EECE 442: Communication Systems

Fourier Transform

EECE442: Communication Systems 1

History
 In 1807, Jean Baptiste Joseph Fourier showed that any periodic signal
could be represented by a series of sinusoidal functions
 Example:
– the composition of the first two functions gives the bottom one

EECE442: Communication Systems 2 2

Fourier Transform
 When we analyse a function using Fourier methods, the function is
decomposed into its frequency components.
 This analysis is used in signal processing, filtering
 Fourier Transform:

G f    g t exp j 2ft dt


 converts a function from the time (or spatial) domain to the frequency
domain
 A plot of the square of the modulus
 of the Fourier transform (H(w) vs w)
 is called the power spectrum.
 It gives the amount the frequency
 contributes to the waveform

EECE442: Communication Systems 3 3
Review: Fourier Transform
 A CT signal g(t) and its frequency domain, Fourier transform signal, G(f), are
related by

G f    g t exp j 2ft dt analysis
 
g t    G  f  exp  j 2ft df synthesis

 This is denoted by:
g t   G  f 
F

 For example:
 at
F 1
e u (t ) 
a  j 2f
 Often you have tables for common Fourier transforms
 The Fourier transform, G(f), represents the frequency content of g(t).
 It exists either when g(t)->0 as |t|->∞ or when g(t) is periodic (it generalizes
the Fourier series)
 Notes on the Amplitude and phase:
– The amplitude of the spectrum is symmetric with respect to the y axis (even
function)
– The phase of the spectrum is symmetric with respect to the origin (odd function)

EECE442: Communication Systems 4

Linearity of the Fourier Transform
 The Fourier transform is a linear function of g(t)

g1 t   G1  f 
F

g 2 t   G2  f 
F

ag1 t   bg 2 t   aG1  f   bG2  f 

F

 This follows directly from the definition of the Fourier transform (as the
integral operator is linear) & it easily extends to an arbitrary number of
signals
 Like impulses/convolution, if we know the Fourier transform of simple
signals, we can calculate the Fourier transform of more complex signals
which are a linear combination of the simple signals

EECE442: Communication Systems 5

Fourier Transform of a Time Shifted Signal
 We’ll show that a Fourier transform of a signal which has a simple time
shift is:

𝐹𝑇 𝑔 𝑡 − 𝑡0 = 𝑒 −𝑗2𝜋𝑓𝑡0 𝐺(𝑓)

 i.e. the original Fourier transform but shifted in phase by –wt0

 Proof
– Consider the Fourier transform synthesis equation:

+∞

𝐹𝑇 𝑔 𝑡 − 𝑡0 = න 𝑔(𝑡 − 𝑡0 )𝑒 −𝑗2𝜋𝑓𝑡 𝑑𝑓
−∞
+∞ +∞

𝐹𝑇 𝑔 𝑡 − 𝑡0 = න 𝑔(𝑢)𝑒 −𝑗2𝜋𝑓(𝑢+𝑡0) 𝑑𝑢 = 𝑒 −𝑗2𝜋𝑓𝑡0 න 𝑔(𝑢)𝑒 −𝑗2𝜋𝑓𝑢 𝑑𝑢

−∞ −∞

EECE442: Communication Systems 6

Example: Linearity & Time Shift
 Consider the signal (linear sum of two time shifted rectangular pulses),
where x1(t) is of width 1, x2(t) is of width 3,
centred on zero (see figures)
x1(t)
x(t )  0.5 x1 (t  2.5)  x2 (t  2.5)
t
 Using the FT of a rectangular pulse L

sin(2𝜋𝑓) sin(3𝜋𝑓) x2(t)

𝑋1 𝑓 = , 𝑎𝑛𝑑𝑋2 𝑓 = 2
2𝜋𝑓 2𝜋𝑓
t
x (t)

 Then using the linearity and time shift Fourier transform

properties
t
sin(2𝜋𝑓) sin(3𝜋𝑓)
X 𝑓 = 𝑒 −𝑗5𝜋𝑓 0.5 + 2
2𝜋𝑓 2𝜋𝑓

EECE442: Communication Systems 7

Fourier Transform of a Derivative
 By differentiating both sides of the Fourier transform synthesis equation
with respect to t:

+∞

𝑥 𝑡 = න 𝑋(𝑓)𝑒 𝑗2𝜋𝑓 𝑑𝑓
−∞
+∞
𝑑𝑥(𝑡)
= න 𝑗2𝜋𝑓𝑋(𝑓)𝑒 𝑗2𝜋𝑓 𝑑𝑓
𝑑𝑡
−∞
 Therefore noting that this is the synthesis equation for the Fourier
transform 𝑗𝜔𝑓 𝑋(𝑓)
dxt 
 j 2f X  f 
dt

 This is very important, because it replaces differentiation in the time

domain with multiplication (by j𝜔𝑓) in the frequency domain.

EECE442: Communication Systems 8

Convolution in the Frequency Domain
 We can easily solve ODEs in the frequency domain:
F
y (t )  h(t ) * x(t )  Y ( f )  H ( f ) X ( f )
 Therefore, to apply convolution in the frequency domain, we just have to
multiply the two Fourier Transforms.

