Discrete Time Signal Processing

BY

Mohd Farhan

Assistant Professor, Lokmanya Tilak College of Engineering

Discrete Time Signal Processing
Mohd Farhan

Copyright © by Ques10. All rights reserved. No part of this publication may be reproduced, copied, or stored in a
retrieval system, distributed or transmitted in any form or by any means, including photocopy, recording or other
electronic or mechanical methods without the prior written permission of the publisher.

This book is sold subject to the condition that it shall not, by the way of trade or otherwise, be lent, resold, hired
out, or otherwise circulated without the publisher's prior written consent in any form of building or cover other
than which it is published and without a similar condition including this condition being imposed on the
subsequent purchaser and without limiting the rights under copyright reserved above.

Edition: May 2020

This edition is for sale in India only. Sale and purchase of this book outside of this country is unauthorized by the
publisher.

Published by:
Ques10
Shop No. 18, Tapovan Arcade, Nahur (W), Mumbai - 400078.
Email: [email protected]
Website: www.ques10.com
Preface
People say, "Don't judge a book by its cover". But don't we?

As engineering students, we are master procrastinators. We only want to study stuff that is absolutely necessary.

So just days before the exams, when no-head-no-tail-heavy-weight reference books stop making any sense, we
start looking for something which does. We run to the nearest bookstore and buy whatever everyone else has.

But, while studying we do notice that they are not up to the mark. No clear explanation, tons of mistakes, tough
sentences.. you know what we are talking about. That's why we have made Ques10 books.

We thought what if we can write books which are concise and to-the-point?

So we approached academically proficient faculties who have been teaching for years and asked them to write
decent books. And we're proud of what we've come out with.

It needs to be mentioned that many have helped us unconditionally along the way. We can't name them all here
because that'd be a really long list. We thank them for their support. We also thank those who have dedicatedly
mailed us for their priceless comments and reviews.

That’s pretty much it. We hope you like our books and benefit from them.

Team Ques10
Syllabus
1.0 Discrete Fourier Transform & Fast Fourier Transform

1.1 Definition and Properties of DFT, IDFT, Circular convolution of sequences using DFT and IDFT. Filtering of
long data sequences: Overlap-Save and Overlap-Add Method for computation of DFT

1.2 Fast Fourier Transforms (FFT), Radix-2 decimation in time and decimation in frequency FFT algorithms,
inverse FFT, and introduction to composite FFT.

2.0 IIR Digital Filters

2.1 Types of IIR Filters (Low Pass, High Pass, Band Pass, Band Stop and All Pass), Analog filter approximations:
Butterworth, Chebyshev I, Elliptic.

2.2 Mapping of S-plane to Z-plane, impulse invariance method, bilinear transformation method, Design of IIR
digital filters (Butterworth and Chebyshev-I) from Analog filters with examples.

2.3 Analog and digital frequency transformations with design examples.

3.0 FIR Digital Filters

3.1 Characteristics of FIR digital filters, Minimum Phase, Maximum Phase, Mixed Phase and Linear Phase Filters.
Frequency response, location of the zeros of linear phase FIR filters.

3.2 Design of FIR filters using Window techniques (Rectangular, Hamming, Hanning, Blackmann, Kaiser), Design
of FIR filters using Frequency Sampling technique, Comparison of IIR and FIR filters

4.0 Finite Word Length effects in Digital Filters

4.1 Quantization, truncation and rounding, Effects due to truncation and rounding, Input quantization error,
Product quantization error, Coefficient quantization error, Zero-input limit cycle oscillations, Overflow limit cycle
oscillations, Scaling.

4.2 Quantization in Floating Point realization of IIR digital filters, Finite word length effects in FIR digital filters.

5.0 DSP Processors

5.1 Introduction to General Purpose and Special Purpose DSP processors, fixed point and floating point DSP
processor, Computer architecture for signal processing, Harvard Architecture, Pipelining, multiplier and
accumulator (MAC), Special Instructions, Replication, On-chip memory, Extended Parallelism.

5.2 General purpose digital signal processors, Selecting digital signal processors, Special purpose DSP hardware,
Architecture of TMS320CX fixed and floating DSP processors.

6.0 Applications of Digital Signal Processing

6.1 Application of DSP for ECG signals analysis.

6.2 Application of DSP for Dual Tone Multi Frequency signal detection.

6.3 Application of DSP for Radar Signal Processing.

V0516th20201
 Chapter 1
Discrete Fourier Transform and Fast Fourier
Transform
Content

DFT
IDFT
FFT
Compare DFT and FFT
Circular convolution
Parserval Theorem

Chap 1 / DFT
1. Explain Quantization effect in computation of DFT
Let x(n) be a finite duration complex-valued sequence of length N.
nk
By definition, X(k) = DFT[x(n)] = ∑ Nx(n) ⋅ W N , where k = 0, 1, 2, …N − 1 and W is the twiddle factor.

nk
Let the real and imaginary components of x(n) and W N be represented by b bits with fixed-point arithmetic. Each
complex multiplication involves 4 real multiplications. Each real multiplication is rounded from 2 b bits to b bits,
and hence there are four quantization (round-off) errors for cach complex-valued multiplication.

In the direct computation of the DFT, there are N complex-valued multiplications for each point in the DFT. So,
there are 4 N real multiplications for each point. Hence, there are 4 N quantization errors.

Round-off errors in DFT multiplication are analyzed using Additive White Noise Model.

Following assumptions are made about the statistical properties of the quantization errors.
−Δ Δ
Quantization (round-off) errors are uniformly distributed random variables in the range 2
to 2
where
Δ = 2 − b.
4 N quantization errors are mutually uncorrelated.
4 N quantization errors are uncorrelated with the sequence x(n).

2 Δ2 2 − 2b
Variance of each quantization error σ e = 12
= 12

2 N
Variance of the quantization errors from the 4 N multiplications σ q = 3
2 − 2b

2 1
If N = 2 m then σ q = 3 2 − 2 ( b − m / 2 )

Hence the variance of the quantization error is proportional to the size of DFT. For every four-fold increase in the
size N of the DFT requires an additional bit in computational precision to maintain same quantization error.
Moreover, while computing DFT another Quantization error is occurs because of overflow due to addition To
prevent overflow due to addition, the input sequence to the DFT must to scaled.
N−1
To satisfy the scaling condition ∑ n = 0 | x(n) | < 1, each point in the input sequence should be divided by N.

Chap 1 / DFT
2. Compute the DFT of the sequence x(n) = {0, 1, 2, 1}
nk
For N = 4, Twiddle factor is W 4 = e − j2πkn / 4

Twiddle factor Matrix

0 0 0 0
W4 W4 W4 W4

W 04 W 14 W 24 W 34
W=
W 04 W 24 W 44 W 64
[ ]
W 04 W 34 W 64 W 94

0
Consider, W 4 = e − 0 / 4 = 1

π π
W 14 = e − jπ2 / 4 = cos 2 − jsin 2 = 0 − j × 1 = − j

2
W 4 = e − jπ4 / 4 = cosπ − jsinπ = − 1 − 0 = − 1

3 3π 3π
W 4 = e − j6π / 4 = cos 2
− jsin 2
=0−j× −1=j

W 44 = e − j8π / 4 = cos2π − jsin2π = 1 − 0 = 1

W 64 = e − j12π / 4 = cos3π − jsin3π = − 1 − 0 = − 1

9π 9π
W 94 = e − j18π / 4 = cos 2
− jsin 2
=0−j×1= −j

1 1 1 1
1 −j −1 j
∴W=
[ 11 −1
j
1
−1
−1
−j
]
Let x(n) = {0, 1, 2, 1}}

∴ DFT[x(n)] = X(k) = W × x(n)

 1 1 1 1 0
1 −j −1 j 1
∴ X(k) = ×
[ 11 −1
j
1
−1
−1
−j
] [ 2
1
]
0+1+2+1
0 − 1j − 2 + 1j
=
[ 00+−1j1 +− 22 −− 11j ]
4
−2
=
[ −02 ]
Hence, X(k) = {4, − 2, 0, − 2}

Chap 1 / DFT
3. Find DFT of the following sequence using DIT FFT algorithm.
x(n) = {1, 1, 1, 1, 1, 1, 1, 0}
Solution:

N = 8 = 23

Given, x(n) = {1, 1, 1, 1, 1, 1, 1, 0}

Step 1: x(n) is written in bit reversed order i.e. {1, 1, 1, 1, 1, 1, 1, 0} is the input for step 1.
0
The phase factor for step 1 is W 2 = e 0 = 1

The butterfly computations for step 1 are:

Output of step 1 is V(k) = {2, 0, 2, 0, 2, 0, 1, 1}

Step 2:Output of step 1 forms the input for second step

2π 2π
The phase factor for step 2 W 04 = e 0 = 1 and W 14 = e − j2π / 4 = cos 4
− jsin 4
=0−j= −j

The butterfly computations for step 2 are:

Output of step 2 is F(k) = {4, 0, 0, 0, 3, − j, 1, j}

Step 3: Output of step 2 forms the input for third step.

0
Phase factors for step 3 are W 8 = e 0 = 1

1 2π 2π 1 1
W 8 = e − j2π / 8 = cos 8
− jsin 8
= −j
√2 √2
2 4π 4π
W 8 = e − j4π / 8 = cos 8
− jsin 8
=0−j×1= −j

3 6π 6π −1 1
W 8 = e − j6π / 8 = cos 8
− jsin 8
= −j
√2 √2
The butterfly computations for step 3 are:
 −1 1 1 1 1 1 −1 1
Output of step 3 = 7,
{ √2
−
√2
j, − j,
√2
−
√2
j, 1,
√2
+
√2
j, j,
√2
+
√2
j
}
Hence, the DFT of given sequence x(n) is
−1 1 1 1 1 1 −1 1
X(k) = DFT{x(n)} = 7,
{ √2
−
√2
j, − j,
√2
−
√2
j, 1,
√2
+
√2
j, j,
√2
+
√2
j
}

Chap 1 / DFT
4. Find the DFT of the following sequence using DIT-FFT ,X{n} =
{1,1,1,1,1,1,0,0,}
Given, x(n) = {1, 1, 1, 1, 1, 1, 0, 0} and N = 8 = 2 3

Step 1: x(n) is written in bit reversed order i.e. {1, 1, 1, 0, 1, 1, 1, 0} is the input for step 1.
0
The phase factor for step 1 is W 2 = e 0 = 1

The butterfly computations for step 1 are:

Output of step 1 is V(k) = {2, 0, 1, 1, 2, 0, 1, 1}

Step 2: Output of step 1 forms the input for second step

2π 2π
The phase factor for step 2 W 04 = e 0 = 1, W 14 = e − j2π / 4 = cos 4
− jsin 4
=0−j= −j

The butterfly computations for step 2 are:

Output of step 2 is F(k) = {3, − j, 1, j, 3, − j, 1, j}

0
Step 3: Output of step 2 forms the input for third step Phase factors for step 3 are W 8 = e 0 = 1

1 2π 2π 1 1
W 8 = e − j2π / 8 = cos 8
− jsin 8
= −j
√2 √2

2 4π 4π
W 8 = e − j4π / 8 = cos 8
− jsin 8
=0−j×1= −j

3 6π 6π −1 1
W 8 = e − j6π / 8 = cos 8
− jsin 8
= −j
√2 √2
The butterfly computations for step 3 are:
Output of step 3 =
{6, − 0.7071 − 1.7071j, 1 − j, 0.7071 + 0.2929j, 0, 0.7071 − 0.2929j, 1 + j, − 0.7071 + 1.7071j}

Hence, the DFT of given sequence x(n) is

X(k) = DFT{x(n)} = {6, − 0.7071 − 1.7071j, 1 − j, 0.7071 + 0.2929j, 0, 0.7071 − 0.2929j, 1 + j, − 0.7071 + 1.7071

Chap 1 / DFT
5. Compute 4- point DFT of a casual four sample sequence given by X(n) =
{j, 0, j, 1}

Chap 1 / DFT
6. Find DFT of the following sequence using DIT FFT algorithm. X(n)={-1,
-1, 2, 0, 2, 0, 2, 0} and sketch the magnitude and phase response.

Chap 1 / DFT
7. Let x be a finite sequence with DFT

X = DFT[x] = [0, 1 + j, 1, 1 − j]

Using the properties of the DFT determine the DFT's of the following:

i) y[n] = e j ( π / 2 ) nx(n)

ii) y[n] = cos(π / 2)n x(n)

iii) y[n] = x[(n − 1) 4]

iv y[n] = [0, 0, 1, 0] ⊕ x[n] with ⊕ denoting circular convolution

No answer yet!

Chap 1 / DFT
8. Find DFT of the image
0121
1232
2343
1232

Step1: Perform Row-wise Transform using FFT Flowgraph

For row1 = [0, 1, 2, 1]

For row2 =row4= [1, 2, 3, 2]

For row3 = [2, 3, 4, 3]

Result of row-wise transform is,

4 −2 0 −2
8 −2 0 −2
R=
[ 128 −2 0
−2 0
−2
−2
]
Step2: Perform Column-wise Transform using FFT Flowgraph

For column1 = [4, 8, 12, 8]

For column2=column4 = [-2, -2, -2, -2]

For column3 =[0, 0, 0, 0]

32 −8 0 −8
−8 0 0 0
Result of column-wise transform,C =
[ −08 0
0
0
0
0
0
]
1
Step3: Scale by
N

1
DFT[f(x, y)] =
N[C]

8 −2 0 −2
−2 0 0 0
DFT[f(x, y)] =
[ −02 0
0
0
0
0
0
]
Chap 1 / DFT
9. State and prove shifting properly of DFT
Appeared in exams: Once

If x(n)( ↔ ) DFT X(k),then

− j2 π kl
x(n − l)( ↔ ) DFTX(k). e N

Proof:We have IDFT equation

1 − j2 π kn
x(n) = ∑ N − 1X(k). e
N k=0
N

− j2 π kl − j2 π kl j2 π kn
N−1
IDFT[X(K). e N ] = 1 / N ∑ k = 0 X(k). e N .e N

−j(2πk(n−l) )
1 / N ∑N −1
k=0
X(K). e N

Comparing with DFT equation we have

− j2 π kl
x(n − l) < − DFT > X(K). e N

Chap 1 / DFT
10. Check whether 1-D DFT is an example of a unitary transform or not.
The Fourier transform can be represented as

F = Wf where f is the input and W is the DFT matrix.

