SIGNALS AND SYSTEMS

Academic year 2024-2025. Instructor: Dr. Marco Fabris.

Solved exercises - Part 3
Notice: u(t) = 1l(t)
1. Compute the discrete-time convolution y(n) = h(n) ∗ x(n), where
(
1, if 0 ≤ n ≤ 10
h(n) =
0, otherwise
and x(n) = u(n).

Solution. The convolution of the signals h(n) and x(n) is computed as

+∞
X n
X
y(n) = h(n − k) x(k) = u(k),
k=−∞ k=n−10

being h(n − k) = 1 for 0 ≤ n − k ≤ 10, i.e., for n − 10 ≤ k ≤ n, and h(n − k) = 0

otherwise. Noting that the unit step signal u(k) = 1 if k ≥ 0 and u(k) = 0 otherwise, we
obtain
 0, n < 0;



 n
X


 1 = n + 1, 0 ≤ n ≤ 10;
y(n) =  k=0

 n
X
1 = 11, 10 < n




k=n−10

As always, one can help themself in calculating the convolution by sketching the plot of
x(k) and h(n − k) as functions of k, for various values of n.
2. Compute the continuous-time convolution y = h ∗ x, where
(
1, if 0 ≤ t < 1;
h(t) = x(t) =
0, otherwise.

Solution. Z +∞
y(t) = h(t − τ )x(τ )dτ,
−∞
where h(t − τ ) and x(τ ) are unit-height rectangles with support (the set where a function
is non-zero) [t − 1, t] and [0, 1] on the τ -axis respectively. Thus, we find:
0, if t ≤ 0;



 R t 1 · dτ = t,

if 0 ≤ t ≤ 1;
y(t) = R01
1 · dτ = 2 − t, if 1 ≤ t ≤ 2;
 t−1



0, if t ≥ 2.

3. Compute the response to the input x(t) = cos(2t − π) δ(3t) − 51l(t) of an LTI system with
impulse response h(t) = 1l(t − 1).
1
Solution. Recall that δ(at) = |a| δ(t) and that f (t)δ(t) = f (0)δ(t) to find cos(2t−π) δ(3t) =
(cos(−π)) 13 δ(t) = − 31 δ(t). Thesystem response is then given by
y(t) = h(t) ∗ x(t) = u(t − 1) ∗ − 13 δ(t) − 5u(t) = − 13 u(t − 1) − u(t − 1) ∗ 5u(t).
R∞
We compute separately w(t) = u(t − 1) ∗ 5u(t) = −∞ 5u(t − τ )u(τ − 1)dτ . The support
(on the integration axis τ ) of the factor u(t − τ )u(τ − 1) is empty if t < 1, Rwhile it
matches the interval [1, t] if t ≥ 1. Thus, w(t) = 0 for t < 1, while w(t) = 1t 5dτ =
5(t − 1) for t ≥ 1. In compact form, w(t) = 5(t − 1)u(t − 1). The system output is
y(t) = − 31 u(t − 1) − 5(t − 1)u(t − 1) = (− 13 − 5(t − 1))u(t − 1).
4. Consider the cascade of two LTI systems
x- y- z-
Σ1 Σ2
 n
j
Let h1 (n) = cos(−3n+2) and h2 (n) = 5
1l(n). Find z(n) when x(n) = 5j δ(n−1)−δ(n).

Solution. The block diagram suggests to compute z as z = h2 ∗(h1 ∗x), but comparing the
analytical expressions of h2 and x should clarify that it is better to perform the calculation
in the form z = h1 ∗ (h2 ∗ x).
It is a useful exercise to derive this expression from the commutative and associative
properties of convolution: z = h2 ∗ (h1 ∗ x) = (h2 ∗ h1 ) ∗ x = (h1 ∗ h2 ) ∗ x = h1 ∗ (h2 ∗ x).
Substituting the
 definitions
 of the signals 
j n
  n  n
h2 (n) ∗ x(n) = 5 u(n) ∗ 5j δ(n − 1) − δ(n) = 5j u(n) ∗ 5j δ(n − 1) − 5j u(n) ∗ δ(n) =
 n+1  n  n  n  n
j j j j j
5
u(n) ∗ δ(n − 1) − 5
u(n) ∗ δ(n) = 5
u(n − 1) − 5
u(n) = 5
(u(n − 1) −
u(n)) = −δ(n)
Finally, we need to compute z(n) = h1 (n) ∗ (−δ(n)) = − cos(−3n + 2) = cos(3n + π − 2).
5. Compute the response to the input signal
x(t) = u(t) − u(t − 2), t ∈ R,
for an LTI system with impulse response
h(t) = 2δ(t) + e−t u(t), t ∈ R.

