.317z −2

(b) H(z) = 0.235 + 0.840z −1 + 0.128z −2 / 1 − 0.136z −1 + 0.563z −2

(c) H(z) = 0.235 + 0.840z −1 + 0.128z −2 / 1 − 0.896z −1 + 0.317z −2

(d) H(z) = 0.105 + 0.210z −1 + 0.105z −2 / 1 − 0.136z −1 + 0.563z −2

We apply s = 1 − z −1 / 1 + z −1 , to obtain H(z).

Solution: A

9. Given a Low pass filter design with equi-ripple pass-band and monotonic
stop-band with following unnormalised frequency specs,
Pass band edge fp : 20KHz
Stop band edge fs : 25KHz
Pass band tolerance δ1 : 0.1
Stop band tolerance δ2 : 0.05
Sampling rate: 100KHz
Which of the following values of epsilon parameter (ϵ) satisfy these specifi-
cations?

3
(a) ϵ = 0.5
(b) ϵ = 0.4
(c) ϵ = 0.65
(d) ϵ = 0.48

Solution: B,D
The filter design involved is the Chebyshev filter, as the pass-band is equi-
ripple and stop-band is monotonic. To normalise the frequency, we have the
relation
2πf
ω=
fs
Hence,
2πfp 2π
ωp = =
fs 5
and,
2πfp π
ωs = =
fs 2
Using the Bilinear Transformation Ω = tan ω2 ,

Ωp = 0.726
and,
Ωs = 1
The design parameter constraints for the chebyshev filter are (i) Parameter
Epsilon ϵ
1
ϵ2 ≤ −1
(1 − δ1 )2
Substituting the values,
1
ϵ2 ≤ −1
(0.9)2
19
ϵ2 ≤
81
ϵ ≤ 0.484
Therefore, the maximum value of parameter epsilon can be 0.484 (3 decimal
points). For values of ϵ greater than this value, the pass-band tolerance is
not correctly realised. Hence, filter does not meet
required specifications if parameter epsilon doesn’t satisfy this range. Hence,
Options A and C are incorrect.

4
10. Given a Low pass filter design with equi-ripple pass-band and monotonic
stop-band with following unnormalised frequency specs,
Pass band edge fp : 20KHz
Stop band edge fs : 25KHz
Pass band tolerance δ1 : 0.1
Stop band tolerance δ2 : 0.05
Sampling rate: 100KHz
Which of the following values of epsilon parameter (ϵ) satisfy these specifi-
cations?
(a) ϵ = 0.5 and N = 12
(b) ϵ = 0.4 and N = 5
(c) ϵ = 0.65 and N = 4
(d) ϵ = 0.48 and N = 7

Solution: D

Order of the filter N

 √ 
−1 (D2 )
 cosh  ϵ
N ≥   
 −1 Ω
cosh s

Ωp


1
Where, D2 = (δ2 )
2 − 1 = 399, Ωs = 1, Ωp = 0.726. Hence,
 √ 
−1 (399)
 cosh ϵ 
N ≥ 
−1

 cosh (1.375) 

 

Substituting the values as per given options, (B) for epsilon = 0.4, The value
of N is N ≥ ⌈5.47⌉. From the ceil function, the required value of N is the
next(bigger on number line) closest integer which is 6 . Since, given order is
not greater than equal to 6 , the filter does not meet the specifications.
(D) for epsilon = 0.48, The value of N is N ≥ ⌈5.25⌉. From the ceil function,
the required value of N is the next(bigger on number line) closest integer
which is 6 . Thus, the given order filter can meet the specifications and is
therefore, the correct option.

5
11. Which of the following is/are true regarding Chebyshev Filter (CF) and
Butterworth filters (BF)?
(a) CF allows rippes in magnitude response but BF does not allow ripples
in magnitude resonse.
(b) CF have a smaller transition band, whereas in BF transition band is
comparitively larger.
(c) CF poles lie on ellipse, whereas BF poles lie on a circle.
(d) CF have a smaller order than BF.

