Brrs(c)

- (s) - to - 044 J

B. Tech Degree VI Semester (Supplementary) Examination

September 2010
CS/EE 602 DIGITAL SIGNAL PROCESSING

Time : 3 Hours

(2006 Scheme)

Maximum Marks : 100

PART A

(Answer ALL questions)

(8 x 5 = 40)
I.

(a)

Find the circular convolution of the sequences x[n]= [1,2,3] and

h[n] [1,0, 21. Derive the matrix H that represents the circular
convolution operator.
Given y[n]- 0.5 yin - 1] = 2x[n], find its causal impulse response and
anticausal impulse response.
Compute the total number of additions and multiplications required for
computing the 16 pt DFT of a sequence. How will the multiplication and
addition be reduced if a 16 pt radix 4 FFT algorithm is used? Ignore all
trivial multiplications. (multiplication by 1,T- j etc.).
Draw the signal flow graph for 8 point DIT FFT. Hence calculate the DFT

of x[n] = [1 2 3 4 5 67 8].
Explain the window method for FIR filter realization. What is the
disadvantage of rectangular window?
Realize the filter H(*)=

(h)

1+3-ZH

in Direct Form

+8*-

and Direct Form 2.

Explain Limit cycle oscillations.
Compare and contrast General Purpose microprocessors and DSPs.
PART B

(4 x 15= 60)
II.

(a)

Using long division method, find the inverse -Z- transform of

1

x(*)=1+-z--I+*--

for causal and anticausal X [rd.

(5)

Find the inverse * transform of X (*) = log(1 +

For the sequence

-1- " u[n] 4-1x[n]=(: ) u[-n -1],

3
2

compute

(5)

X (*) and plot

ROC. Will x[n] have a Fourier Transform?

OR

III.

(a)

Check the linearity, causality and stability of the following systems.

(i)
y[n]. 5 x[n]+ 3

y[n] = x2 [n]

y[n1= 3y[n +11-1-x[nl+ x[n -11

(b)
(c)

Compute the linear convolution of x[n] = [1, 30] and h[n] = [1,4,31
using circular convolution method.
Define Systems Transfer function. Hence for a system given by
y[n]= y[n-11+ x[n] , write down the system transfer function and then
compute the impulse response.

(5)

2
Derive the 8 point D.I.F FFT. Hence compute the DFT of the sequence
x[n] = [1,4,6,8,8,6,4,11 using the DIF FFT algorithm.
Show that DFT (x * [n]) =

[N - k] where N is the length of the

sequence and X [k] is the DFT of x[n].

OR

Compute the DFT of the sequence x

V.

[1,8,4,2,2,4,8,1]

given a

4 point FFT transformer (it can compute FFT of a 4 point sequence only).
The FFT transformer may be used only once.

VI.

(a)
(b)

Compare FIR and BR filters.

Design using rectangular windowing method, a LPF FIR filter with
following specifications.
Cut off frequency
=
500 Hz
Sampling frequency =
2kHz
10
Order of the filter
=

(15)

(5)

(10)

OR

VII.

(a)

Realize the filter

1+ 2+ -1

H () =(1- -E-1(1-

(b)

Direct Form 1
Direct Form 2
(iii)
Cascade Form, where each stage is implemented as a second order
system.
Realize the filter
+ 4 2 +5+ -3 + 4+-4 +3+ -5 ++ -6 in Direct Form.
H (-0)=1+

(8)

Is a linear phase response possible? Thereafter, realize the filter so as to

minimize the number of multiplications.

(7)

(6)
(9)

VIII.

(a)
(b)

Derive the expression for mean square quantization error.

Draw the block diagram of a typical fixed point DSP processor.

IX.

(a)
(b)

Draw the block diagram of a typical floating point DSP processor.

Explain in detail an application of DSP to image processing.

