9/26/2024

Signal System and Digital Signal Processing (AExE08)

8.1. Signal and system

8.2 Linear time invariant system

Signal:
The physical quantity that contains the information. The signal may be
one dimensional or multidimensional
Example:
I(t) , V(t) , X(t1,t2,t3)

1
 9/26/2024

Signal Classification (Types of Signals)

1. Continuous time signal and discrete time signal
Continuous Time Signal
A continuous time signal may be defined as mathematical continuous
function.
A signal that can be defined at every instant of time.
There are infinite possible values for time ‘t’ and instantaneous
amplitude, x(t).

1. Continuous time signal and discrete time signal

Discrete Time Signal:

Discrete time signal is defined only at integer time instant.
Signal that can be defined at discrete instant of time.
The number of elements in the set as well as the possible values of
each element is finite and countable and represented with bits.
Represented by x[n].

2
 9/26/2024

2. Deterministic and Random signal

Deterministic signals:
• Signals which can be completely specified for any given time.
• Nature and amplitude at any time be predicted.
• Pattern of this type of signal is regular and be characterized
mathematically.
• Thus a deterministic signal can be modeled by a known function of
time.
• For eg. x(t)= mt, x(t)= Acos(wt)

2. Deterministic and Random signal

Random signals:
Also called non-deterministic signals.
Those signals that take random values at any given time.
Pattern of this type of signal is random in nature and can’t be described
mathematically.
For eg. flickering of video signal in TV, ECG,

3
 9/26/2024

3. Periodic and Aperiodic signal

Periodic Signal:
A signal which has a definite pattern and repeats itself after a fixed
time period.
They are deterministic signal.
Examples: sine, cosine, square wave
A signal is periodic if
For Continuous: X(t) = X (t + KT0)
for all integer K
and positive value T0

For Discrete: X[n] = X (n + KN0)

3. Periodic and Aperiodic signal

Aperiodic Signal:
Signals which does not repeat itself after a specific interval of time.
Also called non-periodic signal.
They are random signals.
They cannot be represented in form of mathematical equations.
Examples: sound signals from radio noise signals.

4
 9/26/2024

4. Causal, Non Causal and Anti-Causal Signals

Causal Signal:
Signal that begins from time instant zero and move towards + ∞
is casual signal. (Present and Past values)
Casual signals are signals that are zero for all negative time.
i.e. x(t) = 0 , t<0
Anti-Causal Signal:
signals which occurs only in the –ve time instants are called Anti-
Causal signal are signals. (Future values)
Signals that are zero for all +ve values of time.
i.e. x(t) = 0 , t>0
Non casual Signal:
Signals which appears both in –ve and +ve time instant.
signals that have non zero values in both –ve and +ve values of
time.
9

Example : Causal and Noncausal.

yn 
1
xn  1  xn  xn  1
3
Causal or non-causal?
Solution:
Non-causal; the output signal y[n] depends on a future value of the
input signal, x[n+1]
yn 
1
xn  1  xn  xn  1
3

Causality is required for a system to be capable of operating in real

time.

10

5
 9/26/2024

5. Even and odd Signals:

• If x(-t)= x(t) then the signal is even
• And if x(-t)= - x(t) then the signal is odd

11

6. Energy and Power signal

A signal is energy signal if total energy of signal satisfies the condition
0<E<∞ And Pav = 0

A signal is a power signal if and only if average power of signal satisfies the
condition
0<P<∞ and E=∞

12

6
 9/26/2024

Special types of Signals

1. Harmonic signal

13

2. Unit step function, signum function and

Impulse function
• A signal which exist only for positive side and is zero for negative side.
The unit step signal is denoted by u(t).
• u(t) = 1 for t ≥ 0
0 for t < 0

14

7
 9/26/2024

• Signum function is the signal used to define the sign of the signal
Sgn(t) = 1 for t> 0
-1 for t<0
0 for t= 0

Impulse function is the mathematical model to represent the physical

phenomenon that takes place in very short period. It is also known as
delta signal.

