ECE 45 Heath

Lecture 6

More convolution
Preview of today’s lecture
◆ Convolution
 Determine the output of an LTI system using the convolution
 Compute the continuous-time convolution between two signals

◆ Convolution properties
 Summarize key properties of convolution
 Leverage properties of convolution to simplify its calculations

2
Review
◆ Output of an LTI system is completely characterized by the impulse
response of the system

LTI system

◆ Input and output of an LTI system are related through convolution

 Connections back to ECE 45
Lectures 2 - 3 working with signals

Lectures 4 - 7 LTI systems in the time domain

Lectures 11-12 LTI systems in the frequency domain

Lectures 8 - 10 Fourier series

Lectures 13 - 17 Fourier transform Fourier
4
 Date Theme Topic Readings Out In
1 1/7 Signals Signals, systems, circuits and phasors 1.1 HW1
Rectangle, step functions, signal transformations, periodic, even and
2 1/9 Signals 1.2 HW2 HW1
odd
3 1/14 Signals Exponential, sinusoids, complex exponentials, Dirac Delta 1.3 - 1.4
4 1/16 LTI in time Systems, linearity, time invariance 1.6.5, 1.6.6 HW3 HW2
5 1/21 LTI in time Impulse response and convolution 2.2
6 1/23 LTI in time Convolution with a sinusoid, connection to phasors 2.2 HW4 HW3
7 1/28 LTI in time Convolution properties 2.3
8 1/30 Fourier series Fourier series 3.1 - 3.3 HW5 HW4
2/4 Midterm 1
9 2/6 Fourier series Fourier series convergence and properties 3.4 HW6 HW5
10 2/11 Fourier series Fourier series properties 3.5
11 2/13 LTI in frequency Frequency response of LTI systems 3.9 HW7 HW6
12 2/18 LTI in frequency Filters, bode plots 3.10, 6.2.3
13 2/20 Fourier transform Fourier transform 4.1-4.2 HW8 HW7
2/25 Midterm 2
14 2/27 Fourier transform Fourier transform properties 4.3 HW9 HW8
15 3/4 Fourier transform Rectangle and sinc functions 4.3
16 3/6 Fourier transform Convolution property 4.4 HW10
17 3/11 Fourier transform Multiplication property 4.5
18 3/13 Sampling Sampling theorem 7.1 HW10
3/19 Final exam Tuesday 3-6pm
5
 ECE 45 Heath

Convolution

Learning objectives
o Determine the output of an LTI system using the convolution
o Compute the continuous-time convolution between two signals
 No please not
convolution!!

https://commons.wikimedia.org/wiki/File:The_Scream.jpg with additions by Prof. Heath

Uncovering the convolution
◆ Consider an LTI system

h(t)

◆ The output can be computed from the convolution

Basic convolution properties
Shorthand notation
◆ Commutative

◆ Associative Choose option that makes it easy!

◆ Distributive

Use properties to simplify convolutions (more next lecture)

 This is a typical example using two finite length signals. Rectangles

CT convolution example #1 and triangle functions are common in examples / HW as they give
results that are easy to integrate. It is important here to
understand the different intervals in the convolutions.

◆ Find the output of a system with impulse response

1 t

for the input

1 t
CT convolution example #1: Graphical solution
◆ Plot one signal versus 
◆ Flip the second signal and shift it by t
 Here, plot it to the left of h()
 So plotted t has a negative value, usually

Important

◆ There are 4 intervals (why?):

CT convolution example #1: First interval

◆ First interval: there is no overlap!

CT convolution example #1: Second interval

◆ Second interval
CT convolution example #1:Third interval
◆ Third interval

Integrating the same

function, but with
different integration limits
CT convolution example #1: Fourth and final interval
◆ Fourth interval

◆ Sanity checks
 Check at t=1, the output should be the same for intervals 2 & 3 in
order for it to be continuous
 Same for t = 0 and t = 2 (should be zero there)
 Duration of output should be T1 + T2 = 1+1 = 2
CT convolution example #1: Plot of y(t)

0.5

0.4

y(t) 0.3

0.2

0.1

0
-1 -0.5 0 0.5 1 1.5 2 2.5 3

t
Length / duration of a convolution in continuous time

Length is T1 Length is T2

t T2 t
T1

Length is T1+T2

T1+T2
Length / duration of a convolution in continuous time

Length is T0+T1
Length is T2

-T0 t T2 t
T1

Length is T0 + T1+T2

-T0 T1+T2
Animation example

http://www.cse.yorku.ca/~asif/spc/ConvolutionIntegral_Final3.swf
 This is an example with a simple input that consists of a few delta

CT convolution example #2 functions. The convolution is easy to compute using the LTI
property of convolution and the definition of impulse response.

h(t)
◆ Determine and sketch the convolution of the following input

and system with impulse response

20
CT convolution example #2: Solution approach
◆ Recall the property

h(t)

◆ Therefore

h(t)
 CT convolution example #2: Solution sketch

x(t) 2
h(t)
1
1
y(t)
1
0 1 t
0 t
+ 0 1 2 t

2h(t-1)
2

0 1 2 3 t
 This is an important example involving the convolution of two

CT convolution example #3 rectangles of different widths. The solution will be different

depending on the widths of the rectangles. Notice what happens in
each interval.

