Digital Signal Processing, Homework 2, Spring 2013, Prof. C.D.

Chung

1. (0.5%) Page 167, Problem 3.2

Determine the z-transform of the sequence

n; 0 n N 1,
x [n] =
N; N n.

Sol:
n; 0 n N 1,
x [n] = = nu [n] (n N ) u [n N ]
N; N n.
d d 1
nx [n] , z X (z) ) nu [n] , z jzj > 1
dz dz 1 z 1
z 1
nu [n] , jzj > 1
(1 z 1 )2
z N 1
x [n n0 ] , X (z) z n0 ) (n N ) u [n N ] , jzj > 1
(1 z 1 )2

Therefore,
z 1
z N 1 z 1 1 z N
X (z) = =
(1 z 1 )2 (1 z 1 )2

2. (0.5%) Page 168, Problem 3.7

The input to a causal LTI system is
n
1
x [n] = u [ n 1] + u [n] .
2

The z-transform of the output of this system is

1
2
z 1
Y (z) = 1 1 (1 +
.
1 2
z z 1)

(a) Determine H (z), the z-transform of the system impulse response. Be sure to
specify the ROC.
(b) What is the ROC for Y (z)?
(c) Determine y [n].
Sol:
(a)
n
1
x [n] = u [ n 1] + u [n]
2
1 1 1
) X (z) = 1
+ 1 < jzj < 1
1 z 1 2
z 1 2

1
 Now to …nd H (z) we simply use H (z) = Y (z) =X (z); i.e.,
1 1
Y (z) 2
z 1 (1 z 1) 1 2
z 1 (1 z 1 )
H (z) = = 1 1 =
X (z) 1 2
z 1 (1 + z 1) 2
z 1 (1 + z 1 )
H (z) causal ) ROC jzj < 1.
(b) Since one of the poles of X (z), which limited the ROC of X (z) to be less than
1, is cancelled by the zero of H (z), the ROC of Y (z) is the region in the z-plane
that satis…es the remaining two constraints jzj > 21 and jzj > 1. Hence Y (z)
converges on jzj > 1.
(c)
1 1
3 3
Y (z) = 1 + 1
jzj > 1
1 2
z 1 1+z
Therefore,
n
1 1 1
y [n] = u [n] + ( 1)n u [n]
3 2 3
3. (0.5%) Page 170, Problem 3.16
When the input to an LTI system is
n
1
x [n] = u [n] + (2)n u [ n 1] ,
3
the corresponding output is
n n
1 2
y [n] = 5 u [n] 5 u [n] .
3 3

(a) Find the system function H(z) of the system. Plot the pole(s) and zero(s) of H(z)
and indicate the ROC.
(b) Find the impulse response h [n] of the system.
(c) Write a di¤erence equation that is satis…ed by the given input and output.
(d) Is the system stable? Is it causal?
Sol:
(a) To determine H(z), we …rst …nd X (z) and Y (z):

1 1
X (z) = 1 1
1 3
z 1 1 2z
5
3
z 1 1
= 1
, < jzj < 2
1 3
z 1 (1 2z 1) 3

5 5
Y (z) = 1 2
1 3
z 1 1 3
z 1
5
3
z 1 2
= 1 2
, jzj >
1 3
z 1 1 3
z 1 3

2
 Now
Y (z)
H (z) =
X (z)
1
1 2z 2
= jzj >
1 23 z 1 3
The pole-zero plot of H(z) is plotted below.

(b) Taking the inverse z-transform of H(z), we get

n n 1
2 2
h [n] = u [n] 2 u [n 1]
3 3
n
2
= (u [n] 3u [n 1])
3
(c) Since
1
Y (z) 1 2z
H (z) = = 2 1
,
X (z) 1 3
z
we can write
2 1
Y (z) 1 z = X (z) 1 2z 1 ,
3
whose inverse z-transform leads to
2
y [n] y [n 1] = x [n] 2x [n 1]
3
(d) The system is stable becase the ROC includes the unit circle. It is also causal
since the impulse response h [n] = 0 for n < 0.