Taking N = 4, we form a DFT matrix

Chap 1 / DFT
11. Derive the relationship between DFT and DTFT
Appeared in exams: 2 times

Discrete Time Fourier Transform is given by

X(ω) = ∑ n∞= − ∞ x(n). e − j ω n ………(1)

We know that DFT is given by

− j2 π kn
X(k) = ∑ N −1
n=0
x(n). e N ……….(2)

By comparing equation (1) and (2) we get

X(k) = X(ω) | ω=
2πk
N

This equations are equal if

i) Infinite summation is replaced by finite

ii) Continuous frequency variable is replaced by finite number of frequencies located at 2πk/NTs

Chap 1 / DFT
12. Prove that 2D DFT matrix is an unitary matrix.
1 1 1 1
1 1 −j −1 j
Consider DFT matrix for N=4,A =
[1
√4 1 − 1
j
1
−1
−1
−j
]
If AA*=I

Then A is a unitary matrix

For N=4,

1 1 1 1 1 1 1 1
1 1 −j −1 j 1 1 j −1 −j
AA ∗ =
[1
√4 1 − 1
j −1
1
−j
] [
− 1 √4 1
1
−1
−j
1
−1
−1
j
]
4 0 0 0
1 0 4 0 0
= 4
[ 00 0
0
4
0
0
4
]
1 0 0 0
0 1 0 0
=
[ 00 0
0
1
0
0
1
]
AA*= I

Therefore DFT matrix A is a unitary matrix.

Chap 1 / DFT
13. (i) Determine the DFT of the sequence x(n) (ii) Also Find the DFT of the
following sequences, using the result obtained in (i)
i)

x(n) = 1, 0 < n < 3

x(n) = 0, 4 < n < 7

ii)

x 1(n) = 1, n = 0
x 1(n) = 0, 1 < n < 4

x 1(n) = 1, 5 < n < 7

x 2(n) = 0, 0 < n < 1

x 2(n) = 1, 2 < n < 5

x 2(n) = 1, 6 < n < 7

x(n)={1,1,1,1,0,0,0,0}

DFT by DIF-FFT

Output of stage-1

S 1(0) = x(0) + x(4) = 1 + 0 = 1

S 1(1) = x(1) + x(5) = 1 + 0 = 1

S 1(2) = x(2) + x(6) = 1 + 0 = 1

S 1(3) = x(3) + x(7) = 1 + 0 = 1

S 1(4) = [x(0) − x(4)]W 08 = (1 − 0)(1) = 1

S 1(5) = [x(1) − x(5)]W 18 = (1 − 0)(0.707 − j0.707)

S 1(6) = [x(2) − x(6)]W 28 = (1 − 0)( − j) = − j

 3
S 1(7) = [x(3) − x(7)W 8 = (1 − 0)( − 0.707 − j0.707)

Output of stage-2

S 2(0) = S 1(0) + S 1(2) = 1 + 1 = 2

S 2(1) = S 1(1) + S 1(3) = 1 + 1 = 2

0 0
S 2(2) = S 1(0) + S 1(2)]W 8 = (1 − 1)W 8 = 0

2 2
S 2(3) = S 1(1) + S 1(3)]W 8 = (1 − 1)W 8 = 0 = 0

S 2(4) = S 1(4) + S 1(6) = 1 − j

S 2(5) = S 1(5) + S 1(7) = 0.707 − j0.707 − 0.707 − j0.707 = − 1.141j

S 2(6) = [S 1(4) − S 1(6)]W 08 = (1 + j)(1) = 1j = 1 + j

S 2(7) = [S 1(5) − S 1(7)]W 28 = (0.707 − j0.707 + 0.707 + j0.707)( − j) = − 1.414j

Final Output

X(0) = S 2(0) + S 2(1) = 2 + 2 = 4

X(4) = S 2(0) − S 2(1) = 2 − 2 = 0

X(2) = S 2(2) + S 2(3) = 0 + 0 = 0

X(6) = S 2(2) − S 2(3) = 0 − 0 = 0

X(1) = S 2(4) + S 2(5) = (1 − j) + ( − 1.414j) = 1 − 2.414j

X(5) = S 2(4) − S 2(5) = (1 − j) − ( − 1.414j) = 1 + 0.414j

X(3) = S 2(6) + S 2(7) = (1 + j) + ( − 1.414j) = 1 − 0.414j

X(7) = S 2(6) − S 2(7) = (1 + j) − ( − 1.414j) = 1 + 2.414j

X(K) = 4, 1 − 2.414j, 0, 1 − 0.414j, 0, 1 + 0.414j, 0, 1 + 2.414j

ii)x 1(n) = 1, 0, 0, 0, 0, 1, 1,

x 1(n) = x(n + 3)

By the shifting property

x(n + 3)( ↔ ) DFTX(k). )W 8− 3K

0
X 1(0) = X(0)W 8 = 4(1) = 4
X 1(1) = X(1)W 8− .3 = (1 − 2.414j)( − 0.707 + j0.707)

−6
X 1(2) = X(2)W 8 = 0

−9
X 1(3) = X(3)W 8 = (1 − 0.414j)(0.707 + 0.707j)

X 1(4) = X(4)W 8− 12 = 0

X 1(5) = X(5)W 8− 15 = (1 + 0.414j)(0.707 − 0.707j)

X 1(6) = X(6)W 8− 18 = 0

− 21
X 1(7) = X(7)W 8 = (1 + 2.414j)( − 0.707 − 0.707j)

X(k) = 4, 1 + 2.414j, 0, 1 + 0.414j, 0, 1 − 0.414j, 0, 1 − 2.414j

ii)x 2(n) = 0, 0, 1, 1, 1, 1, 0, 0

x 2(n) = x(n − 2)

By time shift property,

2K
x(n − 2)( ↔ ) DFTX(k). )W 8

X 2(0) = X(0)W 08 = 4(1) = 4

X 2(1) = X(1)W 28 = (1 − 2.414j)( − j) = − 2.414 − j

X 2(2) = X(2)W 48 = 0

X 2(3) = X(3)W 68 = (1 − 0.414j)(j) = 0.414 + j

8
X 2(4) = X(4)W 8 = (0)(1) = 0

10
X 2(5) = X(5)W 8 = (1 + 0.414j)( − j) = 0.414 − j

X 2(6) = X(5)W 12
8
=0

X 2(7) = X(7)W 14
8
= − 2.414 + j

X 2(k) = 4, − 2.414 − j, 0, 0.414 + j, 0, 0.414 − j, 0, − 2.414 + j

Chap 1 / DFT
14. Justify DFT as a linear transformation.
If x 1(n)( < − > ) DFTx 1(k) And

x 2(n)( < − > ) DFTx 2(k) Then

a 1x 1(n) + a 2x 2(n)( < − > ) DFTa 1x 1(k) + a 2x 2(k)

Proof:- By the definition

x(k) = ∑ x(n) (W n) Kn

here x(n) = a 1x 1(n) + a 2x 2(n)

=> X(K) = ∑[a 1x 1(n) + a 2x 2(n)](W n) Kn

X(K) = a 1X 1(k) + a 2X 2(k)

Hence proved.

Chap 1 / DFT
15. x(n)=4γ(n)+3γ(n-1)+2γ(n-2)+γ(n-3)is a six point sequence. i) Find p(n),if
P(K)= W_N^2K X(K) ii) If Q(K)=X(K-3),Find q(n)
Appeared in exams: Once

x(n)={4,3,2,1,0,0}
2
i) Find p(n)=ifP(K)= W NKX(K)

Above statement signifies, circular time shift properly.

2K
x(n − l)( ↔ ) DFTW 6 X(K)

By comparison we have,

l=2 ,N=6

p(n)=x(n-2)

P(n)={0,0,4,3,2,1}
ii) If Q(K)=X(K-3),Find q(n)

Above statement signifies circular frequency shift.

− j2 π Kl
x(n)e (N ( ↔ ) DFTX(K − l)

By comparison we have,

l=3
− j2 π ln
q(n) = x(n)e 6

q(n) = x(n). e − j π n

q(0) = x(0)e 0 = 4

q(1) = x(1)e − j π = − 3

q(2) = x(2)e − 2j π = 2

q(3) = x(3)e − 3j π = − 1

q(4) = x(4)e − 4j π = 0

q(5) = x(5)e − 5j π = 0

q(n)={4,-3,2,-1,0,0}

Chap 1 / DFT
16. If x(n)={2,3,4,5} i) Find DFT of x(n)using DIT-FFT ii) If y(n)=x(n-
1)Find DFT of y(n) iii) m(n)=x(n)+jy(n). Find DFT of m(m)using above
results only.
Appeared in exams: Once

Output of stage 1

S 1(0) = x(0) + x(2) = 2 + 4 = 6

S 1(1) = x(0) − x(2) = 2 − 4 = − 2

S 1(2) = x(1) + x(3) = 3 + 5 = 8

S 1(3) = x(1) − x(3) = 3 − 5 = − 2

Final Output

X(0) = S 1(0) + S 1(2)W 04 = 6 + 8

1
X(1) = S 1(1) + S 1(3)W 4 = − 2 + 2j

0
X(2) = S 1(0) − S 1(2)W 4 = 6 − 8

X(3) = S 1(1) − S 1(3)W 14 = − 2 − 2j

X(K) = 14, − 2 + 2j, − 2, − 2 − 2j

ii) y(n)=x(n-1), Find DFT of y(n)

by circular shifting property

− j2 π kl
x(n − l)( ↔ ) DFTX(K). e N

x(n − 1)( ↔ ) DFTX(k). W K

4

Y(0) = X(0). W 04 = (14 × 1) = 14

Y(1) = X(1). W 14 = ( − 2 + 2j)( − j) = + 2 + 2j

Y(2) = X(2)W 24 = ( − 2)( − 1) = 2

3
Y(3) = X(3)W 4 = ( − 2 − 2j)( + j) = 2 − 2j

Y(k) = 14, 2 + 2j, 2, 2 − 2j

iii) m(n)=x(n)+jy(n)

By linearly property

M(K)=X(K)+jY(K)

M(0)=14+14j

M(1)=(-2+2j)+j(+2+2j)=-4+4j

M(2)=-2+2j

M(3)=(-2-2j)+j(2-2j)=0

M(K)={14+14j,-4+4j,-2+2j,0}
Chap 1 / DFT
17. List the properties of 2D-DFT.
Translation
Distributive and scaling
Rotation
Periodicity and Conjugate Symmetry
Separability (kernel separating)
Linearity
Convolution and Correlation

Chap 1 / DFT
18. Find the DFT using MATRIX method x(n) = {0, 1, 2, 1}.

Chap 1 / DFT
19. The first points of 8-point DFT of real valued sequence are{0.25,0.125-
j0.3018,0,0.125-j0.0518,0} Find the remaining three points.
By Symmetry property we have,

X(k) = X ∗ (N − k)
We have

X(0)=0.25

X(1)=0.125-j0.3018

X(2)=0

X(3)=0.125-j0.0518

X(4)=0

k = 5, X(5) = X ∗ (8 − 5)

=X ∗ (3)

=0.125+j0.0518

k = 6, X(6) = X ∗ (8 − 6)

= X ∗ (2)

=0

k = 7, X(7) = X ∗ (8 − 7)

= X ∗ (1)

=0.125+j0.3018

Chap 1 / DFT
20. State the following DFT properties:
1. Linearity property

Statement: If x 1(n) ← FT → X 1(ω)and x 2(n) ← FT → X 2(ω)

then, a 1x 1(n) + a 2x 2(n) ← FT → a 1X 1(ω) + a 2X 2(ω)

Fourier transform of a linear combination of two or more signals is equal to the same linear combination of fourier
transform of individual signal.

2. Periodicity

Statement: The periodicity is defined as X(ω) = X(ω + 2ωk).

If x(n) ← FT → X(k)

Then, 1.x(n) = x(n + N), 2.X(k) = X(k + N)

3.Time shift

Statement: If x(n) ← FT → X(ω)then, x(n − k) ← FT → e − j ω kX(ω)

If a signal is shifted in time domain by k samples then the magnitude spectrum is unchanged but the phase
spectrum is unchanged by amount ( − ωk).

4. Convolution

Statement: If x 1(n) ← FT → X 1(ω)andx 2(n) ← FT → X 2(ω)

then, x 1(n) ∗ x 2(n) ← FT → X 1(ω) ∗ X 2(ω)

Convolution of two signals in time domain is equivalent to multiplication in frequency domain.

5. Time Reversal

Statement: If x(n) ← FT → X(ω)then, x( − n) ← FT → X( − ω)

If we fold the sequence in time domain then the magnitude spectrum is unchanged but the polarity of phase
spectrum is unchanged.

Chap 1 / DFT
21. What is the formula of 1-D DFT and 2-D DFT along with its inverse?

Chap 1 / DFT
22. Compute the circular convolution of x(n){2,1,2,1}and h(n)={1,2,3,4,}by
using FFT-IFFT method.
Calculate DFT of x(n)by DIT-FFT
Calculate DFT of h(n)by DIT-FFT

We know that convolution in time domain is equivalent to multiplication in frequency domain

Y(k)=X(k).H(k)

={60 ,0,-4,0}

Now, calculate IDFT of Y(k)by IDIT-FFT

y(n)={14,16,14,16}

Chap 1 / IDFT
23. Compute IDFT of the following sequence using inverse FFT algorithm.
x(k) = {3, 0, 3, 0, 3, 0, 3, 0}
Solution:

N = 8 = 23

The computation of 8 − point DFT using radix- 2 DITFFT involves three steps.

