SECA1301 Notes

SCHOOL OF ELECTRICAL AND ELECTRONICS
DEPARTMENT OF ELECTRONICS AND COMMMUNICATION

ENGINEERING
UNIT - I
SIGNALS AND SYSTEMS – SECA1301
1
1.INTRODUCTION TO SIGNALS
Signal: Signals are represented mathematically as functions of one or more independent
variables. It mainly focuses attention on signals involving a single independent variable.
For convenience, this will generally refer to the independent variable as time. It is defined
as physical quantities that carry information and changes with respect to time.
Ex: voice, television picture, telegraph.
Continuous Time signal – If the signal is defined over continuous-time, then the signal is a
continuous- time signal.
Ex: Sinusoidal signal, Voice signal, Rectangular pulse function
Discrete Time signal – If the time t can only take discrete values, such as t=kTs is called
Discrete Time signal
Unit Step Signal:
The Unit Step Signal u(t) is defined as
Graphically it is given by
2
Ramp Signal:
Graphically it is given by
Pulse Signal:
A signal is having constant amplitude over a particular interval and for the
remaining interval the amplitude is zero.
Impulse Signal:
Impulse Signal DT representation
Impulse Signal CT representation

3
Exponential Signal:
Exponential signal is of two types. These two type of signals are real exponential signal and
complex exponential signalwhich are given below.
Real Exponential Signal: A real exponential signal is defined as x(t)=Ae σt

Complex exponential Signal: The complex exponential signal is given by x(t)=Ae st where
s=σ+jω
Basic Operations on signals:
Several basic operations by which new signals are formed from given signals are familiar
from the algebra and calculus of functions.
1. Amplitude Scaling :y(t)= a x(t), where a is a real (or possibly complex) constant. C
x(t) is a amplitude scaled version of x(t) whose amplitude is scaled by a factor C.
2. Amplitude Shift: y(t)= x(t)+ b, where b is a real (or possibly complex) constant
4
3. Signal Addition: y(t)= x1(t) + x2(t)
5
As seen from the diagram above,
-10 < t < -3 amplitude of z(t) = x1(t) + x2(t) = 0 + 2 = 2
-3 < t < 3 amplitude of z(t) = x1(t) + x2(t) = 1 + 2 = 3
3 < t < 10 amplitude of z(t) = x1(t) +x2(t) = 0 + 2 = 2
4. Signal Multiplication: y(t)= x1(t). x2(t)
As seen from the diagram above,
-10 < t < -3 amplitude of z (t) = x1(t) ×x2(t) = 0 ×2 = 0
-3 < t < 3 amplitude of z (t) = x1(t) ×x2(t) = 1 ×2 = 2
3 < t < 10 amplitude of z (t) = x1(t) × x2(t) = 0 × 2 = 0
5. Time Shift: If x(t) is a continuous function, the time-shifted signal is y(t)  x(t  t0 ) .
defined as
6
If t0 > 0, the signal is shifted to the right, and if t0< 0, the signal is shifted to the left.
x(t ± t0) is time shifted version of the signal x(t).

x (t + t0) →→ negative shift
x (t - t0) →→ positive shift
6. Time Reversal: If x(t) is a continuous function, the time-reversed signal is defined

as y(t) = x(-t). x(-t) is the time reversal of the signal x(t).
7. Time Scaling: If x(t) is a continuous function, a time-scale version of this signal is defined as y(t)
= x(at). If a>1, the signal y(t) is a compressed version of x(t), i.e., the time interval is compressed
1
to . If 0<a< 1, the signal y(t) is a stretched version of x(t), i.e., the time interval is stretched by
a
1
. When operating on signals, the time-shifting operation must be performed first, and then the
a
time-scaling operation is performed. x(At) is time scaled version of the signal x(t). where A is
always positive.
|A| > 1 →→ Compression of the signal
|A| < 1 →→ Expansion of the signal
7
1. A triangular pulse signal x(t) is depicted below.
x(t)
1
-1 0 1 t
Sketch each of the following signals:
(a) x(3t)
(b) x(3t  2)
(c) x(2t 1)
(d) x(0.5t 1)
2. Draw the waveform x(-t) and x(2-t) of the signal x(t) = t 0≤t≤3
0 t>3
X(t)
X(-t)
0 1 2 3 t
-3 -2 -1 0 t
X(2-t)
-1 0 1 2 t
8
Classification of DT and CT Signals:
1. Even and Odd signal

2. Deterministic and Random Signal
3. Periodic and Aperiodic signal
4. Energy and Power signal
Even and Odd Signal:
An even signal is any signal 'x' such that x(t) = x(-t). Odd signal is a signal 'x' for which
x(t) = -x(-t). The even and odd parts of a signal x(t) are
The even and odd parts of a signal are given by
1
x (t)  x(t)  x(t)

e
2
1
x (t)  x(t)  x(t)

o
2
Here xe(t) denotes the even part of signal x(t) and xo(t) denotes the odd part of signal x(t).
Deterministic Signal:
Deterministic signals are those signals whose values are completely specified for any given
time. Thus, a deterministic signals can be modeled exactly by a mathematical formula are
known as deterministic signals.
Random (or) Nondeterministic Signals:
Nondeterministic signals and events are either random or irregular. Random signals are
also called non deterministic signals are those signals that take random values at any given
time and must be characterized statistically.Random signals, on the other hand, cannot be
described by a mathematical equation they are modeled in probabilistic terms.
Periodic signal:
A CT signal x(t) is said to be periodic if it satisfies the following property: x(t)=x(t+T) at all
time t, where T=Fundamental Time Interval (T=2π/ω)
Ex:
1. x(t)=sin(4πt). It is periodic with period of 1/2

2. x(t)=cos(3πt). It is periodic with period of 2/3
9
Aperiodic Signal:
A CT signal x(t) is said to be periodic if it satisfies the following property: x(t)≠x(t+T) at all
time t, where T=Fundamental Time Interval
Energy Signal:
The Energy in the signal is defined as :
Power Signal:
The Power in the signal is defined as
If 0<E<∞ then the signal x(t) is called as Energy signal. However there are signals where
this condition is not satisfied. For such signals we consider the power. If 0<P<∞ then the
signal is called a power signal. Note that the power for an energy signal is zero (P=0) and
that the energy for a power signal is infinite (E=∞). Some signals are neither energy nor
power signals.
1. Draw the signal x(n) = u(n) – u(n-3)
u(n)
u(n-3)
1
1
0 1 2 3 4 5 6 7 n
0 1 2 3 4 5 6 7 8 9 n
X(n)=u(n)-u(n-3)
0 12 n
10
2. What is the total energy of the discrete time signal x(n) which takes the value of
unity at n= - 1,0,1?
Energy of the signal is given as,

∞ 1
E = ∑ │x(n)│2 = ∑ │x(n)│2
n = -∞ n = -1
= │x(-1)│2 + │x(0)│2 + │x(1)│2 = 3
3. Determine if the following signals are Energy signals, Power signals, or neither, and evaluate E
and P for each signal a(t )  3sin(2t ),   t   ,
 
Ea   | a(t ) | dt   | 3sin(2t ) |2 dt
2
 

1
9
2
1 cos(4 t )dt

 
1
9 dt  9  cos(4t )dt

2 
 J
1 1
1
Pa   | a(t ) 2
| dt  | 3sin(2t ) |2 dt
10 0
1
11
2 1 cos(4t )dt
1
9
0
0 1
 9 1 dt  9cos(4t )dt
0
2 0
1
9  9 
   sin(4 t ) 
2  4 0
9
 W
2
So, the energy of that signal is infinite and its average power is finite (9/2). This means that it
is a power signal as expected. It is a power signal.
Real and Complex signals:
Exponential signal is of two types. These two type of signals are real exponential signal and
complex exponential signalwhich are given below.
Real Exponential Signal: A real exponential signal is defined as x(t)=Aeσt
Complex exponential Signal:The complex exponential signal is given by x(t)=Ae st where

s=σ+jω
TEXT / REFERENCE BOOKS

1. P.Ramesh Babu et al., “Signals and Systems”, 4th Edition, Scitech Publishers, 2017.
2. Rodger E. Ziemer , William H Tranter, D. R. Fannin,”Signals and Systems: Continuous
and Discrete”, 4th Edition, Pearson Education India, 2014.
3. Haykin S. and Van Been B., “Signals and Systems”, 2nd Edition, John Wiley and Sons,
2015.
4. H.P. Hsu, "Signals and Systems", 2nd Edition, Tata McGraw Hill, 2017.
12
ENGINEERING
UNIT - II
1
II. ANALYSIS OF CONTINUOUS TIME SIGNALS
Continuous Time Fourier Transform
Any continuous time periodic signal x(t) can be represented as a linear combination of
complex exponentials and the Fourier coefficients ( or spectrum) are discrete. The
Fourier series can be applied to periodic signals only but the Fourier transform can also
be applied to non-periodic functions like rectangular pulse, step functions, ramp
function etc. The Fourier transform of Continuous Time signals can be obtained from
Fourier series by applying appropriate conditions.
The Fourier transform can be developed by finding Fourier series of a periodic function
and the tending T to infinity.
Representation of Aperiodic signals: Starting from the Fourier series
representation for the continuous-time periodic square wave:
2.1
The Fourier coefficients akfor this square wave is
2.2
2
or alternatively 2.3
where 2 sin(wT1 ) / w represent the envelope of Tak
When T increases or the fundamental frequency w 0 2p / T decreases, the envelope is

sampled with a closer and closer spacing. As T becomes arbitrarily large, the original
periodic square wave approaches a rectangular pulse.
Tak becomes more and more closely spaced samples of the envelope, as T →∞, the
Fourier series coefficients approaches the envelope function.
This example illustrates the basic idea behind Fourier’s development of a representation for
aperiodic signals.
Based on this idea, we can derive the Fourier transform for aperiodic signals.
From this aperiodic signal, we construct a periodic signal (t) , shown in the figure below.
2
2.4
2.5
2.6
Define the envelope X ( jw ) of Tak as,
2.7
3
2.8
Equation 2.8 becomes representation of x(t). In addition the right hand side of equation
becomes an integral.
This results in the following Fourier Transform.
2.9
2.10
Convergence of Fourier Transform
If the signal x(t ) has finite energy, that is, it is square integrable,
Then we guaranteed that X( jw ) is finite or equation 2.10 converges. If
We have
An alterative set of conditions that are sufficient to ensure the convergence:
Contition1: Over any period, x(t ) must be absolutely integrable, that is
Condition 2: In any finite interval of time, x(t ) have a finite number of maxima and minima.
4
Condition 3: In any finite interval of time, there are only a finite number of discontinuities.
Furthermore, each of these discontinuities is finite.
Examples of Continuous-Time Fourier Transform
If a is complex rather than real, we get the same result if Re{a}>0
The Fourier transform can be plotted in terms of the magnitude and phase, as shown in
the figure below.
5
Example
Calculate the Fourier transform of the rectangular pulse signal
The inverse Fourier Transform of the sinc function is
Comparing the results we have,
FT
Square wave Sinc Function
FT-1
This means a square wave in the time domain, its Fourier transform is a sinc function.
However, if the signal in the time domain is a sinc function, then its Fourier transform is a
square wave. This property is referred to as Duality Property.
We also note that when the width of X( jw) increases, its inverse Fourier transform x(t) will
be compressed. When W → ∞, X( jw) converges to an impulse. The transform pair with
several different values of W is shown in the figure below.
6
The Fourier Transform for Periodic Signals
Properties of Fourier Transform
1. Linearity
If x(t)FX( jw) y(t)FY( jw)
then
ax(t)  by(t)FaX ( jw)  bY ( jw)

2. Time Shifting
If x(t)FX( jw)
Then
F
x(t-t0) e-jwt0X(jw)
7
3. Conjugation and Conjugate Symmetry
F
If x(t) X(jw)
Then
F
x*(t) X*(-jw)
4. Differentiation and Integration
5. Time and Frequency Scaling
From the equation we see that the signal is compressed in the time domain, the spectrum will
be extended in the frequency domain. Conversely, if the signal is extended, the corresponding
spectrum will be compressed.
8
If a = -1, we get from the above equation,
𝐹
X(-t) 𝑋(𝑗𝜔)
That is reversing a signal in time also reverses its Fourier transform.
6. Duality
The duality of the Fourier Transform can be demonstrated using the following
example
For any transform pair, there is a dual pair with the time and frequency variables
interchanged.
9
Parseval’s Relation
If x(t) 𝑋(𝑗𝜔),
∞ ∞
1
𝑥(𝑡) 2𝑑𝑡 = 𝑋(𝑗𝑤) 2 𝑑𝜔
We have, −∞ 2𝜋 −∞
Parseval‟s relation states that the total energy may be determined either by computing
2
the energy per unit time 𝑥(𝑡) and integrating over all time or by computing
2
the energy per unit frequency 𝑥(𝑗𝜔) 2𝜋 and integrating over all
10
frequencies. For this reason, 𝑥(𝑗𝜔) 2 is often referred toas the energy density spectrum.
The Parseval's theorem states that the inner product between signals is preserved in
going from time to the frequency domain. This is interpreted physically as “Energy
calculated in the time domain is sameas the energy calculated in the frequency domain”
The convolution properties
𝑦 (𝑡 )= ℎ (𝑡) ∗ 𝒙 (𝑡)= 𝑌 (𝑗𝜔) = 𝐻 𝑗𝜔 𝑋(𝑗𝜔)
The equation shows that the Fourier transform maps the convolution of two signals
into product of their Fourier transforms.
H(j𝜔), the transform of the impulse response is the frequency response of the LTI
system, which also completely characterizes an LTI system.
Example
The frequency response of a differentiator.

