www.auupdates.

com
CS2403 DIGITAL SIGNAL PROCESSING

DEPT.: CSE SEM.: VII

UNIT I SIGNALS AND SYSTEMS

Basic elements of DSP – concepts of frequency in Analog and Digital Signals

– sampling theorem – Discrete – time signals, systems – Analysis of discrete
time LTI systems – Z transform – Convolution (linear and circular) –
Correlation.

UNIT II FREQUENCY TRANSFORMATIONS

Introduction to DFT – Properties of DFT – Filtering methods based on DFT

– FFT Algorithms Decimation – in – time Algorithms, Decimation – in –
frequency Algorithms – Use of FFT in Linear Filtering – DCT.

UNIT III IIR FILTER DESIGN

Structures of IIR – Analog filter design – Discrete time IIR filter from analog
filter – IIR filter design by Impulse Invariance, Bilinear transformation,
Approximation of derivatives – (HPF, BPF, BRF) filter design using
frequency translation

UNIT IV FIR FILTER DESIGN

Structures of FIR – Linear phase FIR filter – Filter design using windowing
techniques, Frequency sampling techniques – Finite word length effects in
digital Filters

UNIT V APPLICATIONS

Multirate signal processing – Speech compression – Adaptive filter – Musical

sound processing – Image enhancement.
www.auupdates.com
PART B (Q&A)

UNIT I SIGNALS AND SYSTEMS

Q 1 What are the basic elements of DSP and its requirements

Ans. The basic elements of digital signal processing system are shown in fig. below.

is measured as

The different blocks of this system as follows.

1 I/P Signal It is the signal generated from some transducer or from some communication
system It may be biomedical signal like ECG or EEC Generally input signal is analog in
nature It is denoted as x(t)

2. Anti Aliasing filter: It is basically a low pass filter. It is used for the following
purposes.

(a) It removes the high frequency noise contain in input signal.

(b) It avoids aliasing effect that means it is used to band limit the signal.

3. Sample and Hold circuit. It takes samples of I/P signal. It keeps the voltage level of I/P
signal relatively constant which is the requirement of ADC

4 Analog to digital converter It is used to convert analog signal into digital form This is
required because digital signal processor accepts the signal which is digital in nature.

5. Digital Signal Processor : It processes input signal digitally. Input signal making
modifying the signal as per requirement. For this purpose DSP processors like ADSP
2100 or TMS 320 are used.

6 Digital to analog converter (DAC) The output of digital signal processor is digital in
nature. But the required final output is analog in nature. So to convert digital signal into
analog signal DAC is used.
www.auupdates.com
7 Reconstruction Filter Output signal of DAC is analog that means it is continous signal.
But it may contain high frequency components. Such high frequency components are
unwanted. To remove these components reconstruction filter is used.

Q. 2. What are the advantages of digital over analog signal processing?

Ans. The advantages of digital over analog signal processing are given below

1. Accuracy : To design analog system analog components like resistors, capacitors and
mductors are used The tolerance of these components reduce accuracy of analog system.
While in case of DSP, much better accuracy is obtained.

2. Versatility: Digital system can be reprogrammed for other applications (where

programmable DSP chips are used). Moreover digital systems can be ported to different
hardware

3. Repeatability Digital systems can be easily duplicated These systems do not depend
upon component tolerances and temperature

4. Simplicity: It is easy to built any digital system as compared to an analog one.

5. Easy upgradations: Because of use of software, digital signal processing systems can
be easily upgraded compared to analog system

6. Compatibility: In case of digital systems, generally all applications needs standard

hardware. Thus operations of dsp system is mainly dependent on software. Hence
universal compatibility is possible compared to analog systems

7. Cheaper In many applications, the digital systems are comparatively cheaper than
analog systems.

8. Remote processing: Analog signals are difficult to store because of problems like noise
and distortion while digital signal can be easily stored on storage media like magnetic
tapes, disks etc Thus compared to analog signals, digital signals can be easily transposed
So remote processing of digital signal can be done easily.

9. Implementation of algorithms The mathematical processing algorithms can be easily

implemented in case of digital signal processing But such algorithms are difficult to
implement m case of analog signals

Q. 3. Show that for LTI discrete-time system to stable, all the poles should lie within
the unit circle
www.auupdates.com
Ans. Stability Criteria for a Discrete-Time LTI Systems : The stability of discrete- time
LTI system is equivalent to its impulse response h(n) being absolutely summable,

i.e.

In this case the DTFT of h(n) converges and the ROC of transfer function H(z)
must include the unit circle (z) =1

A discrete-time LTI system is stable if and only if the ROC of its transfer function
H(z) includes the unit circle z = 1

However it is perfectly possible for a system to be stable but not causal For causal
systems, stability can easily be checked by examining the locations of poles in transfer
function H(z) For a causal discrete-time LTI system with rational transfer function H(z),
the ROC is outside the outmost pole. A causal discrete time LTI system with rational
transfer function H(z) is stable if and only if all of the poles of H(z) lie inside the
unitcircle IzI =1.

For example Check the stability of the causal discrete-time LTI system with
transfer function as under:

Note that the given transfer function

has a pole at z A. Now, this system will be stable if its

pole is inside the unit circle. Iz = 1, i.e., A <1.

Determine of Impulse Response h (n):

We know that
www.auupdates.com
If h(n) is absolutely summable for I A I <1, then the system is stable.

Q. 4. Find the even and odd parts of the function g[n] = U [n] - U[n -4].

Ans.

The calculations of even and odd parts are shown in fig below
www.auupdates.com

Q. 5. Show that h (n) is equal to the convolution of the following signals.

Ans.
www.auupdates.com

Q. 6. Compute convolution of y (n) of the signals.

Ans.
www.auupdates.com
www.auupdates.com
www.auupdates.com
www.auupdates.com

Q. 7. Define the stability conditions for a linear time invarient system. Determine the
range of values of ‘a’ for which the LTI system with impulse response h(n) as
defined below is stable.

Ans. LTl system is stable if its inpulse response is absolutely summable.

Consider a linear time invarient system having impulse response h(n). Let x(n) be input
applied to this system. Now according to the definition of convolution, the output of such
system is expressed as

The input x(n) will be bounded if I x(n) is less than some finite number. Let us denote
this finite number by M x. Thus for input signal to be bounded

Here Mx is finite number, so its values should be less than infinity. Thus eq. (2) can be
written as
www.auupdates.com
Now taking absolute value of both sides of eq. (1)

We will read R.H.S. of eq. (4) as absolute value of summation of terms. If we take

sign outside then the term become

This summation of absolute

Thus:

values of terms. Always absolute values of sum of terms is less than or equal to the
sum of absolute value of terms.

Here x(n — k) is delayed input signal. If input is bounded then its delayed version is
also bounded. This is because delay or qlding is related to time shifting operations By
performing these operations the magrjitu4e is not changed. Now for bounded input,
www.auupdates.com

We know that Mx is finite number. We want 0/P y(n) to be bounded. That means

I y(n) should be finite. So eq. (8) to obtain finite output we have

Here h(k)= h(n) is the impulse response of LTI system. Thus eq. (9) gives the conditions

of stability in terms of impulse response of the system.

From above, it is abvious that given system is stable if a <1.

Q. 8. Determine the output y(n) of linear time invariant system units impulse
response,

when the input is a unit step sequence x(n) = U(n).

Ans. In this case bth h(n) and x (n) are infinite duration sequences. We use the form

of the convolution formula

www.auupdates.com
x(2—k) h(k) are also illustrated in Fig.
www.auupdates.com
www.auupdates.com

Q 9 Let an LTI system be characterised by the difference equation y(n) = x(n) + ax

(n -1)

Ans. LTI systems are characterised by input output relations called as difference equation
with constant co-efficients.

A recursive system shown in fig below

www.auupdates.com
The equation Of output
is

Note that we have obtained output y(n) by putting only positive values of n. That means
for . So these are the equations of causal system. Every equation contains the term
y(-1). It is assumed that, we are starting the operation of system at n = 0. Then the term
y(—l) is called as initial condition of a system.

We know the equation (8) is representing the behaviour of causal system. It consists of
two parts as follows

(1) The first term y(—l) is called initial condition of the system.

(2) The second term is the response of system to the input x(n).

Now depending upon initial conditions, two types of responses are obtained.
www.auupdates.com
1. Zero state response or forced response : If initial condition y(—1) of system is
zero then the response of the system to the input signal is called is zero state
response. it is denoted by y zs (n).

2.Zero input response or natural response: Input is not applied to the system and
initial condition . Thus this kind of output without applying input x(n)
is called zero input response or natural response. It is denoted by yzi

Equation (1) is basically a linear constant co-efficient difference equation. The general
form of this equation can be expressed as

Here ak and bk are constant co-efficients. ‘N’ is the order of difference equation. That

means ‘N’ represents the order of system.

Q. 10. An LTI system is described by . Find the

response of this system for an input of x(n) = 10 cos (0.05 3m).

Ans. Given
www.auupdates.com
www.auupdates.com
www.auupdates.com

Q. 11. Plot the derivative of the function x(t) = sin c (t).

Ans.
www.auupdates.com

Q. 12. Differentiate between a recursive and non recursive system. Determine if the
recursive system defined by the difference equation y(n) a y(n 4) + x (n) is linear.

Ans. Recursive System: A discrete time system in which output y(n) depends on present
input, past inputs as well as previous outputs is called a recursive system. Example of the
recursive systems, is given below:

Consider a system represented by block schematic as shown in fig below.

www.auupdates.com
y(n) = x(n) + ay (n -1)

As shown in fig a connection is drawn from output y(n) to the input side. This is a
feedback. This feedback signal y(n) is passed through an unit delay (z -1) block So the
output of delay block is y(n-1). This is past output compared to y(n). The signal y(n -1) is
multiplied by constant multiplier ‘a’ to obtain ay(n-1) . An adder is used to add input x(n)
and signal ay(n -1). Now the addition of x(n) and ay(n-1) produces the output. Thus the
output of system is expressed as

Non-recursive system : It do not require any past output sample to calculate the present
output, for example consider the equation of FIR causal system which is described by N.

To calculate present output we need present and past inputs. We can write eq. (1) as
follows

This equation does not contain any precious 0/P term like y(n-1), y(n - 2)....etc. So this is
equation of non-recursive system.

In case of non-recursive system, past output terms are not required. Past output terms are
obtained by - taking feedback from output to the input. Thus non-recursive system do not
require any feedback. The representation of non recursive system is shown below.
www.auupdates.com

So at any instant present output depends on past output. Hence it is recursive system. The
representation of non-recursive system is shown in figure below.

As shown in fig above and given by eq. (2) output is a linear combination of input x(n)
and delayed output. So it is linear system.

