ECE 301: Signals and Systems

Homework Assignment #7
Due on December 11, 2015

Professor: Aly El Gamal

TA: Xianglun Mao

1
Aly El Gamal ECE 301: Signals and Systems Homework Assignment #7 Problem 1

Note: Homework 7 will have only 4 problems since this is the last homework. The first 2 problems are
assigned 10 pts, and the last 2 problems are assigned 15 pts. Enjoy the last homework!

Problem 1
A signal x(t) with Fourier transform X(jw) undergoes impulse-train sampling to generate
∞
X
xp (t) = x(nT )δ(t − nT )
n=−∞

where T = 10−4 . For each of the following sets of constraints on x(t) and/or X(jw), does the sampling
theorem guarantee that x(t) can be recovered exactly from xp (t)?
(a) X(jw) = 0 for |w| > 5000π

(b) X(jw) = 0 for |w| > 15000π

(c) Re{X(jw)} = 0 for |w| > 5000π

(d) x(t) real and X(jw) = 0 for w > 5000π

(e) x(t) real and X(jw) = 0 for w < −15000π

(f) X(jw) ∗ X(jw) = 0 for |w| > 15000π

(g) |X(jw)| = 0 for w > 5000π

Solution

(a) The Nyquist rate for the given signal is 2 × 5000π = 10000π. Therefore, in order to be able to recover
2π
x(t) from xp (t), the sampling period must at most be Tmax = 10000π = 2×10−4 sec. Since the sampling
period used is T = 10−4 < Tmax , x(t) can be recovered from xp (t).

(b) The Nyquist rate for the given signal is 2 × 15000π = 30000π. Therefore, in order to be able to recover
2π
x(t) from xp (t), the sampling period must at most be Tmax = 30000π = 0.66 × 10−4 sec. Since the
−4
sampling period used is T = 10 > Tmax , x(t) cannot be recovered from xp (t).

(c) Here, Im{X(jw)} is not specified. Therefore, the Nyquist rate for the signal x(t) is indeterminate.
This implies that one cannot guarantee that x(t) would be recoverable from xp (t).

(d) Since x(t) is real, we may conclude that X(jw) = 0 for |w| > 5000π. Therefore, the answer to this
part is identical to that of part (a).

(e) Since x(t) is real, we may conclude that X(jw) = 0 for |w| > 15000π. Therefore, the answer to this
part is identical to that of part (b).

(f) If X(jw) = 0 for |w| > w1 , then X(jw) ∗ X(jw) = 0 for |w| > 2w1 . Therefore, in this part, X(jw) = 0
for |w| > 7500π. The Nyquist rate for this signal is 2 × 7500π = 15000π. Therefore, in order to be
2π
able to recover x(t) from xp (t), the sampling period must at most be Tmax = 15000π = 1.33 × 10−4 sec.
−4
Since the sampling period used is T = 10 < Tmax , x(t) can be recovered from xp (t).

(g) If |X(jw)| = 0 for w > 5000π, then X(jw) = 0 for w > 5000π. Therefore, the answer to this part is
identical to the answer of part (a).

2
Aly El Gamal ECE 301: Signals and Systems Homework Assignment #7 Problem 2

Problem 2
π
A signal x[n] has a Fourier transform X(ejw ) that is zero for 4 ≤ |w| ≤ π. Another signal
∞
X
g[n] = x[n] δ[n − 1 − 4k]
k=−∞

is generated. Specify the frequency response H(ejw ) of a lowpass filter that produces x[n] as output when
g[n] is the input.

Solution
P∞
Let p[n] = k=−∞ δ[n − 1 − 4k]. Then we obtain
∞ ∞
2π X π X −j 2πk
P (ejw ) = e−jw δ(w − 2πk/4) = e 4 δ(w − 2πk/4)
4 2
k=−∞ k=−∞

Therefore,
Z π
1
G(ejw ) = P (ejθ X(ej(w−θ) dθ))
2π −π
3
1 X −j2πk/4
= e X(ej(w−2πk/4 )
4
k=0

Since X(ejw ) = 0 for π/4 ≤ |w| ≤ π, G(ejw ) is as shown in Figure 1.