 To solve for the differential/convolution equation using Fourier

transforms:
– Calculate Fourier transforms of x(t) and h(t): X(f) by H(f)
– Multiply H(f) by X(f) to obtain Y(f)
– Calculate the inverse Fourier transform of Y(f)

 H(f) is the LTI system’s transfer function which is the Fourier transform of
the impulse response, h(t). Very important in the remainder of the course
(using Laplace transforms)
 This result is proven in the appendix

EECE442: Communication Systems 9

Example 1: Response to a Filter
 Calculate the response of a CT LTI system with impulse response:
h(t )  e btu (t ) b0
 to the input signal:
x(t )  e  at u (t ) a0
 Taking Fourier transforms of both signals:
1 1
H( f )  and X ( f ) 
(b  j 2f ) (a  j 2f )
 gives the overall frequency response:
1
Y( f ) 
(b  j 2f )(a  j 2f )

 to convert this to the time domain, express as partial fractions:

1  1 1  Assume ba
Y( f )    
b  a  (a  j ) (b  j ) 

 Therefore, the CT system response is:


y (t )  b 1 a e  at u (t )  e  bt u (t ) 
EECE442: Communication Systems 10
Example 2: Design a Low Pass Filter
 Consider an ideal low pass filter in frequency domain:
H(j)
1 |  | c
H ( j )  
0 |  | c
 X ( j ) |  | c c c 
Y ( j )  
 0 |  | c

 The filter’s impulse response is the inverse Fourier transform

c sin(c t )
ce d  t
j t
h(t )  1
2

 which is an ideal low pass CT filter. However it is non-causal, so this

cannot be manufactured exactly & the time-domain oscillations may be
undesirable
h(t)

0 t

EECE442: Communication Systems 11

Summary (1/2)
 The Fourier transform is widely used for designing filters. You can design
systems with reject high frequency noise and just retain the low frequency
components. This is natural to describe in the frequency domain.

 Important properties of the Fourier transform are:

F
– Linearity and time shifts ax(t )  by(t )  aX ( f )  bY ( f )
dxt 
– Differentiation  j 2f X  f 
dt
F
– Convolution y (t )  h(t ) * x(t )  Y ( f )  H ( f ) X ( f )
  
– Energy Theorem
E  x(t ) dt   then  x(t ) dt   X ( f ) df
2 2 2
if
  
 Some operations are simplified in the frequency domain, but there are a
number of signals for which the Fourier transform does not exist – this
leads naturally onto Laplace transforms. Similar properties hold for
Laplace transforms & the Laplace transform is widely used in engineering
analysis.
EECE442: Communication Systems 12
Summary (2/2)
 If the time-domain description of a signal changes then the frequency
domain representation changes inversely and vice versa.
 If a signal is strictly limited in frequency domain then the time domain
description of the signal will trail or it will be indefinite in time domain (it
might decrease but still exists)
 A signal is strictly limited in frequency if its spectrum is zero outside a
given range of frequencies
– The bandwidth gives a measure of the extent (content) of the non-zero
frequencies or spectral content of the signal for positive frequencies
– The bandwidth of a low-pass signal is specified by the spectral content which
separates the frequency at which the amplitude attains its maximum and the
positive frequency at which the amplitude spectrum drops by sqrt(2).
– Time bandwidth relation of a pulse:
• (duration).(bandwidth)=constant  first principle expressed in this slide

EECE442: Communication Systems 13

Exercises
 Theory
– Using linearity & time shift calculate the Fourier transform of

x(t )  5e 3(t 1)u (t  1)  7e 3(t 2)u(t  2)

– Use the FT derivative relationship and the Fourier series/transform expression for
sin(w0t) (L10-S3) to evaluate the FT of cos(w0t).

– Calculate the FTs of the systems’ impulse responses

– Calculate the system responses y(t) when the following input signal is applied

y (t )
 3 y (t )  x(t )
t

x(t )  e 5t u (t )

EECE442: Communication Systems 14

Periodic signals
 Periodic signal representation: Any periodic signal could be represented


c exp  j 2nf 0 t 
as
g T0 (t ) 
– n  
n
(Fourier Series)
T0 / 2

(t ) exp  j 2nf 0 t dt

1
– with cn 
T0 g
T0 / 2
T0

 But: gT0 (t )   g (t  kT )
k  
0 with g (t ) is a generating function
g T0 (t )
F
 Let g (t )  G ( f )
– then: cn  f 0Gnf0 
 Properties of periodic signals: (for all periods)
 0 0 0
g T0 (t )   f Gnf exp j 2nf t 
n  
0 0 0
g (t )
 

 g (t  kT )  f  Gnf   f  nf 
F
gT0 (t )  0 0 0 0
k   n  

– discretization with different weights in the frequency domain

EECE442: Communication Systems 15

Ideal Sampling Function
 It consists of an infinite sequence of uniformly spaced delta functions. It is
 denoted as: 
 T (t ) 
0   (t  kT )
k  
0

 Using the properties of the periodic Signal:

 

  (t  kT )  f    f  nf 
F
gT0 (t )  0 0 0
k   n  

EECE442: Communication Systems 16

Discrete Fourier Transform: Like Fourier Series
There is another, only slightly different, way to write a discrete-time periodic signal as a
sum of complex exponentials of frequency ω0k . The discrete Fourier series is defined as:
N 1
1
x(k ) 
N
 X ( n
k 0
) e ik0 n

 The discrete Fourier transform (DFT) and inverse DFT (IDFT):

N 1  j 2
kn

X n    x(k )e N

n 0

N 1 j 2
kn
1
x(k ) 
N
 X ( n)e
k 0
N

EECE442: Communication Systems 17

Readings
 Revision of the Fourier Transform, Fourier Series (continuous and Discrete)
 Causality and stability: need to know the principle
 Filter design and specification: ripple, attenuation, order, etc

EECE442: Communication Systems 18