VALLIAMMAI ENGINEERING COLLEGE

SRM Nagar, Kattankulathur – 603 203

Question Bank

DEPARTMENT OF ELECTRONICS AND COMMUNIUCATION

M.E. COMMUNICATION SYSTEMS

AP5152-ADVANCED DIGITAL SIGNAL PROCESSING

(Regulation: 2017)

Academic Year: 2017 – 2018 (Odd Semester)

Prepared By
Mr. M. SELVARAJ, Assistant Professor / ECE
 UNIT I - DISCRETE RANDOM SIGNAL PROCESSING

Wide sense stationary process – Ergodic process – Mean – Variance - Auto-correlation and Auto-correlation matrix
- Properties -Weiner Khitchine relation - Power spectral density – filtering random process, Spectral
Factorization Theorem, Finite Data records, Simulation of uniformly distributed/Gaussian distributed white
noise – Simulation of Sine wave mixed with Additive White Gaussian Noise.
PART A (2 Marks)

BT
Q.No Questions Competence
Level

1 State Parseval’s theorem BTL 1 Remembering

2 Define statistical variance and covariance BTL 1 Remembering
3 What is wide sense stationary process BTL 1 Remembering
4 List the three cases of interest to derive ap(k) of p x p non-symmetric Toeplitz
BTL 1 Remembering
matrix
5 Can you recall the minimum error of all-pole and all-zero modeling? BTL 1 Remembering
6 Recall the definition for ensemble average? BTL 1 Remembering
7 Compute autocorrelation of sequencex(n) = (0.5)nu(n). BTL 2 Understanding
8 Illustrate all modeling using covariance method. BTL 2 Understanding
9 Write the relationship between rx(k) and Sx(w) BTL 2 Understanding
10 What can you say about periodicity of WSS random process? BTL 2 Understanding
11 Solve the power spectrum of wide sense stationary random process for given
BTL 3 Applying
autocorrelation sequence. rx(k)=2δ(k)+j δ(k-1)-jδ(k+1).
12 Develop all pole modeling using autocorrelation method BTL 3 Applying
13 Contrast linear prediction error and linear prediction coefficient BTL 3 Applying
14 Find the autocorrelation sequence for given power spectral density.
BTL 4 Analyzing
Px(ejw)=3+2cosw
15 Examine the classification of the energy of a discrete signal in time and
BTL 4 Analyzing
frequency domain?
16 Inspect the given matrix to be a valid autocorrelation matrix?
1 1+𝑗 BTL 4 Analyzing
[ ]
1−𝑗 1
17 Enumerate two properties of autocorrelation. BTL 5 Evaluating
18 Interpret Wiener Khintchine relation BTL 5 Evaluating
19 Compile the drawbacks of least square modeling method BTL 6 Creating
20 Formulate the normal equations of prony method of modeling a signal as unit
BTL 6 Creating
sample response of a LSI having p poles and q zeros
PART B (13 Marks)

1 i) Define filtering random process. (6)

ii) Derive the Wiener Khitchine relations in spectral analysis of random
BTL 1 Remembering
process(7)
2 i) Summarize the necessary condition to say a process as WSS. (3)
ii) Give the expression for autocorrelation & auto covariance function, BTL 1 Remembering
state the relationship between them. (10)
3 i) Write short note: Mean, Variance. (4)
ii) What is WSS process? Explain in detail. (6) BTL 1 Remembering
iii) List the properties of autocorrelation matrix. (3)
4 Obtain the expressions::
i) Power spectral density (5) BTL 1 Remembering
ii) Power spectral density factorization (8)
5 i) State and prove spectral factorization theorem.(7)
ii) Elaborate on the role of stochastic models designed for discrete random signal BTL 2 Understanding
processing (6)
6 Demonstrate all pole modeling using auto correlation and covariance method
BTL 2 Understanding
(13)
7 i) Outline the properties of auto correlation matrices. (7)
ii) Summarize the steps in the determination of the autocorrelation and power BTL 2 Understanding
spectrum of a random process. (6)
8 i) Solve the filter to generate a random process with a power spectrum P x(ejw)
= (5+4cosw) / (10+6cosw)
from white noise (7) BTL 3 Applying
ii) Find the power spectrum of the wide sense stationary random process that
have correlation sequence rx(k)=δ(k)2(0.5)|k| (6)
9 i) How will you build a filter using Iterative filtering? (7)
BTL 3 Applying
ii) Develop the concept of FIR least square inverse filter. (6)
10 i) Explain the physical significance of Spectral factorization (7)
ii) Calculate unit sample response of linear shift-invariant filter whose power BTL 4 Analyzing
spectrum is of the form Px(ejw) = (9+4cos2ω)/(15+6cosω) (6)
11 Analyze about the finite data processing methods
i) Autocorrelation method. (7) BTL 4 Analyzing
ii) Auto covariance method. (6)
12 Analyze the process of filtering in random process(13) BTL 4 Analyzing
13 Evaluate the autocorrelation matrix for WSS process and explain its each
BTL 5 Evaluating
property. (13)
14 Determine the autocorrelation, power spectral density of the sequence
BTL 6 Creating
𝑥 (𝑛) = {3,2,1,2}
PART – C(15 Marks)
1 Let x(n) be the random process that is generated by filtering white noise
w(n) with a first order linear shift-invariant filter having a system
function
H(z)=1/1-0.25z-1. Design a suitable filter BTL 3 Applying

