DSP Additonal Solved Examples

Additional Solved Examples
Chapter 1
Example 1.1. The following systems have input x(n) and output y (n) . For each system,
determine whether it is memoryless, stable, causal, linear, or time invariant.
y ( n ) =  k = x ( k )
2n
(a)
 0 n<0
y (t ) = 
(b)  x (n)  x(n  2) n  0
 0 x ( n) < 0
y ( n) = 
(c)  x(n)  x(n  2) x(n)  0
(d) y (n) = {x(n  1)}
(e) y(n) = x(n )
2
Solution:(a) (i) This system accumulates the input from  to 2n . The present output depends
on the past, present, and future inputs. We may conclude, this system is a with-memory system.
(ii) Assume that the input signal x(n) satisfies the condition
| x(n) | Bx <  foralln
We then find that
2n
| y (n) |=  x(k )
k = 
2n
=  | x(k ) |
k = 
2n
= B
k = 
x
=
B
We conclude that in this system if any input is bounded by an arbitrary positive number x , the
corresponding output is unbounded. Thus, the given system is unstable.
(iii) In this system, the present output anticipates with the future inputs (the upper limit of
the summation is 2n ). Consequently, the given system is noncausal.
(iv) To determine whether or not a system is linear, we consider two arbitrary inputs
x1 (n) x (n)
and 2 .
2n
x1 (n)  y1 (n) =  x (k )
k = 
1
2n
x2 (n)  y2 (n) =  x (k )
k = 
2
© Oxford University Press, All rights reserved.

2n
x3 ( n)  y3 ( n) =  x (k )
k = 
3
x3 (n) x ( n) x (n)
Let be a linear combination of 1 and 2 . That is,
x3 (n) = ax1 (n)  bx2 (n)
where a and b are arbitrary scalars. If the system is linear, then
y3 (n) = ay1 (n)  by2 (n)
Considering the LHS of this equation,
2n
y3 ( n) =  x (k )
k = 
3
2n
=  [ax (k )  bx (k )]
k = 
1 2
2n 2n
= a  x1 ( k )  b  x2 (k )
k =  k = 
= ay1 (n)  by2 (n) = RHS

Since LHS = RHS , we conclude that the system is linear.
x ( n)
(v) Let 1 be an arbitrary input to the system and let
2n
y1 (n) =  x (k )
k = 
1
x1 (n)
be the corresponding output. Then consider a second input obtained by shifting in time:
x2 (n) = x1 (n  no )
The output corresponding to this input is
2n
y2 ( n ) =  x (k )
k = 
2
2n
=  x (k  n )
k = 
1 o
2 n  no 2n
= 
k = 
x1 (k )Now,considery1 (n) =  x (k )
k = 
1
2( n  no )
y1 (n  no ) = 
k = 
x1 (k )
y2 (n)  y1 (n  no )
We observe that , and therefore, this system is not time invariant.
(b) (i) This system has memory, since the value of the output signal y (n) at time n
depends on the present and one past input signal x(n) .
 0 n<0
| x( n) | Bx <  foralln.Wethenfindthat | y (n) |= 
| x( n)  x( n  2) | n  0

 0 n<0

| x(n) |  | x(n  2) | n  0
 0 n<0

2 Bx n  0
B
corresponding output is guaranteed to be bounded. Thus, the given system is stable.
(iii) This system is causal, since the present output y (n) at time n depends on the
present and past values of the input signal x(n) .
x1 (n) x (n)
and 2 .
 0 n<0
x1 (n)  y1 (n) = 
 x1 (n)  x1 (n  2) n  0
 0 n<0
x2 (n)  y2 (n) = 
 x2 (n)  x2 (n  2) n  0
 0 n<0
x3 (n)  y3 (n) = 
 x3 (n)  x3 (n  2) n  0
x (n) x ( n) x (n)
Let 3 be a linear combination of 1 and 2 . That is,
x3 (n) = ax1 (n)  bx2 (n)
y3 (n) = ay1 (n)  by2 (n)
 0 n<0
y3 (n) = 
 x3 (n)  x3 (n  2) n  0
 0 n<0

=
[ax (n  2)  bx ( n  2)] n  0
 1 2
 0 n<0

=  a[ x1 (n)  x1 (n  2)]
  b[ x (n)  x ( n  2)] n  0
 2 2
= ay1 (n)  by2 (n) = RHS

x ( n)

 0 n<0
y1 (n) = 
 x1 (n)  x1 (n  2) n  0
x1 (n)
x2 (n) = x1 (n  no )
 0 n<0
y2 ( n ) = 
 x2 (n)  x2 (n  2) n  0
0 n < 0  0 n<0
= Now,considery1 (n) = 
 n0  x1 (n)  x1 ( n  2) n  0
 0 n  no < 0

y1 (n  no ) =  x1 (n  no )
 x (n  n  2) n  n  0
 1 o o
 0 n < no

=  x1 (n  no )
  x (n  n  2) n  n
 1 o o
y (n)  y1 (n  no )
We see that 2 , and therefore, this system is not time invariant.
(d) (i) This system has memory, since the value of the output signal y (n) at time n
depends on the present and one past input signal x(n) .
 0 x ( n) < 0
| x( n) | Bx <  foralln.Wethenfindthat | y (n) |= 
| x( n)  x( n  2) | x( n)  0
 0 x ( n) < 0

| x( n) |  | x( n  2) | x( n)  0
 0 x ( n) < 0

 2 Bx x( n)  0
B
(iii) This system is causal, since the present output y (n) at time n depends on the
present and past values of the input signal x(n) .
x1 (n) x (n)
and 2 .
 0 x1 ( n) < 0
x1 (n)  y1 ( n) = 
 x1 (n)  x1 (n  2) x1 ( n)  0

