EE-210.

Signals and Systems

Homework 4∗
Spring 2010

Exercise Due Date

5th April.

Problems
Q1 Consider the LTI system initially at rest shown below.
S1 S4
x(t) w(t) z(t) + y(t)
h1 (t) w(t) = 2 dz(t)
dt + z(t)
+

S2 S3
v(t) u(t)
h2 (t) h3 (t)

where the differential system S4 is causal, and the impulse responses h1 (t), h2 (t) and h3 (t)
are as shown below.
h1 (t)

1 t

h2 (t)

1 t

h3 (t)
1

1 0.5 t

∗ LUMS School of Science & Engineering, Lahore, Pakistan.

1
 (a) Compute and sketch the overall impulse response h(t) of the system S, i.e. from x(t)
to y(t). Justify all calculations.
SOLUTION First, we compute the impulse response of h4 (t) of S4 .

Step 1
Set up the problem of calculating the impulse response of the differential equation:
dh4 (t)
2 + h4 (t) = δ(t)
dt
Step 2
Find the initial condition of the homogeneous equation at t = 0+ by integrating from
t = 0− to t = 0+ Z 0+
dh4 (τ )
2 dτ = 2h4 (0+ ) = 1
0− dt
1
hence h4 (0+ ) = is our initial condition for homogeneous equation
2
dh4 (t)
2 + h4 (t) = 0
dt
Step 3
1
The characteristic polynomial is p(s) = 2s + 1 and has one zero at s = − , which
2
means that the homogeneous response has the form h4 (t) = Ae−t/2 for t > 0. The
initial condition allows us to determine constant A:

h4 (0+ ) = A = 0.5

So that,
h4 (t) = 0.5e−t/2 u(t)
Next we compute the impulse response of the upper cascade system S1 , S4 , which is
given by h1 (t) ∗ h4 (t)

SOLUTION FOR CONVOLUTION h1 (t) ∗ h4 (t)

t < 0: no overlap so h1 (t) ∗ h4 (t) = 0

0 ≤ t < 1: overlap over 0 < τ < t:
Z t Z t
h1 (t) ∗ h4 (t) = h4 (τ )dτ = 0.5 e−τ /2 dτ = 1 − e−t/2
0 0
t ≥ 1: overlap over t − 1 < τ < t:
Z t Z t
h1 (t) ∗ h4 (t) = h4 (τ )dτ = 0.5 e−τ /2 dτ = (e0.5 − 1)e−t/2
t−1 t−1

Adding all three intervals together, we get impulse response of h1 (t) ∗ h4 (t)

2
 The impulse response of the lower cascade system S2 and S3 is a triangular func-
tion of height 2 from -1 to 0.
Finally, the total impulse response of the system will be:

h(t) = h1 (t) ∗ h4 (t) + h2 (t) ∗ h3 (t)

REMARK: The impulse response of the system can also be found by using
Fourier Transform technique. In this case the total impulse response will be:

H(jω) = H1 (jω)H4 (jω) + H2 (jω)H3 (jω)

h(t) can be found by taking Inverse F.T. of the above equation, which will turn out
to be the same as calculated earlier.
(b) Is the overall system S in (a) Stable? Is it causal? Justify your answers.
SOLUTION The system is not causal as its impulse response h(t) 6= 0 at negative times.
However, the system is BIBO stable, because the impulse response is absolutely in-
tegrable as shown below:

First part of h(t) between -1 and 1 has a finite area, so we only need to check for t > 1:
Z +∞ Z +∞
−0.5
|h(t)|dt = (e − 1) |e−0.5t |dt = . . . = 2(1 − e−0.5 ) < +∞
1 1

Q2 Determine if the following discrete-time systems are:

• Time-invariant
• Linear
• Stable
• Causal
Pn x[k]
(a) y1 [n] = k=−∞
n−k
SOLUTION • It is time-invariant. Let
n
X x[k − N ]
ya [n] = x[n − N ] =
n−k
k=−∞

We have
n−N n
X x[k] X x[m − N ]
y1 [n − N ] = = = y1 [n]
n − N − k m=−∞ n − m
k=−∞

• It is linear. Using principle of superposition, Let

n
X x1 [k]
y1 [n] = x1 [n] = (1)
n−k
k=−∞

3
 and
n
X x2 [k]
y2 [n] = x2 [n] = (2)
n−k
k=−∞

Then for x[n] = ax1 [n] + bx2 [n], we have

n n n
X ax1 [k] + bx2 [k] X x1 [k] X x2 [k]
y[n] = =a +b = ay1 [n] + by2 [n]
n−k n−k n−k
k=−∞ k=−∞ k=−∞

• It is unstable. For a given bound |x[n]| < B, the output can not be bounded for
negative n going to −∞. For example, x1 [n] = 1 is a bounded input leading to
the output
n +∞
X 1 X 1
y[n] = = = +∞ (3)
n − k m=0 m
k=−∞

which is unbounded.
• It is causal. To compute y[n], the system only needs past and current values of
the input