2 Consider the following 3 x 3 symmetry matrix

BTL 6 Creating
 Determine the Eigen values, Eigen vectors, Determinant, spectral
factorization.
3 Find autocorrelation function of the signal x(n) = an u(n) ; 0<a<1 from
the r(x) find the power spectral density. (15) BTL 6 Creating
4 The power spectrum of a wide-sense stationary process x(n) is Px(ejw)=25
– 24 cos ɷ/26 – 10cos ɷ. Find the whitening filter H(z) that produces unit
variances white noise when the input is x(n). (15) BTL 5 Evaluating

UNIT II - SPECTRUM ESTIMATION

Bias and Consistency of estimators - Non-Parametric methods - Correlation method - Co-variance estimator
- Performance analysis of estimators – Unbiased consistent estimators - Periodogram estimator - Barlett
spectrum estimation - Welch estimation.
PART A (2 Marks)

Q.No Questions BT Competence

Level
1 Compare Parametric and Non-Parametric methods of spectral estimation BTL 1 Remembering
2 Define Periodogram? How can it be smoothed BTL 1 Remembering
3 What are the difficulties in FFT based power spectral estimation methods? BTL 1 Remembering
4 List the disadvantages of non-parametric methods BTL 1 Remembering
5 Mention the relation between correlation and spectral density BTL 1 Remembering
6 List the demerits of the periodogram? BTL 1 Remembering
7 Outline bias and consistency BTL 2 Understanding
Find the mean and variance of the periodogram of white noise power spectral
8 BTL 2 Understanding
density is unity.
Compare performance measures for the Parametric and Non-Parametric
9 BTL 2 Understanding
methods of spectral estimation
Determine the frequency resolution of the Barlett, Welch methods of power
10 spectrum estimates for a quality factor Q=15. Assume that overlap in Welch’s BTL 2 Understanding
method in 50% and length of the sample sequence is 1000
11 Compare the non-parametric methods. BTL 3 Applying
12 Developthe bias equation of modified periodogram BTL 3 Applying
13 Explain sample autocorrelation function. Give the mean value of this estimate BTL 3 Applying
14 Test for Biasing of periodogram and modified periodogram BTL 4 Analyzing
15 Infer the variance of Barlett method and Welch’s method of periodogram? BTL 4 Analyzing
16 Illustrate the importance of parametric methods? BTL 4 Analyzing
17 How the biased estimator used in signal processing? BTL 5 Evaluating
18 Deduce the Barlett spectrum estimation expression. BTL 5 Evaluating
19 Design spectrum estimation using periodogram. BTL 6 Creating
20 Predict the principle behind Welch method to estimate power spectrum? BTL 6 Creating
PART B (13 Marks)
1 Tell about the following methods to measure be spectrum of long duration
signals.
BTL 1 Remembering
(i) (i) Correlation method. (7)
(ii) Covariance method. (6)
