Telecommunication - I Spring 2023

Lecture:1

1 Signals and Spectra

The frequency domain description is called the spectrum. The spectral analysis of signals using Fourier series and Fourier
transforms is one of the fundamental methods of communication engineering.

1.1 FOURIER SERIES AND DISCRETE SPECTRA

Complex Exponential Fourier Series:
Let x(t) be a periodic signal with fundamental period T0 . Then we define the complex exponential Fourier series of x(t) as
∞
X 2π
x(t) = cn ejnw0 t w0 = (1)
n=−∞
T0
Z t0 +T0
1
cn = x(t)e−jnw0 t dt (2)
T0 t0

T0
for some arbitrary t0 . Setting t0 = − , we have
2
Z T0 /2
1
cn = x(t)e−jnw0 t dt (3)
T0 −T0 /2

The coefficients cn are called the Fourier coefficients of x(t). These are, in general, complex numbers and can be expressed as

cn = |cn |ejθn (4)

where |cn | is the amplitude and θn is the phase angle of cn .

Frequency Spectra:
A plot of |cn | versus the angular frequency w = 2πf is called the amplitude spectrum of the periodic signal x(t). A
plot of θn , versus w is called the phase spectrum of x(t). These are referred to as frequency spectra of x(t) . Since the
index n assumes only integers, the frequency spectra of a periodic signal exist only at the discrete frequencies nw0 . These
are therefore referred to as discrete frequency spectra or line spectra.

If the periodic signal x(t) is a real function of time, then

c−n = c∗n = |cn |e−jθn (5)

This means that, for a real periodic signal, the positive and negative coefficients are conjugate, that is,

|c−n | = |cn | θ−n = −θn (6)

Hence, the amplitude spectrum is an even function of w and the phase spectrum is an odd function of w.

Power Content of a Periodic Signal and Parseval’s Theorem:

The power content of a periodic signal ×(t) is defined as the mean square value over a period:
Z T0 /2
1
P = |x(t)|2 dt (7)
T0 −T0 /2

Parseval’s theorem for the Fourier series states that if x(t) is a periodic signal with period, T0 , then
Z T0 /2 ∞
1 X
|x(t)|2 dt = |cn |2 (8)
T0 −T0 /2 n=−∞

1
1.2 FOURIER TRANSFORMS AND CONTINUOUS SPECTRA

To generalize the Fourier series representation (equation 1) to a representation valid for non-periodic signals in frequency
domain, we introduce the Fourier transform. Let x(t) be a non-periodic signal. Then the Fourier transform of x(t), symbolized
by F is defined by Z ∞
X(w) = F [x(t)] = x(t)e−jwt dt (9)
−∞

The inverse Fourier transform of X(w), symbolized by F −1 is defined by

Z ∞
1
x(t) = F −1 [X(w)] = X(w)ejwt dw (10)
2π −∞

Equations (9) and (10) are often called the Fourier transform pair denoted by
F
x(t) ←−−−→ X(w)

Frequency Spectra
In general, the Fourier transform X(w) is a complex function of angular frequency w, so that we may express it in the form

X(w) = |X(w)|ejθ(w) (11)

where |X(w)| is called the continuous amplitude spectrum of x(t), and θ(w) is called the continuous phase spectrum
of x(t). Here, the spectrum is referred to as a continuous spectrum because both the amplitude and phase of X(w) are
functions of continuous frequency w. If x(t) si a real function of time, we have

X(−w) = X ∗ (w) = |X(w)|e−jθ(w) (12)

or
|X(−w)| = |X(w)| θ(−w) = −θ(w) (13)
Thus, just as for the complex Fourier series, the amplitude spectrum |X(w)| is an even function of w, and the phase
spectrum θ(w) is an odd function of w.

Energy Content of a Signal and Parseval’s Theorem:

The normalized energy content E of a signal x(t) is defined as
Z ∞
E= |x(t)|2 dt (14)
−∞

If E is finite (E < ∞), then we call x(t) as an energy signal. If E = ∞, then we define the normalized average power P by

1 T /2
P = lim lim |x(t)|2 dt (15)
T →∞ T −T /2

If P is finite (P < ∞), then x(t) is referred to as a power signal. Note that a periodic signal is a power signal if its energy
per period is finite.