Solution The output is given by the convolution

y(t) = h(t) ∗ x(t) = (2δ(t) + e−t u(t)) ∗ (u(t) − u(t − 2)).
By the linearity of convolution and the properties of δ(t), we have
y(t) = 2(u(t) − u(t − 2)) + w(t),
where w(t) = e−t u(t) ∗ (u(t) − u(t − 2)). By direct calculation, we obtain:

 0,
 if t ≤ 0;
R t −τ
w(t) = 0 e dτ = 1 − e−t , if 0 ≤ t ≤ 2;
 Rt −τ −t+2
− e−t , if 2 ≤ t < ∞.

t−2 e dτ = e

6. Compute the discrete-time convolution y = h ∗ x, where

(
1, if 2 ≤ n ≤ 8;
h(n) =
0, otherwise;
and x(n) = u(n − 1).

Solution. From the definition of discrete-time convolution, we obtain



 0, if n < 3;
n−1


 X
∞ 8

1 = n − 2, if 3 ≤ n < 9;
X X 

y(n) = h(k)x(n − k) = u(n − k − 1) = k=2
8

k=−∞ k=2 
 X
if 9 ≤ n.




 1 = 7,
k=2

As always, one can help themself in calculating the convolution by sketching the plot of
h(k) and x(n − k) as a function of k for various values of n.
7. Compute the response to the input signal x(t) = u(t−2) for a continuous-time LTI system
with impulse response h(t) = e−3|t| u(t) − (cos t) δ(2t).

Solution. We need to compute the continuous-time convolution y = h∗x = (h1 −h2 )∗x =
h1 ∗ x − h2 ∗ x, with h1 (t) = e−3|t| u(t) and h2 (t) = (cos t) δ(2t) = 12 δ(t), where the latter
equality follows from the properties of the δ impulse: that one already cited in the solution
1
of Exercise 3 and that one stating that δ(αt) = |α| δ(t) for any real α ̸= 0.
Thus, we compute y1 = h1 ∗ x, obtaining
Z ∞ Z ∞
y1 (t) = x(s)h1 (t − s) ds = e−3|t−s| u(t − s) ds = 0, t<2
−∞ 2

and Z t
−3(t−s) 1 − e−3(t−2)
Z t−2
y1 (t) = e ds = e−3τ dτ =, t > 2.
2 0 3
Next, we obtain y2 = h2 ∗ x = 12 x, recalling that δ = 2h2 is the neutral element for the
convolution or formally by integrating:
Z ∞ Z ∞
1 1
y2 (t) = x(s)h2 (t − s) ds = δ(t − s) ds = u(t − 2), t ∈ R.
−∞ 2 2 2
In conclusion, we have

1 + 2e −3(t−2) 
 0, if t < 2;
y(t) = y1 (t) − y2 (t) = − u(t − 2) =  −3(t−2)
1 + 2e
6  − , if t > 2.
6

Even in continuous time, it is helpful to visualize the convolution calculation by sketching

the plot of x(s) and h(t − s) as a function of s, for various values of t.

8. For an LTI system with impulse response

h(t) = δ(t − 2) − δ(t + 2),

compute the output y(t) corresponding to the input

(
1 − t2 , if |t| < 1;
x(t) =
0, otherwise.

Solution. y(t) = h(t) ∗ x(t) = (δ(t − 2) − δ(t + 2)) ∗ x(t) = x(t − 2) − x(t + 2).
Since the support of x(·) has length 2, there is no overlap between the supports of x(t − 2)
and x(t + 2). Thus,

(t + 2)2 − 1, if −3 ≤ t ≤ −1;



y(t) = 1 − (t − 2)2 , if 1 ≤ t ≤ 3;


0, otherwise.

9. Compute the continuous-time convolution y = h ∗ x, where x(t) = u(t) − 2δ(t − 2) and

(
1, if |t| < 1;
h(t) =
0, otherwise.

and sketch the plot the resulting signal y(t).

 Solution. First, we compute the convolution y1 = h ∗ x1 , with x1 (t) = u(t), obtaining

Z ∞ Z t+1  0,
 t < −1;
R t+1
y1 (t) = h(t − s)x1 (s) ds = u(s) ds = 0 ds = t + 1, −1 < t < 1;
−∞ t−1  t+1
 R
t−1 ds = 2, t > 1.

As for the convolution y2 = h ∗ x2 , with x2 (t) = −2δ(t − 2), the formal properties of the
continuous-time signal δ allow us to directly conclude that
(
−2, if 1 < t < 3;
y2 (t) = −2h(t − 2) =
0, otherwise.

Consequently, from x(t) = x1 (t) + x2 (t), by linearity, we obtain

t < −1;

0,

t + 1, −1 < t < 1;


y(t) = y1 (t) + y2 (t) =
 0,
 1 < t < 3;

2, t > 3.


whose plot is shown below:

y(t) 6

-
−2 −1 1 2 3 4 t

10. Compute the discrete-time convolution z(n) = x(n) ∗ y(n), where


2,
n = 0;
x(n) = −1, |n| = 1; y(n) = u(n − 1) − u(n − 5).


