Signals and Systems: Solution for Assignment 5

Problem 9.36. [40pt] In this problem, we consider the construction of various types of block diagram repre-
sentations for a causal LTI system S with input x(t), output y(t), and system function

2s2 + 4s − 6
H(s) = .
s2 + 3s + 2
To derive the direct-form block diagram representation of S, we first consider a causal LTI system S1 hat has
the same input x(t) as S, but whole system function is
1
H1 (s) = .
s2 + 3s + 2
With the output of S1 denoted by y1 (t), the direct-form block diagram representation of S1 is shown in Fig. 1.
The signals e(t) and f (t) indicated in the figure represent respective inputs into the two integrators.

Figure 1

(a) Express y(t) (the output of S) as a linear combination of y1 (t), dy1 (t)/dt, and d2 y1 (t)/dt2 .

Sol. For the system S1 , we have Y1 (s) = H1 (s)X(s). Then, for the system S,

Y (s) = H(s)X(s) = (2s2 + 4s − 6)H1 (s)X(s) = (2s2 + 4s − 6)Y1 (s). (1)

Taking inverse Laplace Transform on both sides of Eq. (1),

d2 y1 (t) dy1 (t)

y(t) = 2 2
+4 − 6y1 (t). (2)
dt dt

(b) How is dy1 (t)/dt related to f (t)?

dy1 (t)
Sol. From Fig. 1, Y1 (s) = F (s) · 1s . Therefore, f (t) = dt .

(c) How is d2 y1 (t)/dt2 related to e(t)?

df (t) d2 y1 (t)
Sol. From Fig. 1, F (s) = E(s) · 1s . Therefore, e(t) = dt = dt2
.

(d) Express y(t) as a linear combination of e(t), f (t) and y1 (t).

Sol. Plugging in the relations of Item (b) and Item (c) to Eq. (2), y(t) = 2e(t) + 4f (t) − 6y1 (t).

1
Problem 7.4. [20pt] Let x(t) be a signal with Nyquist rate ω0 . Determine the Nyquist rate for each of the
following signals:
Sol. If x(t) has a Nyquist rate of ω0 , its Fourier transform X(jω)=0 for |ω| > ω0 /2.
dx(t)
(a) dt
dx(t) F
Using the differentiation property of Fourier Transform, y(t) = dt ←
→ Y (jω) = jωX(jω).
dx(t)
Then, we can guarantee Y (jω) = 0 for |ω| > ω0 /2. Therefore, Nyquist rate of dt is also ω0 .

(b) x(t) cos ω0 t

jω0 t +e−jω0 t
Let use denote y(t) = x(t) cos ω0 t = x(t) e 2 = 12 x(t)ejω0 t + 21 x(t)e−jω0 t .
Using the frequency-shifting property of Fourier transform, Y (jω) = 21 X(j(ω − ω0 )) + 21 X(j(ω + ω0 )).
We can guarantee Y (jω) = 0 for |ω| < 3ω0 /2. Therefore, Nyquist rate of x(t) cos ω0 t is 3ω0 .

Problem 7.22. [10pt] The signal y(t) is generated by convolving a band-limited signal x1 (t) with another
band-limited signal x2 (t), that is,
y(t) = x1 (t) ∗ x2 (t)
where

X1 (jω) = 0 for |ω| > 1000π

X2 (jω) = 0 for |ω| > 2000π.

Impulse-train sampling is performed on y(t) to obtain

n=+∞
X
yp (t) = y(nT )δ(t − nT ).
n=−∞

Specify the range of values for the sampling period T which ensures that y(t) is recoverable from yp (t).
Sol.
Using the convolution property of Fourier transform, Y (jω) = X1 (jω)X2 (jω).
Since Y (jω) = 0 for |ω| > 1000π, the Nyquist rate of y(t) is 2000π.
Then, the sampling period T can be at most 2000π 2π
= 10−3 seconds. Therefore, we have to use T < 10−3
seconds in order to recover y(t) from yp (t).

2
Problem 7.23. [30pt] Shown in Fig. 2 is a system in which the sampling signal is an impulse train with
alternating sign. The Fourier transform of the input signal is as indicated in the figure.

Figure 2

(a) For ∆ < π/(2ωM ), sketch the Fourier transform of xp (t) and y(t).

Sol. First, let us denote a new signal and its Fourier transform
∞ ∞
X F
X π π
p1 (t) = δ(t − k · 2∆) ←
→ P1 (jω) = δ(ω − k · ).
∆ ∆
k=−∞ k=−∞

Then, the impulse train p(t) is rewritten as p(t) = p1 (t) − p1 (t − ∆). Using the time-shifting property of

3
 FT, its Fourier transform P (jω) is derived as
P (jω) = P1 (jω) − e−jω∆ P1 (jω)
∞
π X π
= (1 − e−jω∆ )δ(ω − k · )
∆ ∆
k=−∞
∞
π X π
= (1 − e−jkπ )δ(ω − k · ) (∵ equivalence property of δ(ω))
∆ ∆
k=−∞
∞
( !
2π X π −jkπ 1 k = 2l
= δ(ω − (2l + 1) · ) ∵e = ,l ∈ Z
∆ ∆ −1 k = 2l + 1
l=−∞

From Fig. 2, xp (t) = x(t)p(t) and thus using the multiplication property of FT,
1
Xp (jω) = X(jω) ∗ P (jω).
2π
Using the convolution property of FT, the FT of output y(t) is
Y (jω) = H(jω)Xp (jω).
The sketch of P (jω). Xp (jω) and Y (jω) is shown in Fig. 3.

Figure 3

(b) For ∆ < π/(2ωM ), determine a system that will recover x(t) from y(t). x(t) can be recovered using a
system plotted in Fig. 4(a) where the frequency response of low-pass filter is given as
(
∆ |ω| < ωM ,
Hlp (jω) =
0 otherwise.
Note the recovery process depicted in Fig. 4(c).
(c) What is the maximum value of ∆ in relation to ωM for which x(t) can be recovered from either xp (t) or
y(t)?
Sol. From Fig. 3, to avoid aliasing in Xp (jω) and Y (jω), following relation should be satisfied
π 3π
+ ωM ≤ − ωM ,
∆ ∆
π
ωM ≤ .
∆
Therefore, the maximum value of ∆max is π/ωM .

4
Figure 4: (a) System diagram; (b) Frequency response Hlp (jω) of the low-pass filter; (c) Recovery process from
y(t) to x(t)