OR

***

(8)
(7)

Brrs(c)

- (s) - to - 044 J

B. Tech Degree VI Semester (Supplementary) Examination

September 2010
CS/EE 602 DIGITAL SIGNAL PROCESSING

Time : 3 Hours

(2006 Scheme)

Maximum Marks : 100

PART A

(Answer ALL questions)

(8 x 5 = 40)
I.

(a)

Find the circular convolution of the sequences x[n]= [1,2,3] and

h[n] [1,0, 21. Derive the matrix H that represents the circular
convolution operator.
Given y[n]- 0.5 yin - 1] = 2x[n], find its causal impulse response and
anticausal impulse response.
Compute the total number of additions and multiplications required for
computing the 16 pt DFT of a sequence. How will the multiplication and
addition be reduced if a 16 pt radix 4 FFT algorithm is used? Ignore all
trivial multiplications. (multiplication by 1,T- j etc.).
Draw the signal flow graph for 8 point DIT FFT. Hence calculate the DFT

of x[n] = [1 2 3 4 5 67 8].
Explain the window method for FIR filter realization. What is the
disadvantage of rectangular window?
Realize the filter H(*)=

(h)

1+3-ZH

in Direct Form

+8*-

and Direct Form 2.

Explain Limit cycle oscillations.
Compare and contrast General Purpose microprocessors and DSPs.
PART B

(4 x 15= 60)
II.

(a)

Using long division method, find the inverse -Z- transform of

1

x(*)=1+-z--I+*--

for causal and anticausal X [rd.

(5)

Find the inverse * transform of X (*) = log(1 +

For the sequence

-1- " u[n] 4-1x[n]=(: ) u[-n -1],

3
2

compute

(5)

X (*) and plot

ROC. Will x[n] have a Fourier Transform?

OR

III.

(a)

Check the linearity, causality and stability of the following systems.

(i)
y[n]. 5 x[n]+ 3

y[n] = x2 [n]

y[n1= 3y[n +11-1-x[nl+ x[n -11

(b)
(c)

Compute the linear convolution of x[n] = [1, 30] and h[n] = [1,4,31
using circular convolution method.
Define Systems Transfer function. Hence for a system given by
y[n]= y[n-11+ x[n] , write down the system transfer function and then
compute the impulse response.

(5)

2
Derive the 8 point D.I.F FFT. Hence compute the DFT of the sequence
x[n] = [1,4,6,8,8,6,4,11 using the DIF FFT algorithm.
Show that DFT (x * [n]) =

[N - k] where N is the length of the

sequence and X [k] is the DFT of x[n].

OR

Compute the DFT of the sequence x

V.

[1,8,4,2,2,4,8,1]

given a

4 point FFT transformer (it can compute FFT of a 4 point sequence only).
The FFT transformer may be used only once.

VI.

(a)
(b)

Compare FIR and BR filters.

Design using rectangular windowing method, a LPF FIR filter with
following specifications.
Cut off frequency
=
500 Hz
Sampling frequency =
2kHz
10
Order of the filter
=

(15)

(5)

(10)

OR

VII.

(a)

Realize the filter

1+ 2+ -1

H () =(1- -E-1(1-

(b)

Direct Form 1
Direct Form 2
(iii)
Cascade Form, where each stage is implemented as a second order
system.
Realize the filter
+ 4 2 +5+ -3 + 4+-4 +3+ -5 ++ -6 in Direct Form.
H (-0)=1+

(8)

Is a linear phase response possible? Thereafter, realize the filter so as to

minimize the number of multiplications.

(7)

(6)
(9)

VIII.

(a)
(b)

Derive the expression for mean square quantization error.

Draw the block diagram of a typical fixed point DSP processor.

IX.

(a)
(b)

Draw the block diagram of a typical floating point DSP processor.

Explain in detail an application of DSP to image processing.

OR

***

(8)
(7)

BTS(C) VI 10 006 J

B.

