You are on page 1of 5

ENGG 2310-B: Principles of Communication Systems 201718 First Term

Assignment 1 Solutions
Instructor: Wing-Kin Ma 5:30pm, October 9, 2017

Problem 1 (40%)

(a) Figure 1.(a) shows a pulse signal which is a half-cosine function. Express this pulse signal by
a mathematical formula. Find its Fourier transform.
(b) Show the amplitude spectrum of the signal in Problem 1.(a). You may either sketch it, or
plot it using software like MATLAB.
(c) Find the Fourier transforms of the pulses shown in Figure 1.(b)(c).

(a) (b)

(c)

Figure 1: Problem 1.

Solution:

(a) The signal g(t) can be formulated as


t
 
A cos T , |t| < T /2
g(t) = (1)
0, |t| T /2.

1
and the corresponding Fourier transform is
    
AT 1 1
G(f ) = sinc f T + sinc f T + .
2 2 2

The Fourier transform result above can be obtained by solving the Fourier integral directly.
Alternatively, we can use properties. The signal g(t) can be written as
   
t t
g(t) = A rect cos .
T T

Let h(t) = rect Tt . By the known result rect(t) sinc(f ) and by the time scaling property,


we have H(f ) = T sinc(f T ). Then, we get


        
A t jt t jt
G(f ) = F rect e T + F rect e T
2 T T
    
A 1 1
= H f +H f +
2 2T 2T
    
AT 1 1
= sinc f T + sinc f T + .
2 2 2

In fact, the proof above is exactly the same as that for the RF pulse in Handout 2.

(b) See Figure 2.(a). You may sketch it manually, or use MATLAB to generate the plot. For the
latter, here is a code insert for your reference.

A= 1; T= 1;
f= -5:0.01:5; % [-5,5] is the range; 0.01 is the resolution
G= (A*T/2)*( sinc(f*T-0.5) + sinc(f*T+0.5) );
plot(f,abs(G));

(c) To facilitate the explanation of the results, let h(t) denote the half-cosine function in Problem
1.(a).
T

For Figure 1.(b), we can write g(t) = h t 2 . By the time shifting property, we get

AT ejf T
    
jf T 1 1
G(f ) = e H(f ) = sinc f T + sinc f T + .
2 2 2

T

For Figure 1.(c), we can write g(t) = h t + 2 . It follows that

AT ejf T
    
jf T 1 1
G(f ) = e H(f ) = sinc f T + sinc f T + .
2 2 2

2
0.7

0.6

0.5

0.4
|G(f)|

0.3

0.2

0.1

0
5 4 3 2 1 0 1 2 3 4 5
f

Figure 2: The amplitude spectra corresponding to Figure 1.(a). A = 1; T = 1.

Problem 2 (30%)

(a) Show the Fourier transform property

dg(t)

j2f G(f ).
dt
d
R R f (x,y)
Hint: use the Leibniz integral rule, which says that dx f (x, y)dy = x dy (this
version of the Leibniz integral rule requires some assumptions with f (x, y), but as an engi-
neering course you can ignore it.)

(b) Consider a system which performs time differentiation. Discuss why this system is linear
time-invariant. Sketch the amplitude spectrum of the system frequency response.

Figure 3: Problem 2.

Solution:

3
(a) We have

d
Z
dg(t)
= G(f )ej2f t df
dt dt
Z
d[G(f )ej2f t ]
= df
dt
Z
= G(f )(j2f )ej2f t df

1
=F [j2f G(f )] .

(b) By applying the Fourier transform to y(t), we see that

Y (f ) = j2f X(f ).

Let H(f ) = j2f . By applying the inverse Fourier transform to the above equation, we
further observe
y(t) = h(t) x(t),
where h(t) = F 1 [H(f )] and recall that denotes the convolution operation. Since the
input-output relationship of the system is given by the convolution formula, it follows that
the system is linear time-invariant.
The amplitude spectrum of the system is |H(f )| = |j2f | = 2|f |.

Problem 3 (30%) Figure 4 shows a system diagram of an AM detector. The incoming signal
s(t) is an AM signal, i.e., s(t) = Ac [1 + ka m(t)] cos(2fc t), and we assume that 1 + ka m(t) 0 for
all t. The full-wave rectifier performs the following operation:

v(t) = |s(t)|;

(in circuits this can be done by a diode bridge). Prove that for an appropriately chosen lowpass
filter, the system can exactly recover the message signal. In particular, provide your suggestion on
how the frequency response of the lowpass filter should look like in order to achieve exact recovery.

4
Figure 4: Problem 3.

Solution: The derivations are similar to those in Handout 4+. We may express v(t) as

v(t) = b(t) w(t),

where b(t) = 1 + ka m(t), and w(t) is a periodic signal with period To = 1/(2fc ) and has

To To
w(t) = Ac cos(2fc t), t ;
2 2
(sketch the time-domain waveform of w(t) by yourself to get some understanding of how w(t)
looks like and why we can put through the above expression). Since w(t) admits a Fourier-series
representation
X
w(t) = cn ej4nfc t ,
n=

we have

" !#
X
j4nfc t
V (f ) = F b(t) cn e
n=

X
= cn F [b(t)ej4nfc t ]
n=
X
= cn B(f 2nfc ).
n=

By sketching out V (f ) (which I skip here and please do it by yourself), we see the following: using
an ideal low-pass filter with the passband being the same as the signal bandwidth W , and assuming
W fc , we get V0 (f ) = c0 B(f ). Moreover, it can be verified that c0 = Ac /. Hence, we get the
desired result v0 (t) = c0 b(t) = c0 (1 + ka m(t)).

You might also like