2 Derive the variance of the periodogram using Blackman-Tukey method. (13) BTL 1 Remembering
3 Define periodogram. How DFT is used for its computation. (13) BTL 1 Remembering
4 Discuss the Welch method of periodogram averaging. (13) BTL 1 Remembering
5 Outline power spectrum estimation using Barlett method. (13) BTL 2 Understanding
6 (i) Summarize in detail about smoothed spectral estimation. (7)
(ii) Use Levinson recursion to find predictor polynomial corresponding the BTL 2 Understanding
autocorrelation sequence R=[2, 1, 1, -2]T (6)
7 (ii)Explain in detail about the modified periodogram spectrum estimation(6)
(ii) Compare and contrast Bartlett spectrum estimation with Welch estimation BTL 2 Understanding
(7)
8 Let rx(k) be a complex autocorrelation sequence given by r x=[2, 0.5(1+j),0.5j]T.
Use the Levinson-Durbin recursion to solve the autocorrelation normal BTL 3 Applying
equations for a second order all-pole model. (13)
9 Identify how the unbiased consistency estimator used to model the signal. (13) BTL 3 Applying
10 The estimated autocorrelation sequence of a random process are rx(0)=1,
rx(1)=0.5, rx(2)=0.5, rx(3)=0.25. Use Levinson Durbin recursion to determine BTL 4 Analyzing
the autocorrelation equations of a third order filter. (13)
11 Examine how biased and unbiased estimator plays role in signal modeling. (13) BTL 4 Analyzing
12 Verify the bias and consistency of periodogram method of spectrum BTL 4 Analyzing
estimation(13)
13 Develop the Blackman Tukey method of power spectrum estimation. (13) BTL 5 Evaluating
14 Derive the bias and variance of modified periodogram (13) BTL 6 Creating
PART – C(15 Marks)
1 Analyze about various types of periodogram estimation with necessary
equation. (15) BTL 4 Analyze
2 Elaborately about various nonparametric method estimation with necessary
examples. (15) BTL 5 Evaluating
3 Using Barlett’s method estimate the power spectrum from a sequence of
N=2000 samples.
(a) What is the minimum length L that may be used for each sequence if
we are to have a resolution of Δf=0.005? (10)
(b) Explain why it would not be advantageous to increase L beyond the BTL 5 Evaluating
value found in (a). (5)
4 i) Barlett’s method is used to estimate the power spectrum of a process from a
sequence of N=2000 samples. (4)
ii) What is the minimum Length L that may be used for each sequence if we are
to have a resolution of ∆f =0.005? (4)
BTL 6 Creating
iii) Explain why it should not be advantageous to increase beyond L beyond the
value found in (ii) (7)
 UNIT III - LINEAR ESTIMATION AND PREDICTION
Model based approach - AR, MA, ARMA Signal modeling – Parameter estimation using Yule-Walker
method. Maximum likelihood criterion - Efficiency of estimator - Least mean squared error criterion -
Wiener filter - Discrete Wiener Hoff equations - Mean square error