Given, X(k) = {3, 0, 3, 0, 3, 0, 3, 0}

X ∗ (k) = {3, 0, 3, 0, 3, 0, 3, 0}

Step 1: The four pairs of X ∗ (k) in bit reversed order is the input for step 1
0
The phase factor for step 1 is W 2 = e 0 = 1

The butterfly computations for step 1 are:

Output of step 1 is V(k) = {6, 0, 6, 0, 0, 0, 0, 0}

Step 2: The output of step 1 forms the input for second step.

0 1 2π 2π
The phase factor for step 2 are W 4 = e 0 = 1& W 4 = e − j2π / 4 = cos 4
− jsin 4
=0−j×1= −j

The butterfly computations for step 2 are:

Output of step 2 is F(k) = {12, 0, 0, 0, 0, 0, 0, 0}

Step 3: The output of step 2 forms the input for third step.
0
The phase factor for step 3 are W 8 = e 0 = 1;

2π 2π 1 1
W 18 = e − j2π / 8 = cos 8
− jsin 8
= −j
√2 √2

2 4π 4π
W 8 = e − j4π / 8 = cos 8
− jsin 8
=0−j×1= −j

3 6π 6π −1 1
W 8 = e − j6π / 8 = cos 8
− jsin 8
= −j ;
√2 √2
The butterfly computations for step 3 are:

Output of step 3 is q(n) = {12, 0, 0, 0, 12, 0, 0, 0}

∴ x(n) = IDFT[X(k)]
1
= N
q ∗ (n)

1
= 8 {12, 0, 0, 0, 12, 0, 0, 0}
 1
= 8 {12, 0, 0, 0, 12, 0, 0, 0}

= {1.5, 0, 0, 0, 1.5, 0, 0, 0}

Hence, the IDFT of X (k) is

= {1.5, 0, 0, 0, 1.5, 0, 0, 0}

Chap 1 / FFT
24. Explain the speed improvement in calculating the DFT using FFT.
Computational benefits of FFT over DFT are

Complex Multiplication Complex Addition

2
DFT N N(N − 1)
N
FFT 2
log 2N Nlog 2N

4N 2 2N
Speed Improvement Factor for complex multiplication in Radix-2 FFT = 2Nlog 2 N
= log 2 N
= 2Nlog N2

For example, Consider a computer which can execute 1 multiplication operation in 1 nano-seconds. We assume
that the amount of time to compute a DFT is determined by the amount of time to perform all the multiplications.

Let us consider a DT signal having total of N = 10 9 points Time required for evaluating DFT by direct method on
the given Computer = N 2 = 10 9 ( ) 2
= 10 18 Nano-sconds, which is approximately 31 years.

N 10 9
Time required for evaluating FFT = 2
log 2N = 2
log 210 9 ≈ 15 × 10 9Nano-seconds approximately 15 secs.

Chap 1 / FFT
25. Find the IDIF-FFT for a given sequence X(k)={26, -2+2j, -2, -2-2j}.
Appeared in exams: Once

Output stage – 1
S 1(0) = X(0) + X(2) = 26 − 2 = 24

S 1(1) = X(1) + X(3) = − 2 + 2j + ( − 2 − 2j) = − 4

S 1(2) = [X(0) − X(2)]W 04 = (26 + 2)(1) = 28

−1
S 1(3) = [X(1) − X(3)]W 4 = ( − 2 + 2j − ( − 2 − 2j))( + j) = − 4

Final Output
1 1
x(0) = 4 [S 1(0) + S 1(1)] = 4 [24 − 4] = 5

1 1
x(2) = 4 [S 1(0) − S 1(1)] = 4 [24 + 4] = 7

1 1
x(1) = 4 [S 1(2) + S 1(3)] = 4 [28 − 4] = 6

1 1
x(3) = 4 [S 1(2) − S 1(3)] = 4 [28 + 4] = 8

x(n) = 5, 6, 7, 8

Chap 1 / FFT
26. Develop composite radix DIT-FFT flow graph for N=6=2*3
Appeared in exams: 4 times

We have generated equation for composite radix.

N −1 m nk N −1 ( nm 1 + 1 ) k N −1 ( nm 1 + m 1 − 1 ) k
X(k) = ∑ n =1 0 x(nm 1)W N 1 + ∑ n =1 0 x(nm 1 + 1)W N + ∑ n =1 0 x(nm 1 + m 1 − 1)W N

For N=6=2×3

=m 1 × N 1

i.e N 1 = 3, m 1 = 2

2 2nk 2 ( 2n + 1 ) k
X(k) = ∑ n = 0x(2n)W 6 + ∑ n = 0x(2n + 1)W 6

2 2nk 2 ( 2nk ) k
= ∑ n = 0x(2n)W 6 + ∑ n = 0x(2n + 1)W 6 W6

k
Let, X(k) = X 1(k) + W 6X 2(k)……………………………….(1)

2 2nk
X 1(k) = ∑ n = 0x(2n)W 6

2k 4k
X 1(k) = x(0) + x(2)W 6 + x(4)W 6

X 1(0) = x(0) + x(2) + x(4)

X 1(1) = x(0) + x(2)W 26 + x(4)W 46

4 8
X 1(2) = x(0) + x(2)W 6 + x(4)W 6

Similarly,
2 2nk
X 2(k) = ∑ n = 0x(2n + 1)W 6

X 2(k) = x(1) + x(3)W 2k

6
+ x(5)W 4k
6

X 2(0) = x(1) + x(3) + x(5)

2 4
X 2(1) = x(1) + x(3)W 6 + x(5)W 6

4 8
X 2(2) = x(1) + x(3)W 6 + x(5)W 6

Substitute all values in equation 1

0
X(0) = X 1(0) + W 6X 2(0)……………………….(i)

1
X(1) = X 1(1) + W 6X 2(1)…………………….(ii)

2
X(2) = X 1(2) + W 6X 2(2)…………………….(iii)

X(3) = X 1(3) + W 36X 2(3)

= X 1(0) + W 36X 2(0)…………………..(iv)

4
X(4) = X 1(4) + W 6X 2(4)

3
= X 1(1) + W 6X 2(1)………………….(v)

5
X(5) = X 1(5) + W 6X 2(5)

= X 1(2) + W 56X 2(2)………………..(vi)

Now, develop the algorithm flow diagram as per equations (i) to (vi)
Chap 1 / Compare DFT and FFT
27. If x(n) = {1, 2, 3, 4, 5, 6, 7, 8}, Find X(k) using DIT-FFT algorithm.
Compare the computational complexity of above algorithm with DFT.
Appeared in exams: 3 times

Output of stage – 1

S 1(0)=x(0)+x(4)=1+5=6

S 1(1)=x(0)-x(4)=1-5=-4

S 1(2)=x(2)+x(6)=3+7=10
S 1(3)=x(2)-x(6)=3-7=-4

S 1(4)=x(1)+x(5)=2+6=8

S 1(5)=x(1)-x(5)=2-6=-4

S 1(6)=x(3)+x(7)=4+8=12

S 1(7)=x(3)-x(7)=4-8=-4

*Output of stage 2 *

S 2(0) = S 1(0) + W 08S 1(2) = 6 + (1)(10) = 16

2
S 2(1) = S 1(1) + W 8S 1(3) = − 4 + ( − j)( − 4) = − 4 + 4j

0
S 2(2) = S 1(0) − W 8S 1(2) = 6 − (1)(10) = − 4

S 2(3) = S 1(1) − W 28S 1(3) = − 4 − ( − j)( − 4) = − 4 − 4j

S 2(4) = S 1(4) + W 08S 1(6) = 8 + (1)(12) = 20

S 2(5) = S 1(5) + W 28S 1(7) = − 4 + ( − j)( − 4) = − 4 + 4j

0
S 2(6) = S 1(4) − W 8S 1(6) = 8 − (1)(12) = − 4

2
S 2(7) = S 1(5) − W 8S 1(7) = − 4 − ( − j)( − 4) = − 4 − 4j

Final output
0
X(0) = S 2(0) + W 8S 2(4) = 16 + (1)(20)

1
X(1) = S 2(1) + W 8S 2(5) = − 4 + 4j + (0.707 − j0.707)( − 4 + 4j)

2
X(2) = S 2(2) + W 8S 2(6) = − 4 + ( − j)( − 4)

3
X(3) = S 2(3) + W 8S 2(7) = − 4 − 4j + ( − 0.707 − j0.707)( − 4 − 4j)

6
X(4) = S 2(0) − W 8S 2(4) = 16 − (1)(20)

1
X(5) = S 2(1) − W 8S 2(5) = − 4 + 4j − (0.707 − j0.707)( − 4 + 4j)

2
X(6) = S 2(2) − W 8S 2(6) = − 4( − j)( − 4)

X(7) = S 2(3) − W 38S 2(7) = − 4 − 4j − ( − 0.707 − j0.707)( − 44j)

X(K) = 36, − 4 + 9.65j, − 4 + 4j, − 4 + 1.65j, − 4, − 4 − 1.65j, − 4 − 4j, − 4 − 9.65j

Chap 1 / Compare DFT and FFT
28. Compute the DFT of the sequence x(n) = {1, 2, 2, 2, 1, 0, 0, 0} using DIF-
FFT algorithm. Compare the computational complexity of above algorithm
with DFT.
Appeared in exams: 2 times

Output of stage-1:

S 1(0) = x(0) + x(4) = 1 + 1 = 2

S 1(1) = x(1) + x(5) = 2 + 0 = 2

S 1(2) = x(2) + x(6) = 2 + 0 = 2

S 1(3) = x(3) + x(7) = 2 + 0 = 2

0 0
S 1(4) = [x(0) − x(4)]W 8 = (1 − 1)W 8 = 0

S 1(5) = [x(1) − x(5)]W 18 = (2 − 0)(0.707 − 0.707j) = 1.414 − 1.414j

2
S 1(6) = [x(2) − x(6)]W 8 = (2 − 0)( − j) = − 2j

S 1(7) = [x(3) − x(7)] = (2 − 0)( − 0.707 − j0.707) = − 1.414 − 1.414j

Output of stage- 2
S 2(0) = S 1(0) + S 1(2) = 2 + 2 = 4

S 2(1) = S 1(1) + S 1(3) = 2 + 2 = 4

S 2(2) = [S 1(0) − S 1(2)]W 08 = (2 − 2)(1) = 0

2
S 2(3) = [S 1(1) − S 1(3)]W 8 = (2 − 2)( − j) = 0

S 2(4) = [S 1(4) + S 1(6)] = (0 − 2j) = − 2j

S 2(5) = [S 1(5) + S 1(7)] = (1.414 − 1.414j − 1.414 − 1.414j)( = 2.818j

0
S 2(6) = [S 1(4) − S 1(6)]W 8 = (0 + 2j)(1) = 2j

2
S 2(7) = [S 1(5) − S 1(7)]W 8 = (1.414 − 1.414j + 1.414 + 1.414j)( − j) = − 2.828j

Final Output

X(0) = S 2(0) + S 2(1) = 4 + 4 = 8

X(4) = S 2(0) − S 2(1) = 4 − 4 = 0

X(2) = S 2(2) + S 2(3) = 0 + 0 = 0

X(6) = S 2(2) − S 2(3) = 0 − 0 = 0

X(1) = S 2(4) + S 2(5) = − 2j − 2.828j = − 4.828j

X(5) = S 2(4) − S 2(5) = − 2j + 2.828j = 0.828j

X(3) = S 2(6) + S 2(7) = 2j − 2.828j = − 0.828j

X(7) = S 2(6) − S 2(7) = 2j + 2.828j = 4.828j

X(K) = 8, − 4.828j, 0, − 0.828j, 0, 0.0828j, 0, 4.828j

Computational Complexity [N=8]

For DFT

No. ofcomplexmultiplication = N 2 = 64

No. ofcomplexaddition = N(N − 1) = 56

For FFT
N
No. ofcomplexmultiplication = 2
log 2N

= 4log 2(8) = 12

No. ofcomplexaddition = Nlog 2N

= 8log 28 = 24

Chap 1 / Circular convolution

29. Find the circular convolution of the sequences x(n) = {1, 2, 1, 2} and
h(n) = {4, 0, 4, 0}
In matrix form,

[y (n) ]
p 4×1 [
= x p(n) ] [h (n) ]
4×4 p 4×1

y p(0) x p(0) x p(3) x p(2) x p(1) h p(0)

y p(1) x p(1) x p(0) x p(3) x p(2) h p(1)
= ×
y p(2) x p(2) x p(1) x p(0) x p(3) h p(2)
[ y (3) ] [ x (3) x p(2) x p(1) x (0)
] [ h (3) ]
p p p p

1 2 1 2 4
2 1 2 1 0
= ×
[ 12 2 1
1 2
]1 [ 40 ]
2

4+0+4+0
8+0+8+0
=
[ 48 ++ 00 ++ 48 ++ 00 ]
y p(0)
8
y p(1) 16
=
[ y (3) ] [ 16 ]
y (2)
p 8

Chap 1 / Circular convolution

30. Compute the circular convolution of the sequence using DFT and IDFT,
x1(n)={1, 2, 0} and x2(n)={2,2,1,1}.
Let N = 4

We make the length of x(n) and h(n) equal to 4 by zero padding.

