book 2012/2/17 1:44 page 1 #1

P T

E R

Signals

Contents
1-1
1-2
1-3
1-4
1-5

Overview, 2
Types of Signals, 3
Signal Transformations, 6
Waveform Properties, 9
Nonperiodic Waveforms, 11
Signal Power and Energy, 21
Chapter 1 Summary, 24
Problems, 25

Objectives
Learn to:

Perform transformations on signals.

Use step, ramp, pulse, and exponential waveforms

to model simple signals.

Model impulse functions.

Calculate power and energy contents of signals.

Signals come in many forms: continuous, discrete, analog,

digital, periodic, nonperiodic, with even or odd symmetry
or no symmetry at all, and so on. Signals with special
waveforms include ramps, exponentials, and impulses.
This chapter introduces the vocabulary, the properties, and
the transformations commonly associated with signals in
preparation for exploring in future chapters how signals
interact with systems.

book 2012/2/17 1:44 page 2 #2

CHAPTER 1

Overview

SIGNALS

This book is about how signals interact with systems. More

precisely, it is about how a system transforms input signals
(excitations) into output signals (responses) to perform a
certain operation (or multiple operations). A system may
be as simple as the voltage divider in Fig. 1-1, wherein
the divider scales down input voltage i to output voltage
o = [R2 /(R1 + R2 )]i , or as complex as a human body
(Fig. 1-2). Actually, the human body is a system of systems;
it includes the respiratory, blood circulation, and nervous
systems, among many others. Each can be modeled as a system
with one or more input signals and one or more output signals.
When a persons fingertip touches a hot object (Fig. 1-3), a
nerve ending in the finger senses the elevated temperature
and sends a message (input signal) to the central nervous

Respiratory
system

R1
i

+
R2

Figure 1-1: A voltage divider is a simple system.

system (CNS), consisting of the brain and spinal cord. Upon

processing the input signal, the CNS (the system) generates
several output signals directed to various muscles in the persons
hand, ordering them to remove the finger away from the hot
object.

Nervous
system

Muscular/skeletal
system

Circulatory
system

Immune
system

Excretory
system

Reproductive
system

Endocrine system

Digestive system

Figure 1-2: The human body is a system of systems.

book 2012/2/17 1:44 page 3 #3

1-1 TYPES OF SIGNALS

SYSTEM

Imager

SYSTEM

Signal denoising

SYSTEM

Image
deblurring

Figure 1-3: Finger-CNS-muscle communication.

SYSTEM

By modeling signals and systems mathematically, we can use

the system model to predict the output resulting from a specified
input. We can also design systems to perform operations of
interest. A few illustrative examples are depicted in Fig. 1-4.
Signals and systems are either continuous or discrete. Both
types are treated in this book, along with numerous examples
of practical applications.
To set the stage for a serious study of signals and systems
and how they interact with one another, we devote the
current chapter to an examination of the various mathematical
models and attendant properties commonly used to characterize
physical signals, and then we follow suit in Chapter 2 with a
similar examination for systems.

Heart monitor

SYSTEM

Music transcriber
400
100

300
200

OVEN

350 degrees

Temperature control

1-1.1 Continuous vs. Discrete

Figure 1-4: A system transforms a continuous input signal x(t)

into an output signal y(t) or a discrete input signal x[n] into
a discrete output signal y[n]. Such system transformations
exist not only in the few examples shown here but also
in countless electrical, mechanical, biological, acoustic, and
financial domains, among many others.

The acoustic pressure waveform depicted in Fig. 1-5(a) is a

continuous-time signal carrying music between a source (the
trumpet) and a receiver (the listeners ear). The waveform
varies with both spatial location and time. At a given instant in
time, the waveform is a plot of acoustic pressure as a function
of the spatial dimension x, but to the listeners eardrum, the
intercepted waveform is a time-varying function at a fixed value
of x.

! Traditionally, a signal has been defined as any quantity

that exhibits a variation with either time, space, or both.
Mathematically, however, variation of any quantity as a
function of any independent variable would qualify as a
signal as well. "

1-1 Types of Signals

65 beats/minute

book 2012/2/17 1:44 page 4 #4

CHAPTER 1

Acoustic pressure waveform

(a) Continuous-time signal

Brightness across discrete row of pixels

(b) Discrete-spatial signal
Percent
20

% of total unemployed by age group

15

SIGNALS

a time record of blood pressure, or fuel consumption per

kilometer as a function of car speed wherein each is plotted
as a function of an independent variableor it may represent
a non-physical quantity such as a stock market index or the
distribution of unemployed workers by age group [Fig. 1-5(c)].
Moreover, in some cases, a signal may be a function of two
or more variables, as illustrated by the two-dimensional (2-D)
X-ray image in Fig. 1-5(d).

1-1.2

Causal vs. Noncausal

Real systemsas opposed to purely conceptual or mathematical constructs that we may use as learning tools even though we
know they cannot be realized in practiceare called physically
realizable systems. When such a system is excited by an
input signal x(t), we usually define the time dimension such
that t = 0 coincides with when the signal is first introduced.
Accordingly, x(t) = 0 for t < 0, and it is called a causal signal.
By extension, if x(t) = 0 for t < 0, it is called noncausal, and
if x(t) = 0 for t > 0, it is called anticausal.
! A signal x(t):
causal if x(t) = 0 for t < 0 (starts at t = 0)

10

noncausal if x(t) = 0 for any t < 0

(starts before t = 0)

5
16-19 20-24 25-29 30-34 35-44 45-54 55-64 65+
Age group
(c) Independent variable is age group

X-ray image

(d) 2-D spatial signal

Figure 1-5: Examples of continuous and discrete signals.

In contrast with the continuous-time signal shown in Fig. 1-5(a),

the brightness variation across the row of pixels on the computer
display of Fig. 1-5(b) constitutes a discrete-space signal
because the brightness is specified at only a set of discrete
locations. In either case, the signal may represent a physical
quantity, such as the altitude profile of atmospheric temperature,

anticausal if x(t) = 0 for t > 0

(ends at or before t = 0) "
Even though (in practice) our ultimate goal is to evaluate the
interaction of causal signals with physically realizable systems,
we will occasionally use mathematical techniques that represent
a causal signal in terms of artificial constructs composed of sums
and differences of causal and anticausal signals.

1-1.3 Analog vs. Digital

Consider an electronic sensor designed such that its output
voltage is linearly proportional to the air pressure p
surrounding its pressure-sensitive capacitor. If the sensors
output is recorded continuously as a function of time
[Fig. 1-6(b)], the resulting voltage record (t) would be
analogous to the pattern of the actual air pressure p(t). Hence,
(t) is regarded as an analog signal representing p(t). The term
analog (short for analogue) conveys the similarity between the
measured signal and the physical quantity it represents. It also
implies that because both and t are continuous variables, the

book 2012/2/17 1:44 page 5 #5

1-1 TYPES OF SIGNALS

5
which the dependent variable continues to enjoy infinite
resolution in terms of its own magnitude but not along the
independent variable t.

