Chapter 2

Discrete-Time Signals and Systems

1. Problem P2.1:
(a) x1 (n) = 10
m=0 (m + 1) [ (n

2m

1)

(n

2m)] ; 0  n  25.

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P2.1ac);
%
% x1(n) = sum_{m=0}{10} (m+1)*[delta(n-2*m)-delta(n-2*m-1)]
n1 = [0:25]; x1 = zeros(1,length(n1));
for m = 0:10
x1 = x1 + (m+1)*(impseq(2*m,0,25) - impseq(2*m+1,0,25));
end
subplot(2,1,1); stem(n1,x1);
axis([min(n1)-1,max(n1)+1,min(x1)-2,max(x1)+2]);
xlabel(n); ylabel(x1(n)); title(Sequence x1(n));
ntick = [n1(1):n1(length(n1))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plots of x1 (n) is shown in Figure 2.1.
(b) x2 (n) = n2 [u (n + 5)

u (n

6)] + 10 (n) + 20 (0:5)n [u (n

4)

u (n

10)].

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P2.1be);
%
% (b) x2(n) = (n2)*[u(n+5)-u(n-6)]+10*delta(n)+20*(0.5)n*[u(n-4)-u(n-10)]
n2 = -5:10; % Overall support of x2(n)
x2 = (n2.2).*(stepseq(-5,-5,10)-stepseq(6,-5,10))+10*impseq(0,-5,10)+...
20*((0.5).n2).*(stepseq(4,-5,10)-stepseq(10,-5,10));
subplot(2,1,1); stem(n2,x2);
axis([min(n2)-1,max(n2)+1,min(x2)-2,max(x2)+2]);
xlabel(n); ylabel(x1(n)); title(Sequence x2(n));
ntick = [n2(1):n2(length(n2))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plots of x2 (n) is shown in Figure 2.3.
(c) x3 (n) = (0:9)n cos (0:2n + =3) ;

0  n  20.

% x3(n) = (0.9)n*cos(0.2*pi*n+pi/3); 0<=n<=20

n3 = [0:20];
3

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

A PRIL 98

Homework1 : Problem 1
Sequence x1(n)
10

x1(n)

5
0
5
10
0

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

n
Sequence x3(n)
1

x3(n)

0.5
0
0.5
1

10 11 12 13 14 15 16 17 18 19 20

Figure 2.1: Problem P2.1 Sequence Plots

x3 = ((0.9).n3).*cos(0.2*pi*n3+pi/3);
subplot(2,1,2); stem(n3,x3);
axis([min(n3)-1,max(n3)+1,-1,1]);
xlabel(n); ylabel(x3(n)); title(Sequence x3(n));
ntick = [n3(1):n3(length(n3))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plots of x3 (n) is shown in Figure 2.1.
(d) x4 (n) = 10 cos 0:0008n2
[ 1; 1].

+ w (n) ; 0

 n  100 where w (n) is a random sequence uniformly distributed between

clear; close all;

% (d) x4(n) = 10*cos(0.0008*pi*n.2)+w(n); 0 <= n <= 100; w(n)uniform[-1,1]
w = 2*(rand(1,101)-0.5);
n4 = [0:100]; x4 = 10*cos(0.0008*pi*n4.2)+w;
subplot(2,1,2); stem(n4,x4); axis([min(n4)-1,max(n4)+1,min(x4)-2,max(x4)+2]);
xlabel(n); ylabel(x4(n)); title(Sequence x4(n));
ntick = [n4(1):10:n4(length(n4))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plot of x4 (n) is shown in Figure 2.2 from which we observe that it is a noisy sinusoid with increasing frequency
(or a noisy chirp signal).


(e) x5 (n) =

:::;

1; 2; 3; 2; 1; 2; 3; 2; 1; : : : . Plot 5 periods.

"

% (e) x5(n) = {...,1,2,3,2,1,2,3,2,1,...}periodic. 5 periods

n5 = [-8:11]; x5 = [2,1,2,3];
x5 = x5*ones(1,5); x5 = (x5(:));
subplot(2,1,2); stem(n5,x5);
axis([min(n5)-1,max(n5)+1,0,4]);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

Sequence x4(n)
10

x4(n)

5
0
5
10
0

10

20

30

40

50

60

70

80

90

100

Figure 2.2: Plot of the sequence x4 (n) in Problem P2.1d.

xlabel(n); ylabel(x5(n)); title(Sequence x5(n));
ntick = [n5(1):n5(length(n5))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plots of x5 (n) is shown in Figure 2.3.
Problem P2.1 : Sequences
Sequence x2(n)
25

x1(n)

20
15
10
5
0
5

10

n
Sequence x5(n)
4

x5(n)

3
2
1
0

8 7 6 5 4 3 2 1

10

11

Figure 2.3: Problem P2.1 Sequence Plots

2. Problem P2.2: The sequence x (n) = f1; 2; 4; 6; 5; 8; 10g is given.
(a) x1 (n) = 3x(n + 2) + x(n

4)

2x (n).

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P2.2ab);

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

A PRIL 98

n = [-4:2]; x = [1,-2,4,6,-5,8,10]; % given seq x(n)

%
% (a) x1(n) = 3*x(n+2) + x(n-4) - 2*x(n)
[x11,n11] = sigshift(3*x,n,-2);
% shift by -2 and scale by 3
[x12,n12] = sigshift(x,n,4);
% shift x(n) by 4
[x13,n13] = sigadd(x11,n11,x12,n12);
% add two sequences at time
[x1,n1] = sigadd(x13,n13,2*x,n);
% add two sequences
subplot(2,1,1); stem(n1,x1);
axis([min(n1)-1,max(n1)+1,min(x1)-2,max(x1)+2]);
xlabel(n); ylabel(x1(n)); title(Sequence x1(n));
ntick = [n1(1):1:n1(length(n1))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10);
The plot of x1 (n) is shown in Figure 2.4.
(b) x2 (n) = 5x (5 + n) + 4x (n + 4) + 3x (n).
% (b) x2(n) = 5*x(5+n) + 4*x(n+4) +3*x(n)
[x21,n21] = sigshift(5*x,n,-5);
[x22,n22] = sigshift(4*x,n,-4);
[x23,n23] = sigadd(x21,n21,x22,n22);
[x2,n2] = sigadd(x23,n23,3*x,n);
subplot(2,1,2); stem(n2,x2);
axis([min(n2)-1,max(n2)+1,min(x2)-0.5,max(x2)+0.5]);
xlabel(n); ylabel(x2(n)); title(Sequence x2(n));
ntick = [n2(1):1:n2(length(n2))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plot of x2 (n) is shown in Figure 2.4.
Homework1 : Problem 2
Sequence x1(n)

x1(n)

30
20
10
0

n
Sequence x2(n)

x2(n)

60
40
20
0
9

Figure 2.4: Problem P2.2 Sequence Plots

(c) x3 (n) = x (n + 4) x(n

1) + x(2

n)x (n).

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P2.2cd);
n = [-4:2]; x = [1,-2,4,6,-5,8,10]; % given seq x(n)
%
% (c) x3(n) = x(n+4)*x(n-1) + x(2-n)*x(n)
[x31,n31] = sigshift(x,n,-4);
% shift x(n) by -4
[x32,n32] = sigshift(x,n,1);
% shift x(n) by 1
[x33,n33] = sigmult(x31,n31,x32,n32); % multiply two sequences
[x34,n34] = sigfold(x,n);
% fold x(n)
[x34,n34] = sigshift(x34,n34,2);
% shift x(-n) ny 2
[x34,n34] = sigmult(x34,n34,x,n);
% shift x(-n) ny 2
[x3,n3] = sigadd(x33,n33,x34,n34);
% add two sequences
subplot(2,1,1); stem(n3,x3);
axis([min(n3)-1,max(n3)+1,min(x3)-2,max(x3)+2]);
xlabel(n); ylabel(x3(n)); title(Sequence x3(n));
ntick = [n3(1):1:n3(length(n3))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10);
The plots of x3 (n) is shown in Figure 2.5.
(d) x4 (n) = 2e0:5n x (n) + cos (0:1n) x (n + 2) ;

10  n  10.

% (d) x4(n) = 2*exp(0.5*n)*x(n)+cos(0.1*pi*n)*x(n+2); -10 <= n <= 10

n4 = [-10:10]; x41 = 2*exp(0.5*n4); x412 = cos(0.1*pi*n4);
[x42,n42] = sigmult(x41,n4,x,n);
[x43,n43] = sigshift(x,n,-2);
[x44,n44] = sigmult(x412,n4,x43,n43);
[x4,n4] = sigadd(x42,n42,x44,n44);
subplot(2,1,2); stem(n4,x4);
axis([min(n4)-1,max(n4)+1,min(x4)-0.5,max(x4)+0.5]);
xlabel(n); ylabel(x4(n)); title(Sequence x4(n));
ntick = [n4(1):1:n4(length(n4))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
The plots of x4 (n) is shown in Figure 2.5.
(e) x5 (n) = 5k=1 nx (n

k) where x (n) = f1; 2; 4; 6; 5; 8; 10g.

clear; close all;

n = [-4:2]; x = [1,-2,4,6,-5,8,10]; % given seq x(n)
% (e) x5(n) = sum_{k=1}{5}n*x(n-k);
[x51,n51] = sigshift(x,n,1); [x52,n52] = sigshift(x,n,2);
[x5,n5] = sigadd(x51,n51,x52,n52);
[x53,n53] = sigshift(x,n,3); [x5,n5] = sigadd(x5,n5,x53,n53);
[x54,n54] = sigshift(x,n,4); [x5,n5] = sigadd(x5,n5,x54,n54);
[x55,n55] = sigshift(x,n,5); [x5,n5] = sigadd(x5,n5,x55,n55);
[x5,n5] = sigmult(x5,n5,n5,n5);
subplot(2,1,2); stem(n5,x5); axis([min(n5)-1,max(n5)+1,min(x5)-2,max(x5)+2]);
xlabel(n); ylabel(x5(n)); title(Sequence x5(n));
ntick = [n5(1):1:n5(length(n5))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10);
The plot of x5 (n) is shown in Figure 2.6.
3. Problem P2.3: A sequence x (n) is periodic if x (n + N ) = x (n) for all n. Consider a complex exponential sequence
e j0 n = e j2 f0 n .