∴ x(n) = {1, 2, 0, 0}

[ ]
By Definition, DFT x 1(n) = X 1(k) = W × x 1(n)

where W is the Twiddle Factor Matrix for N = 4

 1 1 1 1 1
1 −j −1 j 2
∴ X(k) = ×
[ 11 −1
j
1
−1
−1
−j
] [ 0
0
]
1+2+0+0
1 − 2j + 0 + 0
=
[ 11 +− 2j2 ++ 00 ++ 00 ]
3
1 − 2j
=
[ 1 −+12j ]
Hence, X(k) = {3, 1 − 2j, − 1, 1 + 2j}

Similarly, h(n) = {2, 2, 1, 1}

1 1 1 1 2
1 −j −1 j 2
∴ H(k) = ×
[ 11 −1
j
1
−1
−1
−j
] [ 1
1
]
2+2+1+1
2 − 2j − 1 + j

[ 22 −+ 22j+−11−−1j ]
6
1−j
=
[ 1 0+ j ]
Hence, H(k) = {6, 1 − j, 0, 1 + j}

Let, y(n) = x(n) ⊗ h(n)

By Circular Convolution Property of DFT,

DFT [y(n)] = DFT[x(n) ⊗ h(n)]

∴ Y(k) = X(k)H(k)

For k = 0, Y(0) = X(0)H(0) = 3 × 6 = 18

For k = 1, Y(1) = X(1)H(1) = (1 − 2j)(1 − j) = − 1 − 3j

For k = 2, Y(2) = X(2)H(2) = ( − 1) × 0 = 0

For k = 3, Y(3) = X(3)H(3) = (1 + 2j)(1 + j) = − 1 + 3j

Hence, Y(k) = {18, − 1 − 3j, 0, − 1 + 3j}
1
By Definition, IDFT [Y(k)] = y(n) = N
W ∗ × Y(k)

1 1 1 1 18
1 1 j −1 −j − 1 − 3j
∴ y(n) = ×
4
[ 11 −1
−j −1
1 −1
j
] [ − 1 0+ 3j ]
18 + ( − 1 − 3j) + 0 + ( − 1 + 3j)
1 18 + j( − 1 − 3j) − 0 − j( − 1 + 3j)
= 4
[ 1818−−j(( −− 11 −− 3j) + 0 − ( − 1 + 3j)
3j) − 0 + j( − 1 + 3j)
]
16
1 24
=
[ 12 ]
4 20

4
= 5
3
[ ]
Hence, the circular convolution of the sequences x(n) and h(n) is y(n) = {4, 6, 5, 3}

Chap 1 / Circular convolution

31. Find the circular convolution of the two finite duration sequences
x1(n) = {1, − 1, − 2, 3, − 1} x2(n) = {1, 2, 3}
Let period of x 1(n) and x 2(n) be 5

x 1(n) = {1, − 1, − 2, 3, − 1} and x 2(n) = {1, 2, 30, 0}

In matrix form, [y(n)] 5 × 1 = x 1(n) [ ] [x (n) ]

5×5 2 5×1

x 1(0) x 1(4) x 1(3) x 1(2) x 1(1) x 2(0)

y(0)
y(1) x 1(1) x 1(0) x 1(4) x 1(3) x 1(2) x 2(1)
y(2) = x 1(2) x 1(1) x 1(0) x 1(4) x 1(3) × x 2(2)

[ ] [
y(3) x 1(3) x 1(2) x 1(1) x 1(0) x 1(4)
] [ x (3) ]
2
y(4) x 1(4) x 1(3) x 1(2) x 1(1) x 1(0) x 2(4)
 1 −1 3 −2 −1 1
−1 1 −1 3 −2 2
= −2 −1 1 −1 3 × 3
[
3 −2 −1 1 −1 ] [ ]
0
−1 3 −2 −1 1 0

1−2+9+0+0
−1 + 2 − 3 + 0 + 0
= −2 − 2 + 3 + 0 + 0
[
3−4−3+0+0 ]
−1 + 6 − 6 + 0 + 0

y(0) 8
y(1) −2
∴ y(2) = − 1
[ ] [ ]
y(3) −4
y(4) −1

Chap 1 / Circular convolution

32. Find circular convolution of x1(n) ={5, 6, 2, 1} and x2(n)= {3, 2, 1, 4} by
computing DFT of x 1(n) and x 2(n).
By Convolution Theorem:

Convolution of two signals in time domain is equivalent to multiplication in frequency domain.

x 1(n) ∗ x 2(n) ← FT → X 1(k). X 2(k)

where, X 1(k) & X 2(k) are DFTs of x 1(n) &. x 2(n) respectively.

X 1(k) ={14, 3-5j, 0, 3+5j}

X 2(k) = {10, 2+2j, -2, 2-2j}

x 1(n) ∗ x 2(n) ≡ X 1(k). X 2(k) = {14, 3-5j, 0, 3+5j}.{10, 2+2j, -2, 2-2j}

y(n) = x 1(n) ∗ x 2(n) = {140, 16-4j, 0, 16+4j}

Result: y(n) ={140, 16-4j, 0, 16+4j}

Chap 1 / Circular convolution

33. If x(n)={2,3,4,5}and h(n)={5,2,3,4} i) Find circular convolution by time-
domain method ii) Find circular convolution by frequency domain method.
iii) Determine linear convolution
Appeared in exams: 4 times

Given x(n)={2,3,4,5}and h(n)={5,2,3,4}

i) By time domain method

y(n)=

[2 5 4 3][5]
|3 2 5 4| |2|
|4 3 2 5| |3|
[5 4 3 2][4]

[10 + 10 + 12 + 12]
| 15 + 4+ 15 + 16 |
| 20 + 6+ 6+ 20 |
[25 + 8+ 9+ 8]

[44]
| 50 |
=
| 52 |
[50]

y(n)={44,50,52,50}

ii) By frequency domain method

Calculate DFT of x(n)

X(k) =

[1 1 1 1][2]
|1 −j −1 +j | | 3 |
|1 −1 1 −1 | | 4 |
[1 +j −1 − j][5]
=

14
− 2 + 2j
−2
− 2 − 2j

Calculate DFT of h(n)

H(k) =

[1 1 1 1][5]
|1 −j −1 +j | | 2 |
|1 −1 1 −1 | | 3 |
[1 +j −1 − j][4]

14
2 + 2j
2
2 − 2j

We know that convolution in time domain is equivalent to multiplication in frequency domain

Y(k)=X(k).H(k)

={196,-8,-4,-8}

Now calculate IDFT , to get y(n)

y(n) =

[1 1 1 1][196]
1 |1 +j −1 −j | | − 8 |
*
N |1 −1 1 −1 | | − 4 |
[1 −j −1 + j][ − 2]
1
=4*

176
200
208
200

y(n)={44,50,52,50}

iii) Linear Convolution

y(n)={10,19,32,50,34,31,20}

Chap 1 / Parserval Theorem

34. State and prove Parserval's theorem. Verify it for
x(n) = 1, 2, 3, 4

Appeared in exams: 2 times

If x(n)( ↔ ) DFTX(k) & y(n)( ↔ ) DFTY(k)then

N−1 1 N−1
∑ n = 0 x(n). y ∗ (n) = N
∑ n = 0 X(k). Y ∗ (k)

Proof:-We have
N−1
r xy(m) = ∑ n = 0 x(n). y ∗ (n − m) N

At

m = 0, r xy(m) = ∑ N −1
n=0
x(n). y ∗ (n) ………(1)

By DFT, we have

DFTr xy(m) = X(K). Y ∗ (k)

r xy(m) = IDFTX(k). Y ∗ (k)

By IDFT eqn
1 j2 π km
N−1
r xy(m) = ∑
N k=0
X(k). Y ∗ (K). e N

at m=0
1 N−1
r xy(0) = N
∑ k = 0 X(k). Y ∗ (k) ……………(2)

By comparing eq (1)& (2)

N−1 1 N−1
∑ n = 0 x(n). y ∗ (n) = N
∑ k = 0 X(k). Y ∗ (k)

N−1 1 N−1
∑ n = 0 (x(n)) 2 = N
∑ k = 0 ( | X(k) | ) 2
Given x(n)={1,2,3,4}

Consider LHS

∑N −1
n=0
( | x(n) | ) 2 = ∑ 3n = 0( | x(n) | ) 2

= (1) 2 + (2) 2 + (3) 2 + (4) 2

=30

Consider RHS
1 N−1 1 3
N
∑ k = 0 ( | X(k) | ) 2 = 4
∑ k = 0( | X(k) | ) 2

Calculate DFT of x(n)

X(K) = 10, − 2 + 2j, − 2, − 2 − 2j

1
= 4 [(10) 2 + ( − 2 + 2j) 2 + ( − 2) 2 + ( − 2 − 2j) 2]

1
= 4 100 + ( + 8) + (4) + (8)

1
= 4 [100 + 8 + 4 + 8]

= 30
 Chapter 2
IIR Digital Filters
Content

IIR Filters
Mapping of S-plane to Z-plane
Design of IIR digital filters
Frequency transformation
Digital filter

Chap 2 / IIR Filters

1. Explain the frequency warping in Bilinear transformation.
Here, we will obtain the relationship of jΩ axis in s-plane to the unit circle in the z-plane (r = 1)
2 ( 2rsin ω )
Ω= × ………………(1)
Ts ( r 2 + 2rcos ω + 1 )

For the unit circle,r=1. Thus putting r=1in the equation 1, we get
2 ( 2sin ω )
Ω= Ts
× ( 1 + 2cos ω + 1 )

2 ( 2sin ω )
∴Ω= Ts
× ( 2 + 2cos ω )

2 sin ω
∴Ω= Ts
× ( 1 + cos ω )
…………….….(2)

We have trigonometric identities

ω ω ω
sinω = 2sin 2 . cos 2 and2cos 2 2

1 + cosω

Thus equation (1) becomes,

ω ω
2 2sin 2
cos 2
Ω= Ts
× ω
2cos 2 2

ω
2 2sin 2
Ω= Ts
× ω
2cos 2

2 ω
Ω= Ts
tan 2
 Ω Ts
ω = 2tan ( − 1) 2
…..(2)

Now for different values of ΩT s ; the graph of ΩT s versus ω is as shown in below Fig.

Chap 2 / IIR Filters

2. A digital filter has following transfer function. Identify the type of filter
and justify it.
1
H(z) = ( 1 + 0.9z ( − 1 )

Appeared in exams: 2 times

Putz = e j ω
b
H(ω) =
1 − 0.9e − j ω

1
H(ω) = ( ( 1 − 0.9cos w ) − j0.9sinw

w H(w)
0 0.52
0.25π 0.55
0.5π 0.74
0.75π 1.28
π 10
From pass band it is observed that the given filter is “High pass filter”

Chap 2 / IIR Filters

3. Compare IIM and BLT
Appeared in exams: 2 times

Sr.
IIM BLT
No.
Poles are transferred by using the equation Poles are transferred by using the equation s=
1 1 1 2(z−1)
→
( S − Pk ) ( 1 − e PkTsz − 1 Ts ( z + 1 )
2 Mapping is many to one Mapping is one to many
3 Aliasing effect is present Aliasing effect is not present
It is not suitable to design high pass filter and band reset High pass filter and band reset filter can be
4
filter designed
5 Only poles of the system can be mapped Poles as well as zeros can be mapped
6 No frequency warping effect Frequency warping effect is present

Chap 2 / Mapping of S-plane to Z-plane

4. Show the mapping from s-plane to z-plane using Impulse Invariant
method and explain its limitation. Using this method determine H(z).
Assume T= 0.1 sec.
3
H(s) = ( (s+2) (s+3) )
Appeared in exams: 4 times

Limitation:
π −π
We know that Ω is analog frequency and its range is from Ts
to Ts
. While the digital frequency ωvaries
π −π
from -π to π. That total means from Ts
to Ts
.ω maps from -π to π. Let K be any integer. Then we can write
π π
the general range of Ω as (K − 1) T to (K + 1) T ; But for this range also , ω maps from -π to π Thus mapping
s s
from analog from analog frequency Ω to digital frequency ωis many to one. This mapping is not one to one.
Analog filters are not band limited so there will be aliasing due to the sampling process. Because of this
aliasing, the frequency response of resulting digital filter will not be identical to the original frequency
response of analog filter.
The change in the value of sampling time (T s) has no effect on the amount of aliasing

Now,
3
H(s) = ( (s+2) (s+3) )
………………(i)

By using partial fraction,

3 A B
( (s+2) (s+3) )
= (S+2)
+ ( S + 3 ) ………………(ii)

3 = A(s + 3) + B(s + 2)

Put s=-3; B=-3

Put s=-2; A=3

Put the value of A and B in equation (i)

3 3
H(s) = (s+2)
− (s+3)

We have transformation equation

1 1
→
( s − Pk ) ( 1 − e P kT SZ − 1
 1 1 1
→ − 2 ( 0.1 ) −1 =
(s+2 1−e Z ( 1 − 0.818Z − 1

1 1 1
→ − 3 ( 0.1 ) −1 =
(s+3) (1−e Z ( 1 − 0.740Z − 1

3Z 3Z
H(z) = ( Z − 0.818 )
− ( Z − 0.740 )

Chap 2 / Mapping of S-plane to Z-plane

( s + 0.1 )
5. An analog filter has transfer function H(s) = Deterine the
( s + 0.1 ) 2 + 16
transfer function of digital filter using bilinear transformation. The digital
filter should have specification
π
ωr = 2

( s + 0.1 ) ( s + 0.1 )
H(s) = H(s) = From above equation we can say that, Ω=4. And The value of ω r is
( ( s + 0.1 ) 2 + 16 ) ( ( s + 0.1 ) 2 + ( 4 ) 2 )
π 2 ω 2 π 2 π
given as ω r = 2 Now we know that Ω = Ts
tan( 2 ) ∴ 4 = Ts
tan( 4 ) ∴ T s = 4 tan( 4 ) ∴ T s = 0.5sec Using
2 ( (z−1)
bilinear transformation H(z) can be obtained by putting , s = Ts ( z + 1 ) )
in the equation of H(s)
(z−1) ( 4z − 4 )
(4( ( (z+1)
) ) + 0.1 ) ( (z+1)
+ 0.1 ) ( ( 4.1z − 3.9 ) ( z + 1 ) )
∴ H(z) = (z−1) ∴ H(z) = ( 4z − 4 + 0.1z + 0.1 ) ∴ H(z) =
( [4( 2
) + 0.1 ] + 16 ) ([ 2
] + 16 ) ( 16.81z 2 − 31.98z + 15.21 + 16 ( z + 1 ) 2 )
(z+1) (z+1)
( 4.1z 2 + 4.1z − 3.9z − 3.9 ) ( 4.1z 2 + 0.2z − 3.9 )
∴ H(z) = ∴ H(z) =
( 32.81z 2 + 0.02z + 31.21 ) ( 32.81z 2 + 0.02z + 31.21 )