(n + 1)x[n], x[n] ≥ 0
(b) y2 [n] =
−(n + 1)x[n], x[n] < 0
SOLUTION • Not time-invariant
• Non-linear
• Unstable
• Non-causal

Q3 Compute the response y[n] of the discrete-time LTI system described by its impulse re-
sponse h[n] = (−2)2 u[−n − 1] + (0.8)n u[n] to the input signal x[n] = u[n] − u[n − 4]. Sketch
x[n],h[n] and y[n]

x[n] h[n]

1
... ...
0 0
1 2 3 4 n 1 2 3 4 5 n

SOLUTION Let us divide the problem into two sub-problems, each with three intervals for n.
Let h+ [n] = (0.8)n u[n], and h− [n] = (−2)n u[−n − 1], so that

y[n] = h+ [n] ∗ x[n] + h− [n] ∗ x[n] = y+ [n] + y− [n]

Starting with y+ [n] = h+ [n] ∗ x[n]:

For n < 0: x[n-k] is zero for k > 0, hence g[k] = h[k]x[n − k] = 0∀k and y[n] = 0.

4
 For 0 ≤ n ≤ 3: Then g[k] = h[k]x[n − k] 6= 0 for k = 0, . . . , n. We get
n n
X X 1 − (0.8)n+1
y[n] = h[k] = 0.8k = = 5[1 − (0.8)n+1 ]
1 − 0.8
k=0 k=0

For n > 3: Then g[k] = h[k]x[n − k] 6= 0 for k = n − 3, . . . , n. We get

n 3
X X 1 − (0.8)4
y[n] = k
0.8 = 0.8m+n−3 = 0.8n−3 = 5[(0.8)−3 − 0.8]0.8n
m=0
1 − 0.8
k=n−3

Now for y− [n] = h− [n] ∗ x[n]:

For n > 2: x[n-k] is zero for k < 0, hence g[k] = h[k]x[n − k] = 0∀k and y[n] = 0.

For 0 ≤ n ≤ 2: Then g[k] = h[k]x[n − k] 6= 0 for k = n − 3, . . . , −1. We get

−1 −1 2−n
X X X 1 − (−2)3−n 1
y[n] = h[k] = (−2)k = (−2)m+n−3 = (−2)n−3 = [(−2)n−3 −1]
m=0
1 − (−2) 3
k=n−3 k=n−3

For n ≤ −1: Then g[k] = h[k]x[n − k] 6= 0 for k = n − 3, . . . , n. We get

n 3
X X 1 − (−2)4
y[n] = (−2)k = (−2)m+n−3 = (−2)n−3
m=0
1 − (−2)
k=n−3
1 5
= [(−2)−3 − (−2)](−2)n = − (−2)n
3 8

Q4 (a) Compute the impulse response h[n] of the following causal LTI second-order difference
system initially at rest:
√
y[n] − 0.8 2y[n − 1] = −0.64y[n − 2] + x[n − 1] − x[n − 2]

Simplify your expression of h[n] to obtain a real function of time.

SOLUTION √
y[n] − 0.8 2y[n − 1] + 0.64y[n − 2] = δ[n]
Initial conditions for the homogeneous equation for n > 0 are y[0] = 1, y[−1] = 0
characteristic polynomial and zeros:
√ 3π 3π
p(z) = z 2 − 0.8 2z + 0.64 = (z − 0.8ej 4 )(z − 0.8e−j 4 )
3π 3π
The zeros are z1 = 0.8ej 4 , z2 = 0.8e−j 4
The homogeneous response for n > 0 is given by
3π 3π
ha [n] = A(0.8ej 4 )n + B(0.8e−j 4 )n

5
 Use initial conditions to compute the coefficients A and B:
3π 3π
ha [−1] = 0 = A(0.8ej 4 )−1 + B(0.8e−j 4 )−1
ha [0] = 1 = A + B
6π
From the first equation, we get B = −ej 4 = jA, and from the second equation:
1 1 1
A= = − j
1+j 2 2
1 1
B= + j
2 2
and the homogeneous response is
    
1 1 j 3π n 1 1 −j 3π n
ha [n] = −j (0.8e ) +4 +j (0.8e 4 ) u[n]
2 2 2 2
    
1 −j π j 3π n 1 jπ −j 3π n
= √ e 4 (0.8e ) + √ e
4 4 (0.8e 4 ) u[n]
2 2
( π! )
1 −j j 3π
n 1 n 3π π o
= 2< √ e 4 (0.8e 4 ) u[n] = 2 √ (0.8)n < ej( 4 n− 4 ) u[n]
2 2
√
 
3π π
= 2(0.8)n cos n− u[n]
4 4

The impulse response is:

h[n] = ha [n − 1] − ha [n − 2]
√ √
   
n−1 3π π n−2 3π π
= 2(0.8) cos (n − 1) − u[n − 1] − 2(0.8) cos (n − 2) − u[n − 2]
4 4 4 4
(4)

