Department of EEE, BUBT

EEE 301
Continuous Signals and Linear Systems

Lecture Notes
on
Fourier Transform

Prepared by
Md. Abul Hasnat
Lecturer, Dept. of EEE, BUBT
We have found the Fourier Series
Coefficients of an Rectangular Pulse Train
 𝑻/𝟐
𝟏 −𝒋
𝟐𝒏𝝅
𝒕
𝒄𝒏 = න 𝒙(𝒕)𝒆 𝑻 𝒅𝒕
𝑻
−𝑻/𝟐

𝝉/𝟐
𝟏 𝟐𝒏𝝅
−𝒋 𝑻 𝒕
= න 𝟏. 𝒆 𝒅𝒕
𝑻
−𝝉/𝟐

𝟏 𝑻 −𝒋
𝟐𝒏𝝅 𝝉
. 𝒋
𝟐𝒏𝝅 𝝉
.
= . 𝒆 𝑻 𝟐 −𝒆 𝑻 𝟐
𝑻 −𝒋𝟐𝒏𝝅
𝟐𝒏𝝅 𝝉 𝟐𝒏𝝅 𝝉
𝒋 . −𝒋 .
𝒆 𝑻 𝟐 −𝒆 𝑻 𝟐 𝒏𝝅𝝉
𝒋𝟐 sin 𝝉 𝒏𝝉
= = 𝑇 = 𝒔𝒊𝒏𝒄
𝒏𝝅 𝒏𝝅 𝑻 𝑇
𝑭𝒐𝒓 𝝉 = 𝟏𝒎𝒔 𝒂𝒏𝒅 𝑻 = 𝟓𝒎𝒔
𝑭𝒐𝒓 𝝉 = 𝟏𝒎𝒔 𝒂𝒏𝒅 𝑻 = 𝟏𝟎𝒎𝒔
𝑭𝒐𝒓 𝝉 = 𝟏𝒎𝒔 𝒂𝒏𝒅 𝑻 = 𝟐𝟎𝒎𝒔
 From Fourier Series to Fourier Transform

𝑻/𝟐
𝟏
𝒄𝒏 = න 𝒙(𝒕)𝒆−𝒋𝒏𝝎𝟎 𝒕 𝒅𝒕
𝑻
−𝑻/𝟐
2𝜋 𝟏 𝒅𝝎
When, 𝑻 → ∞, → 𝑑𝜔, ∴ →
𝑇 𝑻 𝟐𝝅
and 𝒏𝝎𝟎 → 𝝎 (𝒄𝒐𝒏𝒕𝒊𝒏𝒖𝒐𝒖𝒔 𝒗𝒂𝒓𝒊𝒂𝒃𝒍𝒆)
∞
𝑑𝜔
𝒄𝒏 = න 𝒙(𝒕)𝒆−𝒋𝝎𝒕 𝒅𝒕
2𝜋
𝒕=−∞
 From Fourier Series to Fourier Transform
Now, ∞

𝒙 𝒕 = ෍ 𝒄𝒏 𝒆𝒋𝒏𝝎𝟎 𝒕
𝒏=−∞
becomes,
∞ ∞
𝒅𝝎
𝒙 𝒕 = න න 𝒙(𝒕)𝒆−𝒋𝝎𝒕 𝒅𝒕 𝒆𝒋𝝎𝒕
𝟐𝝅
𝝎=−∞ 𝒕=−∞
∞ ∞
𝟏
= න න 𝒙(𝒕)𝒆−𝒋𝝎𝒕 𝒅𝒕 𝒆𝒋𝝎𝒕 𝒅𝝎
𝟐𝝅
𝝎=−∞ 𝒕=−∞
∞
𝟏
= න 𝑿(𝝎) 𝒆𝒋𝝎𝒕 𝒅𝝎
𝟐𝝅
𝝎=−∞
Synthesis and Analysis Equations
Previously we have seen………………….. ++

1 1
x (t )  1. sin( 2 * 50 * t )  sin( 2 * 3 * 50 * t )  sin( 2 * 5 * 50 * t )
3 5
BASIS
sin (2π*50*t) + (1/3)*sin (2π*3*50*t) + (1/5)*sin
(2π*5*50*t)
Exponential Fourier Series
In cases of Aperiodic Signals……………

These Gaps
are filled
Synthesis and Analysis Equations
Example: Find the Fourier Transform of the signal,
𝒙 𝒕 = 𝒆−𝒂𝒕 𝒖(𝒕), 𝒂 is real and 𝒂 > 𝟎

∞ ∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 = න 𝒆−𝒂𝒕 𝒖 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕

−∞ −∞
∞ ∞
−𝒂𝒕 −𝒋𝝎𝒕 −(𝒂+𝒋𝝎)𝒕
𝟏
=න𝒆 𝒆 𝒅𝒕 = න 𝒆 𝒅𝒕 =
𝒂 + 𝒋𝝎
𝟎 𝟎
𝟏 −𝟏 𝝎
𝑿 𝝎 = ∠ 𝑿 𝝎 = tan −
𝒂𝟐 +𝝎𝟐 𝒂
 Magnitude
Spectrum

Time Domain
Signal
Phase Spectrum
Example: Find the Fourier Transform of the signal, 𝒙 𝒕 = 𝒆−𝒂 𝒕 ,
𝒂 is real and 𝒂 > 𝟎
∞ 𝟎 ∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 = න 𝒆𝒂𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 + න 𝒆−𝒂𝒕 𝒆𝒋𝝎𝒕 𝒅𝒕

