Chapter 9:

Continuous-Time Signals and Systems

In Chapters 4−8, we focused on DT sequences, mainly

for effective computer representation. However, most
signals originate in continuous time.
In this chapter, we revisit many of the concepts
we have learned for DT signals and systems. There
will be many similarities, as well as some important
differences, compared to DT systems.
notation: x(t) where t is the independent variable

Objectives

 good bye to DT systems for a while

 the UNIT IMPULSE signal
 definition
 properties
 continuous-time (CT) signals and systems
 example systems
 review: Linearity and Time-Invariance
 convolution integral: impulse response
 evaluating convolutions (with examples)
 causality and stability
 cascaded/parallel connections of LTI systems

1
 Analog Signals x(t)

 INFINITE LENGTH
 SINUSOIDS: (t = time in secs)
 PERIODIC SIGNALS
 ONE-SIDED, e.g., for t>0 or for t<0
(could be right- or left-sided)
 UNIT STEP: u(t)
 FINITE LENGTH
 SQUARE PULSE
 IMPULSE SIGNAL: (t)

 DISCRETE-TIME: x[n] is list of numbers

2
 periodic vs. aperiodic (non-periodic) signals
Periodic signals are characterized by a repeating
pattern that repeats itself infinitely many times.
expressing this mathematically:

x(t + nTo) = x(t) ∀t ∈ R ∀n ∈ Z and some To

To: period of the signal (the duration of one cycle)

The shortest time interval over which the pattern
repeats itself is called the fundamental period. Its
reciprocal is the fundamental frequency fo.

CT signals: PERIODIC
sinusoidal signal x(t)  10cos(200 t)

square wave

INFINITE DURATION

3
 Special Input Signals:
Recall the unit-step sequence u[n] in DT:
(
1 for n ≥ 0
u[n] =
0 for n < 0
where n is an integer.
In CT, an important special input signal is the
unit-step signal/function defined in a similar way:
(
1 for t ≥ 0
u(t) =
0 for t < 0
time shifted unit-step signal:
(
1 for t ≥ to
u(t − to) =
0 for t < to
time shifted and scaled unit-step signal:
(
A for t ≥ to
Au(t − to) =
0 for t < to
• The step discontinuity can be anywhere (to) on
the time axis and the step can be of any height A.
• Apart from being an important special input sig-
nal to systems, u(t) is also very useful for compact
representation of signals (e.g., right-sided, left-sided,
finite-duration signals).
4
 CT signals: ONE-SIDED
unit-step signal
1 t  0
u(t)  
0 t  0

one-sided
sinusoid

“suddenly applied”
v(t)  e t u(t) exponential

CT signals: FINITE LENGTH

square pulse signal p(t)  u(t  2)  u(t  4)

sinusoid multiplied
by a square pulse

5
 In DT, another important special input signal was
the unit-sample sequence δ[n]:
(
1 for n = 0
δ[n] =
0 for n 6= 0
where n is an integer.
What is an impulse in CT?
As a first attempt, we try to create an impulse as
follows:
(
1 for t = 0
δ(t) = NO!
0 for t 6= 0
However, such a signal has no strength, no energy,
no intensity. If we integrate it, we get nothing.
Design δ(t) to be zero everywhere except at t = 0,
but also satisfying:
Z ∞
δ(t) dt = 1
−∞
Define the impulse to be the limit of a sequence of
CT functions that become increasingly concentrated
at one point (the origin):
(
1
for − ∆ < t < ∆
δ∆(t) = 2∆ YES!
0 otherwise

6
 • This is a rectangular pulse of width 2∆ and height
1
2∆ .
• Total area of δ∆(t) remains constant at 1 regard-
less of the size of ∆.
• As ∆ → 0, the pulse becomes narrower but higher.
In the limit as ∆ → 0:
lim δ∆(t) = δ(t)
∆→0
• Two important properties:
δ(t) = 0 for t 6= 0
Z ∞
δ(t) dt = 1
−∞
• δ(t) is not a function in the normal mathematical
sense. It is also called the Dirac delta.
• The δ(t) can be time shifted and scaled as needed.
More generally, one can have Aδ(t − to) which is an
impulse at t = to of area (or intensity) A:
Z ∞ Z ∞
Aδ(t) dt = A δ(t) dt = A
−∞ −∞
Z ∞ Z∞
Aδ(t − to) dt = A δ(t − to) dt = A
−∞ −∞
• A is called the area or the size or the intensity
of the impulse.
7
 What is an Impulse?
 A signal that is concentrated at one point.


