Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Convolution properties

• Commutative property
a[n] b[n] = b[n] a[n]

• Associative property
 (a[n] b[n]) c[n] = a[n] (b[n] c[n])

• Distributive property
 (a[n] b[n])+(a[n] c[n]) = a[n] (b[n]+ c[n])

1 1
Convolution – cascade, parallel connection
• Cascade connection Useful when we have many sub-systems in a
system and we want to determine the final
set of impulse responses for the system

• Note that the ordering of the systems in the cascade has no


effect on the overall impulse response because of the
commutative property

• Parallel connection

2
Convolution – cascade, parallel connection
• Example

• Simplifying

3
Convolution example (continuous)

4
Convolution example (continuous)
Stage 1 t < 0 and h(t-) does not overlap x() (Fig 4(d))
As the functions do not overlap x()h(t-) = 0 and there is no contribution to the
convolution integral

Stage 2 0 < t  2 and partial overlap occurs between h(t-) and x() (Fig 4 (e))
Over this range:

 
=t =t
y(t) = x( )h(t −  )d = 3  2d
=0 =0

y(t) = 6 0 = 6t
t
for 0 t 2

Note that the integration is with respect to  and not t. In this integration t behaves like a
constant and can be taken outside the integral sign

5
Convolution example (continuous)
Stage 3 2  t  3 and there is complete overlap of h(t-) and x() (Fig 4 (f, g))
Over this range of t:
y(t) =  x( )h(t −  )d =  3  2d
=t =t
 = t−2  = t−2

y(t) = 6 t−2 = 6(t − t + 2) = 12


t
for 2 t  3

Stage 4 3  t  5 This is another partial overlap region as shown in Figure 4 (h)


 
=3 =3
y(t) = x( )h(t −  )d = 3  2d
= t−2 = t−2

y(t) = 6 t−2 = 6(5 − t) = 30 − 6t


3
for 3 t 5

Stage 5 t > 5 As seen in Figure 4 (i) this is a second region of no overlap,


again with no contribution to the convolution integral.

6
Convolution example (continuous)

0 < t  2 y(t) = 6t
2  t  3 y(t) = 12
3  t  5 y(t) = 30 – 6t

The convolution output

Output length for continuous case = input length + impulse response length

7
Convolution Examples
MATLAB example
- convolution useful in filtering
1 1

0.8 0.8

0.6 0.6

0.4 0.4

i=1:1000; 0.2

0
0.2

x1=sin(2*pi*(5/1000)*i); -0.2 -0.2

plot(x1) -0.4

-0.6
-0.4

-0.6

x2=sin(2*pi*(400/1000)*i); -0.8 -0.8

plot(x2)
-1 -1
0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 900 1000

y=x1+x2; 2

plot(y) 1.5

h=[1 1];
1

0.5

z=conv(h,y); 0

figure, plot(z) -0.5

freqz(h)
-1

-1.5

-2
0 100 200 300 400 500 600 700 800 900 1000

20
Magnitude (dB)

-20 3

-40
2
-60
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized Frequency ( rad/sample) 1

0
0
Phase (degrees)

-50 -1

-2
-100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Normalized Frequency ( rad/sample)
-3
0 200 400 600 800 1000 1200

8
Convolution Examples
MATLAB examples

i=1:1000;
h= [0,4,-2,3]
x = [0,1,0,2,1] x1=sin(2*pi*(5/1000)*i);

y=conv(x,h) x2=sin(2*pi*(400/1000)*i);

i=1:1000; y=x1+x2;

x=sin(2*pi*(5/1000)*i);
h=[1 0 0 0 0 0 0 1]
h=[1 -1]
z=conv(y,h);
freqz(h)
plot(z)
z=conv(x,h);
h=[1 0 0 0 1 0 0 1]
plot(z)
z=conv(y,h);
hold on
plot(z)
plot(x, 'r') 9
End effects on finite signals
• In convolution, as one signal is shifted the waveforms no
longer overlap completely
• First few and last few samples in convolution calculations are
based on less information than the data points in between
• Analogous to stabilisation time in analogue circuits – switch
on and off, for example
• See below - an example when convolution is used for filtering
• More on this when we study filtering
1
Amplitude (arbitrary units)

15
0.5

0 10

-0.5

-1
0 200 400 600 800 1000 1200
LPF 5

Sampling points 0

1
Amplitude (arbitrary units)

-5
0.5

0 -10

-0.5
-15
-1 0 200 400 600 800 1000 1200
0 200 400 600 800 1000 1200
Sampling points

10
Summary of topics covered

• Impulse response
• Convolution summation (discrete)
• Convolution integral (continuous)
• Examples of discrete and continuous
calculations
• Properties of convolution
• MATLAB examples

11

You might also like