ELEC 221 Lecture 10
The CT Fourier transform properties:
duality, convolution and multiplication
Thursday 06 February 2025
1 / 28
Announcements
Tutorial tomorrow
Assignment 2 due on Feb 7
Tutorial Assignment 2 due on Feb 7
2 / 28
Last time
We saw su!cient conditions for convergence of Fourier transform.
1. Square integrable (finite energy):
Then the Fourier transform converges to x(t) such that error
signal has zero energy.
2. Dirichlet conditions:
Then the Fourier transform converges to
x(t) where it is continuous,
the average of the values on either side at a discontinuity.
3 / 28
Last time
F
→
↑
Image credit: Oppenheim chapter 4.1
! W
1
x̂(t) = X (jε)e jωt dε
2ω →W
4 / 28
Last time
We computed Fourier transforms of periodic signals.
x(t) = e jω0 t ↓ X (jε) = 2ωϑ(ε ↔ ε0 )
↑
" ↑
"
x(t) = ck e jkω0 t ↓ X (jε) = 2ωck ϑ(ε ↔ kε0 )
k=→↑ k=→↑
Image credit: Oppenheim chapter 4.2
5 / 28
Last time: properties of the Fourier transform
We saw some important properties of the Fourier transform:
Linearity
Behaviour under time shift/scale/reverse/conjugation
6 / 28
Today
Learning outcomes:
Describe the duality between time and frequency domains
Apply the convolution property of the Fourier transform to
characterize LTI system behaviour
Describe the multiplication property of the Fourier transform
and provide an application of its use
7 / 28
Time/frequency duality of the FT
We know
F
→
↑
Now let’s consider a signal whose Fourier transform is
#
1 |ε| < W ,
X (jε) =
0 |ε| > W
Image credit: Oppenheim chapter 4.1
8 / 28
Time/frequency duality of the FT
Compute the inverse Fourier transform:
! ↑
1
x(t) = X (jε)e jωt dε
2ω →↑
! W
1
= e jωt dε
2ω →W
1 jωt W
= e |→W
2ωjt
1 $ jWt %
= e ↔ e →jWt
2ωjt
sin(Wt)
=
ωt
9 / 28
Time/frequency duality of the FT
Image credit: Oppenheim chapter 4.1
10 / 28
Time/frequency duality of the FT
Duality: for any transform pair x(t) ↗ X (jε), there is a dual pair
with the time and frequency variables interchanged.
Image credit: Oppenheim chapter 4.3
11 / 28
Time/frequency duality of the FT
&
1, |t| < T1 F 2 sin(εT1 )
x1 (t) = →
↑ X1 (jε) =
0, |t| > T1 ε
&
sin(Wt) F 1, |ε| < W
x2 (t) = →
↑ X2 (jε) =
ωt 0, |ε| > W
F 2
x3 (t) = e →|t| →
↑ X3 (jε) =
1 + ε2
2 F
x4 (jε) = →
↑ X4 (jε) = 2ωe →|ω|
1 + t2
F
x5 (t) = e jω0 t →
↑ X5 (jε) = 2ωϑ(ε ↔ ε0 )
F
x6 (t) = ϑ(t ↔ t0 ) →
↑ X6 (jε) = e →jω0 t
12 / 28
Convolution and the Fourier transform
Recall complex exponentials are eigenfunctions of LTI systems. If
we input signal x(t) into LTI system with impulse response h(t)
↑
"
x(t) = ck e jkωt
k=→↑
"↑
y (t) = ck H(jkε)e jkωt
k=→↑
'↑ →jωt dt
where H(jε) = →↑ h(t)e
This came from the convolution integral:
! ↑
y (t) = x(t ↔ ϖ )h(ϖ )dϖ = x(t) · H(jε)
→↑
13 / 28
Convolution and the Fourier transform
Let’s express x(t) using the inverse Fourier transform:
! ↑
1
x(t) = X (jε)e jωt dε
2ω →↑
and put this into the convolution integral...
! ↑
y (t) = x(t ↔ ϖ )h(ϖ )dϖ
→↑
! ↑( ! ↑ )
1 jω(t→ε )
= X (jε)e dε h(ϖ )dϖ
→↑ 2ω →↑
! ↑( ! ↑ )
1
= X (jε)e dε e →jωε h(ϖ )dϖ
jωt
→↑ 2ω →↑
( ! ↑ )! ↑
1
= jωt
X (jε)e dε e →jωε h(ϖ )dϖ
2ω →↑ →↑
! ↑
1
= X (jε)H(jε)e jωt dε
2ω →↑
14 / 28
Convolution and the Fourier transform
We have two ways to write y (t):
! ↑
1
y (t) = Y (jε)e jωt dε
2ω →↑
! ↑
1
y (t) = X (jε)H(jε)e jωt dε
2ω →↑
This has an important implication:
y (t) = h(t) ↘ x(t)
Y (jε) = H(jε)X (jε)
15 / 28
Example: convolution
Example: suppose a signal x(t) = sin(ω
ϑt
0 t)
is input into a lowpass
filter with frequency response
#
1, |ε| < εc ,
H(jε) =
0, |ε| ≃ εc
Method 1: inverse FT H(jε) to get h(t), then convolve.