 0 x2 ( n) < 0
x2 (n)  y2 ( n) = 
 x2 ( n)  x2 ( n  2) x2 ( n)  0
 0 x3 ( n) < 0
x3 ( n)  y3 (n) = 
 x3 ( n)  x3 ( n  2) x3 ( n)  0
x3 (n) x ( n) x (n)
x3 (n) = ax1 (n)  bx2 (n)
y3 (n) = ay1 (n)  by2 (n)
 0 x3 (n) < 0
y3 (n) = 
 x3 (n)  x3 (n  2) x3 (n)  0
 0 (ax1 (n)  bx2 (n)) < 0


=
 [ax1 (n  2)
 bx2 (n  2)] (ax1 (n)  bx2 (n))  0
 0 (ax1 (n)  bx2 (n)) < 0
a[ x (n)  x (n  2)]
 1 1
=
  b[ x2 (n)
  x2 (n  2)] (ax1 (n)  bx2 (n))  0
 RHS
Since LHS  RHS , we conclude that the system is non-linear.
x ( n)
 0 x1 ( n) < 0
y1 ( n) = 
 x1 ( n)  x1 ( n  2) x1 ( n)  0
x1 (n)
x2 (n) = x1 (n  no )
 0 x2 (n) < 0
y2 ( n ) = 
 x2 ( n)  x2 ( n  2) x2 ( n)  0
 0 x1 (n  no ) < 0
  0 x1 (n) < 0
=  x1 (n  no ) Now,considery1 (n) = 
 x (n  n  2) x (n  n )  0  x1 (n)  x1 (n  2) x1 (n)  0
 1 o 1 o

 0 x1 (n  no ) < 0

y1 ( n  no ) =  x1 ( n  no )
  x (n  n  2) x ( n  n )  0
 1 o 1 o
y (n) = y1 (n  no )
We see that 2 , and therefore, this system is time invariant.
(d)
x(n  1)  x(  n  1)
y ( n) = {x(n  1)}or,equivalentlyy (n) =
2
(i) This system has memory, since the value of the output signal y (n) at time n depends on the
present and one past input signal x(n) .
x ( n  1)  x (  n  1)
| x ( n) | Bx <  foralln.Wethenfindthat | y ( n) |=
2
x ( n  1) x (  n  1)
 
2 2
 Bx
B
(iii) This system is noncausal, since the present output y (n) at time n depends on the
future values of the input signal x(n) .
x1 (n) x (n)
and 2 .
x (n  1)  x1 (  n  1)
x1 (n)  y1 ( n) = 1
2
x (n  1)  x2 (  n  1)
x2 (n)  y2 (n) = 2
2
x (n  1)  x3 (  n  1)
x3 (n)  y3 (n) = 3
2
x (n) x ( n) x (n)
Let 3 be a linear combination of 1 and 2 . That is,
x3 (n) = ax1 (n)  bx2 (n)
y3 (n) = ay1 (n)  by2 (n)
x (n  1)  x3 (  n  1)
y3 (n) = 3
2
1
= [ax1 (n  1)  bx2 (n  1)]
2

 [ax1 (n  1)  bx2 (n  1)]
= a[ x1 (n  1)  x1 (n  1)]
b[ x2 (n  1)  x2 (n  1)]
= ay1 (n)  by2 (n) = RHS
x ( n)
x ( n  1)  x1 (  n  1)
y1 ( n) = 1
2
x1 (n)
x2 (n) = x1 (n  no )
x (n  1)  x2 (  n  1)
y2 ( n ) = 2
2
x (n  no  1)  x1 (  n  no  1)
= 1 Now,consider
2
x ( n  1)  x1 (  n  1)
y1 ( n) = 1
2
x (n  no  1)  x1 (  n  no  1)
y1 (n  no ) = 1
2
y (n)  y1 (n  no )
We see that 2 , and therefore, this system is not time invariant.
(e) (i) This system has memory, since the value of the output signal y (n) at time n
depends on the future input.
y (n) |n = n = y (no ) = x(no2 )andy (n) |n =  n = y (  no ) = x( no2 )
o o
| x ( n) | Bx <  foralln.Wethenfindthat | y (n) |=| x (n 2 ) |
= Bx
B
(iii) This system is noncausal, since the present output y (n) at time n depends on the
future values of the input signal x(n) .
x1 (n) x (n)
and 2 .
x1 ( n)  y1 ( n) = x1 (n 2 )
x2 ( n)  y2 ( n) = x2 ( n 2 )

x3 ( n)  y3 ( n) = x3 ( n 2 )
x3 (n) x ( n) x (n)
x3 (n) = ax1 (n)  bx2 (n)
y3 (n) = ay1 (n)  by2 (n)
y3 ( n) = x3 ( n 2 )
= ax1 (n 2 )  bx2 (n 2 )
= ay1 (n)  by2 (n) = RHS
x ( n)
y1 (n) = x1 (n 2 )
x1 (n)
x2 (n) = x1 (n  no )
y2 ( n) = x2 ( n 2 )
= x1 (n 2  no )Now,consider
y1 (n) = x1 (n 2 )
y1 ( n  no ) = x1[(n  no ) 2 ]
y2 (n)  y1 (n  no )
We see that and, therefore, this system is not time invariant.
Example 1.2. An arbitrary real-valued discrete-time signal x(n) occupies the entire interval
 < n <  . Show that the energy of the signal x(n) is equal to the sum of the energy of the
x (n) x (n)
even component e and the energy of the odd component o .
Solution: See page 37, property no. 4 of discrete-time even and odd signals.