15

3. Sinc Signal:

16

8
 9/26/2024

Hilbert Transformation

9
 9/26/2024

Let a signal x(t) with Fourier transform X(ω). The Hilbert transform of
x(t) is obtained by the convolution of x(t) and (1/πt), i.e.,

Properties of Hilbert transform

10
 9/26/2024

Inverse Hilbert transform

The process of recovering the original signal x(t) from x^(t)is called the inverse
Hilbert transform. Mathematically, it is defined as,

The equations of functions x(t) and x^(t) together are called Hilbert transform pair.

Fourier Series

To represent any periodic signal x(t), Fourier devel oped a n expression ca lled Fourier
series. This is in terms of a n infinite sum of sines a n d cosines or exponentials. Fourier
series uses orthoganality condition.
Fourier Series Representation of Continuous Time Periodic Signals
A signal is said to b e periodic if it satisfies the condition x (t) = x (t + T) or x (n) = x (n + N).
Where T = fund a me nta l tim e p erio d,
ω 0= fund a me nta l frequenc y = 2π /T
There are two basic periodic signals: x(t)=cosω0t(sinusoidal) & x(t)=ejω0t(complex
exponential)
These two signals are periodic with period T=2π/ω0
A set of harmonically related complex exponentials c a n b e represented as {ϕk(t)}

All these signals are periodic with period T

11
 9/26/2024

Fourier Series

According to orthogonal signal s p a c e approximation of a function x (t) with n, mutually

o rtho g o nal func tio ns is given by

Where ak = Fourier coefficient = coefficient of approximation.

This signal x(t) is also periodic with period T.
Equation 2 represents Fourier series representation of periodic signal x(t).
The term k = 0 is consta n t.
▶ The term k=±1 having fundamental frequency ω0 , is ca lled as 1st harmonics.
▶ The term k=±2 having fundamental frequency 2ω0 , is ca lled as 2 nd harmonics, a n d so
on...
▶ The term k=±n having fundamental frequency nω0, is ca lled as nth harmonics.

Fourier Series

Deriving Fourier Coefficient

We know that

Multiply e−jnω0t on both sides. Then

Consider integral on both sides.

12
 9/26/2024

Fourier Series

by Euler's formula,

He nce in equation 2, the integral is zero for all values of k except at k = n. Put k = n in
equation 2.

Re pla ce n by k

Fourier Series Properties

Properties of Fourier series:

Linearity Property

Time Shifting Property

13
 9/26/2024

Fourier Series Properties

Frequency Shifting Property

Time Reversal Property

Time Scaling Property

Fourier Series Properties

Differentiation and Integration Properties

Multiplication and Convolution Properties

14
 9/26/2024

Fourier Series Properties

Conjugate and Conjugate Symmetry Properties

Trigonometric Fourier Series

Trigonometric Fourier Series (TFS)

sinnω0t a n d sinmω0t are orthogonal over the interval (t0,t0+2πω0). So sinω0t,sin2ω0t forms
a n orthogonal set. This set is not comple te without {cosnω0t } beca use this cosine set is
also orthogonal to sine set. So to comple te this set w e must include both cosine a n d sine
terms. Now the comple te orthogonal set contains all cosine a n d sine terms i.e.
{sinnω0t,cosnω0t } where n=0, 1, 2...

The a b o ve equation represents trigonometric Fourier series representation of x(t).

15
 9/26/2024

Trigonometric Fourier Series

Exponential Fourier Series

Exponential Fourier Series (EFS):

Consider a set of complex exponential functions
which is orthogonal over the interval (t0,t0+T). Where T=2π/ω0 . This is a complete set so it is
possible to represent any function f(t) as shown below

Equation 1 represents exponential Fourier series representation of a signal f(t) over the
interval (t0, t0+T).

16
 9/26/2024

Exponential Fourier Series

Exponential Fourier Series

Relation Between Trigonometric and Exponential Fourier Series:

Consider a periodic signal x(t), the TFS & EFS representations are given below respectively

17
 9/26/2024

ContinuousTime Fourier Transform

INTRODUCTION:
The main dra wba ck of Fourier series is, it is only a pplicable to periodic signals. There are
some naturally produ ce d signals such as nonperiodic or aperiodic, which w e ca nnot
represent using Fourier series. To o ve rcome this shortcoming, Fourier d evel oped a
mathematical model to transform signals between time (or spatial) domain to
frequency domain & vice versa, which is called 'Fourier transform'.
Fourier transform has many applications in physics a n d engineering such as analysis of LTI
systems, RADAR, astronomy, signal processing etc.
Deriving Fourier transform from Fourier series:
Consider a periodic signal f(t) with period T. The complex Fourier series representation of
f(t) is given a s