𝑑𝑥 (𝑡)
y(t)= 𝑑𝑡
From the differentiation property,
Y(j𝜔)= j𝜔𝑋 𝑗𝜔
The frequency response of the differentiator is,
H(j𝜔)= 𝑌(𝑗𝜔 ) =j𝜔

𝑋(𝑗𝜔 )
The Multiplication Property
Multiplication of one signal by another can be thought of as one signal to scale or

modulate the amplitude of the other, and consequently, the multiplication of two signals
is often referred to as amplitude modulation.
Laplace Transform
The Laplace Transform is the more generalized representation of CT complex

exponential signals. The Laplace transform provide solutions to most of the signals
11
and systems, which are not possible with Fourier method. The Laplace transform can
be used to analyze most of the signals which are not absolutely integrable such as the
impulse response of an unstable system. Laplace Transform is a powerful tool for
analysis and design of Continuous Time signals and systems. The Laplace Transform
differs from Fourier Transform because it covers a broader class of CT signals and
systems which may or may not be stable.
Till now, we have seen the importance of Fourier analysis in solving many problems
involving signals. Now, we shall deal with signals which do not have a Fourier
transform. We note that the Fourier Transform only exists for signals which can
absolutely integrated and have a finite energy. This observation leads to generalization
of continuous-time Fourier transform by considering a broader class of signals using the
powerful tool of "Laplace transform". With this introduction let us go on to formally
defining both Laplace transform.
Definition of Laplace Transform
The Laplace transform of a function x(t) can be shown to be,
This equation is called the Bilateral or double sided Laplace transform equation.
This equation is called the Inverse Laplace Transform equation, x(t) being called the
Inverse Laplace transform of X(s).
The relationship between x(t) and X(s) is

LT
x(t) X(s)
Region of Convergence (ROC):
The range of values for which the expression described above is finite is called as the
Region of Convergence (ROC).
Convergence of the Laplace transform
The bilateral Laplace Transform of a signal x (t) exists if
12
X(s)= ∞ 𝑥(𝑡)𝑒−𝑠𝑡 𝑑𝑡
−∞
Substitute s=𝜎 + 𝑗𝜔
∞
X(s) =
−∞
𝑥(𝑡)𝑒−𝜎𝑡 𝑒−𝑗𝜔𝑡 𝑑𝑡
Relationship between Laplace Transform and Fourier Transform
The Fourier Transform for Continuous Time signals is in fact a special case of Laplace
Transform. This fact and subsequent relation between LT and FT are explained below.
Now we know that Laplace Transform of a signal 'x'(t)' is given by:
The s-complex variable is given by
But we consider𝜎 = 0 and therefore „s‟ becomes completely imaginary. Thus we have
s=jΩ. This means that we are only considering the vertical strip at𝜎 = 0.
From the above discussion it is clear that the LT reduces to FT when the complex
variable only consists of the imaginary part. Thus LT reduces to FT along the jΩ axis.
(imaginary axis)
Fourier Transform of x (t) = Laplace Transform of 𝑥(𝑡) 𝑠=𝑗Ω
Laplace transform becomes Fourier transform
if σ=0 and s=jω.
13
Example of Laplace Transform
(1) Find the Laplace transform and ROC of x(t) = 𝑒−𝑎𝑡 𝑢(𝑡)
we notice that by multiplying by the term u(t) we are effectively considering the
unilateral Laplace Transform whereby the limits tend from 0 to +∞
Consider the Laplace transform of x(t) as shown below
(2) Find the Laplace transform and ROC of x(t)=-𝑒−𝑎𝑡 𝑢(−𝑡)
14
If we consider the signals e-atu(t) and -e-atu(-t), we note that although the signals are
differing, their Laplace Transforms are identical which is 1/( s+a). Thus we conclude
that to distinguish L.T's uniquely their ROC's must be specified.
Properties of Laplace Transform
1. Linearity
2. Differentiation in the time domain
3. Time Shift
4. Time Scaling
15
5. Multiplication
6. Time Reversal
When the signal x(t) is time reversed(180° Phase shift)
7. Frequency Shifting
8. Conjugation symmetry
9. Parseval’s Relation of Continuous Signal
It states that the total average power in a periodic signal x(t) equals the sum of the
average in individual harmonic components, which in turn equals to the squared
magnitude of X(s) Laplace Transform.
10. Differentiation in Frequency
When x(t) is differentiated with respect to frequency then,
16
11. Integration Property
When a periodic signal x(t) is integrated, then the Laplace Transform becomes,
12. Convolution Property
13. Initial Value Theorem
The initial value theorem is used to calculate initial value x(0 +) of the given sequence
x(t) directly from the Laplace transform X(S). The initial value theorem does not
apply to rational functions X(S) whose numerator polynomial order is greater than
the denominator polynomial orders.
The initial value theorem states that,

lim 𝑆𝑋 𝑆 = 𝑋(0+)
𝑠→∞
14. Final Value Theorem
It states that,
lim 𝑆𝑋 𝑆 = 𝑋(∞)
𝑠→∞

2. Rodger E. Ziemer , William H Tranter, D. R. Fannin,”Signals and Systems: Continuous and
Discrete”, 4th Edition, Pearson Education India, 2014.
3. Haykin S. and Van Been B., “Signals and Systems”, 2nd Edition, John Wiley and Sons, 2015.
17
ENGINEERING
UNIT - III
1
III.LINEAR TIME INVARIANT CONTINUOUS TIME SYSTEMS
System
A system may be defined as a set of elements or functional blocks which are connected
together and produces an output in response to an input signal. The response or output of
the system depends upon transfer function of the system. Mathematically, the functional
relationship between input and output may be written as
y(t)=f[x(t)]
Types of system
Like signals, systems may also be of two types as under:
1. Continuous-time system
2. Discrete time system
Continuous time System
Continuous time system may be defined as those systems in which the associated signals
are also continuous. This means that input and output of continuous – time system are
both continuous time signals.
For example:
Audio, video amplifiers, power supplies etc., are continuous time systems.
Discrete time systems
Discrete time system may be defined as a system in which the associated signals are also
discrete time signals. This means that in a discrete time system, the input and output are
both discrete time signals.
For example, microprocessors, semiconductor memories, shift registers etc., are discrete
time signals.
LTI system:-
Systems are broadly classified as continuous time systems and discrete time
systems. Continuous time systems deal with continuous time signals and discrete time
systems deal with discrete time system. Both continuous time and discrete time systems
have several basic properties. Out of these several basic properties of systems, two
properties namely linearity and time invariance play a vital role in the analysis of signals
and systems. If a system has both the linearity and time invariance properties, then this
system is called linear time invariant (LTI) system.
2
Characterization of Linear Time Invariant (LTI) system
Both continuous time and discrete time linear time invariant (LTI) systems exhibit one
important characteristics that the superposition theorem can be applied to find the response
y(t) to a given input x(t).
Hence, following steps may be adopted to find the response of a LTI system using
super position theorem:
1. Resolve the input function x(t) in terms of simpler or basic function like impulse
function for which response can be easily evaluated.
2. Determine individually the response of LTI system for the simpler input impulse
functions.
3. Using superposition theorem, find the sum of the individual responses, which will
become the overall response y(t) of function x(t).
From the above discussions, it is clear that to find the response of a LTI system to any given
function, first we have to find the response of LTI system input to an unit impulse called unit
impulse response of LTI system.
Hence, the impulse response of a continuous time or discrete time LTI system is the output
of the system due to an unit impulse input applied at time t=0 or n=0.
x(t)=δ(t ) CT system y(t)=h(t)
Here, δ(t) is the unit impulse input in continuous time and h(t) is the unit impulse response of
continuous time LTI system. Continuous time unit impulse response h(t) is the output of a
continuous time system when applied input x(t) is equal to unit impulse function δ(t)
x(n)=δ(n) DT system y(n)=h(n)
Similarly, for a discrete time system, discrete time impulse response h(n) is the output of a
discrete time system when applied input x(n) is equal to discrete time unit impulse function
δ(n). Here, δ (n) is the unit impulse input in discrete time and h(n) is the unit impulse
response of discrete time LTI system. Therefore, any LTI system can be completely
characterized in terms of its unit impulse response.
Properties of Linear time invariant (LTI) system:-
The LTI system has number of properties not exhibited by other systems. Those are as under:
 Commutative property of LTI systems

 Distributive property of LTI systems
 Associative property of LTI systems
 Static and dynamic LTI systems
 Invertibility of LTI systems
 Causality of LTI systems
 Stability of LTI systems
 Unit-step response of LTI systems
3
Commutative property:
The commutative property is a basic property of convolution in both continuous and

discrete time cases, thus, both convolution integral for continuous time LTI systems and
convolution sum for discrete time LTI systems are commutative. According to the property,
for continuous time LTI system. The output is given by
Thus, we can say that according to this property, the output of a continuous time LTI system
having input x(t) and unit impulse h(t) is identical to the output of a continuous time LTI
system having input h(t) and the unit impulse response x(t).
Distributive property:
The distributive property states that both convolution integral for continuous time LTI
system and convolution sum for discrete time LTI system are distributive.
For continuous time LTI system, the distributive property is
expressed as The output, y(t) = x(t) ∗ [ℎ1(t) + ℎ2(t)]
Or
y(t) = x(t) ∗ ℎ1(t) + x(t) ∗ ℎ2(t)
Thus, the two continuous time LTI systems, with impulse responses h1(t) and h2(t), have
identical inputs and outputs are added as
y1(t) = x(t) ∗ ℎ1(t)
y2(t) = x(t) ∗ ℎ2(t)
The output y(t) = y1(t) + y2(t)
y(t) = x(t) ∗ ℎ1(t) + x(t) ∗ ℎ2(t)
4
h1(t)
input x (t) + outputy(t)
h2(t)
x(t) h1(t)+h2(t) y(t)
Fig 3.1The distributive property of convolution integral for a parallel interconnection of

continuous time LTI system
Associative Property of LTI system:
According to associative property, both convolution integral for continuous time LTI
systems and convolution sum for discrete time LTI systems are associative.
For continuous time LTI system, according to associative property,
The output
y(t) = x(t) ∗ [ℎ1(t) ∗ ℎ2(t)]
or
y(t) = [x(t) ∗ ℎ1(t)] ∗ ℎ2(t)
5
x(t) h1 (t) z ( t ) h2(t) y(t)
X(t)
h( t)=h1(t)+h2(t) y(t)
Fig 3.2: The associative property of convolution integral for a cascade interconnection of continuous
time LTI systems
Here we have y(t)=z(t) *h2(t)But z(t)=x(t)*h1(t). Therefore y(t)=[x(t)*h1(t)*h2(t)]
Static and Dynamic property:
Static systems are also known as memory less systems. A system is known as static if its
output at any time depends only on the value of the input at the same time. A continuous time
system is memory less if its unit impulse response h(t) is zero for t≠0. These memory less LTI
systems are characterized by y (t) =Kx (t) where K is constant. And its impulse response
h(t)=Kδ(t) .If K=1 ,then these systems are called identity systems.
Invertibility of LTI systems:
A system is known as invertible only if an inverse system exists which, when cascade with
the original system, produces and output equal to the input at first system. If an LTI system is
invertible then it will have a LTI inverse system. This means that we have a continuous time
LTI system with impulse response h(t) and its inverse system with impulse response h 1(t)
which results in an output equal to x(t). Cascade interconnection of original continuous time
LTI system with its inverse system is given as identity system.
Thus, the overall impulse response of a system with impulse response h(t) cascaded with its
inverse system with Impulse response h1(t) is given as h(t)*h1(t)=δ(t)
6
Fig3.3: An inverse system for continuous time LTI system
Causality for LTI System
This property says that the output of a causal system depends only on the present and past
values of the input to the system.
A continuous time LTI system is called causal system if its impulse response h(t) is
zero t<0. For a causal continuous time
LTI system, convolution integral is given as
For pure time shift with unit impulse response h(t)=δ(t-t0) is a causal continuous time LTI
system for t≥0. In this case time shift is known as a delay.
Stability for LTI systems
A stable system is a system which produces bounded output for every bounded input.
Condition of Stability for continuous time LTI system:
Let us consider an input x(t) that is bounded in magnitude |x(t)|<M for all values of t
Now, we apply this input to an continuous time LTI system with unit impulse response h(t).
Output of this LTI system is determined by convolution integral and is given by