Solution: A,B,C,D

12. Butterworth filters have a more linear phase response as compared to Cheby-
shev Filter. The given statement is?
(a) True
(b) False

Solution: True

13. The given pole-zero diagram corresponds to a stable filter of

(a) Butterworth Type

6
 (b) Chebyshev Type
(c) Elliptic Type
(d) None of the above

Solution: D
For the filter to be stable the poles must lie inside the unit circle.
14. Which of the following is/are true for Chebyshev filter of order N :
(a) N will increase with the wider transition band.
(b) N will decrease with the wider transition band.
(c) N will not be impacted with wider transition band.
(d) None of the above

Solution: B
This question deals with the engineering question ask more invest more with
denominator analysis of N. N implies the resources needed to improve quality
of the filter and the investment required.
With a wider transition band we are asking for less. This aspect deals with
Engineering needs.

7
 DSP-Assignment-Week6

July 10, 2022

Question 1
The transformation from low pass to band stop is a transformation with order
of atleast
1. 1
2. 3
3. 2
4. 0
Solution : (3)
Since we want frequencies to map to multiple places (eg. stop band) we would
require a transformation of atleast order 2.

Question 2
The following is a filter of type

1. Low Pass
2. Band stop
3. High pass
4. Band pass

1
Solution : (2)
1 sL
H(s) = 1 =
sL + sC 1 + s2 LC
Question 3
Tolerances will get violated with frequency transformation for converting low
pass filter to band stop filter
1. True
2. False
Solution :
In frequency transformations only the independent variable is changed while the
dependant variable is carried over. Hence tolerances will be carried as it is.

Question 4
Choose the correct option

ωBS ωL
(a) 0+ (i) ∞
(b) ωS1 (ii) 0+
B
(c) ωS2 (iii) ωP 2 −ωP 1
B
(d) ωP 1 (iv) ωP 1 −ωP 2
BωS1
(e) ωP 2 (v) 2
ωP 2 ωP 1 −ωS1
BωS2
(f) ∞ (vi) 2
ωP 2 ωP 1 −ωS2

where B = ωP 2 − ωP 1
1. (a)-(ii),(b)-(v),(c)-(vi),(d)-(iii),(e)-(ii),(f)-(i)
2. (a)-(i) ,(b)-(v),(c)-(vi),(d)-(iii),(e)-(ii),(f)-(ii)
3. (a)-(ii),(b)-(vi),(c)-(v),(d)-(ii),(e)-(iii),(f)-(i)
4. (a)-(i) ,(b)-(v),(c)-(vi),(d)-(ii),(e)-(iii),(f)-(ii)
Solution : (i)
BωBS
ωL = 2
ωP 1 ωP 2 − ωBS
Butterworth Filter Design
Consider a design of a butterworth band stop filter with the following specifi-
cations:
• Passband Frequencies : 5kHz and 55kHz
• Stopband frequencies : 10kHz and 50kHz

2
 • Sampling frequencies : 200kHz
• Passband tolerance : 0.25
• Stopband tolerance : 0.2
Question 5
Find the normalized frequency specifications (w)
1. 0.157,0.314,1.571,1.727
2. 0.05,0.1,0.5,0.55

3. 0.157,0.314,1.681,1.927
4. 0.0786, 0.1580, 1, 1.1698
Solution : (1)
2π
To get normalized frequency (w) we multiply the frequencies(in Hz) with fs
Question 6
What are the transformed frequencies in the analog domain (ω)
1. 0.0786, 0.1580, 1.5, 1.1698
2. 0.0786, 0.1580, 1, 1.1698
3. 0.0886, 0.1680, 1, 1.1698

4. 0.0786, 0.1680, 1, 1.1798

Solution : (2)
w
ω = tan( )
2
Question 7
Choose the correct frequency transformation for Band stop filter
Bω
1. ωL = ω 2 −ωo2