   (t)dt  1 lim   (t)   (t)
0


Defining the Impulse

 Assume the properties apply to the limit:

lim   (t)   (t)
0
 One “INTUITIVE” definition is:
 (t)  0, t  0 concentrated at t=0

  ( )d  1 unit area



8
 Sampling Property of the Impulse:
Multiply an arbitrary CT signal or function f (t)
with the impulse δ(t):
f (t)δ(t) = f (t) lim δ∆(t)
∆→0
= lim f (t)δ∆(t)
∆→0
= lim f (0)δ∆(t)
∆→0
= f (0) lim δ∆(t)
∆→0
= f (0)δ(t)
When a CT signal f (t) is multiplied with the im-
pulse δ(t), the value of the CT signal at t = 0 is
sampled out of the signal, and we are left with only
an impulse whose area is scaled by f (0). The area of
the impulse conveys the sample value.
We can integrate the above to get the value of the
signal f (t) at t = 0:
Z ∞ Z ∞ Z ∞
f (t)δ(t) dt = f (0)δ(t) dt = f (0) δ(t) dt = f (0)
−∞ −∞ −∞
More generally, we have:
f (t)δ(t − to) = f (to)δ(t − to) (sampling property)
Z ∞
f (t)δ(t − to) dt = f (to) (sifting property)
−∞

9
Sampling Property of the Impulse

f (t)  (t)  f (0)  (t)

f (t) (t)  f (0) (t)

General Sampling Property

f (t) (t  t0 )  f (t0 ) (t  t0 )

10
 Relationship Between δ(t) and u(t):
generalized area property of the impulse:
Z b (
1 if a < 0 and b ≥ 0
δ(t) dt =
a 0 if a ≥ 0 or b < 0
(assume b > a)
The area is one when the impulse lies between the
limits of the integral.
Applying the generalized area property of the im-
pulse with the limits a = −∞ and b = t, we get:
(
t
if t ≥ 0
Z
1
δ(τ ) dτ = = u(t)
−∞ 0 if t < 0

From the Fundamental Theorem of Calculus,

if we have two arbitrary functions f (t) and g(t) such
that Z t
g(t) = f (τ ) dτ,
−∞
dg(t)
it follows that f (t) = dt .
Applying this theorem to our case above, we have
du(t)
δ(t) = .
dt
This relationship allows us to talk about the deriva-
tive of a function at discontinuities.
11
 More generally,
du(t − to)
δ(t − to) =
dt
Z t
δ(τ − to) dτ = u(t − to)
−∞
We can now take the derivative of x(t) = f (t)u(t).
(discontinuity at t = 0 even when f (t) differentiable)
dx(t) du(t) df (t) df (t)
= f (t) + u(t) = f (t)δ(t) + u(t)
dt dt dt dt
df (t)
= f (0)δ(t) + u(t)
dt

Properties of the Impulse

 (t  t0 )  0, t  t0 concentrated at one time


  (t  t

0 )dt  1 unit area

sampling property
f (t) (t  t 0 )  f (t 0 ) (t  t 0 )

extract one value of f(t)
 f (t) (t  t0 )dt  f (t0 )


du(t)
  (t) derivative of unit step
dt

12
 Continuous-Time Systems

input output

x(t) 
 y(t)
 examples:
 ideal delay y(t)  x(t  t d )
t
 integrator y(t)   x( )d

 modulator y(t)  [A  x(t)]cos  c t

CT Building Blocks

 DIFFERENTIATOR
 INTEGRATOR (CIRCUITS)

 IDEAL DELAY by to

 MODULATOR (e.g., AM radio)