Tech Degree VI Semester Examination, April 2010

CS/EE 602 DIGITAL SIGNAL PROCESSING
(2006 Scheme)
Maximum Marks : 100

Time : 3 Hours
PART A
(Answer ALL questions)

(8 x 5 = 40)
I.

(a)

Show that for a discrete LTI system with zero initial conditions, the o/p is

y [n] = x[n]* h[n],x[n] and h [n] the i/p and impulse response respectively.
Find the stability of the system given by the impulse response
h

[n] =

0.5"

u [n] +

1.5"

u[n] . Plot its ROC.

Show that for a periodic sequence x p [n] with DFS X,, [k], DFS of
Xp * [ n]

X p * [-k] , and DFS of x p * [n] = X p * [k] where * represents complex

conjugation.
Find the circular convolution of the two sequences x1 En1
x2

[n] = [1

[1 2 1]

and

4 2], so that it gives the same result as linear convolution. Verify that both

convolutions give the same result.

Write a short note on the rectangular window method for FIR filter design. What is its
disadvantage?
What is the impulse invariant method for designing HR filters? Using this transformation,
transform the following filter into its discrete equivalent H (s) =
(g)

1
(s +1)(s 1)

If the quantization step size is IV and the number of bits used for quantization m = 3,
find the RMS quantization error. How will it be reduced if m is increased to 8?
Calculate the percentage reduction in the RMS quantization error in the second case.
Enumerate some features that distinguish DSP processors from general purpose processors.
PART B
(4 x 15 = 60)

(a)

For the impulse response below, determine if the systems are causal and stable.

h[n] =

2" u[n]

h[n] = Sin(-7)

(b)

h[n] .

[n] + Sin irn

h[n] =

u[n 1].

(8)

For the systems given by

( i)

y[r]= Co s (x[n])
y[n]=Ax[n]+ B,

check whether the systems are linear.

OR

(7)
(Turn Over)

2
Find the inverse -Z- transform of X (-Z-) =
Determine the causal signal x [n] having the

2-E- 4 +2-z-- ' +1.

transform

X(*) -
(1

)(1--z--1)

Show how the 8 point Radix - 2 Signal flow graph may be used to compute the
DFTs of xi [n] = [

1 4 3 2 2 3 2 2 and x2 [n] = [ 1 4 1 2 2 3 4 2 ],

simultaneously, using the signal flow graph only once. Hence , compute X1 [k]

(a)

(b)
(a)

(b)

and X2 [k] .

(15)

OR
What is the improvement in speed in terms of number of complex additions and
multiplications in calculating the 64 point DFT of a sequence using Direct Computation
and FFT algorithm?
Derive the 8 point D.I.F FFT Signal Flow graph.

(5)
(10)

Design a Low Pass FIR filter for the following specifications :

Cut off frequency
=
500 Hz
21cHz
Sampling frequency
=
.-_
Order of the filter
10, and use the rectangular window for
truncating the impulse response.
For the given FIR filter with

H ()= 1+ 2-z- -1 +3 -z72

(a)

+ 2 -2 +

(8)

, implement the filter using

least number of delay elements. Will the filter have linear phase?
OR
Consider the recursive relation given by

(7)

y[n] = 3y [n 1]+ y [n 2] + 5x [n] + 2x [n 1] . Obtain its

(b)

Direct Form - 1
Direct Form - 2 implementations.
Obtain the linear constant coefficient difference equation of the discrete filter obtained
by discretising the analog filter

(a)

(b)

H (s) =

s+1
(s +3)(s + 5).

Write notes on :
Limit cycle oscillations in a First Order FIR System. Hence calculate
the dead band amplitude if the quantization step size is 0.5 volt and
3 bits are used for binary encoding the quantization levels.
Describe using a suitable diagram an application of a DSP to image processing.
OR
With a neat block diagram explain the architecture of the Texas Instruments TMS
320C - 54 X fixed point processor.
14*

(6)

(9)

(8)
(7)
(15)

BTS(C) VI 10 006 J

B.

