You are on page 1of 96

Communications Engineering

ELEC 341

Chapter 2

SIGNALS AND SIGNAL SPACE

1
Outline
2.1 Size of Signals
2.2 Classification of Signals
2.3 Signal Operations
2.4 Mathematical Signals
2.5 Signal Representation
Trigonometric representation
Phasor representation

2.6 Signal Spectrum


2.7 Fourier Series and Fourier Transform
2.8 Convolution
2
2.1. Size of Signal (Section 2.1 from textbook)
The size of any entity is a number that indicates the largeness or strength of that
entity. The size of a signal is measured in terms of energy and power.
Energy Signal:
We define Signal Energy ( ) of a real value signal g(t) {finite signal called
energy signal} (the energy that the voltage g(t) dissipates on the resistor)

Eg dt energy signal
2
g (t ) g(t)

+ 2
For Complex-valued signal =

0 time
Power Signal: (a) signal with finite energy

The average power is defined by


g(t) power signal
1 T /2
Pg lim g (t ) 2 dt
T / 2
T T

For Complex-valued signal 0 time


1 T /2
Pg lim
2
g (t ) dt

T T T / 2
Figure 2.1 Examples of signals:
(b) signal with infinite energy.
Energy Signal
Therefore, we first square the signal and then find the area under
that curve


2
[]: = <

Energy Signal has finite Energy: <

4
Energy Signal (cont.)
What if the signal does not decay and has the infinite duration?
In this case we have infinite energy for any such signal

Does this mean that a sixty hertz sine wave fed into your
headphones is as strong as the sixty hertz sine wave coming out
of your outlet?

How we can compare signals with infinite energy?

5
Power Signal
Power is a time average of energy [joule/second].
This is useful when the energy of the signal goes to infinity.

1 /2 2
: = lim
/2

Power Signal is finite <


The signal power represents time average of the signal amplitude squared
It is finite only if the signal is periodic or has statistical regularity

6
Power signal & energy signal
A signal with finite energy has zero power
A signal can be either energy signal or power signal, not both!
A signal can be neither energy nor power, for example ramp
signal, r(t)

where k is the rate of ramping.


r(t)
slope = k
ramp signal

time
0

Every signal observed in real life is an energy signal


7
Example 2.1
Determine the suitable measure of the Signals below

Answer:
+
(a) =
=
/ = =
(b) = lim 1/3
/

8
Units of Signal Energy and Power
Joule: units of signal energy
Watt: unit of Power = Joule/second
One ampere at one volt (P = I * V )
Access points generally radiate 30 to 100 mW (20 dBm)
Wireless LAN transmission power in Laptop is equal 32mW (15 dBm)
Cellular phone transmission power from 1000mW (30 dBm) to 500mW (27dBm)
FM radio station transmission power with around 50km range is 100kW (80 dBm)
dB is a logarithmic ratio of powers. So saying 1 = 2 2 is equivalent to

saying 1 = 3 , because 10 log10 1 /2 = 10 log10 2 = 3
2
dB can be used also as units for absolute power level
Also, for voltage ratios, assuming equal impedance levels
2
2
Pout Vout R V P V
g 2 out g[dB ] 10 log10 out 20 log10 out
Pin Vin R Vin Pin Vin
dBm and dBW is a (logarithmic) unit to measure powers:
[]
= 10 log10 = 30
1
1W = 1000mW = 30dBm 1W is equal 0 dBW 1mW = 0.001W = -30dBW

9
2.2. Classification of Signals
Power signal & Energy Signal
Analog signal and Digital signal Continuous time Signal
Based on the signal value
Continuous and Discrete Signals
Based on the time variable
Even symmetric signal & odd Discrete time signal
symmetric signal
Causal signal & anti-causal signal
Periodic and aperiodic (non-periodic)
Deterministic signal & probabilistic
(random stochastic) signal
Deterministic : You know the value of the
signal for sure
Stochastic: you can only guess its value
Orthogonal signal & non-orthogonal
signal
Analog Signal & Digital Signal
Analog signal varies with time in a smooth and continuous fashion
Analog signal: amplitude takes infinite number of values

Digital signal is an ordered sequence of symbols selected from a


finite set of discrete elements
Digital signal: amplitude takes finite number of levels
amplitude

amplitude
analog signal digital signal
signal

signal

time time

11
Continuous time signal & Discrete time signal
Continuous time signals take on real or complex values as a function of an
independent variable that ranges over the real numbers and are denoted as g(t)
Discrete time signals take on real or complex values as a function of an
independent variable that ranges over the integers and are denoted as g[n]
Note the use of parentheses and square brackets to distinguish between
continuous time signal and discrete time signal.
Analog and continuous time Digital and continuous

g[n] g[n]

Analog and discrete-time Digital and discrete time 12


Symmetric signal & asymmetric signal:
Symmetric signal has same magnitude on both sides of y-axis.
Specifically, |x(-t)| = |x(t)|.