−∞ −∞ 𝟎
𝟎 ∞

= න𝒆 𝒂−𝒋𝝎 𝒕 𝒅𝒕 + න 𝒆−(𝒂+𝒋𝝎)𝒕 𝒅𝒕
−∞ 𝟎
𝟏 𝟏 𝟐𝒂
= + = 𝟐
𝒂 − 𝒋𝝎 𝒂 + 𝒋𝝎 𝒂 + 𝝎𝟐

𝟐𝒂
𝑿 𝝎 = ∠𝑿 𝝎 =𝟎
𝒂𝟐 +𝝎𝟐
 Magnitude
Spectrum

Time Domain
Signal

Phase Spectrum
Example: Find the Fourier Transform of the signal,
𝒙 𝒕 = 𝒔𝒈𝒏(𝒕)

∞ 𝟎 ∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 = න(−𝟏). 𝒆−𝒋𝝎𝒕 𝒅𝒕 + න 𝟏. 𝒆−𝒋𝝎𝒕 𝒅𝒕

−∞ −∞ 𝟎
𝟏 𝟏 𝟐
= + =
𝒋𝝎 𝒋𝝎 𝒋𝝎
 Fourier Transform of a rectangular pulse
Find the Fourier Transform of the signal, 𝒙 𝒕 = 𝒓𝒆𝒄𝒕(𝒕|𝝉)
∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕
−∞
𝝉
𝟐
= න 𝟏. 𝒆−𝒋𝝎𝒕 𝒅𝒕
𝝉
−
𝟐
𝝉 𝝉
−𝒋𝝎𝟐 𝒋𝝎𝟐
𝒆 −𝒆
=
−𝒋𝝎
𝝉
𝒋𝝎𝟐
𝝉
−𝒋𝝎𝟐 𝝎𝝉 𝝎𝝉
𝟐 𝒆 −𝒆 𝟐. 𝐬𝐢𝐧 𝒔𝒊𝒏
= . = 𝟐 =𝝉 𝟐 = 𝝉𝑺𝒂 𝝎𝝉 = 𝝉𝒔𝒊𝒏𝒄 𝝎𝝉
𝝎 𝒋𝟐 𝝎 𝝎𝝉 𝟐 𝟐𝝅
𝟐
 𝑭𝑻 𝝎𝝉
𝒙 𝒕 = 𝒓𝒆𝒄𝒕 𝒕|𝝉 𝑿 𝝎 = 𝝉𝑺𝒂
𝟐

Phase Spectrum
Magnitude Spectrum
 Time and Frequency Domain

x(t) X(ω)

ω
t

Frequency
Time Domain
Domain
An Impulse function has all frequency
components with same Magnitude

An Impulse function has all

frequency components with
same Magnitude
DC Signals have 0 frequency only

DC Signal has 0
frequency only
Signals with a single frequency component

Only has a frequency

of ω0
 Existence of Fourier Transform –
Dirichlet Conditions

Dirichlet Conditions are sufficient conditions for the existence

of Fourier Transform, not necessary conditions.

A signal satisfying all the Dirichlet Conditions must have a

Fourier Transform; but signal not satisfying any of the conditions
can also have a Fourier Transform.
 Violation of Dirichlet Condition 1:
Not Absolutely Integrable Signals
∞
න 𝒙 𝒕 𝒅𝒕 ≥ ∞
−∞

Power Signals violates this

condition, but still have Fourier
Transforms.
 Violation of Dirichlet Condition 2:
Infinite number of extrema within an interval

𝟏
𝒙 𝒕 = 𝐬𝐢 𝐧 ,
𝒕
𝟎<𝒕<𝟏
 Violation of Dirichlet Condition 3:
Infinite number of discontinuity within an interval
Linearity – Homogeneity
Linearity – Additivity
Example: Find the Fourier Transform of the signal,
𝒙 𝒕 = 𝐜𝐨𝐬 𝝎𝟎 𝒕

𝑭𝑻
𝒋𝝎𝟎 𝒕
𝒆 𝟐𝝅𝜹(𝝎 − 𝝎𝟎 )
𝑭𝑻
−𝒋𝝎𝟎 𝒕
𝒆 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝑭𝑻
𝒆𝒋𝝎𝟎 𝒕 + 𝒆−𝒋𝝎𝟎𝒕 𝟐𝝅𝜹 𝝎 − 𝝎𝟎 + 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝟏 𝒋𝝎 𝒕 𝑭𝑻 𝟏
𝒆 𝟎 + 𝒆−𝒋𝝎𝟎 𝒕 𝟐𝝅𝜹 𝝎 − 𝝎𝟎 + 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝟐 𝟐
𝑭𝑻
𝐜𝐨𝐬 𝝎𝟎 𝒕 𝝅𝜹 𝝎 − 𝝎𝟎 + 𝝅𝜹(𝝎 + 𝝎𝟎 )
Example: Find the Fourier Transform of the signal,
𝒙 𝒕 = 𝒔𝒊𝒏 𝝎𝟎 𝒕

𝑭𝑻
𝒋𝝎𝟎 𝒕
𝒆 𝟐𝝅𝜹(𝝎 − 𝝎𝟎 )
𝑭𝑻
−𝒋𝝎𝟎 𝒕
𝒆 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝑭𝑻
𝒆𝒋𝝎𝟎 𝒕 − 𝒆−𝒋𝝎𝟎𝒕 𝟐𝝅𝜹 𝝎 − 𝝎𝟎 − 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝟏 𝒋𝝎 𝒕 𝑭𝑻 𝟏
𝒆 𝟎 − 𝒆−𝒋𝝎𝟎 𝒕 𝟐𝝅𝜹 𝝎 − 𝝎𝟎 − 𝟐𝝅𝜹(𝝎 + 𝝎𝟎 )
𝒋𝟐 𝒋𝟐
𝑭𝑻 𝝅 𝝅
𝐬𝐢𝐧 𝝎𝟎 𝒕 𝜹 𝝎 − 𝝎𝟎 − 𝜹(𝝎 + 𝝎𝟎 )
𝒋 𝒋
Example: Find the Fourier Transform of the unit step signal,
𝒖 𝒕
𝑭𝑻 𝟐
𝒔𝒈𝒏(𝒕)
𝒋𝝎
𝑭𝑻
𝟏 𝟐𝝅𝜹 𝝎
𝑭𝑻 𝟐
𝟏 + 𝒔𝒈𝒏 𝒕 𝟐𝝅𝜹 𝝎 +
𝒋𝝎
𝟏 𝑭𝑻 𝟏 𝟐
𝟏 + 𝒔𝒈𝒏 𝒕 𝟐𝝅𝜹 𝝎 +
𝟐 𝟐 𝒋𝝎
𝑭𝑻 𝟏
∴ 𝒖 𝒕 𝝅𝜹 𝝎 +
𝒋𝝎
 Fourier Transform of Periodic Signals
Periodic Signal,
∞

