316 SCP

BIOEN316BiomedicalSignalsandSensors
Spring2015
Solvingconvolutionproblems
PART I: Using the convolution integral

The convolution integral is the best mathematical representation of the physical process
that occurs when an input acts on a linear system to produce an output. If x(t) is the input,
y(t) is the output, and h(t) is the unit impulse response of the system, then continuoustime
convolution is shown by the following integral.
y(t )
x( )h(t )d
In it, is a dummy variable of integration, which disappears after the integral is evaluated.

Example 1: unit step input, unit step response
Let x(t) = u(t) and h(t) = u(t).
y(t )
u( )u(t )d
The challenging thing about solving these convolution problems is setting the limits on t
and . I usually start by setting limits on in terms of t, then using that information to set
limits on t.
The unit step function u() makes the integrand zero for < 0, so the lower bound is 0.
The unit step function u(t-) makes the integral zero for > t, so the upper bound is t.
Once we have used the step functions to determine the limits, we can replace each step
function with 1.
t
y(t ) 1 d
0
This integral produces y(t) = t. However, when we used t to set a limit on , we also created
a limit on t. In this case, y(t) is zero when t < 0, because we have already set 0 < < t and
there is no that satisfies 0 < < 0. Therefore, the answer to the convolution problem is
t , t 0
y(t )
or y(t ) t u(t )
0, t 0
A system with h(t) = u(t) is known as an integrator, and can be made from an amplifier
with a capacitor in it, or pretty much any system that accumulates an input and does not
leak.

Printdate:4/18/2013
Spring2015
Example 2: Unit step input, inverse x response

Let x(t) = u(t) and h(t) = u(t)/(t+1). Convolution is commutative, so we can write the
integral in either of these two ways.
y(t )
u( )
u(t )
d
1
u(t )
u( ) t 1 d
The version on the left looks easier, so lets try it.

The unit step function u() makes the integrand zero for < 0, so the lower limit is 0.
The unit step function u(t) makes the integrand zero for > t, so the upper limit is t.
Once we have used the step functions to determine the limits, we can replace each step
function with 1.
t
y(t )
1
d
1

This integral produces y(t) = ln(t+1). However, the fact that the integrand is nonzero only
when 0 < < t means that y(t) is zero when t < 0. Therefore, the solution is
y(t) = ln(t+1)u(t).

Example 3: pulse input, unit step response.
Let x(t) = u(t) u(t2), h(t) = u(t).
y(t )
[u( ) u( 2)]u(t ) d
The integrand is zero when < 0 and > 2, so that at most the integrand is nonzero when
0 < < 2. It is also zero when > t. This sets up three intervals for t. First, when t < 0 there
is no way that 0 < < 2 and < t. Therefore, for t < 0, y(t) = 0. Next, when 0 < t < 2, the
integrand is 1 when 0 < < t, making 0 and t the limits of integration. Finally, when t > 2,
the value of t does not matter any more and the limits of integration are 0 and 2. Thus:

t 0 y(t ) 0
t
0t 2
y(t ) 1 d t
0
2
2t
y(t ) 1 d 2
0
Printdate:4/18/2013
Spring2015
Problem 1:
Use the convolution integral to find the y(t) = u(t) * exp(t)u(t), where x*h represents the
convolution of x and h.

PART II: Using the convolution sum
The convolution summation is the way we represent the convolution operation for
sampled signals. If x(n) is the input, y(n) is the output, and h(n) is the unit impulse
response of the system, then discrete time convolution is shown by the following
summation.
y(n)
x( k )h(n k )
In it, k is a dummy variable, which disappears when the summation is evaluated.

Discrete signals or functions are often sequences of numbers that are pretty easy to write
in a table, but are not easy to write as a function. A good example is a sequence { 1 1 }, i.e.
h(0) = 1, h(1) = 1, and h(n) = 0 everywhere else. The bold number indicates where n=0.
This could be written as the sum of three step functions, or as two step functions times a
cosine, or something else complicated, or it could just be written { 1, 1 }. Therefore, it
makes the most sense to convolve these signals in a table or graphically rather than as
functions in a sum. A summation with step functions is shown in example 1; a summation
with a table is shown in example 2, and a graphical example is shown in the Lecture 9
slides, starting with slide xx (TBD for 2013).

Example 1: unit step input, unit step response
Let x(n) = u(n) and h(n) = u(n).
y(n)
u( k ) u(n k )
As with the continuous transform, start by setting the limits on n and k. For k < 0 and for
k > n, u(k) = 0. Therefore, the summation has a lower limit of 0 and an upper limit of n. For
0 < k < n, the summand is one.
y(n)
k 0
Of course, when n < 0 then the limits do not permit any summing at all, so y(n) = 0. When
n 0, the sum is n. Therefore, the answer is y(n) = n u(n).

Printdate:4/18/2013
Spring2015
Example 2: input is {1, 1} pulse, response is {+1, 1}

Remember we are summing over k. The input is zero for k < 0, so the lower limit is 0. The
system response is zero for k n, so the upper limit is n, and the output is zero for n < 0.
Therefore we start summing for n = 0.

n
k=0
k=1
sum
0
x(0)h(00)
x(1)h(01)
1
= 1*1 = 1
=1*0 = 0
1
x(0)h(10)
x(1)h(11)
0
= 1*(1) = 1
=1*1=1
2
x(0)h(20) = 1*0 = 0
x(1)h(21) = 1*(1) = 1
1
3
x(0)h(30) = 1*0 = 0
x(1)h(31) = 1*0 = 0
0

The result is that y(n) = { 1 0 1 }, starting at n=0.

Problem 2.
Find the output y(n) = x(n)*h(n), where x(n) = { 1, 1 } and h(n) = { 3, 2, 1 }.
Both x(n) and h(n) are zero for n < 0, and the bold number shows where n = 0.
You may use either a table or the graphical method for finding y(n).

Printdate:4/18/2015

316 SCP

Uploaded by

Copyright:

Available Formats

316 SCP

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

316 SCP

Uploaded by

Copyright:

Available Formats

BIOEN316BiomedicalSignalsandSensors

PART I: Using the convolution integral

Example 2: Unit step input, inverse x response

The version on the left looks easier, so lets try it.

In it, k is a dummy variable, which disappears when the summation is evaluated.

Example 2: input is {1, 1} pulse, response is {+1, 1}

You might also like