Q. 13. Define convolution theorem as applied to discrete time signals.

Find the inverse z-transform of

using convolution theorem.

Ans. Like convolution integral for continuous time LTI systems, the convolution of two
discrete time signals is called discrete convolution. This is used to find the response of a
discrete time LTI system with impulse response h(n) for any arbitrary input x(n).

Let us consider that a discrete time input signal x(n) is applied to a discrete time LTI
system with unit impulse response h(n).

Then the discrete time output signal y(n) of this system may be expressed as
www.auupdates.com
The above expression for discrete time output signal y(n) is called the convolution

sum as against the convolution integral for continuous time LTI systm.

These equation (1) may be expressed as

This equation indicates that the output y(n) is the convolution of the discrete time

input signal x(n) and discrete time unit impulse response h(n).

The convolution process can be summarised into the following steps.

1. Choose an initial value of ‘n’ the starting time for evoluting the output sequence y(n).
If x(n) starts at n = n1 and h(n) starts at n n2 then n = n1 + n2 is good

choice. Then express both sequence in terms of the index k.

2. Folding: Fold the Iz(k) about the origin and obtain h(-k).

3. Time shifting: shift the h(—k) by n unit to right if n is positive and left if n is —ye to
obtain h[-. (k — n)] = h(n — k).

4. Multiplication: Multiply x(k) by h(n - k) to obtain

5. Summation: Sum all the values of the product w(k) to obtain the value of 0/p y(n).

6. Increment the index n, shift the sequence h(n — k) to right by one sample and do step
4.

7. Repeat step 6 until the sum of product is zero all remaining values of n.
www.auupdates.com

Q. 14. What are the various realization techniques of linear time invariant systems?
Mention.

Ans. The linear time invariant system is described by the difference equation of form.
www.auupdates.com
Here ak and bk are constants with a0 = 1.

The methods of realizing digital systems can be divided into the two classes namely

recursive and non-recursive. For the linear time invariant system the non recursive
realization has the form

The current output sample y(n) is a function of past output and present and past

inputs samples. For the LTI system the recursive realization has the form of

First order system block diagram representation.

let us consider a first order system given by

Which is realised as in fig. below. The term y(n —1) is obtained as the output of delay
element with y(n) as input, x (n —1) is obtained as the output of delayed element with
x(n) as input. This realization uses separate delay for both input and output signal
samples. This structure is called as direct form I structure. This systeth can be viewed as
two LTI systems in cascade.
www.auupdates.com

The first is non-recursive system described by

and second is recursive system described by

The block diagram of Fig.(a) can be rearranged or modified in a variety of ways without
changing the overall system function. From this figure we obtain two different equations.

The difference equation (8) and (9) are equivalent to single difference eq. (5)

Fig. (b) reveals that the two elements contain the same input o(n) and hence the same
output w(n -1). Therefore, these two elements can be merged into one delay as shown in
Fig. (c). In comparison to direct form I realization, the new realization required only one
delay for co (n), therefore it is more efficient in term of memory requirements. The
realization of Fig. (c) is called as Direct form II structures.
www.auupdates.com
Q. 15. Determine the impulse response for the cascade of two linear time invariant
systems having impulse responses.
www.auupdates.com
Ans. Let us first determine the homageneoüs solution of the given difference equation.
We assume the solution to be form of

The particular solution is assumed to be an exponential sequence of the some form as

x(n). Let us assume a solution of the form

Since the particular solution y, (n) is already contained in homogeneous solution, so that
this particular solution is redundant. We treat this situation. if the characteristics equation
has multiple roots. Therefore, we assume
www.auupdates.com

Q. 16. List various properties of z-transform.

Ans. The various properties of z-transform are given below.

1. Linearity

1. Shifting:

1. Multiplication (by nm). [or differentiation in z-domain]

www.auupdates.com
4. Scaling in z-domain: [Multiplication by a n ]

5.Time Reversal:

6. Conjugation:

7. Convolution:

8. Initial value:

9. Final value:

10. Correlation of two sequence.

11. Parseval’s theorem:

where the contour of integration must be overlap of the regions of convergence of

www.auupdates.com
Q. 17. State final value’ theorem of Z-transform.

Ans. If

Acc. to the definition of Z-transform for causal sequence we have,

Consider the term x(n) — x(n — 1) using eq. (1), z-transform of this term can be

expressed as.
www.auupdates.com

Hence proved.

Q. 18. Determine the z-fransform of the signal

Ans. By using Euler’s identity the signal x (n) can be expressed as

www.auupdates.com
By simply linearity property.

After some simple algebraic manipulations we obtain the desired result namely.

Q. 19. Find the z-transform and region of convergence for the following sequence.

Apply initial value theorem and check the z-transform whether it is correct or not.

Ans.
www.auupdates.com
www.auupdates.com

Thus calculated z transfer is corrects.

Q. 20. Find the inverse z transform of

Ans. Given:
www.auupdates.com
By using long division method.

Q. 21. Find a causal sequence whose z transform is given by

Ans. Given :

We have a series expansion formula.

www.auupdates.com

Q. 22. Find all possible inverse z-transform of the following.

Ans. Given:

In partial fraction expansion form,

www.auupdates.com

Q. 23. Determine the causal signal x (n) if its z-transforms X (z) is given by
www.auupdates.com
Ans. Given:

Converting x (z) into positive power of z.

In terms of partial fraction expansion we canwrite.

www.auupdates.com
Q. 24. Find the impulse response of the causal system

where I a I < 1 and compute the frequency response.

Ans. To find the impulse response sequence we write the homogeneous equation.

Therefore the impulse response is in the form

We solve for the constant-A, from the original solution, we prefer the initial condition

To find the frequency response,

www.auupdates.com
Using geometric series summation.

Q. 25. Consider a signal x (n) given by

Ans.

The sum in convergences then

www.auupdates.com

For convergence of X (z) both sum must converge

which requires that

1Thus the ROC is because the ROC is the common region where both sum

are finite. The fig shows the ROC of the z-transform of each of the individual terms & for
combined signal.

Q. 26. Find the one sided z-transform of the following discrete sequence
www.auupdates.com
Ans. The one sided z-transform of a n is given by

Q. 27. Find the one sided z-transform of discrete sequence generated by

mathematically sampling the continuous time function

Ans. Given
www.auupdates.com
The discrete sequence is generated by replacing t by kT, where T is the sampling period.

Q. 28. (a) If

Find the steady state value of y (n) if exists.

1. FindY(n) if X (z) is given by

www.auupdates.com
Ans. (a) By inspection, we can find that ROC for Y (z) is has no

poles on or outside the unit circle, so the conditions of the final value theorem satisfied.
Therefore the steady state value exists and is given by:

In this case X (z) has a pole on the unit circle. So x (n) is not defined.

Q. 29. Find the z-transform of

Ans. Given
www.auupdates.com
Q. 30. Find the inverse Z-transform of the function,

Ans. Given

From above equation we can say that if a complex sinusoidal signal is given as input
signal to LTI system, then the output is a sinusoid of the same frequency modified by
H(w). Hence H(w) is called the frequency response of the system. The H(w) produces a
change in the amplitude and phase of the input signal.
www.auupdates.com
The frequency response H(w) of LTI system is also given by the ratio of Fourier
transform of output to Fourier transform of input.

Q. 31. Determine the inverse z-transform of

www.auupdates.com
Ans.(a) Since the ROC is exterior of a circle, we except x (n) to be a causal sequence.
This we divide so as to obtain a series in negative power of z. Carrying out the long
division, we obtain

(b)When the ROC is interior of the circle, the signal x (n) is anticausal signal. Thus we
divide so as to obtain a series in power of z as follows.
www.auupdates.com

Q. 32. Find the inverse Z-transform of the function,

Ans. Given
www.auupdates.com

From above equation we can say that if a complex sinusoidal signal is given as input
signal to LTI system, then the output is a sinusoid of the same frequency modified by
H(w). Hence H(w) is called the frequency response of the system. The H(w) produces a
change in the amplitude and phase of the input signal.

The frequency response H(w) of LTI system is also given by the ratio of Fourier
transform of output to Fourier transform of input.
www.auupdates.com

Q. 33. Determine the inverse z-transforms of

Ans. Given
www.auupdates.com

Q. 34. Determine the one sided z-transform of

www.auupdates.com
Ans. Taking z-transform of difference equation.

System function

Let us defined the function H (z) by

www.auupdates.com
The function H (z) is known as the transfer function of a system or system function.

When the input to the system is impulse signal then

Q. 35. Determine the z-transform of the signal

Ans.

The ROC is now the interior of a circle having radius I . This is shown in fig

below.
www.auupdates.com

UNIT II FREQUENCY TRANSFORMATIONS

Q. 1. In what respects does DFT differ from continuous Fourier transform?

Ans. The Fourier transform of a continuous time a periodic signal is given by

The conditions for the existence of the Fourier transform is follows the Dirichlet
conditions which are:

I .The signal x(t) has a finite no. of discontinuities.

2.The signal x(t) has a finite no. of maxima and minima.

3.The signal x(t) is absolutely integrable that is

The discrete time Fourier transform of finite energy aperiodic signal x(n) is
representation of signal in terms of complex exponential sequence where w is a
real frequency variable.

The discrete time Fourier transforms of x(n) is defined as

www.auupdates.com
where X(w) represents the frequency content if signal x(n). Physically X(w) is a
decomposition of x(n) into its frequency components.

Q. 2. State the Goertzel algorithm and give its importance?

Ans. There are certain applications in which selected number of DFT values are
required, but not the entire DFT. In such cases the direct computation of DFT is more
efficient than FFT algorithms. The direct computation of DFI for selected values can be
realized using linear filtering approach. This type of computation can be performed using
this algorithm called as Goertzel algorithm.

In this algorithm the periodicity property of twiddle factor is used. The

definition of DFT.

Note that here we have used notation x (m) for input sequence.

Now the twiddle factor is given by.

www.auupdates.com
Now according to the linear convolution.

Eq. (2) & (7) are identical. Thus we can conclude that DFT, X(k) can be obtained as
output of LTI system at n = N. But in this case will get X(k) only at one value of k. So we
need to use parallel systems to obtain X(k) at different values of k. Goertzel algorithm is
efficient when X(k) is to be computed at points less than log2N.

Q.3. What are quantization errors in FF1’ algorithm?

Ans. Round off error in FFT algorithm are more quantization errors. In FFT, the

basic operation is butterfly computation. For radix-2 DIT FFT algorithm butterfly is
characterized as. -

Where x & y are the inputs to butterfly, x’ and y’ are outputs, Wk is a twiddle factor.
www.auupdates.com
Thus all these variables are complex valued. Butterfly computation is carried out using
real arithmetic.