Chap 2 / Design of IIR digital filters

6. Design a digital Butterworth IIR filter that satisfies the following
constraint using BLT. Assume T= 0.1sec
0.6≤|H(ω) |≤1 ; 0≤ω≤0.35π

|H(ω) |≤0.1 ; 0.7π≤ω≤π

Appeared in exams: 3 times

Step-1: Identification of filters specification

A p = 0.6; A p = 0.1; ω p = 0.35π; ω s = 0.7π; T = 0.1sec

Now,

2 ωp
Ωp = T
tan( 2
) = 12.25rad / sec

2 ωs
Ωs = T
tan( 2
) = 39.25rad / sec
Step-2: Calculation of order of filter

The order of filter is given by

1
−1
1 As 2
2
log [ 1
]
−1
Ap 2
N> Ωs
log ( Ωp
)

N≥1.72≅2

Step-3: Calculation of cut off frequency

Ωp
Ωc = 1 1
( 2
) 2N
Ap − 1

Ω c = 10.60rad / sec

Step-4: Calculation of poles

π
P k = Ω ce j ( N + 2k + 1 ) 2N

when k=0;

∴ P o = − 7.49 + j7.49

when k=1;

∴ P 1 = − 7.49 − j7.49

Step-5: Calculation of Transfer function H(s)

( Ω c )N
H(s) = ( ( s − Po ) ( s − P1 ) )

( 10.60 ) 2
= ( ( s + 749 − j7.49 ) ( s + 7.49 + j7.49 ) )

112.36
∴ H(s) =
( ( s + 7.49 ) 2 + ( 7.49 ) 2 )

Conversion of analog Transfer function to digital Transfer function:

2 (z−1)
H(z) = H(s) ( s = T (z+1)

112.36
H(z) = ( 20 ( ( z − 1 ) + (7.49) 2 + (7.49) 2
(z+1)

Chap 2 / Design of IIR digital filters

7. The transfer function of digital causal system is given as follows:
 1−z −1
H(z) =
( 1 − 0.2z − 1 − 0.15z − 2 )

1.Find the difference equation

2.Draw Direct Form-I and Direct Form-II realization structure.

3.Draw cascade form, parallel form realization.

Appeared in exams: 2 times

The difference can be obtained by taking IZT if H(z)

(Y(z) )
H(z) = (X(z) )

(Y(z) ) (1−z −1)

=
(X(z) ) ( 1 − 0.2z − 1 − 0.15z − 2 )

Y(z)(1 − 0.2z − 1 − 0.15z − 2) = X(z)(1 − z − 1)

Y(z) − 0.2z − 1Y(z) − 0.15z − 2Y(z) = X(z) − z − 1X(z)

Y(z) = X(z) − z − 1X(z) + 0.2z − 1Y(z) + 0.15z − 2Y(z)

Taking inverse Z transform

∴ y(n) = x(n) − x(n − 1) + 0.2y(n − 1) + 0.15y(n − 2)

ii) Direct form-I and Direct form-II realization structure:

iii) Cascade form, parallel form realization cascade form:

1−z −1
H(z) =
( 1 − 0.5z − 1 ) ( 1 + 0.3z − 1 )

In cascade form

H(z) = H 1(z). H 2(z)

Here

(1−z −1)
H 1(z) =
( 1 − 0.5z − 1 )

1
H 2(z) =
( 1 + 0.3z − 1 )
Parallel Form:-

(1−z −1) (1−z −1)

H(z) = −1 −2 =
( 1 − 0.2z − 0.5z ) ( ( 1 − 0.5z − 1 ) ( 1 + 0.3z − 1 ) )

(z−1)
( z
) (z−1)z
∴ H(z) = ( ( z − 0.5 ) ( z + 0.3 ) ) = ( ( z − 0.5 ) ( z + 0.3 ) )
(
z2 )

H(z) (z−1)
∴ z
= ( ( z − 0.5 ) ( z + 0.3 ) )

Now, By partial fraction

(z−1) A B
( ( z − 0.5 ) ( z + 0.3 ) )
= ( ( z − 0.5 ) )
+ ( ( z + 0.3 ) )

∴ z − 1 = A(z + 0.3) + B(z − 0.5)

Put z=0.5 ; A=-0.625

Put z=-0.3 ; B=1.625

Now,
H(z) ( − 0.625 ) 1.625
z
= ( z − 0.5 )
+ ( z + 0.3 )

( − 0.625 ) 1.625
H(z) = ( +
( 1 − 0.5z − 1 ) ) ( 1 + 0.3z ( − 1 ) )

Chap 2 / Design of IIR digital filters

8. Sketch the frequency response and identify the following filter based on
their passband.
−1 (z −1−a)
i)h(n) = 1, ii)H(z) =
2 ( 1 − az − 1 )

Solution:
−1
h(n) = 1, 2

Taking Z transform
1
H(z) = ∑ n = 0h(n)z − n

1
= 1 − 2z −1

PutZ = e jw
1
H(w) = 1 − 2 e − jw

1
= 1 − 2 [cosw − jsinw]

1 w
H(w) = 1 − 2 cosw − jsin 2

w H(w)
0 0.5
π
0.736
4
π
1.11
2
3π
1.39
4
π 1.5

From pass band it is observed that given filter is High pass filter.

(z −1−a)
ii)H(z) =
( 1 − az − 1 )
Putz = e jw
( e − jw − a )
H(w) =
( 1 − ae − jw )

( ( cos w − jsinw ) − a )
= ( 1 − a ( cos w − jsinw ) )

( ( cos w − a ) − jsin w )
= ( ( − acos w ) + jasin w )

Assuming a=1
( ( cos ω − 1 ) − jsin w )
H(w) = ( ( 1 − cos w ) + jsinw )

w H(w)
0.1π -1
0.2π -1
0.3π -1
0.4π -1
0.5π -1

All the value of w gives same response, Hence it is all pass filter.

Chap 2 / Design of IIR digital filters

9. Design digital low pass IIR Butterworth filter for the following
specification:
Pass band ripple: ≤1dB Pass band Edge: 4KHz

Stop band Attenuation: ≥40dB; Stop band Edge: 6KHz

Sampling Rate: 24KHz. Use bilinear transformation

Appeared in exams: Once

Step-1: Identification of specification of filter

A p = 1dB; A s = 40dB; F PB = 4KHz; F SB = 6KHz; F S = 24KHz; T S = 41.66µsec

Now, we have
 F SB
F SB = FS
= 0.25

ω S = 2πF SB = 1.57

And
F PB
F PB = FS
= 0.166

ω P = 2πF PB = 1.043

Now,

( − A) p(dB) = 20log(A p)

− 1 = 20log(A p)

A p = 0.891

And

( − A) s(dB) = 20log(A s)

− 40 = 20log(A s)

∴ A s = 0.01

2 ωP
ΩP = T
tan( 2
) = 27.583Krad / sec

2 ωS
ΩS = T
tan( 2
) = 47.969Krad / sec

Step-2: Order of filter

Order of filter is given by

1
N ≥ 2 (19.08)

∴ N = 9.54 ≅ 10

Step-3: Cut of frequency

ΩP
Ωc = 1
( 1
( A 2P ) − 1 ) 2N

Ωc = 29.506Krad / sec

Step-4: Calculation of poles

π

P k = Ωce ( j ( N + 2k + 1 ) 2N
When k=0

∴ P o = − 4615 + j29142

Whenk = 1; P 1 = − 13395 + j26290

Whenk = 2; P 2 = − 20863 + j20863

When = 3; P 3 = − 26290 + j13395

When = 4; P 4 = − 29142 + j4615

When = 5; P 5 = − 29142 − j4615

When = 6; P 6 = − 26290 − j13395

Whenk = 7; P 7 = − 20863 − j20863

When = 8; P 8 = − 13395 − j26290

When = 9; P 9 = − 4615 − j29142

Step-5: Calculation of Transfer Function

( Ω c )N
H(s) = ( ( S − P0 ) ( S − P1 ) ( S − P2 ) ( S − P3 ) ( S − P4 ) ( S − P5 ) ( S − P6 ) ( S − P7 ) ( S − P8 ) ( S − P9 ) )

Chap 2 / Design of IIR digital filters

10. Design a digital Butterworthlow pass filter that satisfies the following
constraint using IIM. Assume T=1 sec
0.707≤|H(w) |≤1 ; for 0<ω<0.3π

|H(w) |≤0.2 ; for 0.75π<ω<π

Appeared in exams: Once

Step-1: Identify the filters specification

A p = 0.707; A s = 0.2; ω p = 0.3π; ω s = 0.75π; T = 1sec

Now,
ωp
Ωp = T
= 0.942rad / sec

ωs
Ωs = T
= 39.25rad / sec

Step-2: Calculation of order of filter

The order of filter is given by

 2
( 1 / ( As ) − 1 )
log [ 2
]
1 ( 1 / ( Ap ) − 1 )
N≥ 2 Ωs
log ( Ωp

N ≥ 1.73 ≅ 2

Step-3: Calculation of cut off frequency

Ωp
Ωc = 1 1
( 2
) 2N
( Ap − 1 )

Ω c = 0.941rad / sec

Step-4: Calculation of poles

π
P k = Ω c = e j ( N + 2k + 1 2N

Now,

when k=0;

∴ P o = − 0.665 + j0.665

when k=1;

∴ P 1 = − 0.665 − j0.665

Step-5: Calculation of Transfer function H(s)

( Ω c )N
H(s) = ( ( s − Po ) ( s − P1 ) )

0.885
= ( ( s + 0.665 − j0.665 ) ( s + 0.665 + j0.665 ) )

0.885
∴ H(s) =
( ( s + 0.665 ) 2 + ( 0.665 ) 2 )

Step-6: Conversion of analog Transfer function to digital Transfer function

We know that,

b ( e − aTs [ sin bTs ] z − 1 )

2 2 = − aTs
( (s+a ) +b ) ( 1 − 2e [ cos bTs ] z − 1 + e − 2aTsz − 2 )

( e − 0.665Ts [ sin 0.665Ts ] z − 1 )

∴ H(z) = − 0.665Ts
( 1 − 2e [ cos 0.665Ts ] z − 1 + e − 1.33Tsz − 2 )

Chap 2 / Frequency transformation

11. Frequency transformation on IIR filters.
There are two methods of Frequency Transformation in Infinite Impulse Response (IIR) filters.
Impulse Invariance Technique (IIT):

In IIT the impulse response of the CT system is sampled to produce the impulse response of the DT system.

∞
[
Frequency Response of DT system H(ω) = F s ∑ k = − ∞H a (ω − 2πk)F s ]
The frequency response H(ω) of the DT system is a sum of shifted copies of the frequency response H a(ω) of the
CT system. If the CT system is band-limited to a frequency less than the Nyquist frequency F s of the sampling,
then H(ω) will be approximately cqual to H a(ω) for frequencies below the F s.

Ω
In IIT, the frequency of Digital filter ω = ΩT s = Fs
, where Ω is the analog frequency.,

The relation between CT and DT frequency is linear. So, except for aliasing, the shape of the frequency response is
preserved. The mapping of points from the s-plane to the z − plane is given by the relation z = e sT. In IIT there is
many to one mapping of poles from s-plane to z-plane.

Bilinear Transformation Technique (BLT):

BLT is a conformal mapping which converts imaginary axis of s-plane into unit circle in z-plane. It is one to one
mapping between s-plane and z − plane. There is no aliasing effect in BLT.

In BLT , the relation between the analog frequency (Ω) and corresponding digital frequency (ω) is
2 ω ΩT
Ω= T
tan 2 or ω = 2tan − 1
( )
2

However "Tan inverse" being non-linear function causes nonlinear compression of the frequency axis. This non-
linear mapping which introduces a distortion in the frequency axis, which is called Frequency Warping.

So, the design of discrete-time filters using the BLT is useful only when this distortion can be tolerated or
compensated for, as in the case of filters that approximate ideal piecewise constant magnitude response
characteristics.

Due to Frequency Warping, phase response of analog filter cannot be preserved but magnitude response can be
preserved by pre-warping analog frequencies.

Chap 2 / Frequency transformation

12. Explain frequency sampling method of designing FIR filter?
The frequency sampling method is use to design recursive and non-recursive FIR filters for both standard
frequency selective filters and with arbitrary frequency response. The main idea of the frequency sampling design
method is that a desired frequency response can be approximated by sampling it at N evenly spaced points and
then obtaining N-point filter response.

A continuous frequency response is then calculated as an interpolation of the sampled frequency response. The
approximation error would then be exactly zero at the sampling frequencies and would be finite in frequencies
between them. The smoother the frequency response being approximated, the smaller will be the error of
interpolation between the sample points.

There are two distinct types of Non-Recursive Frequency Sampling method of FIR filter design, depending on
where the initial frequency sample occurred. The type 1 designs have the initial point at ω = 0 , whereas the type 2
 1 π
designs have the initial point at f = 2N
or ω = N

Procedure for Type-1 Design:

1) Choose the desired frequency response H d(ω)

2πk
2) Sample H d(ω) at N -points by taking ω = ω k = N
where k = 0, 1, 2, 3, …. . . (N − 1), generate the sequence
H(k). To obtain a good approximation of the desired frequency response, a sufficiently large number of the
frequency samples should be taken. H(k) = H d(ω) | ω = 2πk / N
for k = 0, 1, …(N − 1)

3) The N-point inverse DFT of the sequence H(k) gives the impulse response of the filter h( n). For practical
realization of the filter, samples of impulse responsed should be real. This can happen if all the complex terms
appear in conjugate pairs.