Symmetric signals can be further classified into even


symmetric signal or odd symmetric signal.

A signal that is not symmetric is asymmetric signal such that


|x(-t)| |x(t)|.

symmetric signal x(t)

0 time

13
Symmetric signal & asymmetric signal:
Even symmetric signal () is defined as follows:
() = ().
Odd symmetric signal xo(t) is defined as follows:
() = ().

If the signal is even, it is composed of cosine waves. If the


signal is odd, it is composed of sine waves.
If the signal is neither even nor odd, it is composed of both
sine and cosine waves.
14
Causal signal & anti-causal signal
A causal signal is any signal that is zero prior to time zero.
Thus, if g(t) denotes the signal amplitude at time t
he signal g(t) is said to be causal if g(t) = 0 for all t < 0.
Simply, a causal signal is zero for t < 0.

An anti-causal signal is zero for t > 0.

g(t) g(t)

causal signal anti-causal signal

0 time 0 time

15
Periodic signal & Aperiodic (non-periodic) signal
A signal g(t) is periodic if there is positive constant 0 such that
= + 0 for all

A signal is aperiodic if it is not periodic

g(t)

periodic signal

0 T0 time

g(t)

aperiodic signal

0 time

16
Properties of Periodic Signals
The signal is periodic if
0 = < < +
where m is any integer. In this case, the signal can be constructed by
combining signal segments of length T0:

The length of the periodic signals is infinite, therefore, the signals in the
practical system can not be strictly periodical.
However, many finite-length signals in practical systems correspond
very accurately to the pure periodic signals.
17
Deterministic signal & random (stochastic) signal
A signal g(t) is called deterministic, if it is completely known
and can be described mathematically. There is no uncertainty
at all.

A signal g(t) is called random, if it takes random value at any


time. Random signal can be described only by probabilistic
and statistic terms, such as
Distribution
Mean value
Squared mean value
Standard deviation

18
Orthogonal signal & non-orthogonal signal
Two signals x(t) and y(t) are orthogonal over a given interval T if
/2
() = 0
/2
Orthogonal signals include those, which are disjoint in time
and/or frequency.

Two signal are non-orthogonal if they are not orthogonal.

19
Example:
Show that cos(t) and sin(t) are orthogonal within interval [-T/2,
T/2], where T is the period.

/2
Poof orthogonality by showing /2
cos() sin() =0

/2
1 /2
= 2
/2 2 /2

1 2
= cos 2 =0
4
2

20
Signal Operations
Time-shifting
Time scaling
Time-inversion / Time-reversal / Reflection
Time shifting
Time-shifting a signal x(t) by T means the time parameter is replaced
by t-T, where T can be a positive or negative number. The time-shifted
signal g(t) = x(t-T).
x(t) 1 0
=
0

0 time The signal is


shifted to the
delayed signal g(t) = x(t-|T|) right if T is
positive.
Hence, the
signal is ( 2)
delayed.
0 time
T T
The signal is
advanced signal g(t) = x(t+|T|) shifted to the
left if T is
negative.
Hence the 0
2 =
signal is 0
0 time advanced.
Time Scaling
Time-scaling a signal g(t) by a scaler k means the time parameter is replaced by
kt, where k is a positive real number. The time-scaled signal () = ().

() = () ( )

k >1 Compression
(/) = (/) (/ )

0< k <1 Expansion

Time-scaling compresses or dilates a signal by multiplying the time variable by some quantity.
Examples of time compression and time expansion of signals
Time-inversion (reflection) of signal
Time-inversing (Time-reversing or Reflection) a signal g(t) means
the time parameter t is replaced by -t. The time-inversed (time-
reversed) signal = ()
Mathematical Signals
Unit step signal
Unit rectangular pulse
Unit impulse
Unit triangular pulse
Unit signum signal
Unit sinc pulse

Causal exponentiel eatu(t).