Each butterfly computation requires for multiplications and six real additions. Hence each
product will require approx. twice as many bits to represent as the operand themselves.
For example, if the variables are represented as 16—bits numbers, then after
multiplication each product will require 32-bits to represent. To overcome this problem,
truncating or rounding each product back to 16 bits is done. But it produces an error
called round off error.

In this case, we can associate four round off noise sources to each butterfly. One for

each product and so the round off noise power at the output of each butterfly is given

as

where

& the system wordlength is B bits. This noise generated by a butterfly at one stage is fed
into subsequent stages. Consider implementation of 8 points radix-2.DIT FFT algorithm.
In this it is seen that each FFT output X(k) can be traced back to (N—i> butterflies. Each
FFT output is connected to all N/2 butterflies in stage 1, N/4 butterflies in stage 2, N/8 in
stage 3 and so on. Assuming that each butterfly generates identical but uncorrected
errors, the maximum noise power at each FF1 0/P is

Thus the noise power is directly proportional to the transform size. Doubling N, which is
equivalent to adding a stage o the FFT, doubles the noise power. To retain the same noise
power we increase the word length by 1 bit since the noise power is proportional to N and
to

The signal to noise ratio is given by.

www.auupdates.com
For butterflies that have non-trivial twiddle factors such as a lower

noise power will be produced as a result of round off error.

Q. 5. Define circular convolution. How can linear convolution be realized using

circular convolution ?

Ans. Consider we have two finite duration sequences of lengths N, x1 (n) & x2 (n).

Their respective N-point DFT are

. If we multiply the two DFT’s together, the results in a DFT say X3 (k) of a sequence
x3(n) of length N. Let us determine the relationship between x 3(n) and the sequences
x1(n) & x2(n).

Suppose that we substitute for is (4) using the DFT’s given in (1) & (2).
Thus we obtain.
www.auupdates.com
The inner sum in brackets in (5) has form.

where a is defined as

We observe that a = I when m - n - 1 is multiple of N. On the other hand ar” = I for any
value of Consequently (6) reduces to

If we substitute the result (7) in (6) we obtam the desired expression for x3 (m) in ‘form

The expression (8) has form of a convolution sum. However it is not the ordinary linear
convolution which relates the output sequence y (n) of a linear system to input sequence
x (n) and the impulse response h (n). Instead the convolution sumin (8) involves the index
((m—n))N and is called circular convolution.

The basic difference between these two types of convolution is that in circular
convolution the folding & shifting operations are performed in circular fashion by
computing the index of one of the sequences module N. In linear convolution, there is no
module N operation. Basically circular convolution y (n) contains same number of
samples as that of x (n) & h (n). In case of linear convolution no. of samples are given by

N.

N=L+M-1
www.auupdates.com
L = No. of samples in x (n)

M = No. of samples in h (n)

N = No. of samples in the result of linear convolution.

That’s why the result of linear and circular convolution are not same.

To obtain the same result both convolutions, the following, steps are used

1 Calculate the value of N that means no of samples contained in linear convolution

2. By doing zero padding make the length of every sequence equal to N. That means in
this case we need to add zeros in x (n) as well as Ii (n).

3. Perform the circular convolution. The result of circular convolution & linear
convolution will be same.

Q. 6. Define circular symmetric of a sequence in DFT.

Ans. Circular Symmetries of a sequence:

We have studied the periodicity property of DFT. Suppose input discrete time sequence is
x(n) then, the periodic sequence is denoted by xp(n). The period of xp(n) is ‘N’ which
means after ‘N’ the sequence x(n) repeats itself. Now we can write the sequence x p (n) as,

We will consider one example. Let x (n) {1, 2, 3, 4). This sequence is shown in Fig. The
periodic sequence xp(n) is shown in fig.

We will delay the periodic sequence xp(n) by two sample as shown in Fig. This sequence
is denoted by xp(n —2). Now the original signal is present in the range n = 0 to n = 3. In
the same range we will write the shifted signal as shown in fig. This signal is denoted by
x’(n).
www.auupdates.com
www.auupdates.com

Now from fig. we can write every sequence as follows

www.auupdates.com
Notation

This relation of circular shift is denoted by,

Now observe Equations (2) and (5). We can say that the sequence x’ (n) is obtained
circularly shifting sequence x(n), by two samples. That means x’ (n) is related to x(n)
circular shift. i.e. called circular symmetry of the sequence.

It means that divide (n - k) by N and retain the remainder only. We can also use the short
hand notation as follows:

Q. 7. Describe the symmetry properties of DFT.

Ans. The symmetry properties for DFT can be obtained by applying the methodology as
used for fourier transform. Let us assume that the N-point sequence x (n) and its DFT are
both complex valued. Then the sequences can be expressed as
www.auupdates.com

Real valued sequences : If the sequence x(n) is real, it follows directly from DFT that

Futhermore, x1 (n) = 0 and therefore x(n) can be determined from (5) which is another
form for the IDFT

Real and even sequences: If x(n) is real and even that is

www.auupdates.com

Q. 8. Discuss various properties of DFT.

Ans. Properties of the DFT.

1. Linearity : The DFT obeys the law of linearity. If

www.auupdates.com
then for any two constants a and b,

1. Periodicity

3. Circular shift of a sequence : This property is analogous to the time shifting

property of the DFT, but with some difference.

Let us consider a sequence x (n) of length N which is defined for 0 n N - 1. The sample
value of such sequence is zero for n <0 and n N. For any arbitrary integer k, the shift
sequence x1 (n) = x (n — k) is no longer defined for the range 0 n N - 1. Therefore we
need to define another type of shift that will always keep the shifted sequence in the
range 0 n N - 1.

This shift is known as circular shift that can be represented as the index modulo N. Thus
we can write,
www.auupdates.com
where x (n) is represented the circular shift of x (n). Or more generally we can

define

Equations (6.4) and (6.5) tell us how to construct x (n).

1. Circular convolution and multiplication of two DFTs : Consider two finite

duration sequences both of length, with their N - point DFTs
and i.e.,

Here we wish to determine the sequence x3 (n) for which the DFT is,

Using equations (6.6) and (6.7) in (6.8), we have,

www.auupdates.com
The brackets term in eqn. (6.9) has the form,

If we substitute the eqn. (6.11) into eqn. (6.12), we have,

The equation (6.12) has the form of convolution sum. However it differs from a linear
convolution of x1 (n) and x2 (n) as defined in unit-TI. In linear convolution the
computation of the sequence x3 (n) involves multiplying one sequence by a folder and
linear shifted version of the other and then summing the values of the product x1 (m) x2
(n - m) for all values of n.

Instead the convolution sum in eqn. (6.12), the second sequence is circularly time
reversed and circularly shifted w.r.t. of first. The equation (6.12) is called circular
convolution of two finite duration sequences.
www.auupdates.com
Thus we conclude that multiplication of the DFTs of two sequences is equivalent to the
circular convolution of the two sequences in time domain. The operation of forming

where, denotes the circular convolution of two N length sequences, x1 (n)

and x2 (n).

1. Time reversal of a sequence:

Hence, when the N-point sequence is reverse in time, it is equivalent to reversing the
DFT values. The time reversal is illustrated in Fig.
www.auupdates.com
If we change the index from n to m by defining m = N - n, then,

7. Circular time shift:

www.auupdates.com
Proof.

If we change the index from n to m = N + n - 1, then

Equation (6.18) can be written as

8. Multiplication of two sequences:

www.auupdates.com
The proof of this property is similar to circular convolution.

9. Circular correlator : For complex valued sequence x (n) and y (n).

where is the circular cross-correlation sequence, given as,

Proof.

We know circular convolution of the two sequences is just equal to the multiplication

of the their DFTs and from complex conjugate property

www.auupdates.com
10. Parseval’s theorem:

For the complex-valued sequence x (n) and y (n), if

11. DF of even and odd sequences : The DFT of an even sequence is purely real.
www.auupdates.com
The DFT of an odd sequence is purely imaginary. Therefore DFT can be evaluated using
cosine and sine transforms for even and odd sequences respectively.

For even sequence,

For odd sequence

Q. 9. Develop a Radix-2, 8-point DIF FFT algorithm with neat flow chart.

Ans. Decimation in frequency stands for splitting the sequences in terms of frequency.
That means we have split output sequences into smaller subsequences. This decimation is
done as follows.

First stage of decimation:

We can divide the stmmation into two parts as follows.

Now consider the second summation that means

www.auupdates.com

Putting this value in 2equatioñ (2), we get

Taking the summation common we get

www.auupdates.com
Here we have no split the sequence in terms of frequency. So we will split X(k) in terms
of even numbered and add numbered DFT.co-efficients. Let X(2r) represents even
numbered DFT and X(2r + 1) represents odd numbered DFT.

Thus puffing k = 2 r in eq (4), we will get even numbered sequence

By puffing k = 2r + 1 in eq. (4), we will get odd numbered sequence

Here ‘r’ is an integer similar to k and it varies from 0 to N/2 -1

Puffing these values in eq (5) & (6) we get

www.auupdates.com
Puffing these values in eq. (9) & (10) we get

Putting these values in eq. (13) & (14) we get

Note that at this stage we have decimated the sequence of ‘N’ pomt DFT mto two N/2

point DFT’s given by eq (17) & (18) Let us consider an example of 8 pomt OFT That

means N = 8 So considermg (17) & (18) (that means N/2=4) we can obtain N (8-point)
www.auupdates.com
DFT. This is first stage of decimation. Note that eq. (17) indicate 4 (N/2) point OFT of
IN\

g(n) and eq (18) indicates 4 (N/2) pomt DFT of h(n) For 8 pomt OFT eq (15) becomes

Here we are computing ‘4’ point DFT, So range of ‘n’ is n=0 to n=3. Putting these

values in eq. (19), we get

Using equations (20) & (22) & eq. (17) & (18) we can draw the flow graph of the

first stage of decimation as shown in fig. below.

www.auupdates.com

Second stage of decimation: In the first stage of decimation we have used 4-point DFT.
We can further decimate the sequence by using 2 point DFT. The second stage of
decimation is shown in fig below.

Third stage of decimation : In the second stage of decimation we have used 2- point DFT.
So further decimation is not possible. Now we will use a butterfly structure to obtain 2 -
point DFT. Thus the total flow graph of 8 point DIF-FFT is shown below.
www.auupdates.com

Q. 10.Discuss DIT and DIF algorithms and also compare the two algorithms.

Ans. Radix-2 Decimation in Time (DIT) Algorithm (DIT FFT) : To decimate means to
break into parts. Thus DIT indicates dividing (splitting) the sequence in time domain. The
different stages of decimation are as follows:

First stage of decimation Let x(n) be the given input sequence containing ‘N’ samples.
Now for decimation in time we will divide. x (n) into even and odd sequences.

Input sequence x(n) has ‘N’ samples. So after decimation;f1(m) and f2(m) will contain N

j- samples.