Chapter 6
Example 6.1. Determine the z -transform of the signal
n 0  n  N 1
x ( n) = 
N N n
Solution: Given that
n 0  n  N 1
x ( n) = 
N N n
= n[u(n)  u (n  N )]  Nu(n  N )
= nu (n)  nu (n  N )  Nu (n  N )
x(n) = nu(n)  (n  N )u (n  N )
We know that (Eq. 6.34)
z 1
nu (n)  , | z |> 1
(1  z 1 ) 2
Using time-shifting property, we obtain
z 1
(n  N )u (n  N )  z  N , | z |> 1
(1  z 1 ) 2
Therefore,
z 1 N z 1
X ( z) =  z
(1  z 1 )2 (1  z 1 ) 2
z 1 (1  z  N )
X ( z) = , | z |> 1
(1  z 1 ) 2
Example 6.2. Let

x ( n) = ( 1) n u ( n)  a n u (  n  n0 )

Determine the constraints on the complex number a and the integer n0 , given that the ROC of
X ( z ) is the region 1 <| z |< 2 .
x ( n) = ( 1) n u ( n)  a n u (  n  n0 )
We know that
1
( 1) n u (n)  , | z |> 1
1  z 1
and
a z 1 n0 1
a u(n  n )  
n
, | z |<| a |
1  az 1
0
Therefore,
X ( z ) =   ( 1) n u ( n)     a n u (  n  n0 ) 
 
n 1
1 a 1 z 0
X ( z) =  , 1 <| z |<| a |
1  z 1 1  az 1
Given that the ROC of X ( z ) is the region 1 <| z |< 2 ; therefore, a = 2 and n0 can take on any
value.
Example 6.3. Determine the inverse z -transform of
 
1 1024  z 10 
X ( z) =  , | z |> 0
1024  1  1 z 1 
 2 

 
1 1024  z 10 
X ( z) =  , | z |> 0
1024  1  1 z 1 
 2 
1 1 z 10
X ( z) = 
1 1
1  z 1 1024 1  z 1
2 2
The inverse z -transform of this equation yields

n n 10
 1 1  1
x ( n) =   u ( n)    u (n  10)
 2 1024  2 
n
 1
=   u ( n)
 2
10 n
1  1  1
     u (n  10)
1024  2  2
n
 1
x(n) =   [u (n)  u (n  10)]
 2
Example 6.4. A causal signal x(n) has the z -transform
X ( z ) = sin( z 1 )(1  3 z 2  2 z 4 )
Find x(11) .
X ( z ) = sin( z 1 )(1  3 z 2  2 z 4 )
z 3 z 5 z 7
= ( z 1   
3! 5! 7!
z 9 z 11
   )(1  3 z 2  2 z 4 )
9! 11!

= x ( n ) z  n
n =0

Since x(11) is simply the coefficient in front of z in this power series expansion of X ( z ) ,
11
therefore
1 3 2
x(11) =   
11! 9! 7!
Example 6.5. Determine the signal x(n) with z -transform

1
X ( z) = ez  e z

1
X ( z) = e  e
z z
1
= ez  ez
 z 2 z3 
= 1  z    
 2! 3! 
 z 2 z 3 
  1  z 1    
 2! 3! 
 1
1    1 
= 1   z  n     z  n 
 n =  n!   n =0 n! 
  1 1
1 
= 1    z n   z n 
 n =0 n! n =  n! 

1
X ( z) = 1   n! z
n = 
n
1
x ( n) =  ( n) 
n!

Example 6.6. Determine all possible signals that can have the following z -transform.
1
X ( z) =
1  1.5 z  0.5 z 2
1
1
X ( z) =
1  1.5 z  0.5 z 2
1
z2
= 2
z  1.5 z  0.5
X ( z) z
=
z ( z  0.5)( z  1)
Using partial-fraction expansion, we obtain
X ( z) 1 2
= 
z z  0.5 z  1
z 2z
X ( z) = 
z  0.5 z  1
1 2
X ( z) =  1

1  0.5 z 1  z 1
X ( z ) has two poles at z = 0.5 and z = 1 . There are three possible choices of the ROCs: (a)
| z |> 1 (b) 0.5 <| z |< 1 (c) | z |< 0.5 .
Case I: ROC | z |> 1 (shown in Fig. 6.1a). Since the ROC, | z |> 1 , is the region in the z -
plane outside the outermost pole, all the poles correspond to causal (right-sided) signals. Now
consider
1 2
X ( z) =  1
 , | z |> 1
1  0.5 z 1  z 1
The inverse z -transform yields
x(n) = (0.5) n u (n)  2u (n)
= [2  (0.5) n ]u (n)

Case II: ROC 0.5 <| z |< 1 (shown in Fig. 6.1b). The pole of the first term is at 0.5 . The
ROC has a radius greater than the pole at z = 0.5 , so this pole corresponds to causal (right-sided)
signal. Therefore,
1
 (0.5) n u (n)  
1  0.5 z 1
The second term has a pole at z = 1 . The ROC has a radius less than the pole at z = 1 , so this
pole corresponds to the anticausal (left-sided) signal. Therefore,
2
2u (  n  1) 
1  z 1
and, hence, we obtain
x(n) =  (0.5) n u (n)  2u (  n  1)
Case III: ROC | z |< 0.5 (shown in Fig. 6.1c). Since the ROC, | z |< 0.5 , is the region in the z -
plane inside the innermost pole, all the poles correspond to anticausal (right-sided) signals. Now
consider
1 2
X ( z) =  1
 , | z |> 1
1  0.5 z 1  z 1
The inverse z -transform yields
x(n) = (0.5)n u (  n  1)  2u (  n  1)
= [2  (0.5)n ]u (  n  1)

