Signal and System

Fourier Transform
Fourier Transform
› Representing aperiodic signal
Fourier Transform
An aperiodic signal can be thought of as
periodic with infinite period.
Let 𝑥(𝑡) represent an aperiodic signal.

Periodic extension: 𝑥𝑇 𝑡 = ∞
𝑘=−∞ 𝑥(𝑡 + 𝑘𝑇)

Then 𝑥 𝑡 = lim 𝑥𝑇 (𝑡)

𝑇→∞
Fourier Transform
Represent 𝑥𝑇 𝑡 by its Fourier series.
Fourier Transform
Doubling periods doubles # of harmonics
in a given frequency interval.
Fourier Transform
As 𝑇 → ∞, discrete harmonics amplitudes 
a continuum 𝐸(𝜔).
Fourier Transform
As 𝑇 → ∞, synthesis sum → integral
Fourier Transform
Replacing 𝐸 𝜔 by 𝑋(𝑗𝜔) yields the Fourier
transform relations.
𝐸 𝜔 = 𝑋(𝑗𝜔)
Fourier transform

(“analysis” equation)

(“synthesis” equation)

Form is similar to that of Fourier series

provides alternate view of signal.
Relation between FT and LT
If the Laplace transform of a signal exist and
if the ROC includes the 𝑗𝜔 axis, then the
Fourier transform is equal to the Laplace
transform evaluated on the 𝑗𝜔 axis.

Laplace transform:

Fourier transform
Relation between FT and LT
Fourier transform inherits properties of
Laplace transform.
Relation between FT and LT
There are also important differences.
Compare FT and LT of 𝑥 𝑡 = 𝑒 −𝑡 𝑢(𝑡).

Laplace transform

a complex-valued function of complex domain

Fourier transform

a complex-valued function of real domain

Laplace Transform
The Laplace transform maps a function of
time 𝑡 to a complex-valued function of
complex-valued domain 𝑠.
 Fourier Transform
The Fourier transform maps a function of
time 𝑡 to a complex-valued function of real-
valued domain 𝜔.

Frequency plots provide intuition that is difficult to otherwise obtain.

Check Yourself
Fourier Transform
Compare the LT and FT of a square pulse.

Laplace transform
function of 𝑠 = 𝜎 + 𝑗𝜔

Fourier transform
function of 𝜔
Check Yourself
Laplace Transform
Laplace transform: complex-valued function
of complex domain.
Fourier Transform
The Fourier transform is a function of real
domain: frequency 𝜔.
Time representation (domain):

Frequency representation (domain):

Check Yourself
Check Yourself
Check Yourself
Fourier Transform
Stretching time compresses frequency.
Check Yourself
Check Yourself
Stretching time compresses frequency.
Find a general scaling rule.
Let 𝑥2 𝑡 = 𝑥1 (𝑎𝑡).

If time is stretched in from 𝑥1 to 𝑥2 , is 𝑎 > 1

or 𝑎 < 1?
𝑥2 2 = 𝑥1 1
𝑥2 𝑡 = 𝑥1 𝑎𝑡

Therefore 𝑎 = 1 2, or more generally, 𝑎 < 1.

Check Yourself
Fourier Transform
Find a general scaling rule.
Let 𝑥2 𝑡 = 𝑥1 𝑎𝑡 .

Let 𝜏 = 𝑎𝑡, (𝑎 > 0).

If 𝑎 <0 the sign of 𝑑𝜏 would change along with the limit

of integration. In general,

If time is stretched 0 < 𝑎 < 1 then frequency is

compressed and amplitude increases (preserving area).
Moment
The value of 𝑋(𝑗𝜔) at 𝜔 = 0 is the integral
of 𝑥(𝑡) over time 𝑡.
Moment
The value of 𝑥(0) is the integrals of 𝑋(𝑗𝜔)
divided by 2𝜋.
Moment
The value of 𝑥(0) is the integrals of 𝑋(𝑗𝜔)
divided by 2𝜋.
Stretching to the Limit
Stretching time compresses frequency and
increase amplitude (preserving area).

