Department o f E l e c t r i c a l & E l e c t r o n i c s E n g g

Subject: Signals and Systems-I

 SYLLABUS OF SIGNALS & SYSTEMS-I (3-1-0)
MODULE-I (10 HOURS)
Introduction of Signals, Classification of Signals, General Signal Characteristics, Signal
energy & Power, Continuous-Time Signals , Discrete-Time Signals
Basic System Properties, Systems with and without memory, Invertibility, casuality,
Stability, Time invariance, Linearity, Linear Time Invariant (LTI) Systems, Discrete
Time LTI Systems, Convolution
Representation of Linear Time-Invariant Discrete-Time Systems Convolution of
Discrete-Time
Signals Convolution Representation of Linear Time-Invariant Continuous-Time Systems
Convolution of Continuous-Time Signals, Properties of LTI Systems, Casual systems

MODULE-II (10 HOURS)

Fourier Representations for Signals: Representation of Discrete Time Periodic signals,
Continuous Time Periodic Signals, Discrete Time Non Periodic Signals, Continuous
Time Non-Periodic Signals, Properties of Fourier Representations,
Frequency Response of LTI Systems, Fourier Transform representation for Periodic and
discrete time Signals, Sampling, reconstruction, Discrete Time Processing of Continuous
Time Signals, Fourier Series representation for finite duration Nonperiodic signals.

MODULE-III (10 HOURS)

Modulation Types and Benefits, Full Amplitude Modulation, Pulse Amplitude
Modulation, Multiplexing, Phase and Group delays
Representation of Signals using Continuous time Complex Exponentials: Laplace
Transform, Unilateral Laplace Transform, its inversion, Bilateral Laplace Transform,
Transform Analysis of Systems

MODULE-IV (10 HOURS)

Representation of Signals using Discrete time Complex Exponentials: The Z-Transform,
Properties of Region of convergence, Inverse Z-Transform, Transform Analysis of LTI
Systems, Unilateral Z Transform.

BOOKS
[1] Simon Haykin and Barry Van Veen, “Signals and Systems”, John Wiley & Sons.
[2] Alan V. Oppenheim, Alan S. Willsky, with S. Hamid, S. Hamid Nawab, “Signals
and Systems”, PHI.
[3] Hwei Hsu, “Signals and Systems”, Schaum's Outline TMH
[4] Edward w. Kamen and Bonnie s. Heck, “Fundamentals of Signals & systems using
Web and MATLAB”,

***
 Contents
Lecture 1- Introduction of Signals and system

Lecture 2- Classification of Signals

Lecture 3- Classification of Signals (continued)

Lecture 4- General Signal Characteristics

Lecture 5- Operation on signals

Lecture 6- Fundamentals of Systems

Lecture 7- System properties

Lecture 8- System properties (continued)

Lecture 9- Linear Time Invariant System

Lecture 10- Convolution of Linear Time-Invariant Discrete-Time Signals

Lecture 11- Convolution Representation of Linear Time-Invariant Continuous-Time

Systems

Lecture 12- Properties of LTI Systems, Casual systems

Lecture 13- Fourier Representations for Signals:

Lecture 14- Fourier Representations of Continuous Time Periodic Signals

Lecture 15- Fourier Representations of Discrete Time Periodic signals

Lecture 16- Fourier Representations of Continuous Time Non Periodic Signals

Lecture 17- Fourier Representations of Discrete Time Non Periodic Signals

Lecture 18- Properties of Fourier Representations

Lecture 19- Properties of Fourier Representations (continued)

Lecture 20- Frequency Response of LTI Systems

Lecture 21- Fourier Transform representation for Periodic and discrete time Signals

Lecture 22- Sampling

Lecture 23- Reconstruction

Lecture 24- Discrete Time Processing of Continuous Time Signals

Lecture 25- Fourier Series representation for finite duration Nonperiodic signals.
Lecture 26- Modulation Types and Benefits

Lecture 27- Full Amplitude Modulation

Lecture 28- Pulse Amplitude Modulation

Lecture 29- Multiplexing

Lecture 30- Phase and Group delays

Lecture 31- Representation of Signals using Continuous time Complex Exponentials:

Laplace Transform

Lecture 32- Unilateral Laplace Transform, its inversion

Lecture 33- Laplace Transform Properties

Lecture 34- Representation of Signals using Discrete time Complex Exponentials:

The Z-Transform

Lecture 35- Properties of Region of convergence

Lecture 36- Inverse Z-Transform

Lecture 37- Inverse Z-Transform (continued)

Lecture 38- Z-Transform Properties

Lecture 39- Z-Transform Properties (continued)

Lecture 40- Transform Analysis of LTI Systems

 Introduction to Signals:

What is a Signal ?

 A signal is formally defined as a function of one or more variables that conveys

information on the nature of a physical phenomenon.
 When the function depends on a single variable, the signal is said to be one
dimensional. E.g.; Speech signal (Amplitude varies with respect to time)
 When the function depends on two or more variables, the signal is said to be
multidimensional. E.g.; Image – 2D (Horizontal & vertical coordinates of the
images are two dimensional)

What is a System ?

 A system is formally defined as an entity that manipulates one or more signals to

accomplish a function, thereby yielding new signals.

i/p signal System o/p signal

e.g.; In a communication system the input signal could be a speech signal or

computer data. The system itself is made up of the combination of a transmitter,
channel and a receiver. The output signal is an estimate of the information
contain in the original message.
Message signal Transmitted signal Received signal Estimate of message
signal
Transmitter Channel Receiver

The examples of other systems are control systems, biomedical signal processing
system, audio system, remote sensing system, microelectro mechanical system etc.

General signal characteristics:

(a) Multichannel & multidimensional signals:
 A signal is described by a function of one or more independent variables.
 The value of the function (dependent variable) can be real valued scalar quantity,
a complex valued quantity or perhaps a vector.
Real valued signal x1 (A) = A sin3πt
Complex valued signal x2(A) = Ae j3 πt = A cos3πt + jAsin3πt
 In some applications, signals are generated by multiple sources or multiple
sensors. Such signals can be represented in vector form and we refer such a
vector of signal as a multichannel signal.
E.g.; In electrocardiography, 3-lead & 12-lead electrocardiograms (ECG) are
often used, which result in 3-channel & 12-channel signals.
One dimensional: If the signal is a function of a single
independent variable, the signal is called 1-D signal. Amp
e.g.; Speech signal

Time
Multidimensional signal: Signals can be functions
of more than one variable, e.g., image signals (2D),
Colour image (3D), etc.

Classification of signals
Broadly we classify signals as:
1
1. Continuous-time signal: A signal x(t), is
said to be continuous-time signal if it is
defined for all time t, where t is a
real-valued variable denoting time. t
Ex: x(t) = e-3tu(t)

Discrete-time signal: A signal x(n), is said

to be discrete-time signal; if it is defined only X[n] e-3
at discrete instant of time, where n is an e-6
integer-valued variable denoting the discrete
e-9
samples of time. We use square brackets [·] to
denote a discrete-time signal.
Ex: x[n] = e-3nu[n]
0 1 2 3
2. Even and odd signal:

A continuous-time signal x(t) is even, if x(-t) = x(t)

and it is odd if x(-t) = -x(t).
A discrete-time signal x[n] is even if x[-n] = x[n]
and is odd if x[-n] = -x[n].
Example 1: x(t) = t2 - 40 is even.
Example 2: x(t) = 0.1t3 is odd.
Example 3: x(t) = e0.4t is neither even nor odd.

(a) (b) (c)

Figure: Illustrations of odd and even functions. (a) Even; (b) Odd; (c) Neither.
Decomposition Theorem
Every continuous-time signal x(t) can be expressed as:

x(t) = 𝑦 𝑡 + 𝑧 𝑡

where 𝑦 𝑡 is even, and 𝑧 𝑡 is odd.

𝑥 𝑡 + 𝑥(−𝑡)
𝑦 𝑡 =
2
and
𝑥 𝑡 − 𝑥(−𝑡)
𝑧 𝑡 m=
2

3. Periodic & non-periodic signals:

A continuous time signal x(t) is periodic if there is a constant T > 0, such that
x(t) = x(t + T ), for all t
A discrete time signal x[n] is periodic if there is an integer constant N > 0, such that
x[n] = x[n + N ], for all n

Signals do not satisfy the periodicity conditions are called non-periodic signals.
Note: The smallest value of T (N) that satisfies the above equations is called fundamental period
Example: Determine the fundamental period of the following signals:
(a) ej3πt/5
(b) ej3πn/5
Solution:
(a) Let x(t) = ej3πt/5 . If x(t) is a periodic signal, then there exists T > 0
such that x(t) = x(t + T ). Therefore,
x(t) = x(t + T )
j3πt/5 j3π(t+T)/5
⇒e =e
j3πT/5
⇒1=e
j2kπ j3πT/5
⇒e =e
10
⇒T= 3
(k = 1)

(b) L et x[n] = ej3πn/5 . If x[n] is a periodic signal, then there exists an integer N > 0
such that x[n] = x[n + N ]. So,
x[n] = x[n + N ]
j3πn/5 j3π(n+N)/5
⇒e =e
j3πN/5
⇒ 1=e
j3πN/5
⇒ ej2kπ = e
⇒ T = 10 (k = 3)
4. Energy signals and power signals:

In electrical systems, a signal may represent a voltage or a current. Consider a voltage

v(t)developed across a resistor R, producing a current i(t). The instantaneous power
dissipated in this resistor is defined by
Define the total energy of the continuous-time signal x( t) as
𝑇/2
𝐸 = lim 𝑥 2 𝑡 𝑑𝑡
𝑇→∞ −𝑇/2

∞
= lim 𝑥 2 𝑡 𝑑𝑡
𝑛→∞ −∞

and its time-averaged, or average, power as

𝑇/2
1
𝑃 = lim 𝑥 2 𝑡 𝑑𝑡
𝑇→∞ 𝑇 −𝑇/2

From above equation, we readily see that the time-averaged power of a periodic signal
x(t) of fundamental period T is given by

𝑇/2
1
𝑃= 𝑥 2 𝑡 𝑑𝑡
𝑇 −𝑇/2

The square root of the average power P is called the root mean-square (rms) value of the
periodic signal x( t).
In the case of a discrete-time signal x[n], the integrals in above equations are replaced by
corresponding sums. Thus, the total energy of x[ n] is defined by
∞

𝐸= 𝑥 2 [𝑛]
𝑛 =−∞
and its average power is defined by
𝑁
1
𝑃 = lim 𝑥 2 [𝑛]
𝑛→∞ 2𝑁 + 1
𝑛=−𝑁

A signal is referred to an energy signal if and only if the total energy is finite .i.e.,
0<E<∞

A signal is referred to an power signal if and only if the average power is finite .i.e.,
0<P<∞
Note: Energy signal has zero time average power and power signal has infinite energy.
Example: x(n) = (- 0.5) nu[n]

Solution:

∞ 2 ∞ 𝑛 1 4
𝐸= 𝑛=−∞ 𝑥 [𝑛] = 𝑛=0 0.25 = 1−0.25
= 3
<∞

1 𝑁 2 1 𝑁 𝑛 1 𝑁 𝑛
𝑃 = lim𝑛→∞ 2𝑁+1 𝑛=−𝑁 𝑥 [𝑛] = lim𝑛 →∞ 2𝑁+1 𝑛=0 0.25 = 2∞+1 𝑛=0 0.25 =0

We got power zero and finite energy. Hence it is an energy signal.

5. Deterministic signals and random signals:

The deterministic signal is a signal about which there is no uncertainty with respect to
its value at any time. The deterministic signals may be modeled as completely specified
function of time.
Example: x(t) = cos2(2πt)
A random signal is a signal about which there is uncertainty before it occurs.
Example: The electrical noise generated in the amplifier of a radio or television
receiver.

Basic Operations of Signals

Operation performed on independent variable:

Time Shift
For any t0 and n0 , time shift is an operation defined as
x(t) → x(t - t0 )
x[n] → x[n -n0].
If t0 > 0, the time shift is known as “delay”. If t0 < 0, the time shift is known
as “advance”.
Example. In Fig. gi ven below, the left image shows a continuous-time signal
x(t). A time- shifted version x(t - 2) is shown in the right image.

Figure: An example of time shift.

Time Reversal
Time reversal is defined as x(t) → x(-t)
x[n] → x[-n],

which can be interpreted as the “flip over the y-axis”.

 Example:

Figure: An example of time reversal.

Time Scaling
Time scaling is the operation where the time variable t is multiplied by a constant a:
x(t) → x(at), a>0

If a > 1, the time scale of the resultant signal is “decimated” (speed up).
If 0 < a < 1,
the time scale of the resultant signal is “expanded” (slowed down).

Figure : An example of time scaling.

Decimation and Expansion

Decimation and expansion are standard discrete-time signal processing operations.

Decimation is defined as
yD [n] = x[M n], for some integers M

Where, M is the decimation factor.

Expansion is defined as
𝑛
x[ 𝐿 ] , n = integer multiple of L
yE [n] =
0, otherwise.

Where, L is the expansion factor.

 Figure 1.8: Examples of decimation and expansion for M = 2
and L = 2.

Combination of Operations
Generally, linear operation (in time) on a signal x(t) can be expressed as
y(t) = x(at-b). The recommended method is “Shift, then Scale”.

Example: The signal x(t) shown in Figure of sketch x(3t - 5).