𝑥 𝑡 = ෍ 𝑐𝑛 𝑒 𝑗𝑛𝜔0𝑡
𝑛=−∞

𝑭𝑻
Now, 𝒆𝒋𝝎𝟎 𝒕 𝟐𝝅𝜹 𝝎 − 𝝎𝟎
𝑭𝑻
𝒆𝒋𝒏𝝎𝟎𝒕 𝟐𝝅𝜹 𝝎 − 𝒏𝝎𝟎
𝑭𝑻
𝒄𝒏 𝒆𝒋𝒏𝝎𝟎 𝒕 𝟐𝝅𝒄𝒏 𝜹 𝝎 − 𝒏𝝎𝟎
∞ ∞
𝑭𝑻
෍ 𝒄𝒏 𝒆𝒋𝒏𝝎𝟎 𝒕 ෍ 𝟐𝝅𝒄𝒏 𝜹 𝝎 − 𝒏𝝎𝟎
𝒏=−∞ 𝒏=−∞
∞
𝑭𝑻
𝒙 𝒕 ෍ 𝟐𝝅𝒄𝒏 𝜹 𝝎 − 𝒏𝝎𝟎
𝒏=−∞
Symmetry of Fourier Transform
∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕
−∞
∞

𝑿 −𝝎 = න 𝒙 𝒕 𝒆𝒋𝝎𝒕 𝒅𝒕
−∞
∞

𝑿∗ 𝝎 = න 𝒙 𝒕 𝒆𝒋𝝎𝒕 𝒅𝒕
−∞
𝑿 −𝝎 = 𝑿 𝝎 ∠𝑿 −𝝎 = −∠𝑿 −𝝎
Even Symmetry of Magnitude Spectrum

Valid Magnitude
Spectrum of Real
Signal

NOT Valid
Magnitude
Spectrum for
real signals
Odd Symmetry of Phase Spectrum

Valid Phase
Spectrum of Real
Signal

NOT Valid
Phase
Spectrum for
real signals
 Scaling

𝑭𝑻
𝒙 𝒕 𝑿(𝝎)

𝑭𝑻 𝟏 𝝎
𝒙 𝒂𝒕 𝑿
𝒂 𝒂
 Time-Domain Shifting

 Time delay in a signal causes a linear phase shift in its

spectrum.

 The delayed signal 𝒙 𝒕 − 𝒕𝟎 can be synthesized by the same

sinusoidal components, each delayed by 𝒕𝟎 seconds.
 cos 𝜔 𝑡 − 𝑡0 = cos 𝜔𝑡 − 𝜔𝑡0
𝐴 𝑠𝑖𝑛𝑢𝑠𝑜𝑖𝑑 𝐴 𝑝ℎ𝑎𝑠𝑒 𝑠ℎ𝑖𝑓𝑡
𝑑𝑒𝑙𝑎𝑦𝑒𝑑 𝑏𝑦 𝜔𝑡0
𝑏𝑦 𝑡0 𝑠𝑒𝑐𝑜𝑛𝑑𝑠

To achieve same time delay, higher frequency sinusoids must

undergo proportionally higher phase shifts.
Example: Find the Fourier Transform of the signal,
𝒙 𝒕 = 𝒓𝒆𝒄𝒕 𝒕 − 𝟑|𝟔

𝑭𝑻 𝝉𝝎
𝒓𝒆𝒄𝒕(𝒕|𝝉) 𝛕 𝑺𝒂
𝟐
𝑭𝑻 𝟔𝝎
𝒓𝒆𝒄𝒕(𝒕|𝟔) 𝟔 𝑺𝒂 = 𝟔 𝑺𝒂 𝟑𝝎
𝟐
𝑭𝑻
𝒓𝒆𝒄𝒕(𝒕 − 𝟑|𝟔) 𝟔 𝑺𝒂 𝟑𝝎 𝒆−𝒋𝟑𝝎
 Magnitude
Spectrum

Phase Spectrum
Time Domain
Signal
Example: Find the Fourier Transform of the signal,
𝐟 𝒕 = 𝒆−𝒂 𝒕−𝒕𝟎 , 𝒂 is real and 𝒂 > 𝟎
𝑭𝑻 𝟐𝒂
𝒆−𝒂 𝒕
𝒂𝟐 + 𝝎 𝟐

𝑭𝑻 𝟐𝒂
𝒆−𝒂 𝒕−𝒕𝟎 𝒆 −𝒋𝝎𝒕𝟎
𝒂𝟐 + 𝝎 𝟐
 Reflection

𝑭𝑻
𝒙 𝒕 𝑿(𝝎)

