Introduction to Convolution

Definition Convolution is an operator that takes an input

signal and returns an output signal, based on knowledge
about the system’s unit impulse response h[n].
x[n] = [n] System y[n] = h[n]

x[n] System: h[n] y[n]

The basic idea behind convolution is to use the system’s

response to a simple input signal to calculate the
response to more complex signals
This is possible for LTI systems because they possess the
superposition property (lecture 3):
x[n]  k ak xk [n]  a1 x1[ n]  a2 x2 [n]  a3 x3 [n]  

y[n]  k ak yk [n]  a1 y1[n]  a2 y2 [n]  a3 y3[n]  

EE-2027 SaS, L4: 1/17

 Discrete Impulses & Time Shifts
Basic idea: use a (infinite) set of of discrete time impulses to
represent any signal.
Consider any discrete input signal x[n]. This can be written
as the linear sum of a set of unit impulse signals:
x[1] n  1
x[1] [n  1]  
 0 n  1 x[1]  [n  1]
x[0] n  0
x[0] [n]  
 0 n0
x[1] n  1
x[1] [n  1]   actual value Impulse, time
 0 n 1 shifted signal
Therefore, the signal can be expressed as:
x[n]    x[2] [n  2]  x[1] [n  1]  x[0] [n]  x[1] [n  1]  
In general, any discrete signal can be represented as:

x[n]   x[k ] [n  k ]
k  
The sifting property

EE-2027 SaS, L4: 2/17

 Example
The discrete signal x[n]

Is decomposed into the following

additive components
x[-4][n+4] +
x[-3][n+3] + x[-2][n+2] + x[-1][n+1] + …

EE-2027 SaS, L4: 3/17

 Discrete, Unit Impulse System Response
A very important way to analyse a system is to study
the output signal when a unit impulse signal is used
as an input
[n] System:  h[n]

Loosely speaking, this corresponds to giving the system

a kick at n=0, and then seeing what happens
This is so common, a specific notation, h[n], is used to
denote the output signal, rather than the more
general y[n].
The output signal can be used to infer properties about
the system’s structure and its parameters .

EE-2027 SaS, L4: 4/17

 Types of Unit Impulse Response
Causal, stable, finite impulse response
y[n] = x[n] + 0.5x[n-1] + 0.25x[n-2]
Looking at unit impulse
responses, allows you to
determine certain system
properties

Causal, stable, infinite impulse response

y[n] = x[n] + 0.7y[n-1]

Causal, unstable, infinite impulse response

EE-2027 SaS, L4: y[n] = x[n] + 1.3y[n-1] 5/17
 Linear, Time Varying Systems
If the system is time varying, let hk[n] denote the response
to the impulse signal [n-k] (because it is time varying,
the impulse responses at different times will change).
Then from the superposition property (Lecture 3) of linear
systems, the system’s response to a more general input
signal x[n] can be written as:
Input signal 
x[n]   x[k ] [n  k ]
k  

System output signal is given by the convolution sum


y[n]   x[k ]h [n]
k  
k

i.e. it is the scaled sum of impulse responses

EE-2027 SaS, L4: 6/17
 Example: Time Varying Convolution
x[n] = [0 0 –1 1.5 0 0 0]
h-1[n] = [0 0 –1.5 –0.7 .4 0 0]
h0[n] = [0 0 0 0.5 0.8 1.7 0]

y[n] = [0 0 1.4 1.4 0.7 2.6 0]

EE-2027 SaS, L4: 7/17

 Linear Time Invariant Systems
When system is linear, time invariant, the unit impulse
responses are all time-shifted versions of each other:
hk [n]  h0  n  k 
It is usual to drop the 0 subscript and simply define the
unit impulse response h[n] as:
h[n]  h0  n
In this case, the convolution

sum for LTI systems is:
y[n]   x[k ]h[n  k ]
k  

It is called the convolution sum (or superposition sum)

because it involves the convolution of two signals x[n]
and h[n], and is sometimes written as:
y[n]  x[n] * h[n]

EE-2027 SaS, L4: 8/17

 System Identification and Prediction
Note that the system’s response to an arbitrary input signal is
completely determined by its response to the unit impulse.
Therefore, if we need to identify a particular LTI system, we
can apply a unit impulse signal and measure the system’s
response.
That data can then be used to predict the system’s response
to any input signal
x[n] y[n]
System: h[n]

Note that describing an LTI system using h[n], is equivalent to

a description using a difference equation. There is a direct
mapping between h[n] and the parameters/order of a
difference equation such as:
y[n] = x[n] + 0.5x[n-1] + 0.25x[n-2]

EE-2027 SaS, L4: 9/17

 Example 1: LTI Convolution
Consider a LTI system with the
following unit impulse response:
h[n] = [0 0 1 1 1 0 0]

For the input sequence:

x[n] = [0 0 0.5 2 0 0 0]

The result is:

y[n] = … + x[0]h[n] + x[1]h[n-1] + …
=0+
0.5*[0 0 1 1 1 0 0] +
2.0*[0 0 0 1 1 1 0] +
0
= [0 0 0.5 2.5 2.5 2 0]

EE-2027 SaS, L4: 10/17

 Example 2: LTI Convolution
Consider the problem
described for example 1
Sketch x[k] and h[n-k] for any
particular value of n, then
multiply the two signals and
sum over all values of k.
For n<0, we see that x[k]h[n-k]
= 0 for all k, since the non-
zero values of the two
signals do not overlap.
y[0] = kx[k]h[0-k] = 0.5
y[1] = kx[k]h[1-k] = 0.5+2
y[2] = kx[k]h[2-k] = 0.5+2
y[3] = kx[k]h[3-k] = 2
As found in Example 1

EE-2027 SaS, L4: 11/17

 Example 3: LTI Convolution
Consider a LTI system that has a step
response h[n] = u[n] to the unit
impulse input signal
What is the response when an input
signal of the form
x[n] = nu[n]
where 0<<1, is applied?
For n0: n
y[n]    k
k 0

1   n 1

1
Therefore,
 1   n 1 
y[n]   u[n]
 1 

EE-2027 SaS, L4: 12/17

 Discrete LTI Convolution in Matlab
In Matlab to find out about a command, you can search the help
files or type:
>> lookfor convolution
at the Matlab command line. This returns all Matlab functions that
contain the term “convolution” in the basic description
These include:
conv()
To see how this works and other functions that may be appropriate,
type:
>> help conv
at the Matlab command line
Example:
>> h = [0 0 1 1 1 0 0];
>> x = [0 0 0.5 2 0 0 0];
>> y = conv(x, h)
>> y = [0 0 0 0 0.5 2.5 2.5 2 0 0 0 0 0]

EE-2027 SaS, L4: 13/17

 Lecture 4: Summary
Any discrete LTI system can be completely determined by
measuring its unit impulse response h[n]
This can be used to predict the response to an arbitrary input
signal using the convolution operator:

y[n]   x[k ]h[n  k ]
k  

The output signal y[n] can be calculated by:

• Sum of scaled signals – example 1
• Non-zero elements of h – example 2
The two ways of calculating the convolution are equivalent

Calculated in Matlab using the conv() function (but note that

there are some zero padding at start and end)

EE-2027 SaS, L4: 14/17

 Lecture 4: Exercises
Q2.1-2.7, 2.21

Calculate the answer to Example 3 in Matlab, Slide 14

EE-2027 SaS, L4: 15/17