Unit step signal
Unit step signal u(t) is defined as follows:

1, t 0
u (t )
0, t 0
It is a signal that transits from zero to unit height at time t = 0.

u(t)
1

0 t
Unit rectangular pulse
Unit rectangular pulse rect(t) is a simple step function with a
unit height and a limited unit width.
0, | t | 1 / 2
rect(t )
1, | t | 1 / 2

rect(t)

1 Symbol (t) are sometime used


in the literature to represent
rectangular pulse, such that
(t) = rect(t).
-1/2 1/2 t
Unit rectangular pulse (cont.)
Unit rectangular pulse rect(t) can be expanded by a factor .
The expanded signal is represented by rect(t/), and illustrated
below:

rect(t/)

t 0, | t | / 2
1
rect
1, | t | / 2
-/2 /2 t

Note that the expansion factor is the pulse width.


Unit impulse
Unit impulse is also called Dirac delta function (t) is defined
as follows:

0 <0<
=
0,
where () is any function that is continuous at t = 0.

Practically, (t) = 0 for t 0, and

(t) (t)
1/ (t) can be seen as a
rectangular pulse with
very small width and
height 1/, for 0.
0 time -/2 /2 time
Unit triangular pulse
Unit triangular pulse tria(t) is a triangle with unit height and
unit area, as illustrated below:
0, || 1
tria = =
1 , || < 1
tri(t)
1

-1 1 t
Unit triangular pulse can be expanded by a factor . The expanded
signal is represented by / , and illustrated below:

tria(t/)
1 0, ||
/ =
1 , || <
- t

Unit signum signal:
The signum function is also known as the sign function,
because if t is positive, the signum function is +1; if t is
negative, the signum function is -1.

Signum function sgn(t) is defined as follows:


1, t 0
1, t 0
sgn( t ) 0, t 0 ,or sgn( t )
1, t 0 1, t 0

sgn(t)
sgn(t) is related to u(t).
1
Comparing u(t) to 0.5
sgn(t), they are the same
0 t
but only with 0.5 offset in
-1 magnitude.
Page 32
Unit sinc pulse:
Sinc pulse sinc(t) is a mathematical function defined as
follows:
1, t 0 There is another definition of sinc
sinc (t ) sin(t ) pulse as follow, which we do not
t , t 0 use in the lecture notes.
1, t 0
sinc (t ) sin( t )
t , t 0
sinc(t)

t
Some properties of sinc(t):
Crossings occur at nonzero integer values of t.
It has unit area.
sinc (t )dt 1.

It is even symmetric.
Dr. A. Al-Dweik CMME 360, L-01 Page 33
Signal Representation
Signals can be represented in time domain & frequency domain.
Oscilloscope provides time domain view.
Spectrum analyzer provides frequency domain view.

In time domain, signals are represented by:


Trigonometry (sinusoidal), OR
Phasor (complex exponential, complex sinusoidal).

In frequency domain, signals are represented by:


Amplitude spectrum, AND
Phase spectrum.
Signals and Spectrum
In the following we consider the representation of signals in the
time and frequency domains, and the relationship between the
representations
can be expressed as a sum of its impulse components
+
=

In the frequency domain, we view the signal as consisting of


sinusoidal or exponential components at various frequencies.

The mathematical definition of the frequency domain


representation, that is, the spectrum is determined by the
Fourier transform.

The spectrum for periodic signals is obtained using the Fourier


series.
Notice: singular: spectrum; plural: spectra 35
Trigonometric representation:
The simplest form of a single-tone sinusoidal
= cos 20 + = cos(0 + )

A A cos()

t
/(2f0) -A 1/f0
A sinusoidal waveform () = cos(20 + )

where A is the amplitude, w0 is the angular frequency


(f0 is the frequency) and is the phase.
2
This is a periodic signal whose period is 0 = = 1/0
0
One peak of the signal is at = /0

36
In trigonometric representation, the phase angle is always in
degree regardless of either 2f0t is in degree or radian.
Trigonometric representation of signals is always in the terms
of cosine. When a signal is given in terms of sine, we need to
convert sine to cosine: sin () = cos ( - 90).
The sinusoidal signal in trigonometric form can be
alternatively represented by phasor.