Now according to the definition of DFT,

www.auupdates.com
Since we have divided x(n) into two parts, wq can write separate summation for even and
odd sequences as follows:

The first summation represents even sequence. So we will put n = 2m in first summation.
While the second summation represents odd sequence, so we will put n =(2m+1)

(2m +1) in second summation. Since even and odd sequences contain N/2 samples each;

the limits of summation will be from m = 0 to N/2-1

But x (2m) is even sequence, so it is f1(m) and x(2m +1) is odd sequence, so it is f2

(m).

Comparing each summation with the definition of DFT,

www.auupdates.com
We will consider an example of 8 point DFT. That means N = 8.

Now F1(k) and F2(k) are 4-point (N/2) DFTs. They are periodic with period N/2.

Using periodicity property of DFT we can write,

www.auupdates.com

We are considering an example of 8 point DFT (N = 8). So in Equations (15) and (16), k
varies from 0 to 3. Now putting k = 0 to 3 in Equations (15) and (16) we get,
www.auupdates.com

The graphical representation of first stage of decimation shown in Fig.

for 8 point DFT is as

In Fig. input sequence are

That means each sequence contains N/2 samples.

Second stage of decimation:

www.auupdates.com
In the first stage of decimation; we obtained the sequences of length N/2 . That

means for 8-point DFT (N = 8); the length of each sequence is ‘4’ as given by Equations
(19) and (20). We discussed that we have to continue this process till we get ‘2’ point
sequence.

We can further decimate f1(m) into even and odd samples. Let g11(n)= f1(2m), which
contains even samples and let g12(n) = f1(2m +1), which contains odd samples of f1(m).

Note that here range of ‘n’ and ‘m’ is from 0 to N/4-1

Now recall equations (15) and (16). We obtained sequences X(k) and X(k+N/2)

from F1(k) and F2(k). The length of each sequence was N/2. Here in the second stage of

decimation. We are further dividing the sequences into even and odd parts. So similar to
Equations (15) and (16) we can write; For F1(k),

Thus, for N = 8 we have the range of K, from K = 0 to K = iHere G11(k) is DFT of

g11(n) and G12(k) is DFT of g12(n) Now putting the values of ‘K’ in Equation (21) we
get,

Here the values of K are ‘0’ and 1. That means it is 2-point DFT. Thus Equations (23)
and (24) shows that we can obtain 4-point DFT by combining two 2-point DFTs. The
graphical representation is shown in Fig.
www.auupdates.com

Note that here,

Now similar to Equations (21) and (22) we can write equations for F2(k) as follows:

Similarly from Equation (27), we get

www.auupdates.com
The graphical representation of Equations (28) and (29) is shown in Fig.

Note that here.

Combination of first and second stage of decimation:

Combining Fig. we get the combination of first and second stage of decimation. It is
shown in Fig.

At this stage we N/4 have that means 2 point sequences. So further decimation is

not possible. As shown in Fig. we have to compute 2-point DFT.

www.auupdates.com

Computation of 2-point DFT: According to the basic definition of DFT,

We will use Equation (31) to compute 2-point DFT. From Fig. consider the first block of
2-point DFT. It is separately drawn as shown in Fig.

Here input sequences are g11(0) and g11(1). We can denote it by g11(n); where n
www.auupdates.com

varies from 0 to 1. Now the output sequences are G 11(0) and G11(1). We can denote it by
G11(k); where ‘k’ varies from 0 to 1. Here G 11(k) is DFT of g11(n).

Thus for G11 (k) we can write Equation (31) as,

Note that this is 2 point DFT, so we have put N = 2.

Now puffing values of k in Equation (32) we get,

Expandin the summation we get,

Expanding the summation we get,

www.auupdates.com

Using Equations (33) and (35), we can represent the computation of 2-point DFT

as shown in Fig. This structure looks like a butterfly. So it is called as FFT butterfly

structure.

Now we know that . Thus we can modify Equation (33) and (35) as

Follows:

This modified butterfly structure is shown in Fig.

www.auupdates.com
Similarly for other 2-point DFTs we can draw the butterfly structure.

Total signal flow-graph for 8-point DIT FF1’

The total signal flow graph is obtained by interconnecting all stages of decimation. In this
case, it is obtained by interconnecting first and second stage of decimation. But the
starting block is the block used to compute 2-point DFT (butterfly structure). The total
signal flow graph is shown in Fig.
www.auupdates.com

Q. 11. Draw a 8 point radius 2 FF1’ DIT flow graphs and obtain DF1’ of the
following

sequence

Ans.
www.auupdates.com

Q. 12. Compute 4-point DFT of causal three sample sequence given by.
www.auupdates.com
Ans. By definition of N-point DFT, the kth is complex co-efficient of X (k) for

is given by

Here N = 4, therefore the 4-point DFF is

The values of X (k) can be evaluated for k = 0, 1, 2, 3.

www.auupdates.com

The three sample sequence and its periodic extension are shown in Fig. below.

Q. 13. Discuss the properties of Radix-2 decimation in time algorithm. Find 8- point
DFT of sequence.

or Define Radix 2-FFT algorithm?

www.auupdates.com
Ans. While calculating DFL we have discussed that we always calculate ‘N’ point DFT.
The number N can be factorised as

Here ‘r’ is called as radix and v as a number of stages.

Now radix means base and if its value is ‘2’ then it is called as radix 2 Thus when

r = 2, the equation become

Thus if we are computing 8 point DFT then N = 8

So for 8 point DFT, there are three stages of FFT algorithm. While computing FFT,

divide number of input samples by 2, till we reach minimum two samples. Based on this
division there are two algorithms as follows.

1. Radix-2 decimation in Time (DIT) algorithm

2. Radix-2 decimation in Frequency (DIF) algorithm

In D1T the input is bit reversed while the output is in natural order. DYE can be go from
normal to shuffled data or vice versa.

If the observe the basic computation performed at every stage of DIT rodix 2, we

can arrive at the following conclusion.

(1) In each computation two complex numbers a and b are considered.

(2) The complex number b is multiplied by a phase factor

www.auupdates.com
(3) The product b is added to the complex number a to form new complex number
A. -

(4) The product b is subtracted from complex number a to form new complex
number B.

The above basic computation can be expressed by a signal flow graph shown in fig

below.

Given sequence is

Thus we have
www.auupdates.com

Q. 14. A finite duration sequence of x (n) of length 8 has 8-point DFT X(k) as shown
in fig. A new sequence y(n) of the length is defined by.

find the 16-point DFT of y(n).

Ans. We have a finite duration sequence x(n) of length 8. Suppose we interpolate it by a

factor of 2. That is we wish to double the size of x(n) by inserting zeros at all values of n
for
www.auupdates.com

Therefore, 16 point DFT of interpolated signal y(n) contains two copies of 8 point DFT
of x(n). Since for Y (k), N 8 therefore Y (k) is periodic with period 8.

15. Compute the DFT of sequence defined by : for

(a) N = even

(b)N= odd.

Ans.
www.auupdates.com

1. For N even

(B)For N odd

Since N is odd, therefore, no k exist so we can write.

www.auupdates.com
Q. 16. A finite duration sequence of length L is given as

Determine N-point DF1’ of this sequence for

Ans. Length L is given as

of this sequence for

Form of this sequence is

The magnitude and phase of are illustrated in Fig for L = 10. The N point of

x(n) is simply evaluated at the set of N equally spaced frequencies

www.auupdates.com

Q. 17. By means of the DFT & IDFT, determine the sequence x3 (n) corresponding
to the circular convolution of the sequence x1 (n) and x2(n).

Ans. First we compute the DFT’s of x1 (n) and x2(n).

The four point DFT of x1 (n) is

www.auupdates.com

Q. 18. Find the inverse DFT of X(k) = {1, 2, 3, 4}.

Ans. We know that the inverse DFT is expressed as

www.auupdates.com

Q. 19. By means of the DFT & IDFT, determine the response of the FIR filter with
impulse response.

Ans. The input sequence has length L = 4 and the impulse response has length M = 3.
Linear convolution of these two sequences produces a sequence of length N = 6.
Consequently, the size of the DFT must be at least six.

For simplicity we compute eight point DFT’s. We should also mention that the efficient
computation of the DFT via the Fast Fourier Transform (FFT) algorithm is usually
performed for a length N, that is power of 2. Hence eight point DFT of x(n) is
www.auupdates.com
www.auupdates.com
Q. 20. An 8-point sequence is given by x (n) = {2, 2, 2, 2, 1, 1, 1, 1). Compute 8-point
DFT of X(n) by radix -2 DIT FFT. Also sketch the magnitude and phase spectrum.

Ans. The given sequence is first arranged in the bit reversed order.

For 8-point DFT by radix 2 FFF we require 3 stages of computation wits 4 butterfly
computations in each stage. The sequence rearranged in the bit reversed order forms the
input to the first stage. For other stages of computation the 0/P of previous stage will be
I/P for current stage.

First stage computation:

The I/P time sequence {2, 1, 2, 1, 2, 1, 2, 1)

The butterfly computations of first stage are shown below.

www.auupdates.com

Input sequence = {3, 1, 3, 1, 3, 1, 3, 1)

The phase factors involved in second stage computation are W & W.

The 0/P OFT sequence = {3, 1, 3, 1, 3, 1, 3, 1} Second stage computation

are

The butterfly computation are shown below.

www.auupdates.com
The butterfly computations of third stage are shown below
www.auupdates.com

Q. 21. An 8-point sequence is given by x(n); x(n) = (2, 2, 2, 2, 1, 1, 1, 1).

Compute 8 point DFT of x(n) by radix -2 DIF-FFT.

Ans. For 8 point DFT by radix 2 FF1 we require 3-stages of computation with 4 butterfly
computation in each stage.

The given sequence is the input to the first stage. For other stages of computation, the
output of the previous stage will be the input for the current stage.

First stage of computation

www.auupdates.com
The input sequence = {2, 2, 2, 2, 1, 1, 1, 1)

The phase factors involved in first stage of computation are

The output sequence of first stage of computation.

Second stage of computation

The input sequence of 2nd stage

www.auupdates.com
The phase factors involved in 2nd stage of computation are

The butterfly computation of second stage are shown above.

Q. 22. Discuss Linear filtering approach for the computation of DFT.

Ans. Linear Filtering Methods Based on the DFT : Since the DFT provides a discrete
frequency representation of a finite-duration sequence in the frequency domain, it is
interesting to explore its use as a computational tool for linear system analysis and
especially, for linear filtering. We have already established that a system with frequency
response H (), when excited with an input signal that has a spectrum X (w), possesses an
output spectrum Y (w) = X (w) H (w). The output sequence y (n) is determined from its
spectrum via the inverse. Fourier transform. Computationally, the problem with this
frequency domain approach is that X (w). H (w) and Y (cv) are functions of the
continuous variable cv. As a consequence, the computations cannot be done on a digital
www.auupdates.com
computer, since the computer can only store and perform computations on quantities at
discrete frequencies.