C
40

Atmospheric temperature

T(t) 50
t

0
(a) Atmospheric temperature in C
V
14
12
10
(t) 8
6
4
2
0

(t)
t
(b) Sensor voltage in volts

V
14
12
10
[n] 8
6
4
2
0

11

13 12

9 10

[n]

8
3

1 2 3 4 5 6 7 8 9 10 11 12
(c) Discrete version of (b)

1011 1101 1100 0100 0011 0000 0001 0101 1001 1010 0111 0110

1
0
(d) Digital signal
Figure 1-6: The atmospheric temperature temporal profile in (a)
is represented in (b) by the continuous signal (t) measured by a
pressure sensor. The regularly spaced sequence [n] in (c) is the
discrete version of (t). The discrete signal [n] is converted
into a digital sequence in (d) using a 4-bit encoder.

resolution associated with the recorded (t) is infinite along

both dimensions.
Had the pressure sensor recorded at only a set of equally
spaced, discrete values of time, the outcome would have looked
like the discrete-time signal [n] displayed in Fig. 1-6(c), in

! To distinguish between a continuous-time signal (t)

and a discrete-time signal [n], the independent variable t
in (t) is enclosed in curved brackets, whereas for
discrete-time signal [n], the index n is enclosed in square
brackets. "
If, in addition to discretizing the signal in time, we were to
quantize its amplitudes [n] using a 4-bit encoder, for example,
we would end up with the digital discrete-time signal shown in
Fig. 1-6(d). By so doing, we have sacrificed resolution along
both dimensions, raising the obvious question: Why is it that the
overwhelming majority of todays electronic and mechanical
systemsincluding cell phones and televisionsperform their
signal conditioning and display functions in the digital domain?
The most important reason is so that signal processing can be
implemented on a digital computer. Computers process finite
sequences of numbers, each of which is represented by a finite
number of bits. Hence, to process a signal using a digital
computer, it must be in discrete-time format, and its amplitude
must be encoded into a binary sequence.
Another important reason for using digital signal processing
has to do with noise. Superimposed on a signal is (almost
always) an unwanted random fluctuation (noise) contributed
by electromagnetic fields associated with devices and circuits
as well as by natural phenomena (such as lightning). Digital
signals are more immune to noise interference than their analog
counterparts.
The terms continuous-time, discrete-time, analog, and digital
can be summarized as follows:
A signal x(t) is analog and continuous-time if both x and
t are continuous variables (infinite resolution). Most realworld signals are analog and continuous-time (Chapters 1
through 6).
A signal x[n] is analog and discrete-time if the values
of x are continuous but time n is discrete (integer-valued).
Chapters 7 and 8 deal with analog discrete-time signals.
A signal x[n] is digital and discrete-time if the values of x
are discrete (i.e., quantized) and time n also is discrete
(integer-valued). Computers store and process digital
discrete-time signals. This class of signals is outside the
scope of this book.
As was stated earlier, a signals independent variable may
not always be time t, and in some cases, the signal may depend

book 2012/2/17 1:44 page 6 #6

CHAPTER 1

SIGNALS

6
4
x(t + 10)

10

x(t)

0
0

x(t 10)

10

14

20

30

t (s)

Figure 1-7: Waveforms of x(t), x(t 10), and x(t + 10). Note that x(t 10) reaches its peak value 10 s later than x(t), and x(t + 10)
reaches its peak value 10 s sooner than x(t).

on more than one variable (as in 2-D images and 3-D X-ray
tomographs). Nevertheless, in the interest of brevity when
introducing mathematical techniques, we will use the symbol t
as our independent variable exclusively. This does not preclude
using other, more appropriate, symbols when applying the
techniques to specific applications, nor does it limit expanding
the formulation to 2-D or 3-D when necessary.

a manufacturing robot, using the information extracted from

multiple input signals.

Concept Question 1-1: What is the difference between

y(t) = x(t T ),

a continuous-time signal and a discrete-time signal?

Between a discrete-time signal and a digital signal?
Concept Question 1-2: What is the definition of a causal

signal? Noncausal signal? Anticausal signal?

1-2 Signal Transformations

! A system transforms an input signal into an output
signal. The transformation may entail the modification
of some attribute of the input signal, the generation of
an entirely new signal (based on the input signal), or the
extraction of information from the input signal for display
or to initiate an action. "
For example, the system may delay, compress or stretch out the
input signal, or it may filter out the noise accompanying it. If the
signal represents the time profile of a cars acceleration a(t), as
measured by a microelectromechanical sensor, the system may
perform
an integration to determine the cars velocity: (t) =
!t
a(
)
d
. In yet another example, the system may be an
0
algorithm that generates signals to control the movements of

1-2.1 Time-Shift Transformation

If x(t) is a continuous-time signal, a time-shifted version with
delay T is given by
(1.1)

wherein t is replaced with (t T ) everywhere in the expression

and/or plot of x(t), as illustrated in Fig. 1-7. If T > 0, y(t)
is delayed by T seconds relative to x(t); the peak value of the
waveform of y(t) occurs T seconds later in time than does the
peak of x(t). Conversely, if T < 0, y(t) is advanced by T
seconds relative to x(t), in which case the peak value of y(t)
occurs earlier in time.
! While preserving the shape of the signal x(t), the timeshift transformation x(t T ) is equivalent to sliding the
waveform to the right along the time axis when T is
positive and sliding it to the left when T is negative. "

1-2.2 Time-Scaling Transformation

Figure 1-8 displays three waveforms that are all similar (but
not identical) in shape. Relative to the waveform of x(t), the
waveform of y1 (t) is compressed along the time axis, while
that of y2 (t) is expanded (stretched out). Waveforms of signals
y1 (t) and y2 (t) are time-scaled versions of x(t):
y1 (t) = x(2t),

(1.2a)

book 2012/2/17 1:44 page 7 #7

1-2

SIGNAL TRANSFORMATIONS

10

y1(t) = x(2t) x(t)

y2(t) = x(t / 2)

x(t)

"

5t for 0 t 2
0 otherwise
"
5 2t = 10t for 0 2t 2
y1 (t) =
0
otherwise
"
5t/2 = 2.5t for 0 t/2 2
y2 (t) =
0
otherwise
x(t) =

10

10

The time-reversal transformation is expressed as

and

y(t) = x(t),

(1.2b)

Mathematically, the time-scaling transformation can be

expressed as
y(t) = x(at),
(1.3)
where a is a compression or expansion factor depending
on whether its absolute value is larger or smaller than 1,
respectively. For the time being, we will assume a to be positive.
As we will see shortly in the next subsection, a negative value
of a causes a time-reversal transformation in addition to the
compression/expansion transformation.
! Multiplying the independent variable t in x(t) by a
constant coefficient a results in a temporally compressed
replica of x(t) if |a| > 1 and by a temporally expanded
replica if |a| < 1. "

1-2.3 Time-Reversal Transformation

! Replacing t with t in x(t) generates a signal y(t)
whose waveform is the mirror image of that of x(t) with
respect to the vertical axis. "

(1.4)

and is illustrated by the waveforms in Fig. 1-9.

1-2.4
y2 (t) = x(t/2).

x(t)

Figure 1-9: Waveforms of x(t) and its time reversal x(t).

Figure 1-8: Waveforms of x(t), a compressed replica given

by y1 (t) = x(2t), and an expanded replica given by y2 (t) =
x(t/2).

Combined Transformation

The three aforementioned transformations can be combined

into a generalized transformation:
y(t) = x(at b)
# #
$$
b
=x a t
a
= x(a(t T )),

(1.5)

where T = b/a. We recognize T as the time shift and a as

the compression/expansion factor. Additionally, the sign of a
( or +) denotes whether or not the transformation includes a
time-reversal transformation.
The procedure for obtaining y(t) = x(a(t T )) from x(t)
is as follows:
(1) Scale time by a:
If |a| < 1, then x(t) is expanded.
If |a| > 1, then x(t) is compressed.
If a < 0, then x(t) is also reflected.
This results in z(t) = x(at).

book 2012/2/17 1:44 page 8 #8

CHAPTER 1

(2) Time shift by T :

x(t)

If T > 0, then x(t) shifts to the right.

If T < 0, then x(t) shifts to the left.

This results in z(t T ) = x(a(t T )) = y(t).

3 2 1

t (s)

z(t) = x(2t)

(1) Time shift by b.

(2) Time scale by a.

e 2t
0

2 1

For signal x(t) profiled in Fig. 1-10(a), generate the

corresponding profile of y(t) = x(2t + 6).

y(t) = x 2 t

6
2

= x(2(t 3)).

t (s)

2 1

$$

Compression factor

y(t)

Solution:
We start by recasting the expression for the dependent variable
into the standard form given by Eq. (1.5),
#

(b) z(t)

Example 1-1: Multiple Transformations

t (s)

(c) y(t)
Figure 1-10: Waveforms of Example 1-1.