Bω
2. ωL = ωo2 −ω 2

ω 2 −ωo2
3. ωL = Bω
B
4. ωL = ω−ωo

Solution (1):
Bω
ωL =
ω2 − ωo2
Question 8
Find the value of B (write ans upto 3 decimal places)

3
(Numerical ans )

Solution : (3)

B = ωP 2 − ωP 1
Question 9
Find the value of o (write ans upto 3 decimal places)
(Numerical ans )

Solution : (3)
√
ωo = ωP 1 ωP 2
Question 10
For design using the butterworth approximation, Find the best value of N.
(Numerical ans )
Solution :
1
D1 = − 1 = 0.5625
1 − δ2
1
D2 = 2 − 1 = 24
δ
q
log( D2 D1 )
N≥ ωL S
log( ωL P )

where, ωLP = 1 ωLS = min( ωBω BωS2

2 −ω 2 , ω 2 −ω 2 ) = 0.437 Thus N = 3
S1
S1 o S2 o

Question 11
Find the frequency response of the rectangular window function of size 2N cen-
tered at 0.
sin(2N w)
1. sin(w/2)

sin(N w)
2. sin(w)

sin(N w)
3. sin(w/2)

sin(2N w)
4. sin(w)

Solution : (3)
The window function is such that: v[n] = 1 f or n ∈ −N + 1/2, −N + 3/2, .... − 1/2, 1/2, 3/2......N − 1/2
v[n]=0 otherwise X
V (w) = v[n]e−jwn
n
−jw(−N +0.5) −jw(−N +1+0.5)
=e +e +.....e−jw(0.5) +ejw(0.5) +.....ejw(N −1−0.5) +ejw(N −0.5)

4
 e−jw(2N ) − 1
= e−jw(−N +0.5)
e−jw − 1
e−jw(N ) − ejwN
= e−0.5jw
e−jw − 1
sin(N w)
=
sin(w/2)
Question 12
A non-causal FIR filter can always be made causal by introducing finite delay.
1. True
2. False
Solution :
An FIR filter has finite no. of terms. Therefore it can be made causal by
introducing a finite delay.
Question 13
Causal rational IIR filters can have linear phase
1. True
2. False
Solution : (2)
The frequency response of any discrete-time IIR filter with lonear phase has the
following form
H(ejw ) = Φ(w)e−jwτ
where ω ∈ [−π, π] and τ ∈ R Here Φ(w) can be either purely real and even or
purely imaginary and odd to ensure that ther is no additional phase contribution
from Φ(w) We take Φ(w) ∈ R

h[n] = ϕ[n] ∗ i[n]

i[n] = IDT F T [e−jwτ ]

sin(π(n − τ ))
=
π(n − τ )
= δ[n − τ ]
¯
Also,ϕ[−n] = ϕ[n]
Proof : Z π
1
ϕ[−n] = Φ(w)e−jwn dw
2π −π
Z π
1 ¯ dw
= Φ(w)ejwn
2π −π
1 π ¯ jwn
Z
= Φ(w)e dw
2π −π

5
 ¯
= ϕ[n]
When ϕ[n] has infinite length (it is IIR), h[n] cannot be made causal by finite
shifts of the impulse response.
h[n[= ϕ[n] ∗ δ[n − τ ]
Therefore an IIR filter cannot be causal and have linear phase at the same time.
Question 14
Which of the following statements about FIR Filters are correct?
1. The transition bandwidth of the FIR Filter is most affected by the first
sidelobe width of the window function.
2. The tolerance of the FIR Filter Impulse Response is most affected by the
sidelobe area of the window function.
3. The FIR Filter obtained using a Triangular Window Function has the
minimum L2-Norm.
4. The rectangular window function of width (2N+1) centered at n=0 is
given by sin(N w/2)
sin(w/2)

Solution :
Option 1
The transition bandwidth occurs when the mainlobe of the frequency response
of the window function passes through the edge of the ideal low pass rectangular
filter. Thus, the transition bandwidth will mainly be affected by the mainlobe
width, not the sidelobe width.
Hence, the given statement is False.