(b) Compute the response y[n] of the system in (a) to the input signal x[n] shown below.

x[n]

1
-5 -4 -3 -2 -1 2 3 4 5 n

-1

6
SOLUTION The input signal is composed of two time-shifted impulses, i.e., x[n] = δ[n+1]−δ[n−1],
thus
y[n] = h[n + 1] − h[n − 1]
Plug in the values from the eq. 4 to get y[n]

Q5 A periodic signal with period T is said to have a half-wave symmetry if the two halves of
one period of the signal are identical in shape. The signal is “even half-wave symmetry” if
it satisfies the condition x(t ± T /2) = x(t). On the other hand if it satisfies the condition,
x(t ± T /2) = −x(t), then it is said to be “odd half-symmetry symmetry”

Exponential form

FS coefficients in exponential form can be expressed as

Z t1 + T2   
1 T
X(k) = x(t) + (−1)k x t + e−jkω0 t dt (5)
T t1 2

Trigonometric form

FS coefficients in trigonometric can be expressed as

2 t1 +T
Z
ak = x(t) cos (kω0 t)dt
T t1
T
2 t1 + 2 2 t1 +T
Z Z
= x(t) cos (kω0 t)dt + x(t) cos (kω0 t)dt
T t1 T t1 + T2

Let α = t − T /2 in the second integral. This gives

"Z  #
t1 + T2 Z t1 + T2    
2 T T
ak = x(t) cos (kω0 t)dt + x α+ cos kω0 α + dα
T t1 t1 2 2
(6)

(a) Find the trigonometric and exponential form of the FS coefficients of even-symmetry
signal and show that odd-indexed FS harmonics are zero.
SOLUTION Trigonometric form

For an even half-wave symmetric signal, eq. 6 will be:

"Z #
t1 + T2 Z t1 + T2
2
ak = x(t) cos (kω0 t)dt + x(α) cos (kω0 α)dα
T t1 t1

The odd-indexed FS coefficients are zero. The even-indexed FS coefficients are given
by "Z #
t1 +T /2
4
ak = x(t) cos (kω0 t)dt
T t1

7
 Similarly, "Z #
t1 +T /2
4
bk = x(t) sin (kω0 t)dt
T t1

Exponential form

From eq. 5 we can see that the odd-indexed FS coefficients are zero. The even-
indexed FS coefficients are given by
Z t1 + T2
2
X(k) = e−jkω0 dt k = 0, 2, 4, . . .
T t1

(b) Find the trigonometric and exponential form of the FS coefficients of odd-symmetry
signal and show that even-indexed FS harmonics are zero.
SOLUTION Trigonometric form

For an odd half-wave symmetric signal, eq. 6 will be:

"Z #
t1 + T2 Z t1 + T2
2
ak = x(t) cos (kω0 t)dt + −x(α)[− cos (kω0 α)]dα
T t1 t1

The even-indexed FS coefficients are zero. The odd-indexed FS coefficients are given
by "Z #
t1 +T /2
4
ak = x(t) cos (kω0 t)dt
T t1

Similarly, "Z #
t1 +T /2
4
bk = x(t) sin (kω0 t)dt
T t1

Exponential form

From eq. 5 we can see that the even-indexed FS coefficients are zero. The odd-
indexed FS coefficients are given by
Z t1 + T2
2
X(k) = e−jkω0 dt k = 1, 3, 5, . . .
T t1

using the results, find the Fourier series for the periodic signal.

-8 -6 -4 -2 0 2 4 6 8 10 12 14 t

-1

8
 π
SOLUTION T0 = 8, ω0 = 4 , a0 = 0 (by inspection). Also the signal is odd half-wave symmetric. Hence
Z 2 
1 t kπ
ak = cos tdt
2 0 2 4
 
4 kπ kπ kπ
= 2 2 cos + sin −1 (n odd)
k π 2 2 2
 
4 kπ kπ
= 2 2 sin −1 (n odd)
k π 2 2
Similarly,
Z 2
1 t kπ
bk = sin tdt
2
0 2 4
 
4 kπ kπ kπ
= 2 2 sin − cos
k π 2 2 2
4 kπ
= 2 2 sin (n odd)
k π 2
Q6 Find the Fourier transform of the following periodic signal using the convolution property
in Table 4.1

−T −T1 T1 T t

SOLUTION This signal (lets say x(t))is the convolution of two rectangular pulses (e.g. v(t)) of total
width T1 , convoluted with an impulse train of period T:
+∞
X
x(t) = δ(t − kT ) ∗[v(t) ∗ v(t)]
k=−∞
| {z }
p(t)

where
1
v(t) = [u(t + T1 /2) − u(t − T1 /2)]
T1
By the convolution property,
+∞
2π X 2π T1 ω
X(jω) = P (jω)V 2 (jω) = δ(ω − k )T12 sinc2 ( )
T T 2π
k=−∞
+∞
X T12 T1 2π
= 2π sinc2 (k )δ(ω − k )
T T T
k=−∞