EC2302

DIGITAL SIGNAL PROCESSING

UNIT I DISCRETE FOURIER TRANSFORM 9 DFT and its properties, Relation between DTFT and DFT, FFT computations using Decimation in time and Decimation in frequency algorithms, Overlap-add and save methods. UNIT II INFINITE IMPULSE RESPONSE DIGITAL FILTERS 9 Review of design of analogue Butterworth and Chebyshev Filters, Frequency transformation in analogue domain Design of IIR digital filters using impulse invariance technique Design of digital filters using bilinear transform pre warping Realization using direct, cascade and parallel forms. UNIT III FINITE IMPULSE RESPONSE DIGITAL FILTERS 9 Symmetric and Antisymmetric FIR filters Linear phase FIR filters Design using Hamming, Hanning and Blackmann Windows Frequency sampling method Realization of FIR filters Transversal, Linear phase and Polyphase structures. UNIT IV FINITE WORD LENGTH EFFECTS 9 Fixed point and floating point number representations Comparison Truncation and Rounding errors - Quantization noise derivation for quantization noise power coefficient quantization error Product quantization error - Overflow error Roundoff noise power - limit cycle oscillations due to product roundoff and overflow errors - signal scaling. UNIT V MULTIRATE SIGNAL PROCESSING 9 Introduction to Multirate signal processing-Decimation-Interpolation-Polyphase implementation of FIR filters for interpolator and decimator -Multistage implementation of sampling rate conversion- Design of narrow band filters - Applications of Multirate signal processing. L = 45, T = 15, TOTAL : 60 TEXT BOOKS: 1. John G Proakis and Manolakis, Digital Signal Processing Principles, Algorithms and Applications, Pearson, Fourth Edition, 2007. 2. S.Salivahanan, A. Vallavaraj, C. Gnanapriya, Digital Signal Processing, TMH/McGraw Hill International, 2007 REFERENCES: 1. E.C. Ifeachor and B.W. Jervis, Digital signal processing A practical approach, Second edition, Pearson, 2002. 2. S.K. Mitra, Digital Signal Processing, A Computer Based approach, Tata McGraw Hill, 1998. 3. P.P.Vaidyanathan, Multirate Systems & Filter Banks, Prentice Hall, Englewood cliffs, NJ, 1993. 4. Johny R. Johnson, Introduction to Digital Signal Processing, PHI, 2006.

JEPPIAAR ENGINEERING COLLEGE

DEPARTMENT OF E.C.E QUESTION BANK EC 2302 DIGITAL SIGNAL PROCESSING
UNIT -1 DISCRETE FOURIER TRANSFORM (FFT) PART-A
1. Determine the DTFT of a sequence x(n) = an u(n). Nov/Dec 2006 Solution: x(n) = an u(n) j X(e ) = x(n) e -jn n=- j X(e ) = an u(n) e -jn n=- j X(e ) = an e -jn n=0 X(e ) = (a e j)n n=0
j

X(ej) = 1 / (1-a-ej ) 2. What is FFT? Nov/Dec 2006 The Fast Fourier Transform is a method or algorithm for computing the DFT with reduced number of calculations. The computational efficiency can be achieved if we adopt a divider and conquer approach. This approach is based on decomposition of an N-point DFT in to sucessively smaller DFTs. This approach leads to a family of an efficient computational algorithm is known as FFT algorithm.

3. The first five DFT coefficients of a sequence x(n) are X(0) = 20, X(1) = 5+j2,X(2) = 0,X(3) = 0.2+j0.4 , X(4) = 0 . Determine the remaining DFT coefficients. May/June 2007 Solution: X (K) = [20, 5+j2, 0, 0.2+j 0.4 , 0,X(5),X(6),X(7)] X (5) = 0.2 j0.4 X (6) = 0 X (7) = 5-j2 4. What are the advantages of FFT algorithm over direct computation of DFT? May/June 2007 1. Reduces the computation time required by DFT. 2. Complex multiplication required for direct computation is N2 and for FFT calculation is N/2 log 2 N. 3. Speed calculation. 5. State and prove Parsevals Theorem. Nov/Dec 2007 Parsevals theorem states that If x(n) X(K) and y(n) Y(K) , Then N-1 N-1 x(n) y*(n) = 1/N X(K) Y*(K) n=0 K =0 When y(n) = x(n), the above equation becomes N-1 N-1 2 x(n) = 1/N X(K)2 n=0 k=0 6. What do you mean by the term bit reversal as applied to FFT? Nov/Dec 2007 Re-ordering of input sequence is required in decimation in time. When represented in binary notation sequence index appears as reversed bit order of row number.

7. Define the properties of convolution. April/May 2008. 1. Commutative property: x(n)*h(n) = h(n) *x(n) 2. Associative Property: [x(n)*h1(n)] *h2(n) = x(n)*[h1(n)*h2(n)] 3. Distributive Property: x(n)*[ h1(n)+ h2(n)] = [x(n)* h1(n)]+ [x(n)* h1(n)] 8. Draw the basic butterfly diagram of radix -2 FFT. April/May 2008. 1 a 1 1 WN b -1 9. Distinguish between DIT and DIF FFT algorithm. Nov/Dec 2008 S.No DIT FFT Algorithm 1. The input is in bit reversed order; the output will be normal order. 2. Each stage of computation the phase factor are multiplied before add subtract operation. DIF FFT Algorithm The input is in normal order; the output will be bit reversed order. Each stage of computation the phase factor are multiplied after add subtract operation.
nk

1 A = a+ WNnk b

B = a - WNnk b

10.If H(K) is the N-point DFT of a sequence h(n) , Prove that H(K) and H(N-K) are comples conjugates. Nov/Dec 2008 This property states that, if h(n) is real , then H(N-K) = H*(K) = H(-K) Proof: By the definition of DFT; N-1 X(K) = x(n) e (j2nk)/N n=0 Replace K by N-K N-1 X(N-K) = x(n) e (j2n(N-K))/N X(N-K) n= X*(K) =

11.State Discrete Fourier Transform. The DFT is defined as N-1 X (K) = x(n) e (j2nk)/N ; K = 0 to N-1 n=0 The Inverse Discrete Fourier Transform (IDFT) is defined as N-1 x (n) = X(K) e (j2nk)/N ; n = 0 to N-1 K=0 12.Distinguish between linear & circular convolution. s.no Linear convolution 1 The length of the input sequence can be different. 2 Zero Padding is not required. circular convolution The length of the input sequence should be same. Zero padding is required if the length of the sequence is different.

13.Define Zero padding? Why it is needed? Appending zeros to the sequence in order to increase the size or length of the sequence is called zero padding.In circular convolution , when the two input sequence are of different size , then they are converted to equal size by zero padding. 14.State the shifting property of DFT. Time shifting property states that DFT {x(n-n0)} = X(K) e (j2n0k)/N 15.Why do we go for FFT? The FFT is needed to compute DFT with reduced number of calculations.The DFT is required for spectrum analysis on the sinals using digital computers. 16.What do you mean by radix-2 FFT? The radix -2 FFT is an efficient algorithm for coputing N- point DFT of an N-point sequence .In radix-2 FFT the n-point is decimated into 2-point sequence and the 2-point DFT for each decimated sequence is computed. From the results of 2-point DFTs, the 4-point DFTs are computed. From the results of 4 point DFTs ,the 8-point DFTs are computed and so on until we get N - point DFT.

17.Give any two application of DFT? 1. The DFT is used for spectral analysis of signals using a digital computer. 2. The DFT is used to perform filtering operations on signals using digital computer. 18.How many multiplications & addition are involved in radix-2 FFT? For performing radix-2 FFT, the value of Nshould be such that, N= 2m. The total numbers of complex additions are Nlog 2 N and the total number of complex multiplication are (N/2) log 2 N. 19.What is Twiddle factor? Twiddle factor is defined as WN = e j2/N. It is also called as weight factor. 20.What is main advantage of FFT? FFT reduces the computation time required to compute Discrete Fourier Transform.