Time-shift

We need to apply the following transformations:

(1) Scale time by 2t: This causes the waveform to reflect
around the vertical axis and then compresses time by a factor
of 2. These steps can be performed in either order. The result,
z(t) = x(2t), is shown in Fig. 1-10(b).
(2) Delay waveform z(t) by 3 s: This shifts the waveform to
the right by 3 s (because the sign of the time shift is negative).
The result, y(t) = z(t 3) = x(2(t 3)), is displayed in
Fig. 1-10(c).

_t

(a) x(t)

The procedure for obtaining y(t) = x(at b) from x(t)

reverses the order of time scaling and time shifting:

Reversal

SIGNALS

If signal y(t) is obtained from x(t)

by applying the transformation y(t) = x(4t 8),
determine the values of the transformation parameters a
and T .

Exercise 1-1:

Answer:

a = 4 and T = 2. (See

Concept Question 1-3: Is the shape of a waveform

Exercise 1-2: If x(t) = t 3 and y(t) = 8t 3 , are x(t) and

altered or preserved upon applying a time-shift

transformation? Time-scaling transformation? Timereversal transformation?

Answer: Yes, because y(t) = 8t 3 = (2t)3 = x(2t).

(See )

y(t) related by a transformation?

book 2012/2/17 1:44 page 9 #9

1-3 WAVEFORM PROPERTIES

x(t)

Exercise 1-3: What type of transformations connect

x(t) = 4t to y(t) = 2(t + 4)?

10

Answer: y(t) = x( 21 (t + 4)), which includes a

time-scaling transformation with a factor of 1/2 and a
time-shift transformation with a time advance of 4 s.
(See )

x(t)

1-3 Waveform Properties

1-3.1 Even Symmetry

1 x(t)
2

In contrast, the waveform in Fig. 1-11(c) has odd symmetry.

Equivalently,

xo(t)

1-3.2 Odd Symmetry

1 x(t)
2

4
0

1 x(t)
2

5
(c) xo(t)

Figure 1-11: Signal x(t) and its even and odd components.

(odd symmetry).

(1.7)

A signal has odd symmetry if reflection about the vertical axis

(followed by reflection about the horizontal axis) leaves its
waveform unaltered. Examples of odd signals include sin(t)
and t n for odd integer n.

0
(b) xe(t)

A signal has even symmetry if reflection about the vertical axis

leaves its waveform unaltered.
The signal displayed in Fig. 1-11(b) is an example of a
waveform that exhibits even symmetry. Other examples include
cos(t) and t n for even integers n.

x(t) = x(t)

1 x(t)
2

(1.6)

! A signal exhibits odd symmetry if the shape of its

waveform on the left-hand side of the vertical axis is the
inverted mirror image of the waveform on the right-hand
side. "

xe(t)

The shape of the waveform on the left-hand side of the vertical

axis is the mirror image of the waveform on the right-hand side.
Mathematically, a signal x(t) has even symmetry if
(even symmetry).

10

(a) x(t)

! A signal x(t) exhibits even symmetry if its waveform

is symmetrical with respect to the vertical axis. "

x(t) = x(t)

We should note that if signal y(t) is equal to the product of

two signals, namely
y(t) = x1 (t) x2 (t),

(1.8)

then y(t) will exhibit even symmetry if x1 (t) and x2 (t) both
exhibit the same type of symmetry (both even or both odd)

book 2012/2/17 1:44 page 10 #10

10

CHAPTER 1

SIGNALS

Table 1-1: Signal transformations.

Transformation

Expression

Consequence

Time shift

y(t) = x(t T )

Time scaling
Time reversal
Generalized

y(t) = x(at)
y(t) = x(t)
y(t) = x(at b)

Even/odd synthesis

x(t) = xe (t) + xo (t)

Waveform is shifted along +t direction if T > 0 and along t direction if

T < 0.
Waveform is compressed if |a| > 1 and expanded if |a| < 1.
Waveform is mirror-imaged relative to vertical axis.
Timeshift T = b/a, compression/expansion factor a, time reversal if a is
negative.
xe (t) = 21 {x(t) + x(t)}, xo (t) = 21 {x(t) x(t)}.

and y(t) will exhibit odd symmetry if x1 (t) and x2 (t) exhibit
different forms of symmetry. That is,
(even) (even) = even,
(even) (odd) = odd,

1-3.4

A signals waveform may be periodic or nonperiodic (also

called aperiodic).
! A periodic signal x(t) of period T0 satisfies the
periodicity property:

and

x(t) = x(t + nT0 )

(odd) (odd) = even.

In Chapter 5, we will find it easier to analyze a signal if it

possesses even or odd symmetry than if it possesses neither. In
that case, it may prove advantageous to synthesize a signal x(t)
as the sum of two component signals, one with even symmetry
and another with odd symmetry:
x(t) = xe (t) + xo (t),

(1.9)

with
1
[x(t) + x(t)],
2
1
xo (t) = [x(t) x(t)].
2
xe (t) =

(1.11)

for all integer values of n and all times t. "

1-3.3 Even/Odd Synthesis

(1.10a)
(1.10b)

As the graphical example shown in Fig. 1-11 demonstrates,

adding a copy of its time reversal, x(t), to any signal x(t)
generates a signal with even symmetry. Conversely, subtracting
a copy of its time reversal from x(t) generates a signal with odd
symmetry.
Table 1-1 provides a summary of the linear transformations
examined thus far.

Periodic vs. Nonperiodic

The periodicity property states that the waveform of x(t) repeats

itself every T0 seconds. Examples of periodic signals are
displayed in Fig. 1-12.
Note that if a signal is periodic with period T0 , it is also
periodic with period 2T0 , 3T0 , etc. The fundamental period of
a periodic signal is the smallest value of T0 such that Eq. (1.11)
is satisfied for all integer values of n. In future references, the
term period shall refer to the fundamental period T0 .
The most important family of periodic signals are sinusoids.
A sinusoidal signal x(t) has the form
x(t) = A cos(0 t + ),

t ,

where
A = amplitude;

xmax = A

and

0 = angular frequency in rad/s,

xmin = A,

= phase-angle shift in radians or degrees.

Related quantities include

f0 = 0 /2 = circular frequency in Hertz,

T0 = 1/f0 = period of x(t) in seconds.

book 2012/2/17 1:44 page 11 #11

1-4

NONPERIODIC WAVEFORMS

11

x(t)

Concept Question 1-4: Define even-symmetrical and

odd-symmetrical waveforms.

x(t) = A sin (2t/T0)

T0/2

T0/4

Concept Question 1-5: State the periodicity property.

3T0/2
T0

2T0

t
Exercise 1-4: Which of the following functions
have even-symmetrical waveforms, odd-symmetrical
waveforms, or neither? (a) x1 (t) = 3t 2 , (b) x2 (t) =
sin(2t), (c) x3 (t) = sin2 (2t), (d) x4 (t) = 4et , (e)
x5 (t) = | cos 2t|.

A
(a)
x(t)

Answer: (a), (c), and (e) have even symmetry; (b) has
odd symmetry; (d) has no symmetry. (See )

xm
0

T0/2

3T0/2

T0

2T0

xm

x(t) = A cos2 (2t/T0)

T0/2

Nonperiodic Waveforms

! A nonperiodic signal is any signal that does not satisfy

the periodicity property. "

(b)
x(t)

1-4

3T0/2

T0

2T0

Many real-world signals are often modeled in terms of a core

set of elementary waveforms which includes the step, ramp,
pulse, impulse, and exponential waveforms, and combinations
thereof. Accordingly, we will use this section to review their
properties and mathematical expressions and to point out the
connections between them.

(c)
Figure 1-12: Examples of periodic waveforms.

Another important periodic signal is the complex exponential

given by
x(t) = Aej 0 t
= |A|e

j (0 t+)

where, in general, A is a complex amplitude given by

j

A = |A|e .
By Eulers formula,
x(t) = |A|ej (0 t+)

= |A| cos(0 t + ) + j |A| sin(0 t + ).