 ADDER & MULTIPLIER

13
 Impulse Responses of Basic LTI Systems:
engineering reality:
In practice, we may have to resort to measuring a
system’s impulse response experimentally by using a
very short pulse that approximates the impulse.
integrator:
Z t
y(t) = x(τ ) dτ
−∞

Letting x(t) = δ(t) to get the impulse response:

Z t
h(t) = δ(τ ) dτ = u(t)
−∞
Integrator is an LTI system; therefore, convolution is
applicable:
y(t) = x(t) ∗ h(t) = x(t) ∗ u(t) = x(−1)(t)
Therefore, convolving any arbitrary CT signal x(t)
with u(t), we get its first anti-derivative or integral.
differentiator:
dx(t)
y(t) = = x(1)(t)
dt
Letting x(t) = δ(t) to get the impulse response:
dδ(t)
h(t) = = δ (1)(t) (doublet)
dt
14
Differentiator is also an LTI system; therefore, con-
volution is applicable:
(1) dx(t)
y(t) = x(t) ∗ h(t) = x(t) ∗ δ (t) = = x(1)(t)
dt
Therefore, convolving any arbitrary CT signal x(t)
with the doublet δ (1)(t), we get its first-order deriva-
tive.
What is the result of convolving u(t) with the dou-
blet?
(1) du(t)
u(t) ∗ δ (t) = = δ(t)
dt
ideal delay:
y(t) = x(t − to)
Letting x(t) = δ(t) to get the impulse response:
h(t) = δ(t − to)
Ideal delay system is also LTI; therefore, convolution
is applicable:
y(t) = x(t) ∗ h(t) = x(t) ∗ δ(t − to) = x(t − to)
Therefore, convolving any arbitrary CT signal x(t)
with a shifted impulse δ(t − to) delays the signal by
to.
x(t) ∗ δ(t − to) = x(t − to) (time-shift property)

15
 Convolution of Impulses:
Impulses do not have to be regularly spaced, unlike
in DT.
Using the time-shift property with x(t) = δ(t − t1)
and renaming to as t2, we get:

δ(t − t1) ∗ δ(t − t2) = δ(t − (t1 + t2))

ex: Given an LTI system with impulse response

h(t) = δ(t − 1) + 0.5 δ(t − 2), find the output y(t) of
the system and make a sketch of it.

y(t) = x(t) ∗ h(t)

= x(t) ∗ [δ(t − 1) + 0.5 δ(t − 2)]
= x(t) ∗ δ(t − 1) + x(t) ∗ 0.5 δ(t − 2)
= x(t − 1) + 0.5 x(t − 2)

This result is valid for any arbitrary input x(t).

16
For the specific input given as follows:

The output is:

17
 Ideal Delay System:

 mathematical definition:
y(t)  x(t  t d )
 to find the IMPULSE RESPONSE, h(t),
let x(t) be an impulse, so
h(t)   (t  td )

Output of Ideal Delay of 1 sec

x(t)  e t u(t)

y(t)  x(t 1)  e(t1)u(t  1)

18
 Integrator:

 useful model for a variety of CT systems

such as capacitors and op-amps
in electrical systems tt
 mathematical definition: y(t)   x( )d
y(t)   x( )d
running integral 

 to find the IMPULSE RESPONSE, h(t),
let x(t) be an impulse, so t

h(t)    ( )d  u(t)



Integrator:

 integrate the impulse

t

  ( )d  u(t)

 if t<0, we get zero
 if t>0, we get one
 by definition, this function is the unit step u(t).
 thus, the impulse response of the integrator is:
h(t) = u(t).

19
 Graphical Representation
du(t)
 (t) 
dt

t 1 t  0
u(t)    ( )d  
 0 t  0

t
y (t )   x( )d
Output of Integrator 
 x (t )  u (t )
x (t )  e 0.8t u(t )

t
0.8
y (t )  e u ( )d

 0 t0
t
  0.8
e u( )d t  0

 
 1.25(1  e 0.8t )u(t )

20
 Differentiator:

 mathematical definition:
dx(t)
y(t) 
dt
 to find h(t), let x(t) be an impulse, so
d (t)
h(t)    (1) (t) doublet
dt

dx(t)
y(t) 
Differentiator Output: dt
x (t )  e 2( t 1)u(t  1)

d 2( t 1)
y (t ) 
dt
e u(t  1) 
 2e 2( t 1)u(t  1)  e 2( t 1) (t  1)
 2e 2( t 1)u(t  1)  1 (t  1)

21
 Linear and Time-Invariant
(LTI) Systems

 If a continuous-time system is both linear and

time-invariant, then the output y(t) is related to
the input x(t) by a convolution integral


y(t)   x( )h(t   )d  x(t)  h(t)

where h(t) is the impulse response of the system.