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

A PRIL 98

Problem P2.2 : Sequences

Sequence x3(n)
60
40

x3(n)

20
0
20
40
8

n
Sequence x4(n)
50

x4(n)

40
30
20
10
0
10 9 8 7 6 5 4 3 2 1

Figure 2.5: Problem P2.2 Sequence Plots

(a) Analytical proof: The above sequence is periodic if
e j2 f0 (n+N ) = e j2 f0 n
or
e j2 f0 N

=1

) f0 N = K (an integer)

which proves the result.

(b) x1 (n) = cos(0:3n);

20  n  20.

% (b) x1(n) = cos(0.3*pi*n)

x1 = cos(0.3*pi*n);
subplot(2,1,1); stem(n,x1);
axis([min(n)-1,max(n)+1,-1.1,1.1]);

Sequence x5(n)
100

x5(n)

80
60
40
20
0
3

Figure 2.6: Plot of the sequence x5 (n) in Problem P2.5e.

10

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP USING M ATLAB

ylabel(x1(n)); title(Sequence cos(0.3*pi*n));

ntick = [n(1):5:n(length(n))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10);
Since f0 = 0:3=2 = 3=20 the sequence is periodic. From the plot in Figure 2.7 we see that in one period of 20
samples x1 (n) exhibits three cycles. This is true whenever K and N are relatively prime.
(c) x2 (n) = cos(0:3n);

20  n  20.

% (b) x2(n) = cos(0.3*n)

x2 = cos(0.3*n);
subplot(2,1,2); stem(n,x2);
axis([min(n)-1,max(n)+1,-1.1,1.1]);
ylabel(x2(n)); title(Sequence cos(0.3*n));
ntick = [n(1):5:n(length(n))];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10);
In this case f0 is not a rational number and hence the sequence x2 (n) is not periodic. This can be clearly seen from
the plot of x2 (n) in Figure 2.7.

Problem 2.3
Sequence cos(0.3*pi*n)
1

x1(n)

0.5
0
0.5
1
20

15

10

10

15

20

10

15

20

Sequence cos(0.3*n)
1

x2(n)

0.5
0
0.5
1
20

15

10

Figure 2.7: Problem P2.3 Sequence Plots

4. Problem P2.5: Even-odd decomposition of complex-valued sequences.
(a) M ATLAB function evenodd:
function [xe, xo, m] = evenodd(x,n)
% Complex-valued signal decomposition into even and odd parts
% ----------------------------------------------------------% [xe, xo, m] = evenodd(x,n)
%
[xc,nc] = sigfold(conj(x),n);
[xe,m] = sigadd(0.5*x,n,0.5*xc,nc);
[xo,m] = sigadd(0.5*x,n,-0.5*xc,nc);

10

S OLUTIONS M ANUAL

FOR

(0:4n)

(b) Even-odd decomposition of x(n) = 10e

DSP

USING

M ATLAB

A PRIL 98

0  n  10.

n = 0:10; x = 10*exp(-0.4*pi*n);
[xe,xo,neo] = evenodd(x,n);
Re_xe = real(xe); Im_xe = imag(xe);
Re_xo = real(xo); Im_xo = imag(xo);
% Plots of the sequences
subplot(2,2,1); stem(neo,Re_xe);
ylabel(Re{xe(n)}); title(Real part of Even Seq.);
subplot(2,2,3); stem(neo,Im_xe);
xlabel(n); ylabel(Im{xe(n)}); title(Imag part of Even Seq.);
subplot(2,2,2); stem(neo,Re_xo);
ylabel(Re{xo(n)}); title(Real part of Odd Seq.);
subplot(2,2,4); stem(neo,Im_xo);
xlabel(n); ylabel(Im{xo(n)}); title(Imag part of Odd Seq.);
The M ATLAB verification plots are shown in Figure 2.8.
Real part of Odd Seq.
1.5

1
Re{xo(n)}

Re{xe(n)}

Real part of Even Seq.

10

6
4
2
0
10

0.5
0
0.5
1

1.5
10

10

0.5

0.5

0
0.5
1
10

10

Imag part of Odd Seq.

Im{xo(n)}

Im{xe(n)}

Imag part of Even Seq.

0
0.5

0
n

10

1
10

0
n

10

Figure 2.8: Plots in Problem P2.5

5. Problem P2.12: Properties of linear convolution.
[x1 (n)

x1 (n)  x2 (n)
 x2 (n)]  x3 (n)

x1 (n)  [x2 (n) + x3 (n)]

x (n)  (n n0)

=
=
=
=

x2 (n)  x1 (n)
x1 (n)  [x2 (n)  x3 (n)]
x1 (n)  x2 (n) + x1 (n)  x3 (n)
x ( n n0 )

:
:
:
:

Commutation
Association
Distribution
Identity

(a) Commutation:
x1 (n)  x2 (n)

k=

x 1 (k ) x 2 (n

k) =

| {z }
=m

m=

x 1 (n

m) x2 (m)

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

m=

USING

M ATLAB

11

m ) = x 2 (n )  x 1 ( n )

x2 (m) x1 (n

Association:
[x1 (n)

 x2 (n)]  x3 (n)

"
=

x1 (k) x2 (n k)
k=

x1 ( k ) x2 ( m
m= k=

 x3 (n)

k=

x 1 (k )

x1 (k) [x2 (n

k=

x 2 (m

k)x (n
| {z } 3

m)5

=`

k=

m=

"

m)

x 1 (k ) 4

k ) x 3 (n

m=

x2 (`) x3 (n
k )  x 3 (n

`)

k)] = x1 (n)  [x2 (n)  x3 (n)]

Distribution:
x1 (n)  [x2 (n) + x3 (n)]

x1 (k) [x2 (n

k ) + x 3 (n

x1 (k) x2 (n

k) +

k=

k=

Identity:
x (n)  (n
since (n

n0

k) = 1 for k = n

x (k) (n

k=

n0 and zero elsewhere.

(b) Verification using M ATLAB:

n1 = -10:20; x1 = n1;
n2 = 0:30; x2 = cos(0.1*pi*n2);
n3 = -5:10; x3 = (1.2).n3;
% Commutative Property
[y1,ny1] = conv_m(x1,n1,x2,n2);
[y2,ny2] = conv_m(x2,n2,x1,n1);
ydiff = max(abs(y1-y2))
ydiff =
4.2633e-014
ndiff = max(abs(ny1-ny2))
ndiff =
0
% Associative Property
[y1,ny1] = conv_m(x1,n1,x2,n2);
[y1,ny1] = conv_m(y1,ny1,x3,n3);
[y2,ny2] = conv_m(x2,n2,x3,n3);
[y2,ny2] = conv_m(x1,n1,y2,ny2);
ydiff = max(abs(y1-y2))

k=

x1 (k) x3 (n

x 1 (n )  x 2 ( n ) + x 1 (n )  x 3 ( n )

n0 ) =

k)]

n0

k) = x (n

n0 )

k)

12

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

ydiff =
6.8212e-013
ndiff = max(abs(ny1-ny2))
ndiff =
0
% Distributive Property
[y1,ny1] = sigadd(x2,n2,x3,n3);
[y1,ny1] = conv_m(x1,n1,y1,ny1);
[y2,ny2] = conv_m(x1,n1,x2,n2);
[y3,ny3] = conv_m(x1,n1,x3,n3);
[y2,ny2] = sigadd(y2,ny2,y3,ny3);
ydiff = max(abs(y1-y2))
ydiff =
1.7053e-013
ndiff = max(abs(ny1-ny2))
ndiff =
0
% Identity Property
n0 = fix(100*(rand(1,1)-0.5));
[dl,ndl] = impseq(n0,n0,n0);
[y1,ny1] = conv_m(x1,n1,dl,ndl);
[y2,ny2] = sigshift(x1,n1,n0);
ydiff = max(abs(y1-y2))
ydiff =
0
ndiff = max(abs(ny1-ny2))
ndiff =
0
6. Problem P2.13: Linear convolution as a matrix-vector multiplication. Consider the sequences
x (n) = f1; 2; 3; 4g and h (n) = f3; 2; 1g
(a) The linear convolution of the above two sequences is
y (n) = f3; 8; 14; 20; 11; 4g
(b) The vector representation of the above operation is:
2
6
6
6
6
6
6
4

3
8
14
20
11
4

7 6
7 6
7 6
7=6
7 6
7 6
5 4

| {z }

3
2
1
0
0
0

0
3
2
1
0
0

{z

0
0
3
2
1
0

0
0
0
3
2
1

3
2
7
7
76
76
74
7
5
|
}

1
2 7
7
3 5
4

{z }
x

(c) Note that the matrix H has an interesting structure. Each diagonal of H contains the same number. Such a matrix is
called a Toeplitz matrix. It is characterized by the following property
[H]i; j = [H]i j

which is similar to the definition of time-invariance.