Tech Degree VI Semester Examination, April 2010

CS/EE 602 DIGITAL SIGNAL PROCESSING
(2006 Scheme)
Maximum Marks : 100

Time : 3 Hours
PART A
(Answer ALL questions)

(8 x 5 = 40)
I.

(a)

Show that for a discrete LTI system with zero initial conditions, the o/p is

y [n] = x[n]* h[n],x[n] and h [n] the i/p and impulse response respectively.
Find the stability of the system given by the impulse response
h

[n] =

0.5"

u [n] +

1.5"

u[n] . Plot its ROC.

Show that for a periodic sequence x p [n] with DFS X,, [k], DFS of
Xp * [ n]

X p * [-k] , and DFS of x p * [n] = X p * [k] where * represents complex

conjugation.
Find the circular convolution of the two sequences x1 En1
x2

[n] = [1

[1 2 1]

and

4 2], so that it gives the same result as linear convolution. Verify that both

convolutions give the same result.

Write a short note on the rectangular window method for FIR filter design. What is its
disadvantage?
What is the impulse invariant method for designing HR filters? Using this transformation,
transform the following filter into its discrete equivalent H (s) =
(g)

1
(s +1)(s 1)

If the quantization step size is IV and the number of bits used for quantization m = 3,
find the RMS quantization error. How will it be reduced if m is increased to 8?
Calculate the percentage reduction in the RMS quantization error in the second case.
Enumerate some features that distinguish DSP processors from general purpose processors.
PART B
(4 x 15 = 60)

(a)

For the impulse response below, determine if the systems are causal and stable.

h[n] =

2" u[n]

h[n] = Sin(-7)

(b)

h[n] .

[n] + Sin irn

h[n] =

u[n 1].

(8)

For the systems given by

( i)

y[r]= Co s (x[n])
y[n]=Ax[n]+ B,

check whether the systems are linear.

OR

(7)
(Turn Over)

2
Find the inverse -Z- transform of X (-Z-) =
Determine the causal signal x [n] having the

2-E- 4 +2-z-- ' +1.

transform

X(*) -
(1

)(1--z--1)

Show how the 8 point Radix - 2 Signal flow graph may be used to compute the
DFTs of xi [n] = [

1 4 3 2 2 3 2 2 and x2 [n] = [ 1 4 1 2 2 3 4 2 ],

simultaneously, using the signal flow graph only once. Hence , compute X1 [k]

(a)

(b)
(a)

(b)

and X2 [k] .

(15)

OR
What is the improvement in speed in terms of number of complex additions and
multiplications in calculating the 64 point DFT of a sequence using Direct Computation
and FFT algorithm?
Derive the 8 point D.I.F FFT Signal Flow graph.

(5)
(10)

Design a Low Pass FIR filter for the following specifications :

Cut off frequency
=
500 Hz
21cHz
Sampling frequency
=
.-_
Order of the filter
10, and use the rectangular window for
truncating the impulse response.
For the given FIR filter with

H ()= 1+ 2-z- -1 +3 -z72

(a)

+ 2 -2 +

(8)

, implement the filter using

least number of delay elements. Will the filter have linear phase?
OR
Consider the recursive relation given by

(7)

y[n] = 3y [n 1]+ y [n 2] + 5x [n] + 2x [n 1] . Obtain its

(b)

Direct Form - 1
Direct Form - 2 implementations.
Obtain the linear constant coefficient difference equation of the discrete filter obtained
by discretising the analog filter

(a)

(b)

H (s) =

s+1
(s +3)(s + 5).

Write notes on :
Limit cycle oscillations in a First Order FIR System. Hence calculate
the dead band amplitude if the quantization step size is 0.5 volt and
3 bits are used for binary encoding the quantization levels.
Describe using a suitable diagram an application of a DSP to image processing.
OR
With a neat block diagram explain the architecture of the Texas Instruments TMS
320C - 54 X fixed point processor.
14*

(6)

(9)

(8)
(7)
(15)