= +
Page 37
The complex representation of sinusoidal signal
Usually signals are real-valued. However, the concept of
complex signals is a useful tool in telecommunication.
Most cases can be handled by using real signals, however,
complex signals are widely used in spectral analysis.
Use of complex signal models arises from the properties of the
Fourier transform (the basis functions are complex).

The following Euler's equations are often needed:


= cos sin
On the other hand, sine and cosine are given by

Notice that |ejx| = 1 for any real x.


38
The complex representation of sinusoidal signal (cont.)
e j jw0t e j jw0t
cos( w0t ) e e
2 2

a) Unit length complex (b) Unit length complex v


variable with positive frequency ariable with negative frequency
(rotating counterclockwise) (rotating clockwise)

39
The complex representation of sinusoidal signal (cont.)
The following notations and conventions are used within this course:
The spectrum is a function of one variable,
the frequency f (Hz)
the angular frequency = 2f (rad/s).
f0 , f1 etc. are used for some fixed frequencies
The phase angle is measured with respect to cosine waves or,
equivalently with respect to the positive real axis of the phasor
diagram.
Amplitude is always positive: =

=
= +

40
Phasor Representation of Real Signal
This is a phasor diagram that shows
a rotating vector in a complex plane
Real signal corresponds to the real
part of the phasor:
A cos( w0t ) Re[ Ae j ( w0t ) ]
Re[ Ae j e j 2f0t ]
Phasor representation eases frequency analysis of the signal.
Frequency spectrum consists: (a) Amplitude spectrum, and (b)
Phase spectrum.

This is called one-sided line spectrum. It covers only positive


frequency. 41
Phasor Representation of Complex Signal
The phasor diagram for the two-sided spectrum of a sinusoid consists of two
vectors whose phase and direction of rotation are reversed. The resultant
vector is a real signal.
( +) ( +)
cos 0 + = 0 + 0
2 2

42
Linear combination of sinusoidal signals
Example: w(t ) 7 10 cos(40t 60) 4 sin120t

Time domain representation of w(t)

One-sided line spectrum

This can be rewritten in the form:


w(t ) 7 cos 2 0t 10 cos(2 20t 120 ) 4 cos(2 60t 90 )
which can be utilized to plot the one-sided line spectrum
43
Two-sided Spectrum
One-sided spectrum could be used for real signals. In the following
two-sided spectrum is used because it allows to handle also complex signals.
In the case of real signals, the two-sided spectrum is obtained by using the

substitution: cos 0 + = 0 + 0
2 2
Example: Plot the two-sided spectrum for:
= 7 10 cos 40 60 + 4 sin(120)
= 7 20 + 5 120 220 + 5 120 220 +2 90 260 + 2 90 260

Here, the basis functions are complex exponentials.

44
Signal Spectrum
Line spectrum vs continuous spectrum.
Line spectrum: Spectrum components exist only at
discrete values of frequencies.
Continuous spectrum: Spectrum components exist over a
continuous range of frequencies.

One-sided spectrum vs two-sided spectrum.


One-sided spectrum: Spectrum components exist only in
positive frequencies.
Two-sided spectrum: Spectrum components exists in both
positive and negative frequencies.

45
Fourier Series and Fourier Transform
[Section 3.1-3.3]:

An arbitrary periodic signal can be decomposed into a sum of


cosine functions, or complex sinusoidals. This sum of functions
is called Fourier series.

Fourier transform is the Fourier series equivalence for


aperiodic (non-periodic) signals.

Source: Praising Dr. Arafat Al-Dweik for Slides


Page 46
Section 2.7: Fourier Series:
A periodic power signal can be decomposed into a sum of
sinusoidals or a sum of phasors. The process of decomposition
is called Fourier series expansion. The sum is called Fourier
series.

Fourier series can be written in two different forms:


Exponential Fourier series.
Trigonometric Fourier series.

Page 47
Exponential Fourier series:
Signal v(t) is a sum of complex exponentials.

v(t ) n Cn e j 2nf0t

n is an integer
v(t) is a periodic power signal with period To = 1/f0,
f0 is called the fundamental frequency.

The Fourier series coefficient Cn is determined as follows:

Cn T1 v(t )e j 2nf0t dt
To / 2

0 To / 2

Page 48
Exponential Fourier series (cont.):
For generality, let coefficient Cn be a complex quantity
Cn Cn e j arg( Cn )

where arg(x) is a function that returns the angle of x.