Fig. 6.1 Possiible ROCs for Example 6.1 (aa) Case I (b) C
Case II (c) Casee III

Example 6.7. (a) State and prove the initial-value theorem for anticausal signals (i.e., x(n) = 0 ,
n > 0 ).
(b) Show that if x(n) is causal [i.e., x(n) = 0 for n < 0 ], then
x(1) = limz[ X ( z )  x (0)]

z 
Solution: (a) If x(n) is anticausal [i.e., x(n) = 0 for n > 0 ], then
x (0) = lim X ( z )
z0
Proof If x(n) is anticausal [i.e., x(n) = 0 for n > 0 ], then its z -transform is given by
0
X ( z) =  x ( n) z
n = 
n
= x(0)  x( 1) z  x( 2) z  
lim X ( z ) = lim[ x (0)  x ( 1) z  x ( 2) z  ]

z0 z0
= x(0)
(b) If x(n) is causal [i.e., x(n) = 0 for n < 0 ], then its z -transform is given by

X ( z ) = x ( n ) z  n
n =0
= x(0)  x(1) z 1  x(2) z 2
 x(3) z 3 
X ( z )  x(0) = x(1) z 1  x(2) z 2
 x(3) z 3 
z[ X ( z )  x(0)] = x(1)  x(2) z 1
 x(3) z 2 

1
limz[ X ( z )  x(0)] = lim[ x(1)  x(2) z
z  z 
 x(3) z 2 ]
limz[ X ( z )  x (0)] = x (1)

z 
Example 6.8. Let y(n) be a signal that is generated from a signal x(n) as follows:
n
y ( n) =  kx(k )
k = 
(a) Show that y(n) satisfies the time-varying difference equation
y(n) = y(n  1)  nx(n)
and show that
 z 2 dX ( z )
Y ( z) =
z  1 dz
where X ( z ) and Y ( z ) are the z -transforms of x(n) and y(n) , respectively.
(b) Use this property to find the z -transform of
n k
 1
y ( n) = k   , n0
k =0
 3
Solution: (a) From the definition of y(n) , we see that

n 1
y (n  1) =  kx(k )
k = 
Given that
n
y ( n) =  kx(k )
k = 
n 1
= nx(n)   kx(k )
k = 

y(n) = nx(n)  y(n  1)
y(n)  y(n  1) = x(n)
Taking the z -transform on both sides, we get
dX ( z )
Y ( z )[1  z 1 ] =  z
dz
 z dX ( z )
Y ( z) =
1  z 1 dz
 z 2 dX ( z )
=
z  1 dz
(b) To find the z -transform of the given signal, note that

n
y ( n) =  kx(k )
k = 
where
n
 1
x( n) =   u ( n)
 3
Because the z -transform of x(n) is
1 1
X ( z) = , | z |>
1 1 3
1 z
3
then
 z 2 dX ( z )
Y ( z) =
z  1 dz
1
 z 2
z2 3
=
z  1  1 1  2
1  z 
3

1 1
z
Y ( z) = 3 , | z |> 1
2
1  1 1 
(1  z ) 1  z 
 3 
Example 6.9. Let the input to an LTI system is

n
 1
x ( n) =   u (n)  (2) n u (  n  1)
 3
and 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 poles and zeros of H ( z) and indicate
the ROC.
(b) Find the impulse response h(n) of the system.
(c) Write a difference equation that is satisfied by the given input and output.
(d) Is the system stable? Is it causal?
Solution: (a) Given that

n
 1
x ( n) =   u (n)  (2) n u (  n  1)
 3
Taking the z -transform of this equation, we obtain
1 1 1
X ( z) =  , <| z |< 2
1 1 1  2 z 1 3
1 z
3
5
 z 1
= 3
 1 1  1
 1  z  (1  2 z )
3
Now consider the given output

n n
 1  2
y ( n) = 5   u ( n)  5   u ( n)
 3  3
he z -transfo
Taking th orm of this equation, we obtain
5 5 2
Y ( z) =  , | z |>
1 1 2 1 3
1 z 1 z
3 3
5
 z 1
= 3
 1 1   2 1 
1  z  1  z 
3 3
Now
Y ( z ) 1  2 z 1 2
H ( z) = = , | z |>
X ( z ) 1  2 z 1 3
3
The pole-zero plot off H ( z) is sh

hown in Fig. 6.2
Fig. 6.2 Pole-zero plot of H(z) of 6.9a

(b) Consider
1  2 z 1 2
H ( z) = , | z |>
2 1 3
1 z
3
1 2 z 1
= 
2 2
1  z 1 1  z 1
3 3

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
Y ( z ) 1  2 z 1
H ( z) = =
X ( z ) 1  2 z 1
3
 2 
Y ( z ) 1  z 1  = X ( z ) 1  2 z 1 
 3 
2 1
Y ( z)  z Y ( z ) = X ( z )  2 z 1 X ( z )
3
The inverse z -transform of this equation leads to
2
y ( n)  y (n  1) = x( n)  2 x( n  1)
3
(d) The system is stable because the ROC includes the unit circle. It is also causal since the
impulse response h(n) = 0 for n < 0 .

1
X ( z ) = e z , with ROC all z except |z|=0

x
Solution: According to the power series expansion of e ,

xn
e =
x
n =0 n!
Therefore,
n
1
1 
X ( z) = e z =  z
n =0 n!

z n
=
n =0 n!

1
=  n! u (n) z
n = 
n
Comparing this equation with


X ( z) =  x ( n) z
n = 
n
we get
1
x( n) = u (n)
n!