7
Magnitude of output y(t) is given as
From the above equation we can conclude that if the impulse response h(t) is absolutely
integerable then output of a continuous time LTI system is bounded in magnitude, and thus,
the system is bounded input, bounded output(BIBO) stable.
Unit step response of an LTI system:
Unit step response is the output of a LTI system for input is equal to unit step function or
sequence. Unit step response of continuous time LTI system is found by convolution integral
of u(t) with unit impulse response h(t) and is expressed as
g(t)=u(t)*h(t)=h(t)*u(t)
according to commutative property. Therefore, unit step response g(t) may be viewed as the
response to the input h(t) of a continuous time LTI system with unit impulse response u(t).
Input x(t)=u(t) output y(t)=g(t)
Continuous time LTI

system
Fig 3.4: Continuous time LTI system
8
Classification of CT LTI system:-
The systems are classified into two types: continuous time and discrete time systems, Now
these two broad types of systems are further classified on the basis of system properties as
under:
 Causal system and non causal system
 Time invariant and time variant system
 Stable and unstable system
 Linear and Non-linear system
 Static and Dynamic systems
 Invertible and noninvertible system
9
Causal systems and Non-causal systems
A system is causal if the response or output does not begin before the input function is
applied. This means that if input is applied at t=t 0,then for causal system, output will depend
on values of input x(t) for t≤t0.
Mathematically,
y(t0)=f[x(t),t≤t0].
In other words we can say that, the response or output of the causal system to an input does
not depend on future values of that input but depends only on the present or past values of
the input.This means that all the real-time systems are also causal systems since these systems
cannot know the future values of the input signal when it constructs output signal.Thus,
causal systems are physically realizable. For example a resister is a continuous time causal
systembecause voltage across it is given by the expression v(t)=R.i(t) and output v(t),i.e.,
voltage depends only on the input i(t) i.e., current at the present time.
Time invariant and time variant system
A system is said to be time invariant if its input –output characteristics do not change with
time. H{x(t)}=y(t) implies that ,H{x(t-t0)}=y(t-t0) for every input signal x(t) and every time
shift t0A system is said to be time variant if its input- output characteristics changes with
time.
Procedure to Test for Time Invariance:-
1. Delay the input signal by t0 units of time and determine the response of the system for
this delayed input signal. Let this response be y(t-t0).
2. Delay the response of the system for undelayed input by t0 units of time. Let this
delayed response be yd(t).
3. Check whether y(t-t0)=yd(t). If they are equal then the system is time invariant.
Otherwise the system is time variant.
Stable and unstable system
A system is called bounded input, bounded output(BIBO) stable if and only if every bounded
input results in a bounded output. The output of such a system does not diverge or does not
grow unreasonably large.
Condition of Stability for continuous time LTI system:
Let us consider an input x(t) that is bounded in magnitude
|x(t)|<M<∞ for all values of t

10
Now, we apply this input to an continuous time LTI system with unit impulse response h(t).
Output of this LTI system is determined by convolution integral and is given by
Magnitude of output y(t) is given as
From the above equation we can conclude that if the impulse response h(t) is absolutely
integerable then output of a continuous time LTI system is bounded in magnitude, and thus,
the system is bounded input, bounded output(BIBO) stable.
The systems not satisfying the above conditions are unstable.
Linear and nonlinear system
A linear system is one that satisfies the superposition principle. The principles of
superposition requires that the response of the system to a weighted sum of the signals is
equal to the corresponding weighted sum of the responses of the system to each of the
individual input signals.
A system is linear if
H{a1x1(t)+a2x2(t)}=a1H{x1(t)}+a2H{x2(t)} for any arbitrary input sequences x1(t) and x2(t) and
for any arbitrary constants a1 and a2.
If a relaxed system does not satisfy the super position principle as given by the above
definition, then the system is nonlinear.
11
Static and dynamic system
A continuous time system is called static or memory less if its output at any instant t
depends on present input but not on the past or future samples of the input. These systems
contain no energy storage elements. This means that the equation relating its output signal to
its input signal contains no derivative, integrals or signal delays.
As an example, consider the system described by the following relationship
Y(t)=x2(t) this system is memory less because the value of the output signal y(t) at time t
depends only on the present value of the input signal x(t).In any other case the system is said
to be dynamic or to have memory. Dynamic systems have one or more energy storage
elements. Input output relationship of a dynamic continuous time system is described by its
differential equation.
Invertible and non invertible system
A system is said to be invertible if there is a one to one correspondence between its
input and output signals. If a system is invertible, then an inverse system exists. The
cascading of and invertible system and its inverse system is equivalent to the identity system.
12
The frequency response of an inverse system is basically reciprocal of the frequency response
of the original system or invertible system.
An example of an invertible continuous time system is given by
Convolution integral:-
The output of any general input may be found by convolving the given input signal x(t)
with the LTI systems unit impulse response h(t).
 Commutative property
 Distributive property
 Associative property
Commutative property:
13
The commutative property is a basic property of convolution in both continuous and
discrete time cases, thus, both convolution integral for continuous time LTI systems and
convolution sum for discrete time LTI systems are commutative. According to the property,
for continuous time LTI system. The output is given by
Thus, we can say that according to this property, the output of a continuous time LTI system
having input x(t) and unit impulse h(t) is identical to the output of a continuous time LTI
system having input h(t) and the unit impulse response x(t).
The distributive property:
The distributive property states that both convolution integral for continuous time LTI
system and convolution sum for discrete time LTI system are distributive.
For continuous time LTI system, the distributive property is
expressed as The output, y(t) = x(t) ∗ [ℎ1(t) + ℎ2(t)]
Or
y(t) = x(t) ∗ ℎ1(t) + x(t) ∗ ℎ2(t)
Thus, the two continuous time LTI systems, with impulse responses h1(t) and h2(t), have
identical inputs and outputs are added as
y1(t) = x(t) ∗ ℎ1(t)
y2(t) = x(t) ∗ ℎ2(t)
The output y(t) = y1(t) + y2(t)
y(t) = x(t) ∗ ℎ1(t) + x(t) ∗ ℎ2(t)
14
input x (t) h1(t) outputy(t)
h2(t)
x(t) h1(t)+h2(t) y(t)
Fig3.1: The distributive property of convolution integral for a parallel interconnection of

continuous time LTI system
Associative Property of LTI system:
According to associative property, both convolution integral for continuous time LTI
systems and convolution sum for discrete time LTI systems are associative.
For continuous time LTI system, according to associative property,
Fig 3.2:The associative property of convolution integral for a cascade interconnection of

continuous time LTI systems
15
Here we have y(t)=z(t) *h2(t). But z(t)=x(t)*h1(t). Therefore y(t)=[x(t)*h 1(t)*h2(t)]
Linear constant coefficient differential equation:
The continuous time linear time invariant (LTI) systems are described by their linear
constant coefficient differential equations. For this, let us consider a first order differential
equation as under
Where x (t) and y(t) are the input and output of the continuous time LTI system. A is a
constant value. The first order differential equation can be extended for higher order
differential equations. A general Nth order linear constant coefficient differential equation
can be given by
The complete solution of differential equation consists of the sum of particular solution y p(t)
and homogenous solution yh(t).
The homogeneous solution of a differential equation is possible by substituting
This solution to differential equation is also known as natural response of the system.
A particular case of differential equation is determined by putting N=0, we obtain
16
Transfer function:
Transfer functions are commonly used in the analysis of systems such as single-input single-output filters,
typically within the fields of signal processing, communication theory, and control theory. The term is often
used exclusively to refer to linear, time-invariant systems (LTI). Most real systems have non-linear
input/output characteristics, but many systems, when operated within nominal parameters have behavior
that is close enough to linear that LTI system theory is an acceptable representation of the input/output
behavior.
The descriptions below are given in terms of a complex variable, , which bears a brief
explanation. In many applications, it is sufficient to define (and ), which reduces
the Laplace transforms with complex arguments to Fourier transforms with real argument ω. The
applications where this is common are ones where there is interest only in the steady-state response of an
LTI system, not the fleeting turn-on and turn-off behaviors or stability issues. That is usually the case
for signal processing and communication theory.
Thus, for continuous-time input signal x(t) and output y(t0, the transfer function H(s) is the
linear mapping of the Laplace transform of the input, X(s)=L{x(t)}, to the Laplace transform
of the output Y(s)=L{y(t)}:
Y(s)=H(s)X(s)
17
Conditions required for transfer function:
(i) System should be in unloaded condition (initial conditions are zero)
(ii) The system should be linear time invariant.
Impulse Response:
In signal processing, the impulse response, of a dynamic system is its output when presented
with a brief input signal, called an impulse. More generally, an impulse response refers to the
reaction of any dynamic system in response to some external change. In both cases, the
impulse response describes the reaction of the system as a function of time. In all these cases,
the dynamic system and its impulse response may be actual physical objects, or may be
mathematical systems of equations describing such objects. Since the impulse function
contains all frequencies, the impulse response defines the response of a linear time- invariant
system for all frequencies. The impulse can be modeled as a Dirac delta function for
continuous- time systems, or as. The Dirac delta represents the limiting case of a pulse made
very short in time while maintaining its area or integral. While this is impossible in any real
system, it is a useful idealization. In Fourier theory, such an impulse comprises equal portions
of all possible excitation frequencies, which makes it a convenient test probe. Any system in a
large class known as linear, time-invariant (LTI) is completely characterized by its impulse
response. That is, for any input, the output can be calculated in terms of the input and the
impulse response. The impulse response of a linear transformation is the image of Dirac's
delta function under the transformation, analogous to the fundamental solution of a partial
differential operator. It is usually easier to analyze systems using transfer functions as
opposed to impulse responses. The transfer function is the Laplace transform of the impulse
response. The Laplace transform of a system's output may be determined by the
multiplication of the transfer function with the input's Laplace transform in the complex
plane, also known as the frequency domain. An inverse Laplace transform of this result will
yield the output in the time domain. To determine an output directly in the time domain
requires the convolution of the input with the impulse response. When the transfer function
and the Laplace transform of the input are known, this convolution may be more complicated
than the alternative of multiplying two functions in the domain. It is obtained by taking
inverse Laplace transform of transfer function H(s).
18
Frequency Response:
Frequency response is the quantitative measure of the output spectrum of a system or device
in response to a stimulus, and is used to characterize the dynamics of the system. It is a
measure of magnitude and phase of the output as a function of frequency, in comparison to
the input.It is obtained from the transfer function by substituting s=jω in transfer function.
Systems respond differently to inputs of different frequencies. Some systems may amplify
components of certain frequencies, and attenuate components of other frequencies. The way
that the system output is related to the system input for different frequencies is called the
frequency response of the system.
The frequency response is the relationship between the system input and output in the Fourier
Domain.
In this system, X(jω) is the system input, Y(jω) is the system output, and H(jω) is the
frequency response. We can define the relationship between these functions as:
Y(jω)=H(jω)X(jω)
Since the frequency response is a complex function, we can convert it to polar notation in the
complex plane. This will give us a magnitude and an angle.
19
Amplitude Response:
For each frequency, the magnitude represents the system's tendency to amplify or attenuate
the input signal.
Phase Response:
The phase represents the system's tendency to modify the phase of the input sinusoids.
The phase response, or its derivative the group delay, tells us how the system delays the input
signal as a function of frequency.
2. Rodger E. Ziemer , William H Tranter, D. R. Fannin,”Signals and Systems: Continuous and

Discrete”, 4th Edition, Pearson Education India, 2014.
20
ENGINEERING
UNIT - IV
1
IV ANALYSIS OF DISCRETE TIME SIGNALS AND LTI DISCRETE
TIME SYSTEMS
Discrete Time Fourier Transform (DTFT)

The discrete•time Fourier transform (DTFT) of a real, discrete•time signal x[n] is a
complex•valued function defined by
for any (integer) value of n.