PART-B
1. a) i) Calculate the DFT of the sequence x(n) = {1,1,-2,-2} ii) Determine the response of LTI system by radix -2 DIT FFT. Nov/Dec 2006 Ans:i) X(K) = { 0, -1-j,6,-1+j} ii) Ref Pg.No 320-328 , DSP by Salivahanan . 2. a) i) Derive the equation for Decimation in time algorithm for FFT. ii) How do you linear filtering by FFT using Save add method? Nov/Dec 2006 & April /May 2008 & Nov/Dec 2008 Ans:i) Ref Pg.No 320-328 , DSP by Salivahanan . ii) Ref Pg.No 369, DSP by Salivahanan. 3. a)i) Prove the following properties of DFT when H(k) is the DFT of an N-point sequence h(n). 1. H(k) is real and even when h(n) is real and even. 2. H(k) is imaginary and odd when h(n) is real and odd. ii) Compute the DFT of x(n) = e-0.5n , 0 n 5. May/June 2007 Ans: i) Ref Pg.No 309, DSP by Salivahanan. ii) X(K) = { 2.414, 0.87-j0.659, 0.627-0.394j, 1.202, 0.62-j0.252, 0.627-j0.252}.

4. a) i) From first principles obtain the signal flow graph for Computing 8-point using radix -2 DIF FFT algorithm. ii) Using the above signal flow graph compute DFT of x(n) = cos (n/4) ,0 n 7. May/June 2007 & Nov/Dec 2007 & Nov/Dec 2008 Ans: i) Ref Pg.No 334-340, DSP by Salivahanan. ii) X(K) = {0, 3, 0, 2.7-j0.7, 0, 1, 0, 1.293-j0.7} 5. a) Two finite duration sequence are given by x(n) = sin (n/2) for n = 0,1,2,3 h(n) = 2 n for n = 0,1,2,3 Determine circular convolution using DFT &IDFT method. Nov/Dec 2007 Ans: X(K) = {0, -2j, 0, 2j} H(K) = {15, -3+6j, -5, -3-6j} y(n) = {6, -3, -6, 3} 6. a) i) Discuss in detail the important properties of the DFT. ii) Find the 4-point DFT of the sequence x(n) = cos (n/4) iii) Compute an 8-point DFT using DIF FFT radix -2 algorithm. x(n) = { 1,2,3,4,4,3,2,1} April /May 2008 Ans: i)Ref Pg.No 308-311, DSP by Salivahanan. ii) X(K) = {1, 1-j1.414, 1, 1+j1.414} iii) X(K) = {20,-5.8-j2.4, 0, 0.17-j0.414, 0, -0.17+j0.414, 0, -5.82+j2.414}.

UNIT -2 INFINITE IMPULSE RESPONSE DIGITAL FILTERS PART-A

1. Give any two properties of Butterworth and Chebyshev filter.
Nov/Dec 2006

Properties of Butterworth: 1. The butterworth filters are all pole design. 2. The filter order N completely specifies the filter 3. The magnitude is maximally flat at the origin. 4. The magnitude is monotomically decreasing function of ohm. Properties of Chebyshev: 1. The magnitude reponse of the filter exhibits ripples in the pass band or stop band 2. The pole of the filter lies on an ellipse. 2. Find the digital transfer function H(Z) by using impulse invariant method for the analog transfer function H(S) = 1/ (S+2).Assume T=0.5sec May /June 2007 &Nov/Dec 2007 Solution: H(S) = 1/ (S+2). H(Z) = 1/[1-e-1 Z-1] H(Z) = 1/ [1-0.368Z-1] 3. What is the relationship between analog and digital frequency in impulse invariant transformation? April/May 2008 Digital Frequency: = T = analog frequency T= Sampling interval 4. What is Prewarping? Why is it needed? Nov/Dec 2008 In IIR design using bilinear transformation the conversion of specified digital frequencies to analog frequencies is called Pre-warping. The PreWarping is necessary to eliminate the effect of warping on amplitude response.

5. Compare FIR & IIR filter. S.No FIR filter IIR filter 1. Only N samples of All the infinite samples of impulse response are impulse response are considered. considered. 2. Linear phase Linear phase characteristics characteristics can be can not be achieved achieved 6. Define Frequency warping. The non linear relation ship between analog and digital frequencies introduced frequency distortion which is called as frequency warping.

PART-B
1. a) Design a digital Butterworth filter satisfying the constraints 0.707 | H()| 1.0 ; 0 /2 | H()| 0.2 ; 3/4 . Nov/Dec 2006 Ans: Ref Pg.No 435-437, DSP by Salivahanan. 2. a) Design a digital Butterworth filter satisfying the constraints 0.8 | H()| 1.0 ; 0 /4 | H()| 0.2 ; /2 . Apply Bilinear transformation method. May/June2007 & Nov/Dec 2008 Ans: Ref: Pg.No: 359-362, DSP by Nagoorkani. 3. a)i) Design a digital BUTTERWORTH filter that satisfies the following constraint using BILINEAR Transformation. Assume T = 1 sec. 0.9 | H()| 1 ; 0 /2 | H()| 0.2 ; (3 /4) ii) Determine the magnitude response of the FIR filter (M=11) and show that phase and group delay are conatant. iii) The desired frequency response of a low pass filter is given by j3 Hd() ={ e ; -3/4 3/4 0 ; other wise.

Determine H(ej) for M= 7using HAMMING window. iv) For the analog transfer function H(S) = 1/ (S+1)(S+2) . Determine H(Z) using impulse invariant technique. April /May 2008 Ans: a) i) Ref Pg.No 437-439, DSP by Salivahanan. ii) Ref Pg.No 383-384, DSP by Salivahanan. iii) Ref Pg.No 400-401, DSP by Salivahanan. iv) Ref Pg.No 426, DSP by Salivahanan.

UNIT -3 FINITE IMPULSE RESPONSE DIGITAL FILTERS PART-A

1. Obtain the block diagram representation of a FIR system?
Nov/Dec 2006 X(Z) Z-1 Z-1 h(1) h(0) h(2) h(N-2) Z-1 Z-1 h(N-1)

Y(Z) + + + +

2. Show that the filter with h(n) = [-1,0,1] is a linear phase filter. May /June 2007 & Nov/Dec 2008 Solution: h(n) = [ -1,0,1] h(0) = -1 = -h(N-1-n) = -h(3-1-0) = -h(2) h(1) = 0 = -h(N-1-n) = -h(3-1-1) = -h(1) h(2) = 1 = -h(N-1-n) = -h(3-1-2) = -h(0) It is a linear phase filter.

3. In the design of FIR digital filter, how is Kaiser Window different from other windows? Nov/Dec 2007 In all other windows a trade off exists between ripple ratio and main lobe width. In Kaiser Window both ripple ratio and main lobe width can be varied independently. 4. What are the merits and demerits of FIR filter? April/May 2008 Merits : 1. Linear phase filter. 2. Always Stable Demerits: 1. The duration of the impulse response should be large 2. Non integral delay. 5.What are the advantages of FIR filter? 1. They can have an exact linear phase. 2. They are always stable 3. They can be realised efficiently in hardware 4. The design methods are generally stable.

