ECE 318 - Analog & Digital Communication Systems

Fall 2018

Tutorial 2

TA: Subhajit Majhi (smajhi@[Link])

Office Hours: Mondays, Wednesdays & Thursdays from 5:30 pm to 6:30 pm (DC 2638)
Textbook: Fundamentals of Communication Systems, 2nd edition, J.G. Proakis and M. Salehi

We will review
• the Fourier transform of aperiodic signals
• linear time invariant systems.

Fourier Transform
Like the Fourier-series decomposes a periodic signal in terms of complex exponentials, the Fourier-
series applies so to non-periodic signals.
Consider a non-periodic signal x(t). Its Fourier Transform is
Z ∞
X(f ) = x(t)e−j2πf t dt, (or in short) X(f ) = F[x(t)].
−∞

X(f ) is referred to as the spectrum of the signal x(t). Also, X(f ) is no longer discrete, and covers
a continuous range of frequencies.
X(f ) is generally a complex function where |X(f )| and ∠X(f ) respectively represent the magnitude
and phase of various frequency components in x(t).
The original signal x(t) can be recovered from X(f ) by applying the Inverse Fourier Transform:
Z ∞
x(t) = X(f )ej2πf t df, (or in short) x(t) = F −1 [X(f )].
−∞

The following shorthand notation is used to denote the pair of x(t) and X(f )

x(t) ⇐⇒ X(f ).

Problem 1: Textbook Example 2.3.1 (Page-60)

Determine the Fourier Transform of the signal x(t) = Π(t) defined in eqn.1 and shown in Figure 1

1, − 1 ≤ t ≤ 1 ,

2 2
Π(t) = (1)
0, otherwise.


1
 Figure 1: The Rectangular Pulse

Solution:
The Fourier Transform can be directly found from the definition as follows
Z ∞
X(f ) = x(t)e−j2πf t dt
Z−∞∞
= Π(t)e−j2πf t dt
−∞
Z 1
2
= (1)e−j2πf t dt
− 12
1
1 −j2πf t 2
= e
−j2πf − 12
1 1 1
=− [e−j2πf ( 2 ) − e−j2πf (− 2 ) ]
j2πf
1 ejπf − e−jπf
= ( )
πf 2j
sin(πf )
=
πf

The last step follows from Euler’s Formula sin(θ) = (ejθ − e−jθ )/2j.

Therefore,

X(f ) = F[Π(t)] = sinc(f) (2)

where the sinc signal is


 sin(πx) , x 6= 0,

πx
sinc(x) =
1,
 x = 0,

and shown below in Figure 2.

2
It is used extensively in communication literature such as in sampling theory etc., due to its desirable
properties, e.g., a decaying sin type function that reaches zero regularly.

Figure 2: Fourier Spectrum X(f ) of the Rectangular Pulse.

Problem 2
Determine the Fourier Transform of the complex exponential x(t) = ej2πat .

Solution:
The Fourier Transform is
Z ∞
X(f ) = ej2πaτ e−j2πf τ dτ
Z−∞
∞
= ej2π(a−f )τ dτ (used τ instead of t as the dummy variable) (3)
−∞

Now, from F[δ(t)] = 1, and the shifting property, we have

F[δ(t − a)] = e−j2πf a F[δ(t)] = e−j2πf a . (4)

We now look at the inverse Fourier transform from two perspectives:

for eqn.(4), the inverse Fourier transform gives

F −1 [e−j2πf a ] = δ(t − a), (5)

whereas from the definition of the inverse Fourier transform, we have

Z ∞ Z ∞
−1 −j2πf a −j2πf a j2πf t
F [e ]= e e df = ej2π(t−a)f df.
−∞ −∞

Therefore,
Z ∞
δ(t − a) = ej2π(t−a)f df
−∞

3
 Signal x(t) of Example 2.3.15

0.8

0.6

x(t)
0.4

0.2

-0.5 -0.3 -0.1 0.1 0.3 0.5

t
(a) (b)

Figure 3: (a) The signal x(t) of Example 2.3.15.. (b) The Rectangular Pulse.

Z ∞
= ej2π(t−a)τ dτ (used τ instead of f as the dummy variable) (6)
−∞

Since t is just a dummy variable in eqn.(6), we can exchange t with f , and write
Z ∞
δ(f − a) = ej2π(f −a)τ dτ
Z−∞∞
= e−j2π(a−f )τ dτ (7)
−∞

Finally, comparing the r.h.s of eqn.(3) and eqn.(7), it is clear that

X(f ) = δ(−(f − a))

= δ(f − a).