Figure: An example of Shift, then Scale

Operation performed on dependent variable :

Amplitude scaling:
Let x(t) denote a continuous time signal
By amplitude scaling, we get y(t) = cx(t)
Where, c is the scaling factor.
Example: An electronic amplifier, a device that performs amplitude scaling.
For discrete time signal y[n] = cx[n]
Amplitude addition:
Let x1(t) and x2(t) is a pair of continuous time signal
By adding these two signals, we get y(t) = x1(t) + x2(t)
Example: An audio mixture
For discrete time signal, y[n] = x1[n] + x2[n]
Amplitude multiplication:
Let x1(t) and x2(t) is a pair of continuous time signal
By multiplying these two signals, we get y(t) = x1(t) x2(t)
Example: An AM radio signal, in which
x1(t) is an audio signal
x2(t) is an sinusoidal carrier wave
For discrete time signal, y[n] = x 1[n] x2[n]
Differentiation:
𝑑
y(t) 𝑑𝑡 𝑥(𝑡)
𝑑
Example: Voltage across an inductor L  v(t) 𝑑𝑡 𝑖(𝑡)
Integration:
𝑡
y(t)= −∞
𝑥 𝜏 𝑑𝜏
1 𝑡
Example: Voltage across a capacitor C  y(t)= 𝐶 −∞
𝑖 𝜏 𝑑𝜏
Elementary Signals

Several elementary signals feature prominently in the study of signals and systems. These are
exponential and sinusoidal signals, the step function, the impulse function, and the ramp
function, all of which serve as building blocks for the construction of more complex signals
Exponential Signals

A real exponential signal, in its most general form, is written as

x(t) = Beat,
where both B and a are real parameters. The parameter B is the amplitude of the exponential
signal measured at time t = 0. Depending on whether the other parameter a is positive or
negative, we may identify two special cases:

Fig: Growing exponential, for a > 0 Decaying exponential, f o r a < 0

In discrete time, it is common practice to write a real exponential signal as x[n] = Brn
 Fig: Growing exponential for r > 1 Decaying exponential for 0 < r <1

Impulse functions

The discrete-time version of the unit impulse is defined by

1, n=0
δ [n] =
0, n≠0

Fig: Discrete time form of unit impulse

The continuous-time version of the unit impulse is defined by the following pair of relations:
δ (t)
δ (t) = 0 for t ≠ 0
∞ 1
and −∞
𝛿 𝑡 𝑑𝑡 = 1

0 t

Fig: Continuous time form of unit impulse

Above equation says that the impulse δ (t) is zero everywhere except at the origin. Equation says that
the total area under the unit impulse is unity. The impulse δ (t) is also referred to as the Dirac delta
function.

Step function:
The discrete-time version of the unit- step function is defined by:
 1, t>0
The continuous-time version of the unit- step function is defined by: u(t) =
0, t<0

Ramp function:
The integral of the step function u(t) is a ramp function of unit slope.

Fig: Ramp function of unit slope

n, n≥0
The discrete-time version of the unit- ramp function is defined by: r[n] = 0, n<0

Introduction to Systems

 Systems are used to process signals to allow modification or extraction of

additional information from the signal.
 A system may consist of physical components (hardware realization) or an
algorithm (operator) that computes the output signal from the input signal.
 A physical system consists of inter-connected components which are characterized
by their input-output relationships.
Figure 2.1: Continuous-time and discrete-time systems: Here H & T are operators.

H = Hx(t) T = Tx(n)
Properties of systems: (Classification of systems):

1 Static (Memoryless) & Dynamic (with memory):

Static: A system is static if the output at time t (or n) depends only on the input at time t (or
n).
Examples:
1. y(t) = (2x(t) - x2 (t))2 is memoryless, because y(t) depends on x(t) only. There is no x(t -
1), or x(t + 1) terms, for example.
2. y[n] = x2[n] is memoryless. In fact, this system is passing the input to output directly,
without any processing.
1
3. Current flowing through a resistor i.e., i(t) = v(t)
𝑅
Dynamic: A system is said to possess memory if its output signal depends on past or future values of
input.
Example:
1. Inductor and capacitor, since the current flowing through the inductor at time „t‟ depends on the all
1 𝑡 1 𝑡
past values of the voltage v(t) i.e., i(t) = −∞ 𝑣(𝜏)𝑑𝜏 and v(t) = −∞ 𝑖(𝜏)𝑑𝜏
𝐿 𝐶
1
2. The moving average system given by y(n)= (x[n]+x[n-1]+x[n-2])
3

2 Stable & unstable system:

 A system is said to be bounded-input, bounded-output (BIBO) stable if and only

if every bounded input results in a bounded output, otherwise it is said to be
unstable.
 If for |x(t)| ≤ Mx < ∞ for all t, output is |y(t)| ≤ My < ∞ for all t; where Mx & My
are some finite positive number.

Example: 1. y(t) = x (t-3) is a stable system.

2. y(t) = t x(t) is an unstable system.
3. y[n] = ex[n] is a stable system.

Assume that |x(n)| ≤ Mx < ∞, for all „t‟

y[n] = ex[n] = eMx = finite  Stable

4. y[n] = rn x[n], where r > 1

Assume that |x(n)| ≤ Mx < ∞, for all „t‟, then

|y[n]| = |rn x[n]| = |rn| | x[n]|
as „n‟  ∞ |rn| ∞
so y[n]∞ hence unstable.

3 Causal and non-Causal system:

Causal: A system is said to be causal if the present value of output signal depends
only on the present or past values of the input signal. A causal system is also known as
physical or non-anticipative system.
1
Example:1. The moving average system given by y(n)= 3 (x[n]+x[n-1]+x[n-2])
2. y(t) = x(t)cos 6t
Note: i) Any practical system that operates in real time must necessarily be causal.
ii) All static systems are causal.

Non-Causal: A system is said to be non-causal if the present value of output

signal depends on one or more future values of the input signal.
1
Example:1. The moving average system given by y(n)= (x[n]+x[n-1]+x[n+2])
3

4 Time invariant and time variant system:

Time invariant: A system is time-invariant if a time-shift of the input signal results in
the same time-shift of the output signal.
That is, if
x(t) → y(t),
then the system is time-invariant if

x(t - tO ) → y(t - tO ), for any tO .

Figure 2.2: Illustration of a time-invariant system.

Example 1.
The system y(t) = sin[x(t)] is time-invariant
Proof. Let us consider a time-shifted signal x1 (t) = x(t - tO ). Correspondingly, we let y1 (t) be
the output of x1 (t). Therefore,

y1 (t) = sin[x1 (t)] = sin[x(t - tO )].

Now, we have to check whether y1 (t) = y(t - tO ). To show this, we note that

y(t - tO ) = sin[x(t - tO )],

which is the same as y1 (t). Therefore, the system is time-invariant.

Time variant: A system is time-variant if its input –output characteristic changes with time.

Example 2:
The system y[n] = nx[n] is time-variant.

Proof: Output for a time shifted input is

y[n] | x(n-k) = nx(n-k)
then the same time shifted output is
y(n-k) = (n-k)x(n-k)
the above two equations are not same. Hence it is time variant.
4 Linear and non-linear system:

Linear system: A system is said to be linear if it satisfies two properties i.e.; superposition &
homogeneity.
Superposition: It states that the response of the system to a weighted sum of signals
be equal to the corresponding weighted sum of responses (Outputs of the system to
each of the individual input signal.
For an input x(t) = x1(t), the output y(t) = y1(t)
and input x(t) = x2(t), the output y(t) = y2(t)
then, the system is linear if & only if
T [a1x1(t) + a2x2(t)] = a1T [x1(t)] + a2T [x2(t)]
Homogeneity: If the input x(t) is scaled by a constant factor „a‟, then the output y(t) is also scaled by
exactly the same constant factor „a‟.

For an input x(t)  output y(t)

and input x1(t) = ax (t)output y1(t) = ay(t)
Example 1:
The system y(t) = 2πx(t) is linear. To see this, let‟s consider a signal

x(t) = ax1 (t) + bx2 (t),

where y1 (t) = 2πx1 (t) and y2 (t) = 2πx2 (t). Then

ay1 (t) + by2 (t) = a (2πx1 (t)) + b (2πx2 (t))

= 2π [ax1 (t) + bx2 (t)] = 2πx(t) = y(t).

Example 2.
The system y[n] = (x[2n])2 is not linear. To see this, let‟s consider the signal

x[n] = ax1 [n] + bx2 [n],

where y1 [n] = (x1 [2n]) and y2 [n] = (x2 [2n])2 . We want to see whether y[n] =
2

ay1 [n] + by2 [n]. It holds that

ay1 [n] + by2 [n] = a (x1 [2n])2 + b (x2 [2n])2 .

However,

y[n] = (x[2n])2 = (ax1 [2n] + bx2 [2n])2 = a2 (x1 [2n])2 + b2 (x2 [2n])2 + 2abx1 [n]x2 [n].

5 Invertible and non-invertible system:

A system is said to be invertible if the input of the system can be recovered from the
output.
Let the set of operations needed to recover the input represents the second system
which is connected in cascade with the given system such that the output signal of the
second system is equal to the input applied to the given system.
 Let H the continuous time system
x(t) input signal to the system
y(t) output signal of the system
Hinv the second continuous time system

x(t) y(t) Hinv x(t)

H

The output signal of the second system is given by

Hinv{y(t)} = Hinv{Hx(t)}
= HinvH{x(t)}
For the output signal to equal to the original input, we require that

H Hinv = I

Where „I‟ denotes the identity operator.

The system whose output is equal to the input is an identity system. The operator Hinv must satisfy the
above condition for H to be an invertible system. Cascading a system, with its inverse system, result in an
identity system.
Example:
An inductor is described by the relation
1 𝑡
y(t) = 𝐿 −∞
𝑥(𝜏)𝑑𝜏 is an invertible system

because, by rearranging terms, we get

𝑑
x(t) = L 𝑑𝑡 y(t),

which is the inversion formula.

Note:i) A system is not invertible unless distinct inputs applied to the system produce distinct outputs.
ii) There must be a one to one mapping between input and output signal for system to be invertible.
Non-invertible System: When several dibfferent inputs results in the same output, it is impossible to
obtain the input from output. Such system is called a non-invertible system.
Example: A square-law system described by the input output relation
y(t) = x2(t), is non-invertible,
because distinct inputs x(t) & -x(t) produce the same output y(t) [not distinct output].

Linear –time convolution system (LTI)

Linear time invariant (LTI) systems are good models for many real-life systems, and they have
properties that lead to a very powerful and effective theory for analyzing their behavior. The LTI
systems can be studied through its characteristic function, called the impulse response. Further, any
arbitrary input signal can be decomposed and represented as a weighted sum of unit sample sequences.
As a consequence of the linearity and time invariance properties of the system, the response of the
system to any arbitrary input signal can be expressed in terms of the unit sample response of the
system. The general form of the expression that relates the unit sample response of the system and the
arbitrary input signal to the output signal, called the convolution sum, is also derived.

Resolution of a Discrete-time signal into impulses:

Any arbitrary sequence x(n) can be represented in terms of delayed and scaled impulse sequence δ(n).
Let x(n) is an infinite sequence as shown in figure below.

●●● ●●●

Figure 1.13: Representing of a signal x[n] using a train of impulses δ[n - k].

The sample x(0) can be obtained by multiplying x(0), the magnitude, with unit impulse
δ(n)
𝑥(0), 𝑛=0
i.e., x[n] δ[n] = 𝑛≠0
0,
Similarly, the sample x(-3) can be obtained as shown in the figure.
𝑥(−3), 𝑛=−3
i.e., x[-3] δ[n+3] =
0, 𝑛≠−3
In the same way we can get the sequence x[n] by summing all the shifted and scaled
impulse function
i.e., x[n] = …. x[-3] δ[n+3] + x[-2] δ[n+2] + …. + x[0] δ[n] + ….+ x[4 ] δ[n-4]…
∞
= 𝑘=−∞ 𝑥 𝑘 𝛿(𝑛 − 𝑘)

Impulse response and convolution sum:

Impulse response: A discrete-time system performs an operation on an input signal

based on predefined criteria to produce a modified output signal. The input signal x[n] is
the system excitation, and y[n] is the system response. The transform operation is shown
in the figure below.

x[n] y[n]=T[x[n]]
T

If the input to the system is the unit impulse i.e., x[n] = δ[n], then the output of the
system is known as impulse response represented by h[n] where
h[n] = T [δ[n]]
Response of LTI system to arbitrary inputs: The convolution sum

From the above discussion, we get the response of an LTI system to an unit impulse as
the impulse response h[n] i.e.,
δ[n] h[n]
δ[n-k] h[n-k], by time invariant property
x(k)δ[n-k] x(k)h[n-k], by homogeneity principle
∞ ∞
𝑘=−∞ 𝑥(𝑘)𝛿[𝑛 − 𝑘] 𝑘=−∞ 𝑥(𝑘)ℎ[𝑛 − 𝑘], by super position

As we know the arbitrary input signal is a weighted sum of impulse, the LHS = x[n]
having a response in RHS = y[n] known as convolution summation.
i.e., x[n] y[n]

In other words, given a signal x[n] and the impulse response of an LTI system
h[n], the convolution between x[n] and h[n] is defined as

𝑦[𝑛] = 𝑥(𝑘)ℎ[𝑛 − 𝑘]
𝑘=−∞

We denote convolution as y[n] = x[n] ∗ h[n].