Hence, the complex exponential is periodic with period T0 =

2/0 .

1-4.1

Step-Function Waveform

The waveform of signal u(t) shown in Fig. 1-13(a) is an (ideal)

unit step function: It is equal to zero for t < 0, at t = 0 it
makes a discontinuous jump to 1, and from there on forward it
remains at 1. Mathematically, u(t) is defined as
u(t) =

"

0
1

for t < 0,
for t > 0.

(1.12)

Because u(t) does not have a unique value at t = 0, its derivative

is infinite at t = 0, qualifying u(t) as a singularity function.
! A singularity function is a function such that either
itself or one (or more) of its derivatives is (are) not finite
everywhere. "
Occasionally, it may prove more convenient to model the unit
step function as a ramp over an infinitesimal interval extending

book 2012/2/17 1:44 page 12 #12

12

CHAPTER 1

Step Functions
u(t)

t
(a) u(t)

u(t)
u(t)

1
Slope =

1 0.5
2

0
(b) Gradual step model

u(t 2)

2
0
(c) Time-shifted step function with T = 2.

The slope of u(t) still is not finite at t = 0, consistent with the

formal definition given by Eq. (1.12), which describes the unit
step function as making an instantaneous jump at t = 0. As will
be demonstrated later in Section 1-4.4, the alternative definition
for u(t) given by Eq. (1.13) will prove useful in establishing the
connection between u(t) and the impulse function (t).
The unit time-shifted step function u(t T ) is a step function
that transitions between its two levels when its argument
(t T ) = 0:
"
0 for t < T ,
u(t T ) =
(1.15)
1 for t > T .
! For any unit step function, its value is zero when its
argument is less than zero and one when its argument is
greater than zero. "

u(t 2)
1

Extending this definition to the time-reversed step function, we

have
"
1 for t < T ,
u(T t) =
(1.16)
0 for t > T .
By way of examples, Figs. 1-13(c) and (d) display plots of
u(t 2) and u(1 t), respectively.

u(1 t)
u(1 t) 1

1-4.2
0
1
(d) Time-reversed step function with T = 1.

Ramp-Function Waveform

The unit ramp function r(t) and the unit time-shifted ramp
function r(t T ) are defined as
r(t) =

"

r(t T ) =

"

Figure 1-13: Unit step function.

between and +, as shown in

u(t) can be defined as

$*
) 1 # t
+1
u(t) = lim
0
2

SIGNALS

With this alternative definition, u(t) is a continuous function

everywhere, but in the limit as 0 its slope in the interval
(, ) is
) #
$*
# $
d 1 t
1
+ 1 = lim
. (1.14)
u (t) = lim
0 dt 2
0 2

u(t)

Fig. 1-13(b). Accordingly,

for t
for t
for t ,

(1.13)

0
t

for t 0,
for t 0,

(1.17a)

and
0
for t T ,
(t T ) for t T .

(1.17b)

Two ramp-function examples are displayed in Fig. 1-14. In

each case, the ramp function is zero when its argument (t T )

book 2012/2/17 1:44 page 13 #13

1-4

NONPERIODIC WAVEFORMS

13

Ramp Functions

(V)
12

Slope = 3 V/s
6V

9
6

3r(t 1)

3V
3 2 1 0

t (s)

0 1

5 4 3 2 1

(a)

t (ms)

(a) Original function

2V

Composite
waveform

(V)

t (s)
3 2 1 0 1 2 3 4
Slope = 2 V/s
2 V
2r(t + 1)
4 V

12
9

1(t) = 3r(t + 2 ms)

(b)

Figure 1-14: Examples of ramp functions.

5 4 3 2 1 0 1
3

is smaller than zero, and equal to its own argument when the
value of t is such that the argument is greater than zero. The
slope of a ramp function x(t) = ar(t T ) is specified by the
constant coefficient a.
Because the time-derivative of r(t T )i.e., its slope
is discontinuous at t = T , the ramp function qualifies as a
singularity function.
The unit ramp function is related to the unit step function by

t (ms)

6
9
12

2(t) = 3r(t 2 ms)

(b) As sum of two time-shifted ramp functions

Figure 1-15: Step waveform of Example 1-2.

r(t) =

+t

u( ) d = t u(t),

(1.18)

and for the time-shifted case,

r(t T ) =

+t

u( T ) d

= (t T ) u(t T ).

(1.19)

Example 1-2: Synthesizing a Step-Waveform

For the (realistic) step waveform (t) displayed in Fig. 1-15,

develop expressions in terms of ramp and ideal step functions.
Note that (t) is in volts (V) and the time scale is in
milliseconds.

Solution:
The voltage (t) can be synthesized as the sum of two timeshifted ramp functions [Fig. 1-15(b)]: One starts at t = 2 ms
and has a positive slope of 3 V/ms and a second starts at t = 2 ms
but with a slope of 3 V/ms. Thus,
(t) = 1 (t) + 2 (t)

= 3r(t + 2 ms) 3r(t 2 ms)

V.

In view of Eq. (1.19), (t) also can be expressed as

(t) = 3(t + 2 ms) u(t + 2 ms)

3(t 2 ms) u(t 2 ms)

V.

book 2012/2/17 1:44 page 14 #14

14

CHAPTER 1
general rectangular function is defined as

Rectangular Pulses
1
1/2

rect

rect(t)

rect

( )
tT

( )
t+2
2

t (s)

0
= 1

for t < (T /2),

for (T /2) < t < (T + /2),
for t > (T + /2).
(1.20)

Example 1-3: Rectangular and Trapezoidal Pulses

Develop expressions in terms of ideal step functions for (a) the

rectangular pulse a (t) in Fig. 1-17(a) and (b) the more-realistic
trapezoidal pulse b (t) in Fig. 1-17(b).
Solution:
(a) The amplitude of the rectangular pulse is 4 V, its duration
is 2 s, and its center is at t = 3 s. Hence,

(b)

8 rect

t T

We note that because the rectangular function is discontinuous

at its two edges (namely at t = T /2 and t = T + /2), it
is a bona fide member of the family of singularity functions.

(a)

t (s)

1/2

SIGNALS

0
3 2 1
8
T = 2

T=3
4

2
8 rect

t 3
a (t) = 4 rect
2

t (s)

( )
t3
2

V.

The sequential addition of two time-shifted step functions, 1 (t)

at t = 2 s and 2 (t) at t = 4 s, as demonstrated graphically
in Fig. 1-17(c), accomplishes the task of synthesizing the
rectangular pulse in terms of two step functions:

(c)
Figure 1-16: Rectangular pulses.

a (t) = 1 (t) + 2 (t)

= 4[u(t 2) u(t 4)]

V.