ContinuousTime Fourier Transform

18
 9/26/2024

ContinuousTime Fourier Transform

In the limit as T→∞,Δf approache s differential df, kΔf becomes a continuous variable f,
a n d summation be come s integration

Fourier transform of a signal

Inverse Fourier Transform is

Fourier Transform of Basic Functions

FT of GATE Function

FT of Impulse Function:

19
 9/26/2024

Fourier Transform of Basic functions

FT of Unit Step Function:

FT of Exponentials:

FT of Signum Function :

ContinuousTime Fourier Transform

Conditions for Existence of Fourier Transform:

Any function f(t) c a n b e represented by using Fourier transform only when the function
satisfies Dirichlet‟s conditions. i.e.
▶ The function f(t) has finite number of maxima a n d minima.
▶ Theremust b e finite number of discontinuities in the signal f(t),in the
given interval of time.
▶ It must b e absolutely integrable in the given interval of time i.e.

20
 9/26/2024

Fourier Transform Properties

Linearity Property:

Then linearity property states that

Time Shifting Property:

Then Time shifting property states that

Fourier Transform Properties

Frequency Shifting Property:

Then frequency shifting property states that

Time Reversal Property:

Then Time reversal property states that

21
 9/26/2024

Fourier Transform Properties

Time Scaling Property:

Then Time scaling property states that

Differentiation and Integration Properties:

Then Differentiation property states that

a n d integration property states that

Fourier Transform Properties

Multiplication and Convolution Properties:

Then multiplication property states that

a n d convolution property states that

22
 9/26/2024

Discrete Time Fourier Transform

Discrete Time Fourier Transforms (DTFT)

Here w e take the exponential signals to b e where „w‟is a real number. The
representation is motivated by the Harmonic analysis, but instead of following the
historical development of the representation we give directly the
defining equation.
Let {x[n]} b e discrete time signal such that , that is sequence is absolutely
summable.
The sequence {x[n]} c a n b e represented by a Fourier integral of the form,

Where,

Discrete Time Fourier Transform

Equation (1) a n d (2) give the Fourier representation of the signal.

Equation (1) is referred as synthesis equation or the inverse discrete time Fourier transform
(IDTFT) a n d equation (2)is Fourier transform in the analysis equation.
Fourier transform of a signal in general is a complex valued function, w e c a n write,

w here is m a gnitud e a nd is the p ha se.

We also use the term Fourier spectrum or simply, the spectrum to refer to. Thus is called
the magnitude spectrum a n d is called the phase spectrum.

Interchanging the order of integration,

23
 9/26/2024

Discrete Time Fourier Transform

Example: Let
Fourier transform of this sequence will exist if it is absolutely summable. We have

Discrete Time Fourier Transform

24
 9/26/2024

Discrete Time Fourier Transform

Discrete Time Fourier Transform

Fourier transform of Periodic Signals

For a periodic discrete-time signal,

its Fourier transform of this signal is periodic in w with period 2∏ , a n d is given

Now consider a periodic sequence x[n] with period N a n d with the Fourier series
representation

The Fourier transform is,

25
 9/26/2024

Discrete Time Fourier Transform

Discrete Time Fourier Transform

26
 9/26/2024

Discrete Time Fourier Transform

Discrete Time Fourier Transform

Properties of the Discrete Time Fourier Transform:

Let {x[n]}and {y[n]} b e two signal, then their DTFT is denoted by and. The notation

is used to say that left ha nd side is the signal x[n] whose DTFT is given a t right ha nd side.

1.Periodicity of the DTFT:

27
 9/26/2024

Discrete Time Fourier Transform

2. Linearity of the DTFT:

3.Time Shifting and Frequency Shifting:

Discrete Time Fourier Transform

4.Conjugation and Conjugate Symmetry:

From this, it follows that Re{X(e jw )} is a n even function of w a n d Im{X (e jw )} is a n o d d

function of w . Similarly, the magnitude of X(e jw ) is a n even function a n d the phase
angle is a n o d d function. Furthermore,

28
 9/26/2024

Discrete Time Fourier Transform

5.Differencing and Accumulation

The impulse train on the right-hand side reflects the d c or avera ge value that c a n result
from summation.

