c 3189

B.E./B.Tech. DE GREE EXAN{INATION, MAY/JUNE 2007.

Fifth Semester

(Regulation 2004)

Electronics and Communication Engineering

EC 1302- DIGITAL SIGNAL PROCESSiNG

(Commonto B.E. (Part-Time)Fourth Semester- Regulation2005)

Time : Three hours Maximum: 100 marks

m
Answer ALL questions.

co
PARTA-(10 x2=20 marks)
N.
1. The first frve DFT coefficientsof a sequencex[n] are X(0) = 20, Xft) = 5 + i2,
va
x(2)=0, X ( 3 ) = 0 . 2 +j 0 . 4 , X ( 4 ) = 0 . Determine the remaining DFT
na

coefficients.
aa

2. What are the advantagesof FFT algorithm over direct computation of DFT?
M
w.

3. Show that the filter with h[nl = i - 1, 0, 1] is a linear phasefilter.

ww

4. Find the digrtal transfer function H Q) by using irnpulse invariant method for

the analog transfer function I/ (s) = Assume T = 0.5 sec.

*

5. Identify the various factors which degrade the performance of the digrtal filter
implementation when finite word length is used.

6. What is meant by iimit cycle osciilation in digital filters?

7. Define power spectrai density and crosspower spectraldensity.

8. What are the disadvantages of non-parametric methods of power spectral
estimation?

9. Differentiate between Von Neumann and Harvard architectures.

10. State the merit and demerit of multiported mernories.

PARTB-(Sx16=80marks)

11. (a) (i) Prove the following properties of DFT when H [kl is the DFT of an
N-point sequenceh [n] .

(1) H lltl is real and even when h[nl is real and even.

(2) H Vzlis imaginary and odd when hlnl is real and odd. (8)

m
(ii) Computethe DFT of x[nl = s-05n,0 < n < 5 . (8)

co
Or N.
(b) (i) From first principles obtain the signal flow graph for computing
va
8-point DFT using radix-2 decimation-in-frequencyFFT algorithm.
(8)
na

(ii) Using the

aa

above signal flow graph compute DFT of

x [ n ] = c o s ( nn / 4 ) ' 0 < n < 7 ' (8)
M

12. (a) A bandpass FIR fiiter of length 7 is required. It is to have iower and
w.

upper cut-off frequenciesof 3 kHz and 6kHz respectively and is intended

ww

to be used with a sampling frequency of 24 kHz. Determine the filter

coefficientsusing Hanning window. Consider the filter to be causal. (16)

Or

(b) Determine H (z) for a Butterworth filter satisfying the following

specifications:

0 . 8< l , H( e l u) l < 1 , 0 < r o< z / 4

l H ( n t ' ) l < o ' 2 ,n / 2 < a < n

Assume T = 0.1 sec.Apply bilinear transformation method.

c 3189
13. (a) (i) Consider the truncation of negative fraction numbers represented
in (f +1)-bit fixed point binary form including sign bit. Let (P*b)
bits be truncated. Obtain the range of truncation errors for signed
magnitude, 2's compiementand 1's complementrepresentationsof
the negativenumbers. (8)

(ii ) The coefficientsofa system definedby

H(z)= -t are representedin a number with a

1 - 0 . 4z ) ( t - 0 . 5 5z - t )
sign bit and 3 data bits. Determine the new pole locations for
(1) direct realizatton and (2) cascade realization of first order
systems.Compare the movementsof the new poles away frorn the
original onesin both the cases. (8)

Or

m
(b) (i) Consider (b+l)-bit (including sign bit) bipolar A/D converter.

co
Obtain an expressionfor signal to quantization noise ratio. State
the assumptionsmade. (8)

(ii )
N.
A causal IIR filter is defined by the difference equation
y [ n ] = x l n l - 0 . 9 y [ n - 1 ] . T h e u n i t s a m p l er e s p o n s eh l n ) i s c o m p u t e d
va

such that the computed values are rounded to one decimal place.
na

Show that the filter exhibits dead band effect. Determine the dead
band range. (8)
aa

L4. (a) (i) Compute the autocorrelation and power spectral density for the
M

signai X(t)=Acos(2zfot+O) whenA and fo are constants.4p is a

w.

random variable which is uniformly distributed over the interval

(-n, n) . (8)
ww

(ii) Expiain briefly the periodogram method of power spectral

estimation. (8)

Or

(b) (i) Describe briefly Bartlett, Welch and Blackman-Tukey methods of

power spectral estimation. (10)

(ii) Determine the frequency resolution of the Bartlett, Welch and

Blackmann-Tukey methods of power spectral estimation for a
quality factor Q = 20. Assume that overlap in Welch's method is
50Voand the length of the sample is 2000. (6)

c 3189
15. (a) (i) Expiain what is meant by instruction pipelining. Explain with an
example, how pipelining increasesthe through put efficiency. (8)

(ii) Explain the operationof TllM serial ports in P-DSPs. (8)

Or

(b) (i) With a suitable diagram describe the functions of multiplier/adder

unit of TMS 320C54X. (8)

(ii) Explain the operationof CSSU of TNIS320C54X and explain its use
consideringthe Viteri operator. (8)

m
co
N.
va
na
aa
M
w.
ww

4 c 3189