EECE 442 Communication Systems Sep.

13, 2021
American University of Beirut, Fall 2021–2022 Handout #3

Problem Set 1 – Solutions

Problem 1.1 (Fourier Transforms)

(a) Using tools acquired in previous classes, the Fourier and Inverse Fourier Transforms
(IFT) when the pulsation is used as a parameter
Z ∞
S(jω) = s(t)e−jωt dt
−∞
Z ∞
1
s(t) = S(jω)ejωt dω.
2π −∞
Since we are using the frequency f instead of ω, we perform a change of variable of
f = ω/(2π) in the second equation,
Z ∞ Z ∞
1 jωt
s(t) = S(jω)e dω = S(f )ej2πf t df.
2π −∞ −∞

(b) The FT of s(t) is equal to

Z ∞
S(f ) = s(t)e−j2πf t dt
Z−∞
∞
= δ(t − τ )e−j2πf t dt = e−j2πf τ .
−∞

(c) Since both s(t) and u(t) are additions and subtraction of functions of the form
ej2πfc t , we first find the FT of such functions.
Z ∞ Z ∞
j2πfc t −j2πf t
e e dt = e−j2π(f −fc )t dt
−∞ −∞
= δ(f − fc )
Z ∞
e−j2πfc t e−j2πf t dt = δ(f + fc )
−∞

Therefore,
1  j2πfc t
+ e−j2πfc t

s(t) = cos(2πfc t) = e
2
1
S(f ) = [δ(f − fc ) + δ(f + fc )]
2
1  j2πfc t
− e−j2πfc t

u(t) = sin(2πfc t) = e
2j
1
U (f ) = [δ(f − fc ) − δ(f + fc )]
2j

1
 (d) Since s(t) is a “rect” function,
1/2
∞ 1/2
1
Z Z 
S(f ) = s(t)e−j2πf t dt = e−j2πf t dt = e−j2πf t 

−∞ −1/2 −j2πf 
−1/2
1 1
ejπf − e
 −jπf

= = sin(πf )
j2πf πf
= sinc(f ).

(f) Let us denote by v(f ) the FT of u(t), i.e.,

Z ∞
v(f ) = u(τ )e−j2πf τ dτ.
−∞

Viewed as a function of time,

Z ∞
v(t) = u(τ )e−j2πtτ dτ.
−∞

Taking the FT of v(t),

Z ∞ Z ∞ Z ∞ 
−j2πf t
v(t)e dt = u(τ )e −j2πtτ
dτ e−j2πf t dt
−∞
Z−∞ −∞
∞ Z ∞
= u(τ )e−j2πtτ e−j2πf t dτ dt
Z−∞
∞
−∞
Z ∞ 
−j2πtτ −j2πf t
= u(τ ) e e dt dτ
−∞ −∞
Z ∞
= u(τ ) [δ(τ + f )] dτ = u(−f ).
−∞

(e) Using the results of (f) and (d), the FT of sinc(t) is

rect(−f ) = rect(f ).

Problem 1.2

(a) Since the signal has a bandwidth of B, by the Nyquist sampling theorem we need
to sample at a rate faster than 2B samples per second, i.e. T < 1/2B seconds.

(b) To compute the energy Es of s̃(t), we use Parseval’s relation:

Z ∞ Z ∞
2
Es = |s̃(t)| dt = |S̃(f )|2 df.
−∞ −∞

2
 We will compute this integral over the left-half side and right-hand side separately.
Z 0 Z B/2  2
2 6
|S̃(f )| df = f df
−∞ 0 B
B/2
36 f 3 
= ·
B2 3  0
12 B 3 3
= 2
· = B.
B 8 2

∞ 3B/2  2
6
Z Z
2
|S̃(f )| df = f df
0 0 3B
3B/2
4 f 3 
= ·
B 2 3 0
4 27B 3 9
= 2
· = B.
3B 8 2

In conclusion,

Es = 6 B.

(c) To compute the energy of s(t), we follow the same approach as before.
Observing the Fourier Transform of s(t),
1h i
S(f ) = S̃(f − fc ) + S̃(−f − fc ) ,
2
and assuming that fc >> B, its energy is Es /2.