Option 2
The tolerance of the FIR Filter, for both the stopband and the passband is
obtained by the convolution of the sidelobes along with the Ideal LP Filter. So,
the tolerance is obtained by the area under the sidelobes, which lie within the
Low pass filter. So, the sidelobe area mainly effects the tolerance of the Filter.
Hence, the given statement is True.

Option 3
As we had seen in the lecture about Advantages of FIR Filters, the rectangular
window has an advantage over other types of FIR filters that it minimizes the
L2-norm of the tolerances.
So, the given statement is false.

Option 4
The DTFT of the impulse response of the rectangular window function can be
given by:
XN
X(w) = ejwn
n=−N

6
 1 − e−jw(2N +1)
X(w) = e−jwn
1 − e−jw
e−jw(N +1)/2 (ejw(N +1)/2 − e−jw(N +1)/2 )
X(w) = e−jwn
e−jw/2 (ejw/2 − e−jw/2 )
2jsin(jw 2N2+1
X(w) =
2jsin( jw
2 )

sin(jw 2N2+1
X(w) =
sin( jw
2 )
Hence, the last option is also False.

7
 Week 10-Assignment Questions

July 16, 2022

1 Paragraph Q1-Q3
A window function (also known as an apodization function or tapering function)
in signal processing is a mathematical function that is zero-valued outside of a
given interval, normally symmetric around the middle of the interval, near a
maximum in the middle, and tapering away from the middle. When another
function or waveform/data-sequence is ”multiplied” by a window function, the
product is also zero-valued outside the interval: all that remains is the overlap,
the ”view through the window.” In practise, the segment of data within the
window is isolated first, and then only that data is multiplied by the window
function values. Thus, the primary purpose of window functions is tapering
rather than segmentation.
Given 2 window functions
i)Rectangular( Dirichlet window)
ii)Bartlett (Triangular window)

1
 Figure 1: Window Function

The main lobe width of a rectangular window determines the width of the
filter transition band, whereas the PSL(Peak Side Lobe) of the window affects
the ripples in the pass-band of the filter. It was shown that as we reduce the
main lobe width and the P SL, the achieved filter gets closer to the ideal re-
sponse.

Q1. Choose the correct statements from the below given Options(Multiple Op-
tions can be Correct)-
(a)In Rectangular window function the Main lobe width is independent of M.
(b)In Rectangular window function the Main lobe width increases as M in-
creases.
(c) In Rectangular window function the Main lobe width decreases as M in-
creases.
(d) Main lobe peak is independent of M .

Ans-c)

2 )
sin( ωM
( )
sin( ω
ω ̸= 0
WRectangular,M (ω) = 2)
M ω=0

Q2. Consider a rectangular and triangular window of length 11,21 respec-

tively.Peak side lobe of rectangular function be x dB below Peak Main lobe
and Peak side lobe of triangular function be y dB below Peak Main lobe. Find
x + y.(Round off to nearest integer)

Ans = 39 dB

2
    
1 1
wBartlett,21 [n] = √ wRectangular,11 [n] ∗ √ wRectangular,11 [n]
11 11

1 2
wBartlett,21 (ω) = (wRectangular,11 (ω))
11

20 log(2.4/11) = 13 dB

Q3. Match the following for Rectangular window function of size M.

1)Amplitude of the envelope roll of (in dB ) (a)6dB/octave
2) Amplitude of the envelope roll of (as function of w) (b) 12 dB/ octave
3)Main side lobe width(in terms of M) (c) 1/w
4)First side lobe width(in terms of M) (d) 1/w2
(e)2π/M
(f)π/M
Ans 1-a,2-c,3-e,4-e

Q4. With respect to Gibbs phenomenon in FIR Filter mark the correct state-
ment
(a) As number of taps increases number of ripple increases
(b) As number of taps increases width of ripple increases
(c)drop off in the main lobe gets less steep as filter length increases
(d)height of lobes(other than main lobe) increases as filter length increases