Figure 1: The signal of X(ejw ) and G(ejw ).

Clearly, in order to isolate just X(ejw ) we need to use an ideal lowpass filter with cutoff frequency π/4 and
passband gain of 4. Therefore, in the range |w| < π,

4, |w| ≤ π/4
H(ejw ) =
0, π/4 ≤ |w| ≤ π

3
Aly El Gamal ECE 301: Signals and Systems Homework Assignment #7 Problem 3

Problem 3
π
Let x(t) be a band-limited signal such that X(jw) = 0 for |w| ≥ T.

(a) If x(t) is sampled using a sampling period T , determine an interpolating function g(t) such that
∞
dx(t) X
= x(nT )g(t − nT ).
dt n=−∞

(b) Is the function g(t) unique?

Solution

(a) Let us denote the sampled signal by xp (t). We have

∞
X
xp (t) = x(nT )δ(t − nT ).
n=−∞

Since the Nyquist rate for the signal x(t) is 2π

T , we can reconstruct the signal xp (t). We then know
that
x(t) = xp (t) ∗ h(t),
where
sin(πt/T )
h(t) = .
πt/T
Therefore,
dx(t) dh(t)
= xp (t) ∗ .
dt dt
dx(t)
Denoting dt by g(t), we have
∞
dx(t) X
= xp (t) ∗ g(t) = x(nT )g(t − nT ).
dt n=−∞

Therefore,
dh(t) cos(πt/T ) T sin(πt/T )
g(t) = = − .
dt t πt2
(b) The function g(t) is not unique. We can easily contruct a counterexample which shows that it is dif-
ferent than the g(t) shown above. However, the proof is omitted.

4
Aly El Gamal ECE 301: Signals and Systems Homework Assignment #7 Problem 4

Problem 4
In this problem we develop the dual to the time-domain sampling theorem, whereby a time-limited signal
can be reconstructed from frequency-domain samples. To develop this result, consider the frequency-domain
sampling operation in Figure 2.

Figure 2: The frequency-domain sampling operation.

(a) Show that

x̃(t) = x(t) ∗ p(t)
where x̃(t), x(t), and p(t) are the inverse Fourier transforms of X̃(jw), X(jw), and P (jw), respectively.
π
(b) Assuming that x(t) is time-limited so that x(t) = 0 for |t| ≥ w0 , show that x(t) can be obtained from
x̃(t) through a ”low-time windowing” operation. That is,

x(t) = x̃(t)w(t)

where (
w0 , |t| ≤ wπ0
w(t) =
0, |t| > − wπ0

π
(c) Show that x(t) is not recoverable from x̃(t) if x(t) is not constrained to be zero for |t| ≥ w0 .

Problem 4 continued on next page. . . 5

Aly El Gamal ECE 301: Signals and Systems Homework Assignment #7 Problem 4 (continued)

Solution

(a) Since,
X̃(jw) = X(jw)P (jw),
we have
x̃(t) = x(t) ∗ p(t).

(b) Taking the inverse Fourier transform of P (jw), we have

∞
1 X 2πk
p(t) = δ(t − ).
w0 w0
k=−∞

From part (a), we have

x̃(t) = x(t) ∗ p(t).

∞
1 X 2πk
= x(t − ).
w0 w0
k=−∞

Noting that x(t) is time-limited so that x(t) = 0 for |t| > wπ0 . we assume that x(t) is as shown in
Figure 3. Then x̃(t) is as shown in the figure next to it. Clearly, x(t) can be recovered from x̃(t) by
multiplying it with the function (
w0 , |t| ≤ wπ0
w(t) =
0, othwise

(c) If x(t) is not contrained to be zero for |t| > wπ0 , then x̃(t) is as shown in Figure 3. Clearly, there is
”time-domain aliasing” between the replicas of x(t) in x̃(t). Therefore, x(t) cannot be recovered from
x̃(t).

Figure 3: The signal of x(t) (top left), x̃(t) obtained in part (b) (top right), and x̃(t) obtained in part (c)
(bottom) .