2
X ( z ) = e z , with ROC all z except |z|=
x
Solution: According to the power series expansion of e ,

xn
e =
x
n =0 n!
Therefore,

( z 2 )n
X ( z) = e =  z2
n =0 n!

0
( z 2 ) n
= 
n =  (  n)!

1
X ( z) =  (n)! u(n) z
n = 
2 n
m
n=
A change of variables is performed by letting m = 2 n , which also yields 2 , m =  as
n =  , and m =  as n =  . Therefore,

1  m
X ( z) = 
m =  
u    z m
m  2 
   !
2

1  n
=  u    z n
n  2
   !
n = 
2
Comparing this equation with


X ( z) =  x ( n) z
n = 
n
we get
1  n
x ( n) = u 
 n  2
   !
2
Since
 n  0 n > 0 or n odd
u  = 
 2  1 otherwise
we get
 0 n > 0 or n odd
 1
x ( n) =  otherwise
   n !
 
  2 

Example 6.12. Let x(n) be a discrete-time signal with x(n) = 0 for n  0 and z -transform
X ( z ) . Furthermore, given x(n) , let the discrete-time signal y(n) be defined by
1
 x (n) n>0
y ( n) =  n
 0 otherwise
(a) Compute Y ( z ) in terms of X ( z ) .
(b) Using the result of part (a), find the z -transform of
1
w(n) = u (n  1)
n   ( n)
Solution: (a) Given that
1
 x (n) n>0
y ( n) =  n
 0 otherwise
Therefore, for n > 0 , we have
x(n)
y ( n) =
n
ny(n) = x(n)
Taking the z -transform of this equation, we obtain
dY ( z )
z = X ( z)
dz
dY ( z )
=  z 1 X ( z )
dz
Y ( z ) =  z 1 X ( z )dz
(b) Given that

1
w(n) = u (n  1)
n   ( n)
For n > 0 , we obtain
1
w( n) = u ( n  1)
n
To apply the result of part (a), we let x(n) = u (n  1) and w(n) = y(n) . Therefore,
W ( z ) =  z 1 X ( z )dz
z 1
=   z 1 dz
1  z 1
1
=  dz
z ( z  1)
 1 1 
W ( z) =    dz
 z z  1
1 1
=  dz   dz
z z 1
= ln( z )  ln( z  1)
 z 
W ( z ) = ln 
 z  1
Example 6.13. Find the z -transform of the digital signal obtained by sampling the analog
4 t
signal x(t ) = e sin(4t )u (t ) at intervals of T = 0.1 second.
Solution: The signal x(n) is obtained by sampling the continuous-time signal
x(t ) = e 4t sin(4t )u (t )
every T = 0.1 sec. Therefore,
x(t ) |t = nT = x(nT )

x(n) = e 4 nT sin(4nT )u (nT )
= e 4 nT sin(4nT )u (n)
x(n) = e 0.4 n sin(0.4n)u (n)
We know that (Eq. 6.19)
[sin(0.4n)u (n)]
z 1sin(0.4)
= , | z |> 1
1  2 z 1cos(0.4)  z 2
Now using scaling in z -domain property (Eq. 6.26), we obtain
[(e 0.4 ) n sin(0.4n)u (n)]
e 0.4 z 1sin(0.4)
= , | z |> 1
1  2e 0.4 z 1cos(0.4)  e 0.8 z 2
Example 6.14. Find the z -transform and its ROC for the signal
x(n) = a n [u (n)  u (n  N )], a  0
In addition, determine and sketch the poles and zeros of the z -transform for N = 4 .
Solution: Consider the given signal
x(n) = a n [u (n)  u (n  N )], a  0
= a n u (n)  a n u ( n  N )
x ( n) = a n u ( n )  a N a n  N u ( n  N )
Taking the z -transform of this equation and using the time-shifting property, we obtain
1 zN
X ( z) =  a N
1  az 1 1  az 1
 
N
1 az 1
= 
1  az 1 1  az 1

 
N
1  az 1
X ( z) =
1  az 1
For N = 4 , we have
 
4
1  az 1
X ( z) =
1  az 1
 
1  az 1 2  1  az 1 2   
= 
1

1  az

 

1  az 1 2  (1  az 1 )(1  az
a 1 )
=
1  az 1
 
X ( z ) = 1  az 1  (1  az 1 )
2
 
( z 2  a 2 )( z  a )
=
z3
d z =  ja , and
It has thrree zeros at z =  a , and a three pooles at z = 0 . The pole--zero diagram
m for
N = 4 is shown in Fiig. 6.3.
Fig o plot for N = 4 for Exampple 6.14

g. 6.3 Pole-zero

Chapter 7
Example 7.1. Let x(n) be a causal N -point signal that is zero outside the range 0  n  N  1 .
When x(n) is the input to the causal LTI system represented by the difference equation
1 1
y ( n)  y (n  2) = x(n  2)  x(n)
4 4
the output is y(n) , also a causal N -point signal.
(a) Show that the causal LTI system described by this difference equation represents an
all-pass filter.
(b) Given that