Ans-a)

For FIR Filters-3 • When the number of taps of the FIR filter increases, num-
ber of ripples increases but the width of ripples decreases. Also, the height
of the largest ripples are independent of the length of the filter. And, as filter
length increases height of all other ripples decreases. - The drop-offs in the main
lobes get steeper because their widths gets narrower as the filter length increases

2πk


Q5. Consider a hamming window: a + b cos 2N

   
2πk 4πk
blackmann window : c + d cos + e cos
2N 2N
choose from the given options below the value of expression (a + b)(c + d) :

3
 (a) 1
(b) 0.92
(c) 0.08
(d)1.02

Ans: b) 0.92

Q6. Time domain Impulse response h[n] of digital differentiator whose fre-
quency response is given by:

H ejω = jω


−π ≤ω ≤π

(a) sin(nπ)
n , n ̸= 0
cos(nπ)
(b) nn , n ̸= 0
(c) (−1)
n , n ̸= 0
sin(nπ)
(d) nπ , n ̸= 0

Answer -b)

Q7. Given P h[n] above find whether the above impulse response is stable by
n=∞
calculating n=−∞ h[n]
(a) 1
(b) 0
(c) π6
(d)diverges

Ans-b) 0

Q8.For a FIR filter of length N(odd) to have linear phase delay which of the
following is true:
(a) h(n) = h(N − 1 − n)
(b) h(n) = −h(N − 1 − n)
(c) h(n) = h((N − 1)/2 − n)
(d) h(n) = h((N − 1)/2 + n)

Ans-a), b)

Answer the following Questions with respect to designing of Ideal High Pass
filter with Hamming window
π
Hd ejω = 1


− ≤ |ω| ≤ π
4

4
 0 otherwise

 
2πk N −1 N −1
HammingW indow = wh [n] = 0.54+0.46 cos − ≤k≤
N −1 2 2

h[n] = wh [n]hd [n]

N = 11
Q9)Choose the correct expression for H(z)z transform of h[n] (time domain
filter response):
Pn=11
(a) H(z) = n=−11 h[n]z n
Pn=10
(b) H(z) = n=−10 h[n]z n
Pn=5
(c) H(z) = n=−5 h[n]z n
Pn=22
(d) H(z) = n=−22 h[n]z n

Ans-c)

Q10)Find the correct option about the window function coefficients H[n] =
wh [n]hd [n]
(a) wh [0] = 1, wh [1] = 0.208, wh [2] = 0.116
(b) wh [0] = 0.5, wh [4] = 0.75, wh [5] = 0.04
(c) wh [0] = 0.75, wh [1] = 0.208, wh [4] = 0
(d) wh [1] = 1, wh [2] = 0.0035, wh [3] = 0.0044

Ans-c)

Q11)Find the sum of h[0] + h[1] + h[2] − (h[−1] − h[−2])

(a) 1
(b) 0
(C) −1
(d) 0.75

Ans d)

Q12)
Statement 1:All causal systems are rational

5
Statement 2:All rational systems are causal
Statement
P 3:To prove:Rational and Causal systems can always be written in the
m=M
m b z −m
form m=0
P n=N
1− an z −n
n=1
a)only 1&3 is true
b)only 2&3 is true
c) Only 3 is true
d) Only 1 is true
e) Only 2 is true
f) None of the above

Ans c)

Q13)In which of the following Gibbs phenomenon occurs W1 = rectangular

wave base width T
W2 triangular wave base width T
W3 trapezoidal wave base width T
W4 saw tooth wave base width T
a) W1.W2
b) W1.W3
c) W1
d) W 2
e) W4

Ans - c),e)