𝑭𝑻
𝒙 −𝒕 𝑿∗ (𝝎)
 Practice Problem
1. Determine the Fourier transforms of the signal 𝑓 𝑡 , 𝑥(𝑡) and 𝑣(𝑡) in the
Figure using the properties of Fourier transform. Given
𝝎𝝉
𝑭𝑻 𝐬𝐢𝐧 𝟐
that,𝒓𝒆𝒄𝒕 𝒕 𝝉 𝝉 𝝎𝝉
𝟐
Example: The Fourier Transform of the signal, 𝒙(𝒕) is
𝝎𝟐 +𝒋𝟒𝝎+𝟐
𝑿 𝝎 =
−𝝎𝟐 +𝒋𝟒𝝎+𝟑
Find the Fourier Transforms of 𝒙 −𝟐𝒕 + 𝟏 .
𝟏
𝒙 −𝟐𝒕 + 𝟏 = 𝒙 −𝟐 𝒕 −
𝟐
𝑭𝑻
𝒙(𝒕) 𝑿(𝝎)
𝑭𝑻
𝒙(−𝒕) 𝑿∗ (𝝎)
𝑭𝑻 𝟏 ∗ 𝝎
𝒙(−𝟐𝒕) 𝑿( )
𝟐 𝟐
𝟏 𝑭𝑻 𝟏 ∗ 𝝎 −𝑱𝝎𝟏
𝒙 −𝟐 𝒕 − 𝑿 ( )𝒆 𝟐
𝟐 𝟐 𝟐
Multiplication with a sinusoid (Modulation) /
Frequency-Domain Shifting
 Multiplication with a sinusoid (Modulation) /
Frequency-Domain Shifting
• Message Signal • Carrier Wave
 Multiplication with a sinusoid (Modulation) /
Frequency-Domain Shifting
• Amplitude Modulated Signal
 Modulation with a Cosine Wave
𝑭𝑻
𝒙(𝒕) 𝑿(𝝎)
𝑭𝑻
𝒋𝝎𝟎 𝒕
𝒙(𝒕)𝒆 𝑿(𝝎 − 𝝎𝟎 )
𝑭𝑻
𝒙(𝒕)𝒆−𝒋𝝎𝟎𝒕 𝑿(𝝎 + 𝝎𝟎 )
𝑭𝑻
𝒋𝝎𝟎 𝒕 −𝒋𝝎𝟎 𝒕
𝒙(𝒕)𝒆 + 𝒙(𝒕)𝒆 𝑿 𝝎 − 𝝎𝟎 + 𝑿(𝝎 + 𝝎𝟎 )
𝟏 𝑭𝑻 𝟏
𝒙 𝒕 𝒆𝒋𝝎𝟎 𝒕 + 𝒆−𝒋𝝎𝟎 𝒕 𝑿 𝝎 − 𝝎𝟎 + 𝑿(𝝎 + 𝝎𝟎 )
𝟐 𝟐
𝑭𝑻 𝟏
𝒙(𝒕) 𝐜𝐨𝐬 𝝎𝟎 𝒕 𝑿 𝝎 − 𝝎𝟎 + 𝑿(𝝎 + 𝝎𝟎 )
𝟐
 Modulation with a Sine Wave
𝑭𝑻
𝒙(𝒕) 𝑿(𝝎)
𝑭𝑻
𝒋𝝎𝟎 𝒕
𝒙(𝒕)𝒆 𝑿(𝝎 − 𝝎𝟎 )
𝑭𝑻
𝒙(𝒕)𝒆−𝒋𝝎𝟎𝒕 𝑿(𝝎 + 𝝎𝟎 )
𝑭𝑻
𝒋𝝎𝟎 𝒕 −𝒋𝝎𝟎 𝒕
𝒙(𝒕)𝒆 −𝒙 𝒕 𝒆 𝑿 𝝎 − 𝝎𝟎 − 𝑿(𝝎 + 𝝎𝟎 )
𝟏 𝑭𝑻 𝟏
𝒙 𝒕 𝒆𝒋𝝎𝟎 𝒕 − 𝒆−𝒋𝝎𝟎 𝒕 𝑿 𝝎 − 𝝎𝟎 − 𝑿(𝝎 + 𝝎𝟎 )
𝟐𝒋 𝟐𝒋
𝑭𝑻 𝟏
𝒙(𝒕) 𝐬𝐢𝐧 𝝎𝟎 𝒕 𝑿 𝝎 − 𝝎𝟎 − 𝑿(𝝎 + 𝝎𝟎 )
𝟐𝒋
• In the generation of communication signals, often two signals
such as 𝒎 𝒕 = 𝟐𝐜𝐨𝐬(𝟐𝟎𝟎𝝅𝒕) and 𝒄 𝒕 = 𝟓𝐜𝐨𝐬(𝟏𝟎𝟎𝟎𝝅𝒕) are
multiplied together to produce
𝝋 𝒕 = 𝟏𝟎𝒄𝒐𝒔(𝟐𝟎𝟎𝝅𝒕) 𝒄𝒐𝒔(𝟏𝟎𝟎𝟎𝝅𝒕)
Sketch the spectra of 𝒎 𝒕 , 𝒄 𝒕 & 𝝋(𝒕).
Differentiation – High Pass Filtering
 𝟐
Example: The Fourier transform of 𝒔𝒈𝒏(𝒕) is . Find the
𝒋𝝎
Fourier Transform of 𝜹(𝒕).

𝑭𝑻 𝟐
𝒔𝒈𝒏(𝒕)
𝒋𝝎
𝒅 𝑭𝑻 𝟐
𝒔𝒈𝒏(𝒕) 𝒋𝝎.
𝒅𝒕 𝒋𝝎
𝑭𝑻
𝟐𝜹 𝒕 𝟐
𝑭𝑻
𝜹 𝒕 𝟏
 Integration – Low Pass Filtering
𝑭𝑻
𝒙 𝒕 𝑿(𝝎)
𝒕
𝑿(𝝎)
න 𝒙 𝝉 𝒅𝝉 = + 𝝅 𝑿(𝟎)𝜹(𝝎)
𝒋𝝎
−∞
If 𝒙(𝒕) has no DC component, then