Hence, ej2πat ⇐⇒ δ(f − a).

Problem 2: Textbook Example 2.3.15 (Page-70)

Determine the Fourier Transform of the signal given in Equation 8 and shown in Figure 3a.


cos(πt) |t| ≤ 1

2
x(t) = (8)
0
 otherwise

Solution:
Instead of the definition, we find the Fourier transform using its properties. Note that x(t) of

4
Equation 8 can be considered as a multiplication of the rectangular pulse in Equation 9 and shown
in Figure 3b with a cosine signal.

1 − 1 ≤ t ≤
 1
2 2
Π(t) = (9)
0 otherwise


Therefore, x(t) can be expressed as:

x(t) = Π(t)cos(πt)
h ejπt + e−jπt i
= Π(t)
2
1 1
= Π(t)e + Π(t)e−jπt
jπt
2 2

We know the Fourier Transform pair: Π(t) ⇐⇒ sinc(f)

And, we know the Fourier Transform properties:

• Modulation (Shift in Frequency Domain): ej2πf0 t x(t) ⇐⇒ X(f − f0 )

• Linearity: αx1 (t) + βx2 (t) ⇐⇒ αX1 (f ) + βX2 (f )

Using these two properties, we have

1  1 1  1
X(f ) = sinc f − + sinc f + (10)
2 2 2 2

where f0 = 21 .

Problem 4: Textbook Example 2.47

Determine and sketch the Fourier Transform of the following signal:

x1 (t) = −x(t) + x(t) cos(2000πt) + 2x(t) cos2 (3000πt) (11)

where the Fourier transform of the signal x(t), i.e., X(f ), is given in Figure 4.
Solution:
Equation 11 can be re-written as:

x1 (t) = x(t)cos(2000πt) + x(t)[2cos2 (3000πt) − 1] (12)

Applying the trigonometric identity cos(2x) = 2 cos2 (x) − 1 to the above equation, we have

x1 (t) = x(t) cos(2000πt) + x(t) cos(6000πt) (13)

5
 Figure 4: X(f ) is the Fourier Transform of the signal x(t) of Problem 2.47.

1/2

-3000 -1000 0 1000 3000 Frequency (f)

Figure 5: The Fourier Transform of the cosine signals.

Figure 6: X1 (f ) is the Fourier transform of signal x1 (t).

1 1
Spectra components 2
X(f − 1000) in green and 2
X(f + 1000) in yellow represent the Fourier
3
transform of x(t) cos(2π10 t).
1 1
Spectra components 2
X(f − 3000) in blue and 2
X(f + 3000) in orange represent the Fourier
3
transform of x(t) cos(2π3 × 10 t).

ejθ +e−jθ
Since, cos(θ) = 2
, the above equation becomes:

1 1 1 1
x1 (t) = ej2000πt x(t) + e−j2000πt x(t) + ej6000πt x(t) + e−j6000πt x(t). (14)
2 2 2 2

We now apply the following two properties of Fourier Transform

• Modulation (Shift in Frequency Domain): ej2πf0 t x(t) ⇐⇒ X(f − f0 )

• Linearity: αx1 (t) + βx2 (t) ⇐⇒ αX1 (f ) + βX2 (f )

6
to eqn.(14), and get
1 1 1 1
X1 (f ) = X(f − 1000) + X(f + 1000) + X(f − 3000) + X(f + 3000) (15)
2 2 2 2
where X(f ) is the Fourier Transform of the signal x(t).

The sketch of X1 (f ) is shown in Figure 6.

Linear Time Invariant (LTI) System Problems

The response of a system to an impulse signal δ(t) is called the system’s impulse response h(t).
Due to the Linear and Time Invariant nature of LTI systems, the output response y(t) for an input
signal x(t) is defined as the convolution of the input x(t) and the system’s impulse response

y(t) = x(t) ∗ h(t) (16)

By the convolution property of the Fourier Transform, in the frequency domain we have

Y (f ) = X(f )H(f ) (17)

(convolution) ⇐⇒ (multiplication)

Since, convolution is a much more difficult operation as compared to multiplication, we can use
Fourier transform to calculate the output y(t) of a LTI system indirectly through Y (f ).
Problem 5: Textbook Problem 2.59 Part(a)
Find the output of an LTI system with impulse response h(t) = sinc(t) when driven by the input
x(t) = sinc(t).
Solution:
In time domain, the output y(t) is

y(t) = x(t) ∗ h(t) = sinc(t) ∗ sinc(t) (18)

Since convolution of two sinc functions is difficult to perform, we resort to the Fourier transform.