Fourier coefficient Cn gives the frequency spectrum of signal


v(t):
|Cn| gives the amplitude spectrum
arg (Cn) gives the phase spectrum

Page 49
Section 2.7: Exponential Fourier Series (cont.)
The set of exponentials exp(jn0t) is orthogonal over any interval T0=2/0
0 m n
T e e dt T e
jm0t

jn0t *
dt j ( m n ) 0 t

0 0
T0 m n
A periodic signal can be written by using the exponential Fourier series

1 1
g (t )
n
Cne j 2nf0t T0
where Cn
f0
T0 T0
g (t )e j 2nf 0t dt
The complex coefficients Cn can be expressed using the polar form:
Cn Cn e j arg Cn
| Cn| is the value of the amplitude spectrum at nf0 and
argCn is the corresponding value of the phase spectrum.

The exponential Fourier series determines the two-sided spectrum for a periodic
signal. It consists of the harmonics (i.e., integer multiples) of the frequency.

50
Properties of the line spectrum
1. All frequencies are integer multiples or harmonics of the fundamental
frequency f0.
2. The DC component C0 (zeroth harmonic) equals the average value of the
signal:
1
C0 g (t ) g (t )dt
3. If g(t) is real, then T0 T0

C n Cn
arg C n arg Cn
which means that:
the amplitude spectrum has even symmetry
the phase spectrum has odd symmetry.

51
Visual interpretations

https://en.wikipedia.org/wiki/Fourier_series#/media/File:Fourier_series_and_transform.gif
https://en.wikipedia.org/wiki/Fourier_series#/media/File:Periodic_identity_function.gif
Page 52
The Role of the Phase Spectrum
The phase spectrum plays an equally important role in wave-shaping

The production of any signal f(t) is achieved by using a proper


combination of amplitudes and phases of various sinusoids
53
Visual interpretations

https://en.wikipedia.org/wiki/Fourier_series#/media/File:Fourier_series_and_transform.gif

https://en.wikipedia.org/wiki/Fourier_series#/media/File:Periodic_identity_function.gif

Page 54
Fourier Series Convergence: Visualization
One can visualize convergence of the Fourier Series by incrementally
adding terms.
2
Example: square waveform t = =1 sin(2)

2
Up to 1st Harmonic: = sin(2) Up to 3rd Harmonic

55
Fourier Series Convergence: Visualization
Up to 5th Harmonics Up to 7th Harmonics


Up to the 39th Harmonics: = = ()

56
Example:
Find exponential Fourier series of v(t) = cos(2f0t).

Two methods: (a) use the definition, and (b) use trigonometric
identities and Eulers theorem.
From Eulers theorem,
v(t ) cos(2f 0t )


2

1 j 2f0t
e e j 2f0t
Hence, C1 = , C-1 = , and Cn = 0 for n 1 and n -1.

Exercise: For v(t) = sin(2f0t), show that its exponential


Fourier series has C1 = -j, C-1 = j, and Cn = 0 for n 1
and n -1.

Page 57
Example:
Find exponential Fourier series for signal v(t) = sin(2f0t + ),
v(t ) sin( 2f 0t )


2j
e
1 j ( 2f0t ) j ( 2f0t )
e

2

j j j 2f0t
e e e j e j 2f0t
Hence,
j j j j
C1 e , and C1 e
2 2

Page 58
Trigonometric Fourier series:
Consider a real-valued signal v(t), we group the complex-
conjugate pairs.

v(t ) n Cn e j 2nf0t

C0 C
n 1 n
e
j 2nf 0t
Cn e
j 2nf 0t


C0 n 1

2 | Cn | cos 2nf 0t arg(Cn )
This is called the trigonometric Fourier series.

From the trigonometry Fourier series, we can find the one-


sided line spectrum for a periodic power signal. On the other
hand, exponential Fourier series gives two-sided line
spectrum.