𝒕
𝑿(𝝎) 𝑿(𝝎)
න 𝒙 𝝉 𝒅𝝉 = = ∠ −𝟗𝟎𝟎
𝒋𝝎 𝝎
−∞
Example: Find the Fourier Transform of 𝒙(𝒕) in the following
figure. Then use the result to find the Fourier Transform of
𝒚 𝒕 .
 𝒂 𝒂
𝒙 𝒕 = 𝑨 𝒓𝒆𝒄𝒕 𝒕 + |𝒂 − 𝒓𝒆𝒄𝒕 𝒕 − |𝒂
𝟐 𝟐
Now,
𝑭𝑻 𝝉𝝎
𝒓𝒆𝒄𝒕(𝒕|𝝉) 𝛕 𝑺𝒂
𝟐

𝑭𝑻 𝒂𝝎
𝒓𝒆𝒄𝒕(𝒕|𝒂) 𝒂 𝑺𝒂
𝟐

𝒂 𝑭𝑻 𝒂𝝎 𝒋𝝎𝒂
𝒓𝒆𝒄𝒕 𝒕 + |𝒂 𝒂 𝑺𝒂 𝒆 𝟐
𝟐 𝟐

𝒂 𝑭𝑻 𝒂𝝎 −𝒋𝝎𝒂
𝒓𝒆𝒄𝒕 𝒕 − |𝒂 𝒂 𝑺𝒂 𝒆 𝟐
𝟐 𝟐
 𝒂 𝒂 𝑭𝑻 𝒂𝝎 𝒂
𝒋𝝎𝟐
𝒂
−𝒋𝝎𝟐
𝒓𝒆𝒄𝒕 𝒕 + |𝒂 − 𝒓𝒆𝒄𝒕 𝒕 − |𝒂 𝒂 𝑺𝒂 𝒆 −𝒆
𝟐 𝟐 𝟐

𝒂
𝑨 ቂ𝒓𝒆𝒄𝒕 𝒕 + |𝒂
𝟐
 𝒂
𝑨 ቂ𝒓𝒆𝒄𝒕 𝒕 + |𝒂
𝟐
 𝑭𝑻 𝟒𝑨 𝟐
𝒂𝝎
𝒙(𝒕) 𝒋 𝒔𝒊𝒏
𝝎 𝟐

𝑭𝑻 𝟏 𝟒𝑨 𝟐
𝒂𝝎
𝐳 𝒕 .𝒋 𝒔𝒊𝒏
𝒋𝝎 𝝎 𝟐

𝟐 𝒂𝝎
𝑭𝑻 𝒔𝒊𝒏
𝐳 𝒕 𝟐
𝑨. 𝒂 . 𝟐
𝒂𝝎 𝟐
𝟐
 𝟐 𝒂𝝎
𝟏 𝑭𝑻 𝒔𝒊𝒏
𝐳 𝒕 𝒂. 𝑨 𝟐
𝒂 𝒂𝝎
𝟐

𝟐 𝒂𝝎
𝑭𝑻 𝒔𝒊𝒏 𝒂𝝎 𝒂𝝎
𝒚 𝒕 𝒂. 𝑨 𝟐 = 𝒂. 𝑨𝑺𝒂𝟐
= 𝒂𝑨𝒔𝒊𝒏𝒄𝟐
𝒂𝝎 𝟐 𝟐 𝟐
𝟐
Convolution
Proof
Example: Find the convolution integral of the pair of the
signal:
𝒙 𝒕 = 𝒆−𝟐𝒕 𝒖(𝒕)
𝒉 𝒕 = 𝒆−𝒕 𝒖(𝒕)

−𝟐𝒕
𝑭𝑻 𝟏
𝒙 𝒕 =𝒆 𝒖 𝒕 𝑿 𝝎 =
𝟐 + 𝒋𝝎
𝑭𝑻 𝟏
𝒉 𝒕 = 𝒆−𝒕 𝒖 𝒕 𝑯 𝝎 =
𝟏 + 𝒋𝝎
𝟏 𝟏 𝟏
𝐘 𝝎 =𝑿 𝝎 𝑯 𝝎 = . =
𝟐 + 𝒋𝝎 𝟏 + 𝒋𝝎 𝟐 + 𝒋𝝎 𝟏 + 𝒋𝝎
 𝟏 𝟏 𝟏
𝐘 𝝎 = = −
𝟐 + 𝒋𝝎 𝟏 + 𝒋𝝎 𝟏 + 𝒋𝝎 𝟐 + 𝒋𝝎

𝒚 𝒕 = 𝒆−𝒕 𝒖 𝒕 − 𝒆−𝟐𝒕 𝒖 𝒕 = {𝒆−𝒕 −𝒆−𝟐𝒕 }𝒖 𝒕

A New Way to Analyze LTI System
Another Technique for finding Impulse Response ℎ(𝑡)

FT
𝑥 𝑡 X(ω)
FT
𝑦 𝑡 Y(ω)

𝑌(𝜔)
𝐻 𝜔 =
𝑋(𝜔)

𝐼𝐹𝑇
𝐻 𝜔 ℎ(𝑡)
 Example: Distortion-less Transmission

ℎ(𝑡)

Here, Transmitted signal, 𝑥 𝑡 𝑖𝑛𝑝𝑢𝑡

Received signal, y 𝑡 (𝑜𝑢𝑡𝑝𝑢𝑡)
Transmission channels, ℎ(𝑡) (system impulse response)
For Distortion-less Transmission,
𝑦 𝑡 = 𝐴𝑥 𝑡 − 𝑡0
∴ 𝐶ℎ𝑎𝑛𝑛𝑒𝑙 𝐼𝑚𝑝𝑢𝑙𝑠𝑒 𝑅𝑒𝑠𝑝𝑜𝑛𝑠𝑒,
ℎ 𝑡 = 𝐴𝛿 𝑡 − 𝑡0
Channel Frequency Response,
∞ ∞

𝑯 𝝎 = න 𝒉 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 = න 𝑨𝜹 𝒕 − 𝒕𝟎 𝒆−𝒋𝝎𝒕 𝒅𝒕 = 𝑨𝒆−𝒋𝝎𝒕𝟎