Discrete Time Fourier Transform

6.Time Reversal

7.Time Expansion
For continuous-time signal, w e have

For discrete-time signals, however, a should b e a n integer. Let us define a signal with k a
positive integer,

29
 9/26/2024

Discrete Time Fourier Transform

For k > 1, the signal is spread out a n d slowed down in time, while its Fourier transform is
compressed.

Discrete Time Fourier Transform

8.Differentiation in Frequency

The right-hand side of the a b o ve equation is the Fourier transform of - jnx[n] . Therefore,
multip lying b oth sides by j , w e see tha t
9.Parseval’s Relation

30
 9/26/2024

Parseval’s Theorem for Energy signals:

Proof:

31
 9/26/2024

Energy Spectral Density:

32
9/26/2024

33
 9/26/2024

Parseval’s Theorem for Power Signals:

• This theorem relates average power of a periodic signal to its fourier series
coefficients
• It states the total average power of a periodic signal x(t) is equal to the sum of
average powers of individual fourier coefficients (Cn).
• The average power of a signal x(t) is defined as:

• If the signal can be represented as sum of fourier series components and power is
defined as:

Where Cn is the amplitude of nth harmonic component of fourier series.

Proof:

34
 9/26/2024

Power Spectral Density Function

• In applying frequency-domain techniques to the analysis of random

signals the natural approach is to Fourier transform the signals.

• Unfortunately the Fourier transform of a stochastic process does not,

strictly speaking, exist because it has infinite signal energy.

• But the Fourier transform of a truncated version of a stochastic

process does exist.

35
9/26/2024

36
9/26/2024

37
 9/26/2024

Auto correlation function and Psdf:

38
 9/26/2024

Properties of Autocorrelation function:

System:
A system refers to any physical device that produces output signal in
response to an input signal.

Input  x(t)
Process  h(t)
Output  y(t)

78

39
 9/26/2024

Example of system
 In automatic speaker recognition system; the system is to extract the information
from an incoming speech signal for the purpose of recognizing and identifying the
speaker.
 In communication system; the system will transport the the information contained
in the message over a communication channel and deliver that information to the
destination.

Figure: Elements of a communication system.

Figure : Block diagram representation of a system.

79

Linear and non linear system

• A linear system is any system that obeys the property of scaling and
superposition. i.e. output is linearly proportional to input.

80

40
 9/26/2024

• Non linear system do not posses linearity.

81

Time variant and Time invariant system

• A system is said to be time variant system if input output

characteristics changes with time.
X(n-n0)  Y (n +n0)

• A system is said to be time invariant system if its input output

characteristics do not changes with time.

X(n-n0)  Y (n -n0)

82

41
 9/26/2024

Properties of Linear time invariant system

1. The output of the LTI system is convolution sum of its input and the
impulse response (IR)
2. Commutative property: x(t) * h(t) = h(t) * x(t)
3. Distributive property: x(t) *( h1(t) + h2(t)) = x(t) *h1(t) +x(t) *h2(t)
4. Associative property: x(t) *{h1(t) *h2(t) } = { x(t) * h1(t) }*h2(t)
5. Memory and Memoryless LTI system: In memory LTI system, the output
depends on present and past input and output value.
Y(t) = X(t) + Y(t-1)
In memoryless LTI system, the output depends only on present input.
Y(t) = X(t)

83

IMPULSE RESPONSE h(t)

x(t) LTI y(t)

y(t) is the output of the continuous-time LTI system with input x(t) and
no initial energy.

(t) LTI h(t)

With the unit impulse as an input [i.e., x(t)=(t)], the output is defined
as the IMPULSE RESPONSE and is represented by h(t).
84

42
 9/26/2024

A conceptual view of convolution

Convolution is a mathematical way of combining two signals to form a third signal.