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

13

(d) Note carefully that the first column of H contains the impulse response vector h (n) followed by number of zeros
equal to the number of x (n) values minus one. The first row contains the first element of h (n) followed by the same
number of zeros as in the first column. Using this information and the above property we can generate the whole
Topelitz matrix.
7. Problem P2.14:
(a) The M ATLAB function conv tp:
function [y,H]=conv_tp(h,x)
% Linear Convolution using Toeplitz Matrix
% ---------------------------------------% [y,H] = conv_tp(h,x)
% y = output sequence in column vector form
% H = Toeplitz matrix corresponding to sequence h so that y = Hx
% h = Impulse response sequence in column vector form
% x = input sequence in column vector form
%
Nx = length(x); Nh = length(h);
hc = [h; zeros(Nx-1, 1)];
hr = [h(1),zeros(1,Nx-1)];
H = toeplitz(hc,hr);
y = H*x;
(b) M ATLAB verification:
x = [1,2,3,4]; h = [3,2,1];
[y,H] = conv_tp(h,x); y = y, H
y =
3
8
14
20
11
H =
3
0
0
0
2
3
0
0
1
2
3
0
0
1
2
3
0
0
1
2
0
0
0
1

8. Problem P2.15: Let x(n) = (0:8)n u(n).

(a) Convolution y(n) = x(n)  x(n):
y ( n)

=
=
=

k
u (n k)
k=
k=0 


n
n
k
n
k
u (n) = (0:8)n (8=8)k u (n)
(0:8) (0:8) (0:8)
k =0
k=0
n
n
(0:8) (n + 1) u (n) = (n + 1) (0:8) u (n)

x (k ) x (n

k) = (0:8)k (0:8)n

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P2.15);
% (a) analytical solution: y(n) = (n+1)*(0.8)(n+1)*u(n)
na = [0:50]; ya = (na+1).*(0.8).(na);
subplot(2,1,1); stem(na,ya); axis([-1,51,-1,3]);
xlabel(n); ylabel(ya(n)); title(Analytical computation);
(b) To use the M ATLABs filter funtion we have to represent one of the x (n) sequence by coefficients an equivalent
difference equation. See Example 2.10 on page 32 for this procedure. MATLAB solution using the filter
function:

14

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

% (b) use of the filter function

nb = [0:50]; x = (0.8).nb;
yb = filter(1,[1, -0.8],x);
subplot(2,1,2); stem(nb,yb); axis([-1,51,-1,3])
xlabel(n); ylabel(yb(n)); title(Filter output);
%
error = max(abs(ya-yb))
error =
4.4409e-016
%
% Super Title
suptitle(Problem P2.15);
The analytical solution to the convolution in (8a) is the exact answer. In the filter function approach of (8b), the
infinite-duration sequence x(n) is exactly represented by coefficients of an equivalent filter. Therefore, the filter solution
should be exact except that it is evaluated up to the length of the input sequence. The plots of this solution are shown in
Figure 2.9.
Problem P2.15
Analytical computation
3

ya(n)

2
1
0
1

10

15

20

10

15

20

25
30
n
Filter output

35

40

45

50

35

40

45

50

yb(n)

2
1
0
1

25
n

30

Figure 2.9: Problem P2.15 Plots

Chapter 3

Discrete-Time Fourier Transform

1. Problem P3.1:
A M ATLAB function to compute DTFT:
function [X] = dtft(x,n,w)
% Computes Discrete-time Fourier Transform
% [X] = dtft(x,n,w)
%
% X = DTFT values computed at w frequencies
% x = finite duration sequence over n (row vector)
% n = sample position row vector
% W = frequency row vector
X = x*exp(-j*n*w);
2. Problem P3.2


(a) Part (c): x (n) = 4; 3; 2; 1; 2; 3; 4

"

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.2c);
%
n = 0:6; x = [4,3,2,1,2,3,4];
w = [0:1:500]*pi/500;
X = dtft(x,n,w); magX = abs(X); phaX = angle(X);
%
% Magnitude Response Plot
subplot(2,1,1); plot(w/pi,magX);grid;
xlabel(frequency in pi units); ylabel(|X|);
title(Magnitude Response);
wtick = [0:0.2:1]; magtick = [0;10;20];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,magtick)
%
% Phase response plot
subplot(2,1,2); plot(w/pi,phaX*180/pi);grid;
xlabel(frequency in pi units); ylabel(Degrees);
title(Phase Response); axis([0,1,-180,180])
wtick = [0:0.2:1]; phatick = [-180;0;180];
15

16

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,phatick)
The plots are shown in Figure 3.1. The angle plot for this signal is a linear function of .
Homework1 : Problem 4
Magnitude Response

|X|

20

10

0
0

0.2

0.4
0.6
frequency in pi units
Phase Response

0.8

0.2

0.4
0.6
frequency in pi units

0.8

Degrees

180

180
0

Figure 3.1: Frequency Response Plots in Problem P3.2c



(b) Part (d): x (n) = 4; 3; 2; 1; 1; 2; 3; 4

"

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.2d);
%
% (d) DTFT of x(n) = {4,3,2,1,1,2,3,4}
n = 0:7; x = [4,3,2,1,1,2,3,4];
w = [0:1:500]*pi/500;
X = dtft(x,n,w); magX = abs(X); phaX = angle(X);
%
% Magnitude Response Plot
subplot(2,1,1); plot(w/pi,magX);grid;
xlabel(frequency in pi units); ylabel(|X|);
title(Magnitude Response);
wtick = [0:0.2:1]; magtick = [0;10;20];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,magtick)
%
% Phase response plot
subplot(2,1,2); plot(w/pi,phaX*180/pi);grid;
xlabel(frequency in pi units); ylabel(Degrees);
title(Phase Response); axis([0,1,-180,180])
wtick = [0:0.2:1]; phatick = [-180;0;180];

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

17

set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,phatick)
The plots are shown in Figure 3.2. The angle plot for this signal is a linear function of .

Problem 3.2
Magnitude Response

|X|

20

10

0
0

0.2

0.4
0.6
frequency in pi units
Phase Response

0.8

0.2

0.4
0.6
frequency in pi units

0.8

Degrees

180

180
0

Figure 3.2: Frequency Response Plots in Problem P3.2d

3. Problem P3.3:
(a) x (n) = 3 (0:9)n u (n). The DTFT is given by
X e j

3 (0:9)n e
n=0

3
0:9e

jn

=3

0:9e

j n

n=0

M ATLAB script for magnitude and angle plot:

% Problem P3.3 : Magnitude and Angle Plot of DTFT
clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.3);
w = [0:200]*pi/200; Z = exp(-j*w); ZZ = Z.2;
% (a) x(n) = 3*(0.9)n*u(n)
X = 3*(1-0.9*Z).(-1); X_mag = abs(X); X_pha = angle(X)*180/pi;
subplot(5,2,1); plot(w/pi,X_mag); axis([0,1,0,max(X_mag)]);
title(Magnitude Plots,fontweight,bold); ylabel(a.);
set(gca,YTickMode,manual,YTick,[0,max(X_mag)],FontSize,10);
subplot(5,2,2); plot(w/pi,X_pha); axis([0,1,-180,180]);
title(Angle Plots,fontweight,bold)
set(gca,YTickMode,manual,YTick,[-180,0,180],FontSize,10);

18

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

The plots are given in Figure 3.3.

(b) x (n) = 2 (0:8)n+2 u (n

2). The sequence can be written as

x ( n)

2+4

2 (0:8)n

2) = 2 (0:8)4 (0:8)n

0:8192 (0:8)n 2 u (n

u (n

u (n

2)

2)

Now using properties of DTFT, we obtain

X e j

= 0:8192

e j2
1 0:8e

M ATLAB script for magnitude and angle plot:

% (b) x(n) = 2*(0.8)(n+2)*u(n-2)
X = 0.8192*ZZ./(1-0.8*Z); X_mag = abs(X); X_pha = angle(X)*180/pi;
subplot(5,2,3); plot(w/pi,X_mag); axis([0,1,0,max(X_mag)]); ylabel(b.);
set(gca,YTickMode,manual,YTick,[0,max(X_mag)],FontSize,10);
subplot(5,2,4); plot(w/pi,X_pha); axis([0,1,-180,180]);
set(gca,YTickMode,manual,YTick,[-180,0,180],FontSize,10);
The plots are given in Figure 3.3.
(c) x (n) = n (0:5)n u (n). The DTFT of (0:5)n u (n) is given by 1= 1

(0 5)n e

jn

n=0

0:5e
1
0:5e

, that is,

Differentiating both sides with respect to , we obtain

j n (0:5)n e

jn

=(

1)

n =0

1
(1

0:5e

0:5e

j)2

j)