New way to think about an impulse!

Fourier Transform
One of the most useful features of the
Fourier transform (and Fourier series) is the
simple “inverse” Fourier transform.

(Fourier transform)

(Inverse FT)
Inverse Fourier Transform
Find the impulse response of an “ideal” low
pass filter.

This result is not so easily obtained without

inverse relation.
Fourier Transform
The Fourier transform and its inverse have
very similar forms.

(Fourier transform)

(Inverse FT)

Convert one to the other by

• 𝑡→𝜔
• 𝜔→𝑡
• scale by 2𝜋
Duality
The Fourier transform and its inverse have
very similar forms.
Two differences:
• minus sign: flips time axis (or equivalently,
frequency axis)
• divide by 2𝜋 (or multiply in the other direction)
Fourier Transform Pair
∞ ∞
1
𝑥 𝑡 = 𝑋 𝑗𝜔 𝑒 𝑗𝜔𝑡 𝑑𝜔 𝑋(𝑗𝜔) = 𝑥 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡
2𝜋
−∞ −∞

Sufficient conditions for the existence of the FT are (Dirichlet

conditions):
1. On any finite interval 𝑓 𝑡 (a) is bounded; (b) has a finite
number of maxima and minima; and has a finite number of
discontinuities.
∞
2. Absolutely integrable; that is −∞
𝑓(𝑡) 𝑑𝑡 < ∞

Note that these are sufficient condition and not necessary

conditions.
You can have a function that is not absolutely integrable however
it has Fourier transform like cos(𝜔𝑡).
Fourier Transform Pair
∞ ∞
1
𝑥 𝑡 = 𝑋 𝑗𝜔 𝑒 𝑗𝜔𝑡 𝑑𝜔 𝑋(𝑗𝜔) = 𝑥 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡
2𝜋
−∞ −∞
∞
Absolutely integrable; that is −∞
𝑓(𝑡) 𝑑𝑡 < ∞

Example of function that is not absolutely integrable:

𝑒 −𝑡 , cos 𝜔𝑡 , sin 𝜔𝑡 , 𝑢(𝑡)

𝑒 −𝑡 does not have FT however 𝑒 −𝑡 𝑢(𝑡) does have and

cos 𝜔𝑡 , sin 𝜔𝑡 , 𝑢(𝑡) has FT.
Finding the FT
Exponential
∞
Pulse: 𝑥 𝑡 = 𝑒 −𝑎𝑡 𝑢 𝑡 𝑎>0
𝑋 𝑗𝜔 = 𝑒 −𝑎𝑡 𝑢(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡
−∞∞
−(𝑎+𝑗𝜔)𝑡
1 ∞ 1
= 𝑒 𝑑𝑡 = 𝑒− 𝑎+𝑗𝜔 𝑡
=
−(𝑎 + 𝑗𝜔) 0 (𝑎 + 𝑗𝜔)
0

a 1

1
𝑥 𝑡 = 𝑒 −𝑡 𝑢 𝑡 𝑎>0 ↔ 𝑋 𝑗𝜔 =
(𝑎 + 𝑗𝜔)
FT Pairs
1
𝑥 𝑡 = 𝑒 −𝑡 𝑢 𝑡 𝑎>0 ↔ 𝑋 𝑗𝜔 =
(𝑎 + 𝑗𝜔)
 Example FT Pairs
Find the FT for the following function.

 1  jt 1
e j(1) e j(1)
X a ()  xa (t)e  jt dt  e  jt dt  e


1
 j

 j
1

  2 sin()  2 sin()
j  j
 e  e  2
   e j
 e  j

j    j2   

 2sinc()
Example FT Pairs

 X a ()  2sinc()

X a ()
2


2   2
 Example FT Pairs

Find the Fourier Transform

for the following function

 j (2)  j (2)
 2
e
 jt 2
e e



X ()  x(t)e  jt dt  e  jt dt 
2
 j

 j
2

 4 sin(2)  4 sin(2)
j 2  j 2
e e 
 4  e
j 2
 e
 j 2

   2  2
j  2  j 2  

 4sinc(2)
Example FT Pairs

2sinc()