Convolution is a mathematical operation used to express the relation between input and output of an
LTI system. It relates input, output and impulse response of an LTI system as:
y(t)=x(t)∗h(t)
85

CT CONVOLUTION INTEGRAL

DT CONVOLUTION SUM
y[n]  x[n] h[n]   x[i]h[n  i]
i

86

43
 9/26/2024

Commutative

x(t)  h(t)  h(t)  x(t)

x(t) h1(t) h2(t) y(t)

Same output!

x(t) h2(t) h1(t) y(t)

87

Distributive
x(t) [h1(t)  h2 (t)]  x(t)  h1(t)  x(t)  h2 (t)

88

44
 9/26/2024

Associative

x(t) [h1 (t)  h2 (t)]  [x(t)  h1(t)] h2 (t)

x(t) h1(t) h2(t) h3(t) y(t)

x(t) h1(t)*h2(t) h3(t) y(t) Same output!

x(t) h1(t) h2(t)* h3(t) y(t)

89

90

45
 9/26/2024

SYSTEM MEMORY

A system is said to possess memory if its output signal depend

on past or future values of the input signal.

A system is memoryless if for any time t=t1, the value of the

output at time t1 depends only on the value of the input at time
t=t1. In other words, the value of the output signal depends only
on the present value of the input signal.

91

Example : Memory and Memory-less System.

Below is the moving-average system described by the input-output relation.
Does it has memory or not?
(a) yn  1 xn  xn  1  xn  2
3

(b) yn  x n

2

Solution:
(a) It has memory, the value of the output signal y[n] at time n depends on
the present and two pass values of x[n].

(b) It is memoryless, because the value of the output signal y[n] depends
only on the present value of the input signal x[n].

92

46
 9/26/2024

Inverse Systems
A system is invertible if the input of the system can be recovered
from the output of the system. For example, this concept is important
in communication applications. We will focus on this property with
our echo cancellation lab.

93

Stable LTI system

• A system is said to be BIBO stable, i.e. bounded input and bounded
output if every bounded input produces a bounded output.
• The output doesn’t diverge until and unless input is diverged.

94

47
 9/26/2024

Signal transfer in LTI system

Prove that Impulse response of a system is the output of the system when the input is a delta function

95

System Bandwidth
In the case of a low pass system , the 3-dB bandwidth is
defined as the difference between zero frequency at which the
amplitude response attains its peak value H(0) and the
frequency at which the amplitude response drops to a value
equal to H(0)/√2 .

In the case of band-pass system the 3-dB bandwidth is defined as the difference
between the frequencies at which the amplitude response drops to a value equal to
1/√2 time the peak value H( fc ) at mid band frequency fc .

96

48
 9/26/2024

Signal bandwidth: The bandwidth of a signal provides a measure of the

extent of significant spectral content of the signal for positive
frequencies .
A signal is said to be low pass if its significant spectral content is
centered around the origin, and bandwidth is defined as one half total
width of main spectral lope.

A signal is said to be band pass if its

significant spectral content is central
around ±fc where fc is a non zero
frequency and the bandwidth is defined
as the width of main lope for positive
frequencies.
97

Band pass signals and system

Band pass signal are the signal that posses the pass band frequencies.

It can be represented by inverse Fourier transform

For example, a radio receiver contains a band-pass filter to select the
frequency of the desired radio signal out of all the radio waves picked
up by its antenna.
98

49
 9/26/2024

Ideal low pass filter

• An ideal low pass filter is a filter that passes pass band frequency without any
distortion and complete attenuate stop band frequencies.

X(ω)
δ(t) x(t)
Ideal LPF
H(ω)
ω
ωm

• An ideal low pass filter should pass without any attenuation or distortion , all
signal frequencies below a certain frequency ‘ωm’ in rad/sec . Whereas signal
frequencies above ωm are completely attenuated.
• Thus, the frequencies response (magnitude response) of ideal LPF is a gate
function and phase response is linear and equal to – ωtd .

99

Let us assume we have a desired frequency response of H(jω)

1
This filter will need the following impulse response

- ωc ωc

Where,

Unfortunately, the sinc function has infinite support. That is, it has non-zero values all the
way from −∞−∞ to ∞∞. In other words, our ideal filter requires infinite memory which makes
it non-realizable.
100

50
 9/26/2024

Sampling

Sampling is the process of converting a continuous-time signal into a discrete-time signal

by taking samples at specific intervals. The goal is to capture the essential information of
the continuous signal while reducing it to a format that can be processed digitally.