6.What is the necessary & sufficient condition of linear phase FIR filter? The condition for a linear phase filter is 1. = (N-1)/2 2. h(n) = h(N-1-n) 7. What is Gibbs phenomenon? In Fir filter design using Fourier analysis method for rectangular window method, the infinite duration impulse response is truncated to finite duration impulse response.The abrupt truncation of impulse response introduce a oscillation in the pass band and stop band .This effect is known as Gibbs phenomenon. 8. Compare Rectangular & Hamming window. S.No Rectangular Window 1. The width of the main lobe in window spectrum is 4/N 2. The maximun side lobe Hamming window. The width of the main lobe in window spectrum is 8/N The maximun side lobe

magnitude in window magnitude in window spectrum is -13 dB spectrum is -41 dB 9. Compare Hamming window & Kaiser Window. S.No Kaiser Window 1. The width of the main lobe in window spectrum depends on the value of and N. 2. The maximun side lobe magnitude with respect to peak of main lobe is variable using the parameter . 10.Compare FIR & IIR filter. S.No FIR filter IIR filter 1. Only N samples of All the infinite samples of impulse response are impulse response are considered. considered. 2. Linear phase Linear phase characteristics characteristics can be can not be achieved achieved Hamming window. The width of the main lobe in window spectrum is 8/N

The maximun side lobe magnitude in window spectrum is -41 dB

11.Compare Rectangular Window& Hanning Window. S.No Rectangular Window 1. The width of the main lobe in window spectrum is 4/N 2. The maximun side lobe magnitude in window spectrum is -13 dB Hanning Window The width of the main lobe in window spectrum is 8/N The maximun side lobe magnitude in window spectrum is -31 dB

12.Compare Hamming Window& Hanning Window. S.No Hamming window. 1. The width of the main lobe in window spectrum is 8/N 2. The maximun side lobe magnitude in window spectrum is -41 dB Hanning Window The width of the main lobe in window spectrum is 8/N The maximun side lobe magnitude in window spectrum is -31 dB

13.Compare Hamming Window& BlackmanWindow. S.No Hamming window. 1. The width of the main lobe in window spectrum is 8/N 2. The maximun side lobe magnitude in window spectrum is -41 dB BlackmanWindow The width of the main lobe in window spectrum is 12/N The maximun side lobe magnitude in window spectrum is -58 dB

PART-B 1.a) Design a high pass filter hamming window by taking 9 samples
of w(n) and with a cutoff frequency of 1.2 radians/sec. Nov/Dec 2006 Ans: Ref: Pg.No: 298-301, DSP by Nagoorkani. 2.a) Design a digital Butterworth filter satisfying the constraints 0.707 | H()| 1.0 ; 0 /2 | H()| 0.2 ; 3/4 . Nov/Dec 2006 Ans: Ref Pg.No 435-437, DSP by Salivahanan. 3.a) Design a digital Butterworth filter satisfying the constraints 0.8 | H()| 1.0 ; 0 /4 | H()| 0.2 ; /2 . Apply Bilinear transformation method.

May/June2007 & Nov/Dec 2008 Ans: Ref: Pg.No: 359-362, DSP by Nagoorkani. 4. a) Describe the design of FIR filter using frequency sampling technique. b) The desired frequency response of a low pass filter is given by j2 Hd() ={ e ; -/4 /4 0 ; other wise. Obtain the filter coefficient, h(n) using RECTANGUAR window define by W(n) = { 1; 0 n 4 0; otherwise. Nov/Dec 2007 Ans: a) Ref Pg.No 389-391, DSP by Salivahanan. b) Ref Pg.No 399, DSP by Salivahanan.

5.

a) Design a bandpass filter to pass frequencies in the range

1to2radians/sec using hanning window,with N=5. Nov/Dec 2006 Ans: Ref: Pg.No: 301, DSP by Nagoorkani.

UNIT -4 FINITE WORD LENGTH EFFECTS PART-A

1. Express the fraction 7/8 and 7/8 in sign magnitude, 2s complement and 1s complement. Nov/Dec 2006 Solution: 7/8 = 0.875 = (0.111)2 is sign magnitude 1s Complement = (0.111)2 2s Complement = (0.111)2 - 7/8 = -0.875 Sign magnitude: (1.111)2 1s Complement = (1.000)2 2s Complement = (1.001)2 2. a) What are the quantization error due to finite word length register in digital filter. b) What are the different quantization methods? Nov/Dec 2006 Quantization Error : 1. Input quantization error

2. Coefficient quantization error 3. Product quantization error Quantization methods 1. Truncation 2. Rounding 3. Identify the various factors which degrade the performance of the digital filter implementation when finite word length is used. May /June 2007 & April/May 2008 & Nov/Dec 2008 1. Input quantization error 2. Coefficient quantization error 3. Product quantization error 4. What is meant by limit cycle oscillation in digital filter? May /June 2007 & Nov/Dec 2007 &April/May 2008 In recursive system when the input is zero or same non-zero constant value the non linearities due to finite precision arithmetic operation may cause periodic oscillation in theoutput. Thus the oscillation is called as Limit cycle. 5. Express the fraction (-7/32) in signed magnitude and 2s complement notations using 6 bits. Nov/Dec 2007 &Nov/Dec 2008 In Signed Magnitude: 1.001110 In 2s complement: 1.110010 6. Compare fixed & floating point number representation. S.no 1. 2. Fixed point number The position of the binary Point is fixed. The resolution is uniform throughout the range Floating point number The position of the binary Point is variable. The resolution is variable.

7.

What are the two types of quantization employed in digital system? 1. Rounding 2. Truncation Define Rounding & truncation. Truncation is the process of discarding all bits less significant than least significant bit that is retained. Rounding of a b bit is accomplished by choosing the rounded result as the b bit number closed to the original number unrounded.

8.

9.

What is dead band? In the limit cycle the amplitude of the output are confined to a range of value which is called as dead band of the filter.

PART-B
1. For the given transfer function H(Z) = H1(Z) .H2(Z) ,where 1 1 H1(Z) = and H2(Z) = -1 1 0.5 Z 1 0.6 Z-1 Find the output round off noise power. Nov/Dec 2006 -2b Ans: 2 /12(5.4315) 2. a)i) Explain the characteristics of a limit cycle oscillation w.r.t the system described by the difference equation y(n) = 0.95y(n-1)+x(n).Determine the dead band of the filter. ii) Draw the product quantisation noise model of second order IIR filter. Nov/Dec 2006 & Nov/Dec 2008 Ans: a) i) Dead band = [-10,10] ii) Ref Pg.No 513-514, DSP by Salivahanan. 3. a)i) Consider the truncation of negative fraction number represented in(+1) bit fixed point binary form including sign bit . Let (-b) bits be truncated .Obtain the range of truncation errors for signed magnitude ,2s complement and 1s complement representation of negative numbers. Nov/Dec 2007 ii) The coefficients of a system defined by 1 H(Z) = (1-0.4Z-1)(1-0.55Z-1) are represented in anumber with a sign bit and 3 data bits. Determine the new pole location for 1) Direct realization and 2) Cascade realization of first order systems.Compare the movements of the new pole away from the original ones in both the cases.

iii) Consider the (b+1) bit bipolar A/D converter.Obtain an expression for signal to quantization noise ratio .
May /June 2007& Nov/Dec 2007&April/May2008 & Nov/Dec 2008

Ans: a) i) Ref Pg.No 496-499, DSP by Salivahanan. ii) Direct form: 1/ [1-0.875z-1+0.125Z-2] Cascade form:1/[1-0.375Z-1][1-0.5Z-1] iii) Ref Pg.No 499-503, DSP by Salivahanan.

UNIT -5 MULTIRATE SIGNAL PROCESSING PART-A

1. Define multirate signal processing. The process of converting a signal from a given rate to a different rate is called sampling rate conversion. The system that employs multiple sampling rates in the processing of digital signals are called multirate digital signal processing systems. 2. What are the advantages of multirate digital signal processing. Computational requirements are less Storage for filter coefficients is less Finite arithmetic effects are less Filter order required in multirate application are low and Sensitivity to filter coefficients lengths are less

3. Give the applications of multirate digital signal processing. Communication systems Speech and audio processing systems Antenna systems Radar systems 4. Define Decimation The process of reducing the sampling rate of the signal is called decimation (sampling rate compression). 5. Define Interpolation The process of increasing the sampling rate of the signal is called interpolation (sampling rate Expansion).

6. Define signal flow graph. A signal flow graph is a graphical representation of the relationships between the variables of a set of linear difference equations.