A CT system is linear if when

x1(t) 7→ y1(t)
x2(t) 7→ y2(t),
it follows that
αx1(t) + βx2(t) 7→ αy1(t) + βy2(t) ∀α, ∀β, ∀t
A CT system is time invariant if when
x(t) 7→ y(t),
x(t − to) 7→ y(t − to) ∀to, ∀t
Test procedures for linearity and time invariance in
CT are very similar to those in DT.
22
 Testing for Linearity

y1 (t)
x1 (t)
w(t)
y2 (t)
x2 (t)
x1 (t)
w(t)
x(t) y(t) y(t)
x2 (t)

Testing Time-Invariance

x(t) x(t  t0 ) w(t) w(t) y(t  t0 )

t0

y(t) t 0 y(t  t )
0

23
 y ( t )  x ( t  td )
Ideal Delay:

 linear
ax1 (t  td )  bx2 (t  td )  ay1 (t )  by2 (t )

 and time-invariant
w(t )  x ((t  td )  t0 )
y (t  t0 )  x ((t  t0 )  td )

t
Integrator: y(t)   x( )d

 linear
t
 [ax1 ( )  bx 2 ( )]d  ay1 (t)  by2 (t)

 and time-invariant
t

w(t)   x(  t )d

0 let     t 0

tt 0

 w(t)   x( )d = y(t - t0 )



24
 Modulator: y(t)  [A  x(t)]cos  c t

 not linear, obvious because

[A  ax1 (t)  bx2 (t)] 
[A  ax1 (t)]  [A  bx2 (t)]
 not time-invariant
w(t)  [A  x(t  t0 )]cos  c t  y(t  t0 )

Convolution is Commutative

h(t )  x(t )   h( ) x(t   )d

let   t   and d   d

h(t )  x(t )    h(t   ) x( )d


  h(t   ) x( )d  x(t )  h(t )


25
 Convolution in CT:
For LTI systems in CT:
Z ∞
y(t) = x(τ )h(t − τ )dτ = x(t) ∗ h(t)
Z−∞
∞
= h(τ )x(t − τ )dτ = h(t) ∗ x(t)
−∞

Properties of Convolution:
• commutative: x(t) ∗ h(t) = h(t) ∗ x(t)
• associative:

[x(t) ∗ h1(t)] ∗ h2(t) = x(t) ∗ [h1(t) ∗ h2(t)]

• distributive over addition:

x(t) ∗ [h1(t) + h2(t)] = x(t) ∗ h1(t) + x(t) ∗ h2(t)

• identity element: x(t) ∗ δ(t) = x(t)

• ideal delay: x(t) ∗ δ(t − to) = x(t − to)
• When considered as an operator or transforma-
tion, convolution is linear and time invariant.

26
 Convolution is Linear

 substitute x(t)=ax1(t)+bx2(t)

y (t )   [ax1 ( )  bx2 ( )]h(t   )d

 
 a  x1 ( )h(t   )d  b  x2 ( )h(t   )d
 
 ay1 (t )  by2 (t )

Therefore, convolution is linear.