On the other had, the DFT does lend itself to computation on a digital computer. In the
discussion that follows, we describe how the DFT can be used to perform linear filtering
in the frequency domain. In particular, we present a computational procedure that serves
as an alternative to time-domain convolution. In fact, the frequency-domain approach
based on the DFT, is computationally more efficient than time-domain convolution due
to the existence of efficient algorithms for computing the DFT. These algorithms, which
are described in Chapter 6, are collectively called fast Fourier transform (FFT)
algorithms.

Use of the DFT in Linear Filtering: In the preceding section it was demonstrated that the
product of two DFTs is equivalent to the circular convolution of the corresponding time-
domain sequences. Unfortunately, circular convolution is of no use to us if our objective
is to determine the output of a linear filter to a given input sequence. In this case we seek
a frequency-domain methodology equivalent to linear convolution.

Suppose that we have a finite-duration sequence x (n) of length L which excites an FIR
filter of lenth M. Without loss of enerality, let

where h (n) is the impulse response of the FIR filter.

The output sequence y (n) of the FIR filter can be expressed in the time domain as the
convolution of x (n) and h (n), that is:

Since h (n) and x (n) are finite-duration sequences, their convolution is also finite in
duration. In fact, the duration of y (n) is L + M - 1.

The frequency-domain equivalent to (1) is:

If the sequence y (n) is to be represented uniquely in the frequency domain by samples of

its spectrum Y (co) at a set of discrete frequencies, the number of distinct samples must
equal or exceed L + M - 1. Therefore, a DFT of size , is required to
represent (y (n)) in the frequency domain.

Now if
www.auupdates.com
then ………………..(3)

where {X (k)) and {H (k)) are the N-point DFI’s of the corresponding
sequences x

(n) and h (n), respectively. Since the sequences x (n) and h (n) have a duration less than
N, we simply pad these sequences with zeros to increase their length to N. This increase
in the size of the sequences does not alter their spectra which are
continuous spectra, since the sequences are aperiodic. However, by sampling their spectra
at N equally spaced points in frequency (computing the N-point DFTs), we have
increased the number of samples that represent these sequences in the frequency domain
beyond the minimum number (L or M, respectively).

Since the N = L + M - I point DFT of the output sequence y (n) is sufficient to represent y
(n) in the frequency domain, it follows that the multiplication of the N-point DFTs X (k)
and H (k), according to (3), followed by the computation of the N-Point IDFT, must yield
the sequence {y (n)). In turn, this implies that the N-point circular convolution of x (n)
with h (n). In other words, by increasing the length of the sequences x (n) and h (n) to N
points (by appending zeros), and then circularly convolving the resulting sequences, we
obtain the same result as would have been obtained with linear convolution. Thus with
zero padding, the DFT can be used to perform linear filtering.

Filtering of Long Data Sequences : In practical applications involving linear filtering of

signals, the input sequence x (n) is often a very long sequence. This is especiäffy true in
some real-time signal processing applications concerned with signal monitoring and
analysis.

Since linear filtering performed via the DFT involves operations on a block of data,
which by necessity must be limited in size due to limited memory of a digital computer, a
long input signal sequence must be segmented to fixed-size blocks prior to processing.
Since the filtering is linear, successive blocks can be processed one at a time via the DFT
and the output blocks are fitted together to form the overall output signal sequence.

We now describe two methods for linear FIR filtering a long sequence on a block - by-
block basis using the DFT. The input sequence is segmented into blocks and each block
is processed via the DFT and IDFT to produce a block of output data. The output blocks
are fitted together to form an overall output sequence which is identical to the sequence
obtained if the long block had been processed via time-domain convolution.

The two methods are called the overlap-save method and the overlap-add method. For
both methods we assume that the FIR filter has duration M. The input data sequence is
www.auupdates.com
segmented into blocks of L points, where, by assumption, L>> M without loss of
generality.

Overlap-save method : In this method the size of the input data blocks is N = L + M -1
and the size of the DFTs and IDFT are of length N. Each data block consists of the last

m - 1 data points of the previous data block followed by L new data points to form a

data sequence of length N = L + M - 1. An N-point DFT is computed for each data block.

The impulse response of the FIR filter is increased in length by appending L — 1 zeros

and an N-point DFT of the sequence is computed once and stored. The multiplication of
the two N-point DFTs for the mth block of data yields..

Then the N-point IDFT yields the result

Since the data record is of length N, the first M — 1 points of are corrupted by
aliasing and must be discarded. The last L points of are exactly the same as the
result from linear convolution and, as a consequence,

To avoid loss of data due to aliasing, the last M — 1 points of each data record are saved
and these points become the first M - 1 data points of the subsequent retord, as indicated
above. To begin the processing, the first M - 1 points of the first record are set to zero.
Thus the blocks of data sequences are
www.auupdates.com
and so forth. The resulting data sequences from the IDFT are given by (8), where the first
M - I points are discarded due to aliasing and the remaining L points constitute the
desired result from linear convolution. This segmentation of the input data and the fitting
of the output data blocks together to form the output sequence are graphically illustrated
in fig.

Overlap-add method : In this method the size of the input data block is L points. and the
size of the DFTs and IDFT is N = L + M - 1. To each data block we append M - 1 zeros
and compute the N-points DFT. Thus the data blocks may be represented as:

and so on. The two N-point DF’Ts are multiplied together to form

The IDFT yields data blocks of length N that are free of aliasing since the size of

the DFTs nd IDFT is N = L + M - 1 and the sequences are increased to N-points by

appending zeros to each block.
www.auupdates.com

Since each data block is terminated with M - I zeros, the last M — I points from each
output block must be overlapped and added to the first M - I points of the succeeding
block. Hence this method is called the overlap-add method. This overlapping and adding
yields the output sequence.

The segmentation of the input data into blocks and the fitting of the output data

blocks to form the output sequence are graphically illustrated in fig. 11.

At this point, it may appear to the reader that the use of the DFT in linear FIR filtering is
not only an indirect method of computing the output of an FIR filter, but it may also be
more expensive compositionally since the input data must first be convected to the
frequency domain vai the DFT, multiplied by the DFT of the FIR filter, and

finally, converted back to the time domain via the IDFr. On the contrary, however, by

using the fast Fourier transform algorithm, as will be shown in Chapter 6, the DFTs and

IDFT require fewer computations to compute the output sequence than the direct
www.auupdates.com
realization of the FIR filter in the time

domain. This computational efficiency is the basic advantage of using the DFT to
compute the output of an FIR filter.

UNIT III IIR FILTER DESIGN

Q. 1. Determine the cascade and parallel realizations for the system described by the
system function

Cascaded Realization:

Ans.
www.auupdates.com

Parallel realization:

In terms of positive powers of z, H(z) can be written as

www.auupdates.com
www.auupdates.com

Parallel realization is shown in fig. below.

www.auupdates.com

Q. 2. Obtain the cascade realization of the system characterized by transfer

function.

Ans.
www.auupdates.com
Q. 3. Obtain the block diagram representations of Direct form I and II realizations
of the system with transfer function.

Ans. Given

Realization of H1(z) is shown below.

www.auupdates.com

Taking inverse z-transform

Realization of H2(z) is shown in fig. ahead.

Direct from-I realization is shown below

www.auupdates.com

Direct form II realization is shown below.

Q. 4. Find the block diagram representations of parallel realization of the system

with transfer function.

Ans. Given
www.auupdates.com
www.auupdates.com
The parallel realization is shown in fig. below.

Q. 5. Draw the canonical cascade realization diagram for the system given by:

Ans. Given

Taking z transform of equation given

www.auupdates.com

Canonical realization is obtained using shortcut method as shown below:

Q. 6. Explain with neat sketches, the cascade and parallel realization forms of
digital filters.

Ans. Cascade form structures : To obtain the cascade form realization, the numerator and
denominator of given transfer function H(z) is factored into the product of second order
terms.

Then the total transfer function H(z) is expressed as:

Here H1(z), H2(z) Hk(z) are second order polynomials Then each subtransfer function
(H1, H2 etc.) can be realised using direct form I or direct form II structures. The total
www.auupdates.com
transfer function is obtained by connecting all second order subsystems in series as
shown in fig below.

Now we have the general difference equation for discrete time Lii system given by.

Since H 1(z), H2(z) are second order polynomials.

We can write the second order differential equation for H k(z) by putting M = N =2 in eq.
(2).

We can obtain direct form-Il structure of eq. (3)

www.auupdates.com
So all such subsystems should be connected in series to obtain the cascade form
realization of hR system.

Parallel form structure:•

We have the general difference equation for HR system given by:

By using partial fraction expansion we can express the overall transfer function

H(z)as:

Here C is constant and are second order subsystems.

The general block schematic of parallel realization structure for hR system is shown
below:
www.auupdates.com
Here H1(z), H2(z) etc. can be realised by using direct-I or direct-Il structures. Then all
these structures are connected in parallel to obtain parallel form realization for hR
system.

Q. 7. Obtain direct form-I, direct form-Il, cascade and parallel structure for the
system described by:

Ans. We take z-transform both sides of difference equation, we have

1. Direct form-I

the order of

1. Direct form-Il : To obtain direct form-Il realization, we change H1(z) and

H2(z), then resulting structure is shown below.
www.auupdates.com
The direct form-Il realization is shown below:

1. Cascade form structure:

The cascade form structure is shown in fig.

1. Parallel form structure : To obtain parallel form structure, H(z) must be

expanded in partial fraction i.e.
www.auupdates.com
After some arithmetic calculation, we find that

The parallel form structure is shown below:

Q. 8. What are the limitations of IIR filter design by impulse invariance method ?
How are they over come by bilinear transformation method. Convert the analog
filter with system function.

into digital IIR filter by means of bilinear transformation.

Ans. In impulse invariant method we have replace analog filter by digital filter. The
limitations of impulse invariance method are given below:
www.auupdates.com
1. We know that

s
is analog frequency and its range is from . While

the digital frequency varies from That means from to maps

from

. Let k be any integer. Then we can write the general range of

butforthisrange also maps from . Thus

mapping from analog frequency to digital frequency is many to one. This

mapping is not one to one.

2. Analog filters are not band limited so there will be aliasing due to 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.

3. The change in the value of sampling time has no effect on the amount of

aliasing. Bilinear transformation method improves the limitations of in invariance

method as follows

1. There is one to one transformation from this domain to z domain.

2. The mapping is one to one.

3. There is no aliasing effect.

4. Stable analog filter is transformed into the stable digital ifiter.

Given

From eq. (1) we can say that

www.auupdates.com
The value of is given as

Now we will find the value of sampling time using the relation.

