Fourier_transform_applications

Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

B39SA

Signals and Systems


1. Energy and Power Spectral Densities
2. LTI Systems, System Function
3. Autocorrelation

Alexander Belyaev
[email protected]
Fourier transform
Forward transform, analysis equation

X (ω ) = F [x (t )] = ∫ x ( t ) e − jω t
dt
−∞

Inverse transform, synthesis equation



1
x (t ) = F -1
[X (ω )] = ∫ X (ω ) e jω t

2π −∞

Fourier transform pair: x (t ) ↔ X (ω )


Basic properties of Fourier transform

Linearity: a1 x1 (t ) + a2 x2 (t ) ↔ a1 X 1 (ω ) + a 2 X 2 (ω )

Time Shifting: x (t − t 0 ) ↔ X (ω ) e − jω t0

Frequency Shifting: x (t ) e ↔ X (ω − ω 0 )
jω 0 t

1 ω 
Scaling: x (a t ) ↔ X 
a a
Time-Reversal: x (− t ) ↔ X (− ω )

Duality: X ( t ) ↔ 2π x (− ω )
Properties of Fourier transform
d
Differentiation: x′(t ) = x (t ) ↔ jω X (ω )
dt
t
1
∫−∞
Integration: x (τ ) dτ ↔ π X (0 )δ (ω ) +

X (ω )

Convolution: x (t ) = x1 (t ) ∗ x2 (t ) = ∫ x (τ )x (t − τ ) dτ
−∞
1 2

x1 (t ) ∗ x2 (t ) ↔ X 1 (ω ) X 2 (ω )

1
Multiplication: x1 (t ) x2 (t ) ↔ X 1 (ω ) ∗ X 2 (ω )

Examples of calculating Fourier transforms
Find the Fourier transform of
x1 (t ) = Ae − at u (t ) x2 (t ) ≡ 1 x3 (t ) = cos (2 ω 0 t )
Examples of calculating Fourier transforms
Find the Fourier transform of
x1 (t ) = Ae − at u (t ) x2 (t ) ≡ 1 x3 (t ) = cos (2 ω 0 t )

Solution:
∞ ∞
A
X 1 (ω ) = F [x1 (t )] = ∫ Ae e− at − jω t
dt = A ∫ e − ( a + jω )t
dt =
0 0
a + jω
Examples of calculating Fourier transforms
Find the Fourier transform of
x1 (t ) = Ae − at u (t ) x2 (t ) ≡ 1 x3 (t ) = cos (2 ω 0 t )

Solution:
∞ ∞
A
X 1 (ω ) = F [x1 (t )] = ∫ Ae e− at − jω t
dt = A ∫ e − ( a + jω )t
dt =
0 0
a + jω

F -1 [δ (ω )] = 1 (2π ) ⇒ X 2 (ω ) = F [1] = 2π δ (ω )
Examples of calculating Fourier transforms
Find the Fourier transform of
x1 (t ) = Ae − at u (t ) x2 (t ) ≡ 1 x3 (t ) = cos (2 ω 0 t )

Solution:
∞ ∞
A
X 1 (ω ) = F [x1 (t )] = ∫ Ae e − at − jω t
dt = A ∫ e − ( a + jω )t
dt =
0 0
a + jω

F -1 [δ (t )] = 1 (2π ) ⇒ X 2 (ω ) = F [1] = 2π δ (ω )

F [1] = 2π δ (ω ) ⇒ F [exp ( j 2 ω 0t )] = 2πδ (ω − 2ω 0 )


1
x3 (t ) = cos (2 ω 0 t ) = [exp ( j 2 ω 0t ) + exp (− j 2 ω 0t )]
2
X 3 (ω ) = F [x3 (t )] = π [δ (ω − 2ω 0 ) + δ (ω + 2ω 0 )]
Example: Modulation Theorem Numerous applications in
F digital communication
Show that if x (t ) ←→ X (ω ) then systems
F 1 1
x (t ) cos ω 0t ←→ X (ω − ω 0 ) + X (ω + ω 0 )
2 2