PART A (2 Marks)
BT Competence
Q.No Questions
Level
1 How will you find the Maximum Likelihood estimate BTL 1 Remembering
2 Give the names of special random processes. BTL 1 Remembering
3 What is meant by mean square error? BTL 1 Remembering
4 Compare AR and ARMA BTL 1 Remembering
5 Write the error criterion for LMS algorithm BTL 1 Remembering
6 Draw the spectrum of AR signal modeling.. BTL 1 Remembering
7 Illustrate the maximum likelihood criterion BTL 2 Understanding
8 Demonstrate the properties wiener filter. BTL 2 Understanding
9 Relate between AR and MA process in terms of expressions. BTL 2 Understanding
10 Name any two application of the AR model BTL 2 Understanding
11 How LMS will reduce noise BTL 3 Applying
12 Examine the necessity of estimation and prediction. BTL 3 Applying
13 Build the discrete wiener Hoff equation for signal modeling. BTL 3 Applying
14 Inspect the properties of maximum likelihood criterion BTL 4 Analyzing
15 Illustrate discrete Wiener Hoff equations BTL 4 Analyzing
16 Examine the efficiency of a signal estimator BTL 4 Analyzing
17 How optimum filter can be used in noise cancellation BTL 5 Evaluating
18 Conclude the Yule – Walker method signal estimation. BTL 5 Evaluating
19 Deduce Autoregressive Spectrum estimation using covariance method BTL 6 Creating
20 Design spectrum estimation Autoregressive moving average model. BTL 6 Creating
PART B (13 Marks)
1 Derive the expression power spectral density, auto correlation of ARMA
BTL 1 Remembering
process. (13)
2 How FIR Wiener filter can be used for filtering and prediction(13) BTL 1 Remembering
3 Derive Wiener Hopf equations and the minimum mean square error for the FIR
BTL 1 Remembering
wiener filter(13)
4 Derive Wiener Hopf equations and the minimum mean square error for a non-
BTL 1 Remembering
causal wiener filter(13)
5 i) Outline Wiener Hopf equations and the minimum mean square error for a
causal wiener filter (7)
BTL 2 Understanding
ii) How will you relate maximum likelihood criterion with LMS error criterion.
(6)
6 i) Demonstrate the AR signal modeling with example (10)
BTL 2 Understanding
ii) State about mean square error. (3)
7 Construct Weiner deconvolution for stationary process (13) BTL 2 Understanding
8 i) Inspect the efficiency of the estimator.(7) BTL 3 Applying
 ii) Give the difference between parametric and non parametric estimation with
brief note. (6)
9 Compare and contrast maximum likelihood criterion with LMS error
BTL 3 Applying
criterion(13)
10 Analyze the concept of noise cancellation using wiener filter, Explain with
BTL 4 Analyzing
example. (13)
11 Examine the working of various signal modeling techniques. (13) BTL 4 Analyzing
12 State and explain in detail about the parametric estimation using Yule BTL 4 Analyzing
Walker method. (13)
13 i) Evaluate the ARMA random process (7) BTL 5 Evaluating
ii) Explain widrow’s LMS algorithm with neat diagram (6)
14 Estimate an unknown constant x that are corrupted by uncorrelated, zero BTL 6 Creating
mean white noise and has a variance σv2. (13)
PART – C (15 Marks)
1 Obtain Yule – Walker equation for auto regressive type signal modeling BTL 5 Evaluating
2 The estimated autocorrelation sequence of a random process x(n) for lags
k=0,1,2,3,4 are
Rx(0)=2 rx(1)=1 rx(2)=1; rx(3)=0.5rx(4)=0
Estimate the power spectrum in each of the case Creating
i) x(n) is AR(2) process (4)
ii) x(n) is MA(2) process (4) BTL 6
iii) x(n) is ARMA(1,1) process (4)
iv) x(n) contains a single sinusoid in white noise (3)
Consider an MA(q) process that is generated by difference equation
3

Creating
where w(n) is zero mean white noise with zero variance.
i) Find the unit sample response of the filter that generates y(n) from w(n). (7)
BTL 6
ii) Find the autocorrelation, power spectrum of y(n). (8)

4 Elaborately discuss about various special random process techniques. BTL 5 Evaluating
(15)

UNIT IV - ADAPTIVE FILTERS

Recursive estimators - Kalman filter - Linear prediction – Forward prediction and Backward prediction, Prediction
error - Whitening filter, Inverse filter - Levinson recursion, Lattice realization, Levinson recursion algorithm for
solving Toeplitz system of equations.
PART A (2 Marks)