Hence
X e j

F fn (0:5)n u (n)g = n (0:5)n e

jn

n=0

0:5e
(1

0:5e

j)2

M ATLAB script for magnitude and angle plot:

% (c) x(n) = n*(0.5)n*u(n)
X = (0.5)*Z./((1-0.5*Z).(2)); X_mag = abs(X); X_pha = angle(X)*180/pi;
subplot(5,2,5); plot(w/pi,X_mag); axis([0,1,0,max(X_mag)]); ylabel(c.);
set(gca,YTickMode,manual,YTick,[0,max(X_mag)],FontSize,10);
subplot(5,2,6); plot(w/pi,X_pha); axis([0,1,-180,180]);
set(gca,YTickMode,manual,YTick,[-180,0,180],FontSize,10);
The plots are given in Figure 3.3.
(d) x (n) = (n + 2) ( 0:7)n
x ( n)

u (n
=
=
=
=

2). The sequence x (n) can be arranged as

(n
(
(
(

2 + 4)( 0:7)n
0 :7 ) ( n
0 :7 ) ( n
0 :7 ) ( n

2 + 4)(

2+1

u (n

2)

0:7)n 2 u (n

2) (

0:7)n 2 u (n

2) (

0:7)n 2 u (n

2)

2) + 4 ( 0:7) ( 0:7)n
2)

2:8 (

0:7)n 2 u (n

u (n
2)

2)

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

19

Using properties and the result from part (c), we obtain,

X e j

0 :7 ) e
0:49e

j2 (

0:7) e

j]2

[1 + 0:7e
j3

2
[1 + 0:7e j]

2 :8

e j
1 + 0:7e

2 :8

e j
1 + 0:7e

M ATLAB script for magnitude and angle plot:

% (d) x(n) = (n+2)*(-0.7).(n-1)*u(n-1)
X = (0.49)*(Z.*ZZ)./((1+0.7*Z).2) - (2.8*ZZ)./(1+0.7*Z);
X_mag = abs(X); X_pha = angle(X)*180/pi;
subplot(5,2,7); plot(w/pi,X_mag); axis([0,1,0,max(X_mag)]); ylabel(d.);
set(gca,YTickMode,manual,YTick,[0,max(X_mag)],FontSize,10);
subplot(5,2,8); plot(w/pi,X_pha); axis([0,1,-180,180]);
set(gca,YTickMode,manual,YTick,[-180,0,180],FontSize,10);
The plots are given in Figure 3.3.
(e) x (n) = 5 ( 0:9)n cos (0:1n) u (n). The sequence can be written as
x ( n)

5 ( 0:9)n

5
2

e j0:1n + e
2

0:9e j0:1

n

j0:1n

u (n) +

u (n )
5
2

0:9e

j0:1 n

u (n)

Hence the DTFT is given by

X e j

=
=

5=2
5=2
+
1 + 0:9e j0:1e j 1 + 0:9e j0:1e
1 + 0:9 cos (0:1) e j
1 + 1:8 cos (0:1) e j + 0:81e j2

M ATLAB script for magnitude and angle plot:

% (e) x(n) = 5*(-0.9).n*cos(0.1*pi*n)*u(n)
X = (1+0.9*cos(0.1*pi)*Z)./(1+1.8*cos(0.1*pi)*Z+(0.81)*ZZ);
X_mag = abs(X); X_pha = angle(X)*180/pi;
subplot(5,2,9); plot(w/pi,X_mag); axis([0,1,0,max(X_mag)]); ylabel(e.);
set(gca,XTickMode,manual,XTick,[0,0.5,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[0,max(X_mag)],FontSize,10);
subplot(5,2,10); plot(w/pi,X_pha); axis([0,1,-180,180]);
set(gca,XTickMode,manual,XTick,[0,0.5,1],FontSize,10);
The plots are given in Figure 3.3.
4. Problem P3.4: This problem is solved using M ATLAB.
% Problem P3.4 : DTFT of Rectangular pulse
clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.4);
w = [-100:100]*pi/100;
% x(n) = -N:N;
% (a) N = 5
N = 5; n = -N:N; x = ones(1,length(n)); X = dtft(x,n,w); X = real(X); X = X/max(X);

20

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Problem 3.3
Magnitude Plots

Angle Plots
180

180

a.

30

0
0

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

180

b.

4.096

0
0

0.2

0.4

0.6

0.8

180
180

c.

0
0

0.2

0.4

0.6

0.8

180
180

180

d.

14.7778

0
0

0.2

0.4

0.6

0.8

180

e.

5.4629

0
0

0.5

180

0.5

Figure 3.3: Problem P3.3 DTFT Plots

subplot(2,2,1); plot(w/pi,X); axis([-1,1,min(X),1]);

title(DTFT for N = 5,fontweight,bold); ylabel(X);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (b) N = 15
N = 15; n = -N:N; x = ones(1,length(n)); X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,2); plot(w/pi,X); axis([-1,1,min(X),1]);
title(DTFT for N = 15,fontweight,bold); ylabel(X);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (c) N = 25
N = 25; n = -N:N; x = ones(1,length(n)); X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,3); plot(w/pi,X); axis([-1,1,min(X),1]);
title(DTFT for N = 25,fontweight,bold); ylabel(X);
xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (d) N = 100
N = 100; n = -N:N; x = ones(1,length(n)); X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,4); plot(w/pi,X); axis([-1,1,min(X),1]);
title(DTFT for N = 100,fontweight,bold); ylabel(X);
xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

21

% Super Title
suptitle(Problem 3.4);
The plots are shown in Figure 3.4. These plots show that the DTFT of a rectangular pulse is similar to a sinc function
and as N increases the function becomes narrowrer and narrower.
Problem 3.4
DTFT for N = 5

DTFT for N = 15

0.2234
1

0.2166
1

DTFT for N = 25

DTFT for N = 100

0
0.2074
1

0
0.005
1

frequency in pi units

frequency in pi units

Figure 3.4: Problem P3.4 DTFT Plots

5. Problem P3.5: DTFT of a symmetric traingular pulse



TN = 1

jnj  R
N

N (n )

This problem is solved using M ATLAB. It uses the function dtft described in P3.1 ans is given below.
function [X] = dtft(x,n,w)
% Computes Discrete-time Fourier Transform
% [X] = dtft(x,n,w)
%
% X = DTFT values computed at w frequencies
% x = finite duration sequence over n (row vector)
% n = sample position row vector
% W = frequency row vector
X = x*exp(-j*n*w);
The M ATLAB script for the problem is:
clear; close all;
w = [-100:100]*pi/100;

22

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

% x(n) = (1-abs(n)/N)*R_N(n);
% (a) N = 5
N = 5; n = -N:N; x = 1-abs(n)/N; X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,1); plot(w/pi,X); axis([-1,1,0,1]);
title(DTFT for N = 5,fontweight,bold); ylabel(X);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (b) N = 15
N = 15; n = -N:N; x = 1-abs(n)/N; X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,2); plot(w/pi,X); axis([-1,1,0,1]);
title(DTFT for N = 15,fontweight,bold); ylabel(X);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (c) N = 25
N = 25; n = -N:N; x = 1-abs(n)/N; X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,3); plot(w/pi,X); axis([-1,1,0,1]);
title(DTFT for N = 25,fontweight,bold); ylabel(X);
xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% (d) N = 100
N = 100; n = -N:N; x = 1-abs(n)/N; X = dtft(x,n,w); X = real(X); X = X/max(X);
subplot(2,2,4); plot(w/pi,X); axis([-1,1,0,1]);
title(DTFT for N = 100,fontweight,bold); ylabel(X);
xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[-1,0,1],FontSize,10);
set(gca,YTickMode,manual,YTick,[min(X),0,1],FontSize,10); grid;
% Super Title
suptitle(Problem 3.5);
The DTFT plots are shown in Figure 3.5. These plots show that the DTFT of a triangular pulse is similar to a (sinc)2
function and as N increases the function becomes narrowrer and narrower.
6. Problem P3.7: Consider xe (n) =
obtain

F [xe (n)]

1

2 [x (n) + x (

=
=

n)]. Using F [x ( n)] = X  e j (see page 48, properties 4 and 5) we

1
F
fx (n) + x ( n)g = 12 fF [x (n)] + F [x ( n)]g
2






1
X e j + X  e j , XR e j :
2

Similarly

F [xo (n)]

=
=

1
1
F
f
x (n) x ( n)g = fF [x (n)]
2
2






1
X e j
X  e j , jXI e j :
2

M ATLAB verification using x (n) = e j0:1n [u (n)

clear; close all;
%

u (n

20)]:

F [x ( n)]g

A PRIL 98

S OLUTIONS M ANUAL

DSP

FOR

USING

M ATLAB

23

Problem 3.5
DTFT for N = 5

DTFT for N = 15

0
1

0
1

DTFT for N = 25

DTFT for N = 100

0
1

0
1

frequency in pi units

frequency in pi units

Figure 3.5: DTFT plots in Problem P3.5.

n = 0:20; x = exp(0.1*pi*n);
w = [-100:100]*pi/100; X = dtft(x,n,w);
XR = real(X); XI = imag(X);
[xe,xo,neo] = evenodd(x,n);
Xe = dtft(xe,neo,w); Xo = dtft(xo,neo,w);
diff_e = max(abs(XR-Xe))
diff_e =
5.5511e-017
diff_o = max(abs(j*XI-Xo))
diff_o =
6.9389e-017
7. Problem P3.16
(a) Part (b): A digital filter is described by the difference equation
y (n) = x (n) + 2x (n