4sinc(2)

A(2T )sinc(T)
Example FT Pairs
Definition of 𝑟𝑒𝑐𝑡 𝑡
𝑇

𝑇 𝑇
𝑢 𝑡+ −𝑢 𝑡−
2 2
Example FT Pairs
Example FT Pairs
Find the FT for the delta function 𝑥 𝑡 = 𝛿(𝑡)

  


X ()  x(t)e

dt   (t)e

dt  (1) (t)dt 1
 jt  jt

  

𝑒 −𝑗𝜔𝑡 𝑡=0
𝛿(𝑡)

𝛿 𝑡 ↔ 1
Properties of Fourier Transform
(1)Linearity:
F[a1x1(t)  a1x1(t) ]  a1X1() + a2 X 2()
Proof:



F[a1x1(t)  a1x1(t) ]  [a1 x1(t)  a1 x1(t) ]e  jt dt

 

 
 a1 x1(t)e jt dt + a2 x2 (t)e  jt dt
 
 

 
 jt  jt
 a1 x1(t)e dt + a2 x2 (t)e dt
 

 a1 X1() + a2 X 2()
Example FT Properties

Direct Method:



X ()  x(t)e  jt
dt


2 0 2 4

   
 jt  jt  jt  jt
 (1)e dt  (2)e dt + (2)e dt  (1)e dt
4 2 0 2
Direct Method
Example FT Properties

Using FT Properties:

 

X ()  X1()  X 2()

Properties of Fourier Transform
(2)Time-Scaling (compressing or expanding)
1 𝜔
Let 𝑥 𝑡 ↔ 𝑋(𝜔) then 𝑥(𝑎𝑡) ↔ 𝑋 𝑎
𝑎

Proof ∞

Let 𝑎 > 0 𝐹 𝑥(𝑎𝑡) = 𝑥(𝑎𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡

−∞
Change of variable
∞
𝜏
−𝑗𝜔 𝑎 𝑑𝜏
𝜏 = 𝑎𝑡 𝐹 𝑥(𝑎𝑡) = 𝑥 𝜏 𝑒
𝑎
−∞
∞
1 𝜔
−𝑗 𝑎 𝜏
= 𝑥 𝜏 𝑒 𝑑𝜏
𝑎
−∞
Properties of Fourier Transform
∞
1 𝜔
−𝑗 𝑎 𝜏
𝐹 𝑥(𝑎𝑡) = 𝑥 𝜏 𝑒 𝑑𝜏
𝑎
−∞
∞
𝜔 1
Let Ω = 𝐹 𝑥(𝑎𝑡) = 𝑥 𝜏 𝑒 −𝑗Ω𝜏 𝑑𝜏
𝑎 𝑎
−∞

𝑋(Ω)

1 1 𝜔
𝐹 𝑥(𝑎𝑡) = 𝑋 Ω = 𝑋 𝑎
𝑎 𝑎
Properties of Fourier Transform
Now let 𝑎 < 0 → 𝑎𝑡 = − 𝑎 𝑡
∞

𝐹 𝑥(𝑎𝑡) = 𝑥(− 𝑎 𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡

−∞
Change of variable
𝑑𝜏 𝑡 = ∞ → 𝜏 = −∞
𝜏=−𝑎𝑡 𝑑𝑡 =
−𝑎 𝑡 = −∞ → 𝜏 = ∞
∞ ∞
𝜏 𝑑𝜏 1 𝜔
−𝑗𝜔 −𝑗 𝜏
𝐹 𝑥(𝑎𝑡) = 𝑥(𝜏)𝑒 −𝑎 = 𝑥(𝜏)𝑒 −𝑎 𝑑𝜏
−𝑎 −𝑎
−∞ −∞

1 𝜔
𝐹 𝑥(𝑎𝑡) = 𝑋 −𝑎 𝑋 𝜔
−𝑎 −𝑎

1 𝜔
Since 𝑎 < 0 → − 𝑎 = 𝑎 then  𝐹 𝑥(𝑎𝑡) = 𝑋
𝑎 𝑎
Check Yourself

Let X a ()  2sinc()