Desired filter coefficients

1 N−1
h(n) = InvDFT{H(k)} = N
∑ k = 0 H(k)e j2πkn / N

For linear phase filters, with positive symmetrical impulse response,

{h(0) + 2 ∑ [H(k)e ]},

1 UL
h(n) = Re j2πkn / N
N k=1

when
N−1 N
N is odd UL = 2
and when N is even UL = 2
−1

4) Take z-transform of the impulse response h (n) to get the filter transfer function, H(z)
N−1
H(z) = ∑ n = 0 h(n) ⋅ z − n

Procedure for Type-2 Design:

(Same steps as above except step 2)

2π
2) Sample H d(ω) at N-points by taking ω = ω k = 2N
(2k + 1) where k = 0, 1, 2, 3, …(N − 1) generate the sequence
H(z)

H(k) = H d(ω) | ω = πk ( 2k + 1 ) / N
for k = 0, 1, …N − 1

Type 2 frequency samples give additional flexibility in the design method to specify the desired frequency
response at a second possible set of frequencies.

Advantage

Unlike the window method, this technique can be used for any given magnitude response.
This method is useful for the design of non-prototype filters where the desired magnitude response can take
any irregular shape.
Major advantage of Frequency sampling method lies in the efficient frequency sampling structure, which is
obtained when most of the frequency samples are zero.
Disadvantage

One disadvantage with this method is that the frequency response obtained by interpolation is equal to the desired
frequency response only at the sampled points. At the other points, there will be a finite error present.

Chap 2 / Digital filter

13. State advantages and disadvantages of digital filter.
Appeared in exams: 2 times

Advantages of digital filter:

1. Many input signals can be filtered by one digital filter without replacing the hardware.
2. Digital filter have characteristic like linear phase response. Such characteristics are not possible to obtain in
case of analog filters.
3. The performance of digital filters does not vary with environmental parameter.
4. Unlike analog filters; the digital filters are portable.
5. From unit to unit, the performance of digital filters is repeatable.
6. Digital filters are highly flexible.

Disadvantages of digital filter:

1.Speed Limitation:

In the case of digital filters. ADC and DAC are used, so the speed of digital filter depends on the conversion time
of ADC and the settling time of DAC. Similarly the speed of operation of digital filter depends on the speed of
processor. Thus the band width of the input signal processed is limited by ADC and DAC.

2.Finite Wordlength Effect:

The accuracy of digital filter depends on the word length should be long enough to obtain the required accuracy.
The digital filter also affected by the ADC noise, resulting from the quantization of the continuous signal.

3.Long Design and development time:

An initial design and development time for digital hardware is more than analog filters.
 Chapter 3
FIR Digital Filters
Content

Characteristics of FIR digital filters

Design of FIR filters
Compare IIR - FIR
Digital Resonator
Notch filter

Chap 3 / Characteristics of FIR digital filters

1. Determine the zeros of the following FIR systems and Identify whether
the following system is minimum phase, maximum phase, mixed phase. Also
comment on stability.
1.H1(z) = 6 + z ( − 1) + 6z − 2

2.H2(z) = 1 − z − 1 − 6z − 2
5 3
3.H3(z) = 1 − 2 z − 1 − 2 z − 2

5 2
4.H4(z) = 1 − 2 z − 1 − 3 z − 2

Appeared in exams: Once

i) H(z) = 6 + z − 1 + 6z − 2
1 6
=6+ z
+ z2

( 6z 2 + z + 6 )
=
z2

Poles: 0, 0

Zeros:-0.08±0.99j

z 1 = 0.99∠85.38°

z 2 = 0.99∠85.38°

Both the zeros lie inside the unit circle, poles are also at origin. Hence , the system is “Stable Minimum phase”

ii)H 2(z) = 1 − z − 1 − 6z − 2
 1 6
=1− −
z z2

( z2 − z − 6 )
= z2

Poles: 0, 0 Zeros: 3, -2

Both the zeros lie outside the unit circle, poles are at origin. Hence, the system is “Stable Maximum Phase”
5 3
iii)H 3(z) = 1 − −1 −
2z 2z − 2

5 3
=1− −
2z 2z 2

5 3
z 2 − 2z − 2
=
z2

Poles: 0, 0 Zero: 3, -0.5

One of the zero is inside the unit circle and other is outside the unit circle and poles are at origin. Hence the system
is “Stable Mixed Phase”
5 2
iv) H 4(z) = 1 − −1 −
2z 3z − 2

5 2
=1− 2z
− 3z

5 2
z 2 − 2z − 3
=
z2

Poles: 0, 0 Zeros: 2.74, -0.24

One of the zero is inside the unit circle and other one is outside the unit circle and poles are at origin. Hence, the
system is “Stable Mixed Phase”.

Chap 3 / Characteristics of FIR digital filters

2. Explain phase delay and group delay.
Appeared in exams: 2 times

For FIR filter,

M−1
H(z) = ∑ n = 0 h(n)z − n

To obtain magnitude and phase response put

z = ej ω
M−1
∴ H(e j ω ) ∑ n = 0 h(n)z − j ω n

Here phase response is given by:

 ( Im [ H ( e j ω ) ] )
ϕ(ω) = tan − 1
( Re [ H ( e j ω ) ] )

The group delay is the delayed response of filter as a function of frequency ω.

The phase delay (T_p) and group delay (T_g)are given by,
(−ϕ(ω))
Tp = ω
and

( −dϕ ( ω ) )
Tg = dω

(M−1)
The parameter T is constant phase delay parameter and it is given by 2

If the phase delay and group delay are constant then such filters are called as linear phase filters. The condition for
linear phase in terms of delay parameter is:

ϕ(ω)=-ωT

Similarly in terms of filter length, condition for Linear phase is

h(n)=h(M-1-n)

If only constant group delay is considered then the Linear phase condition is,

h(n)=-h(M-1-n)

Chap 3 / Characteristics of FIR digital filters

3. Write a note on frequency sampling realization of FIR filter.
Appeared in exams: Once

We know that,

DFT of h(n) is given by:

− j2 π kn
N−1
H(k) = ∑ n = 0 h(n)e N …………………(1)

From the equation of IDFT we have,

1 j2 π kn
N−1
h(n) = N
∑ k = 0 H(k)e N …………………(2)

By the definition of Z transform we have,

$H(z)=∑_{n=0}^{N-1}h(n) z^{-n}$…………………(3)

Put the value of h(n) from equation (2) to equation (3), we get
j2 π kn
N−1 1 N−1
H(z) = ∑ n = 0 N ∑ k = 0 H(k)e N z −n
 1 j2 π kn
N−1 N−1
H(z) = N
∑ k = 0 H(k) ∑ n = 0 e N z −n

Now by sum of finite GP series we have;

We know that, e j2 π k = 1

(1−z −N) N−1 H(k)

H(z) = N
∑k = 0 [ j2 π k ]
(1−e N z −1)

(1−z −N) H(0) H(1) H(2) H(N−1)

H(z) = N
[ −1 + j2 π + j4 π + ⋯……………… + ( j2 π ( N − 1 ) ) z − 1)]
(1−z )
(1−e N z −1) (1−e N z −1) (1−e N

Frequency Sampling Realization;

Chap 3 / Design of FIR filters

4. A digital filter is describe by the following difference equation:
1
y(n) = 0.9y(n − 1) + bx(n) i) Determine b such that |H(0) |=1 ii) Determine the frequency at which |H(ω) |= √ 2

iii) Identify the filter type based on passband.

Appeared in exams: 2 times

We have,

y(n) = 0.9y(n − 1) + bx(n)

Taking z transform

Y(z) = 0.9Y(z)z − 1 + bX(z)

Y(z) − 9z − 1Y(z) = bX(z)

(Y(z) ) b
H(z) = =
(X(z) ) ( 1 − 0.9z − 1
Putz = e j ω
b
H(ω) =
( 1 − 0.9e − j ω )

i) Put ω=0
b
H(0) = ( 1 − 0.9 )
=1

b=0.1

ii) We have,
0.1
H(ω) =
( 1 − 0.9e − j ω )

0.1
= ( 1 − 0.9 ( cos ω − jsin ω )

0.1
H(ω) = ( ( 1 − 0.9cos ω ) )

By plotting magnitude response, it is observe that,

π
ω= 2

iii) Identify the type by pass band

Chap 3 / Design of FIR filters

5. Design digital FIR filter for following specification. Use hamming
window and assume M = 7.
Step-1: Identification of specification of filter

H d(e j ω ) = | H(ω) | e − j ∝ ω ; − ω c ≤ ω ≤ ω c

But N=M=7

We know that,
(N−1)
∝= 2
=3

Therefore,

H d(e j ω ) = e − j3 ω ; 0.4π | ω | ≤ 0.6π

=0 ; otherwise

Window Type: Hamming Window

Step-2: Calculation of IFT of H(ω)

1 ω )
h d(n) = ∫ c H (ω)e j ω ndω
2π − ω c d

1 − 0.4 π ) 1 0.6 π
= ∫
2 π − 0.6 π
e j ω ne − j3 ω dω + ∫
2 π 0.4 π
e − j3 ω e j ω ndω

1 − 0.4 π 1 0.6 π
= ∫
2 π − 0.6 π
e j ω ( n − 3 ) dω + ∫
2 π 0.4 π
ej ( n − 3 ) ω d

1 ej ( n − 3 ) ω − 0.4 π 1 e j ( n − 3 ) ω 0.6 π
= [ ]
2 π ( j ( n − 3 ) − 0.6 π
+ [ ]
2 π ( j ( n − 3 ) 0.4 π

1 ( e − j0.4 π ( n − 3 ) − e − j0.6 π ( n − 3 ) ) ( e j0.6 π ( n − 3 ) − e j0.4 π ( n − 3 ) )

= 2π
[ (j(n−3)
+ (j(n−3)
]

1
= π (n−3)
[sin[0.6π(n − 3)] − sin[0.4π(n − 3)]]

[ 0.4 π ( n − 3 ) ] )
h d(n) = (sin[0.6π(n − 3)] − π (n−3)

h d(0) = 0 = h d(6) {by linear phase property}

h d(1) = − 0.187 = h d(5) {by linear phase property}

h d(2) = 0 = h d(6) {by linear phase property}

By L-Hospital’s Rule

h d(n) = 0.2

Step-3: Calculation of window Response W(n)

Window Response for hamming is given by

2πn
W(n) = 0.54 − 0.46cos( ( N − 1 ) )
W(0)=0.08=W(6)

W(1)=0.31=W(5)

W(2)=0.77=W(4)

W(3)=1

Step-4: Calculate the impulse Response of filter

Impulse Response is given by

h(n) = h d(n) ∗ W(n)

h(0)=0=h(6)

h(1)=-0.057=h(5)

h(2)=0=h(4)

h(3)=0.2

Step-5: Calculation of filter Transfer function H(z)

H(z) = ∑ N −1
n=0
h(n)z − n

6
∴ H(z) = ∑ n = h(n)z − n )

∴ H(z) = 0 + ( − 0.057)z − 1 + 0 + 0.2z − 3 + 0 + (0.057)z − 5 + 0

∴ H(z) = 0 − 0.057z − 1 + 0 + 0.2z − 3 + 0 − .057z − 5 + 0

Step-6: Draw the Realization structure

Chap 3 / Design of FIR filters

6. Design a FIR filter using window method for following specification. Use
hamming window of length.
 3π
H(e j ω ) = e − j3 ω ; 0 ≤ | ω | ≤ 4

=0 ; otherwise

Appeared in exams: 2 times

Step-1: Identify the specification of filter

−3π 3π
N = 7 ∝= 3ω c = 4
≤ω≤ 4

Window Type: Hamming

Step-2: Calculate the Inverse Fourier Transform of H(ω)

1 ω
h d(n) = ∫ c H (ω)e j ω ndω
2π − ω c d

3π
1
= ∫4
2π ( −3π )
e − j3 ω e j ω ndω
4

3π
1
= ∫4
2π ( −3π )
e j ( n − 3 ) ω dω
4

3π
−j(n−3) )
1 3π 4
= 2π
[(e j ( n − 3 ) 4 −e (j(n−3) ]

3π
−j(n−3) ))
1 3π 4
= π (n−3)
[(e j ( n − 3 ) 4 −e 2j ]

3π
sin 4
(n−3)
h d(n) = ( π (n−3)

h d(0) = 0.075 = h d(6) {by linear phase property}

h d(1) = − 0.159 = h d(5) {by linear phase property}

h d(2) = 0.225 = h d(4) {by linear phase property}

By L-Hospital’s Rule

h d(3) = 0.75

Step-3: Calculation of window Response W(n)

2πn
W(n) = 0.54 − 0.46cos( ( N − 1 )

W(0)=0.08=W(6)

W(1)=0.31=W(5)

W(2)=0.77=W(4)
W(3)=1

Step-4: Calculation of window Response h(n)

h(0)=0.006=W(6)

h(1)=-0.049=W(5)

h(2)=0.173=W(4)

h(3)=0.75

Step-5: Calculate of filter Transfer Function

N−1
H(z) = ∑ n = 0 h(n)z − n

H(z) = 0.006 − 0.049z − 1 + 0.173z − 2 + 0.75z − 3 + 0.173z − 4 − 0.049z − 5 + 0.006z − 6

Step-6: Realization Structure

Chap 3 / Design of FIR filters

7. Design linear phase FIR low pass filter of length T and cut of frequency 1
rad/sec using hamming window.
Appeared in exams: 3 times

Step-1: Identify the specification of filter

N = 7 ∝= 3ω c = − 1 < ω < 1

Window Type: Hamming

Step-2: Calculate the Inverse Fourier Transform of H(ω)

1
h d(n) = ∫ (ω H (ω)e j ω ndω
2π − ω c c d

1 1
= ∫ e − j3 ω e j ω nd
2π −1
 1 1
= ∫ e j ( n − 3 ) ω dω
2π −1

1 j(n−3) ω )
1
= 2π
[e (j(n−3) ) ] −1

1 −j(n−3) ) )
= ( π (n−3) )
[(e j ( n − 3 ) − e 2j ]

sin ( n − 3 )
h d(n) = ( π (n−3) )

h d(0) = 0.014 = h d(6) {by linear phase property}

h d(1) = 0.144 = h d(5) {by linear phase property}

h d(2) = 0.267 = h d(4) {by linear phase property}

By L-Hospital’s Rule

h d(3) = 0.318

Step-3: Calculation of window response W(n)