−∞ −∞
For distortion-less transmission :
Attenuation (𝜶) will be
constant for all frequencies
Phase (𝜷) should
increase linearly with
frequencies.
 Practice Problem
1. A Hilbert transformer (ideal phase shifter) is a linear time-
invariant (LTI) system with the frequency response:
𝒋𝝅ൗ𝟐
𝒆 , 𝝎<𝟎
𝑯 𝝎 = ൝ −𝒋𝝅
𝒆 ൗ𝟐 , 𝝎>𝟎
i. Sketch the magnitude spectrum and the phase spectrum of the
system function.
ii. Find the output of the system for the inputs 𝐜𝐨𝐬 𝝎𝟎 𝒕 and
𝐬𝐢𝐧 𝝎𝟎 𝒕 using frequency domain analysis.
Multiplication of two signals
Duality
 Fourier Transform Pairs

x(t) X(ω)

ω
t

Frequency
Time Domain
Domain
 Fourier Transform Pairs

x(t) X(ωX(ω
) )

ω
t
t

Frequency
Time Domain
Domain
Fourier Transform Pairs
 𝝎𝑩 𝒕 𝝎𝑩 𝒕
Find the Fourier Transform of 𝒙 𝒕 = 𝑺𝒂 = 𝒔𝒊𝒏𝒄 .
𝟐 𝟐𝝅

𝑭𝑻 𝝉𝝎
𝒓𝒆𝒄𝒕(𝒕|𝝉) 𝛕 𝑺𝒂
𝟐

𝝉𝒕 𝑭𝑻
𝛕 𝑺𝒂 𝟐𝝅 𝒓𝒆𝒄𝒕(−𝝎|𝝉) (𝑫𝒖𝒂𝒍𝒊𝒕𝒚)
𝟐

𝝉𝒕 𝑭𝑻
𝛕 𝑺𝒂 𝟐𝝅 𝒓𝒆𝒄𝒕(𝝎|𝝉)
𝟐
 𝝎𝑩 𝒕 𝑭𝑻
𝝎𝑩 𝑺𝒂 𝟐𝝅 𝒓𝒆𝒄𝒕(𝝎|𝝎𝑩 )
𝟐
𝝎𝑩 𝒕 𝑭𝑻 𝟐𝝅
𝑺𝒂 𝒓𝒆𝒄𝒕(𝝎|𝝎𝑩 )
𝟐 𝝎𝑩
Fourier Transform Pairs

An Impulse function has all

frequency components with
same Magnitude
Fourier Transform Pairs

DC Signal has 0
frequency only
 Fourier Transform Pairs

𝑭𝑻 𝑭𝑻
𝒙 𝒕 𝑿(𝝎) 𝒙 𝒕 𝑿(𝝎)

𝒅𝒙(𝒕) 𝑭𝑻 𝟎 𝑭𝑻 𝒅𝑿(𝝎)
𝒋𝝎𝑿(𝝎) = 𝝎𝑿(𝝎) 𝒆𝒋𝟗𝟎 −𝒋𝒕 𝒙 𝒕
𝒅𝒕 𝒅𝝎

𝒅𝒏 𝒙(𝒕) 𝑭𝑻 𝟎 𝑭𝑻 𝒅𝒏 𝑿(𝝎)
(𝒋𝝎)𝒏 𝑿(𝝎) = 𝝎𝒏 𝑿(𝝎) 𝒆𝒋𝒏𝟗𝟎 𝒏
𝒅𝒕𝒏 −𝒋𝒕 𝒙 𝒕
𝒅𝝎𝒏
 Practice Problem
1. Sketch the spectra of 𝑚 𝑡 , 𝑐 𝑡 , 𝜑 𝑡 and 𝑧 𝑡 considering the
block diagram in the figure. Also, write the expression of 𝑧(𝑡) in
closed form.
Given that,

𝒎 𝒕 = 𝐜𝐨𝐬(𝟐𝟎𝝅𝒕) + 𝐜𝐨𝐬 𝟑𝟎𝝅𝒕

𝒄 𝒕 = 𝟐 𝐜𝐨𝐬(𝟏𝟎𝟎𝝅𝒕) + 𝟒 𝐜𝐨𝐬(𝟐𝟎𝟎𝝅𝒕)

𝐬𝐢𝐧 𝟏𝟐𝟓𝝅𝒕 − 𝐬𝐢𝐧(𝟐𝟐𝟓𝝅𝒕)

𝒉 𝒕 =
𝝅𝒕
2. Draw the magnitude spectra and find the impulse responses of
the following filters:
i. An ideal low-pass filter with cut-off frequency 1 𝑘𝐻𝑍.
ii. An ideal high-pass filter with cut-off frequency 3 𝑘𝐻𝑍.
iii. An ideal band-pass filter with lower cut-off frequency 1 𝑘𝐻𝑍
and upper cut-off frequency 3 𝑘𝐻𝑍.
iv. An ideal band-stop filter with lower cut-off frequency 1 𝑘𝐻𝑍
and upper cut-off frequency 3 𝑘𝐻𝑍.
Given that,
𝐬𝐢𝐧 𝝎𝑩 𝒕 𝑭𝑻
𝒓𝒆𝒄𝒕 𝝎 𝟐𝝎𝑩
𝝅𝒕
3. Voice inversion scrambling is an analog method of obscuring the content of a
transmission. 𝑚(𝑡) is a voice signal which has been scrambled using the block
diagram in the figure. The spectrum of 𝑚(𝑡) is given in the figure. 𝑦(𝑡) is the
scrambled version of 𝑚(𝑡). The useful bandwidth of the voice signal is 𝐵 𝑟𝑎𝑑/𝑠.
𝒔𝒊𝒏 𝑩𝒕
Given that, 𝒉 𝒕 =
𝝅𝒕

i. Draw the spectrum of the

scrambled signal 𝑦 𝑡 .
ii. How can you get back the original
voice signal 𝑚(𝑡) from 𝑦 𝑡 ?
Parseval’s Theorem
Example: Consider the signal, 𝒙 𝒕 = 𝒆−𝒕 𝒖(𝒕). Find the
percentage of the total energy content of the signal that lies in the
𝟐
frequency band 𝟎 𝒕𝒐 𝑯𝒛.
𝝅
Solution:
𝟐
𝟎 𝒕𝒐 𝑯𝒛 ≡ −𝟒 𝒕𝒐 𝟒 𝒓𝒂𝒅/𝒔
𝝅
∞ ∞