PART-B
1. Explain Multi rate Digital signal processing Ref Pg.No.523, DSP by Salivahanan 2. Explain the Decimation process with an example. Ref Pg.No.526, DSP by Salivahanan 3. Explain the Interpolation process with an example. Ref Pg.No.531, DSP by Salivahanan 4. Explain the polyphase decomposition for FIR filter. Ref Pg.No.541, DSP by Salivahanan 5. Explain Multistage Decimators and interpolators. Ref Pg.No.555, DSP by Salivahanan

B.E/B.TECH DEGREE EXAMINATION, NOV/DEC 2006

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Questions 21.Determine the DTFT of a sequence x(n) = an u(n). Solution: x(n) = an u(n) j X(e ) = x(n) e -jn n=- j X(e ) = an u(n) e -jn n=- j X(e ) = an e -jn n=0

 X(e ) = (a e j)n n=0

j

X(ej) = 1 / (1-a-ej ) 22.What is FFT? The Fast Fourier Transform is a method or algorithm for computing the DFT with reduced number of calculations. The computational efficiency can be achieved if we adopt a divider and conquer approach. This approach is based on decomposition of an N-point DFT in to successively smaller DFTs. This approach leads to a family of an efficient computational algorithm is known as FFT algorithm. 23.Obtain the block diagram representation of a FIR system?
X(Z) Z-1 Z-1 h(1) h(0) h(2) h(N-2) Z-1 Z-1 h(N-1)

Y(Z) + + + +

24.Give any two properties of Butterworth and Chebyshev filter. Properties of Butterworth: 5. The butterworth filters are all pole design. 6. The filter order N completely specifies the filter 7. The magnitude is maximally flat at the origin. 8. The magnitude is monotomically decreasing function of ohm. Properties of Chebyshev: 3. The magnitude response of the filter exhibits ripples in the pass band or stop band 4. The pole of the filter lies on an ellipse. 9. Express the fraction 7/8 and 7/8 in sign magnitude, 2s complement and 1s complement. Solution: 7/8 = 0.875 = (0.111)2 is sign magnitude 1s Complement = (0.111)2

2s Complement = (0.111)2 - 7/8 = -0.875 Sign magnitude: (1.111)2 1s Complement = (1.000)2 2s Complement = (1.001)2 10.a) What are the quantization errors due to finite word length register in digital filter. b) What are the different quantization methods?

Quantization Error : 4. Input quantization error 5. Coefficient quantization error 6. Product quantization error Quantization methods 4. Truncation 5. Rounding 11.What is zero padding? Does Zero padding improve the frequency resolution in the spectral estimate? Let the sequence x(n) has a length L.If we want to find the N-point DFT (N>L) of the sequence x(n) we have to add (N-L) zeros to the sequence x(n).This is known as Zero padding. Yes , it will improve the frequency resolution in the spectral estimate. 12.Explain Deterministic and random signal with examples. Deterministic signals are signals that are completely specified in time. Deterministic signals can be predicted. Ex: Sin n Non deterministic signals are defined as the signal that takes on random value at any given instant and it cannot be predicted. It is also called as Random signal. Ex: ECG signal. 13.Give the digital signal processing application with the TMS 320 family. Communication system like voice coder, Speech recognization, Audio signal processing, Control and data acquisition, Biometric information processing and image /video processing. 14.What is the advantage of Harvard architecture of TMS 320 series? Harvard architecture has two or three memory buses allowing access to filter coefficients and input signals in the same cycle. Since it possesses two independent bus system. The Harvard architecture is

capable of Simultaneous reading an instruction code and reading or writing a memory or peripheral as part of the execution of the previous instruction.

PART B (5*16 =80MARKS)

11. a) i) Calculate the DFT of the sequence x(n) = {1,1,-2,-2} ii) Determine the response of LTI system by radix -2 DIT FFT. Ans:i) X(K) = { 0, -1-j,6,-1+j} ii) Ref Pg.No 320-328 , DSP by Salivahanan . (OR) b) i) Derive the equation for Decimation in time algorithm for FFT. ii) How do you linear filtering by FFT using Save add method? Nov/Dec 2006 & April /May 2008 & Nov/Dec 2008 Ans:i) Ref Pg.No 320-328 , DSP by Salivahanan . ii) Ref Pg.No 369, DSP by Salivahanan. a.

a) Design a high pass filter hamming window by taking 9 samples

of w(n) and with a cutoff frequency of 1.2 radians/sec. Ans: Ref: Pg.No: 298-301, DSP by Nagoorkani. (OR) b) Design a digital Butterworth filter satisfying the constraints 0.707 | H()| 1.0 ; 0 /2 | H()| 0.2 ; 3/4 . Ans: Ref Pg.No 435-437, DSP by Salivahanan.

b. a) For the given transfer function H(Z) = H1(Z) .H2(Z) ,where

1 1 0.5 Z Find the output round off noise power. Ans: 2-2b/12(5.4315) (OR) H1(Z) =
-1

1 and H2(Z) = 1 0.6 Z-1

b) i) Explain the characteristics of a limit cycle oscillation w.r.t the

c.

system described by the difference equation y(n) = 0.95y(n-1)+x(n).Determine the dead band of the filter. ii) Draw the product quantisation noise model of second order IIR filter. Ans: a) i) Dead band = [-10,10] ii) Ref Pg.No 513-514, DSP by Salivahanan a) Determine the performance characteristics of Non parametric Power spectrum estimator Welch, Bartlett and Blackman and Tukey. Ans: a) Ref Pg.No 594-603, DSP by Salivahanan. (OR) b) i) Give the key features of the digital signal processor ii) Write Short notes on: a) 32- bit accumulator b) 16*16 bit parallel multiplier c) Shifter Ans: a) i) Ref Pg.No 8-9, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 40, 59,266, Digital Signal Processor by B.Venkataramani &M.Bhaskar

15. a) i) Explain the function of auxillary register in the indirect Addressing mode to point the data memory location. ii) Write a program to use the auxillary register in memory pointing and looping. Ans: Ref Pg.No 48-49, Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) b) i) Write a program to compute the following equation . Y= A*X1 + B* X2 +C*X3 ii) Write a program to perform addition of two 64 bit numbers Ans: Ref Pg.No 265, Digital Signal Processor by B.Venkataramani &M.Bhaskar

B.E/B.TECH DEGREE EXAMINATION, MAY/JUNE 2007

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Question 1. The first five DFT coefficients of a sequence x(n) are X(0) = 20, X(1) = 5+j2,X(2) = 0,X(3) = 0.2+j0.4 , X(4) = 0 . Determine the remaining DFT coefficients Solution: X (K) = [20, 5+j2, 0, 0.2+j 0.4 , 0,X(5),X(6),X(7)] X (5) = 0.2 j0.4 X (6) = 0 X (7) = 5-j2 2. What are the advantages of FFT algorithm over direct computation of DFT? 1. Reduces the computation time required by DFT. 2. Complex multiplication required for direct computation is N2 and for FFT calculation is N/2 log 2 N. 3. Speed calculation. 3. Show that the filter with h(n) = [-1,0,1] is a linear phase filter. Solution: h(n) = [ -1,0,1] h(0) = -1 = -h(N-1-n) = -h(3-1-0) = -h(2) h(1) = 0 = -h(N-1-n) = -h(3-1-1) = -h(1) h(2) = 1 = -h(N-1-n) = -h(3-1-2) = -h(0) It is a linear phase filter. 4. Find the digital transfer function H(Z) by using impulse invariant method for the analog transfer function H(S) = 1/ (S+2).Assume T=0.5sec

Solution: H(S) = 1/ (S+2). H(Z) = 1/[1-e-1 Z-1]

H(Z) = 1/ [1-0.368Z-1] 5. Identify the various factors which degrade the performance of the digital filter implementation when finite word length is used. 10.Input quantization error 11.Coefficient quantization error 12.Product quantization error 6. What is meant by limit cycle oscillation in digital filter? In recursive system when the input is zero or same non-zero constant value the non linearities due to finite precision arithmetic operation may cause periodic oscillation in the output. Thus the oscillation is called as Limit cycle. 7. Define Power spectral density and cross power spectral density? The spectral characteristics of a random process is obtained according to wiener khinchine theorem, by computing the Fourier transform of the autocorrelation function. The power spectral density is given by xx (F) = xx () e j2ft d - The power density spectrum for two jointly stationary random process X(t) and Y(t) gives the cross correlation function xy () .The Fourier transform of xy (F) is xy (F) = xx () e j2ft d. - 8. What are the disadvantages of non-parametric methods of power spectral estimation? 1. It requires long data sequence to obtain the necessary frequency resolution. 2. Spectral leakage effects because of windowing. 9. Differentiate between Von Neumann and Harvard architecture. Von Neumann architecture possesses one bus system. The same bus carries all the information exchanged between the CPU and the peripherals including the instruction codes as well as data processed by the CPU. Harvard architecture possesses two independent bus systems. It is capable

of simultaneous reading an instruction code and reading or writing a memory or peripheral as part of the execution of precious instruction. 10.State the merits and demerits of multi ported memories. A Mutiported memory operates as on chip memory and off chip memory. 1. Higher performance because no wait states are required. 2. Lower cost than external memory. 3. Lower Power than external memory. 4. Higher performance because better flow with in the pipeline at lential arithmetic logic unit. 5. Ability to access large memory space.