Q.No Questions BT Competence

Level
1 How will you find the forward prediction? BTL 1 Remembering
2 Give the basic principle of Levinson recursion BTL 1 Remembering
3 What are FIR systems BTL 1 Remembering
4 Compare IIR and FIR wiener filters BTL 1 Remembering
5 Write the expression for backward prediction. BTL 1 Remembering
 6 Draw the structure of the forward prediction error filters BTL 1 Remembering
7 Illustrate Lattice structure and the advantage of such structure BTL 2 Understanding
8 Demonstrate the properties of prediction error filters BTL 2 Understanding
9 Relate between forward prediction error and backward prediction error BTL 2 Understanding
10 Name some of the adaptive filter technique. BTL 2 Understanding
11 How will you model a whitening filter BTL 3 Applying
12 Examine the application of linear prediction BTL 3 Applying
13 Build a wiener filter using linear prediction BTL 3 Applying
14 Inspect the properties of inverse filter BTL 4 Analyzing
15 Illustrate Levinson recursion equation. BTL 4 Analyzing
16 Examine the applications of Kalman filter BTL 4 Analyzing
17 How FIR wiener filter can be used as noise cancellation BTL 5 Evaluating
18 Estimate the minimum error for a noncausal filter BTL 5 Evaluating
19 Construct the Toeplitz matrix for Levinson recursion BTL 6 Creating
20 Develop causal IIR wiener filter for noise cancellation BTL 6 Creating
PART B (13 Marks)
1 (i) Describe the basics of forward linear prediction. Give the schematic of FIR
filter and Lattice filter for the first order predictor(7)
BTL 1 Remembering
(ii) Derive the recursive predictor coefficients for optimum lattice predictor by
Levinson-Durbin algorithm. (6)
2 How Levinson recursion algorithm used for solving Toeplitz system of
BTL 1 Remembering
equations(13)
3 Derive the necessary expression for forward and backward prediction. (13) BTL 1 Remembering
4 Give detailed note about Whitening filter with adaptive equations. (13) BTL 1 Remembering
5 (i) Outline lattice realization with example. (8)
BTL 2 Understanding
(ii) How will you understand the term ‘recursive’. Explain? (5)
6 Infer how recursive estimation utilized in signal reconstruction. Explain with
BTL 2 Understanding
necessary examples. (13)
7 Construct discrete Kalman filter for recursive estimation process. (13) BTL 2 Understanding
8 (i)Inspect the signal estimation using Kalman filter of a non-stationary process.
(7) BTL 3 Applying
(ii)Give the properties of linear estimators and the Cramer-Rao bound. (6)
9 Compare and contrast various prediction techniques. (13) BTL 3 Applying
10 Analyze in detail about Prony’s method of solving normal equations (13) BTL 4 Analyzing
11 Examine the working of adaptive filters based on steepest descent method BTL 4 Analyzing
(13)
12 State and explain in detail about the Levinson recursion algorithm for BTL 4 Analyzing
solving Toeplitz system of equations. (13)
13 (i) Evaluate the desirability of Kalman filter from wiener filter in the
estimation using signal flow graph. (7) BTL 5 Evaluating
(ii) Explain forward prediction technique. (6)
14 Estimate an unknown system of matrix equation using Levinson BTL 6 Creating
recursion. (13)
PART – C(15 Marks)
1 Let us consider linear prediction in noisy environment. Suppose that a signal is
BTL 5 Evaluating
corrupted by noise. x(n)=d(n)+w(n) , where rw(k)=0.5δ(k) and rdw(k)=0. The
 signal d(n) in an AR(1) process that satisfies the difference equation
d(n)=0.8d(n-1)+v(n) , where v(n) is white noise with variance σv2=0.36. Assume
that w(n) and v(n) are uncorrelated. Design a causal Wiener predictor and
compute mean square error. (15)
2 Use Levinson Durbin recursion to find third-order all-pole model for a
signal having auto-correlation values BTL 6 Creating
rx(0)=1 rx(1)=0.5 rx(2)=0.5 rx(3)=0.25 (15)
Consider the signal x(n)=δ(n)+bδ(n-1)
3 Suppose that we observe x(n) for n=0,1,….,N.
(a) Using autocorrelation method, find the 2nd- order all pole model
for x(n). BTL 6 Creating
(b) Suppose we want to find a p-pole model for x(n). Let Ѓj denote
the jth reflection coefficient of the lattice filter implementation of
all pole model. Find a recursion for the reflection coefficients
that expresses Ѓj in terms of Ѓj-1 and Ѓj-2
4 With the necessary equations derive the expressions for FIR Wiener BTL 5 Evaluating
Filter? (15)

UNIT V - MULTIRATE DIGITAL SIGNAL PPROCESSING

FIR Adaptive filters - Newton's steepest descent method - Adaptive filters based on steepest descent
method - Widrow Hoff LMS Adaptive algorithm - Adaptive channel equalization – Adaptive echo
canceller - Adaptive noise cancellation - RLS Adaptive filters - Exponentially weighted RLS - Sliding
window RLS - Simplified IIR LMS Adaptive filter.