Inverse Discrete Time Fourier Transform (IDTFT)
The function X (ejω) or X () is called the Discrete•Time Fourier Transform(DTFT) of the
discrete•time signal x(n). The inverse DTFT is defined by the following integral:
Properties of DTFT
2
1. Find the DTFT of an impulse function which occurs at time zero.
x[n]   [n]
X e jw   [n]e jwn  1
F
[n]1
F
[n  1](1)  e jw(1)
Discrete Fourier Transform

The DFT is used to convert a finite discrete time sequence x (n) to an N point
frequency domain sequence denoted by X (K). The N point DFT of a finite duration
sequence x (n) is defined as
N•1
X (K) = ∑ x (n) ej2‫ה‬nk/Nfor K=0, 1, 2, …….N•1
n=0
The discrete Fourier transform (DFT) is the Fourier transform for finite•length
sequences because, unlike the (discrete•space) Fourier transform, the DFT has a discrete
argument and can be stored in a finite number of infinite word•length locations. Yet, it
turns out that the DFT can be used to exactly implement convolution for finite•size arrays
Inverse Discrete Fourier Transform
The IDFT is used to convert the N point frequency domain sequence X (K) to an N point
time sequence. The IDFT of the sequence X (K) of length N is defined as
N•1
x (n) =1/N ∑X (K) e+j2‫ה‬nk/N for n=0, 1,2,……N•1
K=0
Properties of DFT
1. Periodicity: X (K+N) =X (K) for all K.
2. Linearity: DFT[a1 x1 (n)+a2 x2(n)]=a1 X1 (K)+a2 X2 (K)
3. DFT of time reversed sequence: DFT[ x(N•n)]=X(N•K)
4. Circular convolution :DFT[x1(n)*x2(n)]=X1(K) X2(K)
5. Shifting: If DFT {x (n)} =X (K), then DFT{x (n•no)} =X (K) e •j2‫ה‬no k/N
6. Symmetry property
Re[X(N•k)]=ReX(k)
This implies that amplitude has symmetry Im[X(N•k)]= • Im[X(k)]
This implies that the phase spectrum is antisymmetric.
7. If x[n] is an even function xe[n] then
3
This implies that the transform is also even
8.If x[n] is odd function xo[n] than

N 1
F xo n  X o k    j  xo nsinknT 
n0
This implies that the transform is purely imaginary and odd
9.Parseval’s Theorem
The normalized energy in the signal is given by either of the following expressions
N 1 1 N 1
 x n N k 0 X k
2
2

n0
10. Delta Function
FnT 1
11. Unit step function

F u[n]    w  2k 
1

1 e jw k 
 

 2  w  w  2k 
F e jw0n  
0
k 
12. Fourier transform of a CT complex exponential is interpreted as an impulse at w=w 0. For
discrete•time we expect something similar but difference is that DTFT is periodic in w with
period 2. This says that FT of x[n] should have impulses at w0, w0 ±2, w0±4 etc.
F
1
nu[n] ( n  1) 
1 e  jw
13. Linear cross•correlation of two data sequences or series may be computed using DFTs.
The linear cross correlation of two finite•length sequences x1[n] and x2[n] each of
length N is defined to be:
4
This circular correlation can be evaluated using DFTs as shown below:
The circular correlation can be converted into a linear correlation by using augmenting
zeros. If the sequences are x1[n] of length N1 and x2[n] of length N2, then their linear
correlation will be of length N1+N2•1.
To achieve this x1[n] is replaced by x1a[n] which consists of x1[n] with (N2•1) zeros added
and x2[n] is augmented by (N1•1) zeros to become x2a[n].
 r 
( j)  F 1 X  k X k 
1a 2a

x1 x2
1. Find the DFT of the following signal x(n)=δ(n)

N•1
X (K) = ∑ x (n) e•j2‫ה‬nk/N for K=0, 1, 2….,N•1
n=0
N•1
X (K) = ∑ δ (n)e•j2‫ה‬nk/N for K=0, 1, 2,…N•1
n=0
X (K) =1
2. Consider a length-N sequence defined for n = 0,1,2,……,(N1) where
 1 n 0
x[n]  
0 otherwise Find the DFT of the given sequence.
N 1
X [k ]  x[n]WN k  0,1,2,....... ,(N 1)
kn
n0
The N•point DFT is equal to 1

Basic Principles of Z Transform:
The z•transform is useful for the manipulation of discrete data sequences and has
acquired a new significance in the formulation and analysis of discrete•time systems. It is
used extensively today in the areas of applied mathematics, digital signal processing,
control theory, population science, economics.
These discrete models are solved with difference equations in a manner that is
analogous to solving continuous models with differential equations. The role played by the
z•transform in the solution of difference equations corresponds to that played by the
Laplace transforms in the solution of differential equations.
5
Types of Z Transform Unilateral Z•transform
Alternatively, in cases where x[n] is defined only for n ≥ 0, the singlesided or unilateral Z
transform is defined as
In signal processing, this definition can be used to evaluate the Z•transform of the unit
impulse response of a discrete•time causal system
Bilateral Z•transform
The bilateral or two•sided Z•transform of a discrete•time signal x[n] is the
formalpower series X(Z) defined as
where n is an integer and z is, in general, a complex number:
z= A ejɸ = A (cos ɸ+j sin ɸ)
where A is the magnitude of z, j is the imaginary unit, and ɸ is the complex

argument (also referred to as angle or phase) in radians.
Inverse Z Transform
1. The z•transform of the sequence

find its Z transform
6
2. Find the z•transform of the unit pulse or impulse sequence .
3. The z•transform of the unit•step sequence is
4. The z•transform of the sequence is .
7
.
5. The z•transform of the exponential sequence is .
6.
1 1
6 n
7
8
8
10
11
12
13
Properties of Z•Transform
Z•Transform has the following properties:
1. Linearity Property
Then linearity property states that
2. Time Shifting Property
Then Time shifting property states that
3. Multiplication by Exponential Sequence Property
Then multiplication by an exponential sequence property states that
9
4. Time Reversal Property
Then time reversal property states that
5. Differentiation in Z•Domain OR Multiplication by n Property
Then multiplication by n or differentiation in z•domain property states that
6. Convolution Property
Then convolution property states that
7. Correlation Property
Then correlation property states that
8. InitialValueand FinalValueTheorems
Initial value and final value theorems of z•transform are defined for causal signal.
10
Initial Value Theorem
For a causal signal x(n), the initial value theorem states that
This is used to find the initial value of the signal without taking inverse z•transform
Final Value Theorem

For a causal signal x(n), the final value theorem states that
This is used to find the final value of the signal without taking inverse z•transform.
Region of Convergence (ROC) of Z•Transform

The range of variation of z for which z•transform converges is called region of convergence
of z• transform.
Properties of ROC of Z•Transforms
 ROC of z•transform is indicated with circle in z•plane.
 ROC does not contain any poles.
 If x(n) is a finite duration causal sequence or right sided sequence, then the ROC is
entire z• plane except at z = 0.
 If x(n) is a finite duration anti•causal sequence or left sided sequence, then the ROC
is entire z• plane except at z = ∞.
 If x(n) is a infinite duration causal sequence, ROC is exterior of the circle with
radius a. i.e. |z| > a.
 If x(n) is a infinite duration anti•causal sequence, ROC is interior of the circle with
radius a. i.e. |z|< a.
 If x(n) is a finite duration two sided sequence, then the ROC is entire z•plane except
at z = 0 & z= ∞.
The concept of ROC can be explained by the following example:
11
Example 1: Find z•transform and ROC of
The plot of ROC has two conditions as a > 1 and a < 1, as the value of ‘a’ is not known.
In this case, there is no combination ROC.
Here, the combination of ROC is from
12
Hence for this problem, z•transform is possible when a < 1.
1. The Z transform of a right sided signal is
For this summation to converge, i.e., for to exist, it is necessary to have , i.e., the
ROC is . As a special case when , and we have
2. The Z•transform of a left sided signal is:
For the summation above to converge, it is required that , i.e., the ROC is .
Comparing the two examples above we see that two different signals can have
identical z•transform, but with different ROCs.
3. Find the inverse of the given z•transform .

Comparing this with the definition of z•transform:
13
we get
In general, we can use the time shifting property
to inverse transform the given above to directly.
Zeros and Poles of Z•Transform
All z•transforms in the above examples are rational, i.e., they can be written as a ratio of
polynomials of variable in the general form
where is the numerator polynomial of order with roots , and
is the denominator polynomial of order with roots .
In general, we assume the order of the numerator polynomial is lower than that of the denominator
polynomial, i.e., . If this is not the case, we can always expand into multiple terms so
that is true for each of terms.
The zeros and poles of a rational

are defined as:
 Zero: Each of the roots of the numerator polynomial for which
is a zero of .
14
If the order of exceeds that of (i.e., ), then , i.e.,
there is a zero at infinity:
 Pole: Each of the roots of the denominator polynomial for which
is a pole of .
If the order of exceeds that of (i.e., ), then , i.e,

there is a pole at infinity:
Most essential behavior properties of an LTI system can be obtained graphically from the ROC and the zeros and
poles of its transfer function on the z•plane
Inverse Z Transform using Contour Integration Method
15
X z
1
,| z || a|
2. Evaluate the inverse z 1  az 1 using the complex inversion
transform of integral.
Long Division Method
The z•transform is a power series expansion,
where the sequence values x(n)are the coefficients of z •n in the expansion. Therefore, if
we can find the power series expansion for X(z), the sequence values x(n)may be found by
simply picking off the coefficients of z –n.
1. Sometimes the inverse transform of a given can be obtained by long division.
By a long division, we get
which converges if the ROC is , i.e., and we get
. Alternatively, the long division can also be carried out as:
which converges if the ROC is , i.e., and we get
2. To understand how an inverse Z Transform can be obtained by long division,

consider the function
16
If we perform long division
we can see that
.
So the sequence f[k] is given by
Upon inspection
3. Find the Inverse Z Transform using Long Division Method
and the sequence f[k] is given by
17
Inverse Z Transform using Residue Method:
Find the solution using the formula
where are the poles of .
Partial fraction method

Inverse Z Transform by Partial Fraction Expansion
This technique uses Partial Fraction Expansion to split up a complicated fraction

into forms that are in the Z Transform table. As an example consider the function
For reasons that will become obvious soon, we rewrite the fraction before expanding it by
dividing the left side of the equation by "z."
Now we can perform a partial fraction expansion
18
These fractions are not in our table of Z Transforms. However if we bring the
"z" from the denominator of the left side of the equation into the numerator of the
right side, we get forms that are in the table of Z Transforms; this is why we
performed the first step of dividing the equation by "z"
So
or
19
LTI – DT Systems:
A DT System which satisfies Linearity and time invariance property is called LTI DT
systems. LTI systems comprise a very important class of systems, and they can be
described by a standard mathematical formalism.
Characterization using difference equation:
Systems described by constant-coefficient, linear difference equations are LTI systems.

In exploring this fact, it is important to keep in mind that our default setting is that all
signals are defined for –∞< n < ∞. The difference equation is a formula for computing an
output sample at time n based on past and present input samples and past output
samples in the time domain. The difference equation is as follows:
Where x(n) is the input signal and y(n) is the output signal and constants
, are called the coefficients. We have a system whose input

and output signals are related by
y[n] + ay[n -1] = bx[n] , - ∞ < n < ∞
where a and b are real constants. This is called afirst-order, constant-coefficient, linear
difference equation. Given an input signal x[n], this can be viewed as an equation that
must be solved for y[n] for each input signal x[n] thereis a unique solution for the output
signal y[n].
∞
y[n] = ∑ (—a)n-kbx[k]
k=-∞
Example 1: Find the solution to the following difference equation by using the z-
transform x(k+2)+3x(k+1)+2x(k)=0, x(0)=0,x(1)=1
20
Solution:
Take the z-transform of both side of the equation, we get z2 X (z) − z2x(0) − z ⋅x(1) + 3z ⋅X
(z) − 3z ⋅x(0) + 2X (z) = 0
Substituting in the initial conditions and simplifying gives
Take the inverse Z transform of above equation we get,
Example 2: Using the z-transform to solve the following difference equation x(k + 2) +
0.4x(k +1) − 0.32x(k) = u(k), where x(0) = 0 and x(1) = 1. The input u(k) is a unit step
input, i.e. u(k) = 1, for k ≥ 0.
Solution:
Take the z-transform of the difference equation we get
Substituting the initial conditions and simplifying, we obtain
The partial fraction expansion of the solution X(z) is
The corresponding time sequence can be obtained by taking the inverse z-transform of the
above equation:
x(k) = 0.926 − 0.3704(−0.8)k − 0.5556(0.4)k, for k = 0, 1, 2, …
Convolution Sum: To each LTI system there corresponds a signal h[n] such that the
input-output behavior of the system is described by
∞
y[n] = ∑ x[k]h[n-k]
k=—∞
21
This expression is called the convolution sum representation for LTI systems. In addition,
the shifting property easily shows that h[n] is the response of the system to a unit-pulse
input signal.
∞ ∞
y[n]= ∑ x[k]h[n — k]= ∑ g[k]h[n — k]= h[n]
k=-∞ k=-∞
Thus the input-output behavior of a discrete-time, linear, time-invariant system is completely

described by the unit-pulse response of the system. If h[n] is known, then the response to
anyinput can be computed from the convolution sum.
The system response to this inputsignal is given by
To rewrite this expression, change the summation index from k to 1 = k — N, to obtain
The convolution representation for linear, time-invariant systems can be developed by

adopting a particular representation for the input signal and then enforcing the properties of
linearity and time invariance on the corresponding response
Example, if n = 3, then the right side is evaluated by the sifting property to verify
∞
∑ x[k]h[3 - k] = x[3]
k=-∞
We can use this signal representation to derive an LTI system representation as follows.
The response of an LTI system to a unit pulse input, x[n] = u[n], is given the special
notation y[n] = h[n]. Then by time invariance, the response to a k—shifted unit pulse, u[n]
= δ[n — k] = h[n — k]. Furthermore, by linearity, the response to a linear combination of
shifted unit pulses is the linear combination of the responses to the shifted unit pulses. That
is, the response to x[n], as written above, is
∞
y[n]= ∑x[k] h[n — k]

k=-∞
Thus have arrived at the convolution sum representation for LTI systems. The convolution
representation follows directly from linearity and time invariance An alternate expression for
the convolution sum is obtained by changing the summation variable from k to 1 = n - k:
∞
y[ n]= ∑h[ l ] x[ n - l]
k=-∞
It is clear from the convolution representation that if the unit-pulse response of an LTI
system is known, then we can compute the response to any other input signal by evaluating
the convolution sum. Indeed, we specifically label LTI systems with the unit-pulse response
in drawing block diagrams, as shown below
X(n) y(n)
LTI h(n)
Example: convolve the following signals using matrix method x(n)={ 1 1 2 2}, h(n)={1 2 3
4} Ans: y(n)={1 3 7 12 14 14 8}
Properties of Convolution – Interconnections of DT LTI Systems
Convolution of two signals given by