Solution:
1 jω 0 t
cos ω 0t =
2
e( )
+ e − jω 0 t

1 1 − jω 0 t  1 1
[ ( ) ]
F x t cos ω 0t = F  x t e ( ) jω 0 t
( )
+ xt e  = X ω − ω 0 + X (ω + ω 0 )
( )
2 2  2 2
Example: Impulse train ∞

Find the Fourier transform


δ T (t ) = ∑ δ (t − nT )
n = −∞
of the unit impulse train

− 2T − T 0 T 2T t
Solution: ∞
It is a periodic function, its Fourier series is given by δ T (t ) = ∑ n
c e
n = −∞
jnω 0 t

T 2 T 2
1 1 1
∫ δ (t ) e ∫ δ (t )e
− jnω 0 t − jnω 0 t
cn = T dt = dt =
T −T 2
T −T 2
T
F
1 ∞
2π 1 → 2π δ (ω )
δ T (t ) =
T
∑e
n = −∞
jnω 0 t
ω0 =
T F
j nω 0 t
e → 2π δ (ω − nω 0 )

2π Mathematical basis for
F [δ T (t )] =
T
∑ δ (ω − nω )
n = −∞
0 approximating continuous (analogue)
signals by discrete (digital) signals
1. Parseval’s theorem for energy
signals and energy spectral density
∞ ∞

∫ ∫ x(t ) x(t ) dt
2
x (t ) dt =
*
E=
−∞ −∞
*

 1 ∞

= ∫ x (t )  ∫ X (ω )e
jω t
dω  dt
 2π  2
Sx ( f ) = X ( f )
−∞ −∞
∞ ∞
1
∫ ∫( ) * − jω t
= x t X (ω ) e dω dt
2π −∞ −∞

1

∞  *
∫−∞ −∫∞ x(t ) e dt  X (ω ) dω
− jω t
=
2π ω = 2π f
∞ ∞ ∞
1 1
∫−∞ X (ω ) X (ω ) dω = 2π ∫ X (ω ) ∫ X(f )
* 2 2
= dω = df
2π −∞ −∞
1. Energy spectral density: An example
1
Find the energy spectral density x = Π (t )
of the rectangular pulse: x (t ) = 20Π (100 t )
− 0 .5 0 .5
Solution:
sin (π α )
sinc (α ) =
x (t ) = AΠ (t τ ) πα
τ 2
2A  ωτ  sin (ωτ 2 )  ωτ 
X (ω ) = ∫τ Ae
− jω t
dt = sin   = Aτ = Aτ sinc  
− 2
ω  2  ωτ 2  2π 

F 1  ω 
x (t ) = 20 Π (100 t ) → sinc   = 0.2 sinc (0.01 f )
5  200 π 
2
X ( f ) = 0.04 sinc 2 (0.01 f )
1. Parseval’s theorem for periodic signals
and power spectral density (PSD)
∞ t 0 + T0 ∞
2π 1
∑c e ∫ x(t ) ∑c
2 2
x (t ) = k
jω 0 t
ω0 = dt = k
k = −∞ T0 T0 t0 k = −∞

∑c
2
Power spectral density: Sx ( f ) = k δ ( f − k f0 )
k = −∞

∞ ∞
 ∞ 
∫−∞S x ( f ) df = −∫∞k∑
2
ck δ ( f − k f 0 ) df
= −∞ 
∞ ∞ 2  ∞
= ∑  ∫ ck δ ( f − k f 0 ) df  = ∑ ck
2

k = −∞  − ∞  k = −∞
1. Power spectral density: An example
Find the power spectral density of x (t ) = 10 cos (300π t )
Solution:

x (t ) = 10 cos (300π t ) = 5e − j 300 π t + 5e j 300 π t


∑c
2
c1 = c−1 = 5 f 0 = 150 Hz Sx ( f ) = k δ ( f − k f0 )
k = −∞
2 2
= c−1 δ ( f + 150 ) + c1 δ ( f − 150 ) = 25 δ ( f + 150 ) + 25 δ ( f − 150 )