PART B (5*16 =80MARKS)

11.a) i) Prove the following properties of DFT when H(k) is the DFT of an N-point sequence h(n). 3. H(k) is real and even when h(n) is real and even. 4. H(k) is imaginary and odd when h(n) is real and odd. ii) Compute the DFT of x(n) = e-0.5n , 0 n 5. Ans: i) Ref Pg.No 309, DSP by Salivahanan. ii) X(K) = { 2.414, 0.87-j0.659, 0.627-0.394j, 1.202, 0.62-j0.252, 0.627-j0.252}. (OR) b) i) From first principles obtain the signal flow graph for Computing 8-point using radix -2 DIF FFT algorithm. ii) Using the above signal flow graph compute DFT of x(n) = cos (n/4) ,0 n 7. Ans: i) Ref Pg.No 334-340, DSP by Salivahanan. ii) X(K) = {0, 3, 0, 2.7-j0.7, 0, 1, 0, 1.293-j0.7} 12.a) Design a digital Butterworth filter satisfying the constraints 0.8 | H()| 1.0 ; 0 /4 | H()| 0.2 ; /2 . Ans: Ref: Pg.No: 359-362, DSP by Nagoorkani. (OR) b) A band pass FIR filter of length 7 is required. It is to have lower and upper cutt frequencies of 3 KHZ and 6 Khz respectively

and indended to be used with a sampling frequency of 24 KHZ Determine the filter coefficient using HANNING window Consider the filter to be causal. (16)

13.a) i) Consider the truncation of negative fraction number represented in(+1) bit fixed point binary form including sign bit . Let (-b) bits be truncated .Obtain the range of truncation errors for signed magnitude ,2s complement and 1s complement representation of negative numbers. ii) The coefficients of a system defined by 1 H(Z) = (1-0.4Z-1)(1-0.55Z-1) are represented in anumber with a sign bit and 3 data bits. Determine the new pole location for 1) Direct realization and 2) Cascade realization of first order systems. Compare the movements of the new pole away from the original ones in both the cases. (OR)

b) Consider the (b+1) bit bipolar A/D converter.Obtain an expression for signal to quantization noise ratio . Ans: a) i) Ref Pg.No 496-499, DSP by Salivahanan. ii) Direct form: 1/ [1-0.875z-1+0.125Z-2] Cascade form:1/[1-0.375Z-1][1-0.5Z-1] b) Ref Pg.No 499-503, DSP by Salivahanan. 14.a)i) Determine the performance characteristics of Non parametric Power spectrum estimator Welch, Bartlett and Blackman and Tukey. ii) Determine the frequency resolution of the Bartlett, Welch & Blackman Tukey methods of power spectrum estimates for a Quality factor Q = 20. Assume that overlap in Welchs method is 50% and the length of the sample sequences is 2000. Ans: a) i) Ref Pg.No 594-603, DSP by Salivahanan. ii)Ref Pg.No 606, DSP by Salivahanan.

(OR) b) i) Compute the auto correlation & power spectral density for the signal x(t) = K cos (2ft + ) . ii) Explain briefly the periodogram method of power spectral Estimation. Ans: b) i) Ref Pg.No 590-591, DSP by Salivahanan. ii) Ref Pg.No 589, DSP by Salivahanan. 15.a) i) Explain what is meant by instruction pipelining. Explain with an example ,how pipelining increase the through put efficiency. ii) Explain the operation of TDM serial port in P-DSPs. Ans: a) i) Ref Pg.No 45-46, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 50-51, Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) b) i) With a suitable diagram describe the function of Multiplier/adder units of TMS 320 C54X. ii) Explain the operation of CSSU of TMS 320 C54X and explain its use considering the viteri operator. Ans: b) i) Ref Pg.No 267-268, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 269-270, Digital Signal Processor by B.Venkataramani &M.Bhaskar

B.E/B.TECH DEGREE EXAMINATION, NOV/DEC 2007

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Question 1. State and prove Parsevals Theorem. Parsevals theorem states that If x(n) X(K) and y(n) Y(K) , Then N-1 N-1 x(n) y*(n) = 1/N X(K) Y*(K) n=0 K =0 When y(n) = x(n), the above equation becomes N-1 N-1 2 x(n) = 1/N X(K)2 n=0 k=0 2. What do you mean by the term bit reversal as applied to FFT? Re-ordering of input sequence is required in decimation in time. When represented in binary notation sequence index appears as reversed bit order of row number. 3. Find the digital transfer function H(Z) by using impulse invariant method for the analog transfer function H(S) = 1/ (S+2).Assume T=0.5sec Solution: H(S) = 1/ (S+2). H(Z) = 1/[1-e-1 Z-1] H(Z) = 1/ [1-0.368Z-1]

4. In the design of FIR digital filter, how is Kaiser Window different from other windows?

In all other windows a trade off exists between ripple ratio and main lobe width. In Kaiser Window both ripple ratio and main lobe width can be varied independently. 5. What is meant by limit cycle oscillation in digital filter? In recursive system when the input is zero or same non-zero constant value the non linearities due to finite precision arithmetic operation may cause periodic oscillation in the output. Thus the oscillation is called as Limit cycle. 6. Express the fraction (-7/32) in signed magnitude and 2s complement notations using 6 bits. In Signed Magnitude: 1.001110 In 2s complement: 1.110010 7. Define the terms: autocorrelation sequence and power spectral density. Autocorrelation sequence: Rxx (n,n+m) = E[X(n) X(n+m)] Power spectral density: j Sxx (e ) = Rxx (m) e-jm m=- 8. Define unbiased estimate and consistent estimate. Expected value of the estimate is equal to actual value is called as unbiased estimate. Consistent estimate: Lim Varience (estimate) = 0. N 9. What are the factors that may be considered when selecting a DSP processor for an application? 1. Architectural features 2. Exection speed 3. Type of arithmetic 4. Word length 10.What is pipelining? A task is broken down in to a number of distinct subtasks which are then overlapped during execution. It is used in digital signal processors to increase speed.

PART B (5*16 =80 MARKS) 11.a) Two finite duration sequence are given by x(n) = sin (n/2) for n = 0,1,2,3 h(n) = 2 n for n = 0,1,2,3 Determine circular convolution using DFT &IDFT method. Ans: X(K) = {0, -2j, 0, 2j} H(K) = {15, -3+6j, -5, -3-6j} y(n) = {6, -3, -6, 3} (OR) i) From first principles obtain the signal flow graph for Computing 8-point using radix -2 DIF FFT algorithm. ii) Using the above signal flow graph compute DFT of x(n) = cos (n/4) ,0 n 7. Ans: i) Ref Pg.No 334-340, DSP by Salivahanan. ii) X(K) = {0, 3, 0, 2.7-j0.7, 0, 1, 0, 1.293-j0.7}

b)

12.a)

i) Describe the design of FIR filter using frequency sampling technique. ii) The desired frequency response of a low pass filter is given by j2 Hd() ={ e ; -/4 /4 1 ; other wise. Obtain the filter coefficient, h(n) using RECTANGUAR Window define by W(n) = { 1; 0 n 4 0; otherwise. Ans: a) i) Ref Pg.No 389-391, DSP by Salivahanan. ii) Ref Pg.No 399, DSP by Salivahanan. (OR)

b) Design a digital Butterworth filter satisfying the constraints 0.7 | H()| 1.0 ; 0 0.2 | H()| 0.004 ; 0.6 . Apply impulse invariant transformation method. Ans: Ref: Pg.No: 359-362, DSP by Nagoorkani. 13. a) i) Consider the truncation of negative fraction number represented in(+1) bit fixed point binary form including sign bit . Let (-b)

bits be truncated .Obtain the range of truncation errors for signed magnitude ,2s complement and 1s complement representation of negative numbers. ii) A 8-bit ADC feeds a DSP system characterised by the following Transfer function H(Z) = 1/(Z+0.5) .Estimate the steady state Quantisation noise power at the output of the system. Ans: a) i) Ref Pg.No 496-499, DSP by Salivahanan. ii) Ref Pg.No 504-505, DSP by Salivahanan. (OR) b) i) The coefficients of a system defined by 1 H(Z) = (1-0.4Z-1)(1-0.55Z-1) are represented in anumber with a sign bit and 3 data bits. Determine the new pole location for 1) Direct realization and 2) Cascade realization of first order systems.Compare the movements of the new pole away from the original ones in both the cases. ii) An IIR causal filter has the system function H(Z)=Z/Z-0.97,Determine the dead band of the filter. Ans: b) i) Direct form: 1/ [1-0.875z-1+0.125Z-2] Cascade form:1/[1-0.375Z-1][1-0.5Z-1] ii) Ref Pg.No 510, DSP by Salivahanan.