2πn
W(n) = 0.54 − 0.46cos( ( N − 1 ) )

W(0)=0.08=ω(6)

W(1)=0.31=ω(5)

W(2)=0.77=ω(4)

W(3)=1

Step-4: Calculate impulse Response of filter

h(n) = h d(n) ∗ W(n)

h(0)=0.0012=h(6)

h(1)=0.044=h(5)

h(2)=0.205=h(4)

h(3)=0.318

Step-5: Calculation of filter Transfer function

N−1
H(z) = ∑ n = 0 h(n)z − n

H(z) = 0.00112 + 0.044z − 1 + 0.205z − 2 + 0.318z − 3 + 0.205z − 4 + 0.044z − 5 + 0.00112z − 6

Step-6: Realization Structure

Chap 3 / Design of FIR filters
8. Design FIR filter using frequency sampling technique for the following
specification:
π
H(e j ω ) = e j3 ω ; 0 ≤ | ω | ≤ 2
H(e j ω ) = 0; otherwise

Appeared in exams: Once

We know that
2πk
H(k) = H(ω) ( ω = N
) ………(1)

H(e j ω ) = e − j3 ω

Here ∝=3
(N−1)
And ∝= 2

∴N=7

Now, By using equation (1)

( − j6 π k ) 2πk π
H(k) = e 7 ;0 ≤ 7
≤ 2

π 2πk
= 0; 2
≤ 7
≤π

( − j6 π k ) 7π
∴ H(k) = e 7 ;0 ≤ k ≤ 4π

7π 7π
= 0; 4π
≤k≤ 2π

( − j6 π k )
∴ H(k) = e 7 ;0 ≤ k ≤ 2

= 0; 2 ≤ k ≤ 4
Now , Here K is vary from 0 to 2

When k=0

∴ H(0) = 1

When k=1
( − j6 π )
∴ H(1) = e 7

When K=2
( − j12 π )
∴ H(2) = e 7

According to transfer function equation of Frequency Sampling Realization:

(1−z −7) H(0) H(1) H(2) )
H(z) = [ ][ + +( ]
7 (1−z −1 j2 π j6 π
(1−e 7 z −1 (1−e 7 z −1)

Chap 3 / Compare IIR - FIR

9. Compare IIR and FIR filters.
Appeared in exams: 5 times

Sr No. Characteristic IIR FIR

1 Stability Depends upon system design Guaranteed
2 Linear Phase No Guaranteed
3 Required Hardware Memory Least Most
4 Availability of design software Good Very Good
5 Support adaptive Filtering Yes Yes

Chap 3 / Digital Resonator

10. Write a short note on digital resonator.
Digital Resonators:

It is a two pole bandpass filter, having a pair of complex conjugate poles located near the unit circle.

The frequency magnitude response resonates near the location of poles. It is having large magnitude response.

The resonant frequency of this filter is determined by angular position of poles.

We can select two zeros & two poles system to represent digital resonator

Select two zeros at the origin.

Select one zero sat +1 and -1

( b oz 2 )
H(z) =
( z − re ( j ω 0 ) ) ( z − re ( − j ω 0 ) )

Here b_ois the gain of system.

Chap 3 / Notch filter

11. Write a short note on Notch filter
No answer yet!

Chap 3 / Notch filter

12. Write a short note on Notch filter
Notch filter

Notch filter are used to remove the specific frequency components from the total frequency response.

The notch filter contains ideally nulls in the frequency response characteristics.

The nulls are obtained at the frequencies ω 0andω 1. To obtain null at ω 0a pair of complex conjugate zero is
introduced on unit circle.

Z 1, 2 = e ± j ω 0

H(z) = b 0(z 1z − 1)(1 − z 2z − 1)

H(z) = b 0(1 − e j ω 0z − 1)(1 − e − j ω 0z − 1)

Limitation:

Bandwidth is relatively large. So that frequency components round the nulls are also affected.
 Chapter 4
Finite Word length effects in Digital Filters
Content

Quantization in FIR
Truncation and Rounding
Cycle oscillations
Floating Point realization
Finite word length effects
Scaling FIR

Chap 4 / Quantization in FIR

1. Explain the effects of coefficient quantization in FIR filters.
Appeared in exams: 2 times

Digital FIR filters are designed such that they have linear phase characteristics in the pass band, if FIR filters
are realized using Direct Form realization in linear phase is maintained even when quantization of the filter is
done.

Quantization does not affect the phase characteristics of FIR filter, but it affects the magnitude response.

To avoid this affect, the cascade form realization should be used and 12 to 14 beats should be used to
represent the coefficients.

Similarly the number of bits for coefficient must be increased to maintain the same and in the frequency
response of characteristics of the filter.

Let us consider each filter coefficient is rounded to (b + 1) bits, then the maximum error in the
coefficient value is bounded as:

− 2 ( b + 1 ) < e h(n) < 2 − ( b + 1 )

The error in the frequency response is given by,

M−1
E M(ω) = ∑ n = 0 e h(n)e − j ω n

Chap 4 / Truncation and Rounding

2. Effect of arithmetic round-off errors.
The difference equation of the FIR filter is given by,
N−1
y(n) = ∑ m = 0h(m)x(n − m)

where, each variable is represented by a fixed number of bits. Typically the input and output samples, x(n-m) and
y(n), are each represented by 12 bit in the coefficients by 16 bits and two's complement format.
• Output of the filter is obtained as a sum of product of h(m) and x(n-m). After each multiplication the product
contains more bit then either h(m) or x(n-m).

• For example, if 12 bit input is multiplied by 16 bit coefficient, result is 28 bit long and will need to be quantize
back to 16 bits before it can be stored in a memory or to 12 bits before it can be sent as an output of the DAC.

• This leads to error was effects are similar to those of ADC noise. The common way to quantize the result of
arithmetic operation is either:

(i) Truncate the result, i.e. to retain the MSB and to discard the LSB.

(ii) To round of the results, i.e. to choose the higher order bits closest to the unrounded results. This is achieved by
adding half an LSB to the results.

• Round-off errors can be minimized by representing all products exactly with the double length register, and then
rounding the results after obtaining the final sum.

Chap 4 / Finite word length effects

3. Short note on finite word length effect in digital filters.
Appeared in exams: 4 times

• Digital Signal Processing the computations like FFT algorithm, ADC and filter designs are associated with
numbers and coefficients.

• These numbers and coefficients are stored in a finite length registers but due to mathematical manipulations
perform with fixed point arithmetic number of errors are present by storing the numbers and coefficients are
required to quantize the different type of number representations are used for this purpose.

• The implementation of digital filters involves the use of finite precision arithmetic. This leads to quantization of
the filter coefficients and the results of the arithmetic operations. Such quantization operations are nonlinear and
cause a filter response substantially different from the response of the underlying infinite-precision model.

• Finite word length of the signals to be processed the finite word length of the filter coefficients does not affect the
linearity of the filter behavior. This effect only amounts to restrictions on the linear filter characteristics, resulting
in discrete grids of pole-zero patterns.

• These effects, which divide into those due to "signal quantization" and those due to "overflow".

Chap 4 / Scaling FIR

4. Short note on dynamic range scaling.
In case of recursive system, a feedback connection is present. So if there is an overflow then it is feedback
and used to come to the next output, where it causes further overflow.

This creates undesired oscillations at the output, hence results in nonlinearity. It becomes difficult to analyze
the digital filter precisely.

To limit this overflow it is required to scale the input signal and unit sample response. This scaling is done
between the input and any internal summing node in the system.

Let us assume that YK denotes the response of the system at kth node, for the input x(n) and let hk(n) be the
impulse response of the system.
 According to the definition of convolution:
∞
y k(n) = ∑ k = − ∞ h k(m). x(n − m)

Taking magnitudes of both sides;

∞
| y k(n) | = | ∑ k = − ∞ h k(m). x(n − m) |

∞
| y k(n) | ≤ A x ∑ m = − ∞ | h k(m) |

1
Ax ≤ ∞
( ∑m = − ∞ | hk ( m ) | )

This is the necessary and sufficient condition to prevent overflow in the system. It means to avoid the overflow,
proper dynamic scaling should be done at that particular node.
 Chapter 5
DSP Processors
Content

Compare DSP and Microprocessor

Need of dsp processor
Addressing modes of dsp processor
Pipelining in DSP Processor
Architecture of DSP
Fixed and Floating point implementation

Chap 5 / Compare DSP and Microprocessor

1. Compare DSP processor and Microprocessor.
Sr
Parameters DSP processor Microprocessor
No.
Instructions are executed in single cycle of Multiple clocks cycles are required for
1 Instruction cycle
the clock execution of one instruction.
Execution of instruction is always
2 Instruction execution Parallel execution is possible.
sequential.
3 Memories Separate data and program memory. No such separate memories are present.
On chip/Off chip Program and Data memories are present on Normally on chip cache memory
4
memories chip extendable off chip. present, main memory is off chip.
Program sequencer and instruction register Program counter take care of flow of
5 Program flow control
take care of program flow execution.
Pipelining is implicate through instruction Queuing is perform explicate by one
6 Pipelining
register and instruction cache. queue register to support pipelining.
Multiple operands can be fetched
7 Operand Fetch Operands are fetched sequentially.
simultaneously.
Address and data bus are not multiplexed.
Address and data bus
8 They are separate on chip as well as off Address and data bus are multiplexed.
multiplexing
chip.
Three separate computational units: ALU,
9 Computational units Only one main unit ALU.
MAC and shifter.
On chip address and Separate address and data bus for program Address and data bus are the two buses
10
data bus and data memory. on the chip
Direct, Indirect, Register, Register
11 Addressing modes Direct and indirect addressing modes. indirect, Immediate addressing mode
etc.
Signal processing, audio processing,
12 Application General Purpose applications.
speech processing and array processing etc

Chap 5 / Need of dsp processor

2. Explain the need of DSP processors.
 DSP, is a specialized microprocessor that has an architecture which is optimized for the fast operational needs
of digital signal processing.
The goal of digital DSP signal processors is usually to measure, filter or compress continuous real-world
analog signals. Most general-purpose microprocessors can also execute digital signal processing
algorithms successfully.
We cannot use a general-purpose microprocessor to process signals very well, Add and subtract
operations are performed quite simply by general-purpose microprocessors in a single or very few clock
cycles. The multiply and divide operations are more complex. A digital multiply operation consists of a
series of shift and add operations. Division, which is more complex.
General-purpose microprocessors are quite slow in performing multiply and divide operations. They will
typically sequentially execute a series of shift, add, and subtract operations from their microcode to
perform a single multiply operation, and may consume many cycles to complete.
The DSP performs multiplication in a single cycle by implementing shift and add operations in parallel.
The circuitry is relatively complex and consumes a considerable number of transistors. The benefit is
very fast multiplication, which is required for processing most digital signals. When general-purpose
DSPs are not fast enough, the signal is either processed using analog circuits (which may have some
drawbacks), or in specialized DSP hardware designed only for that task. This eliminates many of the
benefits of a programmable DSP.
Digital signal processing algorithms typically require a large number of mathematical operations to be
performed quickly and repeatedly on a series of data samples. Signals (perhaps from audio or video
sensors) are constantly converted from analog to digital, manipulated digitally, and then converted back
to analog form.

Chap 5 / Addressing modes of dsp processor

3. Addressing Modes of TMS320C67XX
Appeared in exams: 2 times

DSP processors support various addressing modes for execution of instructions and to access data. The efficient
way of accessing data (signal sample and filter coefficients) can significantly improve implementation
performance, it provides flexible ways to access data helps in writing programs. Data addressing modes enhance
DSP implementation, DSP processors addressing modes are:

• Immediate Addressing Mode

Operand is explicitly known in value, capability to include data as part of the instruction

Instruction Operation
ADD #imm #imm #imm+ A->A

#imm: value represented by imm (fixed number such as filter coefficients is known ahead of time)

A: accumulator register

• Register Addressing Mode

Operand is always in processor register reg, it provides capability to reference data through its register

Instruction Operation
ADD Reg Reg + A->A

Reg: processor register provides operand.

A: accumulator register.
• Direct Addressing Mode

Operand is always in memory location mem, provides capability to reference data by giving its memory location
directly.

Instruction Operation
ADD Mem Mem + A->A

Mem: specified memory location provides operand (e.g., memory could hold input signal value).

A: accumulator register

• Indirect Addressing Mode

Operand memory location is variable, operand address is given by the value of register Addrreg, operand accessed
using pointer Addrreg.

Instruction Operation
ADD *Addrreg *Addrreg + A->A

Addrreg: needs to be loaded with the register location before use.

A: accumulator register.

• Special Addressing Modes

It provides to implement real-time digital signal processing and FFT algorithms

i.Circular Addressing Mode: Circular buffer allows one to handle a continuous stream of incoming data samples;
once the end of the buffer is reached, samples are added to the beginning again.

ii.Bit-Reversed Addressing Mode: Address generation unit can be provided with the capability of providing bit-
reversed indices.

Input Index Output Index

000=0 000=0
001=1 100=4
010=2 010=2
011=3 110=6
100=4 001=1
101=5 101=5
110=6 011=3
111=7 111=7

Chap 5 / Pipelining in DSP Processor

4. Pipelining in DSP Processor
Architectural feature in which an instruction is broken into a number of steps, a separate unit performs each
step at the same time usually working on different stage of data. This can be done by processing number of
instructions simultaneously.
 Generally one complete instruction start from fetching an instruction and ends with execution of instruction.
The total required time period is allocated for different purposes throughout the instructions.

An entire instruction can be divided into number of steps:

1) Fetching instruction from program memory.

2) Decoding of instruction.

3) Operand fetching, which is necessary for execution of instruction.

4) Execution of instruction

5) Save the result.

Now each of the above steps (micro-instruction) can be carried out separately by 5 functional units in CPU.
Suppose each micro instruction requires same amount of execution time.