Using bilinear transformation H(z) can be obtained by putting, in

the equation of H(s)

This is the required transfer function for digital hR filter.

Q 10. Given an analog transfer function as

www.auupdates.com
Obtain H (z) using impulse invariant method. Take T =1s

Ans.

The corresponding digital filter is then

Q. 11. What is the difference between butterworth and chebyshev filters in terms of
frequency response?

Ans. Butterworth filters are all pole filters characterized by the magnitude required
frequency response.
www.auupdates.com
where N is the order of the filter, is its —3dB frequency (àsually called the out- off

frequency), is the passband edge frequency and is the band edge value

(i+e)

of The frequency response characteristics of Butterworth filters are shown in

figure below.

There are two types of chebyshev filters. Type I chebyshev filters are all pole filters that
exhibit equiripple behaviour in the passband and a monotonic characteristic in the
stopbond. On the other hand the family of type II Chebyshev filters contains both poles
and zeros and exhibits a monotonic behaviour in the passband. The magnitude squared of
the frequency response characteristics of a type I Chebyshev filter is given as:

Where e is a parameter of the filter related to the ripple in the passband and

is the Nth order chebyshev polynomial defined as

www.auupdates.com
The frequency response of chebyshev-I filter is given below.

Q. 12. For given analog filter system function into digital HR

filter by means of bilinear z-transformation. Digital filter is to have resonant

frequency

Ans. The given transfer function is:

……….(1)

From eq. (1) we can say that f = 4

The value of is given as

Now we will find out the value of sampling time (T5) using we relation.
www.auupdates.com
Using bilinear transformation H(z) can be obtained by puffing

in the eq. of H (s).

This is the required transfer function for digital hR filter.

Q. 13. An IIR low pass filter is to be designed to meet the following specifications.

(a) Pass-band frequency = 0 to 1.2 kHz

(b) Stop-band edge = 2KHz

www.auupdates.com
(c)Pass-band attenuation 8.5db

(d) Stop-band attenuation 15 db

Using Butterworth approximation and Bilinear transformation obtain the desired

IIR digital filter.

Ans. In the problem the specifications of required digital filter will be given and it will
be asked to design a particular, discrete time butlerworth’s filter. Then the following
steps should be used.

1. For Bilinear transformation method.

2. Calculate the order ‘N’ of filter using

www.auupdates.com

1. Calculate cut off frequency

1. Calculate the poles.

www.auupdates.com

1. Calculate the system transfer function of analog filter using

Q. 14. A chebyshev low pass filter has the following specifications:

(a) Order of the filter = 3

(b) Ripple in pass-band = 1 db

(c) Cut off frequency = 100 Hz

(d) Sampling frequency = 1 kHz.

Determine H(z) of the corresponding hR digital filter using bilinear transformation

technique.

Ans. First we will calculate the values of the edge frequency for analog filter.

Given : Order of filter = N 3

1. Calculation of required design specification of digital filter.

www.auupdates.com
1. Given

3. For normalized filter

4. Calculation of poles

First we calculate parameter

Now we will calculate the values of and R

5. Calculate roles:
www.auupdates.com

6 Calculation of system function

www.auupdates.com
7 Calculation of transfer function of digital filter

Q. 15. Using impulse invariant method, obtain the digital filter realization of the
analog filter shown in fig.

Ans. Let

Let us find y (t) i.e. the ‘impulse response .

The transformed circuit is shown below:

www.auupdates.com

Taking inverse Laplace transform on both sides, we get,

The analog system function is

The unit sample response derived by sampling every second is given by:
www.auupdates.com

Digital frequency response is given by

Digital filter system function is obtained by substituting z for

The digital filter structure is as follows:

Q. 16. Convert the analog filter with system functions

into the digital hR filter by means of the impulse invariance method.

www.auupdates.com
Ans. The partial fraction expansion of is given as:

The corresponding digital filter is then

It should be noted that zero of is not obtained by transforming the

zero at s = -z into a zero at

Q 17 Design a chebyshev filter for the following specification using (a) bilinear
transformation (b) Impulse invariance method.

Ans. Given
www.auupdates.com

Using bilinear transformation

www.auupdates.com
(b) Impulse Invariance Method:

Taking Inverse Laplace transform we obtain

Let

Taking z-Transform
www.auupdates.com
Assume T =1 sec.

Q. 18. Why frequency transformation in analog domain is done? Discuss in detail.

Ans. Frequency Transformations

Uptil now we have studied the design of low pass filter only. Now if it is asked to design
other filter like high pass, band pass or band reject filter then we have to use the
frequency transformation.

If the cut-off frequency of LPF is equal to I that means if then, it is called as

normalized filter. To design the other types of filters; first the system function of
normalized LPF is obtained. Then using frequency transformation we can get the system
function of the required filter. The following formulae are used for the frequency
transformation.

Let, we have a normalized L.P.F. having cut-off frequency

1.Low pass to low pass : Suppose it is asked to design another LPF with new passband
edge frequency Then use the transformation,

That means replace ‘s’ by in the given equation of H(s).

LP

2.Low pass to high pass: Suppose we have to design HPF with cut off frequency

then use,

3.Low pass to bandpass : Suppose we have to design bandpass filter with higher
www.auupdates.com
cutoff frequency and lower cut off frequency then use the transformation,

4.Low pass to bandstop or notch or band reject : Suppose we have to design notch filter
with higher cut-off frequency and lower cut off frequency then use the
transformation,

The following Table gives the summary of frequency transformation

Note : These formulae are applicable for analog frequency transformation. Similarly, we
can transform the LPF in digital domain by using the formulae for digital frequency
transformation.

UNIT IV FIR FILTER DESIGN

Q. 1. Obtain a cascade realization using minimum number of multiplications for the

system.
www.auupdates.com
Ans. We can consider that H(z) is product of factors

These two factors

symmetry. The realization is shown in fig below :
and

are having linear phase

Q. 2. Realize the system function.

by using direct form structure.

Ans. Given that

The realization is shown in fig below:

www.auupdates.com
Q. 3. Discuss signal flow graph representation and lattice form structures for FIR

systems.

Ans. Lattice Structure for FIR Filters : Consider Mth order FIR system with the transfer
function.

Here M denotes the degree of polynomial and is coefficient.

Thus when M = 0 we get

Basically is the transfer function which can be written as

Putting Equation (I) in Equation (3) we get.

Taking IZT of both sides we get,

Let M = I then Equation (5) becomes,

www.auupdates.com
In the simplest way Equation (6) can be realized as shown in Fig. (a).

Now the same output can be obtained by using the structure shown in Fig. (b). This

structure is called as single stage lattice structure. This structure provides two outputs

namely A1(n) and B 1(n).

A0 and B0 are constant multipliers.

In terms of x (n) we can write,

Similarly, the other output can be written as,

www.auupdates.com
In terms of x (n) we can write,

We know that Equation (7 (b)) is obtained by using single stage lattice structure.

Equations (6) and (7 (b)) are same if,

Similarly, Equation (6) and (7 (a)) are matching if,

Here ‘k’ is called as reflection coefficient.

That means the same output can be obtained using single stage lattice structure.

Now for M = 2; Equation (5) becomes,

As M = 2 we have to cascade two stages to obtain two stage lattice structure as shown in
Fig. (c).

From fig. (c) we can write,

Output of first stage:

www.auupdates.com
And the output of second stage is,

Now putting equation (10) in equation (12) we get,

From Equation (11) we can write,

Putting this value in equation (14) we get,

Observe that equations (9) and (15) are matching.

Here k1 and k2 are reflection coefficients The same way we can increase the number of
stages Finally Mth stage lattice structure is obtained as shown in Fig (d)
www.auupdates.com

Q. 4. Discuss quantization of filter coefficients in design of IIR and FIR filter.

Ans. Quantization of Filter Coefficients:

While implementing filter in hardware, the coefficient values must be quantized to the
accuracy of digital number system used in hardware. The filter coefficients should be
adjusted depending upon the number of bits supported by the hardware. Because of this,
the error is automatically introduced; which causes the frequency response of filter to
deviate from the desired response. It may happen that the positions of poles or zeros may
change.

How to minimize this error ? This coefficients quantization error can be minimized by
realizing higher order filter with either single pole or double pole filter sections. The
single pole filter sections have complex poles. This requires complex arithmetic
calculations. To avoid this, complex conjugate poles should be combined to form second
order filter sections.

What is the effect of quantization on FIR filter coefficients?

Generally digital FIR filters are designed such that they have linear phase characteristic
in the passband. If FIR filters are realized using direct form realization then linear phase
is maintained even when the quantization of filter coefficients is done. That means the
quantization does not affect the phase characteristics of FIR filter. But it affects the
magnitude response. To avoid this effect; the cascade form realization should be used and
12 to 14 bits should be used to represent the coefficients. Similarly the number of bits per
coefficient must be increased to maintain the same error in the frequency response
characteristics of the filter.

Let us say each filter coefficient is rounded to (b +1) bits. Then the maximum error in the
coefficient value is bounded as,

The error in the frequency response is given by,

www.auupdates.com
Round-off Effects in Digital filters

If in the realization of digital filter; one or more quantizers are used then, it results in a
non-linear characteristic. This characteristic may be significantly different than that of
ideal characteristic.

As a result of finite precision arithmetic operations performed in the digital filter; some
registers may overflow when the input signal level becomes large. The overflow is
defined as “When two numbers of ‘b’ bits are added and the sum cannot be represented
by b bits because of the overflow, then very large positive number becomes a very large
negative number and vice versa.

In case of recursive system; a feedback connection is present. So if there is an overflow

then it is fedback and used to compute next output, where it causes further overflow. This
creates undesired oscillations at the output. This results in non-linearity.

Because of this non-linearity effect; it becomes difficult to analyze the digital filters
precisely. To perform analysis of .quantization effects, a statistical characterization of
quantization effect is adopted.

Consider a digital filter which is characterized by the set of coefficients X1. With the
infinite precision realization, let the values of these coefficients be XR. Now at the
condition the magnitude of transfer function of the filter is exactly equal to 1.
This shows that the transfer function has zero sensitivity.

Reduction of Product Round-off Errors:

Whenever the multiplication operation is done in designing of digital filter, then the
rounding of the result of the multiplication is done. Because of this the errors are
generated. These errors are called as product round-off errors.

Now the quantization of the product of two numbers cause the overflow and the response
becomes non-linear. In order to obtain the more general result; the quantization errors in
the multiplication is modeled as an additive noise sequence, e (n).