Convolution is Time Invariant

 substitute x(t-t0)


w(t)   h( )x((t   )  t )d o




  h( )x((t  t o )   )d



 y(t  to )

27
 Convolution of Impulses

 convolution of two impulses

 (t  t1 )   (t  t2 )   (t  t1  t2 )
 convolution of unit step and shifted impulse

u(t )   (t  t0 )  u(t  t0 )
 convolution of unit step and derivative of
impulse (doublet)
u(t)   (1) (t)   (t)

28
Evaluating a Convolution (ex.1)

x (t )  u(t  1) h (t )  e  t u (t )


y (t )   h( ) x(t   )d  h(t )  x(t )


“Flipping and Shifting”

x ( )

“flipping”
g( )  x( )  u(  1)

“flipping and shifting” g(  t)  x((  t))  x(t   )

t 1 t

29
 Evaluating the Integral

 0 t 1  0
t 1
y (t )   
 e d t 1  0

0

The Output:

t 1
  t 1
y (t )  e d   e
0
0
 1  e ( t 1) t 1

y (t )  0 t<1

30
 Convolution GUI

Evaluating a Convolution (ex.2)

x (t )  e  at u(t ) h(t )  e  bt u(t ), b  a


y (t )   x( )h(t   )d  x(t )  h(t )


  bt t a b
 e  e e d t0
  e a u( )e b ( t  )u(t   )d  
0
  0 t0

 e at  e bt
 t0 e at  e bt
  ab  u (t )
 b  a
 0 t0

31
 Special Case: b=0--->h(t)=u(t)
x (t )  e  at u(t ), a  0 h (t )  u (t )

y (t )   x( )h(t   )d  x(t )  h(t )

1
 (1  e at )u(t )
a
if a  2
1
y (t )  (1  e 2t )u(t )
2

Convolve two Unit Steps (a=b=0)

x (t )  u (t ) h (t )  u (t )

y (t )   x( )h(t   )d  x(t )  h(t )


 t
 1 d t  0
  u( )u(t   )d   
0
  0 t0

 t t  0
  t u(t ) unit ramp
0 t  0

32
33
34
Cascade of CT LTI Systems

 (t) h1 (t) h1 (t)  h2 (t)

h(t)  h1 (t)  h2 (t)  h2 (t)  h1 (t)

 (t) h2 (t) h2 (t)  h1 (t)

35
 ex:
Cascading an ideal unit delay system with an inte-
grator, both of which are LTI.

ex:
LTI system with impulse response

h(t) = u(t − 1) − u(t − 2)

Rewriting the impulse response as:

h(t) = u(t−1)−u(t−2) = u(t)∗[δ(t−1)−δ(t−2)],

we see that this is the cascade of an ideal integrator

followed by a parallel combination of two ideal delay
systems.

36
 Differentiation and Integration of Convolution:

dy(t)
y1(t) = = y(t) ∗ δ (1)(t)
dt
= [x(t) ∗ h(t)] ∗ δ (1)(t)
= x(t) ∗ [h(t) ∗ δ (1)(t)]
= x(t) ∗ h(1)(t)
dx(t)
y2(t) = ∗ h(t) = x(1)(t) ∗ h(t)
dt
We know that y1(t) = y2(t), therefore,
dy(t)
= y (1)(t) = x(1)(t) ∗ h(t) = x(t) ∗ h(1)(t)
dt
37
 • Differentiation of the input or the impulse re-
sponse differentiates the output of the convolution.
• Replacing the differentiators by integrators and
performing a similar analysis, we have:
Z t
y(τ ) dτ = y (−1)(t) = x(−1)(t) ∗ h(t) = x(t) ∗ h(−1)(t)
−∞

• Integrating the input or the impulse response inte-

grates the output of the convolution.
More generally,

y (m+n)(t) = x(m)(t) ∗ h(n)(t)

where m and n are integers such that:

−∞ < m < ∞ and − ∞ < n < ∞

m, n > 0 for derivatives

m, n < 0 for integrals
The integers m and n may have opposite signs.