Generalizing, a unit rectangular function rect[(t T )/ ] always

can be expressed as

1-4.3 Pulse Waveform

The rectangular function rect(t) is defined as
rect(t) =

for |t| <

for |t| >

1
2
1
2

1
=u t+
2

#
$
1
u t
,
2

and its waveform is displayed in Fig. 1-16(a). Note that the

rectangle is of width 1 s, height 1 unit, and is centered at t = 0.
In general, a rectangular pulse can be described mathematically by the rectangular function rect[(t T )/ ]. Its two
parameters are T , which defines the location of the center of
the pulse along the t-axis, and , which is the duration of the
pulse [Fig. 1-16(b)]. Examples are shown in Fig. 1-16(c). The

t T
rect

,
./
=u t T
2
,
./
u t T +
.
2

(1.21)

(b) The trapezoidal pulse exhibits a change in slope at

t = 0, t = 1 s, t = 3 s, and t = 4 s, each of which can
be accommodated by the introduction of a time-shifted ramp
function with the appropriate slope. Building on the procedure
used in Example 1-2, b (t) can be synthesized as the sum of

book 2012/2/17 1:44 page 15 #15

1-4

NONPERIODIC WAVEFORMS

15

Waveform Synthesis
a(t)

( )

4V
0 1

b(t)

t3
4 rect
2

5V
5

t (s)

b(t)

a(t)
4u(t 2)

4V

4 V

3 4

(b) Trapezoidal pulse

(a) Rectangular pulse

0 1

0 1

2 1

1(t) = 5r(t)
4(t) = 5r(t 4)

5V
t (s)

2 1

4u(t 4)

0 1

3(t) = 5r(t 3)

2(t) = 5r(t 1)

(c) a(t) = 4u(t 2) 4u(t 4)

(d) b(t) = 1(t) + 2(t) + 3(t) + 4(t)

Figure 1-17: Rectangular and trapezoidal pulses of Example 1-3.

x(t)

the four ramp functions shown in Fig. 1-17(d):

b (t) = 1 (t) + 2 (t) + 3 (t) + 4 (t)

10 V

= 5[r(t) r(t 1) r(t 3) + r(t 4)]

= 5[t u(t) (t 1) u(t 1)

(t 3) u(t 3) + (t 4) u(t 4)]

V,

where in the last step, we used the relation given by Eq. (1.19).

Example 1-4: Periodic Sawtooth Waveform

Express the periodic sawtooth waveform shown in Fig. 1-18 in

terms of step and ramp functions.
Solution:
The segment between t = 0 and t = 2 s is a ramp with a slope
of 5 V/s. To effect a sudden drop from 10 V down to zero at

...
0

...
2

t (s)

Figure 1-18: Periodic sawtooth waveform of Example 1-4.

t = 2 s, we need to (a) add a negative ramp function at t = 2 s

and (b) add a negative offset of 10 V in the form of a delayed
step function. Hence, for this time segment,
x1 (t) = [5r(t) 5r(t 2) 10u(t 2)] V,

0 t 2 s.

book 2012/2/17 1:44 page 16 #16

16

CHAPTER 1

By extension, for the entire periodic sawtooth waveform with

period T0 = 2 s, we have
x(t) =
=

n=

n=

(t)

(t T)

x1 (t nT0 )

[5r(t 2n) 5r(t 2 2n)

10u(t 2 2n)] V.

(t)
Area = 1

functions related to the step function?

1/2

The step function u(t) is

considered a singularity function because it makes a
discontinuous jump at t = 0. The ramp function r(t) is
continuous at t = 0. Yet it also is a singularity function.
Why?
Concept Question 1-7:

t (s)

u(t)

1
Slope =
2

u(t)

in terms of unit step or ramp functions.

10

(b) Rectangle model for (t)

Exercise 1-5: Express the waveforms shown in Fig. E1-5

(a) (t) and (t T)

Concept Question 1-6: How are the ramp and rectangle

SIGNALS

0.5

(c) Gradual step model for u(t)

10

Figure 1-19: Unit impulse function.

(a)
x
5
0

t (s)

5
(b)
Figure E1-5

Answer:
(a) x(t) = 10 u(t) 20 u(t 2) + 10 u(t 4), (b) x(t) =
2.5 r(t) 10 u(t 2) 2.5 r(t 4). (See )

Exercise 1-6: How is u(t) related to u(t)?

Answer: They are mirror images of one another (with

respect to the y-axis). (See )

1-4.4

Impulse Function

Another member of the family of singularity functions is the

unit impulse function, which is also known as the Dirac or
delta function (t). Graphically, it is represented by a vertical
arrow, as shown in Fig. 1-19(a). If its location is time-shifted
to t = T , it is designated (t T ). For any specific location T ,

book 2012/2/17 1:44 page 17 #17

1-4

NONPERIODIC WAVEFORMS

17

the impulse function is defined through the combination of two

properties:
(t T ) = 0

for t = T

(1.22a)

(t T ) dt = 1.

(1.22b)

By extension, a scaled impulse k (t) has an area k and

+t

k ( ) d = k u(t).

(1.26)

and
+

Example 1-5: Alternative Models for Impulse Function

! The first property states that the impulse function

(t T ) is zero everywhere, except at its own location
(t = T ), but its value is infinite at that location. The
second property states that the total area under the unit
impulse is equal to 1, regardless of its location. "
To appreciate the meaning of the second property, we can
represent the impulse function by the rectangle shown in
Fig. 1-19(b) with the understanding that (t) is defined in the
limit as 0. The rectangles dimensions are such that its
width, w = 2, and height, h = 1/2, are reciprocals of one
another. Consequently, the area of the rectangle is always unity,
even as 0.
According to the rectangle model displayed in Fig. 1-19(b),
(t) = 1/(2) over the narrow range < t < . For the
gradual step model of u(t) shown in Fig. 1-19(c), its slope also
is 1/(2). Hence,
du(t)
= (t).
(1.23)
dt
Even though this relationship between the unit impulse and unit
step functions was obtained on the basis of specific geometrical
models for (t) and u(t), its validity can be demonstrated to be
true always. The corresponding expression for u(t) is

u(t) =

+t

Show that models x1 (t) and x2 (t) in Fig. 1-20 qualify as unit
impulse functions in the limit as 0.

Solution:
To qualify as a unit impulse function, a function must: (1) be
zero everywhere except at t = 0, (2) be infinite at t = 0, (3) be
even, and (4) have a unit area.
(a) Triangle Model x1 (t)
(1) As 0, x1 (t) is indeed zero everywhere except at t = 0.
(2) lim x1 (0) = lim 1 = ; hence infinite at t = 0.
0

(3) x1 (t) is clearly an even function [Fig. 1-20(a)].

x1(t)

(a) Triangle model x1(t)

x2(t)
( ) d,

(1.24)

and for the time-shifted case,

d
[u(t T )] = (t T ),
dt
+t
u(t T ) =
( T ) d.

(1.25a)
(1.25b)

(b) Gaussian model x2(t)

Figure 1-20: Alternative models for (t).

book 2012/2/17 1:44 page 18 #18

18

CHAPTER 1

(4) Area of triangle =

of .

1
2

1
2
2 1 = 1, regardless of the value

Hence, x1 (t) does qualify as a unit impulse function.

(b) Gaussian Model x2 (t)
(1) Except at t = 0, as 0, the magnitude of the

2
2
exponential et /2 always will be smaller than 2 .
Hence, x2 (t) 0 as 0, except at t = 0.
(2) At t = 0,

1
2
2
lim
et /2
0 2

t=0

1
= lim
0 2

A=

One of the most useful features of the impulse function is

its sampling property. For any function x(t) known to be
continuous at t = T :
+

x(t) (t T ) dt = x(T ).
(sampling property)

1
2
2
et /2 dt.

x(t) (t T ) dt =

Applying the integral formula

+

ea

2x2

(1.29)

Derivation of the sampling property relies on Eqs. (1.22b) and

(1.28):

(4) The area of the Gaussian model is

+

! Multiplication of a time-continuous function x(t) by

an impulse located at t = T generates a scaled impulse
of magnitude x(T ) at t = T , provided x(t) is continuous
at t = T . "

= .

(3) x2 (t) is clearly an even function [Fig. 1-20(b)].

SIGNALS

x(T ) (t T ) dt

= x(T )

dx =
a

(t T ) dt

= x(T ).

leads to A = 1. Hence, x2 (t) qualifies as a unit impulse

function.

1-4.6 Time-Scaling Transformation of (t)

To determine how time scaling affects impulses, let us evaluate
the area of (at):

1-4.5 Sampling Property of (t)

As was noted earlier, multiplying an impulse function by
a constant k gives a scaled impulse of area k. Now we
consider what happens when a continuous-time function x(t)
is multiplied by (t). Since (t) is zero everywhere except at
t = 0, it follows that
x(t) (t) = x(0) (t),

(1.27)

provided that x(t) is continuous at t = 0. By extension,

multiplication of x(t) by the time-shifted impulse function
(t T ) gives
x(t) (t T ) = x(T ) (t T ).