Sx ( f )
25 25
f
− 150 150
2. Linear Time-Invariant (LTI) Systems

y (t ) = Sys[x (t ) ] x (t ) LTI System y (t )

Sys[x1 (t ) + x2 (t ) ] = Sys[x1 (t ) ] + Sys[x2 (t ) ] = y1 (t ) + y 2 (t )

Sys[α x (t ) ] = α Sys[x (t ) ] = α y (t )

Sys[x (t − t 0 ) ] = y (t − t 0 )
2. Impulse response and response to an arbitrary input

y (t ) = Sys[x (t ) ] x (t ) LTI System y (t )


impulse
h(t ) = Sys[δ (t ) ] response If h(t) = 0 for t < 0 the
system is called causal.
∞ ∞
x (t ) = δ (t ) ∗ x (t ) = ∫ δ (t − τ ) x (τ ) dτ = ∫ δ (τ ) x (t − τ ) dτ
−∞ −∞
∞ ∞
Sys[x (t ) ] = ∫ Sys[δ (t − τ )] x (τ ) dτ = ∫ h(t − τ ) x (τ ) dτ = h(t ) ∗ x (t )
−∞ −∞

The response y(t) of an LTI system to an arbitrary


input x(t) is given by
∞ ∞
y (t ) = h(t ) ∗ x (t ) = ∫ x(τ ) h(t − τ ) dτ = ∫ x(t − τ ) h(τ ) dτ
−∞ −∞
2. System function (frequency response)

y (t ) = Sys[x (t ) ] x (t ) LTI System y (t )


∞ ∞
y (t ) = h(t ) ∗ x (t ) = ∫ x(τ ) h(t − τ ) dτ = ∫ x(t − τ ) h(t ) dτ
−∞ −∞

Let us apply the Fourier transform


X (ω ) = F [x (t )] H (ω ) = F [h(t )] Y (ω ) = F [ y (t )]
y (t ) = h(t ) ∗ x (t ) → Y (ω ) = X (ω )H (ω )

Y (ω ) System function (frequency response,


H (ω ) =
X (ω ) frequency transfer function)
2. System 1 H (ω )
function
b b
(frequency
response) δ (t ) h(t )
LTI System
x (t ) y (t ) = x (t ) ∗ h(t )
b b
X (ω ) Y (ω ) = X (ω ) H (ω )

 ∞
 jω t
[ ]
Sys e jω t = ∫ h (τ ) e jω (t −τ )
dτ =  ∫ h (τ ) e − jω τ
dτ  e = H (ω ) e jω t
−∞ −∞ 

[ ]
Sys e jω t = H (ω ) e jω t

x (t ) = ∑ cω e jω t → ∑ cω H (ω ) e jω t
ω ω
2. System function (frequency response): Example 1

+ +
R
x(t ) y (t ) x (t ) →
Sys
y (t )
C
− −

δ (t ) → ?
Sys
u (t ) → ?
Sys
2. System function (frequency response): Example 1
+ + x (t ) →
Sys
y (t )
R
x(t )
Sys
y (t ) δ (t ) → ?
C
− −
( ) Sys
u t → ?
t=0
vC (t ) = A + Be −t ( RC )
R
+
V =1V vC (t ) vC (0 ) = 0 vC (∞ ) = 1
- C vC (t ) = 1 − e −t ( RC )

u (t ) →
Sys
( )
1 − e −t ( RC ) u (t )
du (t ) Sys 1 −t ( RC )
δ (t ) = → e u (t )
dt RC
2. System function (frequency response): Example 1

+ +
R
x (t ) →
Sys
y (t )
x(t ) y (t )
C δ (t ) →
Sys
h(t )
− − 1 −t ( RC )
h(t ) = e u (t )
RC