𝑿 𝝎 = න 𝒙 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕 = න 𝒆−𝒕 𝒖 𝒕 𝒆−𝒋𝝎𝒕 𝒅𝒕

−∞ −∞
∞ ∞
−𝒕 −𝒋𝝎𝒕 −(𝟏+𝒋𝝎)𝒕
𝟏
=න𝒆 𝒆 𝒅𝒕 = න 𝒆 𝒅𝒕 =
𝟏 + 𝒋𝝎
𝟎 𝟎
 2 ∗
𝟏 𝟏 𝟏
𝑿 𝝎 =𝑿 𝝎 𝑿 𝝎 = . =
𝟏 + 𝒋𝝎 𝟏 − 𝒋𝝎 𝟏 + 𝝎𝟐

Total Energy,
∞ ∞
𝟏 𝟐
𝟏 𝟏 𝟏
𝑬= න 𝑿 𝝎 𝒅𝝎 = න 𝒅 𝝎 =
𝟐𝝅 𝟐𝝅 𝟏 + 𝝎𝟐 𝟐
−∞ −∞
 𝟐
Energy in the frequency band, 𝟎 𝒕𝒐 𝑯𝒛 ,
𝝅

𝟒 𝟒
𝟏 𝟐𝒅 𝝎
𝟏 𝟏
∆𝑬 = න𝑿 𝝎 = න 𝟐
𝒅 𝝎 = 𝟎. 𝟒𝟐𝟐
𝟐𝝅 𝟐𝝅 𝟏 + 𝝎
−𝟒 −𝟒

∆𝑬 𝟎.𝟒𝟐𝟐
= = 𝟎. 𝟖𝟒𝟒 = 𝟖𝟒. 𝟒% lies in the the frequency band,
𝑬 𝟐 𝟎.𝟓
𝟎 𝒕𝒐 𝑯𝒛
𝝅

So, if the signal, 𝒙 𝒕 = 𝒆−𝒕 𝒖(𝒕) is transmitted through a Low-

𝟐
pass transmission channel with bandwidth 𝑯𝒛, then 𝟖𝟒. 𝟒% of
𝝅
its energy content passes through the channel. The rest of the
content of the signal is filtered out.
 Practice Problem
𝐬𝐢𝐧(𝒕) 𝐬𝐢𝐧(𝟑𝒕)
1. Find the energy of the signal 𝒙 𝒕 = + using
𝝅𝒕 𝝅𝒕
𝐬𝐢𝐧 𝝎𝑩 𝒕 𝑭𝑻
Parseval’s theorem. Given that, 𝒓𝒆𝒄𝒕 𝝎 𝟐𝝎𝑩
𝝅𝒕
The Fourier transform of an impulse train is an
impulse train in frequency domain
 The Concept of Bandwidth
Bandwidth of an information signal is the difference between the highest
and the lowest frequency contained in that signal.
The bandwidth of a signal is the amount of "frequency'' required to sustain
the signal unmodified.
 𝟐𝝅𝑩
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟐𝝅𝑩 𝒓𝒂𝒅𝒊𝒂𝒏ൗ𝒔 = = 𝑩 𝑯𝒁
𝟐𝝅
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝑩 𝑯𝒁
 SIGNAL BANDWIDTH

𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟏𝟓𝟎 − 𝟓𝟎 = 𝟏𝟎𝟎 𝑯𝒛

 SIGNAL BANDWIDTH

𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟐𝟓𝟎 − 𝟎 = 𝟐𝟓𝟎 𝑯𝒛

𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟏𝟕 x 𝟓𝟎 − 𝟎 = 𝟖𝟓𝟎 𝑯𝒛
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟗𝟗 x 𝟓𝟎 − 𝟎 = 𝟒𝟗𝟓𝟎 𝑯𝒛
So, a Square Wave has Infinite Bandwidth
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = ∞ − 𝟎 = ∞ 𝑯𝒛
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = 𝟐𝒌𝑯𝒛 − 𝟐𝒌𝑯𝒛 = 𝟎 𝑯𝒛
An everlasting signal (𝒇𝒓𝒐𝒎
−𝒕 = ∞ 𝒕𝒐 𝒕 = +∞) can have 0 (Zero),
Finite or Infinite Bandwidth.
𝑺𝒊𝒈𝒏𝒂𝒍 𝑩𝒂𝒏𝒅𝒘𝒊𝒅𝒕𝒉 = ∞ − 𝟎 = ∞ 𝑯𝒛
Time Limited Signal = Everlasting Signal
× Rectangular Window

Multiplication Convolution
with with ‘sinc’
function
Rectangular which makes
Window bandwidth
infinite
The Bandwidth of a time-limited
signal is “theoretically” Infinite
A time-limited signal must not be band-limited.

A band-limited signal must not be time-limited.

A signal can be infinite in both time and frequency domain

This is one of the form of the famous uncertainty principle.