• Equivalent form: Letting m = n - k, we can show that

∞ ∞ ∞

𝑥(𝑘)ℎ[𝑛 − 𝑘] = 𝑥(𝑛 − 𝑚)ℎ[𝑚] = 𝑥[𝑛 − 𝑘]ℎ[𝑘]

𝑘=−∞ 𝑚 =−∞ 𝑘=−∞

Properties of convolution:
The following “standard” properties can be proved easily:
1. Commutative: x[n] ∗ h[n] = h[n] ∗ x[n]

2. Associative: x[n] ∗ (h1 [n] ∗ h2 [n]) = (x[n] ∗ h1 [n]) ∗ h2 [n]

3. Distributive: x[n] ∗ (h1 [n] + h2 [n]) = (x(t) ∗ h1 [n]) + (x[n] ∗ h2 [n])
 How to Evaluate Convolution?
To evaluate convolution, there are four basic steps:
1. Fold 3. Multiply

2. Shift 4. Summation

Example1: Consider the signal x[n] and the impulse response h[n] shown below.

Let‟s compute the output y[n] one by one. First, consider y[0]:
∞ ∞

𝑦[𝑜] = 𝑥 𝑘 ℎ [ 0 − 𝑘] = 𝑥 𝑘 ℎ[−𝑘] = 1
𝑘 =−∞ 𝑘 =−∞

∞
Note that h[-k] is the flipped version of h[k], and 𝑘=−∞ 𝑥 𝑘 ℎ[−𝑘] = 1 is the multiply-
add between x[k] and h[-k].

To calculate y[l], we flip h[k] to get h[-k], shift h[-k] go get h[l-k], and multiply-add
∞
to get 𝑘=−∞ 𝑥 𝑘 ℎ[1 − 𝑘] . Therefore

∞ ∞

𝑦[1] = 𝑥 𝑘 ℎ[1 − 𝑘] = 𝑥 𝑘 ℎ[1 − 𝑘] = 1 × 1 + 2 × 1 = 3

𝑘=−∞ 𝑘=−∞

The calculation is shown in the figure below.

System Properties
With the notion of convolution, we can now proceed to discuss the system
properties in terms of impulse responses.
 Memoryless
A system is memoryless if the output depends on the current input only. An
equivalent statement using the impulse response h[n] is that:

An LTI system is memoryless if and only if

h[n] = aδ[n], for some a.
Invertible
An LTI system is invertible if and only if there exist g[n] such that

h[n] ∗ g[n] = δ[n].

Causal

An LTI system is causal if and only if

h[n] = 0, for all n < 0.

Stable
An LTI system is stable if and only if
∞

ℎ[𝑘] < ∞
𝑘=−∞

∞
Proof: Suppose that 𝑘=−∞ ℎ[𝑘] < ∞. For any bounded signal |x[n]| ≤ B, the output is

|𝑦[𝑛]| ≤ 𝑥[𝑘]ℎ[𝑛 − 𝑘]
𝑘=−∞

= 𝑥[𝑘] . |ℎ[𝑛 − 𝑘]|

𝑘=−∞
∞

≤ 𝐵. |ℎ[𝑛 − 𝑘]|
𝑘=−∞

Therefore, y[n] is bounded.

 Continuous-time Convolution
The continuous-time case is analogous to the discrete-time case. In continuous-
time signals, the signal decomposition is
∞
𝑥 𝑡 = 𝑥 𝜏 𝛿 𝑡 − 𝜏 𝑑𝜏
−∞

and consequently, the continuous time convolution is defined as

∞
𝑡 = 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
−∞

Example: The continuous-time convolution also follows the three step rule: flip, shift,
multiply- add. Let us consider the signal x(t) = e-at u(t) for a > 0, and impulse response
h(t) = u(t). The output y(t) is
Case A: t > 0:
∞
𝑦 𝑡 = 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
−∞

∞
= 𝑒 −𝑎𝑟 𝑢(𝜏) 𝑢 𝑡 − 𝜏
−∞
𝑡
= 𝑒 −𝑎𝑟 𝑑𝜏
0
1
=
−𝑎
[1-e-at]
Case B: t ≤ 0:
y(t) = 0.
Therefore,
1
𝑦 𝑡 = 𝑎 [1 − 𝑒 −𝑎𝑡 ]u(t)

Properties of CT Convolution
The following properties can be proved easily:
1. Commutative: x(t) ∗ h(t) = h(t) ∗ x(t)
2. Associative: x(t) ∗ (h1 (t) ∗ h2 (t)) = (x(t) ∗ h1 (t)) ∗ h2 (t)
3. Distributive: x(t) ∗ [h1 (t) + h2 (t)] = [x(t) ∗ h1 (t)] + [x(t) ∗ h2 (t)]

Continuous-time System Properties

The following results are analogous to the discrete-time case.

Memoryless.
An LTI system is memoryless if and only if
 h(t) = aδ(t), for some a
Invertible.
An LTI system is invertible if and only if there exist g(t) such that
h(t) ∗ g(t) = δ(t).
Causal.
A system is causal if and only if

h(t) = 0, for all t < 0

Stable.
A system is stable if and only if
∞
|ℎ 𝜏 | 𝑑𝜏 < ∞
−∞
Interconnection of LTI systems:

1. Parallel connection of LTI System: Consider two LTI systems with impulse
responses h1(t)and h2(t) connected in parallel as shown in the figure below. The output
of this connections of systems, y(t), is the sum of the outputs of the two systems i.e.,

y(t)= y1(t)+ y2(t)= x(t) ∗ h 1 (t)+ x(t) ∗ h 2 (t)

= x(t) ∗ [h 1 (t)+ h 2 (t)]

Identical results hold for the discrete time case.

x(n) ∗ h 1 (n)+ x(n) ∗ h 2 (n) = x(n) ∗ [h 1 (n)+ h 2 (n)]

Fig: Parallel interconnection of two LTI system & its equivalent systems

2. Cascade connection of LTI System: Consider the cascade connection of two LTI
systems as shown in the figure. The output of this connection of systems

y(t)= {x(t) ∗ h 1 (t) ∗ h 2 (t)}

Using associative property of convolution, we get

y(t)= x(t) ∗{h 1 (t) ∗ h 2 (t)}

 Fig: Cascade Interconnection of two LTI system & its equivalent systems

Step response: Step input response are often used to characterize the response of an
LTI system to sudden changes in the input. It is defined as the output due to a unit step
input signal.
Let h[n] be the impulse response of a discrete-time LTI system and s[n] be the step
response.
Then, s[n] = h[n] ∗ u[n]
∞
= 𝑘 =−∞ ℎ[𝑘]𝑢[𝑛 − 𝑘]

Now, as u[n-k] = 0 for k > n and u[n-k] = 1 for k ≤ n,

∞

𝑠[𝑛] = ℎ[𝑘]
𝑘=−∞

i.e., the step response is the running sum of the impulse response. Similarly, the
step response s(t) of a continuous-time system is expressed as the running
integral of the impulse response:
𝑡
𝑠 𝑡 = ℎ 𝜏 𝑑𝜏
−∞

Note: These relationships may be inverted to express the impulse response in terms of
the step response as

h[n] = s[n] – s[n-1]

𝑑
and, h(t) = 𝑠(𝑡)
𝑑𝑡
Fourier Representations for Signals

In this chapter, the signal is represented as a weighted superposition of complex sinusoids.

If such a signal is applied to an LTI system, then the system output is a weighted
superposition of the system response to each complex sinusoid. Representing signal as
superposition of complex sinusoids not only leads to a useful expression for the system
output, but also provides an insightful characterization of the signals and systems. The
study of signals and systems using sinusoidal representation is known as Fourier analysis
named after Joseph Fourier.
Basing on the periodicity properties of the signal and whether the signal is discrete or
continuous in time, there are four different types of Fourier representations, each
applicable to a different class of signals.

Complex sinusoids and frequency response of LTI systems:

The response of an LTI system to a sinusoidal input leads to a characterization of system

behavior termed as frequency response of the system. This characterization is obtained in
terms of the impulse response by using convolution and a complex sinusoidal input signal.
Let us consider the output of a discrete-time LTI system with impulse response h[n] and
unit amplitude complex sinusoidal input x[n] = e jΩn . This output is given by:
∞

𝑦[𝑛] = ℎ[𝑘]𝑥[𝑛 − 𝑘]
𝑘=−∞
∞

= ℎ[𝑘] 𝑒 𝑗 𝛺(𝑛 −𝑘)

𝑘=−∞

We factor e jΩn from the sum to get

∞
𝑗𝛺𝑛
𝑦[𝑛] = 𝑒 ℎ[𝑘] 𝑒 −𝑗𝛺𝑘
𝑘=−∞

= H(𝑒 𝑗𝛺 )𝑒 𝑗𝛺𝑛
Where we have defined
∞
𝑗𝛺
𝐻(𝑒 ) = ℎ[𝑘] 𝑒 −𝑗𝛺𝑘
𝑘=−∞
Hence, the output of the system is a complex sinusoid of the same frequency as the input,
multiplied by the complex number 𝐻(𝑒 𝑗𝛺 ) . The relationship is shown in figure below:

The complex scaling factor 𝐻(𝑒 𝑗𝛺 ) is not a function of time n, but only is a function of
frequency Ω and is termed the frequency response of the discrete-time system.
The results obtained for continuous-time LTI system is similar to the above.
 jωt
Let the impulse response of such a system be h(t) and the input be x(t) = e . Then the
convolution integral gives the output as

Where we define,

The above equation is referred to as frequency response of the continuous time system.
Writing the complex valued frequency response H(jω) in polar form

H(jω) = |H(jω)|e jυ
Where,
|H(jω)| => magnitude response
And, φ => phase response = arg{H(jω)}

Example: The impulse response of the system given the figure below is

Find an expression for the frequency response and plot the magnitude and phase response.

Solution: Substituting h(t) in equation of H(jω), we get

The magnitude response is:

While the phase response is

arg{H(jω)} = - arctan(ωRC)

Fig: (a) Magnitude-response (b) Phase-response

Eigenvalue and Eigenfunctions of an LTI System

Definition: For an LTI system, if the output is a scaled version of its input, then the
input function is called an eigenfunction of the system. The scaling factor is called the
eigenvalue of the system.
We take the complex sinusoid ψ(t) = e jωt is an eigenfunction of the LTI system H
associated with the eigenvalue λ = H(jω), because ψ satisfies an eigenvalue problem
described by
H{ψ(t)} = λψ (t)
 The effect of the system on an eigenfunction input signal is scalar multiplication. The
output is given by the product of the input and a complex number. This eigen
representation is shown in the figure below.

Fourier representation of four classes of signals:

 There are four distinct Fourier representations, each applicable to a different class of
signals.
 The Fourier series (FS) applies to continuous time periodic signal, and the discrete -
time Fourier series (DTFS) applies to discrete time periodic signal.
 The Fourier transform (FT) applies to a signal that is continuous in time and
nonperiodic.
 The discrete-time Fourier transform (DTFT) applies to a signal that is discrete in
time and nonperiodic.
Relationship between time properties of a signal and the appropriate Fourier representation
is given below:

Continuous-time periodic signals: The Fourier series

Continuous-time periodic signals are represented by the Fourier series (FS). We may write
the FS of a signal x(t) with fundamental period T and fundamental frequency ω0 = 2π/T, as

Where,
are the FS coefficient of the signal x(t). We say that x(t) and X[k] are an FS pair and denote
this relationship as

The Fourier series coefficients are known as the frequency-domain representation of

x(t),because each FS coefficient is associated with a complex sinusoid of different
frequency.
In the representation of the periodic signal x(t) by the Fourier series, the issue arises, is
whether or not the series converges to x(t) for each value of t, i.e., whether the signal x(t)
and its FS representation are equal at each value of t.
The Dirichlet conditions guarantee that the FS will be equal to x(t), except at the value of t
for which x(t) is discontinuous. At these values of t, FS converges to the mid-point of the
discontinuity. The Dirichlet conditions are:

1. The signal x(t) has a finite number of discontinuities in any period.

2. The signal contains a finite number of maxima and minima during any period.
3. The signal x(t) is absolutely integrable (bounded) i.e.,
∞
𝑥 𝑡 𝑑𝑡 < ∞
𝑇
If x(t) is periodic and satisfies the Dirichlet condition, it can be represented in FS.

Direct calculation of FS coefficients:

Example: Determine the FS coefficient for signal x(t)

Solution: Time period T = 2, Hence, ω0 = 2π/2 = π. On the interval 0 ≤ t ≤ 2, one period of

x(t) is expressed as x(t) = e-2t. So,

We evaluate the integral to get

 Fig: Magnitude and phase response of X[k]

Calculation of FS coefficients by inspection:

Example: Determine the FS representation of the signal

x(t) = 3 cos (πt/2 + π/4)

Solution: Time period T = 4, So, ω0 = 2π/4 = π/2. We may write FS of a signal x(t) is,

Using Euler‟s formula to expand the cosine, gives

Equating each term in this expansion to the terms in equation of x(t) gives the FS
coefficient:

The magnitude and phase spectra are shown below.

 Figure: Magnitude and phase spectrum

Example: Find the time domain signal x(t) corresponding to the FS coefficient

Assume that fundamental period T = 2.