Page 59
The compact form trigonometric Fourier series can be
expanded to include both cosine and sine terms.

v(t ) n Cn e j 2nf0t

C0 n1 Cn e j 2nf0t Cn e j 2nf0t

C0 n1 Cn cos(2nf 0t ) j sin( 2nf 0t ) Cn cos(2nf 0t ) j sin( 2nf 0t )

C0 n1 (Cn Cn ) cos(2nf 0t ) j (Cn Cn ) sin( 2nf 0t )

a0 n1 an cos(2nf 0t ) bn sin( 2nf 0t )

where
a0 C0 v(t )dt v(t )
To / 2
1
To To / 2
C n = an j b n
|Cn | = (a2n+ b2n)
an Cn Cn T2 v(t ) cos(2nf 0t )dt
To / 2

o To / 2 arg(Cn) = arc tan (-bn /an)


bn j (Cn Cn ) v(t ) sin( 2nf 0t )dt
To / 2
2
To To / 2

Page 60
Example:
Find Fourier series for a periodic train of rectangular pulses,
v(t) as given below.

Note that the rectangular pulse amplitude is A, pulse duration


is , and pulse period is T0. The ratio / T0 is called the duty
cycle of the pulse train. The signal v(t) is written as follows:
0, | t | / 2
v(t )
A, | t | / 2
With this v(t), there are discontinuities at |t| = /2.

Page 61
Fourier series coefficient, Cn is determined as follows:
To / 2 j 2nf 0t
Cn 1

T0 To / 2
v (t ) e dt
/2 j 2nf 0t
1

T0 / 2
Ae dt
/ 2 j 2nf t
A

T0 / 2
e 0 dt

A
T0 j 2nf 0
e jnf 0
e jnf 0

A sin(nf 0 )
T0 nf 0
A sinc (nf 0 )
T0
Af 0 sinc (nf 0 )

Page 62
The amplitude spectrum of signal v(t) is the amplitude of Cn,
i.e., |Cn |. For duty cycle = , i.e., f0 = .

Co is the DC component of v(t) and it is also the average value


of v(t).

Page 63
The phase spectrum of signal v(t) is the phase of Cn, i.e.,
arg(Cn). For duty cycle = , i.e., f0 = .

For real-valued v(t), amplitude spectrum is even symmetry,


and phase spectrum is odd symmetry.

For v(t), Fourier series coefficients Cn are real-valued, and take


negative values some times. When Cn is positive, the phase is
zero. When Cn is negative, the phase is 180.

Page 64
Parsevals Power Theorem:
Consider a periodic signal v(t), the average power Sv is given as
follows:
Sv
To / 2
1 2
To
| v(t ) | dt
To / 2


To / 2
1 v(t )v * (t )dt
To To / 2

T1
o
To / 2

To / 2

v(t ) * j 2f 0t
n n e
C dt
1 j 2f 0t *
To / 2
To / 2

n To v (t ) e dt Cn

C C
n n n
*


n | C n | 2

Average power of a periodic power signal is the sum of


power of its phasor components (Fourier series coefficients).
This is Parsevals power theorem.

Page 65
Fourier Transform:
In real-life, most signals do not continue forever and thus, are
(aperiodic) not periodic. For aperiodic signal v(t), the Fourier
transform V(f) is given as follows:

V( f ) v(t )e j 2ft dt

The corresponding inversed Fourier transform is as follows:



v(t ) V ( f )e j 2ft df

Symbolically, we can write Fourier transform as follows:


1
V ( f ) F [v(t )], and v(t ) F [V ( f )]
V ( f ) v(t ), and v(t ) V ( f )

Page 66
Time domain Frequency domain
Fourier
transform series


v(t ) n Cn e j 2nf0t v(t )e j 2nf0t dt
To / 2
Cn T
1

0 To / 2
Fourier


v(t ) V ( f )e j 2ft
df V( f ) v(t )e j 2ft dt

Page 67
Example:
Find Fourier transform for impulse v(t) = A (t).
v(t)
A

0 t

From the definition of Fourier transform,


V ( f ) F [v(t )]

v(t )e j 2ft dt V(f)

A

A (t )e j 2ft dt

A 0 f

Page 68
Example:
Find inversed Fourier transform for unit impulse in frequency
domain V(f) = (f). V(f)

0 f

From the definition of inversed Fourier transform,


v(t ) F 1[V ( f )]
v(t)

V ( f )e j 2ft df 1


( f )e j 2ft df 0 t

1 Note that spectrum of a constant is
an impulse. Just imagine the
frequency component of a DC signal.
Page 69
Example:
Find Fourier transform for a rectangular pulse v(t) = A rect(t/).
t
v(t) v(t) Arect
A
0, | t | / 2

A, | t | / 2
- /2 /2 t

From the definition of Fourier transform,


V ( f ) F [v(t )]

v(t )e j 2ft dt

t
Arect e j 2ft dt

/2
A e j 2ft dt
/ 2 continue next page
Page 70
/2
A
V( f ) e j 2ft
j 2f / 2

A e jf e jf

f 2j
A sinc ( f )
Note that V(0) = A is the net area under the graph. For V(f),
zero crossings happen at f = n/, where n is an integer.