PART A (2 Marks)
Q.No Questions BT Competence
Level
1 Why are FIR filters used in adaptive filter application BTL 1 Remembering
2 What is adaptive noise cancellation BTL 1 Remembering
3 Define misadjustment of adaptive filter BTL 1 Remembering
4 State the properties of Widrow-Hopf LMS adaptive algorithm BTL 1 Remembering
5 Recall the step size of LMS adaptive filter BTL 1 Remembering
6 Express the LMS adaptive algorithm. State its properties BTL 1 Remembering
7 Demonstrate the need for adaptive filters BTL 2 Understanding
8 Outline about channel equalization BTL 2 Understanding
9 Show how to avoid echo’s in long distance telephonic circuits. BTL 2 Understanding
10 List some applications of Adaptive filters BTL 2 Understanding
11 Illustrate the principle behind LMS algorithm BTL 3 Applying
12 Identify are the advantages of FIR adaptive filters BTL 3 Applying
13 Why LMS is normally preferred over RLS BTL 3 Applying
14 Relate the order of the filter with the step size in LMS adaptive filter BTL 4 Analyzing
15 Write the difference between LMS algorithm and RLS algorithm BTL 4 Analyzing
16 Express the principle of steepest descent adaptive FIR filter BTL 4 Analyzing
17 Explain the advantage of normalized LMS over LMS adaptive filter BTL 5 Evaluating
18 Deduce error function of exponentially weighted RLS BTL 5 Evaluating
19 Develop error function of sliding window RLS BTL 6 Creating
20 Build time constant for the steepest descent FIR adaptive filter BTL 6 Creating
 PART B (13 Marks)
1 i) What is direct form FIR adaptive filter? (7)
BTL 1 Remembering
ii) Derive the weight vector update equation of the LMS algorithm. (6)
2 Discuss adaptive noise cancellation using LMS algorithm. (13) BTL 1 Remembering
3 i)Define adaptive echo cancellation. (7)
BTL 1 Remembering
ii) Explain adaptive channel equalization. (6)
4 Obtain Widrow-Hoff LMS adaptation algorithm (13) BTL 1 Remembering
5 Explain steepest descent algorithm for FIR adaptive filter. (13) BTL 2 Understanding
6 Outline the sliding window RLS algorithm. (13) BTL 2 Understanding
7 Demonstrate the RLS algorithm with the exponentially weighted factor. Write
BTL 2 Understanding
down the update equations for the tap-weight vector (13)
8 Develop normalized LMS algorithm and the convergence. (13) BTL 3 Applying
9 Develop the first order adaptive filter and explain about LMS adaptation
BTL 3 Applying
algorithm in detail (13)
10 i) Give analytical note about adaptive filter? (6)
ii)Discuss the minimum MSE criterion to develop an adaptive FIR BTL 4 Analyzing
filter.(7)
11 i) Illustrate the principle of operation of adaptive filter used as a noise
canceller in reconstruction. (7) BTL 4 Analyzing
ii) Compare and contrast exponentially weighted RLS with sliding
window RLS (6)
12 Deduce the Windrow Hoff LMS adaptive algorithm with its properties BTL 4 Analyzing
(13)
13 Discuss in detail about the steps involved in the design of adaptive filters BTL 5 Evaluating
based on steepest descent method (13)
14 Build an adaptive channel equalizer using steepest descent principle(13) BTL 6 Creating
PART – C(15 Marks)
1 Analyze the performance of adaptive channel equalization and adaptive BTL 4 Analyzing
echo cancellation. (15)
2 Suppose that the input to an FIR LMS adaptive filter is a first order AR
process with autocorrelation rx(k)=cα |k| . where c>0 and 0<α<1. Suppose
the step size µ is µ=(1/5λ max).
(i) How does the rate of convergence of the LMS algorithm depend upon
the value of α? (4)
(ii) What effect does the value c have on the rate of convergence? (4) BTL 5 Evaluating
(iii) How does the rate of convergence of the LMS algorithm depend upon
the desired signal d(n)? (7)
3 The first three autocorrelations of a process x(n) are r x(0)=1, r x(1)=0.5,
rx(2)=0.5 Design a two-coefficient LMS adaptive linear predictor for x(n) BTL 6 Creating
that has a misadjustment M=0.05 and find the steady-state mean-square
error. (15)
4 Explain in detail and derive the expressions for Exponentially weighted
RLS. (15) BTL 5 Evaluating