Solution: Substituting the values given for X[k] and ω0 = 2π/2 = π into equation x(t) gives

The second geometric series is evaluated by summing from l = 0 to l = ∞, and subtracting

the l = 0 term. The result of summing both infinite geometric series is

Putting the fractions over a common denominator results in

Discrete-time periodic signals: The discrete-time Fourier series

Discrete-time periodic signals are represented by the discrete-time Fourier series (DTFS).
We may write the DTFS of a signal x[n] with fundamental period N and fundamental
frequency Ω0 = 2π/N, as

Where

Are the DTFS coefficient of the signal x[n].We say that x[n] and X[k] are a DTFS pair and
denote this relationship as
The DTFS coefficients are known as the frequency-domain representation of x[n], because
each DTFS coefficient is associated with a complex sinusoid of different frequency.

Direct calculation of DTFS coefficients:

Example: Determine the DTFS coefficient for signal x[n] shown

Solution: The signal has a period N = 5, so Ω0 = 2π/5. As the signal also has odd symmetry,
it can be sum over n =-2 to n =2 in the equation and we get,

Using the values of x[n], we get

From the above equation, we identify one period of DTFS coefficients X[k], k=2 to k=-2, in
rectangular and polar coordinates as

Fig: Magnitude and Phase Response of X[k]

The above figure shows the magnitude and phase of X[k] as functions of the frequency
index k.
Now suppose we calculate X[k] using n = 0 to n =4 for the limits on the sum in eqn.of X[k]
to obtain

Calculation of DTFS coefficients by inspection:

Example: Determine the DTFS coefficients of the signal

x[n] = cos (πn/3 + υ)

Solution: Time period N=6.We expand the cosine by using Euler‟s formula as

Now comparing with the DTFS equation with Ω0= 2π/6=π/3, written by summing from
k= -2 to 3

Equating the terms, we get

The magnitude and phase spectrum is given below

The Inverse DTFS:

Example: Determine the time signal x[n] from the DTFS coefficients given in figure below

Solution: The DTFS coefficients have period = 9, hence Ω0=2π/9. It is convenient to

evaluate x[n] over the interval k = -4 to k = 4 to obtain

Continuous-time nonperiodic signals: The Fourier transform

The Fourier transform is used to represent a continuous-time nonperiodic signal as a

superposition of complex sinusoids. We know that the continuous nonperiodic nature of a
time signal implies that the superposition of complex sinusoids used in the Fourier
representation of the signal involves a continuum of frequencies ranging from -∞ to ∞. So
the FT representation of a continuous-time signal involves an intyegral over the entire
frequency interval; i.e.,

Where,

Example1. FT of a real decaying exponential:

Find the FT of x(t) = e-atu(t) shown in the figure below.

Solution: The FT does not converge for a ≤ 0, since x(t) is not absolutely integrable, i.e.;

∞
0
𝑒 −𝑎𝑡 𝑑𝑡 = ∞ , a ≤ 0

For a > 0, we have

Converting to polar form, the magnitude and phase of X(jω) are respectively given by

and

as shown in figure below.

The magnitude of X(jω) plotted against ω is known as the magnitude spectrum of the
signal x(t) , and the phase of X(jω) plotted as a function of ω is known as the phase
spectrum of x(t).

Example 2: FT of a rectangular pulse:

Consider the rectangular pulse shown figure below and defined as

Find the FT of x(t).

Solution: The rectangular pulse x(t) is absolutely integrable, provided that T0 < ∞. So we
have

For ω = 0, the integral simplifies to 2T0. L‟Hospital‟s rule straightforwardly shows that
2
lim sin 𝜔𝑇0 = 2𝑇0
𝜔 →0 𝜔

Thus, we usually write

2
X(jω) = 𝜔 sin (ω T0)

With the understanding that the value at ω = 0 is obtained by evaluating a limit. In this case
X(jω) is real and is shown in the figure below.
The magnitude spectrum is

and the phase spectrum is

Using sinc function notation, we may write X(jω) as

Inverse FT of a rectangular spectrum:

Example: Find the inverse FT of the rectangular spectrum (figure below) given by

Solution: Using equation for x(t) for inverse FT yields

When t = 0, the integral simplifies to W/π. As

We usually write

or

The value at t = 0 is obtained as a limit. The x(t) is shown in the following diagram.

Discrete-time nonperiodic signals: The Discrete-time Fourier transform

The DTFT is used to represent a discrete-time nonperiodic signal as a super position of

complex sinusoids. As reasoned previously, the DTFT would involve a continuum of
frequencies on the interval – π < Ω < π, where Ω has unit of radians. Thus, the DTFT
representation of a time-domain signal involves an integral over frequency, namely,

Where

is the DTFT of the signal x[n]. As X(ej Ω) and x[n] are a DTFT pair, we can write

.
The transform X(ej Ω) describes the signal x[n] as a function of a sinusoidal frequency Ω
and is called the frequency-domain representation of x[n]. The equation for x[n] is usually
called the inverse DTFT, as it maps the frequency-domain representation back into the
time-domain.
If
∞

𝑥[𝑛] < ∞
𝑛=−∞
i.e.; if x[n] is absolutely summable, then the sum in eqn. X(ej Ω) converges uniformly to a
continuous function of ω.

Example: Find the DTFT of the sequence x[n] = αnu[n].

Solution: Using the above equation, we have

This sum diverges for |α| ≥ 1. For |α| ≤ 1, we have the convergent geometric series

If α is real valued, the denominator of the above equation may be expanded. Using Euler‟s
formula, we get

From this form, we see that the magnitude and phase spectra are given by

and

, respectively.

The magnitude and phase spectra for α = 0.5 and α = 0.9 are shown in the figure below.
The magnitude is given and the phase is odd and both are 2π periodic.

Inverse DTFT

Example: Find the Inverse DTFT of the following rectangular spectrum

Solution: Substituting X(ej Ω) in DTFT representation, we get

For n = 0, the integrand is unity and we have x[0]=W/π. Using L‟Hopital‟s rule, we can
easily show that

And thus we usually write

Fig: Inverse DTFT in Time domain

Properties of Fourier representation:

1. Linearity:
All four Fourier representations satisfy the linearity property.
 In these relationships, we assume that the upper case symbol denotes the Fourier
representation of the corresponding lower case symbol.

2. Symmetry:

3. Convolution:

a) Convolution of nonperiodic signal: Convolution of two signals h(t) & x(t) in the
time domain corresponds to multiplication of their FT, H(jω) & X(jω) in frequency
domain.

A similar property holds for convolution of discrete time non-periodic signal .If
DTFT DTFT
x[n] X(ejΩ) and h[n] H(ejΩ),then
DTFT
y[n] = x[n]∗ ℎ[𝑛] Y (ejΩ) = X (ejΩ) H(ejΩ)

b) Convolution of periodic signal: The periodic convolution of two periodic CT

signals x(t) and z(t), each having period T , as

Substituting the FS representation of z(t) into the convolution integral leads to the
property

Similarly in DTFS

4) Differentiation and Integration:

 a) Differentiation in time: Differentiating a signal in time domain corresponds to
multiplying its FT by jω in the frequency domain i.e;

b) Differentiation in frequency: Differentiation of FT in frequency domain

corresponds to multiplication of the signal by –jt in the time domain i.e;

Commonly used Differentiation and Integration properties:

5) Time Shift:
Let z(t) = x(t-t0) be a time –shifted version of x(t). The goal is to relate the FT of z(t)
to the FT of x(t)

Putting τ = t – t0, we obtain

Time-shift properties of Fourier representation:

 Frequency-shift properties of Fourier representation:

Multiplication property:

For nonperiodic signal:

For periodic signal:

Scaling properties:
Let us consider the effect of scaling the time variable on the frequency-domain
representation of a signal. Beginning with the FT, let z(t) = x(at), where a is a constant. By
definition, we have

We effect the change of variable τ = at to get

These can be combined into a single integral

We can conclude that

Scaling the signal in time introduces the inverse scaling in the frequency-domain
representation and an amplitude change, as shown in the given figure:
Parseval Relationships:

It states that the energy or power in the time-domain representation of a signal is equal
to the energy or power in the frequency-domain representation. So the energy and
power are conserved in the Fourier representation.
The energy in a continuous-time non-periodic signal is
∞
𝑊𝑥 = |𝑥 𝑡 |2 𝑑𝑡
−∞
Where it is assumed that x(t) may be complex valued general. As |x(t)|2 = x(t) x‫(٭‬t),
taking the conjugate of both sides of Eq, we may express x‫(٭‬t) in terms of its FT X(jω)
as

Substituting this formula into the expression for Wx, we obtain

Now we interchange the order of integration:

Observing that the integral inside the braces is the FT of x(t), we obtain

And so conclude that

Hence, the energy in the time-domain representation of the signal is equal to the energy
in the frequency-domain representation, normalized by 2π. The quantity |X(jω)|2 plotted
against ω is known as energy spectrum of the signal.
Analogous results hold for the periodic signal is known as power density spectrum of
the signal.
The other three representations are summarized in the table below:

Duality:
In this chapter, we observed a consistent symmetry between the time and frequency
domain representations of signals. For example, a rectangular pulse in either time or
frequency corresponds to a sinc function in either frequency or time, as shown in the
figure below.

We have also observed symmetries in Fourier representation properties i.e.; convolution

in one domain corresponds to modulation in other domain; differentiation in one domain
corresponds to multiplication by the independent variable in the other domain, and so
on. So by the consequence of this symmetry, we may interchange time and frequency.
This interchangeability property is termed duality.
Duality properties of Fourier representation is summarized in the table below:
Applications of Fourier representations to mixed signal classes

 We now discuss the application of Fourier representations to mixed signals like

- Periodic & nonperiodic signal
- Continuous & discrete time signal
 Such mixing of signals occurs most commonly when one uses Fourier methods to
- Analyze the interaction between signals & systems
- Numerically evaluate properties of signal or the behavior of a system
 For example: If we apply a periodic signal to a stable LTI system, the convolution
representation of the system output involves a mixing of nonperiodic (impulse
response) and periodic (input) signals.
 Another example: A system that samples continuous time-signal involves both
continuous & discrete-time signals.
 In order to use Fourier methods to analyze such interactions, we must build bridges
between the Fourier representations of different classes of signals.
 DTFS is the only Fourier representation that can be evaluated numerically on a
computer.

Fourier transform representation of periodic signals:

Neither FT nor DTFT converges for periodic signals. However, by incorporating impulses
into the FT & DTFT in the appropriate manner, we may develop FT & DTFT
representations of such signals.

Relating FT to FS:

The FS representation of a periodic signal x(t) is

∞

𝑥 𝑡 = 𝑋[𝑘]𝑒 𝑗𝑘𝑡 𝜔 0
𝑘=−∞

Where, ω0 is the fundamental frequency of the signal.

As, , using frequency shift property, the inverse FT of a frequency shifted
impulse δ (ω-kω0) is a complex sinusoid with frequency kω0. i.e,;

If we find FT of 𝑥 𝑡 , then by using linearity property of FT we obtain

Thus the FT of a periodic signal is a series of impulses spaced by the fundamental frequency
ω0. The kth impulse has strength 2πX(k) , where X(k) is the kth FS coefficient.
The shape of X(jω) is identical to that of X(k). The FT obtained from the FS by placing
impulses at integer multiples of ω0 and weighing them by 2π times the corresponding FS
coefficient.
Given an FT consisting of impulses that are uniformly spaced in ω, we obtain the
corresponding FS coefficients by dividing the impulse strengths by 2π.
 Fig: FS and FT representation of periodic continuous-time signals

Relating DTFT to DTFS:

The DTFS expression for a N periodic signal x[n] is

The inverse DTFT of a frequency shifted impulse is a discrete-time complex sinusoid. The
DTFT is a 2π periodic function of frequency. So we may express a frequency shifted
impulse either by expressing one period such as

Or, by using an infinite series of shifted impulses, separated by an interval of 2π, to obtain
the 2π periodic function

The inverse DTFT equation is evaluated by means of the shifting property of the impulse
function. We have

Hence we identify the complex sinusoid and the frequency shifted impulse as a DTFT pair
using linearity property and substituting the above equation in equation of x[n], we get
DTFT of periodic signal x[n]

Since DTFT is 2π periodic, it follows that, DTFT of x[n] consists of a set of N impulses of
strength 2πX(k),k=0,1,2…….,N-1.

Given the DTFS coefficient and the fundamental frequency Ω0, we obtain the DTFT
representation by placing impulses at integer multiples of Ω0 and weighting them by 2π
times the corresponding DTFS coefficient.
 Fig: DTFS & DTFT representation of a periodic discrete-time signal

Fourier transform representation of Discrete-time signals:

In this section we derive, an FT representation of discrete-time signal by incorporating

impulses into the description of the signal the appropriate manner, establishing a
correspondence between the continuous frequency ω and the discrete-time frequency Ω.
Let us define the complex sinusoids
x(t) = ejωt and g[n] = ejΩn
Suppose a force g[n] equal to the samples of x(t) taken at intervals of Ts i.e.;
g[n] = x(nTs)
=> ejΩn = ejωTsn

From which we conclude that Ω = ωTs.

Relating FT to DTFT:

DTFT of an arbitrary discrete-time signal x[n] is

We want to find out an FT pair that corresponds to the DTFT pair

. Substituting Ω = ωTs, we obtain the following function of continuous
time frequency ω.