∞
y[n] = x(n) * h(n) = ∑ x[k]h[n – k]
k=-∞
For any n, the value of y[n] in general depends on all values of the signals x[n] and h[n], y[n]
= x[n]* h[n], for example, y[2] = x[2]* h[2] .
 Commutativity: Convolution is commutative. That is, x(n) * h(n) = h(n) * x(n)

∞ ∞
∑ x[ k] h[ n- k]= ∑h[ k] x[ n- k] , for all n
k—∞ k—∞
∞ ∞
( x* h)[n]= ∑ x[k]h[n- k]= ∑ x[n- q]h[q]
k=-∞ q=∞
= ∑h[q]x[n- q]= h(n)*
x(n) q=-∞
21
Using this result, there are two different ways to describe in words the role of the unit-
pulse response values in the input-output behavior of an LTI system. The value of h[n – k]
determineshow the nth value of the output signal depends on the kth value of the input
signal. Or, the value of h[q] determines how the value of y[n] depends on the value of x[n
– q].
 Associativity:Convolution is associative. That is, 
(x* h1 * h2))[n] = ((x* h1)* h2)[n]
 Distributivity: Convolution is distributive (with respect to addition). That is, (x* (h1
+ h2))[n]= (x* hi)[n]+ (x* h2)[n]
For any constant b,((bx)* h)[n] = b (x* h)[n]
 Shifting Property:This is simply a restatement of the time-invariance property. For any integer
no , if i[n] = x[n — no], then
h[n] = (x* h)[n — no]
 Identity: It is worth noting that the "star" operation has the unit pulse as an
identity element. Namely,
(x* S)[n] = x[n]
This can be interpreted in system-theoretic terms as the fact that the identity system, y[n] =
x[n] has the unit-pulse response h[n] = δ[n] .Also we can write (δ* δ)[n]= g[n]. The unit
pulse is the unit-pulse response of the system whose unit-pulse response is a unit pulse.
These algebraic properties of the mathematical operation of convolution lead directly to

methods for describing the input-output behavior of interconnections of LTI systems. For
example,
x(n) y(n)
LTI Systems
has the same input-output behavior as the system
x(n) h1(n)+h2(n) y(n)
both have the same input-output behavior as the system
x(n) y(n)
LTI (h1*h2)(n)
22
Transfer Function and Impulse Response Sequence
The transfer function for the continuous-time system relates the Z transform of the
continuous-time output to that of the continuous-time input. For discrete-time systems,
the transfer function relates the z- transform of the output at the sample instance to
that of the sampled input. Consider a linear time- invariant discrete-time system
characterized by the following linear difference equation:
where u(k) and y(k) are the system input and output, respectively, at the kth sample
instances. If we take the z-transform and by using the time shift property of the z-
transform, we obtain
which can be written as
where
Consider the response of the linear discrete-time system described by Equation, initially
atrest (y(k) = 0, k < 0), when the input u(k) is the Kronecker delta function δ0(k), i.e.
Since
then
23
Thus, G(z) is the z-transform of the response of the system to the Kronecker delta function
input. The function G(z) is called the transfer function of the discrete-time system. In the
above derivation, the role of the Kronecker delta function in discrete-time system is
similar to that of the unit impulse function (the Dirac delta function) in continuous-time
systems.The inverse transform of G(z) as given by Eq.
is called the impulse response function (sequence).The system described by the difference
equation
where the system is initially at rest (y(k) = 0, k < 0) and the input u(k) = 0, for k < 0, can
be represented by the transfer function G(z).
Example Consider the difference equationy(k + 2) + a y(k +1) + a y(k) = b u(k + 2) + b u(k
+1) + b u(k). Assuming that the system is initially at rest and u(k) = 0 for k < 0, find the
transfer function.
Solution:
The z-transform of the difference equation is
Collect common terms
24
To determine the initial conditions y(0) and y(1), we substitute k = −2 into the original
difference equation and obtain
which implies
By substitute k = −1 into the original difference equation and obtain
which implies y(1) =-a y(0)+ b u(1)+ b u(0)
By substituting Equations we get
Hence, if both y(k) and u(k) are zero for k < 0, then the system’s input and output are
related by the above Equation.The transfer function G(z) = Y(z) / U(z) can be written as
The above Equation is the same transfer function for the system described by the
difference equation y(k) + a y(k − 1) + a y(k − 2) = b u(k) + b u(k − 1) + b u(k −2 ).
Causality and Stability of LTI Systems

A DT system is said to be a causal if the output of the system at any time depends only
on the present input, past input but does not depend on future input and output
Ex: y(n)=x(n), x(n-1), x(n-2)….
A system is said to be stable if and only if every bounded input produces a bounded output
condition for stability is given by
∞
−∞ |ℎ 𝑛 |<∞
Computation of Impulse response and Transfer function using Z Transform.

Example: Assuming that the system is initially at rest, find the impulse response of the
following discrete time system:
25
y(k + 3) = 2u(k + 3) − u(k + 2) + 4u(k +1) + u(k). Find the transfer function
Solution:
Transfer function of the system can be written as
The impulse response of the system with zero initial condition is then the inverse z-
transform of the pulse transfer function,
Hence g(0) = 2, g(1) = −1, g(2) = 4, g(3) = 1, g(k) = 0, for k >3
Frequency Response of Discrete-Time Systems

In order for systems to possess a steady-state response to a sinusoidal input, in must be
stable (all the poles of the transfer function must lie within the unit circle of the
complex z plane). Let the system of interest be defined by
where pi are the complex poles of the system. We further assume that the system is stable,
i.e. pi <1 for all i.
Let the input to the system be a cosine sequence of radian frequency ω, i.e.
The corresponding z-transform of the input sequence is
Substituting the input equations the output Y(z) is given by
A partial fraction expansion of the above equation can be written as
26
Each term in the summation on the right hand side of equation yields a time domain
sequence of the form Di (pi) k, which if pi <1 will vanish when k gets larger and hence
does not contribute to the steady-state response. The coefficients B and C in Eq. can be
evaluated by the following formula
jT
Substituting Y(Z) in the above formula gives
Thus the steady state response YSS(s) is
Since G(z) is a rational function of the complex variable z, G(e jωT) is a complex number
that can be written in polar forma as
where φ is the phase angle of the complex number G(e jωT). With similar reasoning,
G(e −jωT) will have the same magnitude and conjugate phase angle as G(e jωT), i.e.
Substituting the values, the steady-state response can be written as
Taking inverse z-transform of the above equation, we can obtain the time sequence of the
steady state sinusoidal response to be
27
Using the Euler identity, the above equation can be further simplified and the steady-state
sinusoidal response is
From the above Equationwe see that, similar to the continuous-time case, the steady-state
response of the system G(z) to a sinusoidal input is still sinusoidal with the same frequency
but scaled in amplitude and shifted in phase. The amplitude of the steady-state response is
scaled by a factor of G|ejwt|,which will be referred to as the system gain associated with G(z)
at frequency ω . The complex function of ω, G(e jωT), is called the frequency response
function of the system G(z). The frequency response function of a system can be obtained
by replacing the z-transform complex variable z with e jωT, i.e.
As in the continuous-time case, we are usually interested in the magnitude and

phasecharacteristics of this function as a function of frequency. It is interesting to note
that the DC gain of the system corresponds to the magnitude of the frequency response
function at ω = 0,
DC Gain=
This is slightly different from the continuous-time case where the DC gain is evaluated
by substituting the Laplace variable s by 0.
Example: Find the frequency response for the discrete-time system described by the
following difference equation:
y(k) = e−2T y(k −1) + u(k), where T =π/5
Solution:
The impulse transfer function of the system can be found by taking the z-transform of
thedifference equation and assuming zero initial conditions
which implies
28
The frequency response of the system is

2. Rodger E. Ziemer , William H Tranter, D. R. Fannin,”Signals and Systems: Continuous and Discrete”, 4th Edition,
Pearson Education India, 2014.
29
ENGINEERING
UNIT - V
1
V.REAL TIME APPLICATIONS OF SIGNALS AND SYSTEMS
Signal modelling
• Modelling of signal is basically mathematical representation of signal.
• Fourier series, Fourier transform are kind of signal models.
DETERMINISTIC
SIGNALS
INPUT WILL BE
RANDOM STOCHASTIC
PROCESS,WHITE
Figure 1: Signal classificationNOISE
Speech Signal
Time domain
Energy
The amplitude of the speech signal varies over time due to the type of speech sound and to
supra segmental factors. The short-time energy is a convenient way to represent these
amplitude variations. For a digital signal
where w(n) is a window function, which is nonzero over only the analysis time interval.
That is, w(n) selects the interval to be analyzed. (For energy determination, the shape of w(
n) is not critical j it is often a rectangular window, i.e., 1 over its nonzero extent. For
spectral analysis (we shall note that the shape of wen is important.) Because of the
squaring operation, the energy can be sensitive to large signal levels, and the computation
can produce overflow on short word length computers.
For analog signals, it is more convenient to measure magnitude, by means of a rectifier
and low pass filter. Note that the impulse response of the low pass filter corresponds to the
window function w(n) . Energy or magnitude provides a primary basis for separating
speech from silence and, because voiced sounds are generally higher in amplitude than
unvoiced, for distinguishing between voiced and unvoiced intervals. They are also useful
for delineating syllables and for locating the syllable nucleus.
2
Zero Crossings
It is well known that infinitely clipped speech is intelligible (although distorted), so there
must be useful information in the zero-crossings of the signal. The zero-crossing rate
(ZCR) is obtained by counting the number of times the signal changes sign during a fixed-
length analysis interval. This rate gives a very rough measure of the frequency content of
the signal, as may be appreciated by noting that a sinusoid of frequency F gives an average
ZCR of 2F sec- 1 ZCR is sometimes used after broad bandpass filters for a rough formant
estimation, but the fact that the frequency ranges of the formants overlap limits the
accuracy of these estimates. The ZCR also helps distinguish between speech and non
speech and between voiced and unvoiced sounds (the mean ZCR for voiced sounds is about
1500 sec- 1 and for unvoiced sounds 5000 sec- 1, but the distributions overlap (3)).
Pitch Period Estimation

Measurement of the pitch period (or equivalently the fundamental frequency) of voiced
speech and separation of voiced from unvoiced intervals are problems for which many
techniques have been devised. Some of these techniques are oriented toward the time
domain, and these will be described here; others are frequency-domain oriented.
Autocorrelation
The short-time autocorrelation of the signal,
peaks when K is equal to the pitch period, even if the signal is only approximately periodic,
and this is the basis for autocorrelation methods. But the autocorrelation function also
contains other information, enough to represent the entire spectrum. Autocorrelation
peaks due to formant structure can be as large as peaks due to approximate periodicity.
Therefore, before autocorrelating, it is useful to lowpass filter the speech (to remove the
higher formants) and/or to flatten the spectrum (to remove formant information). Spectral
flattening can be done by center-clipping (or linear prediction inverse filtering.
FREQUENCY DOMAIN METHODS

The representation of signals as the sum of sinusoids or complex exponentials often leads
to convenient solutions and insight into physical phenomena. For linear systems, (such as
our model of speech production), it is convenient to determine the system response to such
a representation. A spectral representation often displays characteristics not evident in the
waveform, and we know that a kind of spectral analysis is performed in the ear. For these
reasons, frequency domain methods are important in speech analysis.
It is with frequency-domain methods, in which the shape of the spectrum is relevant, that
the shape of the short-time window function is important. Windowing is multiplication of
the waveform and window function, which appears in the frequency-domain as
convolution of their Fourier transforms. Therefore, a narrow main lobe of the window's
transform (to minimize "smearing" of the spectrum) and low side lobes (to minimize
"leakage" from other parts of the spectrum) are desirable. In these respects, the
rectangular window is inferior to smooth windows such as the Hann, Hamming, or Parzen
windows (3-6). In analog filter bank spectral analysis , the window function is determined
by the shape of the filter bandpass characteristics (a wider bandwidth yields sharper time
resolution). Additional time-averaging is usually introduced by the smoothing of the
3
envelope detector following each filter.
Music signal Analysis and modelling

