EE 220 : Signals and Systems

Department of Electronics and Electrical Engineerng

Indian Institute of Technology, Guwahati
Monsoon 2022

Topic 4 : Continous time LTI systems : Convolution

Instruction and notes by : Manish

1 Linear Time-Invariant (LTI) Systems

Systems that are linear as well as time invariant are commonly known as LTI systems. The
advantages of such systems is that they are easy to model and analyze; and thus play important
role in several engineering problems.
Let’s begin with understanding some common mathematical operations.

1.1 Convolution
For continuous time functions, convolution of two functions x and h, denoted x ∗ h, is defined
as Z ∞
x ∗ h(t) = x(τ )h(t − τ )dτ (1)
−∞
Convolution operation is used extensively in systems theory. The convolution result x ∗ h
evaluated at the point t is simply a weighted superposition on the function x, and the weighting
is given by h, time reversed and shifted by t.

1
 The area under the product signal (f · g) helps in determining the convolution integral.
Convolution is pretty useful. We will learn it in detail by seeing examples and learning its
properties. Let us begin with the basic properties.

• The convolution operation is commutative. That is, h ∗ x = x ∗ h

• The convolution operation is associative. That is, (x ∗ h1 ) ∗ h2 = x ∗ (h1 ∗ h2 )

• The convolution operation is distributive with respect to addition. That is, x∗(h1 +h2 ) =
x ∗ h1 + x ∗ h2

• For any function x, Z ∞

x ∗ δ(t) = x(τ )δ(t − τ )dτ = x(t) (2)
−∞

• Thus, δ is the convolution identity; i.e.

x∗δ =x

1.2 How to evaluate convolution between two functions

Performing convolution may not be always easy due to the involved calculus. There are usually
three methods that one may implement to evaluate convolution between two functions. Let’s
see them one by one.

1.2.1 Analytical method

When both the impulse response h(t) and the input signal x(t) have simple analytical descrip-
tions. Let’s directly see an example:
Example: Find the convolution of two unit-step functions.
Solution: Given: x(t) = u(t); and h(t) = u(t)
Z ∞
x∗h= u(τ )u(t − τ )dτ
−∞

Now, we know unit step function is zero for negative points on time axis and 1 otherwise. So
processing the first unit step function inside the integral:
Z ∞
x∗h= u(t − τ )dτ
0

Now once again, the integrand is zero for t − τ < 0 i.e. for τ > t. So we need to integrate only
between the limits 0 to t (and elsewhere it will be 0)
Z t Z t
x∗h= u(t − τ )dτ = 1dτ = t ∀ t > 0 and x ∗ h = 0 ∀ t≤0
0 0

or in a single line we can write it as:

x ∗ h = t · u(t) = r(t)

Example: Find the convolution between x(t)=1 and u(t).

2
 Answer: Z ∞
x∗h= x(τ )u(t − τ )dτ
−∞

but x(τ ) = 1: Z ∞
x∗h= u(t − τ )dτ
−∞

Once again, whenever τ > t integral would be zero. So we may change limits as well:
Z t
x∗h= 1 · dτ = [t]t−∞
−∞

Of course the conclusion is that x ∗ h in this case is not meaningful for any value of t ! This is
a reminder that convolution expressions must be checked to make sure they are meaningful.

1.2.2 Graphical method

Sometimes a graphical approach is easier for keeping track of the calculations. Basically, we
plot the two signals in the integrand, x(τ ) and h(t − τ ) , versus τ , for the value of t of interest.
Then multiplying the two signals provides the integrand, which is the net area under the curve.

Example: Consider two rectangular pulsesgiven below. Find the convolution between
them,

Solution: Lower limit of the integral (where area overlap begins) at the first graphical
touch point will be 0. So integral from 0 to t (for 0 < t < 2)
Z t
x∗h= 1/6 · dτ = t/6
0

The next step is to integrate between t=2 to t=3. the overlapped area is going to be constant
in this duration. (for 2 ≤ 3) Z 3
x∗h= 1/6 · dτ = 1/6
2

The next step is to integrate from (t-2) to t=3 (where overlapping ends). the overlapped area
is going to be constant in this duration. (for 3 < t < 5)
Z 3
x∗h= 1/6 · dτ = 1/6 · (5 − t)
t−2

The final function will be a trapezoid as shown in the figure.

3
1.2.3 Cleverness method
If one of the signals in the convolution can be written as a linear combination of simple, shifted
signals, then by the properties of linearity and time invariance, the response can be computed
from a single convolution involving the simple signals.
Example: In the previous example given above; x(t) can be written as:
x(t) = 1/2[u(t) − u(t − 2)];
Now, if the following convolution is well defined
Z ∞
ŷ(t) = u(τ )h(t − τ )dτ
−∞

then, we can solve the problem as

y(t) = 1/2[ŷ(t) − ŷ(t − 2)]

2 Convolution and LTI systems

Convolution operation can be utilized to evaluate impulse response and step response of any
system. Let’s see what is impulse response and step response.

2.1 Impulse response

The response of a system to an impulse input δ is called the impulse response of the system.
It is usually denoted by letter h.
For an LTI system (either continuous or discrete) with input x, output y, and impulse
response h, the following relationship holds:
y =x∗h
An LTI system can be completely characterized by its impulse response. It means that if
the impulse response of an LTI system is known, we can determine the response of the system
to any input. The impulse response is an extremely useful quantity and we often want to
determine this quantity.

4
Top row: Block diagram of a simple LTI system. Middle row: Series connections. Bottom row:
Parallel connections.

2.2 Step response

The response of a system to a unit step function (u) input is called the step response of the
system. It is usually denoted by the letter s.
Impulse response h and the step response s are related as:

ds(t)
h(t) =
dt
For a discrete time system, the equivalent equation will be:

h[n] = s[n] − s[n − 1]

The main use of knowing the step response of a system is that we can use it to determine
the impulse response using above equation. Impulse response h is the more important quantity
than the step response s for practical purposes.

3 Convolution of periodic functions

The convolution of two periodic functions is usually not well defined. An alternative notion of
convolution for periodic functions is known as periodic convolution. The periodic convolution
of the T-periodic functions x and h denoted x ⊛ h is defined as:
Z
x⊛h= x(τ )h(t − τ )dτ (3)
T
R
here, T
denotes integral over an interval of length T.

Reference material
1. Textbook: Signals and Systems by Simon Haykin
2. Lecture notes are inspired from the course materials of JHU 520.214; MIT 6.003; Purdue
ECE-301; UVic ECE-260; and Imperial College E2.5