Q14)Find the correct optionwith respect yo the given below direct from I flow
graph(Figure 2)
a) y[n] = a1 y[n − 1] + a2 y[n − 2] + b0 x[n] + b1 x[n − 1] + b2 x[n − 2]
b) y[n] = a1 y[n] + a2 y[n − 1] + b0 x[n] + b1 x[n − 1] + b2 x[n − 2]
c) y[n] = a1 y[n − 1] + a2 y[n − 2] + b0 x[n − 1] + b1 x[n − 2] + b2 x[n − 3]
d) y[n] = a1 y[n] + a2 y[n − 1] + b0 x[n − 1] + b1 x[n − 2] + b2 x[n]

Ans-a)

Q15)Choose the correct options regarding the direct form 2 representation of

the following
Y [n] − 0.75y[n − 1] + 0.125y[n − 2] = x[n] + 0.333x[n − 1]

6
 Figure 2

Figure 3

a)a=1
b)b=-0.75
c)c=0.125
d)None of these

Ans-d)

For a given filter specification study the above filter response of butterworth(B),chebychev(C),elliptic(E)
filters and compare them.

Q16)Order of resources required

(a) E < C < B

7
 (b) C < E < B
(C)B < C < E
(d)B < E < C

Ans-a)

Q17)Arrange them in increasing order of transition band steepness

(a) B > C > E
(b) E > C > B
(c) B > E > C
(d) E > B > C

Ans b)

Q18) Arrange them in increasing order of linearity of phase response

(a) E < C < B
(b) B < C < E
(C)B < E < C
(sd)E < B < C

Ans a)

Q19)
Which of the following is true when compared to E,C,B filters
(a)FIR filter requires the least amount of resources.
(b)Has the most linear phase response
(c)sharpest transition band
(d) None of the above.
Ans b)

Paragraph Q20-Q22
By using Kaiser window attempt is made to reduce the sidelobe level &
control by modifying nth order Bessel function I0(x)& controlling the shape
factor by adjustable parameter α. Few dB sidelobe level response is also reduced.

 0.1102(A − 8.7), A > 50
β= 0.5842(A − 21)0.4 + 0.07886(A − 21), 21 ≤ A ≤ 50,
0.0, A < 21


where A = −20 log10 δ, and

8
 A−8
M=
2.285∆ω

where ∆ω is the transition band width in the design specifications.

Design a FIR linear phase Filter using kaiser window to meet the following
specifications.

0.99 < H(ejw ) < 1.01 |w| < 0.19π

H(ejw ) < 0.01 0.21π < |w| < π

Q20) Calculate A Attenuation parameter

A = −20 log 0.01 = 40

Q21)Calculate M(integer)

M = (A − 8)/2.285∆ω = 223

Q22)Calculate Beta

Beta = 0.5842 ∗ (19)∧ 0.4 + 0.07886 ∗ 19 = 3.395

9
1 Week 11− Assignment Questions
Q1-3 Paragraph Questions Given a three stage lattice filter with coefficients
K1 = 0.25, K2 = 0.5 and K3 = 13 , determine the FIR filter coefficients for the
direct-form structure.

Q1)Find A1 (z)
(a) A1 (z) = 1 + 13 z −1
(b) A1 (z) = 1 + 14 z −1
(c) A1 (z) = 14 + z −1
(d) A1 (z) = 13 + z −1

Answer-(b)

Q2)Find A2 (z)
(a)A2 (z) = 1 + 38 z −1 + 14 z −2
(b)A2 (z) = 1 + 14 z −1 + 38 z −2
(c)A2 (z) = 1 + 34 z −1 + 12 z −2
(d)A2 (z) = 1 + 38 z −1 + 12 z −2

Answer-(d)

Q3)Find A3 (z)
−1
(a) A3 (z) = 1 + 13
24 z + 85 z −2 + 13 z −3
5 −1 −2
(b)A3 (z) = 1 + 8 z − 13 24 z + 13 z −3
1 −1 5 −2 −3
(c) A3 (z) = 1 + 3 z + 8 z + 13 24 z
5 −1 1 −2 13 −3
(d)A3 (z) = 1 + 8 z − 3 z + 24 z

Answer-a)

