Graphical Convolution Example

• Suppose that f(t) = g(t) where f(t) is the rectangular pulse depicted in
figure, of height 1.

1, 0  t  T
f (t )  g (t )  
0, otherwise

0 T t
 Graphical Convolution Example
• Case 1: t < 0

g (t   ) f ( )

t T t 0 T 

y (t )   f ( ) g (t   )d  0

• Case 2: 0 <= t< T

g (t   ) f (t )

t T 0 t T 
t

 1.1d  t
0
• Case 3: T  t  2T

f (t ) g (t   )

0 t T T t 
T

 1.1d  2T  t
t T
• Case 4: t  2T

x( ) h(t   )

0 T t T t 
 Output
This image cannot currently be display ed.

0, t0
t , 0t T

y (t )  
2T  t , T  t  2T
0, t  2T
y(t)

0 T 2T t

Example of a resistor?
 Graphical Convolution Example
• Convolve the following two functions:

f(t) g(t)
2 3

*
t t
2 -2 2

• Replace t with τ in f(t) = -t +2 and g(t)

• Choose to flip and slide g(t)
• Functions overlap like this: 3 g(t-)
 2
f()
y (t ) 

 f ( ) g (t   )d 2

-2 + t t 2+t
 Graphical Convolution Example
3
• Convolution can be divided into 5 parts g(t-)

I. t < -2 2
f()
• Two functions do not overlap

• Area under the product of the 2
-2 + t 2+t
functions is zero
3 g(t-)
II. -2  t < 0
2
• Part of g(t) overlaps part of f(t) f()

• Area under the product of the 

2
functions is -2 + t 2+t

 2t 2t
2
  
y (t )   f ( ) g (t   )d   3(  2)d  3   2 
 0  2 0
32  t 
2
3t 2
  62  t    6
2 2
Graphical Convolution Example
III. 0t<2 3 g(t-)
• Here, g(t) completely overlaps f(t)
2
• Area under the product is just f()
2 
2 2
   2
0 3   2  d   3  
 2  2  6 -2 + t 2+t
  0
IV. 2t<4 3 g(t-)
• Part of g(t) and f(t) overlap
2
• Calculated similarly to -2  t < 0 f()

• 3t2/2 -12t +24 

2
V. t>4 -2 + t 2+t

• g(t) and f(t) do not overlap

• Area under their product is zero
 Example
• Convolution of two gate pulses each of height 1

f( g(

0 1 t 0 2 t

y   f  g t   d 

 Example

f*g Case 1: 0 <= t < 1

Case 2: 1<= t <=2
Case 3: 2 <= t <= 3
Case $: t > 3

0 1 2 3 t
 Properties of LTI systems
• Commutative Property: Roles of the input and impulse response can be
interchanged
• CT Systems: y (t )  x(t )  h(t )  h(t )  x(t )
 
  x( )h(t   )d   h( )x(t   )d
 

• DT systems: y[n]  x[n]  h[n]  h[n]  x[n]

 
  x[k ]h[n  k ]   h[k ]x[n  k ]
 

h[n] x[n]
x[n] y[n] h[n] y[n]
 Properties of LTI systems
• Distributive

• CT systems y (t )  x(t )  ( h1(t )  h 2(t ))

 x(t )  h1(t )  x(t )  h 2(t )

• DT systems y[n]  x[n]  (h1[n]  h 2[n])

 x[n]  h1[n]  x[n]  h 2[n]

x[n] h1[n]
y[n] x[n] y[n]
h1[n] + h2[n]

h2[n]
 Properties of LTI systems
• Associative

• CT systems y (t )  x(t )  (h1(t )  h 2(t ))

 ( x (t )  h1(t ))  h 2(t )
 h1(t )  ( x(t )  h 2(t ))

• DT systems y[n]  x[n]  ( h1[n]  h 2[n])

 ( x[ n]  h1[n])  h2[ n]
 h1[ n]  ( x[n]  h2[ n])

h[n] h[n] h1n]*h2[n]

x[n] y[n] x[n] y[n]
• Memoryless: A LTI system is memoryless if its impulse response is

h[n]  K [n] ; h(t )  K (t )

• Causality: An LTI system is causal if its output does not depend on future
values of input. Or, output at [n] must not depend on k>n.
h[n]  0 for n  0; h(t )  0 for t  0
 n
y[n]   x[k ]h[n  k ]   x[k ]h[n  k ]
k   k  

• Stability: A LTI system is stable if its impulse response

 

 | h[k ] |  ;  | h(t ) | dt  
k   
 Example
• Find if memoryless, causal and stable?
a) h(t) = u(t+1) – u(t-1)
b) h(t) = u(t) – 2u(t-1)
c) h(t) = e-2|t|
d) h(t) = eatu(t)
e) h[n]=2nu[-n]
f) h[n] = e2nu[n - 1]
• Invertibility: A LTI system is invertible if
h[ n]  h 1[n]   [n]
h(t )  h 1 (t )   (t )

• Find a causal inverse system of y[n] = x[n] + ax[n-1]. Recall the echo
problem or multipath communication problem. A signal may travel
through different paths.

• Also find if inverse system is stable?

 RADAR Range measurement
• Suppose we transmit an RF pulse and
determine the round trip time delay

sin(t ),0  t  T
x(t )  
0, otherwise

We need to compute the received signal and

channel’s impulse response.
 Matched filter
• We need to compute β, towards this we need
to match the received signal with the
transmitted signal.

• We can build an LTI system, such that the

impulse response is
 sin t ,T  t  0
h ( t )  
0, otherwise
 Example
If x(t )  e  at u(t ), a  0
and h(t )  u (t )
for t  0, then product of x( τ ) and h(t-τ )  0
e  aτ , 0  τ  t
x( τ )h(t-τ )  
0, otherwise
t
 aτ 1  aτ t 1
y (t )   e dτ   e  (1  e  at )
0 a 0 a
thus for all t, y(t) is
1
y(t)  (1  e  at )u (t )
a
 Example
• If y(t) denote convolution of following two
signals
x(t )  e 2t u (t )
h(t )  u (t  3)
when (t - 3)  0 , the product of x( ) and h(t -  ) is nonzero
for -     t - 3
for t  3, product is non zero for -     0