Pitch
Most musical instruments—including string-based instrumentssuch as guitars,
violins,and pianos,as well as instruments based on vibrating air columns such as flutes,
clarinets, and trumpets—are explicitly constructed to allow performers to produce sounds
with easily controlled, locally stable fundamental periods. Such a signal is well described as
a harmonic series of sinusoids at multiples of a fundamental frequency, and results in the
percept of a musical note (a single perceived event) at a clearly defined pitch in the mind of
the listener. With the exception of unpitched instruments like drums, and a few inharmonic
instruments such as bells, the periodicity of individual musical notes is rarely ambiguous,
and thus equating the perceived pitch with fundamental frequency is common.
Harmony
While sequences of pitches create melodies—the “tune” of a music, and the only part
reproducible by a monophonic instrument such as the voice—another essential aspect of
much music is harmony, the simultaneous presentation of notes at different pitches. Different
combinations of notes result in different musical colors or “chords,” which remain
recognizable regardless of the instrument used to play them. Consonant harmonies (those
that sound “pleasant”) tend to involve pitches with simple frequency ratios, indicating many
shared harmonics. Fig. 2 shows middle C (262 Hz), E (330 Hz), and G (392 Hz) played on a
piano; these three notes together form a C Major triad, a common harmonic unit in western
music. The figure shows both the spectrogram and the chroma representation. The ubiquity
of simultaneous pitches, with coincident or near-coincident harmonics, is a major challenge
in the automatic analysis of music audio: note that the chord in Fig. is an unusually easy case
to visualize thanks to its simplicity and long duration, and the absence of vibrato in piano
notes.
Figure 2. Frequency Spectrum

Medical uses of signal analysis and modelling
Biomedical signal analysis allows us to extract meaningful information from biological
processes, thus enabling the assessment, characterization, and understanding of their
originating mechanisms. Biomedical signals, however, are nonstationary and have statistics
that change over time. As a result, conventional frequency-domain signal analyses have their
limitations and a more powerful analysis technique is needed, particularly one capable of
characterizing the changes in spectral content over time. In this article, we propose one such
spectrotemporal representation—the modulation spectrogram. We start by presenting the
theoretical foundation behind the technique and compare three of the most utilized
approaches to calculate the spectrotemporal representation, namely the short-time Fourier
transform, the continuous wavelet transform, and the Hilbert transform. An open-source
amplitude modulation analysis toolbox is presented to allow the reader to explore amplitude
modulation analysis of different biomedical signals. Lastly, to illustrate the advantages of the
4
modulation spectrum analysis over conventional frequency-domain tools, several biomedical
applications are described, ranging from detecting breathing rate from ECG signal, to
improved Alzheimer’s disease diagnosis using novel features extracted
from electroencephalograms, to artifact removal of wearable electrocardiograms. It is hoped
that this article and its companion open-source toolkit will allow readers to quickly witness
the advantages of the described spectrotemporal representation and explore novel biomedical
signal analysis applications.
Figure 3. Medical Signal Processing and Analysis
Figure 4. Pulse train
Applications of Fourier Transform

• The Fourier transform has many applications in any field of physical science that uses
sinusoidal signals, such as engineering, physics, applied mathematics, and chemistry,
will make use of Fourier series and Fourier transforms
5
Figure No 5 FT of a signal with multiple frequencies
Telecommunication
• It’s a obvious for that there would be digital telecom without Nyquist-Shannon
theorem and Fourier Transform. Both are extremely important. Fourier transform
has greatly improved the way we are sending/collecting data. Note that mp3 file
format uses transform. Transforms are used in file compression: for instance cosinus
transform is also used in jpeg file compression. There are numerous applications in
telecoms, and therefore I decided not to dwell on them and go further.
Automotive
• A car is going to be have an engine test stand. A computer with multiple microphone
is going to measure voice of the engine the car produces. The software analyses it and
gives output that shows us how devastated engine is and what is its real distance. This
is comes in handy.
• A system of Vibro-acoustic signal analysis (acceleration of the unsprung mass of

passenger’s car suspension actuated to vibration by harmonic, kinematic vibration)
that analized the suspension and adjusted it automatically to the current road
condition.
• Adaptive cruise control. The cruise control is a systems that controls your speed car.
There are many improvements to this system. One is that it uses FFT while controlling
speed.
Voice recognition system. Driving a car requires a lot of attention. Sometimes it’s difficult to
put the hands off the steering wheel and change the radio station. If we have to push one
button, it is far less problematic than choosing a street number in a GPS receiver. The
6
automotive companies are working on a voice recognition systems
The Fourier transform of the square wave generates a frequency spectrum that presents the
magnitude of the harmonics that make up the square wave (the phase is also generated, but is
typically of less concern and therefore is often not plotted).
The Fourier transform can also be used to analyze non-periodic functions such as
transients (e.g. impulses) and random functions. With the advent of the modern computer the
Fourier transform is almost always computed using the Fast Fourier Transform (FFT)
computer algorithm in combination with a window function.
Figure 6. Magnitude plot

Hearing devices
• Beltone hearing device
• You have from 9 up to 16 directed microphones, it minimizes and filters out the noise
of a street. It is self adjusting – this means you get different voice intensification
/amplification in different environments. This means that no mater where you are – in
laboratory or in a public place, you can always talk and hear the person you are
talking with. If you have a better one device – you can talk with your phone despite
having the disability. This allows people for normality in their lives. Everything is
packed in a small device, which you can hardly see.
• Voice recognition
• There are attempts to perform automatic voice recognition system in customer

support. The goal is to eliminate the extremely boring procedure in veryfing the right
user asking questions like: what is mother’s maiden name, what is your postal code
etc. The system may verify the user by analyzing its voice using Fourier Transform. It
is right research direction in my opinion and will greatly facilitate our live.
• Military
• Submarine requires extremely reliable communication
7
• The main point of Fourier Transform in military applications is: perform reliable and
safe communication. Using fourier transform you can decode your message in a
special way.
• cosinus tranform is used in jpeg file (->data compression). But the point is to sent a
message under the noise level.
Medical Imaging
Selected methods of image processing have been applied to expert system supporting the
process of identifying medical images. Fourier transform is well known tool for many
applications in the processing of images in many fields of science and technology, also in
medicine. In this case the image processing consists in spatial frequencies analysis of Fourier
transforms of medical images. Skin lesions are studied on the base of their images and it
seems that Fourier transformation is the right toll for such research. The distributions of
selected colors in Fourier transform images are studied.
NOISE REMOVAL IN medical IMAGES

The use of the FT and its inverse to remove unwarranted information from an image. (a) An
image obtained by adding the sine wave and chest radiography images together with its
equivalent Fourier spectrum in (b). The unwanted interference caused by the sinusoidal
brightness pattern can be removed by editing the spatial frequency information as shown by
the blackened areas in (c). The inverse FT then recovers the original chest image largely
undistorted as shown in (d). Further refinement of the editing process would allow complete
restoration of the image quality, in theory.
Figure 7. Noise removal using FT

MEDICAL SIGNALS
The EEG rhythms capture the nonlinear complex dynamic behavior of the brain system and
the nonstationary nature of the EEG signals. This method analyzes common frequency
components in multichannel EEG recordings, using the filter bank signal processing. The
mean frequency (MF) and RMS bandwidth of the signal are estimated by applying Fourier-
8
transform-based filter bank processing on the EEG rhythms, which we refer intrinsic band
functions, inherently present in the EEG signals. The MF and RMS bandwidth estimates, for
the different classes (e.g., ictal and seizure-free, open eyes and closed eyes, inter-ictal and
ictal, healthy volunteers and epileptic patients, inter-ictal epileptogenic and opposite to
epileptogenic zone) of EEG recordings, are statistically different and hence used to
distinguish and classify the two classes of signals using a least-squares support vector
machine classifier.
Figure 7. Fourier transform of EEG signal

• In recent years, detection of cardiovascular abnormalities in patients can be achieved
by using electrocardiogram (ECG) recording. Initially, an effective FFT is used to
extract the feature points in ECG signals, such as PQRST wave’s amplitude and wave
function and then the proposed multi-objective genetic algorithm is used to classify the
abnormality of heart patient. Basically, the ECG behaviour depends on various
factors such as age, physical condition of patients and the surrounding environment.
Figure 8 ECG wave
9
Figure 9 Magnitude and Phase spectrum
The Speech Communication Process

Speech is a complex sound produced by the human vocal apparatus, which consists of
organs primarily used for breathing and eating: the lungs, trachea, larynx, throat, mouth,
and nose. The source of energy for the production of these sounds is the reservoir of air in the
lungs. Sound is generated in two ways. If air is forced through the larynx with the vocal folds
appropriately positioned and tensioned, it sets them into oscillation, so that they release puffs
of air in a quasi-periodic fashion at rates of about 80 to 200 Hz for male speakers, and at
faster rates for women and children. This glottal source is rich in harmonics, and it excites
the acoustic resonances of the vocal tract above the larynx, which filter the sound. These
sharp resonances, called formants, are determined by the shape of the throat, mouth, and, if
the velum (soft palate) is open, the nasal cavity. It is through manipulation of the vocal tract
shape by the articulators (tongue, jaw, lips, and velum) that we control the formant
frequencies that differentiate the various voiced speech sounds. These are the vowels, nasal
consonants, liquids (/rl and Ill), and glides (/wl and Iy/). Only the lowest three or four
formants, up to about 4 kHz, are perceptually significant. The other sound source used in
speech is turbulent noise, produced by forcing air through some constriction (such as
between the tongue and teeth in th-sounds) or by an abrupt release of pressure built up at
some point of closure in the vocal tract (such as behind the lips in aIp/).
10
The spectral peaks associated with these fricated sounds generally lie between 2 and 8 kHz
and are primarily determined by the position and the shape of the constriction. Some sounds,
the voiced fricatives, such as Izl and lvi, have both voiced and turbulent excitation.
Distinguishable differences in the voice signal can be produced by quite small changes in the
way the vocal tract is manipulated, so a very large number of sounds can be produced. For
the communication of language, however, only a restricted number of sounds, or more
accurately, sound classes, are used. Words in English are made up of approximately 40 of
these phonemes, which correspond roughly to the pronunciation symbols in a dictionary.
Since there are relatively few of these elemental sound units, many speech recognition
systems choose phonemes or phoneme-like units as the units of recognition. This concept of
elemental units does not reduce speech recognition to the sequential recognition of 40 or so
fixed pat terns. The acoustic realizations of phonemes in real speech are drastically different
from their characteristics in isolated environments. When phonemes are strung together into
words, the acoustic characteristics of successive phonemes become overlapped, due to the
dynamics of the articulators and the tendency for an articulator to anticipate the position it
will next assume. (For example, when you say the It I of the word "tin", the lips are spread
apart, but in "twin", they are pursed together. Since the lips are not used for the articulation
of the It I , they are free to assume the position for the following phoneme.)
These context effects, along with effects due to linguistic stress, speed of talking, the
size of a particular speaker's vocal tract, and variation in the way a speaker says a given
word from repetition to repetition make the decoding of the speech signal a nontrivial
problem. To be sure, the speech wave does contain much information directly related to the
phonemes intended by the speaker, but the details of the coding are complex . The phonemes
themselves can be grouped into classes according to the ways they can modify and affect one
another when they are in proximity in normal speech. It is convenient to describe these
classes in terms of about fifteen binary distinctive features (9-12), most of which have
straightforward articulatory and acoustic attributes, such as voiced/unvoiced,
nasal/nonnasal, fricated/unfricated, and tongue position: back/front, high/low, etc. Distinctive
features have proved to be quite powerful in describing the phonological changes that
underlie the contextual effects referred to above. For that reason, some speech recognition
systems use them as recognition units or as ways of organizing their recognition of phoneme-
sized units.
11
Analog and Digital Techniques
There is a major division of SSPFE techniques between analog and digital. Analog
signal processing techniques are those performed on the (electrical analog of the) speech
signal by means of electronic circuitry. This requires investment in special-purpose electronic
equipment, which, although frequently modular, is somewhat inflexible, and furthermore,
requires periodic calibration and adjustment. Digital signal analysis, on the other hand, can
be performed on a general purpose digital computer, a microprocessor, or special purpose
digital hardware. When implemented on a computer, such digital instrumentation is very
flexible. Programming turns the computer into a custom-designed instrument; the
characterh:tics of the instrument may be changed or even redesigned without recourse to a
soldering iron or supply cabinet, and there is nothing that can go out of calibration.
Furthermore, many of the digital techniques now available are simply not realizable in the
analog domain. Perhaps the most notable drawback is that while analog instrumentation is
inherently real-time, it is not always possible for digital techniques to be performed in real
time. This is counterbalanced by the potentially far greater functionality of digital
techniques. The same technological advances that have made computers faster and
cheaperhave done the same for digital processing techniques, and we may expect this
tendency to accelerate. The flexibility and capability of digital SSPFE methods, together with
the increasing speed and decreaSing costs of computers and microprocessors, have
established a trend of increaSing importance of digital techniques in all aspects of speech
processing. It represents both analog and digital techniques, but the dominance of the digital
domain is unavoidable. For a signal to be analyzed digitally, it must first be converted from
the form of a continuous signal (an acoustic pressure wave or its electrical analog) to a digital
signal. A digital signal is discrete in two ways: it consists of measurements of the speech signal
amplitude of discrete, regularly spaced instants of time, and each of these samples has been
quantized to a certain number of bits of precision. This transformation from a continuous
waveform to a list of digital numbers is performed by an analog-to-digital converter.
Nyquist's sampling theorem (see, for example, (3-6)) tells us that in order to capture the
information contained in a bandwidth of B Hz, we must sample at a rate of a least 2B samples
per second; furthermore, if the signal contains energy at frequencies higher than B Hz, the
result will be distorted unless we first lowpass filter the signal at or below B Hz. For example,
12
an application in which the 0-5 kHz band is of interest may call for lowpassfiltering at or
below 5 kHz and sampling at 10 ,000 samples per second, with each sample represented by a
12 bit number.
Time and Frequency Domain Approaches Another fundamental division of SSPFE
approaches is between time domain and frequency domain approaches. Time domain
approaches deal directly with the waveform of the signal, and these representations are often
attractive because of their simplicity of implementation. In working with time domain
representations, it must be noted that the shape of the waveform for voiced speech depends
on (among other things) the pitch period, so we generally shouldn't pay attention to details of
the waveform. Also, the perception of speech is only minimally dependent on the phase of the
signal, so phase-independent measures are desirable. Frequency-domain approaches involve
(explicitly or implicitly) some form of spectral representation. A spectral representation often
displays characteristics not evident in the time domain, and we know that a kind of spectral
analysis is performed in the ear. The above observations about pitch and phase have
counterparts in the frequency domain. For voiced speech, the spectrum has both a fine and a
coarse structure, due to the pitch harmonics and the vocal tract transfer function
respectively. We use the magnitude or power spectrum, which does not contain phase
information. A time/frequency dichotomy does not account for all SSPFE approaches. Some
are hybrids of the two, and for others the concept of "time or frequency" seems ill-defined.
Analysis and Synthesis Of Signals
The task of an automatic speech recognition (ASR) system is to convert a speech signal
into a linguistic representation of its content -- to "recognize" the linguistic message
contained in the signal. A common aspect of all ASR systems is that they process the speech
signal in some way in order to produce a representation of the signal that is better suited to
the recognition process. This operation (or set of operations) is the function of the component
that can be called "speech signal processing and feature extraction" (SSPFE). What do we
mean by "speech signal processing" and by "feature extraction"? "Speech signal processing"
refers to the operations we perform on the speech signal (e.g., filtering, digitization, spectral
analysis, fundamental frequency estimation, etc.). "Feature extraction" is a pattern
recognition term that refers to the characterizing measurements that are performed on a
pattern (or signal); these measurements (" features") form the input to the classifier that
"recognizes" the pattern. In many cases in speech recognition, it is impossible to separate
13
these functions; they are one and the same.
Audio/speech processing is an important area of research in Forensic and Defense (speaker
recognition) and in search engines (retrieval). In both cases, it is necessary to extract the
features which describe the speech signal. Though considerable research is carried out in this
area, these techniques are text and language dependent. Extracting the text features is highly
complex. Also language dependency is yet another bottleneck in speech processing. Hence
major research work aims at developing efficient speaker recognition/speech retrieval
technique from the speech samples irrespective of the speech content and language. Major
research challenge lies in the selection of appropriate transform for decomposing the signals,
statistical features for aggregating the co-efficient and a classifier for identifying the speaker.
In case of defense organizations, forensic departments and biometric enabled