4. (0.5%) Page 172, Problem 3.23

An LTI system is characterized by the system function
1
1 2
z 2 1
H (z) = 1 1
, jzj > .
1 2
z 1 1 4
z 1 2

3
 (a) Determine the impulse response of the system.
(b) Determine the di¤erence equation relating the system input x [n] and the system
output y [n].
Sol:
(a)
1
1 2
z 2
H (z) = 1 1 1
1 2
z 1 4
z 1
5 + 27 z 1
= 4+ 3
1 4
z 1 + 18 z 2
2 7
= 4 1 +
1 2
z 1 1 41 z 1

n n
1 1
h [n] = 4 [n] 2 u [n] + 7 u [n]
2 4
(b)
3 1 1
y [n] y [n 1] + y [n 2] = x [n] x [n 2]
4 8 2
5. (0.5%) Page 172, Problem 3.25
Consider a right-sided sequence x [n] with z-transform

1 z2
X (z) = 1
= .
(1 az ) (1 bz 1 ) (z a) (z b)
In Section 3.3, we considered the determination of x [n] by carrying out a partial
fraction expansion, with X (z) considered as a ratio of polynomials in z 1 . Carrry out
a partial fraction expansion of X (z), considered as a ratio of polynomials in z, and
determine x [n] from this expansion.

Sol:
z2 z2
X (z) = =
(z a) (z b) z2 (a + b) z + ab
Obtain a proper fraction:
1
z2 (a + b) z + abj z 2
z2 (a + b) z + ab
(a + b) z ab

(a+b)a ab (a+b)a ab
(a + b) z ab a b b a
X (z) = 1 + =1+ +
(z a) (z b) z a z b
a2 b2
a b a b 1 a2 z 1 b2 z 1
= 1+ =1+ 1 1
z a z b a b 1 az 1 bz

4
 a2 b2
x [n] = [n] + an 1 u [n 1] bn 1 u [n 1]
a b a b
1
= [n] + an+1 bn+1 u [n 1]
a b

6. (0.5%) Page 173, Problem 3.29

A causal LTI system has system function
1 z 1 1 z 1
H (z) = 2
= .
1 0:25z (1 0:5z 1 ) (1 + 0:5z 1 )

(a) Determine the output of the system when the input is x [n] = u [n].
(b) Determine the input x [n] so that the corresponding output of the above system
is y [n] = [n] [n 1].
(c) Determine the output y [n] when the input is x [n] = cos (0:5 n) for 1 < n < 1.
You may leave your answer in any convenient form.
Sol:
1 z 1 1 z 1
H (z) = 2
=
1 0:25z (1 0:5z 1 ) (1 + 0:5z 1 )
1
(a) Given x [n] = u [n], we have X (z) = 1 z 1 , 1 < jzj. Then
Y (z) = H (z) X (z)
1 z 1 1
= 1 1 1
(1 0:5z ) (1 + 0:5z ) 1 z
1
=
(1 0:5z ) (1 + 0:5z 1 )
1
1 1
2 2
= + , 0:5 < jzj .
(1 0:5z 1 ) (1 + 0:5z 1 )
(The ROC for Y (z) includes the intersection of the ROC of H (z) with the ROC
of X (z).)
Inverse z-transform gives
1 1
y [n] = (0:5)n u [n] + ( 0:5)n u [n] .
2 2
(b) If y [n] = [n] [n 1], then Y (z) = 1 z 1 , 0 < jzj. We have
Y (z)
X (z) =
H (z)
1
1 z
= 1 z 1
1 0:25z 2

= 1 0:25z 2 , 0 < jzj .

Inverse z-transform gives
x [n] = [n] 0:25 [n 2] .

5
 (c) Now x [n] = cos (0:5 n), 1 < n < 1. At ! = 0:5 we have
1 e j0:5
H ej0:5 =
1 0:25e j
= 1:13ej 4 .

Then
y [n] = 1:13 cos 0:5 + .
4
7. (0.5%) Page 174, Problem 3.31
Using any method, determine the inverse z-transform for each of the following:
1
(a) X (z) = 2 ,
(1+ 12 z 1 ) (1 2z 1 )(1 3z 1)

(x [n] is a stable sequence)

1
(b) X (z) = ez
z 3 2z
(c) X (z) = z 2
, (x [n] is a left-sided sequence)
Sol:
(a)
1 1
X (z) = < jzj < 2
1 + 21 z 1 2 (1 2z 1 ) (1 3z 1 ) 2
1 58 1568 2700
35 1225 1225 1225
= 2 + +
1+ 1
z 1 1 + 12 z 1 (1 2z 1 ) (1 3z 1 )
2

Therefore,
n n
1 1 58 1
x [n] = (n + 1) u [n + 1] + 2 u [n]
35 2 (35) 2
1568 n 2700 n
+ 2 (2) u [ n 1] 2 (3) u [ n 1]
(35) (35)

(b)
1 z 2 z 3 z 4
X (z) = ez =1+z 1
+ + + + :::
2! 3! 4!
1
Therefore, x [n] = n!
u [n].
(c)
z3 2z 2
X (z) = = z 2 + 2z + 1
jzj < 2
z 2 1 2z
Therefore,
x [n] = [n + 2] + 2 [n + 1] 2 (2)n u [ n 1]