1) + x (n

2)

0:5y (n

1) + 0:25y (n

2)

The frequency response H (e j ) of the filter: Substituting x (n) = e jn and y (n) = H e jn e jn in the above difference equation and simplifying


1 + 2e j + e j2
H e j =
1 + 0:5e j 0:25e j2
clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.16b);
%
% Analytical calculations of Frequency Response using diff eqn
b = [1,2,1]; a = [1,0.5,0.25];
w = [0:1:500]*pi/500; kb = 0:length(b)-1; ka = 0:length(a)-1;

24

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

H = (b*exp(-j*kb*w)) ./ (a*exp(-j*ka*w));
magH = abs(H); phaH = angle(H);
%
% Magnitude Responses Plot
subplot(2,1,1); plot(w/pi,magH);grid; axis([0,1,0,3]);
xlabel(frequency in pi units); ylabel(|H|);
title(Magnitude Response);
wtick = [0:0.2:1]; magtick = [0:3];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,magtick)
%
% Phase response plot
subplot(2,1,2); plot(w/pi,phaH*180/pi);grid;
xlabel(frequency in pi units); ylabel(Degrees);
title(Phase Response); axis([0,1,-180,180])
wtick = [0:0.2:1]; phatick = [-180;0;180];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,phatick)
The magnitude and phase response plots are shown in Figure 3.6.
Homework1 : Problem 5
Magnitude Response
3

|H|

0
0

0.2

0.4
0.6
frequency in pi units
Phase Response

0.8

0.2

0.4
0.6
frequency in pi units

0.8

Degrees

180

180
0

Figure 3.6: Problem P3.16b Plots

(b) Part (c): A digital filter is described by the difference equation
y (n) = 2x (n) + x (n

1)

0:25y (n

1) + 0:25y (n

2)

The frequency response H (e j ) of the filter: Substituting x (n) = e jn and y (n) = H e jn e jn in the above difference equation and simplifying


2 + e j
H e j =
1 + 0:25e j 0:25e j2

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

25

clear; close all;

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P3.16c);
%
% Analytical calculations of Frequency Response using diff eqn
b = [2,1]; a = [1,0.25,-0.25];
w = [0:1:500]*pi/500; kb = 0:length(b)-1; ka = 0:length(a)-1;
H = (b*exp(-j*kb*w)) ./ (a*exp(-j*ka*w));
magH = abs(H); phaH = angle(H);
%
% Magnitude Responses Plot
subplot(2,1,1); plot(w/pi,magH);grid; axis([0,1,0,3]);
xlabel(frequency in pi units); ylabel(|H|);
title(Magnitude Response);
wtick = [0:0.2:1]; magtick = [0:3];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,magtick)
%
% Phase response plot
subplot(2,1,2); plot(w/pi,phaH*180/pi);grid;
xlabel(frequency in pi units); ylabel(Degrees);
title(Phase Response); axis([0,1,-180,180])
wtick = [0:0.2:1]; phatick = [-180;0;180];
set(gca,XTickMode,manual,XTick,wtick)
set(gca,YTickMode,manual,YTick,phatick)
The magnitude and phase response plots are shown in Figure 3.7.

Problem 3.16c
Magnitude Response
3

|H|

0
0

0.2

0.4
0.6
frequency in pi units
Phase Response

0.8

0.2

0.4
0.6
frequency in pi units

0.8

Degrees

180

180
0

Figure 3.7: Problem P3.16c Plots

8. Problem P3.19: Sampling frequency Fs
h (n) = (0:5)n u (n).

100 sam/sec (or sampling interval Ts

0:01 sec/sam) and impulse response

26

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

(a) xa (t ) = 3 cos (20t ). Hence x (n) = xa (nTs ) = 3 cos (0:2n). Therefore, the digital frequency is 0:2 rad/sam.
(b) The steady-state response when x (n) = 3 cos (0:2n): The frequency response is
H e j

n
= F [h (n)] = F [(0:5) u (n)] =

1
:
0:5e j

At = 0:2, the response is

H e j0:2

1
0:5e j0:2

= 0:6969 (

0:2063c) :

Hence
yss (n) = 3 (0:6969) cos (0:2n

0:2363)

which after D/A conversion gives yss (t ) as

yss;a (t ) = 2:0907 cos (20t

0:2363) :

(c) The steady-state DC gain is obtained by setting = 0 which is equal to H e j0

yss;a (t ) = 6.

2. Hence yss (n) = 3 (2) =

(d) Aliased frequencies of F0 for the given sampling rate Fs are F0 + kFs . Now for F0 = 10 Hz and Fs = 100, the aliased
frequencies are 10 + 100k = f110; 210; : : :g. Therefore, two other xa (t )s are
3 cos (220t ) and 3 cos (420t ) .
(e) The prefilter should be a lowpass filter with the cutoff frequency of 50 Hz.
9. Problem P3.20: An analog signal xa (t ) = cos (20t ) ; 0  t

 1 is sampled at Ts = 0 01, 0 05, and 0 1 sec intervals.

:

(a) Plots of x (n) for each Ts . M ATLAB script:

clear; close all;
%
t = 0:0.001:1; xa = cos(20*pi*t);
% (a) Plots of sequences
Ts = 0.01; N1 = round(1/Ts); n1 = 0:N1; x1 = cos(20*pi*n1*Ts);
subplot(3,1,1); plot(t,xa,n1*Ts,x1,o); axis([0,1,-1.1,1.1]);
ylabel(x1(n)); title(Sampling of xa(t) using Ts=0.01);
set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 0.05; N2 = round(1/Ts); n2 = 0:N2; x2 = cos(20*pi*n2*Ts);
subplot(3,1,2); plot(t,xa,n2*Ts,x2,o); axis([0,1,-1.1,1.1]);
ylabel(x2(n)); title(Sampling of xa(t) using Ts=0.05);
set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 00.1; N3 = round(1/Ts); n3 = 0:N3; x3 = cos(20*pi*n3*Ts);
subplot(3,1,3); plot(t,xa,n3*Ts,x3,o); axis([0,1,-1.1,1.1]);
ylabel(x3(n)); title(Sampling of xa(t) using Ts=0.1);
set(gca,xtickmode,manual,xtick,[0:1]);xlabel(t,sec);
The plots are shown in Figure 3.8.
(b) Reconstruction from x (n) using the sinc interpolation. M ATLAB script:
% (b) Reconstruction using sinc function
Ts = 0.01; Fs = 1/Ts;
xa1 = x1*sinc(Fs*(ones(length(n1),1)*t-(n1*Ts)*ones(1,length(t))));
subplot(3,1,1);plot(t,xa1); axis([0,1,-1.1,1.1]);
ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.01);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

27

Sampling of xa(t) using Ts=0.01

x1(n)

1
0

Sampling of xa(t) using Ts=0.05

x2(n)

1
0

Sampling of xa(t) using Ts=0.1

x3(n)

1
0

1
t,sec

Figure 3.8: Plots of x (n) for various Ts in Problem P3.20a.

set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 0.05; Fs = 1/Ts;
xa2 = x2*sinc(Fs*(ones(length(n2),1)*t-(n2*Ts)*ones(1,length(t))));
subplot(3,1,2);plot(t,xa2); axis([0,1,-1.1,1.1]);
ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.05);
set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 0.1; Fs = 1/Ts;
xa3 = x3*sinc(Fs*(ones(length(n3),1)*t-(n3*Ts)*ones(1,length(t))));
subplot(3,1,3);plot(t,xa3); axis([0,1,-1.1,1.1]);
ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.1);
set(gca,xtickmode,manual,xtick,[0:1]);xlabel(t,sec);
The reconstruction is shown in Figure 3.9.
(c) Reconstruction from x (n) using the cubic spline interpolation. M ATLAB script:
% (c) Reconstruction using cubic spline interpolation
Ts = 0.01; Fs = 1/Ts;
xa1 = spline(Ts*n1,x1,t);
subplot(3,1,1);plot(t,xa1); axis([0,1,-1.1,1.1]);
ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.01);
set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 0.05; Fs = 1/Ts;
xa2 = spline(Ts*n2,x2,t);
subplot(3,1,2);plot(t,xa2); axis([0,1,-1.1,1.1]);
ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.05);
set(gca,xtickmode,manual,xtick,[0:1]);
Ts = 0.1; Fs = 1/Ts;
xa3 = spline(Ts*n3,x3,t);
subplot(3,1,3);plot(t,xa3); axis([0,1,-1.1,1.1]);

28

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Reconstruction of xa(t) when Ts=0.01

xa(t)

1
0

Reconstruction of xa(t) when Ts=0.05

xa(t)

1
0

Reconstruction of xa(t) when Ts=0.1

xa(t)

1
0

1
t,sec

Figure 3.9: The sinc interpolation in Problem P3.20b.

ylabel(xa(t)); title(Reconstruction of xa(t) when Ts=0.1);
set(gca,xtickmode,manual,xtick,[0:1]);xlabel(t,sec);
The reconstruction is shown in Figure 3.10.
(d) Comments: From the plots in Figures it is clear that reconstructions from samples at Ts = 0:01 and 0:05 depict the
original freuwncy (excluding end effects) but reconstructions for Ts = 0:1 show the original frequency aliased to
zero. Furthermore, the cubic spline interpolation is a better recosntruction than the sinc interpolation, that is, the
sinc interpolation is more susceptible to boundary effect.