V(f)
A

f
-5/ -4/ -3/ -2/ -1/ 1/ 2/ 3/ 4/ 5/

Page 71
Amplitude spectrum:

Phase spectrum:

Page 72
Rayleighs Energy Theorem:
Rayleighs energy theorem for non-periodic energy signal is
analogous to Parsevals power theorem for periodic power
signal.

The energy of an non-periodic signal can be calculated by


integrating the square its amplitude spectrum over all
frequencies.
E V ( f ) df
2

Page 73
Fourier Transform Pairs:
v(t) V(f)
1. t sinc (f )
rect

2. f
2W sinc (2Wt ) rect
2W
3. t
tria sinc 2 (f )

4. 1
t
e u (t ), 0 j 2f
5. (t ) 1
6. 1 ( f )
7. u (t ) 1 1
( f )
j 2f 2
8. sgn(t ) 1
jf

Page 74
General Properties of Fourier Transform:

Duality
Superposition
Time delay
Time scaling
Frequency translation
Modulation
Differentiation
Integration

Page 75
Duality:
Recall that V(f) is the Fourier transform for an non-periodic
energy signal v(t).
1
v(t ) F [V ( f )]

V ( f )e j 2ft df


V ( )e j 2t d

where is a dummy variable.

Consider a function z(t) that is related to V(f) such that z(t) =


V(t).
Z ( f ) F [ z (t )]

z (t )e j 2ft dt


z ( )e j 2f d continue next page

Page 76

Z( f ) V ( )e j 2 ( f ) d

v( f )

If there exists a time function z(t) related to V(f) such


that z(t) = V(t). Then, the Fourier transform of z(t) is
given as follows: Z ( f ) F [ z (t )]
v( f )

Page 77
Superposition:
Let a1 and a2 be constants. Then,
v(t ) a1v1 a2v2
and,
V ( f ) F [a1v1 (t ) a2v2 (t )]
a1 F [v1 (t )] a2 F [v2 (t )]
a1V1 ( f ) a2V2 ( f )

Generally,
V ( f ) F k ak vk (t )
k akVk ( f )

Page 78
Time Delay:
This Fourier transform property is also called time shifting.

Replacing t by t-td in v(t) produces a time-delayed signal v(t-td).


The signal v(t-td) has the same shape as v(t) but shifted to the
right along time axis for as much as td. The time delay
introduces an additional linear phase with slope -2td.
F v(t td ) V ( f )e j 2ft
d

phase difference

t
td

Page 79
Let x = t td be a dummy variable.

F v(t td ) v(t td )e j 2ft dt


v( x)e j 2f ( x td ) dx

j 2ftd
e v( x)e j 2fx dx

V ( f )e j 2ftd

Page 80
Time Scaling:
Multiplying t in v(t) with a scaling factor ( 0), to produce
v( t) can lead to signal compression, expansion, or time-
reversal.
It is signal expansion if | | < 1
It is signal compression if | |> 1
It is time-reversal if < 0

For compressed and expanded signal,


1 f
F v(t ) V

Compression (expansion) in time domain leads to expansion


(compression) in frequency domain. This is the reciprocal
relationship between time and frequency domains.
Page 81
To proof the scaling property, we consider < 0, such that =
-| |. We further define a dummy variable = t = -| |t.

F v(t ) v( | | t )e j 2ft dt

1
v( )e j 2 ( / ) f d

1
v( )e j 2 ( f / ) d

1 f
V

Note the change in sign in integration range when dummy
variable replaces time variable t.

Page 82
Frequency translation:
Multiplying a time function v(t) by ej2fct causes its spectrum to
be translated (shifted) in frequency domain by +fc.
F [v(t )e j 2fct ] V ( f f c )

From the definition of Fourier transform,



F v(t )e j 2f ct


v(t )e j 2fct e j 2ft dt

v(t )e j 2 ( f fc )t dt

V ( f fc )

In real-life, ej2fct is not a real function and cannot be


generated. Practical frequency shift is achieved by
multiplying a signal with sinusoidal.