8. (0.5%) Page 174, Problem 3.35

For each of the following sequences, determine the z-transform and ROC, and sketch
the pole-zero diagram:

6
(a) x [n] = an u [n] + bn u [n] + cn u [ n 1] ; jaj < jbj < jcj
(b) x [n] = n2 an u [n]
4 4
(c) x [n] = en cos 12
n u [n] en cos 12
n u [n 1]
Sol:
(a)

x [n] = an u [n] + bn u [n] + cn u [ n 1] jaj < jbj < jcj

1 1 1
X (z) = 1
+ 1
jbj < jzj < jcj
1 az 1 bz 1 cz 1
1 2cz 1 + (bc + ac ab) z 2
X (z) = jbj < jzj < jcj
(1 az 1 ) (1 bz 1 ) (1 cz 1 )
Poles: a, b, c,
Zeros: z1 , z2 , 1 where z1 and z2 are roots of numerator quadratic.
(b)

x [n] = n2 an u [n]
1
x1 [n] = an u [n] , X1 (z) = 1
jzj > a
1 az
d az 1
x2 [n] = nx1 [n] = nan u [n] , X2 (z) = z
X1 (z) = jzj > a
dz (1 az 1 )2
d d az 1
x [n] = nx2 [n] = n2 an u [n] , z X2 (z) = z jzj > a
dz dz (1 az 1 )2
az 1 (1 + az 1 )
X (z) = jzj > a
(1 az 1 )3

(c)
4 4
x [n] = en cos n u [n] en cos n u [n 1]
12 12
4
= en cos n (u [n] u [n 1]) = [n]
12
Therefore, X (z) = 1 for all jzj.

7
9. (0.5%) Page 267, Problem 4.5
Consider the system of Figure 4.10, with the discrete-time system an ideal lowpass
…lter with cuto¤ frequency =8 radians/s.

(a) If xc (t) is bandlimited to 5kHz, what is the maximum value of T that will avoid
aliasing in the C/D converter?
(b) If 1=T = 10kHz, what will the cuto¤ frequency of the e¤ective continuous-time
…lter be?
(c) Repeat part (b) for 1=T = 20kHz.
Sol: A plot of H (ej! ) appears below.

(a)
xc (t) = 0, j j 2 5000
1
The Nyquist rate is 2 times the highest frequency. ) T = 10;000
sec. This avoids
all aliasing in the C/D converter.
(b)
1
= 10kHz
T
! = T
1
= c
8 10; 000
c = 2 625rad/sec
fc = 625Hz

8
 (c)
1
= 20kHz
T
! = T
1
= c
8 20; 000
c = 2 1250rad/sec
fc = 1250Hz

10. (0.5%) Page 271, Problem 4.21

Consider a continuous-time signal xc (t) with Fourier transform Xc (j ) shown in Figure
P4.21-1.

(a) A continuous-time signal xr (t) is obtained through the process shown in Figure
P4.21-2. First, xc (t) is multiplied by an impulse train of period T1 to produce
the waveform xs (t), i.e.,
X
+1
xs (t) = x [n] (t nT1 )
n= 1

Next, xs (t) is passed through a low pass …lter with frequency response Hr (j ).
Hr (j ) is shown in Figure P4.21-3.

9
 Determine the range of values for T1 for which xr (t) = xc (t).
(b) Consider the system in Figure P4.21-4. The system in this case is the same as the
one in part (a), except that the sampling period is now T2 . The system Hs (j )
is some continuous-time ideal LTI …lter. We want xo (t) to be equal to xc (t) for
all t, i.e., xo (t) = xc (t) for some choice of Hs (j ). Find all values of T2 for which
xo (t) = xc (t) is possible. For the largest T2 you determined that would still allow
recovery of xc (t), choose Hs (j ) so that xo (t) = xc (t). Sketch Hs (j ).

Sol:
(a) The impulse-train signal xs (t) has spectrum Xs (j ) given by
1 X
1
2
Xs (j ) = X j k .
T k= 1 T1
An example is shown below.

10
 We will have xr (t) = xc (t) provided T1 0
.
(b) We will have x0 (t) = xc (t) under any of the following circumstances:
1. As illustrated above, T2 0
.
1:5 2
2. As illustrated below, 0
T2 0
.

3
3. As illustrated below, T2 = 0
.

The frequency response of the …lter that is needed to recover xc (t) is shown
below.

11