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

29

Reconstruction of xa(t) when Ts=0.01

xa(t)

1
0

Reconstruction of xa(t) when Ts=0.05

xa(t)

1
0

Reconstruction of xa(t) when Ts=0.1

xa(t)

1
0

1
t,sec

Figure 3.10: The cubic spline interpolation in Problem P3.20c.

Chapter 4

The z-transform
1. Problem P4.1
(a) Part (c): The given sequence is x (n) = (4=3)n u (1
X (z)

n). Hence the z-transform is

1  n
4

3


4
3z 1

3z
4

n
=

1
(3z=4)

4
3z

4
3z
3z
4
0
=

n
n

16=9
4
; jzj <
z (z 4=3)
3

M ATLAB verification: Since the sequence is a left-sided sequence, the difference equation (and hence the filter
function) should be run backward in time. This means that X (z) should be a rational function in z. Furthermore,
since x (n) begins at n = 1 and continues (backwards in time) to , we will advance x (n) by one sample and
simulate zX (z) for verification purposes where
zX (z) =

4=3
(3=4) z

from the first term on the right-hand side of X (z) above.

%(c) x(n) = (4/3)n*u(1-n)
b = [4/3]; a = [1, -3/4];
delta = [1, zeros(1,7)];
% filter solution
x = filter(b,a,delta)
x =
Columns 1 through 7
1.3333
1.0000
Column 8
0.1780

0.7500

% simulation of x(n)
n = [1:-1:-6]; x = (4/3).n
x =
Columns 1 through 7
1.3333
1.0000
0.7500
Column 8
0.1780

% Difference equation
% Input sequence

0.5625

0.4219

0.3164

0.2373

0.5625

0.4219

0.3164

0.2373

30

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

M ATLAB

USING

31

(b) Part (d): The given sequence x (n) = 2 jnj + 3 jnj can be rearranged as
x (n)

2 n u (n )

2n u ( n

1)] + 3 n u (n)

3n u ( n

1)]

i. The z-transform is
X (z)

1
2 1z

jzj

{z

2 1

2 + 5z 1
5z 1 + 6z
{z

jzj

{z

jzj

<

>

1
2z

1
3 1z

jzj

{z

3 1

>

2 (5=6)z 1
(5=6)z 1 + (1=6)z

<

{z

jzj

1
3z

{z

jzj

<

(4.1)

2 1

>

which after simplification becomes

X (z)

4:1667z 1 + 11:6667z 2 4:1667z 3 + 0:5z 4

5:8333z 1 + 10:3333z 2 5:8333z 3 + 1:0000z

0:5 < jzj < 2

ii. M ATLAB verification: Since the sequence is two-sided, it is not possible to drive the difference equation in both
directions using the filter function. However we can verify the step in (4.1) by generating positive-time
and negative-time sequences as shown below.
%(d) x(n) = (2)(-|n|)+(3)(-|n|)
R = [1;-1;1;-1];
% residues
p = [1/2;2;1/3;3];
% poles
[b,a] = residuez(R,p,[])
% Difference equation coefficients
b =
0
-4.1667
11.6667
-4.1667
a =
1.0000
-5.8333
10.3333
-5.8333
1.0000
% Forward difference equation
Rf = [1;1]; pf = [1/2;1/3];
[bf,af] = residuez(Rf,pf,[])
bf =
2.0000
-0.8333
af =
1.0000
-0.8333
0.1667
[delta,nf]= impseq(0,0,30);
xf = filter(bf,af,delta);
% Backward difference equation
Rb = [-1;-1]; pb = [2;3];
[bb,ab] = residuez(Rb,pb,[])
bb =
-2
5
ab =
1
-5
6
[delta,nb]= impseq(0,0,29);
xb = filter(fliplr(bb),fliplr(ab),delta);
% Total solution
x1 = [fliplr(xb),xf];
% simulation of x(n)
n = [-fliplr(nb+1),nf];

32

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

x2 = 2.(-abs(n)) + 3.(-abs(n));
% difference
diff = max(abs(x1-x2))
diff =
1.1102e-016
2. Problem P4.2
(a) Part (b):The given sequence x (n) can be rearranged as
 n

x (n)

=
=
=

1
u (n 2) + (0:9)n 3 u (n)
3
 2  n 2
1
1
u (n 2) + (0:9) 3 (0:9)n u (n)
3
3
 
1 1 n 2
1000
n
u (n 2) +
(0 :9 ) u (n )
9 3
729

The z-transform is

 

X (z)

=
=

1 2
1 n
1000
z Z
u (n ) +
Z [(0:9)n u (n)]
9
3
729
!


1 2
1
1000
1
z
+
9
729 1 0:9z 1
1 13 z 1

which after simplification becomes

X (z)

1000
729

1000 1
1
+ 9 z 2 0:1z 3
2187 z
37 1
1 30 z + 0:3z 2

M ATLAB verification:
%
%
%
%
%
%
%

Sequence:
x(n) = (1/3)n*u(n-2) + (0.9)(n-3)*u(n)
Analytical Expression of X(z)
X(z) = ((1000/729) - (1000/2187)*z(-1) + (1/9)*z(-2) -0.1*z(-3)
X(z) = -----------------------------------------------1 - (37/30)*z(-1) + 0.3*z(-2)

% Matlab verification
b = [1000/729, -1000/2187, 1/9, -0.1]; a = [1, -37/30, 0.3];
delta = impseq(0,0,7); format long
xb1 = filter(b,a,delta)
xb1 =
Columns 1 through 4
1.37174211248285
1.23456790123457
1.22222222222222
1.03703703703704
Columns 5 through 8
0.91234567901235
0.81411522633745
0.73037174211248
0.65655724737083
%
% check
n = 0:7;
xb2 = ((1/3).n).*stepseq(2,0,7) + ((0.9).(n-3)).*stepseq(0,0,7)
xb2 =
Columns 1 through 4
1.37174211248285
1.23456790123457
1.22222222222222
1.03703703703704

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

Columns 5 through 8
0.91234567901235
0.81411522633745

USING

M ATLAB

0.73037174211248

33

0.65655724737083

%
error = abs(max(xb1-xb2)), format short;
error =
4.440892098500626e-016
(b) Part (d):The given sequence x (n) can be rearranged as
 n

x ( n)

1
2

cos

 n

 n 1

1
2
 n
1 1
2 2
1
2

=
=

The z-transform is

cos
1

cos

45 u (n

 n

=
=

1 1
1
z Z
2
2
1
1 1
z
2
1 z

1)

(n

 n

X (z)


u (n
4

1)

cos

1 ) u (n


 

1)

n u (n )

1 1
2 z cos (=4)
1 cos (=4) + 1 z 2
4

which after simplification becomes

0:5z
X (z)

p1 z
2

p1 z 2
4 2
1 + 0:25z 2

jzj

>

0:5

M ATLAB verification:
%
%
%
%
%
%
%

Sequence:
x(n) = (1/2)n*cos(pi*n/4-pi/4)*u(n-1)
Analytical Expression of X(z)
0.5*z(-1) - 1/(4*sq(2))*z(-2)
X(z) = -------------------------------1 - 1/sq(2)*z(-1) + 0.25*z(-2)

% Matlab verification
b = [0, 0.5, -1/(4*sqrt(2))]; a = [1, -1/sqrt(2), 0.25];
delta = impseq(0,0,7); format long
xb1 = filter(b,a,delta)
xb1 =
Columns 1 through 4
0
0.50000000000000
0.17677669529664
Columns 5 through 8
-0.04419417382416 -0.03125000000000 -0.01104854345604
%
% check
n = 0:7;
xb2 = ((1/2).n).*cos(pi*n/4-pi/4).*stepseq(1,0,7)
xb2 =
Columns 1 through 4
0
0.50000000000000
0.17677669529664
Columns 5 through 8
-0.04419417382416 -0.03125000000000 -0.01104854345604

0.00000000000000
0.00000000000000

0.00000000000000
0.00000000000000

34

S OLUTIONS M ANUAL

FOR

DSP

M ATLAB

USING

A PRIL 98

%
error = abs(max(xb1-xb2)), format short;
error =
6.938893903907228e-018
3. Problem P4.3
(a) Part (b): The z-transform of x (n) is X (z) = 1 + 2z

2

z 6= 0. Consider

x2 (n) = 1 + n + n x (n) = x (n) + n [x (n)] + n [nx (n)] :

Then

X2 (z)

X (z) +

X (z)

z
z

1 + 2z

d
X (z)
dz


+

d
dz

z


d
X (z)
dz

d
d
d
X (z) + z
z X (z)
dz
dz dz
1

2z

+ 2z

= 1 + 6z



z 6= 0:

(b) Part (c): The z-transform of a sequence x (n) is X (z) = 1 + 2z 1 , z 6= 0. A new sequence is x3 (n) =
Using the time-shift property
Z [x (n 2)] = z 2 X (z) = z 2 + 2z 3
with no change in ROC. Now using the frequency-shift property


 n

1
2

x (n

2)

X3 (z) = z

(2z)

+ 2z

+ 2 (2z)

z!z=0:5

= 0:25z

+ 0:25z

with ROC scaled by (1=2). Since the old ROC is jzj > 0; the new ROC is also jzj > 0.
4. Problem P4.5 (b): The inverse z-transform of X (z) is x (n) = 2 n u (n). Then

zX z



=
=

x(
2

(n + 1)) = x (
(

n 1)

u( n

1)

1) = 2

n+1

u( n

1)

5. Problem P4.9
(a) Part (b): The z-transform X (z) of a sequence is given as follows:
X (z) =

1
1

4z

2 + 4z 3

11 1
13
+ 8z 2
4 z

0:25z

absolutely summable sequence

The partial fractions are computed using the residuez function.

b = [1,-1,-4,4]; a = [1,-11/4,13/8,-1/4];
[R,p,k] = residuez(b,a)
R =
0.0000
-10.0000
27.0000
p =
2.0000
0.5000
0.2500
k =
-16

1 n
x (n
2

2).