Sampling Rate (Frequency): The number of samples taken per unit time. It is usually
measured in Hertz (Hz).

Sampling Interval: The time between successive samples, which is the reciprocal of the
sampling rate (i.e., Ts=1/fs, where fs​ is the sampling frequency).

The sampling rate must be sufficiently high to accurately represent the continuous signal.
This is governed by the Sampling Theorem.

Sampling Theorem

The Sampling Theorem, also known as the Nyquist-Shannon Sampling Theorem, states:

For a signal to be completely represented in its discrete form without loss of

information, it must be sampled at a rate that is at least twice the maximum frequency
component of the signal.

Mathematically: If x(t) is a continuous-time signal with a maximum frequency fmax​, then

it should be sampled at a rate fs such that s​≥2fmax​. This rate is known as the Nyquist
Rate.

51
 9/26/2024

Sampling theorem of low pass signals

Statement of Sampling Theorem:

A b a n d limited signal c a n b e reconstructed exactly if it is sampled at a rate atleast twice
the maximum frequency compone nt in it.“
The following figure shows a signal g(t) that is b a n d limited.

Figure1: Spectrum of b a n d limited signal g(t)

The maximum frequency compone nt of g(t) is fm. To recover the signal g(t) exactly from
its samples it has to b e sampled at a rate fs ≥2fm.
The minimum required sampling rate fs = 2fm is called “Nyquist rate”.

Sampling theorem of low pass signals

Figure 2: (a) Original signal g(t) (b) Spectrum G(ω)

52
 9/26/2024

Sampling theorem of low pass signals

Let g s(t) b e the sampled signal. Its Fourier Transform Gs(ω) is given by

Sampling theorem of low pass signals

53
 9/26/2024

Sampling theorem of low pass signals

Aliasing:
Aliasing occurs when a signal is sampled at a rate lower than its Nyquist Rate Aliasing is a phenomenon
where the high frequency components of the sampled signal interfere with e a c h other
beca use of inadequate sampling ωs < ω m

Aliasing leads to distortion in recovered signal. This is the reason why sampling frequency
should b e atleast twice the bandwidth of the signal.

Sampling theorem of low pass signals

Aliasing:

This results in different signals becoming indistinguishable from each other when sampled,
leading to distortion or misrepresentation of the original signal.

Mathematically: Aliasing can be described by the formula falias= |f - kfs,| where f is the
original frequency, k is an integer, and fs​ is the sampling frequency.

Prevention: To avoid aliasing, you can either sample at a rate higher than the Nyquist Rate
or use an anti-aliasing filter to remove high-frequency components before sampling.

54
 9/26/2024

Sampling theorem of low pass signals

Oversampling:
In practice signal are oversampled, where fs is significantly higher than Nyquist rate to
avoid aliasing.

Sampling
If Rs < 2B, aliasing (overlapping of the spectra) results.

If signal is not strictly band limited, then it must be passed through

Low Pass Filter (LPF) before sampling.

Types of sampling
Ideal Sampling
Natural Sampling
Flat-Top Sampling

55
 9/26/2024

Ideal Sampling ( or Impulse Sampling)

Performed by the multiplication of the signal x(t) by the uniform train of impulses (comb function).
The output is simply the replication of the original signal at discrete intervals.
Amplitude of impulse changes with respect to amplitude of input signal.

Natural Sampling / Practical Sampling

In practice we cannot perform ideal sampling. It is practically difficult to create a train of
impulses
Thus a non-ideal approach to sampling must be used.
We can approximate a train of impulses using a train of very thin rectangular pulses.

The problem with a natural sampled waveform is that the tops of the sample pulses are not flat.
It is not compatible with a digital system since the amplitude of each sample has infinite
number of possible values.
Another technique known as flat top sampling is used to alleviate this problem.

56
 9/26/2024

Flat-Top Sampling
Flat top sampling is obtained by the convolution of the signal obtained after ideal
sampling with a unity amplitude rectangular pulse, p(t).
The top of the samples i.e.pulse is held to a constant height for the whole sample
period
This technique is used to realize Sample-and-Hold (S/H) operation
In S/H, input signal is continuously sampled and then the value is held for as long as it
takes to for the A/D to acquire its value

57