N 1
 | x ( n) | = 5
n =0
2
determine the value of

N 1
 | y ( n) |
n =0
2
Solution: Consider the difference equation
1 1
y ( n)  y (n  2) = x(n  2)  x(n)
4 4
1 1
Y ( z )  z 2Y ( z ) = z 2 X ( z )  X ( z )
4 4
1
Y ( z) z 2 
H ( z) = = 4
X ( z ) 1  1 z 2
4
1
  z 2
H ( z) = 4
1
1  z 2
4
1
Since the poles and zeros (poles at z =  and zeros at z = 2 ) occur in conjugate reciprocal
2

pairs, the system is all-pass. The coefficients of the numerator and denominator polynomial are
in reversed order, and also H ( z ) H ( z 1 ) = 1 .
(b) It is a property of all-pass systems that the output energy is equal to the input energy
(refer to Section 7.7.1). Thus
N 1 N 1
 | x ( n) | =  | y ( n) | = 5
n =0
2
n =0
2
Example 7.2. An IIR LTI discrete-time system is described by the difference equation
y (n)  a1 y (n  1)  a2 y (n  2)
= b0 x(n)  b1 x(n  1)  b2 x(n  2))
where x(n) and y(n) are input and output, respectively. Find the frequency response. For what
values of bi , 1  i  2 , will be the magnitude response constant for all values of  ?
Solution: Consider the given difference equation
y (n)  a1 y (n  1)  a2 y (n  2)
= b0 x(n)  b1 x(n  1)  b2 x(n  2)
Taking DTFT, we get
Y (e j )  a1e  j Y (e j )  a2 e  j 2 Y (e j )
= b0 X (e j )  b1e  j X (e j )  b2 e  j 2 X (e j )
Y ( e j ) j b0  b1e  j  b2 e  j 2
= H (e ) =
X ( e j ) 1  a1e  j  a2 e  j 2
The magnitude response is constant for all frequencies if
b0 = a2 , b1 = a1 , b2 = 1

Chapteer 8
wo sequencees x1 (n) andd x2 ( n) . Thhe value of x2 (n) at n = 3 is

Example 8.1 Figure 8.1 shows tw
unknown
n, but is show
wn as variab 8 shows x3 (n) , the four-point circcular convollution
ble a . Fig. 8.2
of x1 (n) and x2 ( n) . Based on th
he graph of y (n) , can yoou specify a uniquely? Iff so, what is a? If
d yield the seequence x3 (n) as shownn.
not, give two possiblle values of a that would
Fig. 8.1
Fig. 8.2
Fig. 8.3
Solution:: From Fig. 8.1,

8 we can write
w

x2 (n) =  (n  1)  a (n  3)
The four-point circular convolution of x1 (n) and x2 ( n) is given by
x3 (n) = x1 (n)  x2 (n)
= x1 (n)  [ (n  1)  a (n  3)]
x3 (n) = x1 ( n  1 4 )  ax1 ( n  3 4 )
From Fig. 8.1, we can write
x1 ( n) = {1,0,2,1}

x1 ( n  1 4 ) = {1,1,0,2}

x1 ( n  2 4 ) = {2,1,1,0}

x1 ( n  3 4 ) = {0,2,1,1}

ax1 ( n  3 4 ) = {0,2a,a,a}

Thus,
x3 (n) = x1 ( n  1 4 )  ax1 ( n  3 4 )
x3 ( n) = {1,2a  1,a,2  a}

x3 ( n) is shown in Fig. 8.3. Matching this sequence to the one given in Fig. 8.2, we get a = 1 ,
which is unique.
Example 8.2 The even part of a real sequence x(n) is defined by
x ( n)  x (  n)
xe (n) =
2
Suppose that x(n) is real finite-length sequence defined such that x( n) = 0 for n < 0 and
n  N . Let X ( k ) denote the N-point DFT of x(n) .

(a) Is e{ X ( k )} the DFT of xe (n) ?
(b) What is the inverse DFT of e{ X ( k )} in terms of x(n) ?
Solution: (a) If x(n) is of length N , then xe ( n) is of length 2 N  1 .
 x ( n)  x (  n )
 N 1  n  N 1
xe (n) =  2
 0 otherwise
Taking the (2 N  1) -point DFT of xe ( n) , we get (  N  1  k  N  1 )

N 1
X e (k ) = 
n =  N 1
xe (n)W2nkN 1
N 1
 x(n)  x(  n)  nk
= 
n =  N 1

2
 W2 N 1
0
x(  n) nk N 1
x(n) nk
= 
n =  N 1 2
W2 N 1  
n =0 2
W2 N 1
N 1
x(n)  nk N 1 x(n) nk
= W2 N 1   W2 N 1
n =0 2 n =0 2
x(n) j 2 2N1nk N 1 x(n)  j 2 2N1nk

N 1
= e  e
n =0 2 n =0 2
N 1
 2 nk 
X e (k ) = x(n) cos 
n =0
 2 N  1
The N-point DFT of x(n) is given by

N 1
X (k ) = x(n)WNkn
n =0
N 1 2
j
=  x ( n )e
kn
N
n =0
Its real value is given by

N 1
 2 kn 
e{ X (k )} = x(n) cos 
n =0
 N 
So DFT[ xe (n)]  e{ X (k )} .
(b)
X (k )  X * (k )
e{ X ( k )} =
2
*
1 N 1 1  N 1 
= x(n)WNnk  x(n)WNnk 
2 n=0 2  n =0 
1 N 1 1 N 1
= 
2 n =0
x(n)WNnk  x* (n)WN nk
2 n =0
1 N 1 1 N 1
= 
2 n =0
x(n)WNnk  x(n)WN nk
2 n =0
1 N 1
e{ X (k )} = 
2 n =0
[ x(n)  x( n N )]WNnk
 x(n)  x( n  N ) 
e{ X ( k )} = DFT  
 2
Example 8.3 If X ( k ) is the N-point DFT of the sequence x(n) , determine the N-point DFT of
the following sequences in terms of X ( k ) :
 2 
xc (n) = x(n) cos  k0 n ,0  n  N  1
(a)  N 
 2 
xs (n) = x(n) sin  k0 n ,0  n  N  1
(b)  N 
Solution: (a) Taking the DFT of xc (n) and using the multiplication property (Eq. 8.101), we get
 2 
xc (n) = x(n) cos  k 0 n
 N 