! ↑ ! ωc
1 1 sin(εc t)
h(t) = H(jε)e jωt dε = e jωt dε =
2ω →↑ 2ω →ωc ωt
! ↑ ! ↑
sin(ε0 ϖ ) sin(εc (t ↔ ϖ ))
y (t) = x(ϖ )h(t ↔ ϖ )dϖ = dϖ
→↑ →↑ ωϖ ω(t ↔ ϖ )
16 / 28
Example: convolution
Method 2: compute X (jε) then use convolution property.
#
sin(ε0 t) 1, |ε| < ε0 ,
x(t) = ↗F X (jε) =
ωt 0, |ε| ≃ ε0
Y (jε) = H(jε)X (jε) =
y (t) =
17 / 28
Exercise: convolution
Consider an LTI system that sends
1 →|t|
x(t) = e →t u(t) ↑ y (t) = e
2
What is its impulse response?
Y (jε)
Y (jε) = X (jε)H(jε) ↓ H(jε) =
X (jε)
1
X (jε) =
1 + jε
1 1 1 1 1 1
y (t) = e →t u(t) + e t u(↔t) ↓ Y (jε) = +
2 2 2 1 + jε 2 1 ↔ jε
18 / 28
Exercise: convolution
Y (jε) 1 1 1 1
H(jε) = = · (1 + jε) + · (1 + jε)
X (jε) 2 1 + jε 2 1 ↔ jε
( )
1 1 + jε
= 1+
2 1 ↔ jε
( )
1 1 ↔ jε + 1 + jε
=
2 1 ↔ jε
1 2
=
2 1 ↔ jε
1
=
1 ↔ jε
h(t) = e t u(↔t)
19 / 28
The multiplication property
We know that:
y (t) = h(t) ↘ x(t)
Y (jε) = H(jε)X (jε)
Something similar holds when we interchange time and frequency:
r (t) = s(t)p(t)
! ↑
1
R(jε) = S(jϱ)P(j(ε ↔ ϱ))dϱ = S(jε) ↘ P(jε)
2ω →↑
This is the multiplication property.
20 / 28
Example: the multiplication property
We are going to take a much closer look at this when we discuss
communication systems and signal modulation.
For now, here is a taste:
Image credit: Prof. Olivia Di Matteo
21 / 28
Example: the multiplication property
To shift our signal into the frequency range of transmission, we can
multiply it by a carrier signal (amplitude modulation):
Is this doing what we think it is?
Image credit: Prof. Olivia Di Matteo
22 / 28
Example: the multiplication property
Consider the Fourier spectrum of both signals:
Image credit: Prof. Olivia Di Matteo
23 / 28
Example: the multiplication property
The multiplication property tells us
y (t) = x(t)c(t)
! ↑
1
Y (jε) = X (jϱ)C (j(ε ↔ ϱ))dϱ
2ω →↑
We have
F
x(t) →
↑ X (jε)
F
c(t) = e jωc t →
↑ C (jε) = 2ωϑ(ε ↔ εc )
24 / 28
Example: the multiplication property
Let’s convolve them:
! ↑
1
Y (jε) = X (jϱ)C (j(ε ↔ ϱ))dϱ
2ω →↑
! ↑
1
= X (jϱ)2ωϑ((ε ↔ εc ) ↔ ϱ)dϱ
2ω →↑
= X (j(ε ↔ εc ))
25 / 28
Example: the multiplication property
Multiplication with complex exponential carrier signal shifts the
spectrum. We can move it into the desired frequency range.
Image credit: Prof. Olivia Di Matteo
26 / 28
Example: the multiplication property
Image credit: Prof. Olivia Di Matteo
27 / 28
For next time
Content:
Behaviour of the Fourier transform under di!erentiation and
integration
LTI systems based on di!erential equations
Recommended reading:
From today’s class: Oppenheim 4.3-4.6
Suggested problems: 4.12, 4.15, 4.17, 4.19, 4.26, 4.32
For next class: Oppenheim chapters 4.6-4.8
28 / 28