14. a)i) With Suitable relation ,explain briefly the periodogram method of power spectral estimation.Examine the consistency and bias of the periodogram. ii) Explain power spectrum estimation using the bartlett method. Ans: a) i) Ref Pg.No 588-589, DSP by Salivahanan. ii) Ref Pg.No 594, DSP by Salivahanan.

(OR) b) i) Explain how the Black man and Tukey is used in smoothing the periodogram? Derive the mean and variance of the power

spectral estimate of the blackman and Tukey method. ii) Determine the frequency resolution of the Bartlett, Welch and Blackman and Tukey methods of the power spectral estimation for a quality factor Q =15 and sample is 1500. Ans: a) i) Ref Pg.No 596-599, DSP by Salivahanan. ii) Ref Pg.No 606, DSP by Salivahanan.

15.a) i) Explain how Harvard architecture as used by the TMS 320 family differs from the Strict Harvard architecture .Compare this with the architecture of a satndard Von Neumann Processor. ii) Explain the operation of MAC unit. Ans: Ref Pg.No 40-44, Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) i) In relation to DSP processor, explain the following technique: SMID, VLIW. ii) Explain the operation of CSSU of TMS 320 C54X and explain its use considering the viterbi operator. Ans: Ref Pg.No 40-44 &269-270 , Digital Signal Processor by B.Venkataramani &M.Bhaskar

b)

B.E/B.TECH DEGREE EXAMINATION, APRIL /MAY 2008

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Question

1. Define the properties of convolution. 1. Commutative property: x(n)*h(n) = h(n) *x(n) 2. Associative Property: [x(n)*h1(n)] *h2(n) = x(n)*[h1(n)*h2(n)] 3. Distributive Property: x(n)*[ h1(n)+ h2(n)] = [x(n)* h1(n)]+ [x(n)* h1(n)] 2. Draw the basic butterfly diagram of radix -2 FFT. 1 a 1 1 WN b
nk

1 A = a+ WNnk b

B = a - WNnk b

-1 3. What are the merits and demerits of FIR filter? Merits : 3. Linear phase filter. 4. Always Stable Demerits: 3. The duration of the impulse response should be large 4. Non integral delay. 4. What is the relationship between analog and digital frequency in impulse invariant transformation? Digital Frequency: = T = analog frequency T= Sampling interval

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

i. ii. iii.

Input quantization error Coefficient quantization error Product quantization error

6. What is meant by limit cycle oscillation in digital filter? In recursive system when the input is zero or same non-zero constant value the non linearities due to finite precision arithmetic operation may cause periodic oscillation in the output. Thus the oscillation is called as Limit cycle. 7. Define Periodogram. F {rxx (m)} = 1/N X (f)2 The above equation gives the estimation of power spectral density (PSD). This equation is also called as Periodogram. 8. Determine the frequency resolution of the Bartlett method of power spectrum estimates for a quality factor Q =15 .Assume that the length of the sample sequence is 1500. Solution: Q Bart = 1.11N f Frequency resolution: f = 15 /(1.11 *1500) f = 0.0009 9. What is pipelining? A task is broken down in to a number of distinct subtasks which are then overlapped during execution. It is used in digital signal processors to increase speed. 10.What is the principal feature of the Harvard architecture? The program and data memory lies in two separate spaces, permitting full overlap of instruction, Fetch and execute. PART B (5*16 = 80 MARKS) 11.a) i) Disuss in detail the imporatnt properties of the DFT. ii) Find the 4 point DFT of the sequence x(n) = cos n/4. Ans: i)Ref Pg.No 308-311, DSP by Salivahanan. ii) X(K) = {1, 1-j1.414, 1, 1+j1.414} iii) X(K) = {20,-5.8-j2.4, 0, 0.17-j0.414, 0, -0.17+j0.414, 0, -5.82+j2.414}. (OR)

b)i) Using DIT draw the butterfly line diagram for 8-point FFT calculation and explain. ii)Compute an 8-point DFT using DIF FFT algorithm. X(n) = {1,2,3,4,4,3,2,1} Ans: i) X(K) = {20,-5.8-j2.4, 0, 0.17-j0.414, 0, -0.17+j0.414, 0, -5.82+j2.414}. ii) Ref Pg.No 334-340, DSP by Salivahanan. 12. a) i) Determine the magnitude response of an FIR filter (M=11) and show that the phase and group delays are constant. ii) The desired frequency response of a low pass filter is given by j3 Hd() ={ e ; -3/4 3/4 0 ; other wise. j Determine H(e ) for M= 7using HAMMING window. (OR) b)i) For the analog transfer function H(S) = 1/ (S+1)(S+2) . Determine H(Z) using impulse invariant technique. ii) Design a digital Butterworth filter satisfying the constraints 0.9 | H()| 1.0 ; 0 /2 | H()| 0.2 ; 3/2 . Apply Bilinear transformation method. (16) Ans: a) i) Ref Pg.No 437-439, DSP by Salivahanan. ii) Ref Pg.No 383-384, DSP by Salivahanan. b)i) Ref Pg.No 400-401, DSP by Salivahanan. ii) Ref Pg.No 410-413, DSP by Salivahanan. 13.a) i) Discuss in detail the truncation error and round off error for sign Magnitude and 2s complement. ii) Explain the quantization effects in converting analog signal in to digital signal Ans: a) i) Ref Pg.No 496-499, DSP by Salivahanan. ii) Ref Pg.No 495, DSP by Salivahanan. (OR) b) i) A digital system, is characterized by the difference equation y(n) = 0.9y(n-1) +x(n) .Determine the dead band of the filter.

ii) What is meant by coefficient quantization? Explain Ans: b) Ref Pg.No 510, DSP by Salivahanan. 14.a) i) Explain the Bartlett method of averaging Periodogram. ii) What is the relation ship between autocorrelation and power Spectrum? Prove it. Ans: a) i) Ref Pg.No 588-589, DSP by Salivahanan. ii) Ref Pg.No 594, DSP by Salivahanan. (OR) b)i) Derive the mean and variance of the power spectral estimate of the Blackman and Tukey method. ii) Obtain the expression for mean and variance of the autocorrelation function of random signals. Ans: b) Ref Pg.No 596-599, DSP by Salivahanan. 15.a) i) Describe the MAC unit in DSP Processor. ii) Explain the architecture of TMS 320 C5X DSP Processor. Ans: Ref Pg.No 40-44, &256 Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) b) i) Discuss in detail the four phases of the pipeline techniques. ii) Write short notes on: 1. Parallel Logic 2. Circular register. Ans: Ref Pg.No 40-41,56-65,60,48, Digital Signal Processor by B.Venkataramani &M.Bhaskar

B.E/B.TECH DEGREE EXAMINATION, NOV /DEC 2008

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Questions 1. Distinguish between DIT and DIF FFT algorithm. S.No DIT FFT Algorithm 1. The input is in bit reversed order; the output will be normal order. 2. Each stage of computation the phase factor are multiplied before add subtract operation. DIF FFT Algorithm The input is in normal order; the output will be bit reversed order. Each stage of computation the phase factor are multiplied after add subtract operation.