Q4-7 Paragraph Question -Determine the lattice coefficients K1 , K2 , K3 of the

following FIR Filter.
3 1 1
H(z) = 1 + z −1 + z −2 + z −3
4 2 4

Q4)Which of the following is the correct matrix representation of the lattice

structure
(a)     
Am (z) 1 Km Am−1 (z)
=
Bm (z) Km 1 Bm−1 (z)
(b)     
Am (z) Km 1 Bm−1 (z)
=
Bm (z) 1 Km z −1 Am−1 (z)

1
 (c)     
Am (z) 1 Km Bm−1 (z)
=
Bm (z) Km 1 z −1 Am−1 (z)
(d)     
Am (z) 1 Km Am−1 (z)
=
Bm (z) Km 1 z −1 Bm−1 (z)

Answer-(d)

Q5)Find K1
(a) 21
(b) 1
(c) 14
(d) 34

Answer -(a)

Q6)Find K2
a) 12
b) 1
c) 13
d) 34

Answer -c)

Q7)Find K3
(a) 14
(b) 1
(c) 34
(d) 12

Answer a)

Q8-10 Paragraph Questions-Understand the Cascade (Direct form 2) Realiza-

tion of the following system and answer the following questions

1 − z −1
H(z) = = H1 (z)H2 (z)
1 − 0.2z −1 − 0.15z −2
Q8)Find poles of the denominator
a) 0.5, 0.3

2
 Figure 1:

b) −0.5, 0.3
c) −0.5, −0.3
d) 0.5, −0.3

Answer-d)

Q9)Find H2 (z)
1−z −1
(a) 1+0.3z −1
1
(b) 1+0.3z −1
1
(c) 1−0.3z −1
−1
1−z
(d) 1−0.3z −1

Answer b)

Q10)Calculate the value of a + b

(a) −0.2
(b) 0.8
(c) 0.2
(d) 1

Answer-(c)

Q11)Now consider the parallel realization as follows

3
 H(z) = H1 (z) + H2 (z)
Find c-d
(a) 1
(b) 2.3
(c)2
(d)-1

Answer-(b)

4
 Week 12-Assignment Questions

July 11, 2022

1 Paragraph Q1-Q4
Consider 2 finite sequences x = [x[0], x[1], x[2], x[3]], h = [h[0], h[1], h[2], h[3]]
whose DFT’s are given by

X = DF T (x) = [1, j, −1, j]

H = DF T (h) = [0, 1 + j, 1, 1 − j]

Q1. Compute the DFT of the following DF T (x[3], x[0], x[1], x[2]])
(a) [1, j, −1, −j]
(b) [1, −1, 1, −1]
(c) [1, 1, 1, 1]
(d)[1, j, 1, j]

Answer-(c)

Q2.Consider the DFT ([h[0], −h[1], h[2], −h[3]]) = [H′ [0], H′ [1], H′ [2], H′ [3]].Calculate
the sum of H′ [0] + H′ [1] + H′ [2] + H′ [3]
(a) 0
(b)3
(c) 2
(d) 1

Answer-(b)

Q3.Compute the DFT(h ⊛ x)

(a) [0, −1 + j, −1, 1 + j]
(b) [0, 1 + j, −1, 1 + j]
(c) [0, −1 + j, −1, 1 − j]
(d) [0, −1, −1, 1]

Answer- (a)

1
Q4. Let y = [x[0], h[0], x[1], h[1], x[2], h[2], x[3], h[3]].Then which of the following
expression is correct regarding the DF T (y) = Y
(a) Y [k] = ω4k X[k] + H[k],for k = 0, . . . , 3
(b) Y [k] = X[k] + ω4k H[k],for k = 0, . . . , 3
(c) Y [k] = ω8k X[k] + H[k], for k = 0, . . . , 7
(d) Y [k] = X[k] + ω8k H[k], for k = 0, . . . , 7

Answer-(d)