1    2  
X c (k ) =  X (k )  DFT cos  k 0 n  
N   N  
1  N 
=
N  X (k )  2 [ ( k  k0  N )   ( k  k0  N )]
 X (k )  [ ( k  k0  N )   ( k  k0  N )]
1
=
2
 X ( k  k0  N )  X ( k  k0  N ) 
1
X c (k ) =
2
(b) Taking the DFT of xs (n) and using the multiplication property (Eq. 8.101), we get
 2 
xs (n) = x(n) sin  k 0 n
 N 
1    2  
X s (k ) =  X (k )  DFT sin  k 0 n  
N   N  
1  N 
=
N  X ( k )  2 j [ ( k  k0  N )   ( k  k0  N )]
 
 X (k )  [ ( k  k0  N )   ( k  k0  N )]
1
=
2j
 X ( k  k 0  N )  X ( k  k0  N )
1
X s (k ) =
2j
Example 8.4 Let X ( k ) be the N -point DFT of the sequence x(n) , 0  n  N  1 . What is the
N-point DFT of the sequence g ( n) = X (n) , 0  n  N  1 ?
Solution: See duality property (Section 8.5.7).
Example 8.5 Let X ( k ) be the N-point DFT of length-N sequence x(n) . Prove the following
properties of DFT:
(a) X ( k ) is real and circularly even when x(n) is real and circularly even.

(b) X ( k ) is imaginary and circularly odd when x(n) is real and circularly odd.
Solution: (a) See case II of Section 8.5.6 (Eq. 8.77).
(b) See case III of Section 8.5.6 (Eq. 8.77).
Example 8.6 Let x(n) be a real-valued sequence of length-N, and let X ( k ) be its N-point DFT
with real and imaginary parts X R (k ) and X I (k ) respectively.
(a) Show that X R (k ) = X R ( N  k ) , 0  k  N  1
(b) Show that X I (k ) =  X I ( N  k ) , 0  k  N  1
Solution: (a) See case VII of Section 8.5.6 (Eq. 8.87).
(b) See case VII of Section 8.5.6 (Eq. 8.89).
Example 8.7 Show that the DFT values of a real sequence are always complex conjugate.
Solution: See case I of Section 8.5.6 (Eq. 8.76).

Chapter 9
Example 9.1 How many complex additions and complex multiplications are required for the
direct computation of an N-point DFT?
Solution: See Section 9.2
Example 9.2 Briefly explain Goertzel algorithm for efficient DFT computations.
Solution: See Section 9.7.1
Example 9.3 Draw a signal flow graph for 8-point DFT using DIF-FFT algorithm. Obtain its
computational complexity.
Solution: See Section 9.4
Example 9.4 Draw a signal flow graph for 8-point DFT using DIT-FFT algorithm. Obtain its
computational complexity.
Solution: See Section 9.3 .
Example 9.5 What is in-place algorithm and what is the advantage of this algorithm?
Solution: See Section 9.3.2 .
Example 9.6 Compare the computational complexity of DFT and FFT.

Solution: See Section 9.4.1 .
Example 9.7 Explain the inverse FFT algorithm to compute inverse DFT for N = 8 . Draw the
signal flow graph for the same.
Solution: See Section 9.6 .
Example 9.8 Show that the direct computation of the N-point DFT of a length-N sequence
requires 4N real multiplications and (4 N  2) N real additions.
2
Solution: See the four points of Section 9.2 .

Chapter 10
Example 10. 1 Discuss the factors that influence the choice of structure for realization of an LTI
system.
Solution: See Section 10.3.
Example 10.2 What are the various building blocks required in realizations of digital systems?
Solution: See Section 10.4.1.
Example 10.3 Discuss cascade and parallel forms for realization structures of IIR systems.
Solution: See Sections 10.6.3 and 10.6.4.
Example 10.4 Explain the different structures for realization of an IIR system and explain how
conversion can be made from direct form I to direct form II structure.
Example 10.5 Discuss transposed form structures with an example.
Example 10.6 Discuss direct form and cascade and linear-phase realization structures of FIR
filters.
Solution: See Sections 10.5.1, 10.5.2, and 10.5.3.

Chapter 11
Example 11.1 Write the various window functions used for FIR filter design. Compare their
important performance parameter.
Solution: See Sections 11.8.1–11.8.6 for rectangular, Bartlett, Hanning, Hamming, Blackman,
and Kaiser windows. For comparison of their important performance parameter, see Table 11.2.
Example 11.2 Derive the frequency response of a linear-phase FIR filter with symmetric impulse
response and filter length M odd.
Example 11.3 Explain windowing method for the design of FIR filters.
Example 11.4 Describe the features of a window function.

Example 11.5 Derive the condition for constant phase and constant group delay in FIR filters. If
M is even, derive an expression for the frequency response of a linear-phase FIR filter.
Solution: See Section 11.2.2 for constant phase and constant group delay in FIR filters. See
Section 11.3.2 for expression of the frequency response of a linear-phase FIR filter.
Example 11.6 Derive the expression of Type-4 FIR filters.

Example 11.7 Discuss the design of FIR differentiators.

Example 11.8 Give frequency response of an ideal Hilbert transformer. Find its impulse
response.
Example 11.9 Derive the frequency response of a linear-phase FIR filter with symmetric impulse
response and filter length M odd.
Example 11.10 Derive the condition for linear-phase in FIR filters. Why is linear phase
important?
Solution: See Section 11.2.2 and 11.2.
Example 11.11 Briefly explain Alternation Theorem. What is the application of this theorem in
the FIR filter design?
Example 11.12 Explain equiripple approximations for a Type-1 FIR Filter.

Solution: See Example 11.50.