0, otherwise;

Solution. Noting that y(n) = 1 for 1 ≤ n ≤ 4, otherwise y(n) = 0, from the definition of
discrete-time convolution, we obtain

∞
X X 0,
n−1 if n < 0, n = 2, 3, n > 5;


z(n) = x(k)y(n − k) = x(k) = −1, if n = 0, 5;

k=−∞ k=n−4 
1, if n = 1, 4.

As always, it can be helpful to calculate the convolution by plotting the graphs of x(k)
and y(n − k) as a function of k, for different values of n.
Note: The signal h(n) = δ(n + 1) − 2δ(n) + δ(n − 1) (equal to the opposite of the signal
somewhat unfortunately called x(n) in this exercise) is the impulse response of a discrete-
time LTI filter that produces the “second difference” of the input signal as output. In
 fact, convolving h(n) with a generic input x(n) gives the output y(n) = h(n) ∗ x(n) =
x(n + 1) − 2x(n) + x(n − 1) = z(n + 1) − z(n), where z(n) = x(n) − x(n − 1) represents
the “first difference” of x(n). This operation, performed by the LTI filter with impulse
response h(n), is analogous (and actually approximates in the case of sampled signals)
to the computation of the second derivative for continuous-time signals, allowing the
highlighting of changes in “slope” in a sequence of data. Try to interpret the convolution
computed in this exercise based on this observation.

11. Compute the response to the input signal x(t) = δ(t − 5) + e−2t u(t) − (sin t) δ(2t) for a
continuous-time LTI system with impulse response h(t) = u(t − 2).

Solution. First, we note that (sin t) δ(2t) = 12 (sin t) δ(t) = 0, as follows from the identities
1
δ(αt) = |α| δ(t), valid for any α ̸= 0, and f (t)δ(t) = f (0)δ(t), valid for any continuous f (·)
at t = 0 (sifting property).
We are thus required to calculate the convolution y(t) = h(t) ∗ x(t), where x(t) = x1 (t) +
x2 (t), with x1 (t) = δ(t − 5) and x2 (t) = e−2t u(t). Applying the distributive property of
convolution, we get y(t) = y1 (t) + y2 (t), with y1 (t) = h(t) ∗ x1 (t) and y2 (t) = h(t) ∗ x2 (t).
Then, we obtain y1 (t) = h(t − 5) = u(t − 7) by the shift property of the delta function,
and

Z ∞  0, if t < 2;
u(t − τ − 2)e−2τ u(τ ) dτ =
Z t−2
y2 (t) =
−∞  e−2τ dτ = 12 (1 − e−2(t−2) ), if t > 2;
0

thus obtaining

0,

 if t < 2;
1 −2(t−2)
y(t) = y1 (t) + y2 (t) = (1 − e ), if 2 < t < 7;
 21
 −2(t−2)
2
(3 − e ), if 7 < t.

12. Compute the step response of a discrete-time LTI system with impulse response

h(n) = 3−n u(n)

Solution. The step response h−1 is obtained for an LTI system as the convolution of the
impulse response h with the unit step input. In particular, in discrete time, we have
∞
X n
X
h−1 (n) = h(k)u(n − k) = h(k)
k=−∞ k=−∞

so that the step response results as the integral sum of the impulse response. Therefore,
in this example,

n 
 0, n<0
X
−k n −(n+1) −n
h−1 (n) = 3 u(k) =  1−3 3−3
3−k =
X
= , n≥0
k=−∞ 1 − 3−1 2

k=0

∞ ∞
X X 3
In particular, h−1 (n) converges, as n → ∞, to h(k) = |h(k)| = < ∞,
k=−∞ k=−∞ 2
showing that the system is BIBO-stable.
 ∞
X
13. Compute the continuous-time convolution y = h ∗ x, where x(t) = δ(t − 3k) is a
k=−∞
“pulse train” and (
t, if 0 < t < 1;
h(t) =
0, otherwise.
Sketch the plot of the resulting signal y(t).

Solution. Recalling the formal properties of the continuous-time δ signal, in particular

the identity h(t) ∗ δ(t − t0 ) = h(t − t0 ), for any t0 ∈ R, from the linearity of convolution
we get
∞
X ∞
X
y(t) = h(t) ∗ δ(t − 3k) = h(t − 3k), t ∈ R,
k=−∞ k=−∞

that is, the “periodic repetition” y(t) = repT h(t) with period T = 3 of the signal h(t),
whose graph is shown in the figure:

y(t) 6

-
−4 −3 −2 −1 1 2 3 4 t

14. Consider a continuous-time LTI system characterized by the impulse response

(
e−t , if 0 < t < 1;
h(t) =
0, otherwise.

a. Compute the corresponding frequency response H(jω).

b. Determine the response y(t) to the input x(t) = 2 ej2πt .