2. If H(K) is the N-point DFT of a sequence h(n) , Prove that H(K) and H(N-K) are complex conjugates. This property states that, if h(n) is real , then H(N-K) = H*(K) = H(-K) Proof: By the definition of DFT; N-1 X(K) = x(n) e (j2nk)/N n=0 Replace K by N-K N-1 X(N-K) = x(n) e (j2n(N-K))/N X(N-K) n= X*(K) = 3. Show that the filter with h(n) = [-1,0,1] is a linear phase filter. Solution: h(n) = [ -1,0,1] h(0) = -1 = -h(N-1-n) = -h(3-1-0) = -h(2) h(1) = 0 = -h(N-1-n) = -h(3-1-1) = -h(1) h(2) = 1 = -h(N-1-n) = -h(3-1-2) = -h(0) It is a linear phase filter.

4. What is Prewarping? Why is it needed? In IIR design using bilinear transformation the conversion of specified digital frequencies to analog frequencies is called Pre-warping. The PreWarping is necessary to eliminate the effect of warping on amplitude response. 5. Identify the various factors which degrade the performance of the digital filter implementation when finite word length is used. 1. Input quantization error 2. Coefficient quantization error 3. Product quantization error 6. Express the fraction (-7/32) in signed magnitude and 2s complement notations using 6 bits. In Signed Magnitude: 1.001110 In 2s complement: 1.110010 7. What are the disadvantages of non-parametric methods of power spectral estimation? 1. It requires long data sequence to obtain the necessary frequency resolution. 2. Spectral leakage effects because of windowing. 8. Define unbiased estimate and consistent estimate. Expected value of the estimate is equal to actual value is called as unbiased estimate. Consistent estimate: Lim Varience (estimate) = 0. N 9. State the merits and demerits of multi ported memories. A Mutiported memory operates as on chip memory and off chip memory. 1. Higher performance because no wait states are required. 2. Lower cost than external memory. 3. Lower Power than external memory. 4. Higher performance because better flow with in the pipeline at lential arithmetic logic unit. 5. Ability to access large memory space.

10.What are the factors that may be considered when selecting a DSP processor for an application? 1. Architectural features 2. Exection speed 3. Type of arithmetic 4. Word length PART B ( 5*16 = 80 MARKS) 11. a) Two finite duration sequence are given by x(n) = Cos (n/2) for n = 0,1,2,3 h(n) = 0.5 n for n = 0,1,2,3 Determine circular convolution using DFT &IDFT method. Ans: X(K) = {0, -2j, 0, 2j} H(K) = {15, -3+6j, -5, -3-6j} y(n) = {6, -3, -6, 3} (OR) b) i) From first principles obtain the signal flow graph for Computing 8-point using radix -2 DIT FFT algorithm. ii) Using the above signal flow graph compute DFT of x(n) = cos (n/4) ,0 n 7. Ans: i) Ref Pg.No 334-340, DSP by Salivahanan. ii) X(K) = {0, 3, 0, 2.7-j0.7, 0, 1, 0, 1.293-j0.7}

12. a) A band pass FIR filter of length 7 is required. It is to have lower and upper cutt frequencies of 3 KHZ and 5 Khz respectively . and indended to be used with a sampling frequency of 20 KHZ Determine the filter coefficient using HANNING window Consider the filter to be causal. (OR) b) Design a digital Butterworth filter satisfying the constraints 0.8 | H()| 1.0 ; 0 0.2 | H()| 0.2 ; 0.6 . Apply Bilinear transformation method. Ans: Ref: Pg.No: 359-362, DSP by Nagoorkani.

13. a) i) Consider the (b+1) bit bipolar A/D converter. Obtain an expression for signal to quantization noise ratio . ii) Consider the truncation of negative fraction number represented in(+1) bit fixed point binary form including sign bit . Let (-b) bits be truncated .Obtain the range of truncation errors for signed magnitude ,2s complement and 1s complement representation of negative numbers. Ans: a) i) Ref Pg.No 496-499, DSP by Salivahanan. ii) Ref Pg.No 499-503, DSP by Salivahanan. (OR) The coefficients of a system defined by 1 H(Z) = (1-0.3Z-1)(1-0.65Z-1) are represented in a number with a sign bit and 3 data bits. Determine the new pole location for 1) Direct realization and 2) Cascade realization of first order systems. Compare the movements of the new pole away from the original ones in both the cases. Ans: b) Direct form: 1/ [1-0.875z-1+0.125Z-2] Cascade form:1/[1-0.375Z-1][1-0.5Z-1]

b)

14. a)i) With Suitable relation ,explain briefly the periodogram method of power spectral estimation. Examine the consistency and bias of the periodogram. ii) Explain power spectrum estimation using the bartlett method. Ans: a) i) Ref Pg.No 588-589, DSP by Salivahanan. ii) Ref Pg.No 594, DSP by Salivahanan. (OR) b) i) Explain how the Black man and Tukey is used in smoothing the periodogram? Derive the mean and variance of the power spectral estimate of the blackman and Tukey method. ii) Determine the frequency resolution of the Bartlett, Welch and Blackman and Tukey methods of the power spectral estimation for a quality factor Q =15 and sample is 1500.

Ans: a)

i) Ref Pg.No 596-599, DSP by Salivahanan. ii) Ref Pg.No 606, DSP by Salivahanan.

15. a) i) Explain what is meant by instruction pipelining. Explain with an example ,how pipelining increase the through put efficiency. ii) Describe the MAC unit in DSP Processor. Ans: a) i) Ref Pg.No 45-46, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 40-44, &256 Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) b)i) With a suitable diagram describe the function of Multiplier/adder units of TMS 320 C54X. ii) In relation to DSP processor, explain the following technique: SMID, VLIW. Ans: b) i) Ref Pg.No 267-268, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 40-44, Digital Signal Processor by B.Venkataramani &M.Bhaskar

B.E/B.TECH DEGREE EXAMINATION, NOV /DEC 2009

FIFTH SEMESTER (REGULATION 2004) ELECTRONICS AND COMMUNICATION ENGINEERING

EC 1302 DIGITAL SIGNAL PROCESSING

TIME: 3 Hrs Max. Mark: 100

PART A (10*2 =20 MARKS)

Answer all Questions

1. What do you mean by radix-2 FFT? The radix -2 FFT is an efficient algorithm for coputing N- point DFT of an N-point sequence .In radix-2 FFT the n-point is decimated into 2-point sequence and the 2-point DFT for each decimated sequence is computed. From the results of 2-point DFTs, the 4-point DFTs are computed. From the results of 4 point DFTs ,the 8-point DFTs are computed and so on until we get N - point DFT. 2. What is the relationship between z transform and DFT? j2k/N DFT of x(n), x(k)=x(z)/z=e
X(Z) is the Ztransform.

3. What are the properties of FIR Filters? FIR filters are stable. FIR filters have linear phase. They need higher order filters for the same magnitude response compared to IIR Filters. 4. State the limitations of Impulse Invariance mapping technique. In Impulse Invariance method, the mapping from S plane to Z plane is many to one. This produces aliasing effect. Hence it is not suitable for designing high pass filters. 5. Define zero input limit cycle oscillation. For an IIR filter, implanted with infinite precision arithmetic, the output should approach zero in the steady state if the input is zero,and it should approach a constant value, if the input is a constant. Due to finite word length effect, the output may oscillate between positive and negative values.This is called zero input limit cycle.

6. What is called dead band? The amplitude of the output during a limit cycle is confined to a range of values called the dead band of the filter. 7. What is the need for spectrum estimation? To detect the narrow band signals in noise, spectrum estimation is required. 8. Define Periodogram. F {rxx (m)} = 1/N X (f)2 The above equation gives the estimation of power spectral density (PSD). This equation is also called as Periodogram. 9. What is meant by pipelining? A task is broken down in to a number of distinct subtasks which are then overlapped during execution. It is used in digital signal processors to increase speed. 10. What are the desirable features of DSP Processors? (i) DSP processors should have multiple registers. (ii) Multiple operand fetch capacity. (iii) On-chip memory. (iv) Multiprocessing capability. PART B (5*16 = 80 MARKS) 11. (a) (i) Describe the following properties of DFT. (1) Convolution (2) Time Reversal (3) Time Shift (4) Perioddicity. Ans: (1) DFT [xI(n) *x2(n] = X1(k)X2(k) (2) If DFTx(n)=X(K) then DFT[x (-n) ] =X(N-K) (3) DFT[x (n-m) N] =e-j2km/N X(k) (4)If x(n+N)=x(n) for all n, Then X(K+N)=X(K) for all k. and Explanation.