Parseval’s theorem for the Fourier transform states that if x(t) is an energy signal, then
Z ∞
1
|x(t)|2 dt = |X(w)|2 dw (16)
−∞ 2π

2
1.3 PROPERTIES OF FOURIER TRANSFORM
1. Linearity (Superposition):
a1 x1 (t) + a2 x2 (t) ↔ a1 X1 (w) + a2 X2 (w) (17)
2. Time Shifting:
x(t − t0 ) ↔ X(w)e−jwt0 (18)
Equation (18) shows that the effect of a shift in the time domain is simply to add a linear term −wt0 to the original
phase spectrum θ(w).
3. Frequency Shifting:
x(t)ejw0 t ↔ X(w − w0 ) (19)
4. Scaling:
1 w
x(at) ↔ X (20)
|a| a
Equation (20) implies that time compression of a signal (a > 1) results in its spectral expansion and that time expansion
of the signal (a < 1) results in its spectral compression.
5. Time-Reversal:
x(−t) ↔ X(−w) (21)
6. Duality:
X(t) ↔ 2πx(−w) (22)
7. Differentiation:
Time differentiation
d
x0 (t) =x(t) ↔ jwX(w) (23)
dt
Equation (23) shows that the effect of differentiation in the time domain is the multiplication of X(w) jw in the
frequency domain.
Frequency differentiation:
d
(−jt)x(t) ↔ X 0 (w) = X(w) (24)
dw
8. Integration:
If X(0) = 0, then
Z t
1
x(τ )dτ ↔ X(w) (25)
−∞ jw
Equation (25) shows that the effect of integration in the time domain is the division of X(w) by jw in the frequency
domain, assuming that X(0) = 0. Note that by definition
Z ∞
X(0) = x(t)dt (26)
−∞

The more general case pertaining to X(0) 6= 0 is considered later.

9. Convolution:
The convolution of two signals x1 (t) and x2 (t), denoted by x1 (t) ∗ x2 (t), is a new signal x(t) defined by
Z ∞
x(t) = x1 (t) ∗ x2 (t) = x1 (τ )x2 (t − τ )dτ (27)
−∞

x1 (t) ∗ x2 (t) ↔ X1 (w)X2 (w) (28)

Equation (28) is referred to as the time convolution theorem, and it states that the convolution in the time domain
becomes multiplication in the frequency domain.
Note that the operation of convolution is commutative, that is,
x1 (t) ∗ x2 (t) = x2 (t) ∗ x1 (t)

10. Multiplication:
1
x1 (t)x2 (t) ↔ X1 (w)X2 (w) (29)
2π
Equation (29)i s often referred to as the frequency convolution theorem. Thus, the multiplication in the time domain
becomes convolution in the frequency domain.

3
1.4 FOURIER TRANSFORMS OF POWER SIGNALS
To find the Fourier transform of a periodic signal or a power signal, we need to introduce the unit impulse function.

Impulse Function:
The unit impulse function, also known as the Dirac delta function δ(t), is not an ordinary function and is defined in terms
of the following process: Z ∞
φ(t)δ(t)dt = φ(0) (30)
−∞

where φ(t) is any regular function continuous at t = 0. Equation (30) and all subsequent expressions will also apply to the
frequency-domain impulse δ by replacing t by w.

Note that Eq. (30) is a symbolic expression and should not be considered an ordinary Riemann integral. In this sense,
δ(t) is often called a generalized function and φ(t) is known as a testing function. A different class of testing function
will define a different generalized function. Similarly, the delayed delta function δ(t − t0 ) is defined by
Z ∞
φ(t)δ(t − t0 )dt = φ(t0 ) (31)
−∞

where φ(t) is any regular function continuous at t = t0 . Some additional properties of δ(t) are

x(t)δ(t − t0 ) = x(t0 )δ(t − t0 ) (32)

if x(t) is continuous at t = t0 .
x(t)δ(t) = x(0)δ(t) (33)
if x(t) is continuous at t = 0.
1
δ(at) = δ(t) a 6= 0 (34)
|a|
δ(−t) = δ(t) (35)
x(t) ∗ δ(t − t0 ) = x(t − t0 ) (36)
x(t) ∗ δ(t) = x(t) (37)
Note that Eq. (35) can be obtained by setting a = −1 in Eq. (34). Equations (33) and (37)are the special cases of Eqs. (32)
and (36), respectively, for t0 = 0.