Solution. a. The system is BIBO-stable, as the impulse response is absolutely integrable.

Indeed, Z ∞ Z 1
1
|h(t)| dt = e−t dt = −e−t = 1 − e−1 < ∞.
−∞ 0 0

Now, by definition, the frequency response of the system is calculated as the integral
1
Z ∞
−jωt
Z 1
−t −jωt e−(1+jω)t 1 − e−(1+jω)
H(jω) = h(t)e dt = e e dt = − = , ω ∈ R.
−∞ 0 1 + jω 0
1 + jω

b. In continuous-time, for an LTI and BIBO-stable system in sinusoidal steady-state, the

response to the input x(t) = A ejω0 t is y(t) = H(jω0 ) A ejω0 t . Therefore, with A = 2 and
ω0 = 2π, we have

1 − e−(1+j2π) j2πt 1 − e−1 j2πt

y(t) = 2 H(j2π)ej2πt = 2 e =2 e , t ∈ R,
1 + j2π 1 + j2π

having simplified the factor e−j2π = 1.

15. Compute the discrete-time convolution y = h ∗ x, where h(n) = δ(n + 3) − δ(n) + δ(n − 3)
and x(n) = cos( 2π
3
n).

Solution. Recalling that for any discrete-time signal x(n) and shift n0 ∈ Z we have
∞
X
δ(n − n0 ) ∗ x(n) = δ(k − n0 )x(n − k) = x(n − n0 ).
k=−∞

We get
y(n) = x(n + 3) − x(n) + x(n − 3) = x(n) = cos( 2π
3
n),
since the signal x(n) = cos( 2π
3
n) is periodic with period N = 3.

16. Consider a continuous-time LTI system characterized by the frequency response

sin 3ω
H(jω) = , ω ∈ R.
ω
π 2π
Compute the output y(t) corresponding to the input x(t) = ej 6 t + ej 3
t
.

Ak ejωk t
P
Solution. For a continuous-time LTI system, the response to the input x(t) = k
is y(t) = k H(jωk ) Ak ejωk t . Therefore, in our case,
P

π 2π π
y(t) = H(j π6 )ej 6 t + H(j 2π
3
)ej 3
t
= π6 ej 6 t ,

since H(j π6 ) = 6
π
and H(j 2π
3
) = 0.

17. Compute the response to the input signal x(t) = 2(cos 2t) δ(4t)−e−5|t| 1l(t) for a continuous-
time LTI system with impulse response h(t) = 1l(t − 2).

1
Solution. Recall that δ(at) = |a| δ(t) and that f (t)δ(t) = f (0)δ(t) to get 2(cos 2t)δ(4t) =
2(cos 2t) 4 δ(t) = 2 δ(t). Also, note that e−5|t| u(t) = e−5t u(t). The system’s response is
1 1

then given by  
y(t) = h(t) ∗ x(t) = u(t − 2) ∗ 12 δ(t) − e−5t u(t) = 21 u(t − 2) − u(t − 2) ∗ e−5t u(t).
R ∞ −5(t−τ )
Let us compute separately w(t) = u(t − 2) ∗ e−5t u(t) = −∞ e u(t − τ )u(τ − 2)dτ .
The support (on the integration axis τ ) of the factor u(t − τ )u(τ − 2) is empty if t < 2
while it Rcoincides with the interval [2, t] if t ≥ 2. Thus, w(t) = 0 for t < 2, while
w(t) = 2t e−5(t−τ ) dτ = 51 (1 − e−5(t−2) ) for t ≥ 2. In compact form, we write w(t) =
1
5
(1 − e−5(t−2) )u(t

− 2). The system output is y(t) = 21 u(t − 2) − 15 (1 − e−5(t−2) )u(t − 2) =
3
10
+ 15 e−5(t−2) u(t − 2).

18. Consider a continuous-time LTI system characterized by the frequency response

sin 5ω
H(jω) = , ω ∈ R.
ω
π π
Compute the output y(t) corresponding to the input x(t) = 2 + e−j 5 t + ej 10 t .

Solution. From the definition of frequency response, we obtain

π π sin 5(− π ) π π
sin 5( 10 ) jπt
y(t) = H(j0)2ej0t + H(−j π5 )e−j 5 t + H(j 10
π
)ej 10 t = 5 · 2 + − π 5 e−j 5 t + π e
10 =
5 10
sin(−π) −j π t sin( π2 ) π π
10 j 10
10 + − π5
e 5 + π ej 10 t = 10 + π
e t.
10