(d) The bandwidth of the signal is now 2B and by Nyquist’s sampling theorem we need
to sample at a rate faster than 4B samples per second, i.e. T < 1/4B seconds.

(e) This result seems inconsistent with (a) as we obtained essentially for the same signal
a value that is twice the value.
Nevertheless, this is not surprising since in the baseband case we need 2B samples
per second, but these samples are generally complex translating into 4B real samples
per second. In the passband case we have a real signal and we similarly need the
same number of samples!

(f) To obtain s̃(t) from s(t), one needs to keep the right-hand side of the FT (and scale
it). This may be done by shifting the FT down by fc , low-pass filter it and scale it
appropriately, i.e.,

– Construct s(t)e−j2πfc t .
– Low-pass the signal with an (ideal) filter of bandwidth of B.
– Multiply the result by 2.

3
Problem 1.3 (Power of a cosine)

(a) If s(t) is a non-zero periodic signal s(t) with period T , its energy is equal to
Z T /2
Es = lim |s(t)|2 dt.
T →∞ −T /2

Since s(t) is a (continuous) non-zero signal, the integral of its square-amplitude over
a period
Z T /2
|s(t)|2 dt = α,
−T /2

for some positive scalar α. Furthermore,

Z kT /2
|s(t)|2 dt = kα,
−kT /2

and hence
Z T /2 Z kT /2
2
Es = lim |s(t)| dt = lim |s(t)|2 dt = lim kα = ∞.
T →∞ −T /2 k→∞ −kT /2 k→∞

(b) Now consider the signal s(t) = Am cos(2πfm t). Its power is
T /2
1
Z
Ps = lim |s(t)|2 dt.
T →∞ T −T /2

Assuming that the limit exist, we choose T to be a multiple of 1/fm :

k/2fm
fm
Z
Ps = lim |s(t)|2 dt
k→∞ k −k/2fm
k/2fm k/2fm  
fm fm 1 + cos(4πfm t)
Z Z
= lim A2m 2
cos (2πfm t) dt = lim A2m dt
k→∞ k −k/2fm k→∞ k −k/2fm 2
k/2fm
fm 1 fm 2 k
Z
= lim A2m dt = lim Am
k→∞ k −k/2fm 2 k→∞ k 2fm
A2m
= .
2

(c) Similarly to before, assuming that the limit exists, we choose T to be a multiple of

4
 1/fm :
k/2fm
fm
Z
Ps = lim |s(t)|2 dt
k→∞ k −k/2fm
k
fm
Z
2fm
A2m cos2 (2πfm t) + A2n cos2 (2πfn t) + 2Am An cos(2πfm t) cos(2πfn t) dt
 
= lim
k→∞ k −k
2fm
k/2fm  
fm 1 + cos(4πfm t)
Z
= lim A2m dt
k→∞ k −k/2fm 2
fm k/2fm 2 1 + cos(4πfn t)
Z  
+ lim An dt
k→∞ k −k/2fm 2
fm k/2fm 2 2
Z
+ lim Am An [cos(2π(fm + fn )t) + cos(2π(fm − fn )t)] dt
k→∞ k −k/2fm
fm 2 k fm 2 k
= lim Am + lim A
k→∞ k 2fm k→∞ k n 2fm
A2m A2n
= + .
2 2
where we used the fact that the second term has a period of 1/2fn . The integration
is therefore over 1/fm = 2k (1/2fn ), an integer multiple of periods and the “cosine”
term integrates to zero. Similarly, in the last term the first cosine has a period
1/(fm + fn ) = 1/(k + 1)fm and the second cosine a period of 1/(fm − fn ) =
1/(k − 1)fm . Assuming that k 6= 1, both integrate to zero over 1/fm .

(d) The power of s(t) is equal to

T /2
1
Z
Ps = lim |s(t)|2 dt.
T →∞ T −T /2

We note first the energy of s(t) is finite. Indeed, by Parseval’s equality

Z ∞
Es = |s(t)|2 dt
Z−∞∞ Z ∞
2
= |S(f )| df = A2 |rect(f )|2 df
−∞ −∞
Z 1/2
= A2 df = A2 .
−1/2

Therefore, its power (energy per second) is zero.