(1.28)

(at) dt =
=

( )

d
|a|

1
.
|a|

Hence, (at) is an impulse of area 1/|a|. It then follows that

(at) =

1
(t).
|a|

(1.30)

(time-scaling property)

book 2012/2/17 1:44 page 19 #19

1-4

NONPERIODIC WAVEFORMS

This result can be visualized for |a| > 1 by recalling that scaling
time by a compresses the time axis by |a|. The area of the
uncompressed rectangle in Fig. 1-19(b) is
1
[ ()] = 1.
2

Area of (t):

Repeating the calculation for a compressed rectangle gives

Area of (at):

1
2

#
$*

1

.
=
|a|
|a|
|a|

19

Concept Question 1-8: How is u(t) related to (t)?

Concept Question 1-9: Why is Eq. (1.29) called the

sampling property of the impulse function?

Exercise 1-7: If x(t) is the rectangular pulse shown in

Fig. E1-7(a), determine its time derivative x (t) and plot
it.

x(t)
2

Also note that the impulse is an even function because

1
(t)
| 1|
= (t).

(t) =

t (s)

(a) x(t)
x(t)
2 (t 3)

Example 1-6: Impulse Integral

Evaluate

!2
1

t (s)
t 2 (2t 3) dt.

2 (t 4)

Solution:
Using the time-scaling property, the impulse function can be
expressed as
# #
$$
3
(2t 3) = 2 t
2
#
$
1
3
= t
.
2
2
Hence,
+2
1

1
t (2t 3) dt =
2
2

+2
1

#
$
3
t t
dt
2
2

# $2
3
2

1
2

9
.
8

We note that (t (3/2)) = 0 only at t = 3/2, which is included

in the interval of integration, 1 t 2.

(b) x(t)
Figure E1-7
Answer:

1-4.7

x (t) = 2(t 3) 2(t 4). (See

Exponential Waveform

The exponential function is a particularly useful tool for

characterizing fast-rising and fast-decaying waveforms. Figure
1-21 displays plots for
x1 (t) = et/
and
x2 (t) = et/
for > 0. The rates of increase of x1 (t) and decrease of x2 (t)
are governed by the magnitude of the time constant .

book 2012/2/17 1:44 page 20 #20

20

CHAPTER 1

SIGNALS

Exponential Functions
Positive exponential
et/
1

et/

0.37
3

Negative
exponential
t/
2
3

Figure 1-21: By t = , the exponential function et/ decays

to 37% of its original value at t = 0.

et

et/2 et
Shorter time
constant,
faster decay

et/2

Longer time
constant,
slower decay
t

0
(a) Role of time constant
e(t+1) et

e1 2.7

e(t1)

1
! An exponential function with a small (short) time constant rises or decays faster than an exponential function
with a larger (longer) time constant [Fig. 1-22(a)]. "
Replacing t in the exponential with (t T ) shifts the exponential
curve to the right if T is positive and to the left if T is negative
[Fig. 1-22(b)]. Multiplying a negative exponential function by
u(t) limits its range to t > 0 [Fig. 1-22(c)], and by extension,
an exponential that starts at t = T and then decays with time
constant is given by
x(t) = e(tT )/ u(t T ).

e1 0.37
0
(b) Role of time shift T
1
et/ u(t)

(c) Multiplication of et/ by u(t)

1
e(t T )/ u(t T )

0
(d)

If the time constant of a

negative exponential function is doubled in value, will
the corresponding waveform decay faster or slower?

Concept Question 1-11: What is the approximate shape

(t)
V0

Table 1-2 provides a general summary of the shapes and

expressions of the five nonperiodic waveforms we reviewed in
this section.
Concept Question 1-10:

Its waveform is displayed in Fig. 1-22(d).

Occasionally, we encounter waveforms with the shape shown
in Fig. 1-22(e), wherein x(t) starts at zero and builds up as a
function of time towards a saturation value. An example is the
voltage response of an initially uncharged capacitor,
(t) = V0 (1 et/ ) u(t).

V0[1 et/] u(t)

0

(e) (t) = V0(1 et/) u(t)

Figure 1-22: Properties of the exponential function.

of the waveform described by the function (1 e|t| )?

book 2012/2/17 1:44 page 21 #21

1-5

SIGNAL POWER AND ENERGY

21

Table 1-2: Common nonperiodic functions.

Function
Step

Ramp

Rectangle

Expression
u(t T ) =

"
0
1

General Shape

for t < T
for t > T

0 T

t T
= u(t T1 ) u(t T2 )

T1 = T ; T2 = T +
2
2
rect

rect
1

exp[(t T)/] u(t T)

exp[(t T )/ ] u(t T )

Exercise 1-8: The radioactive decay equation for a certain

1-5

T2

(t T )

Signal Power and Energy

material is given by n(t) = n0 et/ , where n0 is the initial

count at t = 0. If = 2 108 s, how long is its half-life?
[Half-life t1/2 is the time it takes a material to decay to
50% of its initial value.]

The instantaneous power p(t) dissipated in a resistor R due to

the flow of current i(t) through it is

Answer: t1/2 = 1.386 108 s 4 years. (See

Additionally, the associated energy expended over a time

interval t1 < t < t2 is

Exercise 1-9: If the current i(t) through a resistor R

decays exponentially with a time constant , what is the

ratio of the power dissipated in the resistor at t = to its
value at t = 0?

Answer: p(t) = i 2 R = I02 R(et/ )2 = I02 Re2t/ ,

p( )/p(0) = e2 = 0.135, or 13.5%. (See )

t T

(t T)

Exponential

Slope = 1
t

0 T1
1

Impulse

0 T
r(t T)

r(t T ) = (t T ) u(t T )

u(t T)

p(t) = i 2 (t) R.

E=

+t2

p(t) dt.

(1.31)

(1.32)

t1

The expressions for power and energy associated with a

resistor can be extended to characterize the instantaneous
power and total energy of any signal x(t)whether electrical
or not and whether real or complexas
p(t) = |x(t)|2

(1.33a)

book 2012/2/17 1:44 page 22 #22

22

CHAPTER 1

and

SIGNALS

x1(t)
+T

E = lim

T
T

|x(t)|2 dt =

|x(t)|2 dt,

(1.33b)

where E is defined as the total energy over an infinite time

interval ( < t < ).
If |x(t)|2 does not approach zero as t , the integral
in Eq. (1.33b) will not converge. In that case, E is infinite,
rendering it unsuitable as a measure of the signals energy
capacity. An alternative measure is the (time) average
power Pav , which is defined as the power p(t) averaged over
all time:

Pav

1
= lim
T T
1
= lim
T T

+T /2

p(t) dt

+T /2

|x(t)|2 dt.

(a) x1(t)

x2(t)
4 cos (2t /10)

4
10

10

15

T /2

T /2

(1.34)

Conversely, if E is finite, Pav becomes the unsuitable measure,

because in view of Eq. (1.33b), we have
Pav

6e(t 2)

Slope = 3

E
= lim
=0
T T

(E finite).

(1.35)

(b) x2(t)

x3(t)
Slope = 2

0 1
(c) x3(t)

! Pav and E define three classes of signals:

(a) Power signals: Pav is finite and E

Figure 1-23: Signal waveforms for Example 1-5.

(b) Energy signals: Pav = 0 and E is finite

(c) Non-physical signals: Pav and E "
For a periodic signal of period T0 , it is not necessary to
evaluate the integral in Eq. (1.34) with T ; integration
over a single period is sufficient. That is,

Pav

1
=
T0

T+0 /2

T0 /2

|x(t)|2 dt

(periodic signal).

(1.36)

Most periodic signals have finite average power; hence, they

qualify as power signals.