industries, speech signals are processed to identify a speaker. On the other hand, search
engines like Google and Yahoo need speech signal retrieval from queries. Both these cases
involve the following steps: speech signal database, signal preprocessing, outlier removal,
feature extraction for speech characterization and decision making. In both cases, Speech
characterization necessitates the researcher to understand physics behind the generation of
speech signals and the anatomy of vocal tract.
Human vocal system consists of lungs, trachea, vocal folds, epiglottis, tongue,
velum and nasal cavity. When there is an urge to speak, the nerve cells receive the commands
from brain and based on that the vocal system generates the speech signal. Air from the lungs
passes to the larynx through the trachea and hence vibrates the vocal chords. It is the
vibration of the vocal chords that result in sound generation. In addition, position of the
tongue also alters the voice signal. Human vocal system is as shown in Figure 9.1.
14
Figure 9.1 Human vocal system
Speech signal consists of voiced sound, unvoiced sound and silence (Figure 9.2). In
case of voiced sound, vocal cords are opened and closed periodically. It chops the airflow into
periodic pulses. Voiced sound occurs during vowel production. Unvoiced sound occurs
during consonants and vocal cords are randomly opened and closed and it results in noisy
turbulent signal. Silence occurs when neither voiced nor unvoiced sound is produced. In
general, speech signal is defined as a one Dimensional signal s(t) where ‘t’ denotes time and
s(t) is the amplitude or intensity at ‘t’. If ‘s’ and ‘t’ are both finite, then it is called a digital
speech signal.
15
Speech Signal
1
0.8
0.6
0.4
0.2
Amplitude 
-0.2
-0.4
Silent
-0.6
Unvoiced
-0.8
-1 Voiced
0 0.5 1 1.5 2 2.5 3 3.5
Samples 5
x 10
Figure 9.2 Voiced, unvoiced and silent regions in speech signal
Owing to the presence of voiced, unvoiced and silent samples in speech, speech
signals are non-stationary in nature. However it can also be viewed as a series of short term
(time) periodic signals. Energy (measure of amplitude of the signal) decreases as signal
traverses from voiced to silent through unvoiced sound. Due to randomness in unvoiced
sound, Zero Crossing Rate (ZCR) of unvoiced sound is higher than that of voiced sound.
Autocorrelation is the measure of repetitiveness in the signal while pitch is an indicator of
fundamental frequency. Having understood the nature of speech signals, the next task is to
develop speaker identification and speech retrieval systems.
Speech Signal Feature Selection and Extraction
Speech Signal
voice signal of speaker 1
0.25
0.2
0.15
0.1
Forward Feature Speaker
amplitude
Aggregation Identification
0.05
-0.05
Transform
-0.1
-0.15
-0.2
0 5 10 15
n 4
x 10
Figure 9.3 Block Diagram of Speaker Recognition System
Speaker Recognition system validates the identity of a person using the features
extracted from their voice signals. The two phases of a speaker recognition system is the
16
feature extraction and speaker identification. In this chapter, various features selected for
development of the proposed speaker recognition system, their significance and the various
transforms used for extraction of those features are discussed. The block diagram of the
proposed system is shown in Figure 9.3.
Feature Selection
Speech waveform is produced by the human speech production system and the
structure of their vocal tract.The voluntary movement of the anatomical structures present in
our human speech production system generates an acoustic sound pressure wave called the
speech waveform. Vocal folds, soft palate, tongue, teeth and lips are the finer anatomical
components involved in the human speech production. These finer anatomical components
are also called as articulators. Based on the movements and position of the articulators, the
modulation occurs in the speech waveform. Larynx plays an important role in speech
production as it provides the periodic excitation to the system for speech sound called as
voice. Since the human system varies with time, the spectral characteristics of the speech
waveform possess non stationary property. The prosodic features involved in the study of
voicing are the fundamental period, fundamental frequency and pitch.Speech signal provides
glottal information of the speaker, which by proper feature extraction gives the identity of the
speaker. Features namely fundamental period, fundamental frequency, pitch and number of
peaks for a smaller co articulation of the speakers along with a suitable feature extraction
technique is responsible to explore the speaker identity. Significance of prosodic and
statistical features for the representation of the speech signals is shown in Table 3.1.
As mentioned in Table 3.1, these features provide a holistic representation of the

speech signal. These features can be used for both speaker recognition and speech ranking
systems. In this research work, these features are determined on the actual coefficients and
also on the transformed coefficients. Transforms are selected based on the appropriateness in
describing the speech signal.
Transform Selection
As speech is a non-stationary signal and the information is present in low

17
frequencies, it necessitates a non-stationary tool that performs multi resolution analysis.
Feature extraction on the components from non-stationary analysis is an efficient way to
develop speaker recognition system. Significance of various transforms on speech signals is
shown in Table 3.2.
Speech Feature Extraction
On the acquired speech signals, both prosodic and statistical features are
determined from the transformed and time domain coefficients. These aggregated features
are used for developiing automated speaker recognition system and feature based speech
ranking system. The tree diagram in Figure 3.2 depicts the various features extracted on
speech signals.
Table 3.1 Significance of Prosodic and Statistical Features
Sl.
Feature Formula Courtesy Significance
No.
1 Pitch f0 fs www.indiana.ed For an adult male the vocal folds open and close
f0
T u/~acoustic/s302/ completely 100 times in one second. For an adult
phonation- female it is 200 times per second, for a child, it
chap4.pdf may be 300 times per second. Larger structures
vibrate more slowly than smaller structures.
N 1
3 Energy E iitg.vlab.co.in, The energy associated with voiced region is
E   x 2 ( n)
n 0 (2011). Short large compared to unvoiced region and silence
Term Time region will not have least energy.
Domain
Processing of
Speech.
4 Zero www- f s If the number of zero crossings are more in a
ZCR  n _ cross 
Crossing N
gth.die.upm.es/p given signal, then the signal is changing rapidly
Rate artners/sony/Ch and accordingly the signal may contain high
(ZCR) apter7_b.doc frequency information.
18
5 Number Size(pks) David Counting the number of positive peaks per
of Peaks GerhardTechnic second in the waveform determines the
NOP) al Report TR-CS frequency of the waveform.
2003-
06November,
2003
6 Mean  N 1 Digital Signal If N is small, the statistical noise in the
µ  xi
 i 0 Processing: A calculated mean of the signal will be very large.
Practical Guide In other words, you do not have access to
for Engineers enough data to properly characterize the signal.
and Scientists The larger the value of N, the smaller the
expected error will become.
7 Standard 1
1 Nhttps://en.wikipe Standard deviation is used to evaluate the
Deviation
s 
N  1 i dia.
0
( xi   ) 2
differences between speaking styles. The higher
org/wiki/Standar the standard deviation, the more lively the voice,
d_ deviation and the more pleasant and interesting to listen
to
8 Skewness E ( x   )3www.itl.nist.gov/ Skewness is used to detect the polarity of the
s
3 div898/handboo speech signal. It also provides vocal tract
k/eda/section3/ed information and estimate on the glottal source.
a35b.htm A positive skewness means the estimated glottal
source will have a negative skewness. Higher
skewness means higher excitation strength of
the vocal tract
Sl. Feature Formula Courtesy Significance
No.
9 Kurtosis E ( x   )4Ali Mansour et Kurtosis is negative during a silent period, and
s
4 al (1998), it becomes positive during the speech transient.
“Kurtosis:
Definition and
properties”,
19
FUSION 98
10 Second m2  E ( x  ) 2http://www1.mat Second order statistics is used to estimate the
Order hs.leeds.ac.uk/ap Number of harmonics present in the signal
Moment plied/news.dir/is
sue2/hos_intro.h
tml
11 Third m3  E ( x  )3http://www1.mat Third order moment has the potential to
Order hs.leeds.ac.uk/ap capture nonlinear information and it is more
Moment plied/news.dir/is immune to Gaussian noise
sue2/hos_intro.h
tml
12 Spectral J.
1 King Saud Music has a higher rate of change in the shape
SF 
Flux ( N  1)( M  1)Vol. 19, of the spectrum, and goes through more drastic
Univ.,
N 1 M 1
  log( x(n, k )   )
Eng. Sci. (1), pp. frame-to-frame changes where speech typically
n 1 k 1 95-133, Riyadh has a more constant rate of change.
(1427H./2006)
 log( x(n  1, k )   ) 2
13 Spectral  kX (kJ.) King Saud Spectral centroid describes the center of

SC  k
 X (k )
centroid Univ., Vol. 19, frequency at which most of the power in the
k Eng. Sci. (1), pp. signal is found. In speech signals the pitch of the
95-133, Riyadh audio signal stays in a more narrow range of
(1427H./2006) low values. As a result, music has a higher
spectral centroid than speech.
20
Table 3.2 Significance of Transforms on Speech Signals
Sl.No. Transform Formula Courtesy Significance
| f |  ( t2)
1 Stockwell  2
f2 Localization S-transform
S ( , f )   h(t ) e ei 2 ft dt
Transform  2 of the uniquely
complex combines a
spectrum: frequency
The S dependent
Transform resolution of
the time-
frequency
space with
absolutely
referenced
local phase
information.