Taking the inverse FT of Xδ(jω), using linearity and the FT pair

yields the continuous time signal description

Hence,
The discrete-signal has values x[n], while the corresponding continuous time signal consists
of a series of impulses separated by Ts, with nth impulse having strength x[n]. The DTFT
X(ejΩ) is 2π periodic in Ω, while the FT Xδ(jω) is 2π / Ts periodic in ω.

Fig: Relationship between FT and DTFT representation of a discrete-time signal

Relating FT to DTFS:

The DTFT representation of an N periodic signal x[n] is given as

Where X[k] is DTFS coefficient.

Substituting Ω = ωTs into this eqn. yields the FT representation

Using scaling property of impulse

We can write

DTFS coefficients X[k] are N-periodic function, which implies that Xδ(jω) is periodic with
period NΩ0 /Ts = 2π / Ts.
Continuous time representation of discrete-time signal as derived in previous chapter

As x[n] is N periodic, so xδ(t) is also periodic with fundamental period NTs

 Fig: Relationship between FT and DTFS representation of a discrete-time signal

Sampling:

- The sampling operation generates a discrete-time signal from the continuous-time

signal in order to manipulate the signal on a computer or microprocessor.
- Such manipulations are common in communication, control and signal processing
systems.
- Sampling is also frequently performed on discrete-time signal to change the effective
data rate, an operation termed subsampling.

Sampling continuous-time signals:

- Let x(t) be a continuous-time signal. To define a discrete-time signal x[n] which is

equal to the samples of x(t) at integer multiples of a sampling interval Ts, i.e.;
x[n] = x[nTs]
- The impact of sampling is elevated by relating the DTFT of x[n] to the FT of x(t).
- The continuous-time representation of discrete-time signal x[n] is given by

- Substituting x[nTs] for x[n] in above eqn. we get

Since

So we may write

Where

- The above eqn. implies that we may mathematically represent the sample signal as
the product of original continuous-time signal and impulse train.
- This representation is commonly termed as impulse sampling and is a mathematical
tool used only to analyze sampling.
- The effect of sampling is determined by relating FT of xδ(t).to FT of x(t).
- As multiplication in the time domain corresponds to the convolution in the
frequency domain, so by multiplication property:

- As impulse train is continuous periodic function, so its FS is given as

𝑇𝑠/2
1 1
𝑃[𝑘] = 𝛿(𝑡)𝑒 −𝑗𝑘 𝜔 0 𝑡 𝑑𝑡 =
𝑇𝑠 −𝑇𝑠/2 𝑇𝑠
By using FT representation of FS
∞

𝑃 𝑗𝜔 = 2𝜋 𝑃[𝑘]𝛿(𝜔 − 𝑘𝜔𝑠 )
𝑘=−∞
We get FT of impulse train as
∞
2𝜋
𝑃 𝑗𝜔 = 𝛿(𝜔 − 𝑘𝜔𝑠 )
𝑇𝑠
𝑘=−∞

Where ωs = 2π/Ts is the sampling frequency.

Now we convolve X(jω) with each of the frequency shifted impulse to get

- The FT of the sampled signal is given by an infinite sum of shifted versions of the
original signal‟s FT.
- The shifted versions are offset by integer multiples of ωs.
- The shifted versions of X(jω) may overlap with each other if ωs is not large enough
compared with the frequency extent or bandwidth of X(jω).
- Let the frequency component of the signal x(t) is assumed to lie within the frequency
band –W < ω < W for the purpose of illustration.
Fig: The FT of a sampled signal for different sampling frequencies (a) Spectrum of a continuous-
time signal, (b) Spectrum of sampled signal when ωs = 3W, (c) Spectrum of sampled signal when ωs
= 2W, (d) Spectrum of sampled signal when ωs = 1.5W

- Note that, as Ts increases and ωs decreases, the shifted replicas of X(jω) move close
together, finally overlapping one another when ωs < 2W.
- Overlap in the shifted replicas of the original spectrum is termed aliasing.
- Aliasing distorts the spectrum of the sampled signal.
- The spectrum of the sampled signal no longer has a one to one correspondence with
that of the original continuous-time signal.
- This means that we cannot use the spectrum of the sampled signal to analyze the
continuous-time signal and we cannot uniquely reconstruct the original continuous-
time signal from its samples.
- The DTFT of the sampled signal is obtained from Xδ(jω) by using the relationship
Ω = ωTs

- The scaling of the independent variable implies that ω= ωs corresponds to Ω = 2π.

- The FTs have period ωs, while DTFTs have period 2π.
Fig: The DTFTs corresponding to the FTs depicted in Fig:1((b)-(d)). (a) ωs = 3W, (b) ωs = 2W, (c) ωs = 1.5W

Reconstruction of continuous-time signals from samples:

- The problem of reconstructing a continuous-time signal from samples involving a

mixture of continuous & discrete-time signals
- A device that performs reconstruction has a discrete-time input signal and a
continuous-time output signal.

- The FT is well suited for analyzing this problem, since it may be used to represent
both continuous & discrete-time signals.
- We first consider the conditions that must be met in order to uniquely reconstruct a
continuous-time signal from its samples.
Sampling Theorem:
- The samples of a signal do not always uniquely determine the corresponding
continuous-time signal.
- For example, the figure below shows, two different continuous-time signals having
the same set of samples x[k] = x1(nTs) = x2(nTs)

Fig: Two continuous-time signals x1(t) (dashed line) and x2(t) (solid line) that have the same set of samples
- Note that the samples do not tell us anything about the behavior of the signal in
between the times it is sampled.
- In order to determine how the signal behaves in between those times, we must
specify additional constraints on the continuous-time signal.
- One such set of constraints, that is very useful in practice, involves requiring the
signal to make smooth transitions from one sample to another.
- The smoothness, or rate at which the time domain signal changes, is directly related
to the maximum frequency that is present in the signal.
- So, constraining smoothness in the time domain corresponds to limiting the
bandwidth of the signal.
- Because there is one to one correspondence between the time domain and frequency
domain representations of a signal, we may also consider the problem of
reconstructing the continuous-time signal in the frequency domain.
- To reconstruct a continuous-time signal uniquely from its samples, there must be a
unique correspondence between the FTs of the continuous-time signal and the
sampled signal.
- The FTs are uniquely related if the sampling process does not introduce aliasing.
- Aliasing distorts the spectrum of the original signal and destroys one-to-one
relationship between the FTs of the continuous-time signal and the sampled signal.
- Prevention of aliasing requires satisfying the sampling theorem.
- Let represent a band-limited signal so that X(jω) = 0 for |ω| > ωm
If ωs > 2 ωm, where ωs = 2π / Ts is the sampling frequency, then x(t) is uniquely
determined by its samples x(nTs), n=0,±1,±2,………
- The minimum sampling frequency, 2 ωm, is termed the Nyquist sampling rate or
Nyquist rate. The actual sampling frequency, ωs, is commonly referred to as the
Nyquist frequency.
- If
fm = ωm / 2π and fs > 2 fm
=>1/Ts > 2 fm
=> Ts <1/2 fm
Ideal Reconstruction:

- The sampling theorem indicates how fast we must sample a signal so that the
samples uniquely represent the continuous-time signal.
- If , then the FT representation of the sampled signal is given by:

- The goal of reconstruction is to apply some operation to Xδ(jω) that converts it back
to X(jω).
- Any such operation must eliminate the replicas, images of X(jω) that are centered at
kωs.
- This is accomplished by multiplying Xδ(jω) by

, as shown in the figure below

Fig: (a) Spectrum of original signal, (b) Spectrum of sampled signal (c) frequency response of
reconstruction filter.
- We then have X(jω) = Xδ(jω) Hr(jω)
- Note that, multiplication by Hr(jω) will not recover X(jω) from Xδ(jω) if the
conditions of the sampling theorem are not met and aliasing occur.
- Multiplication in the frequency domain transforms to convolution in the time
domain.
- Hence, x(t) = xδ(t) * hr(t)
Where, .
Substituting xδ(t) in the above equation, we get

Next we use

𝜔 𝜔 𝑠𝑡
= Ts (2𝜋𝑠 𝑠𝑖𝑛𝑐 )
2𝜋
𝜔 𝑠𝑡
= 𝑠𝑖𝑛𝑐 2𝜋
 So,

- In the time domain, we construct x(t) as a weighted sum of sinc functions shifted by
the sampling interval. The weights correspond to the values of the discrete-tine
sequence.
- The value of the x(t) at t = nTs is given by x[n], because all of the shifted sinc
functions are zero at nTs, except the nth one and its value is unity.
- The operation described by the above equation is referred to as ideal band limited
interpolation, since it indicates how to interpolate in between the samples of a band-
limited signal.

Fig: Ideal reconstruction in the time domain

Fourier series representation of finite duration on periodic signal:

- As DTFS is the only Fourier representation that can be evaluated numerically, so we
apply DTFS and FS to signals that are not periodic to facilitate numerical computation
of Fourier representations.
- Another advantage of this representation is understanding of relationship between the
FT and corresponding FS representation.
Relating the DTFS to DTFT:

- Let x[n] be a finite duration aperiodic signal of length M i.e.;

x[n] = 0 for n < 0 and n ≥ M
- DTFT of this signal is

- Let 𝑥[n] be a periodic discrete-time signal with period N ≥ M such that one period of
𝑥[n] is given by x[n] .
- The DTFS coefficients of 𝑥 [n] are given by

(as 𝑥 [n] = x[n] within one period)

Where Ω0 = 2π / N

(as x[n] = 0 for n ≥ M)

A comparison of 𝑋[𝐾] and X(ejω) reveals that

- The DTFS coefficient of 𝑥[n] are samples of the DTFT of x[n], divided by N and
evaluated by at intervals of 2π / N.
- Although x[n] is not periodic, we define DTFS coefficients using x[n], n = 0,1,….N-1
according to

So X[K] = 𝑋[𝐾] = (1/N) X (ejkω0) (from above two equations)

- DTFS coefficients of x[n] correspond to the DTFS coefficients of periodically extended
signal 𝑥[n].
- The effect of sampling the DTFT of a finite-duration nonperiodic is to periodically
extend the signal in the time domain. i.e.;
 Fig: The DTFS of a finite duration nonperiodic signal

- The above relationship is dual to sampling frequency.

- Sampling a signal in time generates shifted replicas of the original signal in the
frequency domain.
Dual:
- Sampling a signal in frequency generates shifted replicas of the original signal in the
time domain.
- In order to prevent overlap or aliasing, of those shifted replicas in time, we require the
frequency sampling interval Ω0 to be less than or equal to 2π/ M (Ω0 ≤ 2π/ M => N ≥
M)
Relating the FS to the FT:
Let x(t), an aperiodic signal have finite duration T0,
i.e.; x(t) = 0, t < 0 & t ≥ T0
Construct a periodic signal with period t

With T ≥ T0 by periodically extending x(t) , the FS coefficients of 𝑥 (𝑡) are

Where we have used the relationship:
𝑥 𝑡 = 𝑥(𝑡) for 0 ≤ t ≤ T0 and 𝑥 𝑡 = 0 for T0 < t < T
The FT of x(t) is defined by

(as x(t) is finite duration)

Hence, comparing 𝑋 𝑘 with X(jω)

The FS coefficients are the samples the FT, normalized by T.

Modulation:
Modulation is basic to the operation of a communication system. Modulation provides the
means for

1. Shifting the range of frequencies contained in the message signal into another frequency
range suitable for transmission over the channel.

2. Performing a corresponding shift back to the original frequency range after reception of
the signal.

Formally modulation is defined as the process by which some characteristic of a carrier

wave is verified in accordance with the message signal. The message signal is referred to as
the modulating wave, and the result of the modulation process is referred to as the
modulated wave. In the receiver, demodulation is used to recover the message signal from
the modulated wave. Demodulation is the inverse of modulation process.

Types of modulation:

The specific type of modulation used in a communication system is determined by the form
of carrier wave used to perform the modulation. The two most commonly used forms of
carrier are a sinusoidal wave and a periodic pulse train. Correspondingly, there are two
classes of modulation: Continuous-wave (CW) modulation and pulse modulation.

Continuous-wave modulation:

Consider the sinusoidal carrier wave

c(t) = Ac cos(φ(t))

which is uniquely defined by the carrier amplitude Ac and angle φ(t). Depending on the type
of parameter chosen for modulation, two subclasses of CW modulation is identified. i.e.,
  Amplitude modulation, in which the carrier amplitude is varied with the message
signal.
 Angle modulation, in which the angle of carrier is varied with the message signal.

Fig: Amplitude and angle modulated signal

Pulse modulation:

Consider a carrier wave

∞

𝑐 𝑡 = 𝑝(𝑡 − 𝑛𝑇)
𝑛=−∞

that consists of a periodic train of narrow pulses, where „T‟ is the period and p(t) denotes a
pulse of relatively short duration. When some characteristic parameter of p(t) is varied in
accordance with the message signal, we get pulse modulation. Depending on how pulse
modulation is actually accomplished, the two subclasses are
 Analog pulse modulation, in which a characteristic parameter such as amplitude,
duration or position of a pulse is varied continuously with the message signal.
 Digital pulse modulation, in which the modulated signal is represented in coded
form known as pulse code modulation.
 Fig: Pulse amplitude modulation waveform

Benefits of modulation

In communication system, four benefits which results from the use of modulation are:

1. Modulation is used to shift the spectral content of a message signal so that it lies inside
the operating frequency band of a communication channel. It is useful for long distance and
high speed transmission. E.g., The telephonic communication over a cellular radio channel,
where 300-3100 Hz frequency are shifted to 800-900 MHz frequency, which is assigned to
cellular radio system in North America.