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

35

Therefore,
X (z)

16 +

0
2z

10
0:5z

27
0:25z

0:5 < jzj < 2

Hence from the Z -transform table:

10 (0:5)n u (n) + 27 (0:25)n u (n)

x (n) = 16 (n)
M ATLAB verification:

[delta,n] = impseq(0,0,7);
xb1 = filter(b,a,delta)
xb1 =
Columns 1 through 4
1.00000000000000
1.75000000000000 -0.81250000000000
Columns 5 through 8
-0.51953125000000 -0.28613281250000 -0.14965820312500
xb2 = -16*delta - 10*(0.5).n + 27*(0.25).n
xb2 =
Columns 1 through 4
1.00000000000000
1.75000000000000 -0.81250000000000
Columns 5 through 8
-0.51953125000000 -0.28613281250000 -0.14965820312500
error = abs(max(xb1-xb2))
error =
0

-0.82812500000000
-0.07647705078125

-0.82812500000000
-0.07647705078125

(b) Part (d): The z-transform X (z) of a sequence is given as follows:

X (z) =

z
;
z3 + 2z2 + 1:25z + 0:25

jzj

>

The partial fractions are computed using the residuez function.

b = [0,0,1]; a = [1,2,1.25,0.25];
[R,p,k] = residuez(b,a), echo on;
R =
4.0000
0.0000 - 0.0000i
-4.0000 + 0.0000i
p =
-1.0000
-0.5000 + 0.0000i
-0.5000 - 0.0000i
k =
[]
Therefore,
X (z)

4
1+z

0
1 + 0:5z

4
2
(1 + 0:5z 1)

Hence from the Z -transform table:

x (n)

=
=

M ATLAB verification:

4 ( 1)n u (n) + ( 4) 01:5 (n + 1) ( 0:5)n+1 u (n + 1)

4 ( 1)n u (n) + 8 (n + 1) ( 0:5)n+1 u (n)

36

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

[delta,n] = impseq(0,0,7);
xd1 = filter(b,a,delta)
xd1 =
0
0
1.0000
-2.0000
2.7500
xd2 = 4*(-1).n + (8*(n+1)).*((-0.5).(n+1))
xd2 =
0
0
1.0000
-2.0000
2.7500
error = abs(max(xd1-xd2))
error =
0

-3.2500

3.5625

-3.7500

-3.2500

3.5625

-3.7500

6. Problem P4.10: The z-transform X (z) is given as follows:

X (z) =

2 + 3z 1
z 1 + 0:81z

jzj

>

0:9

(a) Sequence x(n) in a form that contains no complex numbers: Compare the denominator 1 z 1 + 0:81z
denominator 1 2az 1 cos 0 + a2 z 2 of the sin(cos) transform pairs (see page 174 of the text).
1

+ 0:81z

or
a = 0:9; cos 0 =

1
1 :8

cos0 + a2z

a2 = 0:81;

2az

) 0 = 0 3125
:

cos 0 =

1
2a

and sin 0 = 0:8315

Now X (z) can be put in the form

X (z)

=
=

2
1
1

1 + 4z 1

2 (0:9) z 1 11:8 + (0:
9)2 z

2 1 0:9z 1 11:8
2 (0:9) z

1 1 + (0:9)2 z 2
1 :8

jzj

>

0 :9

+ 5:3452

0:9 (0:8315) z
1

2 (0:9) z

1 1 + (0:9)2 z 2
1:8

Finally, after table lookup

x (n) = [2 (0:9)n cos (0:3125) + (5:3452) (0:9)n sin (0:3125)] u (n)
For another approach using PFE and residues, see the M ATLAB script below.
clear all
% (a) Use of the residue and transform table
b=[2, 3]; a=[1,-1,0.81];
[R,p,k] = residuez(b,a);
R_real = (real(R))
R_real =
1
1
R_imag = (imag(R))
R_imag =
-2.6726
2.6726
p_magn = (abs(p))
p_magn =
0.9000
0.9000
p_angl = (angle(p))/pi

with the

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

37

p_angl =
0.3125
-0.3125
[delta,n] = impseq(0,0,19);
xa = (p_magn(1).n).*(2*R_real(1)*cos(p_angl(1)*pi*n)-2*R_imag(1)*sin(p_angl(1)*pi*
%
% Print Response
fprintf(1,\n Hence the sequence x(n) is \n)

Hence the sequence x(n) is

fprintf(1,\n\tx(n) = (%1.1f)n * (%1.0f*cos(%1.4f*pi*n) - (%2.4f)*sin(%1.4f*pi*n)\
p_magn(1),2*R_real(1),p_angl(1),2*R_imag(1),p_angl(1));
x(n) = (0.9)n * (2*cos(0.3125*pi*n) - (-5.3452)*sin(0.3125*pi*n)
(b) M ATLAB verification:
xb = filter(b,a,delta);
error = abs(max(xa-xb))
error =
1.5543e-015
7. Problem P4.11
(a) Part (a): The impulse response is h (n) = 2 (0:5)n u (n). The system function representation is
H (z) = Z [h (n)] =

2
0:5z

jzj

>

0 :5 :

The difference equation representation is

Y (z)
X (z)

2
0:5z

) Y (z)

0:5z 1Y (z) = 2X (z)

or
y (n) = 2x (n) + 0:5y (n

1) .

The pole-zero description is given by a zero at z = 0 and a pole at z = 0:5. Finally, to compute output y (n) when
x (n) = (1=4)n u (n) we use the z-transform approach (since the ROCs overlap):
Y (z)

H (z) X (z) =

4
0:5z

2
1

1
1 0:5z
1 0:25z
2
; jzj > 0:5
1 0:25z 1

Hence
y (n) = 4 (0:5)n u (n)

2 (0:25)n u (n) :

(b) Part (d): Impulse response of an LTI system:



h (n) = n [u (n)

u (n

10)] = 0; 1; 2; : : : ; 9

"

i) System function representation:

H (z) = z

+ 2z

+ : : : + 9z

kz

k =1

38

S OLUTIONS M ANUAL

FOR

DSP

M ATLAB

USING

A PRIL 98

ii) Difference equation representation:

9

y (n ) =

kx (n

k) :

k =1

iii) Pole-zero plot M ATLAB script:

clear, close all;
hb = [0:9]; ha = [1,0]; zplane(hb,ha);
The pole-zero plot is shown in Figure 4.1.

Imaginary part

0.5

0.5

1.5

0.5

0
Real part

0.5

1.5

Figure 4.1: Pole-zero plot in Problem P4.11d.

iv) The output y (n) when the input is x (n) = (1=4)n u (n):
Y (z)

=
=
=

H (z) X (z) = z
z

+ 2z

+ : : : + 9z

1
0:25z

jzj

>

0:25

1 + 2z 2 + : : : + 9z 9
1 0:25z 1

3029220
3029220 757304z 1 189324z
0:25z 1
2952z 5 732z 6 176z 7 36z 8

where the PFE was performed using M ATLAB.

clear, close all;
hb = [0:9]; ha = [1,0]; xb = [1]; xa = [1,-0.25];
yb = hb; ya = xa;
[R,p,k] = residuez(yb,ya)
R =
3029220
p =
0.2500
k =

47328z

11828z

A PRIL 98

S OLUTIONS M ANUAL

Columns 1 through 6
-3029220
-757304
Columns 7 through 9
-732
-176
Hence
y (n)

FOR

DSP

USING

-189324

M ATLAB

-47328

-11828

-2952

-36

3029220 (0:25)n u (n) 3029220 (n)

176 (n 7) 36 (n 8) :

39

757304 (n

1)

8. Problem P4.15: A stable system has the following pole-zero locations:

z1 = j;

z2 =

1
1
+j ;
2
2

p1 =

j;

1
2

p2 =

1
2

It ia also known that the frequency response function H e j evaluated at = 0 is equal to 0:8, i.e.,
H e j0

= 0 :8

a) System function H (z) and region of convergence.