Example 1-7: Power and Energy

Evaluate Pav and E for each of the three signals displayed in

Fig. 1-23.
Solution:
(a) Signal x1 (t) is given by

0
x1 (t) = 3t

(t2)
6e

for t 0,
for 0 t 2,
for t 2.

book 2012/2/17 1:44 page 23 #23

1-5

SIGNAL POWER AND ENERGY

23

Its total energy E is

The integration was facilitated by the integral relation

+
(3t) dt + [6e(t2) ]2 dt

E1 =

+2

+2

32
+
9t 3 33
4
+
36e
e2t dt
3 30

9t 2 dt +

= 24 + 36e
= 42.

36e2(t2) dt

3 5
e2t 33
2 32

+
+
(6et )2 dt = 36 e2t dt
0

1
T0

cos2

T0 /2

A2
=
2

2 nt
+
T0

for any sinusoidal

signal of amplitude A .
and nonzero frequency

x3 (t) = 2r(t) =

"

1
=
10
1
=
10
= 8.

#
$*
+5 )
2t 2
4 cos
dt
10

0
2t

for t 0,
for t 0.

1
= lim
T T

+T /2
4t 2 dt
0

3T /2 ;
1 4t 3 33
= lim
T T
3 30
)
*
1
4T 3
= lim

T T
24
) 2*
T
= lim
.
T
6

From the argument of cos(2t/10), the period is 10 s. Hence,

application of Eq. (1.36) leads to

(1.38)

Since E1 is finite, it follows from Eq. (1.35) that Pav1 = 0.

(b) Signal x2 (t) is a periodic signal given by
$

(1.37)

The time-averaged power associated with x3 (t) is

Pav3

2t
x2 (t) = 4 cos
10

1
,
2

dt =

If 0 = 0, Pav = A2 , not A2 /2. Since Pav2 is finite, it follows

that E2 .
(c) Signal x3 (t) is given by

= 18.

Pav2

T+0 /2

which is valid for any value of and any integer value of n

equal to or greater than 1. In fact, because of Eq. (1.37),

Pav

Note that the second integral represents the energy of 6et u(t)
delayed by 2 s. Since delaying a signal does not alter its energy,
an alternative method for evaluating the second integral is

Moreover, E3 as well.
Concept Question 1-12: Signals are divided into three

power/energy classes. What are they?

+5

16 cos2

2t
10

dt

Exercise 1-10: Determine the values

1 of2Pav and E for a

pulse signal given by x(t) = 5 rect

Answer:

t3
4

Pav = 0 and E = 100. (See

book 2012/2/17 1:44 page 24 #24

24

CHAPTER 1

SIGNALS

Chapter 1 Summary
Concepts
A signal may be continuous, discrete, or digital. It may vary with time, space, or some other independent variable and
may be single or multidimensional.
Signals are classified as causal, noncausal, or anticausal, according to when they start and end.
Signals can undergo time-shift, time-scaling, and time-reversal transformations.
A signal may exhibit even or odd symmetry. A signal with neither form of symmetry can be synthesized as the sum of
two component signals: one with even symmetry and the other with odd symmetry.
Real-world signal waveforms often are modeled in terms of a set of elementary waveforms, which include the step, ramp,
pulse, impulse, and exponential waveforms.
A signals energy capacity is characterized by its average power Pav and total energy E. These attributes are defined for
any signal, whether electrical or not.

Mathematical and Physical Models

Signal Transformations
Time shift
Time scaling
Time reversal

Signal Waveforms
See Table 1-2

y(t) = x(t T )
y(t) = x(at)

y(t) = x(t)

Signal Power and Energy

Pav

Signal Symmetry
Even
Odd

x(t) = x(t)

x(t) = x(t)

Even part

xe (t) =

Odd part

xo (t) =

Sum

1
= lim
T T

1
2
1
2

E = lim

{x(t) + x(t)}

+T

T
T

{x(t) x(t)}

+T /2

T /2

|x(t)|2 dt

|x(t)| dt =

|x(t)|2 dt

x(t) = xe (t) + xo (t)

Glossary of Important Terms

Provide definitions or explain the meaning of the following terms:
analog signal
anticausal signal
causal signal
continuous signal
delta function
digital signal
discrete signal

even symmetry
exponential waveform
impulse function
noncausal signal
nonperiodic (aperiodic)
odd symmetry
periodic

physically realizable system

pulse waveform
ramp function
sampling property
signal power
signal energy
singularity

time constant
time reversal
time-scaled
time-shifted
unit rectangular
unit step function

book 2012/2/17 1:44 page 25 #25

PROBLEMS

25

PROBLEMS
Section 1-1: Types of Signals
1.1

x1(t)

4
2

Is each of these 1-D signals:

(a) x1(t)

Analog or digital?
Continuous-time or discrete-time?

10

x2(t)

(a) Daily closes of the stock market

(b) Output from phonograph-record pickup

1.2

10

Is each of these 2-D signals:

(b) x2(t)

Analog or digital?
Continuous-space or discrete-space?

10
5

(a) Image in a telescope eyepiece

(b) Image displayed on digital TV
(c) Image stored in a digital camera

x3(t)
0

(c) x3(t)

10

1.3 The following signals are 2-D in space and 1-D in time,
so they are 3-D signals. Is each of these 3-D signals:

20

30

x4(t)

Analog or digital?

(t /2)2

Continuous or discrete?

Symmetrical

(a) The world as you see it

* (b)

(c) Output from compact-disc pickup

A movie stored on film

(d) x4(t)

(c) A movie stored on a DVD

Figure P1.4: Waveforms for Problems 1.4 to 1.7.

Section 1-2: Signal Transformations

1.4 Given the waveform of x1 (t) shown in Fig. P1.4(a),
generate and plot the waveform of:
(a) x1 (2t)
(b) x1 [2(t 1)]
1.5 Given the waveform of x2 (t) shown in Fig. P1.4(b),
generate and plot the waveform of:
*Answer(s) in Appendix E.

(a) x2 [(t + 2)/2]

(b) x2 [(t 2)/2]

1.6 Given the waveform of x3 (t) shown in Fig. P1.4(c),
generate and plot the waveform of:
* (a)

x3 [(t + 40)]

(b) x3 (2t)

book 2012/2/17 1:44 page 26 #26

26

CHAPTER 1

SIGNALS

Sections 1-3 and 1-4: Waveforms

(a)

3
(b)

Figure P1.11: Waveforms for Problems 1.11 and 1.12.

1.7 The waveform shown in Fig. P1.4(d) is given by

0
for t 0,

1 t 22

for 0 t 2 s,
2
x4 (t) = 1
for 2 t 4 s,

f (t) for 4 t 6 s,

0
for t 6 s.

(a) Obtain an expression for f (t), which is the segment

covering the time duration between 4 s and 6 s.
(b) Obtain an expression for x4 [(t 4)] and plot it.

1.8 If

"

0
for t 2
(2t 4) for t 2,
,
/
1 t+1 2
.
plot x(t), x(t + 1), x 2 , and x (t+1)
2
x(t) =

1.9 Given x(t) = 10(1 e|t| ), plot x(t + 1).

1.10

Given x(t) = 5 sin2 (6t), plot x(t 3) and x(3 t).

1.11 Given the waveform of x(t) shown in P1.11(a), generate

and plot the waveform of:
(a) x(2t + 6)
* (b) x(2t + 6)
(c) x(2t 6)
1.12 Given the waveform of x(t) shown in P1.11(b), generate
and plot the waveform of:
(a) x(3t + 6)
(b) x(3t + 6)
(c) x(3t 6)

1.13 If x(t) = 0 unless a t b, and y(t) = x(ct + d)

unless e t f , compute e and f in terms of a, b, c, and d.
Assume c > 0 to make things easier for you.
1.14 If x(t) is a musical note signal, what is y(t) = x(4t)?
Consider sinusoidal x(t).
1.15 Give an example of a non-constant signal that has the
property x(t) = x(at) for all a > 0.