An alternative definition of δ(t) is provided by the following two conditions:

Z t2
δ(t − t0 )dt = 1 t1 < t0 < t2 (38)
t1

δ(t − t0 ) = 0 t 6= t0 (39)
Conditions (38) and (39) correspond to the intuitive notion of a unit impulse as the limit of a suitably chosen conventional
function having unity area in an infinitely small width. For convenience, δ(t) is shown schematically in Fig. 1(a).

4
 Fourier Transforms of δ(t) and a Constant Signal:

Using Eq. (9) and Eq. (30), the Fourier transform of δ(t) is given by
Z ∞
F [δ(t)] = δ(t)e−jwt dt = 1
−∞

We thus have the Fourier transform pair for δ(t):

δ(t) ↔ 1 (40)
This relation states that the spectrum of δ(t) extends uniformly over the entire frequency interval, as shown in Fig. 1(b).

By applying the duality property [Eq. (22)] to Eq. (40) and noting that the delta function is an even function [Eq. (35)],
we obtain
1 ↔ 2πδ(w) (41)
Equation (41) states that the spectrum of a constant signal (or dc signal) [Fig. 2(a)] is a delta function πδ(w) occurring
at zero frequency, as shown in Fig. 2(b).

Integration Property:

If X(0) 6= 0, then
Z t
1
x(τ )dτ ↔ πX(0)δ(w) + X(w) (42)
−∞ jw

1.5 Problems

Example: 1
Consider the periodic square wave x(t) shown in Figure below. Determine the complex Fourier series of x(t) and plot its
magnitude spectrum for (a) a = T /4 and (b) a = T /8.

∞
X 2π
x(t) = cn ejnw0 t w0 =
n=−∞
T

R T /2
Because of the symmetry of x(t) about t = 0 we use Eq. (3) (cn = T10 −T0 0 /2 x(t)e−jnw0 t dt) to determine the Fourier series
coefficients for x(t).
" #
1 a −jnw0 t 1 a
Z Z
1 −jnw0 a jnw0 a sin nw0 a 2a
cn = e dt = e −e = c0 = dt =
T −a −jnw0 T nπ T −a T

5
a)
 
T π 1  sin(nπ/2) 
a = w0 = c0 = |cn | = 
 
4 2 2 nπ

 

b)
 
T π 1  sin(nπ/4) 
a = w0 = c0 = |cn | = 
 
8 4 4 nπ

 

Example:2
Consider the signal

x(t) = sin w0 t

Find the complex Fourier series of x(t) and plot its frequency spectra.

We could use Eq. (3) to compute the Fourier coefficients, but for this case it is simpler to use Euler’s formula and identify
by inspection the Fourier coefficients. Now we can express x(t) as
1  jw0 t  1 1
x(t) = sin w0 t = e − e−jw0 t = − e−jw0 t + ejw0 t
2j 2j 2j
Thus
1 1 1 1
c−1 = − = ej(π/2) c1 = = e−j(π/2) cn = 0 for n 6= +1, −1
2j 2 2j 2
The frequency spectra of sin w0 t are plotted in Figure 5

6
 Example: 3
Consider the signal below

x(t) = e−at u(t) a>0

where u(t) is the unit step function defined by

(
1 t>0
u(t) =
0 t<0

Find the Fourier transform of x(t) and sketch its magnitude and phase spectra.

From equation below, we have

Z ∞
X(w) = F [x(t)] = x(t)e−jwt dt
−∞

Z ∞ Z ∞
1
X(w) = e−at e−jwt dt = e−(a+jw)t dt =
0 0 a + jw

The amplitude and phase spectra of x(t) are

!
1 −1 w
|X(w)| = √ θ(w) = − tan
a + w2
2 a

which are sketched in Figure below.