38
 Stability:

• The impulse response of an LTI system completely

characterizes and defines the system properties.
• For a CT system, where x(t) 7→ y(t), we can de-
fine stability in two ways: a general definition based
on intuition and also as a precise mathematical con-
dition.
general definition of stability:
A system is stable if and only if every bounded input
produces a bounded output.
mathematical condition for stability:
A system is stable if and only if, for all bounded inputs
such that |x(t)| < B < ∞ for all t, it is also true
that |y(t)| < C < ∞ for all t, where B and C are
constants.
If we specialize to the case of CT LTI systems,
then a necessary and sufficient condition can be
based on the impulse response h(t) as follows:
A CT LTI system is stable if and only if:
Z ∞
|h(τ )|dτ < ∞
−∞

(absolute integrability of the impulse response)

39
 proof of the sufficiency part: ⇒
Z ∞ 
 
|y(t)| =  h(τ )x(t − τ )dτ 
Z −∞
∞
≤ |h(τ )||x(t − τ )|dτ
Z−∞∞
≤ |x(t − τ )||h(τ )|dτ
Z−∞∞
≤ B|h(τ )|dτ
−∞
Z ∞
≤ B |h(τ )|dτ
−∞

R∞
It follows that if −∞ |h(τ )|dτ < ∞, then, |y(t)| <
C < ∞, which means that the system is stable.
The proof that systems satisfying the absolute in-
tegrability condition of the impulse response are the
only LTI systems that are stable (the proof of the ne-
cessity part ⇐) is somewhat tricky and not included
here.
ex: ideal integrator
Is the ideal integrator LTI? Is it stable?
We know that for the ideal integrator, h(t) = u(t).
We also know that the integrator is LTI. Testing the

40
condition for absolute integrability:
Z ∞ Z ∞ Z ∞
|h(τ )|dτ = |u(τ )|dτ = dτ → ∞
−∞ −∞ 0
Therefore, the ideal integrator is an unstable sys-
tem. To make it stable, it is possible to include a loss
factor so that the impulse response is h(t) = e−atu(t)
(a small). This way, it becomes a non-ideal inte-
grator which is stable but lossy due to leakage.
ex: reciprocal system
1
y(t) =
x(t)
Is this system LTI? Is it stable?

Causality:

Causality is concerned with precedence in time.

general definition of causality:
A system is causal if and only if y(to), the output at
the present time to, does not depend on the values of
the input signal x(τ ) at times in the future, i.e., for
τ > to .
In other words, a causal system cannot anticipate
the future of the input signal.
For a CT LTI system with impulse response h(t),
41
the condition for causality simplifies to the following
test on the impulse response:
A CT LTI system is causal if and only if

h(τ ) = 0 for τ < 0 (∗)

necessity: ⇐
We can show that the system is non-causal if h(t) 6= 0
for t < 0. Since h(t) is the output when δ(t) is the
input, having h(t) 6= 0 for t < 0 means that the
output starts before the input does, which violates
the definition of causality.
sufficiency: ⇒
We need to examine the convolution integral expres-
sion for the output of an LTI system at a specific time
to: Z ∞
y(to) = x(τ )h(to − τ ) dτ (∗∗)
−∞
If (∗) is true, then h(to − τ ) is zero for (to − τ ) < 0
which is the same as

h(to − τ ) = 0 for τ > to

Therefore, the integrand in (∗∗) is zero for τ > to

and we can change the upper limit of the integral to

42
τ = to: Z to
y(to) = x(τ )h(to − τ ) dτ
−∞
This integral says that y(to) depends only on values
of the input x(τ ) for τ ≤ to.
Thus, we have demonstrated that (∗) is both nec-
essary and sufficient for causality of an LTI system.

ex: time-reversal (flipping) system

y(t) = x(−t) −∞<t<∞

This system “turns the signal around in time.” Equa-

tion (∗∗) does not apply to this case because the
system is not LTI (linear but not time invariant).
Therefore, we must resort to the general definition
of causality.
From the input/output relation, we see that for neg-
ative t values, −∞ < t < 0, the system requires a
value of the input at a future time. For example,
when t = −5, y(−5) = x(5), which is a future value
of the input. Therefore, the system is not causal.
ex: Is the ideal integrator a causal system?
ex: Is the ideal delay system causal?

43
 Stability

 a system is stable if every bounded input

produces a bounded output.
 a continuous-time LTI system is stable if
and only if

 h(t) dt  

(impulse response is absolute integrable)

Causal Systems

 a system is causal if and only if y(t0)

depends only on x() for t0 .

 an LTI system is causal if and only if

h(t )  0 for t  0

44