5
Problem 1.4

(a) For the capacitor C to charge rapidly we need the time constant of the RC cuircuit
to be small compared to 1/fc , i.e.

Rl 1
C <<
Rl + Rs fc
i.e.
1 Rs + Rl
C << .
f c Rl
(b) For the capacitor C to discharge slowly between positive peaks of the carrier wave,
but not so long that the capacitor voltage will not discharge at the maximum rate
of change of the modulating wave, we need
1
Rl C >>
fc
1
Rl C <<
fm
This is possible because fm << fc .
Therefore, we need
1 1
>> Rl C >>
fm fc
1 1
>> Rl >>
Cfm Cfc

(c) The plot is presented in Figure 1

(d) One possible way to do that is to use a low pass filter at the output of the circuit.

Problem 1.5

(a) First note that cos(2πfm t) is periodic with period 1/fm and that cos(2πfc t) is
periodic with period 1/fc .
If we assume that the ratio fc /fm is rational equal to q/l, where q and l are positive
integers, then the signal s(t) is periodic with period T = l/fm . Indeed, cos(2πfm t)
is periodic over T = l(1/fm ) and cos(2πfc t) is periodic over T = (l/fc )(fc /fm ) =
l(q/l)(1/fc ) = q(1/fc ).

6
 1.5

0.5

−0.5

−1

−1.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Figure 1: Output of Envelope Detector

Using this period T ,

s2 (t) = [1 + ka m(t)]2 cos2 (2πfc t)

1 + 2ka m(t) + ka2 m2 (t) cos2 (2πfc t)
 
=
1 + 2ka cos(2πfm t) + ka2 cos2 (2πfm t) cos2 (2πfc t)
 
=
= cos2 (2πfc t) + 2ka cos(2πfm t) cos2 (2πfc t) + ka2 cos2 (2πfm t) cos2 (2πfc t)

7
 1 T 2
Z
P = s (t)dt
T 0
1 T
Z
= cos2 (2πfc t) + 2ka cos(2πfm t) cos2 (2πfc t)
T 0
+ka2 cos2 (2πfm t) cos2 (2πfc t)dt
1 1 T cos(4πfc t) + 1
Z
= + 2ka cos(2πfm t)
2 T 0 2
2 2 2
+ka cos (2πfm t) cos (2πfc t)dt
1 1 T cos(2πfm t) cos(4πfc t) + cos(2πfm t)
Z
= + 2ka
2 T 0 2
2 2 2
+ka cos (2πfm t) cos (2πfc t)dt
1 1 T cos(2π(fm + 2fc )t) + cos(2π(fm − 2fc )t)
Z
= + 2ka
2 T 0 4
2 2 2
+ka cos (2πfm t) cos (2πfc t)dt
1 1 T ka2
Z
= + [cos(2π(fm + fc )t) + cos(2π(fm − fc )t)]2 dt
2 T 0 4
1 1 T ka2  2
Z
cos (2π(fm + fc )t) + cos2 (2π(fm − fc )t)

= +
2 T 0 4
1 T ka2
Z
+ [2 cos(2π(fm + fc )t) cos(2π(fm − fc )t)] dt
T 0 4
1 ka2 1 T ka2
Z
= + + cos(2π(fm + fc )t) cos(2π(fm − fc )t)dt
2 4 T 0 2
1 ka2 1 T ka2 1 k2
Z
= + + (cos(2π2fm t) + cos(2π2fc t)) dt = + a .
2 4 T 0 4 2 4

Finally, since m(t) = cos(2πfm t), its maximal value is one, and hence the percentage
modulation is pm = 100ka , and finally, P = 1/2 + pm2 /40000.

(b) The power in the sidebands of the signal corresponds actually to that of the time
domain function ka m(t) cos(2πfc t). Based on the previous derivation, the corre-
sponding power is ka2 /4.
Therefore the efficiency of the modulated signal is

ka2 pm2
= .
2 + ka2 20000 + pm2

A plot of the efficiency function of the percentage modulation is presented in Fig-

ure 2

8
0.35

0.3

0.25

0.2

0.15

0.1

0.05

0
0 10 20 30 40 50 60 70 80 90 100

Figure 2: Efficiency Function of the Percentage modulation