2. Modulation provides a mechanism for putting the information content of a message signal
into a form that may be less vulnerable to noise or interference.

3. It permits the use of multiplexing. Multiplexing permits the simultaneous transmission of

information bearing signals from a number of independent sources over the channel and on
to their respective destinations which makes communication channels cost-effective.

4. Modulation makes it possible for the physical size of the transmitting or receiving
antenna to assume a practical value. Electromagnetic theory says that the physical aperture
of an antenna is directly comparable to the wavelength of the radiated or incident
electromagnetic signal. Alternatively, since wavelength and frequency are inversely related
we may say that the aperture of the antenna is inversely proportional to the operating
frequency. Modulation elevates the spectral content of the modulating signal by an amount
equal to the carrier frequency. Hence, the larger the carrier frequency, the smaller will be
the physical aperture of the transmitting as well as the receiving antenna.
Full amplitude modulation:
Let us consider a sinusoidal carrier wave
c(t) = Accos(ωct).
For convenience of presentation, we have assumed that the phase of the carrier wave is zero
in above equation (as the primary emphasis here is on variations imposed on the carrier
amplitude). Let m(t) represent a message signal of interest. Amplitude modulation (AM) is
defined as a process in which the amplitude of the carrier is varied in proportion to a
message signal m(t).
s(t) = Ac[1+kam(t)] cos(ωct). ………………..(1)
where ka is a constant called the amplitude sensitivity factor of the modulator. The
modulated wave s(t) o defined is said to be a „full‟ AM wave. Here, the radian frequency ωc
of the carrier is maintained constant.
Percentage of modulation: Is called the envelope of the AM wave s(t). Using a(t) to denote
this envelope, the equation may be written as
a(t) = Ac|1+kam(t)|…………………….(2)
Two conditions arise, depending on the magnitude of kam(t), compared with unity:
1. Undermodulation, governed by the the condition
|kam(t)| ≤ 1, for all t.
Under this condition, the term 1+kam(t) is always nonnegative. Therefore, the expression for
the envelope of the AM wave may be simplified as
a(t) = Ac[1+kam(t)], for all t.
2. Overmodulation, governed by the weaker condition
|kam(t)| > 1, for some t.
Under this condition, the equation (2) is used in evaluating the envelope of the AM wave.
% modulation = The maximum absolute value of kam(t) × 100.
Accordingly, the first condition corresponds to a percentage modulation ≤ 100%, whereas
the second condition corresponds to a percentage modulation > 100%.
Generation of AM wave:
Various schemes have been devised for the generation of an AM wave. Let us consider a
simple circuit that follows from the defining equation (1). This equation can be rewritten as
:
s(t) = ka [m(t)+B] Ac cos(ωct).
The constant B, equal to 1/ ka, represents a bias that is added to the message signal m(t)
before modulation. The above equation suggests the scheme described in the block diagram
given below for generating an AM wave. Basically it consists of two functional blocks:
 An adder that adds the bias B to the incoming message signal m(t)
 A multiplier that multiplies the adder output [m(t) + B] by the carrier wave Ac
cos(ωct), producing the AM wave s(t).
The percentage modulation is controlled by adjusting the bias B.

Message Multiplier AM Wave

Adder
signal m(t) s(t)

Bias B Carrier
Ac cos(ωct)

Fig: System for generating an AM wave

s(t)

Fig: Waveform of Amplitude modulation for a varying percentage of modulation

Frequency domain description of amplitude modulation:
To develop the frequency description of AM wave s(t), we take the Fourier transform of
both sides of equation (1). Let S(jω) denote Fourier transform of s(t) and M(jω) denote
Fourier transform of m(t).
 The Fourier transform of Ac cos(ωct) is π Ac [δ (ω-ωc) + δ (ω+ωc)]
1
 The Fourier transform of m(t) cos(ωct) is [M (jω-jωc) + M (jω+jωc)]
2

Using these results and invoking the linearity and scaling properties of the Fourier
transform, the Fourier transform of the AM wave is expressed as
1
S(jω) = π Ac [δ (ω-ωc) + δ (ω+ωc)] + 2 ka Ac[M (jω-jωc) + M (jω+jωc)] ……(3)

Let the message signal m(t) be band limited to the interval - ωm ≤ ω ≤ ωm as shown in the
figure below.

(a) (b)

Fig:

We refer to the highest frequency component ωm of m(t) as the message bandwidth,

measured in rad/s. We find from the eqn (3) that the spectrum S(jω) of the AM wave shown
in the figure (b) above for the case where ωc > ωm . This spectrum consists of two impulse
functions weighted by the factor π Ac and occurring at ± ωc, and two versions of the message
1
spectrum shifted in frequency by ± ωc and scaled in amplitude ka Ac. The spectrum in fig
2

(b) described as follows:

a) For positive frequencies, the portion of the spectrum of the modulated wave lying above
the carrier frequency ωc is called upper sideband, where as the symmetric portion below ωc
is called lower sideband. For negative frequencies, this condition is reversed. The condition
ωc > ωm is a necessary condition for the side bands not to overlap.
b) For positive frequencies, the highest frequency component of the AM wave is ωc+ ωm
and the lowest frequency component of the AM wave is ωc- ωm . The difference between
these two frequencies defines the transmission bandwidth ωT for an AM wave which is
exactly twice the message bandwidth ωm, i.e., ωT = 2 ωm. The spectrum of AM wave as
depicted in fig (b) is full, in that the carrier, the upper sideband and the lower sideband are
all completely represented. Hence, this form of modulation is called as full amplitude
modulation.

Demodulation of AM Wave:
Envelope detector is used for demodulation of AM wave, shown in the figure below, which
consists of a diode and a resistor-capacitor filter. The operation of this envelope detector is
as follows:
On the positive half-cycle of the input signal, the diode is forward biased and the capacitor
C charges up rapidly to the peak value of the input signal. When the input signal falls below
this value the diode becomes reverse biased and the capacitor C discharges slowly through
the load resistor Rl . The discharging process continues until the next positive half cycle.
When the input signal becomes greater than the voltage across the capacitor, the diode
conducts again and the process is repeated. Here it is assumed that the diode is an ideal
diode, the load resistance Rl is large compared with the source resistance Rs. During the
charging process, the time constant is effectively equal to RsC. This time constant must be
short compared with the carrier period 2π/ωc, i.e.,
RsC << 2π/ωc

Fig: Circuit diagram of Envelope detector showing its input & output.

Accordingly the capacitor C charges rapidly and thereby follows the applied voltage upto
the positive peak when the diode is conducting. In contrast, when the diode is reverse
biased, the discharging time constant is equal to RlC. This second time constant must be
long enough to ensure that the capacitor discharges slowly through the load resistor Rl
between positive peaks of the carrier wave, but not so long that the capacitor voltage will not
discharge at the maximum rate of change of modulating wave, i.e.,
2π/ωc << RlC << 2π/ωm.

Pulse Amplitude Modulation:

Pulse amplitude modulation (PAM) is a widely used form of pulse modulation. The basic
operation in PAM systems is the sampling that includes the derivation of sampling theorem
and related issues of aliasing and reconstructing the message signal from its sampled
version.

Message Low-pass Sample-

Sampled
signal m(t) antialiasing and-hold
signal
filter circuit

Timing pulse
generator

Fig: System for generating a flat-topped PAM signal

The sampling theorem in the context of PAM is in two equivalent parts as follows:
1) A band-limited signal of finite energy that has no radian frequency components higher
than ωm is uniquely determined by the values of the signal at instant of time separated
by π / ωm seconds.
2) A band-limited signal of finite energy that has no radian frequency components higher
than ωm may be completely recovered from knowledge of its samples taken at the rate
of ωm / π per seconds.
Part 1 of sampling theorem is exploited in the transmitter of a PAM system and part 2, in the
receiver of the system. The special value of the sampling rate ωm / π is referred to as the
Nyquist rate. To combat the effect of aliasing in practice, we use two corrective measures:
 Prior to sampling, a low pass anti-aliasing filter is used to attenuate high frequency
component of the signal which lie outside the band of interest.
 The filtered signal is sampled at a rate slightly higher than the Nyquist rate.
Mathematical description of PAM:
PAM is a form of pulse modulation, in which the amplitude of the pulsed carrier is varied in
accordance with instantaneous sample values of the message signal.

Fig: Wave form of flat topped PAM signal

For a mathematical representation of PAM signal s(t) for a message signal m(t), we may
write
∞

𝑠 𝑡 = 𝑚[𝑛]ℎ(𝑡 − 𝑛 𝑇𝑠 )
𝑛=−∞

Where, Ts  sampling period

m[n]  the value of message signal m(t) at time t=nTs
h(t)  a rectangular pulse of unit amplitude and duration T0
The impulse sampled version of the message signal m(t) is given by
∞

𝑚𝛿 𝑡 = 𝑚[𝑛]𝛿(𝑡 − 𝑛 𝑇𝑠 )
𝑛=−∞

The PAM signal is expressed as

∞

𝑠 𝑡 = 𝑚[𝑛]ℎ(𝑡 − 𝑛 𝑇𝑠 )
𝑛=−∞

= 𝑚𝛿 𝑡 ∗ ℎ(𝑡)

The above equation states that s(t) is mathematically equivalent to the convolution of 𝑚𝛿 𝑡
- the impulse sampled version of m(t) – and the pulse h(t).
Multiplexing:
Modulation provides a method for multiplexing whereby message signal derived from
independent sources are combined into a composite signal suitable for transmission over a
common channel.
In telephone system, the signals from different speakers are combined in such a way that
they do not interfere with each other during transmission and so that they can be separated at
the receiving end.
Multiplexing can be accomplished by separating different message signals either in
frequency, or time, or through the use of coding techniques. Thus, there are three basic types
of multiplexing, viz:
a) Frequency-division multiplexing: In this type of multiplexing, the signals are separated
by allocating them to different frequency bands. FDM favours use of CW modulation, where
each message signal is able to use the channel on a continuous-time basis.

Fig: (a) Frequency-division multiplexing (b) Time-division multiplexing

b) Time-division multiplexing: Here, the signals are separated by allocating them to

different time slots within a sampling interval . TDM favours the use of pulse modulation,
where each message signal has access to the complete frequency response of the channel.
c) Code-division multiplexing: It relies on the assignment of different codes to the
individual users of the channel.
a) Frequency-division multiplexing: The block diagram of FDM system is shown below.
The low pass filters are used for band limiting the input signals. The filtered signals are
applied to modulators that shift the frequency ranges of the signals so as to occupy mutually
exclusive frequency intervals. The band pass filters following the modulators are used to
restrict the band of each modulated wave to its prescribed range. Next, the resulting band
pass filters are summed to form the input to the common channel. At the receiving terminal,
a bank of band pass filters, with their inputs, connected in parallel, is used to separate the
message signal on a frequency occupancy basis. Finally, the original message signals are
recovered by individual demodulators.

Fig: Block diagram of FDM system

b) Time-division multiplexing: The basic operation of a TDM system is the sampling

theorem, which states that we can transmit all the information contained in a band limited
message signal by using samples of the signal taken uniformly at a rate that is usually higher
than the Nyquist rate. The important feature of the sampling process is conservation of time
i.e., the transmission of the message samples engages the transmission channel for only a
fraction of sampling interval on a periodic basis, equal to the width T0 of a PAM modulating
pulse. In this way, some of the time interval between adjacent samples is cleared for use by
other independent message sources on a time shared basis.

Fig: Block diagram of TDM system

The concept of TDM is illustrated by the above block diagram.

 Each input message signal is first restricted in band width by a low pass filter to remove
the frequency that is non essential to an adequate representation of the signal.
 LPF output applied to a commutator that is usually implemented by means of electronic
switching circuitry.
 The function of the commutator is two fold. 1) to take a narrow sample of each of the M
input message signal at a rate 1/Ts i.e., slightly higher than ωc / π, where ωc is the cut off
frequency of LPF. 2) to sequentially interleave these M samples inside a sampling
interval Ts.
 The multiplexed signal is applied to a pulse modulator that transforms it into a form
suitable for transmission over a common channel.
 At receiver, the signal is applied to a pulse demodulator which performs inverse
operation of pulse modulator.
 The narrow samples produced are distributed to the appropriate low pass reconstruction
filter by decommutator.
 Synchronization between timing operation of the transmitter and receiver in a TDM
system is essential for satisfactory performance of the system.
 Synchronization may be achieved by inserting an extra pulse into each sampling
interval on regular basis.
 The combination of M PAM signals and a synchronization pulse combined in a single
sampling period is referred to as a frame synchronization.
Phase and Group delay:
Whenever a signal is transmitted a through a frequency-selective system, such as
communication channel, some delay is introduced into the output signal in relation to the
input signal. The delay is determined by the phase response of the system.
Let the phase response of a dispersive communication channel is represented by:
υ (ω) = arg{H(jω)}, where H(jω) frequency response of the channel
Suppose that a sinusoidal signal is transmitted through the channel at a frequency ωc .
The signal received at the channel output lags the transmitted signal by υ(ωc) radians.
The time delay corresponding to this phase lag, which is known as phase delay (τp):
𝜑(𝜔 𝑐 )
𝜏𝑝 = − 𝜔𝑐
, where minus sign (-) denotes the lag.