2 Paragraph Q5-Q8
Consider,

x[n] = {1, 2, −3, 4}

↑

h[n] = {2, 5}
↑
′ ′
x [n] = { , , , } h [n] = { , , , }
↑ ↑
′ ′ ′ ′ ′
Q5.Find x (n) = {x [0], x [1], x [2], x [3]}
a) x′ [2] = 1
b) x′ [2] = −3
c) x′ [3] = 2
d) x′ [3] = 4

Answer-(a),(c)

Q6. Find h’(n) = {h′ [0], h′ [1], h′ [2], h′ [3]}

a) h′ [1] = 5
b) h′ [2] = 0 c) h′ [0] = 0
d) h’ [3] = 2

Answer-(a),(b),(c),(d)

Q7)Find the trace of Matrix M s.t y = M .h’

(a)-12
(b) 4
(C) 8
(d) 16

2
Answer-(a)

Q8.let y be the circular convolution y = x′ ⊛ h′ , {y ′ [0], y ′ [1], y ′ [2], y ′ [3]}

Find y[0] + y[3]

Answer = −3

Q9.If x(n) is a complex valued sequence given by x(n) = xR (n) + jxl (n), then
what is the DFT of xR (n)?
PN
a) n=0 xR (n) cos 2πkn 2πkn
N + xI (n) sin N
PN
b) n=0 xR (n) cos 2πkn 2πkn
N − xI (n) sin N
PN −1
c) n=0 xR (n) cos 2πkn 2πkn
N − xI (n) sin N
PN −1
d) n=0 xR (n) cos N + xI (n) sin 2πkn
2πkn
N

Answer-(d)
R 6π
2
Q10.x[n] = {−1, 2, −3, 2, −1} Then the value of 0
X ejω dω

Answer-358.14

3 Q11-Q14 Paragraph Question

A butterfly is a part of a fast Fourier transform algorithm that combines the
results of smaller discrete Fourier transforms (DFTs) into a larger DFT, or vice
versa (breaking a larger DFT up into subtransforms). The term ”butterfly”
refers to the shape of the data-flow diagram in the radix-2 case, which is de-
scribed further below.

The butterfly in the radix-2 Cooley-Tukey algorithm is simply a DFT of size-2

that takes two inputs (x0, x1) (corresponding outputs of the two sub-transforms)
and gives two outputs (y0, y1) by the formula (without twiddle factors):

3
 Figure 1:

Below show is the 4 point dft in butterfly form

Figure 2:

The first branch corresponds to matrix multiplication

X = F.x where 
1 1 1 1
1 ω ω2 ω3 
F = 
1 ω2 ω4 ω6 
1 ω3 ω6 ω9
Now answer the following questions

Q11. Number of Multipliers used in ordinary matrix multiplication 4 point DFT

Answer = 16

4
Q12. Number of Multipliers used in total (Considering 1 complex multipli-
cation per butterfly with twiddle factors)

Answer=4= N2 · log N

Q13.Let the twiddle factor in place of question mark be T. Calculate the abs(Real(T))
upto 1 decimal place.

Answer-ω41 = 0.7

Q14.ratio of Total number of additions in ordinary dft to butterfly( 2 adds

per butterfly)
(a) N log N
(b) ( N2 − 1) log N
(c) N
(d) N 2 log N

Answer-(a)

Q15 What is the input order for 8 point df t

(a) x0, x4, x2, x6, x1, x5, x3, x7
(b) x0, x2, x4, x6, x1, x3, x5, x7
(c) x0, x1, x2, x3, x4, x5, x6, x7
(d) x0, x1, x4, x6, x2, x3, x5, x7

Ans -Reverse bit order (a)

Q16 X’=[x’[0],...,x’[7]]]=DFT (DFT(x)) ignore the scaling factor and assume

it is 8 point DFT
(a)x′ [0] = x[0]
(b) x′ [1] = x[5]
(c) x′ [1] = x[7]
(d) x′ [2] = x[6]

Answer-(a),(c),(d)