Example 11.13 Discuss the utility of the Parks–McClellan algorithm.

Chapter 12

Example 12.3 Define an IIR filter and bring out the constraints to be maintained in
conversion of an analog filter into digital filter.
Example 12.4 Discuss how aliasing effect occurs in designing of IIR filters using impulse
invariant method.

Example 12.5 Derive bilinear transformation method and discuss its mapping properties.
Solution: See Section 12.7 and 12.7.1.
Example 12.6 What is warping effect? Discuss influence of warping effect on amplitude
response and phase response of a derived digital filter from a corresponding analog filter.
Solution: See Section 12.7.1 and 12.7.2.
Example 12.7 Compare and contrast bilinear and impulse-invariant transformation technique.
Solution: See Section 12.7.5 and 12.7.6.
Example 12.8 Discuss in detail the procedure of designing an analog filter using Butterworth
approximation technique.
Solution: See Section 12.10 and 12.10.2.
Example 12.9 Derive frequency transformation relation to convert a lowpass filter to a highpass
filter.

Chapter 13
Example 13.1 Explain the quantization errors due to rounding and truncation in fixed-point and
floating-point arithmetic.
Solution: Refer Section 13.4 and 13.5.
Example 13.2 Assuming two’s complement fixed-point number system. Find range of rounding
off error due to quantization. Find mean and variance of this error.
Solution: Refer Section 13.7 and 13.7.1.
Example 13.3 What is the mean and variance of the output noise if the system is h(n) and input
noise mean and variance are μe and σ2e respectively.
Solution: Refer Section 13.7.4.
Example 13.4 For a finite length register with B +1 bits and the rounding off as the quantization
method, find the (i) mean μe of the quantization error e(n), (ii) variance σ2e of the quantization
error e(n). If this error signal is passed through a digital system h(n). What would be the mean
and variance of the output sequence?
Solution: Refer Section 13.7.1 and 13.7.4.
Example 13.5 Discuss in detail about addition overflow errors and their remedies.
Solution: See Example 13.7 and Sections 13.9.4 and 13.9.5.
Example 13.6 What are limit cycles? Explain its types in brief.
Solution: Refer Section 13.9.

Chapter 14
Example 14.1 Determine the z-transform of the output of a decimator.
Solution: See proof of Eq. (14.11) in Section 14.2.2.
Example 14.2 Determine the z-transform of the output of an interpolator.
Solution: See proof of Eq. (14.21) in Section 14.3.2.
Example 14.3 Show that the upsampler and downsampler satisfy the property of commutation if
they are co-prime.

DSP Additonal Solved Examples

Uploaded by

DSP Additonal Solved Examples

Uploaded by

Additional Solved Examples

© Oxford University Press, All rights reserved.

= ay1 (n)  by2 (n) = RHS

© Oxford University Press, All rights reserved.

= ay1 (n)  by2 (n) = RHS

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

Example 6.1. Determine the z -transform of the signal

Solution: Given that

We know that (Eq. 6.34)

Using time-shifting property, we obtain

Example 6.2. Let

© Oxford University Press, All rights reserved.

Solution: Given that

Example 6.3. Determine the inverse z -transform of

Solution: Given that

© Oxford University Press, All rights reserved.

The inverse z -transform of this equation yields

Example 6.4. A causal signal x(n) has the z -transform

Solution: Given that

© Oxford University Press, All rights reserved.

Example 6.5. Determine the signal x(n) with z -transform

Solution: Given that

The inverse z -transform of this equation yields

© Oxford University Press, All rights reserved.

Solution: Given that

Using partial-fraction expansion, we obtain

The inverse z -transform yields

x(n) = (0.5) n u (n)  2u (n)

© Oxford University Press, All rights reserved.

and, hence, we obtain

x(n) =  (0.5) n u (n)  2u (  n  1)

The inverse z -transform yields

© Oxford University Press, All rights reserved.

© Oxford University Press, All rights reserved.

x(1) = limz[ X ( z )  x (0)]

Solution: (a) If x(n) is anticausal [i.e., x(n) = 0 for n > 0 ], then

= x(0)  x( 1) z  x( 2) z  

lim X ( z ) = lim[ x (0)  x ( 1) z  x ( 2) z  ]

= x(0)  x(1) z 1  x(2) z 2

X ( z )  x(0) = x(1) z 1  x(2) z 2

z[ X ( z )  x(0)] = x(1)  x(2) z 1

© Oxford University Press, All rights reserved.

limz[ X ( z )  x (0)] = x (1)

(a) Show that y(n) satisfies the time-varying difference equation

y(n) = y(n  1)  nx(n)

and show that

where X ( z ) and Y ( z ) are the z -transforms of x(n) and y(n) , respectively.

(b) Use this property to find the z -transform of

Solution: (a) From the definition of y(n) , we see that

© Oxford University Press, All rights reserved.

y(n)  y(n  1) = x(n)

Taking the z -transform on both sides, we get

(b) To find the z -transform of the given signal, note that

Because the z -transform of x(n) is

© Oxford University Press, All rights reserved.

Example 6.9. Let the input to an LTI system is

and the corresponding output is

(b) Find the impulse response h(n) of the system.

(d) Is the system stable? Is it causal?

Solution: (a) Given that

Taking the z -transform of this equation, we obtain

Now consider the given output

© Oxford University Press, All rights reserved.

The pole-zero plot off H ( z) is sh

Fig. 6.2 Pole-zero plot of H(z) of 6.9a

© Oxford University Press, All rights reserved.

The inverse z -transform of this equation yields

The inverse z -transform of this equation leads to

Example 6.10. Determine the inverse z -transform of

© Oxford University Press, All rights reserved.

Comparing this equation with