Note: The phase delay is not necessarily the true signal delay.

Let us consider a transmitted signal

s(t) = A cos (ωct) cos (ω0t),

consisting of a DSB-Sc modulated wave with carrier frequency ωc and sinusoidal

modulation frequency ω0.

Expressing the modulated signal s(t) in terms of its upper and lower side frequencies, it may
be written as:
1 1
s(t) =2 A cos (ω1t) + 2 A cos (ω2t)

where, ω1 = ωc + ω0 and ω2 = ωc - ω0

If ω0 << ωc => side frequencies ω1 & ω2 are close together, with ωc between them. Such a
modulated signal is called narrowband signal. The phase response υ (ω) may be
approximated in the vicinity of ω= ωc by the two-term Taylor series expansion

The time delay incurred by the message signal (i.e., the envelope of the modulated signal) is
given by:

The time delay τg is called the group delay or envelope delay. The group delay is defined as
the negative of the derivative of the phase response υ (ω) of the channel with respect to ω ,
evaluated at the carrier frequency ωc.

Note: The time delay is a true signal delay.

For wide-band modulated signal, the frequency components of the message signal are
delayed by different amounts at the channel output. Consequently, the message signal
undergoes a form of linear distortion known as delay distortion. To reconstruct a faithful
version of the original message signal in the receiver, we have to use a delay equalizer. This
equalizer has to be designed in such a way that when it is connected in cascade with the
channel, the overall group delay is constant.

Introduction to the Laplace Transform

 Fourier transforms are extremely useful in the study of many problems of practical
importance involving signals and LTI systems.
 They are purely imaginary complex exponentials est, s=jω
 A large class of signals can be represented as a linear combination of complex
exponentials and complex exponentials are eigenfunctions of LTI systems.
 However, the eigenfunction property applies to any complex number s, not just purely
imaginary (signals).
 This leads to the development of the Laplace transform where s is an arbitrary complex
number.
 Laplace and z-transforms can be applied to the analysis of un-stable system (signals
with infinite energy) and play a role in the analysis of system stability
 The response of an LTI system with impulse response h(t) to a complex exponential
input, x(t)=est, is
y(t )  H (s)e st
where s is a complex number and

H ( s)   h(t )e  st dt

when s is purely imaginary, this is the Fourier transform, H(jω)
when s is complex, this is the Laplace transform of h(t), H(s)
The Laplace transform of a general signal x(t) is:

X ( s)   x(t )e  st dt

and is usually expressed as:
L
x(t )  X (s)
Laplace and Fourier Transform
The Fourier transform is the Laplace transform when s is purely imaginary:
X ( s) s  j  F x(t )
An alternative way of expressing this is when s = σ+jω

X (  j )   x(t )e (  j )t dt






x(t )e e
t  jt
dt

  x' (t )e  jt dt


 F{x' (t )}
The Laplace transform is the Fourier transform of the transformed signal x‟(t) = x(t)e-σt.
Depending on whether σ is positive/negative this represents a growing/negative signal
Example 1: Laplace Transform
 at
Consider the signal x(t )  e u(t )
The Fourier transform X(jω) converges for a>0:
  1
X ( j )   eatu(t )e jt dt   eat e jt dt  , a0
 0 j  a
The Laplace transform is:
 
X ( s)   e atu (t )e  st dt   e ( s  a )t dt
 0

  e (  a )t e  jt dt
0

which is the Fourier Transform of e-(σ+a)tu(t)

1
Or X (  j)  ,  a 0
(  a)  j
L 1
e atu(t )  X ( s)  , Re{s}  a
sa
If a is negative or zero, the Laplace Transform still exists
Example 2:
Consider the signal x(t )  eatu(t )
The Laplace transform is:

X ( s )    e  at e  stu (t )dt


   e ( s  a )t dt
0



1

sa
Convergence requires that Re{s+a}<0 or Re{s}<-a.
The Laplace transform expression is identical to Example 1 (similar but different signals),
however the regions of convergence of s are mutually exclusive (non-intersecting).
For a Laplace transform, we need both the expression and the Region Of Convergence
(ROC).
Example 3:
The Laplace transform of the signal x(t) = sin(ωt)u(t) is:

X (s)  

1
 2 j
e jt

 e  jt u (t )e  st dt
 
 1
2j 0  e ( s  j )t dt  21j  e ( s  j )t dt
0

 e ( s  j )t e ( s  j )t 
 

 1 
  ( s  j )
2j
( s  j ) 0 
 0 
 1 1 
 21j   
 ( s  j ) ( s  j ) 
 2 j 
 1
 2 2 
 s  
2j



s 2
2
Fourier Transform does not Converge …
It is worthwhile reflecting that the Fourier transform does not exist for a fairly wide class of
signals, such as the response of an unstable, first order system, the Fourier transform does
not exist/converge
E.g. x(t) = eatu(t), a>0
∞
𝑋 𝑗𝜔 = 𝑒 𝑎𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡
0
does not exist (is infinite) because the signal‟s energy is infinite
This is because we multiply x(t) by a complex sinusoidal signal which has unit magnitude
for all t and integrate for all time. Therefore, as the Dirichlet convergence conditions say,
the Fourier transform exists for most signals with finite energy.
Region of Convergence:
The Region Of Convergence (ROC) of the Laplace transform is the set of values for s
(=s+jω) for which the Fourier transform of x(t)e-σt converges (exists).
The ROC is generally displayed by drawing separating line/curve in the complex plane, as
illustrated below for Examples 1 and 2, respectively.

Re{s}  a Re{s}  a

The shaded regions denote the ROC for the Laplace transform
Example 4:
Consider a signal that is the sum of two real exponentials:
x(t )  3e2t u(t )  2et u(t )
The Laplace transform is then:

 
X (s)   3e 2t u (t )  2e t u (t ) e  st dt

 
 3 e 2t u (t )e  st dt  2 e t u(t )e  st dt
 

Using Example 1, each expression can be evaluated as:

3 2
X ( s)  
s  2 s 1
The ROC associated with these terms are Re{s}>-1 and Re{s}>-2. Therefore, both will
converge for Re{s}>-1, and the Laplace transform:
s 1
X ( s)  2
s 3s  2
Ratio of Polynomials:
In each of these examples, the Laplace transform is rational, i.e. it is a ratio of polynomials
in the complex variable s.
N ( s)
X ( s) 
D( s )
where N and D are the numerator and denominator polynomial respectively.

In fact, X(s) will be rational whenever x(t) is a linear combination of real or complex
exponentials. Rational transforms also arise when we consider LTI systems specified in
terms of linear, constant coefficient differential equations.
We can mark the roots of N and D in the s-plane along with the ROC

Example 3:

Poles and Zeros:

The roots of N(s) are known as the zeros. For these values of s, X(s) is zero.

The roots of D(s) are known as the poles. For these values of s, X(s) is infinite, the Region
of Convergence for the Laplace transform cannot contain any poles, because the
corresponding integral is infinite.

The set of poles and zeros completely characterise X(s) to within a scale factor (+ ROC for
Laplace transform)
 ( s  zi )
X ( s)  i
 j (s  p j )

The graphical representation of X(s) through its poles and zeros in the s-plane is referred to
as the pole-zero plot of X(s)

Example:
Consider the signal: 4 1
x(t )   (t )  e t u (t )  e 2t u (t )
3 3
By linearity we can evaluate the second and third terms
The Laplace transform of the impulse function is:

L{ (t )}    (t )e  st dt  1


The Laplace transform of the impulse function is:

4 1 1 1
X ( s)  1  
3 s 1 3 s  2
( s  1) 2
 , Re{s}  2
( s  1)( s  2)
which is valid for any s. Therefore,

ROC Properties for Laplace Transform:

Property 1: The ROC of X(s) consists of strips parallel to the jω-axis in the s-plane
Because the Laplace transform consists of s for which x(t)e-σt converges, which only
depends on Re{s} = σ
Property 2: For rational Laplace transforms, the ROC does not contain any poles
Because X(s) is infinite at a pole, the integral must not converge.
Property 3: if x(t) is finite duration and is absolutely integrable then the ROC is the entire s-
plane.
Because x(t) is magnitude bounded, multiplication by any exponential over a finite interval
is also bounded. Therefore the Laplace integral converges for any s.

Inverse Laplace Transform:

The Laplace transform of a signal x(t) is:


X (  j )  F{x(t )e t }   x(t )e t e  jt dt

We can invert this relationship using the inverse Fourier transform
1 
x(t )e t  F 1{ X (  j )} 
2  
X (  j )e jt d
Multiplying both sides by eσt:
1 
x(t ) 
2 

X (  j )e(  j )t d
Therefore, we can recover x(t) from X(s), where the real component is fixed and we
integrate over the imaginary part, noting that ds = jdω
1   j
x(t ) 
2j   j
X (s)e st ds

Inverse Laplace Transform Interpretation:

Just about all real-valued signals, x(t), can be represented as a weighted, X(s), integral of
complex exponentials, est.
1   j
x(t ) 
2j   j
X (s)e st ds
The contour of integration is a straight line (in the complex plane) from σ-j to σ+j (we
won‟t be explicitly evaluating this, just spotting known transformations). We can choose
any s for this integration line, as long as the integral converges
For the class of rational Laplace transforms, we can express X(s) as partial fractions to
determine the inverse Fourier transform.

M
Ai e  ai t u (t ) Re{s}  ai
A
X (s)   i
i 1 s  ai L1{ Ai /( s  ai )}
 Ai e  ait u(t ) Re{s}  ai

Example 1: Inverting the Laplace Transform

Consider when
1
X ( s)  (s)  1
(s  1)( s  2)
Like the inverse Fourier transform, expand as partial fractions
1 A B 1 1
X ( s)     
(s  1)( s  2) (s  1) (s  2) (s  1) ( s  2)
Pole-zero plots and ROC for combined & individual terms
L 1
e t u (t )  , Re{s}  1
s 1
L 1
e 2t u (t )  , Re{s}  2
s2
L 1
x(t )  (e t  e 2t )u (t )  , Re{s}  1
( s  1)( s  2)
Example 2: Consider when
1
X ( s)  Re{s}  2
(s  1)( s  2)
Like the inverse Fourier transform, expand as partial fractions
1 A B 1 1
X ( s)     
(s  1)( s  2) (s  1) (s  2) (s  1) ( s  2)
Pole-zero plots and ROC for combined & individual terms
L 1
 e t u (t )  , Re{s}  1
s 1
L 1
 e 2t u (t )  , Re{s}  2
s2
L 1
x(t )  (e t  e 2t )u (t )  , Re{s}  2
( s  1)( s  2)

Laplace Transform Properties:

a) Linearity property:
L

If x1 (t )  X1 (s) ROC = R1
L
and x2 (t )  X 2 (s) ROC = R2
L
then ax1 (t )  bx2 (t )  aX1 (s)  bX 2 (s) ROC= R1R2

This follows directly from the definition of the Laplace transform (as the integral operator is
linear). It is easily extended to a linear combination of an arbitrary number of signals.

b) Time Shifting property:

If L ROC=R
x(t )  XL(s)
then x(t  t0 )  e st0 X ( s) ROC=R

Proof:
  j
x(t )  1

2j   j X (s)e st ds
Now replacing t by t-t0
  j
x(t  t0 )  1

2j   j X ( s)e s (t t0 ) ds
  j
 1

2j   j (e  st0 X ( s))e st ds

Recognising this as L{x(t  t0 )}  e  st0 X (s)

A signal which is shifted in time may have both the magnitude and the phase of the Laplace
transform altered.
Example: Linear and Time Shift
Consider the signal (linear sum of two time shifted sinusoids)
x(t )  2 x1 (t  2.5)  0.5 x1 (t  4)
where x1(t) = sin(ω0t)u(t).