◆ Determine and sketch , where

𝑥(𝑡)
1

0 1

h(𝑡)
1

0 α
23
CT convolution example #3: Solution
𝑥(𝜏)
ℎ(−τ)
1
1

0 1 𝜏 −α 0 𝜏
𝑡<0
ℎ 𝑡 − τ = ℎ(− 𝜏 − 𝑡 )

𝑡−α 𝑡 𝜏

𝑡 ∈ [0, 𝛼] ℎ 𝑡−τ

𝑡−α 𝑡
𝜏
33
 CT convolution example #3: Solution
𝑥(𝜏)

0
1 𝜏

ℎ 𝑡−τ
𝑡 ∈ [𝛼, 1]

𝑡−α 𝑡
𝜏
ℎ 𝑡−τ
𝑡 ∈ [1, 1 + 𝛼]

𝑡−α 1 𝑡
𝜏
34
 CT convolution example #3: Solution
𝑥(𝜏)

0 1 𝜏
𝑡 ∈ [1 + 𝛼, ∞]

ℎ 𝑡−τ

𝑡−α 𝑡
𝜏

35
CT convolution example #3: Solution (summary)

𝑥(𝑡) h(𝑡)
1 1

0 1 0 α

27
 This is an important example of convolution to causal signals

CT convolution example #4 together. It will be solved by using the graphical approach to help
set up the integrals.

◆ Determine and sketch , where

𝑥 𝑡 = 𝑒 −𝑎𝑡 𝑢 𝑡 , 𝑎>0 ℎ 𝑡 =𝑢 𝑡

28
 CT convolution example #4: Solution

1
𝑢(𝑡 − 𝜏) 𝑡<0

1
𝑒 −𝑎𝜏

29
 CT convolution example #4: Solution
For 𝑡 < 0
1 𝑢(𝑡 − 𝜏)

1 𝑒 −𝑎𝜏

30
 CT convolution example #4: Solution
For 𝑡 ≥ 0

𝑢(𝑡 − 𝜏)
1

1 𝑒 −𝑎𝜏

31
CT convolution example #4: Solution

1/a (e.g. for a>0)

t
32
 This is an example where an anti-causal and a causal signal are

CT convolution example #5 convolved together, unlike the previous examples. Notice the
difference intervals in this case.

◆ Determine and sketch , where

𝑥 𝑡 = 𝑒 2𝑡 𝑢 −𝑡 ℎ 𝑡 =𝑢 𝑡−3

33
 CT convolution example #5: Solution

For t-3 < 0

1
𝑢(𝑡 − 𝜏)

𝜏
𝑡−3

𝑒 2𝑡 𝑢(𝜏)

𝜏
0

34
 CT convolution example #5: Solution

For t-3 > 0

1
𝑢(𝑡 − 𝜏)

𝜏
𝑡−3

𝑒 2𝑡 𝑢(𝜏)

𝜏
0

35
CT convolution example #5: Solution

36
Basic convolution properties
◆ Commutative

◆ Associative

◆ Distributive

Use properties to simplify convolutions

Words of wisdom on convolution
◆ Convolution is a fact of life
 All real world LTI systems generate outputs in the time domain given
by the convolution sum or integral (yes RLC circuits are doing
convolution for you)

◆ You need to practice convolutions

 There are many examples in the book and videos online

◆ In future lectures, we will cover tools and ideas that allow us to

avoid computing convolutions (most of the time)
 But you still need to learn how to do it in the time domain (don’t wait)
38
 ECE 45 Heath

How does research in ECE (especially signal

processing) get done at a university?

Or do professors really write all of those papers they publish?

 Composition of authors on a paper

Graduate research assistant (GRA) Industry collaborator (recent PhD)

Professor Industry collaborator (lab director)

Percent effort
75% 15% 8% 2%
40
 Summary of system

$
National Science Foundation Professor Papers

Department of Defense
$ (summer salary)

Code
Industry $ GSR(s)
(salary, tuition)

Donations
$ Demos
Overhead
(staff, rooms,?) Trained
Students
graduates
University

Note that GSRs (PhD bound) are paid to do the research and attend school
41
How the work gets done 1/2
◆ Professors
 Identify technical directions
 Raise funding to support research
 Provide input, on publications and presentations
 General mentoring
◆ GRAs
 Understand the state-of-the-art
 Identify problems that can be solved Type of work that may be done by a USR
 Develop solutions, simulate results, build prototype, etc
 Create conference paper, journal paper, dissertation, etc

42
 How the work gets done 2/2
◆ Industry collaborators
 Give insight into the state-of-the-art
 Suggest important problems to solve
 Fund the research (sometimes)

◆ Academic collaborators (usually Professor or Professor + GRA)

 Contribute to the development of the research, e.g. bringing a
complementary set of expertise or domain knowledge
 Collaborate on the research results

Collaboration is increasingly important to identify and solve important problems

43
 ECE 45 Heath

Convolution properties

Learning objectives
o Summarize key properties of convolution
o Exploit the properties to simplify convolution
Basic convolution properties
Shorthand notation
◆ Commutative

◆ Associative Choose option that makes it easy!

◆ Distributive

Use properties to simplify convolutions

Recall this convolution
◆ Find the output of a system with impulse response

1 t

for the input

1 t
Solution to the convolution

0.5

0.4

0.3

0.2

0.1

0
-1 -0.5 0 0.5 1 1.5 2 2.5 3

For later use, let us refer to this convolution as

47
Use the properties to solve the following integral

1 t 1 t
Use the properties to convolve the following

1 t -2 -1 t
Use the properties to convolve the following

1 t -2 -1 1 2 t
Use the properties to convolve the following

1 t -2 -1 1 2 t
Summary about properties
◆ Convolution satisfies commutative, associative and distributed
properties
 Properties follow primarily from the integral structure
 Use these properties to simplify the calculation of a convolution

52