1
H (ω ) = F [h(t )] = ∫ h(t )e
− jω t
dt =
−∞
1 + jω RC
2. System function (frequency response): Example 1

+ + x (t ) y (t )
R LTI System
e jω t
1 H (ω ) e jω t
jω C Sys e [ ] = H (ω )e
jω t jω t

− −

Phasor analysis yields the same result:


jω t jω t
e 1 e
H (ω ) e jω t
= =
R + 1 ( jω C ) jω C 1 + jω RC
1
H (ω ) =
1 + jω RC
2. System function (frequency response): Example 2

+ + x (t ) y (t )
R LTI System
x(t ) = Π (t ) 1 y (t )
jω C Sys e [ ] = H (ω )e
jω t jω t

− −
1
H (ω ) = 1
1 + jω RC
x = Π (t )

x (t ) = AΠ (t τ ) − 0 .5 0 .5
τ 2
2A  ωτ  sin (ωτ 2 )  ωτ 
X (ω ) = ∫τ Ae
− jω t
dt = sin   = Aτ = Aτ sinc  
− 2
ω  2  ωτ 2  2π 
1 ω 
Y (ω ) = H (ω ) X (ω ) = sinc  
1 + jω RC  2π 
2. System function (frequency response): Example 2
+ +
1
R
x = Π (t ) x(t ) = Π (t ) 1 y (t )
− 0 .5 0 .5 jω C
− −
1
H (ω ) =
1 + jω RC
1 ω 
Y (ω ) = sinc  
1 + jω RC  2π 
2. System function (frequency response): Example 2
+ +
R
x (t ) = Π (t ) 1 y (t )
jω C
− −
2. System function (frequency response): Example 2
2. System function (frequency response): Example 3

x (t ) y (t )
LTI System

d 2 y (t ) dy (t )
2
+2 + 26 y (t ) = x (t )
dt dt
2. System function (frequency response): Example 3

x (t ) y (t )
LTI System

d 2 y (t ) dy (t )
2
+2 + 26 y (t ) = x (t )
dt dt

( jω )2 Y (ω ) + 2( jω )Y (ω ) + 26Y (ω ) = X (ω )
Y (ω ) 1 1
H (ω ) = = =
X (ω ) ( jω ) + 2( jω ) + 26 (26 − ω 2 ) + j 2ω
2
3. Autocorrelation function

Let x(t) be an energy signal. The autocorrelation function


is defined by ∞
rxx (τ ) = ∫ x(t ) x(t + τ ) dt
−∞

Let x(t) be a periodic signal. The autocorrelation function


is defined by T0 2
1
rxx (τ ) =
T0 ∫ x(t ) x(t + τ ) dt
−T0 2

If x(t) is a non-periodic power signal. The autocorrelation


function is defined by  T0 2 
1
rxx (τ ) = lim 
 0
T0 → ∞ T ∫ x(t ) x(t + τ ) dt 
−T0 2 
3. Properties of the autocorrelation function


rxx (τ ) = ∫ x(t + τ ) x(t ) dt
−∞
rxx (τ ) = rxx (− τ )


F
x (t ) ∗ y (t ) = ∫ x (t − s ) y (s ) ds → X (ω )Y (ω )
−∞
∞ ∞
rxx (t ) = x (t ) ∗ x (− t ) = ∫ x(s − t ) x(s ) ds = ∫ x(λ ) x(λ + t ) dλ
−∞ −∞
∞ ∞
F F
x (t ) → X (ω ) x (− t ) → ∫ x (− t ) e − jω t dt = ∫ x (s ) e jω s
ds = X *
(ω )
−∞ −∞

F 2
rxx (t ) → X (ω ) X (ω ) = X (ω ) = P (ω )
*
Spectral Density
3. Applications of autocorrelation

The auto-correlation
of a random signal is
zero at all lags except
for the lag of zero.
3. Applications of autocorrelation
x = 3*sin(2*pi*f1*t)+2*sin(2*pi*f2*t);
n = 3*randn(size(t));
y = x+n;

Noisy signal

PSD

You might also like