 Practice Problems
1. Find the absolute (theoretical) bandwidth (in 𝐻𝑒𝑟𝑧) of the
following signals:
i. 𝑥1 𝑡 = 2 + 𝑐𝑜𝑠(2000𝜋𝑡) + 𝑠𝑖𝑛(6000𝜋𝑡)
ii. 𝑥2 𝑡 = 𝑢 𝑡 − 𝑢(𝑡 − 3)
iii. 𝑥3 𝑡 = 7 𝑐𝑜𝑠 200𝜋𝑡
3𝜋
𝑐𝑜𝑠 1000𝜋𝑡 , 0 < 𝑡 <
iv. 𝑥4 𝑡 = ቐ 2
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
 Application of Fourier Transform in
Telecommunication
• In Analysis of

 Filtering
 Amplitude Modulation (AM)
 Frequency Division Multiplexing (FDM)
 Sampling
Amplitude Modulation (AM)
 Usefulness of AM

• Transmitting Information Efficiently

• Overcoming Hardware Limitations
• Reducing Limitations
• Utilizing the Electromagnetic Spectrum
Efficiency
 What is Amplitude Modulation ?

• Amplitude Modulation is a process where the

amplitude of a carrier signal is altered
according to information in a message signal.
• The frequency of the carrier signal is usually
much greater than the highest frequency of the
input message signal.
 Technique of Amplitude Modulation
AM is formally defined as a process in which the amplitude of the
carrier wave c(t) is varied about a mean value linearly with a message
signal m(t).

Message Signal: 𝒎 𝒕

Sinusoidal Carrier: 𝒄 𝒕 = 𝒄𝒐𝒔(𝝎𝟎 𝒕)

AM Wave: 𝒚 𝒕 = 𝒎 𝒕 × 𝒄 𝒕 = 𝒎 𝒕 𝐜𝐨𝐬(𝝎𝟎 𝒕)
• 𝑀𝑒𝑠𝑠𝑎𝑔𝑒 𝑆𝑖𝑔𝑛𝑎𝑙 𝑚(𝑡) • 𝐶𝑎𝑟𝑟𝑖𝑒𝑟 𝑊𝑎𝑣𝑒 c(t)
• 𝐴𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒 𝑀𝑜𝑑𝑢𝑙𝑎𝑡𝑒𝑑 𝑆𝑖𝑔𝑛𝑎𝑙 𝑦(𝑡)
 Block Diagram of Amplitude Modulation

𝑚 𝑡 y 𝑡 y 𝑡 z 𝑡 Low-
Transmission m 𝑡
X Channel X Pass
Filter

𝑐 𝑡 𝑐 𝑡

Modulation Demodulation
Amplitude Modulation in Frequency Domain:
𝑭𝑻
𝒎 𝒕 𝑴(𝝎)

𝑭𝑻
𝒄 𝒕 𝑪 𝝎 = 𝝅 [𝜹 𝝎 − 𝝎𝟎 + 𝜹(𝝎 + 𝝎𝟎 )]

𝟏
𝒀 𝝎 = 𝑴 𝝎 ∗ 𝝅 [𝜹 𝝎 − 𝝎𝟎 + 𝜹(𝝎 + 𝝎𝟎 )]
𝟐𝝅

𝟏
𝒀 𝝎 = [𝑴 𝝎 − 𝝎𝟎 + 𝑴(𝝎 + 𝝎𝟎 )]
𝟐
Amplitude Modulation in Frequency Domain:
 Demodulation in Frequency Domain:
𝐳 𝐭 = 𝐲 𝐭 × 𝐜 𝐭 = 𝐲 𝐭 𝐜𝐨𝐬 𝛚𝟎 𝐭

𝟏
∴𝐙 𝛚 = [𝐘 𝛚 ∗ 𝐂(𝛚)]
𝟐𝛑
𝟏
= 𝐘 𝛚 ∗ 𝛑 [𝛅 𝛚 − 𝛚𝟎 + 𝛅(𝛚 + 𝛚𝟎 )]
𝟐𝛑
𝟏
= 𝐘 𝛚 − 𝛚𝟎 + 𝐘 𝛚 + 𝛚𝟎
𝟐
𝟏 𝟏 𝟏
= 𝐌 𝛚 + 𝐌 𝛚 − 𝟐𝛚𝟎 + 𝐌 𝛚 + 𝟐𝛚𝟎
𝟐 𝟒 𝟒

After Filtering Simple Amplifying Output: 𝐌 𝛚

Demodulation in Frequency Domain:
 Modulation & Demodulation in Time Domain
Modulation:
𝒚 𝒕 = 𝒎 𝒕 × 𝒄 𝒕 = 𝒎 𝒕 𝒄𝒐𝒔(𝝎𝟎 𝒕)
Demodulation :
𝒛 𝒕 =𝒚 𝒕 ×𝒄 𝒕
= 𝒎 𝒕 𝒄𝒐𝒔𝟐 (𝝎𝟎 𝒕)
𝟏
= 𝒎(𝒕) (𝟏 + 𝒄𝒐𝒔(𝟐𝝎𝟎 𝒕))
𝟐
𝟏 𝟏
= 𝒎 𝒕+ 𝒎(𝒕)𝒄𝒐𝒔(𝟐𝝎𝟎 𝒕)
𝟐 𝟐
Low pass filtering: Only Message Signal 𝒎 𝒕 remains which is
desired
 Frequency Division Multiplexing (FDM)
Suppose, we want to send several messages at the same
time within the same transmission channel.
How?

Let, Message Signals: 𝒙𝟏 𝒕 , 𝒙𝟐 𝒕 , 𝒙𝟑 𝒕

How can we send 𝒙𝟏 𝒕 , 𝒙𝟐 𝒕 , 𝒙𝟑 𝒕 at the same

time within the same transmission channel.
 Do all the signals have Fourier Transform ?
∞
න 𝑒 𝑎𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = ∞ , (𝑎 > 0)
−∞
∞ ∞ ∞
න 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 ≤ න 𝑥 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = න 𝑥 𝑡 𝑑𝑡
−∞ −∞ −∞

∞
If,
‫׬‬−∞ 𝑥 𝑡 𝑑𝑡 = ∞
Then,
𝑥 𝑡 has no Fourier Transform