1.16 For each of the following functions, indicate if it exhibits

even symmetry, odd symmetry, or neither one.
(a) x1 (t) = 3t 2 + 4t 4

* (b)

x2 (t) = 3t 3

1.17 For each of the following functions, indicate if it exhibits

even symmetry, odd symmetry, or neither one.
(a) x1 (t) = 4[sin(3t) + cos(3t)]
sin(4t)
(b) x2 (t) =
4t
1.18 For each of the following functions, indicate if it exhibits
even symmetry, odd symmetry, or neither one.
(a) x1 (t) = 1 e2t

(b) x2 (t) = 1 e2t

1.19 Generate plots for each of the following step-function

waveforms over the time span from 5 s to +5 s.
(a) x1 (t) = 6u(t + 3)

(b) x2 (t) = 10u(t 4)

(c) x3 (t) = 4u(t + 2) 4u(t 2)

1.20 Generate plots for each of the following step-function

waveforms over the time span from 5 s to +5 s.
(a) x1 (t) = 8u(t 2) + 2u(t 4)

* (b)

x2 (t) = 8u(t 2) 2u(t 4)

(c) x3 (t) = 2u(t + 2) + 2u(t + 4)

1.21 Provide expressions in terms of step functions for the
waveforms displayed in Fig. P1.21.
1.22 Generate plots for each of the following functions over
the time span from 4 s to +4 s.
(a) x1 (t) = 5r(t + 2) 5r(t)

(b) x2 (t) = 5r(t + 2) 5r(t) 10u(t)

* (c)

x3 (t) = 10 5r(t + 2) + 5r(t)

#
$
#
$
t +1
t 3
(d) x4 (t) = 10rect
10rect
2
2
#
$
#
$
t 1
t 3
(e) x5 (t) = 5rect
5rect
2
2

1.23 Provide expressions for the waveforms displayed in

Fig. P1.23 in terms of ramp and step functions.

book 2012/2/17 1:44 page 27 #27

PROBLEMS

27

x1(t)

x2(t)

x3(t)

0 1
2 1
2

t (s)

2 1 0 1
2

(a) Step

t (s)

0 1
2 1
2

(b) Bowl
x5(t)

x6(t)

t (s)

2 1 0 1
2

t (s)

0 1
2 1
2

(e) Hat

(d) Staircase down

t (s)

(c) Staircase up

x4(t)

0 1
2 1
2

t (s)

(f) Square wave

Figure P1.21: Waveforms for Problem 1.21.

x1(t)

x2(t)

t (s)

t (s)

(b) Mesa

4
(a) Vee
x3(t)
4
2
2

t (s)

4
(c) Sawtooth
Figure P1.23: Waveforms for Problem 1.23.

book 2012/2/17 1:44 page 28 #28

28

CHAPTER 1

1.24 For each of the following functions, indicate if its

waveform exhibits even symmetry, odd symmetry, or neither.
(a) x1 (t) = u(t 3) + u(t 3)

(b) x2 (t) = sin(2t) cos(2t)

(c) x3 (t) =

sin(t 2 )

(b) x2 (t) = 10e0.1t u(t)

(c) x3 (t) = 10e0.1t u(t 5)

3

(d) x4 (t) = 10(1 e10 t ) u(t)

(e) x5 (t) =
(f) x6 (t) =

10e0.2(t4)

u(t)

10e0.2(t4)

u(t 4)

1.26 Determine the period of each of the following

waveforms.
(a) x1 (t) = sin 2t
- .
(b) x2 (t) = cos
t
-3 .
(c) x3 (t) = cos2
t
3
* (d) x (t) = cos(4 t + 60 ) sin(4t + 60 )
4
#
$
4

(e) x5 (t) = cos

t + 30 sin(4t + 30 )

5
(a) x1(t) M

1.25 Provide plots for the following functions over a time

span and with a time scale that will appropriately display the
shape of the associated waveform of:
(a) x1 (t) = 100e2t u(t)

SIGNALS

2
2

10

(b) x2(t) triangle

3
2

10

3
(c) x3(t) Haar
Figure P1.27: Waveforms for Problem 1.27.

1.27 Provide expressions for the waveforms displayed in

Fig. 1.27 in terms of ramp and step functions.
1.28 Use the sampling property of impulses to compute the
following.
!
(a) y1 (t) = t 3 (t 2) dt
!
(b) y2 (t) = cos(t) (t /3) dt
! 1
(c) y3 (t) = 3 t 5 (t + 2) dt

1.29 Use the sampling property of impulses to compute the

following.
!
(a) y1 (t) = t 3 (3t 6) dt
!
* (b) y (t) = cos(t) (3t ) dt
2

! 1 5
(c) y3 (t) = 3 t (3t + 2) dt
1.30 Determine the
waveforms.
1
(a) x1 (t) = 6 cos 2
3
1
(b) x2 (t) = 6 cos 2
3
1
(c) x3 (t) = 6 cos 2
3

period of each of the following

2
1 2
t + 7 cos 2 t

2
t + 7 cos( 2 t)
2
1 2
t + 7 cos 23 t

1.31 Determine the period of each of the following functions.

(a) x1 (t) = (3 + j 2)ej t/3
(b) x2 (t) = (1 + j 2)ej 2t/3 + (4 + j 5)ej 2t/6
(c) x3 (t) = (1 + j 2)ej t/3 + (4 + j 5)ej t/2

1.32 If M and N are both positive integers, provide a general

expression for the period of
$
#
$
#
2
2
t + + B cos
t + .
A cos
M
N
Sections 1-5: Power and Energy
1.33 Determine if each of the following signals is a power
signal, an energy signal, or neither.
(a) x1 (t) = 3[u(t + 2) u(t 2)]
(b) x2 (t) = 3[u(t 2) u(t + 2)]
(c) x3 (t) = 2[r(t) r(t 2)]
(d) x4 (t) = e2t u(t)

book 2012/2/17 1:44 page 29 #29

PROBLEMS

29

1.34 Determine if each of the following signals is a power

signal, an energy signal, or neither.
(a) x1 (t) = [1 e2t ] u(t)

* (b)

x2 (t) = [t cos(3t)] u(t)

(c) x3 (t) = [e2t sin(t)] u(t)

1.35 Determine if each of the following signals is a power

signal, an energy signal, or neither.
(a) x1 (t) = [1 e2t ] u(t)
(b) x2 (t) = 2 sin(4t) cos(4t)

total energy of the signal x(t),

average power of the signal x(t) if x(t)
is periodic.

Prove each of the following energy and power properties.

(a)
E[x(t + b)] = E[x(t)]
and
Pav [x(t + b)] = Pav [x(t)]
(time shifts do not affect power or energy).

(c) x3 (t) = (1 |t|) rect(t/2)

1.39 Compute the average power of the following signals.

(a) x1 (t) = ej at for real-valued a
x3 (t) = ej 3 ej 5t

1.40 Prove these energy properties.

(a) If the even-odd decomposition of x(t) is
x(t) = xe (t) + xo (t),
then
E[x(t)] = E[xe (t)] + E[xo (t)].
(b) If the causal-anticausal decomposition of x(t) is x(t) =
x(t) u(t) + x(t) u(t), then
E[x(t)] = E[x(t) u(t)] + E[x(t) u(t)].

E[ax(t)] = |a|2 E[x(t)]

and

(c)

Compute the energy of the following signals.

(a) x1 (t) = eat u(t) for a > 0
(b) x2 (t) = ea|t| for a > 0

* (c)

1.36 Use the notation for a signal x(t):

(b)

1.38

(b) x2 (t) = (3 j 4)ej 7t

(c) x3 (t) = 2 sin(3t) cos(4t)

E[x(t)]:
Pav [x(t)]:

1.37 Use the properties of Problem 1.36 to compute the

energy of the three signals in Fig. P1.27.

Pav [ax(t)] = |a|2 Pav [x(t)]

(scaling by a scales energy and power by |a|2 ).

E[x(at)] =

1
E[x(t)]
a

and
Pav [x(at)] = Pav [x(t)]

if a > 0 (time scaling scales energy by a1 but doesnt affect

power).