In a conventional processor, CPU processes only one instruction at a time, so each functional unit is busy only
for one-fifth time period. But in case of processes which use pipelining all these 5 micro-instructions can be
carried out simultaneously in the CPU

Chap 5 / Architecture of DSP

5. Architecture of TMS320C6XX DSP Processor
•TMS 320C67 X is Fixed as well as Floating point processor. Central Processing Unit consists of 8 functional
units. These functional units are divided into two sides A and B. The Architecture of TMS 320C67 X is shown in
figure below.
•Each side contains units M, L, S and D. These units are basically used to perform various operations; certain
instruction can be executed by using more than one unit for example Add instruction. Each side contains sixteen 32
bit registers interaction with CPU is done using these registers.

•Internal buses consists of following:

i. 32-bit program address bus.

ii. 256-bit data program bus.

iii. To load data buses LD1 and LD2.

iv. Two 32-bit store data buses namely ST1 and ST2.

v. A 32-bit Direct Memory Access (DMA) data bus and 32-Bit DMA address bus.

vi. External memory is accessed through a bit 20-Bit address bus and 32-Bit data bus.

The peripheral on C6 X processes are as follows:

1.EDMA (Enhanced Direct Memory Access): It has 16 Programmable channels and RAM space to hold multiple
configurations. It makes the movement of data from one place in memory to the other place without interfering
with CPU operation.

2.Boot Loader: It boots the code from HPI to internal memory. It is basically used to determine what actions the
DSP performs; when the device is reset.

3.McBSP (Multichannel Buffered Serial Port): It provides high speed multi-channel serial communication link.
This port can buffer serial samples in memory automatically with the help of a EDMA controller. It is also having
multichannel capability which is compatible with various networking standards.

4.HPI (Host Port Interference): It allows the host to access internal memory. The host and CPU can exchange
the data via internal memory.

5.Time and Power down unit: Two 32-Bit general purpose timer and used to time events, count events, general
pulses, interrupt the CPU etc. This unit also sends synchronization event to DMA controller. Power down unit is
used to save the power for duration when CPU is inactive.
6.EMI (External Memory Interface): This block supports an interface to several external devices, like
synchronous burst, asynchronous devices, external shared memory device.

Chap 5 / Fixed and Floating point implementation

6. Compare Fixed Point and Floating Point Implementation.
Appeared in exams: Once

Sr
Fixed Point Implementation Floating Point Implementation
no
It has limited dynamic range. For example It has large dynamic range. For example single bit positive
1 16-Bit integer represents a maximum range number has range 2 − 149 to (3.409 × 10) 38 and double precision
65,536. number has a range ( − 10) − 308 to (10) 308.
Difference between two successive number
The difference between two successive small valued number
2 whether it is small valued or large valued
large valued number is different.
remains same.
Error due to rounding and truncation are
3 Floating point representation gives larger precision.
large; so it has less precision
Output of multiply and add stage produces
4 Due to large dynamic range such errors are not produced.
error in the algorithm.
5 Software implementation is complicated. Software implementation is easy.
6 Less computational power. More computational power.
Addition of two numbers does not affect the
7 Addition of two number, usually affect the precision.
precision.
Rounding and truncation must be a part of
8 It is not necessary to specify rounding and truncation.
program.
Overflow error occurs because the size of
The size of intermediate register is around 80-Bits, so overflow
9 intermediate register is comparatively
error does not occur.
small.
Requires less registers and less number of
10 Requires large registers and more number of input-output pins.
input-output pins.
 Chapter 6
Applications of Digital Signal Processing
Content

Application of DSP

Chap 6 / Application of DSP

1. Write a short note on Dual Tone Multi-Frequency Signal Detection.
Appeared in exams: 4 times

• Dual Tone Multi-Frequency or DTMF is a method for instructing a telephone switching system of the telephone
number to be dialed, or to issue commands to switching systems or related telephony equipment.

• The DTMF dialing system follow the technique proposed by AT&T in the 1950s called MF (Multi-Frequency)
which was deployed within the AT&T telephone network to direct calls between switching facilities using in-band
signaling.

• The DTMF system uses eight different frequency signals transmitted in pairs to represent sixteen different
numbers, symbols and letters. This table shows how the frequencies are organized:

• The frequencies used were chosen to prevent any harmonics from being incorrectly detected by the receiver as
some other DTMF frequency. The transmitter of a DTMF signal simultaneously sends one frequency from the
high-group and one frequency from the low-group.

• This pair of signals represents the digit or symbol shown at the intersection of row and column in the table. For
example, sending 1209Hz and 770Hz indicates that the "4" digit is being sent.
• At the transmitter, the maximum signal strength of a pair of tones must not exceed +1 dBm, and the minimum
strength is -10.5 dBm for the low-group frequencies and -8.5 dBm for the high-group frequencies.

Labeling of DTMF numeric digits

• The DTMF telephone keypad is laid out in a 4×4 matrix of push buttons in which each row represents the low
frequency component and each column represents the high frequency component of the DTMF signal. Pressing a
key sends a combination of the row and column frequencies.

• For example, the key 1 produces a superimposition of tones of 697 and 1209 hertz (Hz). Initial pushbutton
designs employed levers, so that each button activated two contacts. The tones are decoded by the switching center
to determine the keys pressed by the user.

• DTMF was originally decoded by tuned filter banks. By the end of the 20th century, digital signal processing
became the predominant technology for decoding. DTMF decoding algorithms often use the Goertzel algorithm to
detect tones.

Chap 6 / Application of DSP

2. Write a short note on different methods for digital signal synthesis.
Different methods for digital signal synthesis are:

Discrete Wavelet Transform (DWT):

Discrete wavelet transform (DWT) is any wavelet transform for which the wavelets are discretely sampled. As
with other wavelet transforms, a key advantage it has over Fourier transforms is temporal resolution: it captures
both frequency and location information (location in time).

Haar wavelet transform:

It may be considered to pair up input values, storing the difference and passing the sum. This process is repeated
recursively, pairing up the sums to provide the next scale, which leads to (2 n − 1) differences and a final sum.

Z-transform converts a discrete-time signal, which is a sequence of real or complex numbers, into a complex
frequency domain representation.

The bilateral or two-sided Z-transform of a discrete-time signal x[n] is the formal power series X(z) defined as
oo
X(z) = Zx[n] = ∑ n = − oox[n]z − n

The inverse Z-transform is

1
x[n] = Z − 1X(z) = 2nj
X(z)z n − 1dz

where C is a counterclockwise closed path encircling the origin and entirely in the region of convergence (ROC).
In the case where the ROC is causal (see Example 2), this means the path C must encircle all of the poles of X(z).

Frequency domain:

Fourier transform decomposes a function of time (a signal) into the frequencies that make it up, similarly to how a
musical chord can be expressed as the amplitude (or loudness) of its constituent notes. The Fourier transform of a
function of time itself is a complex-valued function of frequency, whose absolute value represents the amount of
that frequency present in the original function, and whose complex argument is the phase offset of the basic
sinusoid in that frequency. The Fourier transform is called the frequency domain representation of the original
signal. The term Fourier transform refers to both the frequency domain representation and the mathematical
operation that associates the frequency domain representation to a function of time

For a square image of size N×N, the two-dimensional DFT is given by:

Time domain:

The most common processing approach in the time or space domain is enhancement of the input signal through a
method called filtering. Digital filtering generally consists of some linear transformation of a number of
surrounding samples around the current sample of the input or output signal. There are various ways to
characterize filters; for example:

(i) A "linear" filter is a linear transformation of input samples; other filters are "non-linear". Linear filters satisfy
the superposition condition, i.e. if an input is a weighted linear combination of different signals, the output is a
similarly weighted linear combination of the corresponding output signals.

(ii) A "causal" filter uses only previous samples of the input or output signals; while a "non-causal" filter uses
future input samples. A non-causal filter can usually be changed into a causal filter by adding a delay to it.

(iii) A "time-invariant" filter has constant properties over time; other filters such as adaptive filters change in time.

(iv) A "stable" filter produces an output that converges to a constant value with time, or remains bounded within a
finite interval. An "unstable" filter can produce an output that grows without bounds, with bounded or even zero
input.

Chap 6 / Application of DSP

3. Short note on music sound processing.
Almost all musical programs are produced in basically two stages. First, sound from each individual
instrument is recorded in an acoustically inert studio on a single track of a multi track tape recorder.

Then, the signals from each track are manipulated by the sound engineer to add special audio effects and are
combined in a mix-down system to finally generate the stereo recording on a two-track tape recorder.

The audio effects are artificially generated using various signal processing circuits and devices, and they are
increasingly being performed using digital signal processing techniques.

Commonly used techniques are:

1. Single Echo Filter:

Echoes are simply generated by delay units. For example, the direct sound and a single echo appearing R sampling
periods later can be simply generated by the FIR filter shown in figure, which is characterized by the difference
equation:

y(n) = x(n) + ax(n − R)

2. Multiple Echo Filter:

To generate a fixed number of multiple echoes spaced R sampling periods apart with exponentially decaying
amplitudes, one can use an FIR filter with a transfer function of the form

H(z) = 1 + az − R + a 2z − 2R + a 3z − 3R + ⋯

3.Reverberation:

The sound reaching the listener in a closed space, such as a concert hall, consists of several components: direct
sound, early reflections, and reverberation. The early reflections are composed of several closely spaced echoes
that are basically delayed and attenuated copies of the direct sound, whereas the reverberation is composed of
densely packed echoes. The sound recorded in an inert studio is different from that recorded inside a closed space,
and, as a result, the former does not sound “natural” to a listener. However, digital filtering can be employed to
convert the sound recorded in an inert studio into a natural-sounding one by artificially creating the echoes and
adding them to the original signal.

4.Flanging

There are a number of special sound effects that are often used in the mix-down process. One such effect is called
flanging. Originally, it was created by feeding the same musical piece to two tape recorders and then combining
their delayed outputs while varying the difference between their delay times. Filter used for generation of flanging
is given in figure below:

One way of varying Δt is to slow down one of the tape recorders by placing the operator’s thumb on the flange of
the feed reel, which led to the name flanging. The corresponding input–output relation is then given by:
y(n) = x(n) + ax[n − β(n)]
Congratulations! you've done it.

Hope you found this book useful. We'd love to hear your comments. Just send them over to [email protected]
.

Question papers included

• "+ lesson_no +" • "+ seq_num +"
" + answer_block.html() + "
"); if( !answer_block.html().match('Found repeated in ') ) { answer_block.html("

Answer:

— Found repeated in "+ paper_year + " • "+ lesson_no +" • "+ seq_num +"

"); } } else { elem_paper.after("NOT FOUND!"); } elem_paper.siblings(".paper-ques-desc").wrap(""); } else

{ elem_paper.after("Not mapped"); } }); function get_seq_num(paper_year, lesson_no) { if(typeof
chap_seq_obj === "undefined") { chap_seq_obj = { "Chap 1": 0, "Chap 2": 0, "Chap 3": 0, "Chap 4": 0,
"Chap 5": 0, "Chap 6": 0, "Chap 7": 0, "Chap 8": 0, "Chap 9": 0, "Chap 10": 0, "Chap 11": 0, "Chap
12": 0, "Chap 13": 0, "Chap 14": 0, "Chap 15": 0 } } var key = paper_year + " " + lesson_no; if(typeof
chap_seq_obj[key] === "undefined") { chap_seq_obj[key] = 0; } return ++chap_seq_obj[key] } // Arrange
questions chapter-wise within paper var chap_seq_arr = ['Chap 1', 'Chap 2', 'Chap 3', 'Chap 4', 'Chap 5',
'Chap 6', 'Chap 7', 'Chap 8', 'Chap 9', 'Chap 10', 'Chap 11', 'Chap 12', 'Chap 13', 'Chap 14', 'Chap 15',];
$(".papers-page").each(function(k,v) { $(v).append("
Solution to the paper
"); $.each(chap_seq_arr, function(k1,v1) { var this_lessno = $(v).find(".paper-question .lessno:contains('"+
v1 +"')"); if(this_lessno.length > 0) { var cflag = true; $.each(this_lessno, function(k2, v2){ var this_question
= $(v2).closest('.paper-question'); if(this_question.html().indexOf("Found repeated") 0) { if(cflag) {
$(v).append("

"+ v1 + ": " + $(".lesson-no:contains('"+ v1 +"')").siblings(".lesson-heading").text() || "" +"

"); cflag = false; } $(v).append(this_question.html()); this_question.find(".answer-block").html("

— " + this_question.find(".bubble.right").text() + "

"); } this_question.find(".bubble.right").remove(); }) } }); // create papers box on last page $(".qp-included-

wrapper").append("

" + $(v).find("h2 .paper-header").text().split("-")[1].trim() + "

"); // create link to the paper header $(v).find("h2 .paper-header").html(""+ $(v).find("h2 .paper-
header").text() +""); }); // show papers and hide lessons $(".qp-included-wrapper").removeClass("hide");
$('.lesson-container').addClass('hide'); } else if(print_type == "chapwise") { // if not solutions // remove
repeated topic names $(".question-block p:contains('Subject')").remove(); $(".question-block
p:contains('subject')").remove(); // remove question metadata from older questions $(".question-block
p:contains('Mumbai University')").remove(); $(".question-block p:contains('Mumbai
university')").remove(); $(".question-block p:contains('mumbai university')").remove(); $(".question-block
p:contains('Year')").remove(); $(".question-block p:contains('year')").remove(); // Map questions in QP to
lessons $(".paper-page-id").each(function(k,v){ var elem_paper = $(v); var page_id =
$.trim(elem_paper.text()); if(page_id != "00") { var elem_ques = $("div[data-page-id="+ page_id +"]");
if(elem_ques.length > 0) { var lesson_no = elem_ques.siblings(".lesson-no").text(); var ques_no =
elem_ques.find(".qno").text().replace(")", ""); elem_paper.after(""+ lesson_no +""+ ques_no +""); } else {
elem_paper.after("NOT FOUND!"); } } else { elem_paper.after("Not mapped"); } }); } });

OceanofPDF.com