Using the sin() Laplace transform example

0
X 1 ( s)  Re{s}  0
s  02
2

Then using the linearity and time shift Laplace transform properties


X ( s)  2e 2.5 s  0.5e 4 s  s  
2
0
2
Re{s}  0
0

c) Convolution property:
The Laplace transform
L also has the multiplication property, i.e.
x(t )  X (s) ROC = R1
L
h(t )  H (s) ROC = R2
L
x(t ) * h(t )  X (s) H (s) ROC  R1R2

roof is “identical” to the Fourier transform convolution

Note that pole-zero cancellation may occur between H(s) and X(s) which extends the ROC
s 1
X ( s)  {s}  2
s2
s2
H ( s)  {s}  1
s 1
X ( s) H ( s)  1    {s}  

Example 1: First order input & First order system impulse response
Consider the Laplace transform of the output of a first order system when the input is an
exponential (decay?)
x(t )  e  atu (t ) (Solved with Fourier transforms when a,b>0)
h(t )  e btu (t )
Taking Laplace transforms
1
X ( s)  Re{s}  a
s 1a
H ( s)  , Re{s}  b
s b
Laplace transform of the output is
1 1
Y ( s)  Re{s}  max{ a,b}
sa sb
Splitting into partial fractions
 1  1 1 
Y ( s)      Re{s}  max{ a,b}
 b  a  s  a s  b 
and using the inverse Laplace transform
y(t )  b1 a e  atu(t )  e btu(t )
Note that this is the same as was obtained earlier, expect it is valid for all a & b, i.e. we can
use the Laplace transforms to solve ODEs of LTI systems, using the system‟s impulse
response
L
h(t )  H (s)
Example 2: Sinusoidal Input

Consider the 1st order (possible unstable) system response with input x(t)
h(t )  e atu(t )
x(t )  cos(0t )u(t )
Taking Laplace transforms
1
H ( s)  Re{s}  a
X ( s)  2s  sa
Re{s}  0
s  02
The Laplace transform of the output of the system is therefore
s 1
Y ( s)  2 Re{s}  max{ 0,a}
s  0 s  a
2

 1  as  02   a  1
  2 
2  2
  2 
2 
 a   0  s   2
0  a   0 sa

and the inverse Laplace transform is

y(t ) 
u (t )
a  0 22

a sin( 0t )  0 cos(0t )  aeat 
d) Differentiation in the Time Domain:
Consider the Laplace transform derivative in the time domain
L
x(t )  X (s) ROC = R
1   j
x(t )  
2j   j
X (s)e st ds
dx(t ) 1   j
2j   j
 sX (s)e st ds
dt
dx(t ) L
 sX ( s)
dt ROC R

sX(s) has an extra zero at 0, and may cancel out a corresponding pole of X(s), so ROC may
be larger

Widely used to solve when the system is described by LTI differential equations
Example: System Impulse Response

Consider trying to find the system response (potentially unstable) for a second order system
with an impulse input x(t)=δ(t), y(t)=h(t)
d 2 y(t ) dy(t )
a 2
b  cy(t )  x(t )
dt dt
Taking Laplace transforms of both sides and using the linearity property

 d 2 y(t )   dy(t ) 
aL 2 
 bL   cLy(t )  L{ (t )}
 dt   dt 
L{ y (t )}(as 2  bs  c)  1
1 1 k1 k2
L{ y (t )}  H ( s)  2   
as  bs  c a( s  r1 )( s  r2 ) ( s  r1 ) ( s  r2 )
where r1 and r2 are distinct roots, and calculating the inverse transform

y(t )  k1er1t u(t )  k2er2t u(t )

The general solution to a second order system can be expressed as the sum of two complex
(possibly real) exponentials.
 The z-Transform

It plays the same role in the analysis of discrete time signals & LTI systems as the
Laplace transform does in the analysis of continuous time signal and LTI systems. The
most important one is, in the z-domain, the convolution of two time domain signals is
equivalent to multiplication of their corresponding z transforms.

Definition:
The z-transform of a discrete-time signal x[n] is:

𝑋 𝑧 = 𝑥 𝑛 𝑧 −𝑛
𝑛=−∞

We denote the z-transform operation as

x[n] ←→ X (z).
In general, the number z in X(z) is a complex number. Therefore, we may write z as
z = rejw ,
where r is the radius of the circle. When r = 1, (7.1) becomes

∞
𝑗𝜔
𝑋 𝑒 = 𝑥 𝑛 𝑒 −𝑗𝜔𝑛
𝑛=−∞
which is the discrete-time Fourier transform of x[n]. Therefore, DTFT is a special
case of the z-transform. Hence, we can view DTFT as the z-transform is evaluated on
the unit circle. (figure below)

Figure 7.1: Complex z-plane. The z-transform reduces to DTFT for values of z on
the unit circle.
 When r ≠ 1, the z-transform is equivalent to
∞

𝑋 𝑟𝑒 𝑗𝜔 = 𝑥 𝑛 ( 𝑟 −𝑛 𝑒 −𝑗𝜔𝑛 )
𝑛=−∞

= [𝑥 𝑛 𝑟 −𝑛 ]𝑒 −𝑗𝜔𝑛
𝑛=−∞

−n
= ℱ [r x(n)],

which is the DTFT of the signal r-n x[n]. However, from the development of DTFT
we know that DTFT does not always exist. It exists only when the signal is square
summable, or satisfies the Dirichlet conditions. Therefore, X (z) does not always
converge. It converges only for some values of r. This range of r is called the region
of convergence.

ROC: The Region of Convergence (ROC) of the z-transform is the value of

z such that X (z) converges, i.e.,

∣ 𝑥 𝑛 ∣ 𝑟 −𝑛 < ∞
𝑛 =−∞

Example1: Consider the signal x[n] = an u[n], with 0 < a < 1. The z-transform of
x[n] is
∞

𝑋 𝑧 = 𝑎𝑛 𝑢[𝑛 ]𝑧 −𝑛
−∞
∞

= ( 𝑎𝑧 −1 )𝑛
𝑛=0

∞ −1 𝑛
Therefore, X(z) converges if 𝑛=0(𝑎 𝑧 ) <∞. From geometric series, we know that

∞ −1 𝑛 1
𝑛=0(𝑟 𝑧 ) = 1−𝑎𝑧 −1

when | az -l | < 1, or equivalently | z| > | a| . So,

1
X (z) = ,
1 - ax-l

with ROC being the set of z such that | z| > | a| .

 Figure 7.2: Pole-zero plot and ROC of Example 1.

Note: ROC of casual and infinite sequence is the exterior of a circle having radius | a| .
Example2: Consider the signal x[n] = -an u[-n -1] with 0 < a <1. The
z-transform of x[n] is
∞

𝑋 𝑧 =− 𝑎𝑛 𝑢[−𝑛 − 1]𝑧 −𝑛
𝑛=−∞

−1

=− 𝑎𝑛 𝑧 −𝑛
𝑛=−∞
∞

=− 𝑎−𝑛 𝑧 𝑛
𝑛 =−1
∞

=1− (𝑎 −1 𝑧)𝑛
𝑛=0

Therefore, X (z) converges when | a-l z| < 1, or equivalently | z| < | a| . In this case,

1 1
X (z) = 1 - = ,
1-a z
-l 1 - az -l
with ROC being the set of z such that | z| < | a| . Here, the z-transform is same as
that of Example 1, the only difference being ROC. Example 2 is just the left-sided
version of Example 1.

Figure 7.3: Pole-zero plot and ROC of Example 2.

Properties of ROC
(1) The ROC is a ring or disk in the z-plane centered at origin.

(2) DTFT of x[n] exists if and only if ROC includes the unit circle.
Proof: By definition, ROC is the set of z such that X (z) converges. DTFT is the z-
transform evaluated on the unit circle. Therefore, if ROC includes the unit circle, then
X (z) converges for any value of z on the unit circle i.e., DTFT converges.

(3) The ROC does not contain any pole.

(4) If x[n] is a right-sided sequence, then ROC extends outward from the outermost
pole.
(5) If x[n] is a left-sided sequence, then ROC extends inward from the innermost pole.
O
(6) If x (n) is a causal sequence, then the ROC is the entire z plane except at z = 0.

(7) If x (n) is a non-causal sequence, then the ROC is the entire z plane except at z = ∞.

(8) If x (n) is an infinite duration, two sided sequence; the ROC will consist of a ring in the
z plane, bounded on the interior or exterior by a pole, not containing any pole.
(9) The ROC of a LTI stable system contains the unit circle.
.
Properties of z-transform
 Linearity:
If x1(n)↔ X1(z) and x2(n) ↔ X2(z), then
axl[n] + bx2[n] ←→ aXl(z) + bX2(z)

 Time reversal:
If x(n) ↔ X(z), then
x(-n) ) ↔ X(z-1)

 Time shift:
If x(n)↔ X(z), then
x[n - nO ] ←→ X (z)z-n0, Where n0 is an integer.

 Multiplication by αn:
If x(n) ↔ X(z), then
[anx(n)] x(n)↔ X(z) X(a-1z)

 Convolution:
If x(n)↔ X(z), h(n)↔ H(z), then
[x(n) * h(n)] ↔ X(z) H(z)

 Differentiation in the z-domain:

If x(n)↔ X(z), then
𝑑
[nx(n)] )↔ -z 𝑑𝑧 X(z)
z-transform Pairs
7.2.1 A. Table 10.2
1. δ[n] ←→ 1, all z

2. δ [n - m] ←→ z -m , all z except 0 when m > 0 and ∞ when m < 0.

3. u[n] ←→ l
, |z| > 1
l-z -l

4. an u[n] ←→ l
, | z| > a
l-az -l

5. -an u[-n - 1] ←→ l
, | z| < | a| .
l-az -l

6. x∗ [n] ←→ X ∗ (z ∗ )

Inverse z-transform:

Three different methods are:

(1) Partial fraction method
(2) Power series method
(3) Long division method

Partial fraction method:

 In case of LTI systems, commonly encountered form of z-transform is
𝐵(𝑧)
𝑋 𝑧 =
𝐴(𝑧)

𝑏0 +𝑏1 𝑧 −1 + ⋯ + 𝑏𝑀 𝑧 −𝑀
𝑋 𝑧 =
𝑎0 +𝑎1 𝑧 −1 + ⋯ + 𝑎𝑁 𝑧 −𝑁
Usually M<N
 If M>N, then long division method is used & X(z) is expressed in the
form
𝑀−𝑁
𝐵(𝑧)
X z = fk 𝑧 −k +
𝐴(𝑧)
𝑘 =0

Where B(z) has now the order one less than the denominator
polynomial and partial fraction method is used to find z transform.
 The inverse z-transform of the terms in the summation are obtained from the
transform pair and time shift property

δ[n] ←→ 1
δ [n - m] ←→ z -m
 If X(z) is expressed as ratio of polynomials in z instead of z-1, then it
is converted to the polynomial of z-1.
 Convert the denominator into the product of first-order terms
 𝑏0 +𝑏1 𝑧 −1 + ⋯ + 𝑏𝑀 𝑧 −𝑀
𝑋 𝑧 =
𝑎0 𝑁 𝑘 =1 (1 − 𝑑𝑘 𝑧
−1 )

Where dk are the poles of X(z).

For distinct poles:

 For all distinct poles, the X(z) can be written as
𝑁
𝐴𝑘
𝑋 𝑧 =
(1 − 𝑑𝑘 𝑧 −1 )
𝑘 =1
 Depending on ROC, the inverse z-transform associated with each term is then is
determined by using the appropriate transform pair. We get

𝐴𝑘
Ak(dk)nu[n] ←→ With ROC z > dk
(1−𝑑 𝑘 𝑧 −1 )

𝐴𝑘
- Ak(dk)nu[-n-1] ←→ With ROC z < dk
(1−𝑑 𝑘 𝑧 −1 )

 For each term the relationship between the ROC associated with X(z) and each
pole determines whether the right sided or left sided inverse transform is selected.
For repeated poles:
For Repeated poles
 If pole di is repeated r times, then there are r terms in the partial fraction expansion
associated with that pole.

𝐴𝑖1 𝐴𝑖1 𝐴𝑖1

… . .
1 − 𝑑𝑖 𝑧 −1 , (1 − 𝑑𝑖 𝑧 −1 )2 (1 − 𝑑𝑖 𝑧 −1 )𝑟
 Here if X(z) with ROC | z| > d i , then the right sided inverse transform is chosen &
if X(z) with ROC |z| < d i , then the left sided inverse transform is chosen.

Example: Find the inverse z transform of

1 + 3𝑧 −1
𝑋 𝑧 = z >2
1 + 3𝑧 −1 +𝑎2 𝑧 −2

Solution: First we eliminate the negative power by multiplying numerator &

denominator by z2.

𝑧( 𝑧+3)
𝑋 𝑧 =
𝑧 2 +3𝑧+2

Dividing X(z) by z we get

𝑋 (𝑧) 𝑧+3
=
𝑧 𝑧+1 (𝑧+2)
The above equation can be written in partial fraction form

𝑋 𝑧 𝐶1 𝐶2
= +
𝑧 𝑧+1 𝑧+2
𝑋(𝑧)
Where, C1 = (z+1) 𝑧
|𝑧=−1 = 2
𝑋(𝑧)
Similarly, C1 = (z+2) 𝑧
|𝑧=−2 = -1
𝑋 𝑧 2 1
Therefore, = −
𝑧 𝑧+1 𝑧+2
𝑧 𝑧
X(z)= 2 𝑧+1
−
𝑧+2
As ROC is z > 2, the sequence is causal and we find
x(n)= 2 (-1)nu(n)-(-2)nu(n)

Power series method:

 Express X(z) as a power series in z-1 or z as given in z-transform equation.
 The values of the signal x[n] are then given by coefficient associated with z-n.
 Main disadvantage is : limited to one sided signals.
 Signals with ROCs of the form | z| > a or | z| < a.
 If ROC is |z| > a, then express X(z) as a power series in z-1 and we get
right sided signal.
 If ROC is |z| < a, then express X(z) as a power series in z and we get
left sided signal.

Long division method:

 Find the z-transform of

2+𝑧−1 1
𝑋𝑧 = 1 −1
, with ROC | z| > 2
1−2𝑧

 Solution: since ROC indicates that x[n] is right sided sequence, use long division method
to write X(z) as a power series in z-1.

 We get
1
X(z) = 2+2z-1+z-2+ 2 z-3+……

 Comparing with z-transform pair, we get

1
x(n)= 2δ(n)+ 2δ(n-1)+ δ(n-2)+ δ(n-3)+………
2
 1
 If we change the ROC to |z| < 2 , then expand X(z) as a power series in z
using long division method , we get

X(z) = - 2 - 8z - 16z2 - 32z-3+……

 Now we get x(n) as

x ( n ) = - 2 δ ( n ) - 8 δ ( n + 1 ) - 1 6 δ ( n + 2 ) + 32δ ( n + 3 ) + … … …