2 Discrete 1  t b  A Wavelet DWT
Wavelet
x (a, b) 
a 

x(t ) 
 a 
 dt
Tour of decomposes
Transform Signal the signal into
Processing: approximation
The Sparse coefficients
Way and detailed
coefficients
Approximation
coefficients
correspond to
low
frequencies
and detailed
coefficients
represent the
21
high frequency
info. present
in the signal.
3 Cepstrum N 1  N 1 2
 j kn   j 2N kn Digital Cepstral
C (n)   log    x(n)e N  e
  n 0 Processing analysis
n 0   
of Speech separates the
Signals speech signal
into
component
representing
excitation
source and
vocal tract
impulse
response. So it
provides
information
about pitch
and vocal tract
configuration.
4 Autocorrelation k 1 Digital One would

A(k )   x ( n) x ( n  k ) Processing expect exact
m0
of Speech similarity at a
Signals time lag of
zero, with
increasing
dissimilarity as
the time lag
increases.
22
5 Empirical n Hilbert- any
Mode
x(t )   c j  rn Huang complicated
j 1
Decomposition Transform data set can be
and Its decomposed
Applications into a finite
and often small
number of
components.
EMD
preserves the
characteristics
of the varying
frequency.
Time Domain Short Time Energy, Flux,

Centroid
Discrete Wavelet
voice signal of speaker 1
Transform
0.25
0.2
0.15
0.1
Mean, Standard Deviation,
Stockwell
amplitude
0.05
-0.05 Transform Skewness, Kurtosis,

-0.1
-0.15 Energy, Zero Crossing

-0.2
0 5 10 15
n 4
Rate, Second Order

x 10
Cepstrum
Moment, Third Order
Emphirical Mode Moment,
Decomposition
Pitch, Period,
Number of Peaks
Autocorrelation
Figure 3.2 Feature Extraction on Speech Signals
23
Text Independent Speaker Recognition System with Ann
The robustness of any speaker recognition system depends on the appropriate selection of
acoustic features, feature extraction technique and the classifier involved in the system.
Initially text independent speaker recognition system using stockwell transform features and
DWT features is proposed with Back Propagation Network (BPN) as classifier.
The block diagram of ANN based speaker recognition system using stockwell transform
features is shown in Figure 5.1. Stockwell Transform is then applied on these signals to
obtain amplitude localized to both time and frequency. These amplitudes are then aggregated
using the statistical parameters namely mean, variance, Skewness and kurtosis. These
features are then used to train and test the ANN based classifier. In the proposed work, Back
Propagation Network (BPN) with three hidden layers and one output layer is chosen. The
number of neurons in the hidden layer is 10,10,5 respectively and one neuron in the output
layer. The learning parameter and the momentum parameters are 0.6 and 0.9 as these are the
optimized values. An exemplar is created with mean, variance, skewness and kurtosis as
input parameters and speaker code as output parameter. The speakers are coded from 1 to
10. Of the 100 exemplars generated, 50 are used for training and the other set of 50 is used
for testing the Network.
Speech
signals
Feature extraction using Feature aggregation using
Stockwell Transform statistical moments
Speaker
BPN based classifier Exemplar Generation
Figure 5.1 Block Diagram of ANN Based Speaker Recognition System using Stockwell
Transform Features
24
Separate set of exemplars are used for training and testing the network. The
performance of the proposed technique is measured in terms of relationship between the
desired and the actual output.
Feature extraction on the components from nonstationary analysisis anefficient way to
develop speaker recognition system. Discrete wavelet transform(DWT)is such a non
stationary analysis tool for converting the signals from time domain to spectral domain. DWT
decomposes the signal into approximation coefficients and detailed coefficients.
Approximation coefficients correspond to low frequencies and detailed coefficients represent
the high frequency information present in the signal. Since its formulation by Stephen Mallat,
various wavelet packets are cited in the literature. Performance of wavelet transform based
approaches is strongly dependent on the choice of the wavelet, order of the wavelet and level
of decomposition. These coefficients are aggregated through statistical parameters namely,
mean, variance, skewness, kurtosisetc. Figure 5.3 shows the various steps involved in an
automatic speaker recognition system. The speech signals are storedin“. Wav”format. DWT is
then applied on these signals to obtain amplitude localized to both time and frequency. These
amplitudes are then aggregated using the statistical parameters namely mean, variance,
Skewness, kurtosis, ZCR, Second order moments, third order moments and energy.These
features are then used to train and test the ANN based classifier.
25
Figure 5.3 ANN Based Speaker Recognition System Using Discrete Wavelet Transform
26
Applications of Fourier Transform
27
28
29
30
31
32
33
Sampling Theorem
A signal has three properties like voltage or amplitude, frequency, phase. The signals
are represented only in an analog form where the digital form of technology is not available.
Analog signals are continuous in time and difference in voltage levels for different periods of
the signal. Here, the main drawback of this is, the amplitude keeps on changing along with
the period of the signal. This can be overcome by the digital form of signal representation.
Here conversion of an analog form of the signal into digital form can be done using the
sampling technique. The output of this technique represents the discrete version of its analog
signal. Here in this article, you can find what is sampling theorem, definition, applications,
and its types.
What is the Sampling Theorem?
A continuous signal or an analog signal can be represented in the digital version in the form
of samples. Here, these samples are also called as discrete points. In sampling theorem, the
input signal is in an analog form of signal and the second input signal is a sampling signal,
which is a pulse train signal and each pulse is equidistance with a period of “Ts”. This
sampling signal frequency should be more than twice of the input analog signal frequency. If
this condition satisfies, analog signal perfectly represented in discrete form else analog signal
may be losing its amplitude values for certain time intervals. How many times the sampling
frequency is more than the input analog signal frequency, in the same way, the sampled
signal is going to be a perfect discrete form of signal. And these types of discrete signals are
well performed in the reconstruction process for recovering the original signal.
Statement: A continuous time signal can be represented in its samples and can be recovered
back when sampling frequency fs is greater than or equal to the twice the highest frequency
component of message signal. i. e.
fs≥2fm.
Proof: Consider a continuous time signal x(t). The spectrum of x(t) is a band limited to f m Hz
i.e. the spectrum of x(t) is zero for |ω|>ωm.
Sampling of input signal x(t) can be obtained by multiplying x(t) with an impulse train δ(t) of
period Ts. The output of multiplier is a discrete signal called sampled signal which is
represented with y(t) in the following diagrams:
34
Here, you can observe that the sampled signal takes the period of impulse. The process of
sampling can be explained by the following mathematical expression:
Sampled signal y(t)=x(t).δ(t)......(1)
35
36
Aliasing Effect
The overlapped region in case of under sampling represents aliasing effect, which can be
removed by
 considering fs >2fm
 By using anti aliasing filters.
Applications
There are few applications of sampling theorem are listed below. They are
 To maintain sound quality in music recordings.

 Sampling process applicable in the conversion of analog to discrete form.
 Speech recognition systems and pattern recognition systems.
 Modulation and demodulation systems
 In sensor data evaluation systems
 Radar and radio navigation system sampling is applicable.
 Digital watermarking and biometric identification systems, surveillance systems.
37
Mathematical Models
• Mathematical models serve as tools in the analysis and design of complex systems
• A mathematical model is used to represent, in an approximate way, a physical
process or system where measurable quantities are involved
• Typically a computer program is written to evaluate the mathematical model of the
system and plot performance curves – The model can more rapidly answer questions
about system performance than building expensive hardware prototypes
• Mathematical models may be developed with differing degrees of fidelity
• A system prototype is ultimately needed, but a computer simulation model may be
the first step in this process
• A computer simulation model tries to accurately represent all relevant aspects of the
system under study
• Digital signal processing (DSP) often plays an important role in the implementation
of the simulation model
• If the system being simulated is to be DSP based itself, the simulation model may
share code with the actual hardware prototype
• The mathematical model may employ both deterministic and random signal models
38
Engineering Applications
Communications, Computer networks, Decision theory and decision making,

Estimation and filtering, Information processing, Power engineering, Quality control,
Reliability, Signal detection, Signal and data processing, Stochastic systems, and others.
39
Random Signals in Practice
 A typical application of random signals concepts involves one or more of the
following:
– Probability
– Random variables
– Random (stochastic) processes
Example : Modeling with Random Processes
 Consider a random or stochastic process of the form
40
x(t) = A cos(2π fct + θ) + n(t)
which is a sinusoidal carrier plus noise
 In this example the carrier phase θ is modeled as a random variable and n(t) is
modeled as an independent stationary random process
 We may be interested in how to recover the sinusoidal carrier from the noisy signal
x(t)
 The power spectral density of a random process allows us to see the spectral content of
a signal
 The power spectral density of a wide sense stationary random process x(t) is given by the
Fourier transform of the autocorrelation function
41
Application Areas
• Control
• Communications
• Signal Processing
Control Applications
• Industrial control and automation (Control the velocity or position of an object)
• Examples: Controlling the position of a valve or shaft of a motor
• Important Tools:
– Time-domain solution of differential equations
– Transfer function (Laplace Transform)
– Stability
42
Communication Applications
• Transmission of information (signal) over a channel
• The channel may be free space, coaxial cable, fiber optic cable
• A key component of transmission: Modulation (Analog and Digital
Communication)
Signal Processing Applications

• Signal processing=Application of algorithms to modify signals in a way to make
them more useful.
• Goals:
– Efficient and reliable transmission, storage and display of information
– Information extraction and enhancement
• Examples:
– Speech and audio processing
– Multimedia processing (image and video)
– Underwater acoustic
– Biological signal analysis
Multimedia Applications
• Compression: Fast, efficient, reliable transmission and storage of data
• Applied on audio, image and video data for transmission over the Internet, storage
• Examples: CDs, DVDs, MP3, MPEG4
• Mathematical Tools: Fourier Transform, Quantization, Modulation Biological
Signal Analysis
• Examples:
– Brain signals (EEG)
– Cardiac signals (ECG)
– Medical images (x-ray, PET, MRI)
• Goals:
– Detect abnormal activity (heart attack, seizure)
– Help physicians with diagnosis
• Tools: Filtering, Fourier Transform
43
2. Rodger E. Ziemer , William H Tranter, D. R. Fannin,”Signals and Systems: Continuous
and Discrete”, 4th Edition, Pearson Education India, 2014.
3. Haykin S. and Van Been B., “Signals and Systems”, 2nd Edition, John Wiley and Sons,
2015.
44

SECA1301 Notes

Uploaded by

Copyright:

Available Formats

SECA1301 Notes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SECA1301 Notes

Uploaded by

Copyright:

Available Formats

SCHOOL OF ELECTRICAL AND ELECTRONICS

DEPARTMENT OF ELECTRONICS AND COMMMUNICATION

Ex: Sinusoidal signal, Voice signal, Rectangular pulse function

Unit Step Signal:

The Unit Step Signal u(t) is defined as

Impulse Signal DT representation

Impulse Signal CT representation

Real Exponential Signal: A real exponential signal is defined as x(t)=Ae σt

Basic Operations on signals:

-10 < t < -3 amplitude of z(t) = x1(t) + x2(t) = 0 + 2 = 2

-3 < t < 3 amplitude of z(t) = x1(t) + x2(t) = 1 + 2 = 3

3 < t < 10 amplitude of z(t) = x1(t) +x2(t) = 0 + 2 = 2

4. Signal Multiplication: y(t)= x1(t). x2(t)

As seen from the diagram above,

-10 < t < -3 amplitude of z (t) = x1(t) ×x2(t) = 0 ×2 = 0

-3 < t < 3 amplitude of z (t) = x1(t) ×x2(t) = 1 ×2 = 2

3 < t < 10 amplitude of z (t) = x1(t) × x2(t) = 0 × 2 = 0

x(t ± t0) is time shifted version of the signal x(t).

6. Time Reversal: If x(t) is a continuous function, the time-reversed signal is defined

|A| > 1 →→ Compression of the signal

|A| < 1 →→ Expansion of the signal

1. Even and Odd signal

Even and Odd Signal:

Random (or) Nondeterministic Signals:

1. x(t)=sin(4πt). It is periodic with period of 1/2

The Power in the signal is defined as

Energy of the signal is given as,

= │x(-1)│2 + │x(0)│2 + │x(1)│2 = 3

Real and Complex signals:

Real Exponential Signal: A real exponential signal is defined as x(t)=Aeσt

Complex exponential Signal:The complex exponential signal is given by x(t)=Ae st where

TEXT / REFERENCE BOOKS

Continuous Time Fourier Transform

Representation of Aperiodic signals: Starting from the Fourier series

representation for the continuous-time periodic square wave:

The Fourier coefficients akfor this square wave is

where 2 sin(wT1 ) / w represent the envelope of Tak

When T increases or the fundamental frequency w 0 2p / T decreases, the envelope is

Define the envelope X ( jw ) of Tak as,

This results in the following Fourier Transform.

Convergence of Fourier Transform

Then we guaranteed that X( jw ) is finite or equation 2.10 converges. If

An alterative set of conditions that are sufficient to ensure the convergence:

Contition1: Over any period, x(t ) must be absolutely integrable, that is

Examples of Continuous-Time Fourier Transform

If a is complex rather than real, we get the same result if Re{a}>0

Calculate the Fourier transform of the rectangular pulse signal

The inverse Fourier Transform of the sinc function is

Comparing the results we have,

Properties of Fourier Transform

If x(t)FX( jw) y(t)FY( jw)

ax(t)  by(t)FaX ( jw)  bY ( jw)

4. Differentiation and Integration

5. Time and Frequency Scaling

spectrum will be compressed.

That is reversing a signal in time also reverses its Fourier transform.

The convolution properties

𝑦 (𝑡 )= ℎ (𝑡) ∗ 𝒙 (𝑡)= 𝑌 (𝑗𝜔) = 𝐻 𝑗𝜔 𝑋(𝑗𝜔)

The frequency response of a differentiator.

From the differentiation property,

Here we have y(t)=z(t) h2(t)But z(t)=x(t)h1(t). Therefore y(t)=[x(t)h1(t)h2(t)]