What is the FT of

since 𝑥(𝑡) = 𝑥𝑎 𝑡
1 2

1     2X (2)
X ()  Xa   a  2(2sinc(2))
1 2 1 2 
 4sinc(2)
Properties of Fourier Transform
(3)Time-Shifting
Let 𝑥 𝑡 ↔ 𝑋 𝜔 then 𝑥 𝑡 − 𝑡0 ↔ 𝑋 𝜔 𝑒 −𝑗𝜔𝑡0
Proof ∞ ∞

Let y 𝑡 = 𝑥 𝑡 − 𝑡0  𝑌 𝜔 = 𝑦 𝑡 𝑒 −𝑗𝜔𝑡 𝑑𝑡 = 𝑥 𝑡 − 𝑡0 𝑒 −𝑗𝜔𝑡 𝑑𝑡

−∞ −∞

Change of variable ∞

𝜏 = 𝑡 − 𝑡0 𝑌 𝜔 = 𝑥 𝜏 𝑒 −𝑗𝜔(𝜏+𝑡0) 𝑑𝑡
−∞
∞ ∞

𝑌 𝜔 = 𝑥 𝜏 𝑒 −𝑗𝜔𝜏 𝑒 −𝑗𝜔𝑡0 𝑑𝑡 = 𝑒 −𝑗𝜔𝑡0 𝑥 𝜏 𝑒 −𝑗𝜔𝜏 𝑑𝑡

−∞ −∞

𝑋(𝜔)

𝑥 𝑡 − 𝑡0 ↔ 𝑋(𝜔)𝑒 −𝑗𝜔𝑡0
Check Yourself x (t )

Find the FT of pulse function

1

t
0 2

From previous example

 X a ()  2sin c(2)

Since 𝑥 𝑡 = 𝑥𝑎 (𝑡 − 1)

 jt0
X ()  X a ()e = 2sinc(2) e
 j (1)

 j
= 2sinc(2) e
Properties of Fourier Transform
(4)Time Transformation
𝜔
1 −𝑗 𝑡
Let 𝑥 𝑡 ↔ 𝑋 𝜔 then 𝑥 𝑎𝑡 − 𝑡0 ↔ 𝑋 𝜔
𝑎
𝑒 𝑎 0
𝑎
Proof ∞

Let 𝑎 > 0 𝐹 𝑥(𝑎𝑡 − 𝑡0 ) = 𝑥(𝑎𝑡 − 𝑡0 )𝑒 −𝑗𝜔𝑡 𝑑𝑡

−∞
Change of variable 𝑑𝜏
𝑡 = 𝜏+𝑡
𝑎
0 𝑑𝑡 =
𝑎
𝜏 = 𝑎𝑡 − 𝑡0 𝑡=∞ → 𝜏=∞
𝑡 = −∞ → 𝜏 = −∞

∞ 𝑡 ∞
−𝑗𝜔 𝑎0
−𝑗𝜔
𝜏+𝑡0 𝑑𝜏 𝑒 𝜏
−𝑗𝜔 𝑎
𝐹 𝑥(𝑎𝑡 − 𝑡0 ) = 𝑥(𝜏)𝑒 𝑎 = 𝑥(𝜏)𝑒 𝑑𝜏
𝑎 𝑎
−∞ −∞
 Properties of Fourier Transform
∞ 𝑡 ∞
−𝑗𝜔 0
−𝑗𝜔
𝜏+𝑡0 𝑑𝜏 𝑒 𝑎 𝜏
−𝑗𝜔 𝑎
𝐹 𝑥(𝑎𝑡 − 𝑡0 ) = 𝑥(𝜏)𝑒 𝑎 = 𝑥(𝜏)𝑒 𝑑𝜏
𝑎 𝑎
−∞ −∞
𝜔 ∞ 𝜔
−𝑗 𝑡
𝑒 𝑎 0 𝜔
−𝑗 𝑎 𝜏 𝑒
−𝑗 𝑎 𝑡0
= 𝑥(𝜏)𝑒 𝑑𝜏 = 𝑋 𝜔
𝑎 𝑎 𝑎
−∞