Page 83
Modulation:
Modulation property is a direct application of the frequency
translation property.

It covers a general case where sinusoidal has a phase offset.


e j e j
v(t ) cos(2f ct ) V ( f fc ) V ( f fc )
2 2

Page 84
Differentiation:
For a differentiable signal v(t),

F
dv(t )
dt j 2f V ( f )
From Fourier transform definition,
dv(t ) d j 2ft
dt


dt
V ( f ) e dt


V( f )
dt
e
d j 2ft

dt

j 2f V ( f ) e j 2ft dt

By iteration,
d n v(t )
F n ( j 2f ) n V ( f )
dt
Page 85
Integration:
Consider a signal v(t) with zero net area, such that

v(t )dt 0
Then,

F v( )d
t V( f )
j 2f

We may relax the zero net area condition. Then,

F v( )d
t V( f ) 1
V (0) ( f )
j 2f 2

Page 86
v(t) V(f)
Duality V(t) v(-f)
Superposition k1v1(t) + k2v2(t) k1V1(f) + k2V2(f)
Time Delay v(t - td) V(f) e -j2ftd
Time Scaling v(t) 1f
V

Frequency v(t) ej2fct V(f fc)
Translation
Modulation v(t) cos(2fct + ) e j e j
V ( f fc ) V ( f fc )
2 2
Differentiation d n v(t )
n ( j 2f ) n V ( f )
dt
t V( f ) 1
Integration v( )d V (0) ( f )
j 2f 2

Page 87
Convolution:
Use symbol * to represent convolution. Then, convolution of
two functions of same variable, says v(t) and w(t) is defined as
follows:
v(t ) * w(t ) v( ) w(t )d


This is called the convolution integral.

Convolution is a special operation that defines the input-


output relationship of a system.

While * is used to indicate convolution operation, it must not


be confused with complex conjugate for a complex-valued
function. As a matter of fact, convolution has completely
nothing to do with the notation of a complex conjugate.

Page 88
In convolution integral, the independent variable is t and
integration is performed on dummy variable . There is one
integration for each value of t. Hence, there is indefinite
number of integrations needed to cover the whole continuous
range of t.

It will be easier to understand convolution integral through a


graphical interpretation. Consider two signals, v(t) and w(t) as
illustrated below:

v(t ) Ae t , 0 t
t
w(t ) , 0t T
T
Page 89
Recalled that convolution integral is performed on dummy
variable . Hence,
t
v( ) Ae , 0 w(t ) , 0 t T
T

To obtain the graph for w(t-), we first flip the original graph
of w() over the vertical axis, and then shift it to the left by as
much as t.

Page 90
Case 1: t < 0

There is no overlap between the two graphs. So, integration


will be zero. Thus, () () = 0 < 0.

Case 2: 0 < t < T

t
v(t ) * w(t ) Ae
t
d
0 T
A

t 1 e t , 0 t T
T Page 91
Case 3: t > T


t
t
v(t ) * w(t ) Ae d
t T
T


A
T

T 1 e T e ( t T ) , t T

Combined all the three cases,

Understand convolution by imagining you injecting an input signal


w(t) to a system represented by v(t). It is this injection that leads to
the time reversal in w(t) while performing integration.
Page 92
Comparison: Convolution vs Correlation

93
Properties of convolution:
Fourier transform of convolution in time domain is
multiplication in frequency domain.
F v(t ) * w(t ) V ( f )W ( f )

Fourier transform of multiplication in time domain is


convolution in frequency domain.
F v(t ) w(t ) V ( f ) *W ( f )
v( ) w( f )d

Page 94
Summary
Size of Signals
Classification of Signals
Signal Operations
Mathematical Signals
Signal Representation
Trigonometric representation
Phasor representation
Signal Spectrum
Fourier Series and Fourier Transform
Convolution
Tutorial 1: Uploaded on Black Board
95
Textbook Recommended Problems
3.1-4
3.1-5
3.1-6
3.1-7
3.2-2
3.2-3
3.31-1
3.3-2
3.3-6
3.3-7
3.3-10

Page 96

You might also like