Consider an additive noise model for the quantization error in a single pole system as
shown in fig.
www.auupdates.com
We can write the difference equation as follows,

The effect of rounding the product is modeled as a noise sequence e(n) and it is
added to the actual output .Equation (1) indicates that there are two

parts in the output. One is the response of system to the input x(n) and the other is the
response of system to the additive quantization noise e(n). Thus we can write Equation
(1) as,

Here y(n) = Actual output of the system = x(n) + av (n —1)

and q(n) Response of the system to the error sequence e (n).

Now as shown in Fig. the error sequence is fedback to the input. This reduces the
amount of noise power at the output.

Limit Cycles in IIR Filters

We have discussed the effect of quantization in the design of digital filter. We have also
studied that the non-linearities occur in the output because of the finite precision
arithmetic operations. This causes the periodic oscillations in the output; even when the
input is zero or constant. Such oscillations in the recursive systems are called as limit
cycles.

These limit cycles occur as a result of quantization effects in the multiplications. When
the input sequence x(n) becomes zero then the output of system enters into the limit
cycle. The output remains in the limit cycle until another input applied to the system to
drive the system out of the limit cycle.

There are two types of limit cycle as follows

www.auupdates.com
1.Granular limit cycle : This limit cycle has low amplitude. This limit cycle occurs when
the input applied to the system is very low.

2.Overflow limit cycle: As the name indicates; this limit cycle occurs because of the
overflow taking place in the implementation of digital filters.

Q. 5. For second order IIRfilter.

Ans. Given that

Direct form realization

Now can quantize the co-efficient by truncation

After quantizing the filter co-efficients, the transfer function of the hR system can

be written as:
www.auupdates.com
The direct form structure of H(z) is shown below:

If we compare the poles of H(z) and we can observe that both quantized and
unquantized poles of deviate very much from the original poles.

Q. 6. Describe the magnitude and phase response of FIR filters. How is linear phase
FIR filter defined.

Ans. The FIR filter can be characterized by its system function.

Which we consider as polynomial of degree N-1 in the variable .The roots of

this polynomial constitute the zeroes of the filter. The frequency response of eq. (1) is
givr’ .by

which is periodic in frequency with period

where is magnitude response and is phase response

www.auupdates.com
For FIR filters with linear phase we can define

where a is a constant phase delay in sample. Symmetric Condition

We can write

which gives us

By taking ratio. (3) to (2), we have

Simplifying the above equation we get

The eq. (4) will be zero when

www.auupdates.com
Therefore, FIR filters will have constant phase and when the impulse response is

symmetrical about , then its phase is piecewise linear. Fig. below shows the
property of. eq. (5), for N = 6 and N = 5.

In both the cases when N = 6 and N = 5 in the general case, the unit impulse

response sequence satisfying eq. (5). is symmetrical about For N odd, there is

one sample , that is not matched to any other sample.

Antisymmetric Condition:

For this

Then
www.auupdates.com
Therefore FIR filters have constant group delay and not constant phase delay when

impulse response is antisymmetric about

The filters that satisfy the above conditions and have a delay of N—i samples but

their impulse response are antisymmetric around the centre of the sequence, as opposed
to the true linear phase sequence that are symmetric around the centre of the sequence.

Magnitude Specifications : It is shown in fig below:

Hence
www.auupdates.com

Magnitude specification of FIR filter can be written as

Q 7 Discuss design of FIR filter using window method Also compare design using
Kaiser and Hanrnng Windows

Ans. FIR Filter Design using Windows Different types of windows are used to design
FIR filter First we will discuss the design of FIR filter using rectangular window The
rectangular window is as shown in fig.

It is denoted by Its magnitude is 1 for the range, n = 0 to M -1 Now ????? the

impulse response having infinite duration If hd (n) is multiplied by then impulse
response is obtained as shown in Fig (a) That means we will get only ??? not all
pulse Since we are truncating the input sequence by using a “process is called as
truncation process Since the shape of window function is react in called as rectangular
window.
www.auupdates.com

Magnitude response of rectangular window:

The rectangular window is defined as,

Let be infinite duration impulse response. We know that the finite impulse
response h(n) is obtained by multiplying

We will denote Fourier transform of . Thus using the definition of

fourier transform we can write,

Q. 8. Discuss the design of FIR filters by rectangular and hamming windows.

Compare their performance.
www.auupdates.com
Ans. The design of FIR filters by rectangular Window technique as given below.

The rectangular window is denoted as and it is defined as

We can obtain the spectrum of by taking fourier transform of equation (1) as

Hence the above equation becomes

Figure below illustrates the plot of equation (2). The transition width of the main labe is
approximate . The first sidelobe will be 13dB down the peak of main lobe and the
roll off will be at 20dB per decade. For a causal rectangular window, the frequency
response can be written as
www.auupdates.com
From equation (2) and (3), it may be noted that the linear phase response of the the causal

filter is written as and the non causal impulse response will have a
zero phase shift.

The design of FIR filters by Hamming Window technique as given below.

The Harmming window function is denoted as and is given by

The hamming window function (non-causal) is expressed as

It may be observed that the non-causal Hamming Window function is related to the

rectangular window function as shown below

The spectrum of Hamming window can be obtained as.

www.auupdates.com
We note that the width of the main lobe is approximately &r/M and the peak of the

first side lobe is at —43dB. The side lobe roll off is 20 dB/decade.

Q. 9. Represent system function using linear phase FIR structure.

Ans. FIR filter is said to be having a linear phase structure if its unit impulse

sequence is either symmetric or antisymmetric about some point in time. That means FIR
filter has linear phase if it satisfies the condition.

Here

The transfer function of FIR filter is given by

Let us split the summation into two points

But for linear phase we have

www.auupdates.com
Puffing this condition in the second summation of eq. (3) we get

eq. (4) can be written as,

Case (1) for even M:

We know that thus eq. (5) becomes

Now expanding the sjimmation we get,

www.auupdates.com ………(6)

Taking IZT of both sides we get,

………(7)

Let M = 6 Thus equation (7) becomes

……….(8)

The realization of eq (8) is shown below m fig

Case H For Odd M: -

When M is odd (let M = 5) we can simply write the difference as

www.auupdates.com
But we have the condition for
symmetry

Here M = 5. Then form eq. (10) we get, -

For

Thus eq. (9) becomes

The realization of eq. (11) is shown in fig below:

Q. 10. Design an ideal low pass filter with a frequency response

www.auupdates.com
Find the values of h (n) for N 1. Also find the filter transfer and frequency
magnitude frequency function.

Sol. Step 1. Draw the desired frequency response:

Given,

From the frequency response, we can find that the given response is a symmetrical

N odd response.

Step 2. To find

In general,
www.auupdates.com
Step 3. To find h(n):

For symmetric response,

So

But we know that

Step 4. To find the filter transfer function.

www.auupdates.com
Step 5. To find the realizable filter transfer function.

From the realizable filter transfer function, we have

Step 6. To find the magnitude frequency response

www.auupdates.com
Q. 11. Design a low pass FIR filter using hamming window to meet the following
specifications.

use 10 tap filter and obtain the impulse response of the desired filter.

Ans. The filter co-efficients are given by :

Given M = 10. The filter co-efficients are:

The hamming window function is

www.auupdates.com
The filter co-efficients of the resultant filter are then

Therefore

The frequency response is given by

Q. 12. Design an ideal band pass filter with a frequency response.

www.auupdates.com
Find the values of h(n) for N 7. Find the realizable filter transfer function and
magnitude function of

Sol. Step 1. Draw the ideal desired frequency response of bandpass filter.

Form the desired frequency response, we can find that the given response is symmetric N
odd

Step 2. To find

Step 3. To find h(n).

For symmetry response

www.auupdates.com

Step 4. To find filter transfer function,

Step 5. To find the realizable filter transfer function

Therefore, the filter co-efficients of the causal filters are,

Step 6. To find the magnitude response of

www.auupdates.com

Q. 13. Design an ideal band reject filter with a desired frequency response

Find the value of h(n) for N = 7. Find H(z) and

Sol. Step 1. Draw the ideal desired frequency response of band reject filtr.

From the desired frequency response, we can find that the given response is symmertric,
N-odd.

Step 2. To find
www.auupdates.com
In general,

Step 3. To find h(n)

For symetric response,

Step 4. To find filter transfer function,

www.auupdates.com
Step 5. To find the realizable filter transfer function.

Therefore, the filter co-efficients of the causal filters are,

Step 6. To find magnitude function

Q. 14. Design an ideal highpass filter with a frequency response

www.auupdates.com
Find the value of h(n) for N = 11 using

(a) Hamming window

(b) Hanning window.

Sol. (a) Hamming Window

Step 1. Draw the desired frequency response of ideal highpass filter.

From the desired frequency response, we can find that the given response is

symmetric.

Step 2. To find

We know that,

Step3. To find the Hamming window sequence.

www.auupdates.com

Step 4. To find the filter co-efficents

www.auupdates.com

Step 5. To find the filter co-efficients using Hamming window sequence.

Step 6. To find the transfer function of the filter.

www.auupdates.com
Step 7. To find transfer function of the realizable filter

The filter co-efficients of causal filters are,

(b) Hanning Window

Step 1. The filter co-efficients can be obtained from part (a), step (2) and step (5)

Step 2. To find the Hanning window sequence

The Hanning window sequence is given by

www.auupdates.com

Step 3. To find the filter co-efficients using Hanning window.

The filter co-efficients using Hanning window are

Step 4. To find the transfer function of the filter.

The transfer function of the filter is given by,

www.auupdates.com
Step 5. To find the transfer function of realizable filter.

Q. 15. Discuss signal flow graph representation and lattice structures for hR
systems.

Ans. It is noted that the computational algorithm of an LTI digital filter is conveniently
represented as block diagram using basic building blocks tepresenting the unit
delay or storage element, the multiplier and the adder. Figure shows these basic
building blocks.

As an example, for a first-order digital system (i.e., filter), we can write the difference
equation as under:

Figure illustrates the basic realization block diagram for equation (4) and also the
corresponding structure of the signal flow graph.
www.auupdates.com

Note : From fig., it is quite obvious that there is a direct correspondence between
branches in the digital realization structure and branches in the corresponding signal flow
graph. However, in the signal flow graph, the nodes represent both branch points and
adders in the digital realization block diagram.

Lattice Structure of an hR System: Let us consider an all-pole system with system

function

……….(1)

The difference equation for this hR system will be

………(3)

for N=1,we have

www.auupdates.com
The equation can be realised in lattice structure as shown in figure from which we

can obtain

Now, let us consider for the case N = 2, then, we have

This output can also be obtained from a two-stage lattice filter as shown in figure

from which we have.

www.auupdates.com

In a similar manner, we have

On comparing equation (9) and equation (17), we get

For a N-stage hR filter realized in lattice structure as shown in figure we get

Conversion from Lattice Structure to Direct form Structure For N = 3 we can write the
difference equation for an hR filter as below
www.auupdates.com
and from the lattice structure, we can write

Substituting m = 2 in equation (20) and in eqation (21), we get

And

Now, substituting equations (27) and equation (29) in eqauation (25), we obtain

comparing equation (23) and equation we get

The equation (32) canbe used to convert lattice structure to direct form.
www.auupdates.com
Q. 16. Discuss various steps for the design of linear phase FIR filters using window
method.