𝜔
𝑋 𝑎
𝜔
−𝑗 𝑡
Similarly 𝑎 < 0 𝑒 𝑎 0
𝐹 𝑥(𝑎𝑡 − 𝑡0 ) = 𝑋 −𝜔
𝑎
𝑎

𝜔
1 𝜔 −𝑗 𝑡0
𝑥 𝑎𝑡 − 𝑡0 ↔ 𝑋 𝑎
𝑒 𝑎
𝑎
Check Yourself
Find the FT for Rect function 𝑔(𝑡).

From FT Pairs Table

 T 
Tsinc 
 2 
Check Yourself
1     j  
 t0
x(at t0)  X  e  a 
|a|  a 
 
sinc 
2

1 𝜔 𝜔
𝑎 −𝑗
= 3rect  0.5t 2  𝐺 𝜔 =3 𝑠𝑖𝑛𝑐 𝑒 0.5 (2)
0.5 2

𝐺 𝜔 = 6𝑠𝑖𝑛𝑐 𝜔 𝑒 −𝑗4𝜔
Properties of Fourier Transform
(5)Duality

If 𝑥(𝑡) ↔ 𝑋 𝜔 then 𝑋 𝑡 ↔ 2π𝑥(−𝜔)

Find the FT of 2𝑊𝑠𝑖𝑛𝑐(2𝑊𝑡) 2W

as an exercise!


t
 
2W 2W
Find the FT of 2𝑊𝑠𝑖𝑛𝑐(2𝑊𝑡) .

2W


t
 
2W 2W

2W

 

2W 2W
 2W

 

2W 2W

FT Pairs
 2W


t
 
2W 2W

FT Pairs
 Check Yourself
Find the FT of following
t  t
rect  rect   
   
Since t 
rect  
 
t 
rect    tri 
t
convolution    

Time
 0 
t
  0 
t   0



t
2 2 2 2

Frequency  sinc( f )
multiplication
 sinc( f )   2sinc2( f )
 System Analysis with FT
impulse response

x (t ) h (t ) y(t) = x (t )h (t )


= x ()h (t ) d 
 convolution in time

multiplication in Frequency

X () H () Y ()  X () H ()

y(t) = x (t )h (t ) Y ()  X () H ()

multiplication in Frequency
convolution in time
Properties of Fourier Transform
Let 𝑥(𝑡) ↔ 𝑋(𝜔) then 𝑥 𝑡 𝑒 𝑗𝜔0 𝑡 ↔ 𝑋(𝜔 − 𝜔0 )

Proof
Let 𝑦 𝑡 = 𝑥 𝑡 𝑒 𝑗𝜔0𝑡
  


 j ( 0 )t
dt  x(t)e

Y ()  y(t)e

 jt j0t  jt
dt  x(t)e e dt
  

Change of variable
∞

Ω = 𝜔 − 𝜔0 𝑌 𝜔 = 𝑥 𝑡 𝑒 −𝑗Ω𝑡 𝑑𝑡 = 𝑋(Ω)
−∞
Let 𝑥(𝑡) ↔ 𝑋(𝜔) then 𝑥 𝑡 𝑒 𝑗𝜔0 𝑡 ↔ 𝑋(𝜔 − 𝜔0 )
Similarly 𝑥 𝑡 𝑒 −𝑗𝜔0𝑡 ↔ 𝑋(𝜔 + 𝜔0 )

𝑥 𝑡 𝑒 ±𝑗𝜔0 𝑡 ↔ 𝑋(𝜔 ± 𝜔0 )
Summary
› Learned most of the basic of Fourier
Transform, properties and manipulation
technique to solve problem in continuous
time.
› Next, we focusses on application of FT in
image processing (discrete time).
– Discrete Fourier Transform (DFT)
– Fast Fourier Transform (FFT)