7
 Example: 4
Find the Fourier transform of the rectangular pulse signal x(t) [Figure (a) below] defined by
(
1 |t| < a
x(t) = pa (t) =
0 |t| > a

Z ∞ Z a
2 sin aw sin aw
X(w) = pa (t)e−jwt dt = e−jwt dt = = 2a
−∞ −a w aw
The Fourier transform of pa (t) is shown in Figure (b).

Example: 5
Consider a box function or rectangular pulse of duration T and amplitude A, as shown in Figure (a). To define this pulse
mathematically in a convenient form, we use the notation

1 − 1 ≤ t ≤ 1

rect(t) = 2 2
1 1
0 t < −
 or t>
2 2

which stands for a rectangular function of unit amplitude and unit duration centered at t = 0. Then, in terms of this
“standard” function, we may express the rectangular pulse of Figure (a) simply as
!
t
g(t) = Arect
T

The Fourier transform of the rectangular pulse g(t) is given by

Z T /2
sin(πf T )
G(f ) = Ae−j2πf t dt = AT
−T /2 πf T

To simplify the notation in the preceding and subsequent results, we introduce another standard function—namely, the sinc
function—defined by
sin(πλ)
sinc(λ) =
πλ

8
 where l is the independent variable. The sinc function plays an important role in communication theory. As shown in
Figure above, it has its maximum value of unity at λ = 0, and approaches zero as λ approaches infinity, oscillating through
positive and negative values. It goes through zero at λ = ±1, ±2, . . . , and so on.

Thus, in terms of the sinc function, we may rewrite

!
t
A rect
AT sinc (f T )
T

The amplitude spectrum |G(f )| is shown plotted in Figure (b). above The first zero-crossing of the spectrum occurs at
f = ±1/T . As the pulse duration T is decreased, this first zero-crossing moves up in frequency. Conversely, as the pulse
duration T is increased, the first zero-crossing moves toward the origin.

Example: 6
A truncated decaying exponential pulse is shown in Figure (a). We define this pulse mathematically in a convenient form
using the unit step function:


 1 t>0
1

u(t) = t=0

 2
0 t<0


We may then express the decaying exponential pulse of Figure (a) as

g(t) = e−at u(t)

Recognizing that g(t) is zero for t < 0, the Fourier transform of this pulse is
Z ∞ Z ∞
1
G(f ) = e−at e−j2πf t dt = e−t(a+2πf ) dt =
0 0 a + j2πf

9
 The Fourier-transform pair for the decaying exponential pulse of Figure (a) is therefore
1
e−at u(t)

a + j2πf

A truncated rising exponential pulse is shown in Figure (b), which is defined by

g(t) = eat u(−t)

Note that u(−t) is equal to unity for t < 0, one-half at t = 0, and zero for t > 0. With g(t) equal to zero for t > 0, the
Fourier transform of this pulse is
Z 0
G(f ) = eat e−j2πf t dt
−∞

Replacing t with −t, we may next write

Z ∞
1
G(f ) = e−t(a−j2πf ) dt =
0 a − j2πf

The Fourier-transform pair for the rising exponential pulse of Figure (b) is therefore
1
e−at u(−t)

a − j2πf
The decaying and rising exponential pulses of Figure above are both asymmetric functions of time t. Their Fourier transforms
are therefore complex valued, as shown in equations above. Moreover, from these Fourier-transform pairs, we readily see that
truncated decaying and rising exponential pulses have the same amplitude spectrum, but the phase spectrum of the one is
the negative of the phase spectrum of the other.

Example:
Consider a double exponential pulse defined below

−at
e
 t>0
g(t) = 1 t=0 = e−a|t|

 at
e t<0

This pulse may be viewed as the sum of a truncated decaying exponential pulse and a truncated rising exponential pulse.
Therefore, using the linearity property and the Fourier-transform pairs of equations given earlier, we find that the Fourier
transform of the double exponential pulse of Figure (a) is

1 1 2a
G(f ) = + = 2
a + j2πf a − j2πf a + (2πf )2

We thus have the following Fourier-transform pair for the double exponential pulse of Figure (a):
2a
e−a|t|

a2 + (2πf )2

10