We know the Fourier transform pair: sinc(t) ⇐⇒ Π(f)

In this case:

• H(f ) = Π(f )

• X(f ) = Π(f )

and therefore:

Y (f ) = X(f )H(f )

7
 = Π(f )Π(f )
= Π(f )

Now, the time domain output y(t) can be found by taking the inverse Fourier transform of Y (f ).

Since sinc(t) ⇐⇒ Π(f),

y(t) = sinc(t) (19)

Problem 6: Textbook Problem 2.60

Can the response of an LTI system to x(t) = sinc(t) be y(t) = sinc2 (t)? Justify your answer.

Solution:
The output of an LTI system is

y(t) = x(t) ∗ h(t) (20)

for some impulse response h(t).

In this case, we are asked if the following is possible

sinc2 (t) = sinc(t) ∗ h(t) possible ? (21)

It is difficult to answer this question in the time domain. However, we can use the convolution
property of the Fourier transform, and infer from the frequency domain.

We know that in an LTI system

Y (f ) = X(f )H(f ) (22)

We also know the following Fourier Transform pairs from Table 2:

• x(t) = sinc(t) ⇐⇒ Π(f) = X(f) (rectangular pulse)

• y(t) = sinc2 (t) ⇐⇒ Λ(f) = Y(f) (triangular pulse)

Therefore, we are asked if the following is possible

Λ(f ) = Π(f )H(f ) possible ? (23)

The input and expected output spectra X(f ) and Y (f ) are shown in Figure 7a and Figure 7b
1
respectively. Note that the input spectrum is zero for |f | > 2
which would require the output
spectrum to be zero for |f | > 12 . However, that is not true.

Hence, there does not exist any H(f ) that gives the expected output given the input in this case.
Therefore, we can conclude that the response of an LTI system to x(t) = sinc(t) cannot be
y(t) = sinc2 (t).

8
 (a) (b)

Figure 7: Fourier Transforms of the input (a) X(f ) = Π(f ) and expected output (b) Y (f ) = Λ(f ).

Table 1: Table of Fourier-Transform Properties (Textbook Table 2.2 )

Signal Fourier Transform

αx1 (t) + βx2 (t) αX1 (f ) + βX2 (f )
X(t) x(−f )
1 f
x(at) |a| X( a )
−j2πf t0
x(t − t0 ) e X(f )
ej2πf0 t x(t) X(f − f0 )
x(t) ∗ y(t) X(f )Y (f )
x(t)y(t) X(f ) ∗ Y (f )
d
dt x(t) j2πf X(f )
dn
dtn x(t) (j2πf )n X(f )
j d


tx(t) 2π df X(f )
j n dn


tn x(t) 2π df n X(f )
Rt X(f )
−∞ x(τ )dτ j2πf + 12 X(0)δ(f )

9
 Table 2: Table of Fourier-Transform Pairs (Textbook Table 2.1 )

Time Domain Frequency Domain

δ(t) 1
1 δ(f )
δ(t − t0 ) e−j2πf t0
ej2πf0 t δ(f − f0 )
1
cos(2πf0 t) 2 δ(f − f0 ) + 21 δ(f + f0 )
sin(2πf0 t) − 2j1 δ(f + f0 ) + 1
2j δ(f − f0 )
Π(t) sinc(f)
sinc(t) Π(f )
Λ(t) sinc2 (f)
sinc2 (t) Λ(f )
e−αt u−1 (t), α > 0 1
α+j2πf

te−αt u−1 (t), α > 0 1

(α+j2πf )2

e−α|t| 2α
α2 +(2πf )2
2 2
e−πt e−πf
1
sgn(t) jπf
1 1
u−1 (t) 2 δ(f ) + j2πf
0
δ (t) j2πf
δ (n) (t) (j2πf )n
1
t −jπsgn(f )
Pn=+∞ 1
Pn=+∞ n
n=−∞ δ(t − nT0 ) T0 n=−∞ δ(f − T0 )

10