(12)

(ii) Compare the computational complexity of Direct DFT computation and FFT computation of a sequence with N=64. (4) 2 2 Ans: Direct DFT computation: N = (64) = 4096 FFT computation: N/2log2N=64/2log264=192

(OR) (b) (i) Explain Decimation in time FFT algorithm for N=8. Ans: (i) Ref Pg.No 320-328 , DSP by Salivahanan . (ii) Determine the 4 point DFT of x(n) =(0,1,2,3). Ans: w4 =
0 x4= 1 2 3

(8)

(8)

X4 = W4 x4 = 6 -2+2j -2

-2-2j 12. (a) (i) Describe the Frequency sampling method of designing FIR Filters. (8) Ans: (i) In frequency sampling method the desired magnitude response is sampled and a linear phase response is specified.The samples of the desired frequency response are identified as DFT coeffients.The filter coefficients are determined as the IDFT of this set of samples. Frequency sampling method is attractive for narrowband frequency selective filters. (ii)And Explanation. (ii) Derive the condition for linear phase in FIR filters. (8) Ans: When h(n) is symmetric about n=N-1/2 h(n)=h(N-1-n) When h(n) is antisymmetric about n=N-1/2 h(n)=-h(N-1-n) (ii)And Derivation. (OR) (b) (i) The desired response of a low pass filter is Hd (ejw) = e-j3w -3 /4 w 3/4 =0 3/4 w Design the filter for M=7 using Hamming Window. (10) (i) Ref Pg.No 400-401, DSP by Salivahanan. (ii) Using Impulse Invariant mapping, convert the analog transfer Function into digital. Assume T=0.1 sec .

2 H(S) = (s+1)(s+2) (6)

Ans:

H(Z)

13. (a) (i) Draw the Quantization noise model for a second order system 1 H(Z) = 1-2rcosz-1+r2z-2 and find the steady state output noise variance. (8) (ii) What are called overflow oscillations? How can it be prevented?(8) Ans: steady state output noise variance v2= v12+ v22 v12=v22 h(n)= rn sin(n+1) u(n)
sin

(OR) (b) (i) Describe the effect of quantization on pole location with an example. (6) (ii) Explain the characteristics of a limit cycle oscillation with respect to the system described by the difference equation y(n) = 0.95y(n-1)+x(n) Determine the dead band of the filter. (10) Ans: b) Ref Pg.No 510, DSP by Salivahanan. Deadband =(-0.625,+0.625) 14. (a) (i) Derive the relationship between autocorrelation and power spectral density of a signal. (8) (ii) Explain the Bartlet method of spectrum estimation. (8) Ans: i) Ref Pg.No 588-589, DSP by Salivahanan. ii) Ref Pg.No 594, DSP by Salivahanan.

(OR) (b) (i) Describe the use of DFT in spectrum estimation. (8) (ii) Explain the Blackman and Tukey method of spectrum estimation. (8) Ans: (i) Ref Pg.No 596-599, DSP by Salivahanan.

15. (a) Write note on: (i) Harvard Arahitecture (ii) Dedicated MAC unit (iii) Multiple ALUs. Ans: a) i) Ref Pg.No 45-46, Digital Signal Processor by B.Venkataramani &M.Bhaskar ii) Ref Pg.No 40-44, &256 Digital Signal Processor by B.Venkataramani &M.Bhaskar (OR) (b) Describe the Advanced modes of DSP processors in detail. Ans: ii) Ref Pg.No 40-44, Digital Signal Processor by B.Venkataramani &M.Bhaskar

(16)

(16)

B.E./B.Tech. DEGREE EXAMINATION, NOVEMBER/DECEMBER 2010 EC 2302 - DIGITAL SIGNAL PROCESSING Maximum : 100 Marks Answer ALL questions http://2freshesworld.com PART A - (10 X 2 = 20 Marks) 1. Obtain the circular convolution of the following sequences x(n) = {1, 2,1} & h(n)= {1, 1 2, 2}.
Using matrix method, Appending zero to the sequence x(n), x(n)={1,2,1.0} h(n) = {1,1,2.2} Answer X(n)= {7,5,5,7}

2. How many multiplications and additions are required to compute N-point DFT using radix-2 FFT?
Complex Addition - Nlog2N Complex Multiplications (N/2)log2N

3. What is prewarping? In IIR design using bilinear transformation the conversion of specified digital frequencies to analog frequencies is called Pre-warping. The PreWarping is necessary to eliminate the effect of warping on amplitude response. 4. What is the advantage of direct form II realization when compared to direct form I realization? In direct form II structure the number of delay elements required is exactly half that for direct form I structure .Hence it requires less amount of memory. 5. Give the equations for Hamming window and Blackman window.
WH(n) = 0.54 - 0.46cos(2n/N-1) ; 0 n N-1 WB(n) = 0.42 0.5 cos(2n/N-1) + 0.08 cos (2n/N-1) ; 0 n N-1

6. Determine the transversal structure of the system function ( ) 1 2 3 1 2 3 4 Hz=+z??z??z? 7. What is truncation? Truncation is the process of discarding all bits less significant than least significant bit that is retained. 8. What is product quantization error? In digital computations the output of multipliers are quantized to finite word length inorder to store the result in the registers 9. What is decimation? The process of reducing the sampling rate of the signal is called decimation

(sampling rate compression). 10. What is sub band coding? It is application of multirate signal processing
http://2freshesworld.comrld.com PART B - (5 X 16 = 80 Marks)

http://2freshesworld. 11. (a) (i) Compute the eight-point DFT of the sequence Using the radix-2 decimation-in-time algorithm. (10) Ref Pg.No 207-209, DSP by A.Nagoor Kani. (ii) Explain overlap-add method for linear FIR filtering of a long sequence. (6) Ref Pg.No 191, DSP by A.Nagoor Kani. (Or) (b) (i) Compute the eight-point DFT of the sequence By using the decimationin-frequency FFT algorithm. (10) Ref Pg.No 215-219, DSP by A.Nagoor Kani. (ii) Summarize the properties of DFT. (6) Ref Pg.No 158-161, DSP by A.Nagoor Kani. 12. (a) Determine the system function H(z) of the Chebyshevs low pass digital filter with the specifications = 1 dB p ? ripple in the pass band 0 ?? ? 0.2 ? = s ? 15 dB ripple in the stop band 0.3? ?? ? ? using bilinear transformation (assume T= 1 sec). (16) Ref Pg.No 365-367, DSP by A.Nagoor Kani. (Or) (b) Obtain the direct form I, direct form II, cascade and parallel form realization for the system y(n) = 0.1 y(n -1) + 0.2 y(n - 2) + 3x(n) + 3.6x(n -1) + 0.6 x(n - 2) (16) Ref Pg.No 54-57, DSP by A.Nagoor Kani.

13. (a) Design an ideal high pass filter with a frequency response Find the values of h(n) for N =11 using hamming window. Find H(z) and determine the magnitude response. (16) Ref Pg.No 298-301, DSP by A.Nagoor Kani. (Or) (b) (i) Determine the coefficients {h(n)} of a linear phase FIR filter of length M =15 which has a symmetric unit sample response and a frequency response that satisfies the condition (10) Ref Pg.No 308-311, DSP by A.Nagoor Kani. (ii) Obtain the linear phase realization of the system function H (z )= + z + z + z + z + z z (6) Ref Pg.No 207-209, DSP by A.Nagoor Kani. 14. (a) Discuss in detail the errors resulting from rounding and truncation. Ref Pg.No 496-499, DSP by Salivahanan Or (b) Explain the limit cycle oscillations due to product round off and overflow errors. (16) Ref Pg.No420-423, DSP by A.Nagoor Kani. 15. (a) Explain the polyphase structure of decimator and interpolator. (16) Ref Pg.No 541-551, DSP by Salivahanan. Or (b) Discuss the procedure to implement digital filter bank using multirate signal processing. (16) Ref Pg.No 578-581, DSP by Salivahanan.