Ans. Design of Linear Phase FIR Filters Using Windows: Let us consider that the digital
filter which is to the designed have the frequency response This is also called the
desired frequency response. As discussed earlier, the required frequency response of a
digital filter is periodic in frequency and may be expanded in a Fourier series, i.e.,

Let the corresponding unit sample response (desired) be

Then, we have

The Fourier Series coefficients of the series h (n) are similar to the impulse response of a
digital filter. Infect, there are two difficulties in the implementation of expression in
equation (33) for designing a digital filter. Firstly, the impulse response is of infinite
duration and secondly, the filter is non-causal and unrealisable. No finite amount of delay
can make the impulse response realisable. Therefore, the filter resulting from a Fourier
series representation of is an unrealisable hR filter.

The finite duration impulse response may be converted to a finite duration impulse
response simply by truncating the infinite series n ± N. However, this results in
undesirable oscillations in the passband and stop band of the digital filter. This is because
of the slow convergence of the Fourier series near the points of discontinuity. These
undesirable oscillations can be reduced by using a set of time, limited weighting
functions, to known as window functions, to modify the Fourier coefficients. The
windowing technique has been illustrated in Fig.
www.auupdates.com

The required frequency response and its Fourier coefficients {h (n)} have been
shown at the top of this figure The finite duration weighting function and is Fourier
transform are shown in the second row. The Fourier transform of the weighting
function contains a main label, which consists of most of the energy of the

window function and side lobes which decay rapidly. The sequence

is obtained to get an FIR approximation of The sequence is exactly zero

outside the interval The sequence and its Fourier transform

www.auupdates.com
are shown In the thfrd row. is nothing but thecircular convolution of
and The realisable causal sequence g (n), which is obtained by shifting is
shown in the !ast row and this can be used as the desired filter impulse response.

A major effect of windowing is that the discontinuities in are converted into transi,
ion bands between values on either side of the discontinuity. The width of these transition
bands depends on the width of the main lobe of produces approximation errors for
all w. Based on the above discussion, the desirable characteristics can be listed as under

(i) The Fourier transform of the window functin must have a small width of main
lobe having as mush of the total energy as possible.

(ii) The Fourier transform of the window function should have side lobeswhich
decrease in energy rapidly as w tends to r. Some of the most frequently used window
functions are described in the following sections to follow

Rectangular Window Function (i.e., Technique)

The rectangular window is denoted as and it is defined as

We can obtain the spectrum of by taking Fourier transform of equation as

www.auupdates.com
Hence, above equation becomes

Figure illustrates the plot of eqation. The transition width of the main lobe is

approximaty . The first sidelobe will be 13dB down the peak of the main lobe and
the roll off will be at 20 d19 per decade. For a causal rectangular window, the frequency
resiponse can be written as:

From above equations, it may be noted that the linear phase response of the causal

filter is written as and the non-causal impulse response will have a

zero phase shift.

Q.17 Discuss about the effects of finite word length precision in digital filters.

Rounding Noise
When a signal is sampled or a calculation in the computer is performed, the
results must be placed in a register or memory location of fixed bit length. Rounding the
value to the required size introduces an error in the sampling or calculation equal to the
value of the lost bits, creating a nonlinear effect. Typically, rounding error is modeled as
a normally distributed noise injected at the point of rounding. This model is linear and
allows the noise effects to be analyzed with linear theory, something we can handle. The
www.auupdates.com
noise due to rounding is assumed to have a mean value equal to zero and a variance given
in Equation 3:

For a derivation of this result, see Discrete Time Signal Processing.1 Truncating
the value (rounding down) produces slightly different statistics. Multiplying two B-bit
variables results in a 2B-bit result. This 2B-bit result must be rounded and stored into a
B-bit length storage location. This rounding occurs at every multiplication point.
Scaling We don't often think about scaling when using floating-point calculations
because the computer scales the values dynamically. Scaling becomes an issue when
using fixed-point arithmetic where calculations would cause over- or under flow. In a
filter with multiple stages, or more than a few coefficients, calculations can easily
overflow the word length. Scaling is required to prevent over- and under flow and, if
placed strategically, can also help offset some of the effects of quantization.

Finite Precision Effects in Digital Filters

Causal, linear, shift-invariant discrete time system difference equation:

Z-Transform:

where is the Z-Transform Transfer Function,

www.auupdates.com
and is the unit sample response

Where:

 Is the sinusoidal steady state magnitude frequency

response

 Is the sinusoidal steady state phase frequency response

is the Normalized frequency in radians

if

then
www.auupdates.com

If the input is a sinusoidal signal of frequency , then the output is a

sinusoidal signal of frequency (LINEAR SYSTEM)

If the input sinusoidal frequency has an amplitude of one and a phase of zero, then
the output is a sinusoidal (of the same frequency) with a magnitude

and phase

So, by selecting and ,

can be determine in terms of the filter order and coefficients:

:
(Filter Synthesis)
www.auupdates.com
If the linear, constant coefficient difference equation is implemented directly:

There are two main effects which occur when finite precision arithmetic is used to
implement a DIGITAL FILTER: Multiplier coefficient quantization, Signal quantization

1. Multiplier coefficient quantization

The multiplier coefficient must be represented using a finite number of bits. To do this
the coefficient value is quantized. For example, a multiplier coefficient:

might be implemented as:

The multiplier coefficient value has been quantized to a six bit (finite precision) value.
The value of the filter coefficient which is actually implemented is 52/64 or 0.8125
AS A RESULT, THE TRANSFER FUNCTION CHANGES!

The magnitude frequency response of the third order direct form filter (with the gain or
scaling coefficient removed) is:

2. Signal quantization
www.auupdates.com
The signals in a DIGITAL FILTER must also be represented by finite, quantized binary
values. There are two main consequences of this: A finite RANGE for signals (I.E. a
maximum value) Limited RESOLUTION (the smallest value is the least significant bit)
For n-bit two's complement fixed point numbers:

If two numbers are added (or multiplied by and integer value) then the result can be
larger than the most positive number or smaller than the most negative number. When
this happens, an overflow has occurred. If two's complement arithmetic is used, then the
effect of overflow is to CHANGE the sign of the result and severe, large amplitude
nonlinearity is introduced.
For useful filters, OVERFLOW cannot be allowed. To prevent overflow, the digital
hardware must be capable of representing the largest number which can occur. It may be
necessary to make the filter internal word length larger than the input/output signal word
length or reduce the input signal amplitude in order to accommodate signals inside the
DIGITAL FILTER.
Due to the limited resolution of the digital signals used to implement the DIGITAL
FILTER, it is not possible to represent the result of all DIVISION operations exactly and
thus the signals in the filter must be quantized.

The nonlinear effects due to signal quantization can result in limit cycles - the filter
output may oscillate when the input is zero or a constant. In addition, the filter may
exhibit dead bands - where it does not respond to small changes in the input signal
amplitude. The effects of this signal quantization can be modeled by:
www.auupdates.com
where the error due to quantization (truncation of a two's complement number) is:

By superposition, the can determine the effect on the filter output due to each
quantization source. To determine the internal word length required to prevent overflow
and the error at the output of the DIGITAL FILTER due to quantization, find the GAIN
from the input to every internal node. Either increases the internal wordlengh so that
overflow does not occur or reduce the amplitude of the input signal. Find the GAIN from
each quantization point to the output. Since the maximum value of e(k) is known, a
bound on the largest error at the output due to signal quantization can be determined
using Convolution Summation. Convolution Summation (similar to Bounded-Input
Bounded-Output stability requirements):

If

then

is known as the norm of the unit sample response. It is a necessary and

sufficient condition that this value be bounded (less than infinity) for the linear system to
be Bounded-Input Bounded-Output Stable.

The norm is one measure of the GAIN.

www.auupdates.com

An alternate filter structure can be used to implement the same ideal transfer function.

Note that the effects of the same coefficient quantization as for the Direct Form filter (six
bits) does not have the same effect on the transfer function. This is because of the
reduced sensitivity of this structure to the coefficients. (A general property of integrator
based ladder structures or wave digital filters which have a maximum power transfer
characteristic.)

Note that all coefficient values are less than unity and that only three multiplications are
required. There is no gain or scaling coefficient. More adders are required than for the
direct form structure.
www.auupdates.com
Of course, a finite-duration impulse response (FIR) filter could be used. It will still have
an error at the output due to signal quantization, but this error is bounded by the number
of multiplications. A FIR filter cannot be unstable for bounded inputs and coefficients
and piecewise linear phase is possible by using symmetric or anti-symmetric coefficients.
But, as a rough rule an FIR filter order of 100 would be required to build a filter with the
same selectivity as a fifth order recursive (Infinite Duration Impulse Response - IIR)
filter.
Effects of finite word length
Quantization and multiplication errors
Multiplication of 2 M-bit words will yield a 2M bit product which is or to an M bit word.
Truncated rounded
Suppose that the 2M bit number represents an exact value then:
Exact value, x' (2M bits) digitized value, x (M bits) error e = x - x'
Truncation
x is represented by (M -1) bits, the remaining least significant bits of x' being discarded

Quantization errors
Quantization is a nonlinearity which, when introduced into a control loop, can lead to or
Steady state error
www.auupdates.com
Limit cycles
Stable limit cycles generally occur in control systems with lightly damped poles detailed
nonlinear analysis or simulation may be required to quantify their effect methods of
reducing the effects are:
- Larger word sizes
- Cascade or parallel implementations
- Slower sample rates
Integrator Offset
Consider the approximate integral term:

Practical features for digital controllers

Scaling
All microprocessors work with finite length words 8, 16, 32 or 64 bits.
The values of all input, output and intermediate variables must lie within the
Range of the chosen word length. This is done by appropriate scaling of the variables.
The goal of scaling is to ensure that neither underflows nor overflows occur during
arithmetic processing
Range-checking
Check that the output to the actuator is within its capability and saturate
the output value if it is not. It is often the case that the physical causes of saturation are
variable with temperature, aging and operating conditions.
Roll-over
Overflow into the sign bit in output data may cause a DAC to switch from a high positive
Value to a high negative value: this can have very serious consequences for the actuator
and Plant.
Scaling for fixed point arithmetic
Scaling can be implemented by shifting
binary values left or right to preserve satisfactory dynamic range and signal to
quantization noise ratio. Scale so that m is the smallest positive integer that satisfies the
condition
www.auupdates.com