H (z) = K

z2 + 1
j ) (z + j )
1
; jzj > p
=K
2
j0:5)(z + 0:5 + j0:5)
z + z + 0:5
2

(z
(z + 0:5

Now @z = e j0 = 1, H (1) = 0:8 is given, hence

0:8 = K

1+1
2
=K
1 + 1 + 0:5
2 :5

)K=1

or
H (z) =

z2 + 1
1
; jzj > p
z2 + z + 0:5
2

b) Difference equation representation.:

H (z) =

z2 + 1
z2 + z + 0:5

1+z 2
1 + z 1 + 0:5z

Y (z)
X (z)

After cross multiplying and inverse transforming

y (n) + y (n

1) + 0:5y (n

2) = x (n) + x (n

1)

c) Steady-state response yss (n) if the input is

x ( n) =
From the z-transform Table

p1

X (z) =
Hence

p1
2

sin


 n 

2
z

z2 + 1

u (n) :

jzj


Y (z) = H (z) X (z)

=
=

>

1


z2 + 1
p1 z2 +z 1
2 + z + 0:5
z
2


1
z
p z2 + z + 0:5 ; jzj > p12
2

Thus the poles of Y (z) are the poles of H (z) which are inside the unit circle. Therefore, there is NO steady-state
response or yss (n) = 0.

40

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

d) Transient response ytr (n): Since yss (n) = 0, the total response y (n) = ytr (n). From the Y (z) expression from part c)
above


p1 p1 z
p
1
z
1
2 2
Y (z) = p
= 2
; jzj > p
 2
2 + z + 0:5
z
2
2
1 + z 1 + p12 z 2
Hence from table lookup we have
y (n) = ytr (n) =

p1

n

sin (0:75n) u (n)

9. Problem P4.16: Digital filter is described by the difference equation:

y (n ) = x ( n ) + x ( n

1) + 0:9y (n

1)

0:81y (n

2)

a) Magnitude and phase of the frequency response: M ATLAB script

clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P4.16a);
% (a) Magnitude and Phase Plots
b = [1, 1]; a = [1,-0.9,0.81];
w = [0:1:500]*pi/500; H = freqz(b,a,w);
magH = abs(H); phaH = angle(H)*180/pi;
subplot(2,1,1); plot(w/pi,magH); axis([0,1,0,12]); grid
title(Magnitude Response); xlabel(frequency in pi units); ylabel(|H|);
subplot(2,1,2); plot(w/pi,phaH); axis([0,1,-180,180]); grid
title(Phase Response); xlabel(frequency in pi units); ylabel(Degrees);
%
w = [pi/3,pi]; H = freqz(b,a,w); magH = abs(H); phaH = angle(H)*180/pi;
fprintf(1,\n At w = pi/3 the magnitude is %1.4f and the phase is %3.4f degrees \n,...
magH(1), phaH(1));
At w = pi/3 the magnitude is 10.5215 and the phase is -58.2595 degrees
fprintf(1,\n At w = pi the magnitude is %1.4f and the phase is %3.4f degrees \n,...
magH(2), phaH(2));
At w =

pi

the magnitude is 0.0000 and the phase is -90.0000 degrees

The frequency response plots are shown in Figure 4.2.

b) Steady-state response: M ATLAB script
n = [0:200]; x = sin(pi*n/3) + 5*cos(pi*n);
y = filter(b,a,x);
n = n(101:201); x = x(101:201); y = y(101:201); % Steady-state section
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P4.16b);
subplot(2,1,1); plot(n,x); title(Input sequence); axis([100,200,-6,6])
xlabel(n); ylabel(x(n));
subplot(2,1,2); plot(n,y); title(Output sequence); axis([100,200,-12,12])
xlabel(n); ylabel(y(n));
The steady-state response is shown in Figure 4.3. It shows that the = frquency is suppressed while the only
component in the output is due to the = =3 frequency which is amplified by about 10.

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

41

Problem 4.16a: Frequency Response Plots

Magnitude Response
10

|H|

8
6
4
2
0
0

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units
Phase Response

0.7

0.8

0.9

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units

0.7

0.8

0.9

Degrees

100
0
100
0

Figure 4.2: Frequency Response Plots in Problem P4.16a

10. Problem P4.17: Difference equation solution using one-sided z-transform approach:
y(n) = 0:5y(n 1) + 0:25y (n
x(n) = (0:8)n u (n)
y ( 1) = 1; y ( 2) = 2

2) + x(n); n  0

Taking one-sided z-transform:



Y + (z) = 0:5 z 1Y + (z) + y ( 1)

or
Y + (z) 1

0:5z

0:25z

+ 0:25

z 2Y + (z) + y ( 2) + z 1 y ( 1)

+X

1
1 0:8z 1
1
1 0:55z 1 0:2z
=
0:8z 1
1 0:8z 1

{z

Equi. I.C. Input

Y + (z)

=
=
=

2 0:55z 1 0:2z 2
(1 0:8z 1) (1 0:5z 1 0:25z 2 )
1 0:55z 1 0:2z 2
1 1:3z 1 + 0:15z 2 + 0:2z 3
65:8702
64
0:1298
+
+
1 0:8090z 1 1 0:8z 1 1 + 0:3090z

Hence
y (n) = 65:8702 (0:8090)n

64 (0:8)n + 0:1298 ( 0:3090)n ; n  0

M ATLAB verification:
b = [1]; a = [1, -0.5, -0.25];
yic = [1,2];

(z)

= 0:5 + 0:5 + 0:25z 1 +

=1 + 0:25z

or

% Difference equation
% Initial conditions

2+1

42

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Problem 4.16b: SteadyState Response

Input sequence
4
x(n)

2
0
2
4
6
100

110

120

130

140

150
160
n
Output sequence

110

120

130

140

170

180

190

200

170

180

190

200

10

y(n)

5
0
5
10
100

150
n

160

Figure 4.3: Steady-state Response in problem P4.16b

n = [0:20]; x = (0.8).n;
% Numerical Solution
V = filtic(b,a,yic), echo on;
V =
1.0000
0.2500
y1 = filter(b,a,x,V);

% Input sequence

% equivalent initial condition input

% Output sequence

% Analytical solution
b1 = conv(V,[1,-0.8])+[1,0,0]
% Num of Y(z)
b1 =
2.0000
-0.5500
-0.2000
a1 = conv([1,-0.8],[1,-0.5,-0.25])
% Denom of Y(z)
a1 =
1.0000
-1.3000
0.1500
0.2000
[R,p,k] = residuez(b1,a1), echo on; % PFE
R =
65.8702
-64.0000
0.1298
p =
0.8090
0.8000
-0.3090
k =
[]
y2 = zeros(1,21); L = length(R);
%
for l = 1:L
% Assemble

A PRIL 98

S OLUTIONS M ANUAL

y2 = y2 + R(l)*(p(l).n);
end
%
error = abs(max(y1-y2))
error =
1.1169e-013

FOR

DSP

USING

%
%

M ATLAB

43

Output
Sequence

% Difference

11. Problem P4.19: A causal, linear, and time-invariant system is given by

y (n) = y (n

1) + y (n

2) + x (n

1)

a) System Function: Taking z-transform of both sides, we obtain

Y (z) = z 1Y (z) + z 2Y (z) + z 1 X (z) ) H (z)

Y (z)
X (z)

def
=

jzj

>

r0

where r0 is the magnitude of the largest pole since the system is causal.
b) Pole-zero pot and the ROC:
H (z) =

z
z

z2

1+ 5
2

1 5
1+ 5
. Hence the ROC is jzj >
2
2

The zero is @ z = 0 and poles are @ z =

= 1:618.

c) Impulse response: Using PFE

H (z) =
z

1+ 5
2

!=

p1

u (n )

p1
5

Hence
h (n ) =

p1
5

1+ 5
2

1+ 5
2

!n

p1

jj

!; z

>

1+ 5
2

!n

u (n )

d) Clearly the system is not stable since h (n) % as n % . For a stable unit sample response the ROC should be

<

Then

p1
5

h (n) =

1+ 5
2

jzj

<

1+ 5
2

!n

u( n

1)

p1
5

!n

u (n )

12. Problem P4.20: The difference eqution is

1
y (n) = y (n
4

1) + x (n) + 3x (n

1) ; n  0;

y ( 1) = 2

with the input x (n) = e jn=4 u (n). Taking one-sided z-transform of the difference equation, we obtain,
Y + (z) =


1 1 +
z Y (z) + y ( 1) +
4
1

1
e j=4 z

+3

z 1
e j=4 z

Substituting the initial condition and rearranging, we obtain



(z)

1
z
4


1

1
1 + 3z 1
+
2 1 e j=4 z

(4.2)

44

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

The second term on the right-hand side provides the zero-state response of the system. Thus
Yzs+ (z)

1 + 3z


1 14 z 1 1
4:4822e j0:8084
1 1
4z

e j=4 z
+

3:8599e j2:1447
1 e j=4 z 1

Hence the zero-state response is



 n

yzs (n) = 4:4822e

1
4

j0:8084


+ 3:8599e

j2:1447 jn=4

u (n ) :

The steady-state part of the total response is due to simple poles on the unit circle. The pole on the unit circle is at
z = e j=4 which is due to the input sequence. From (4.2), the total response is
Y

(z)

1
1

1 1
4z

!

1
1 + 3z 1
+
2 1 e j=4 z


1

3=2 + (2:6464 j0:3536) z

1 14 z 1 1 e j=4 z 1

4:4822e j0:8084 3:6129e j2:0282

+
:
1 e j=4 z 1
1 14 z 1

The steady-state response is given by the second-term on the right-hand side. Thus
yss (n) = 3:6129e j2:0282 e jn=4 u (n) = 3:6129e j(n=4

2:0282)

u (n) .

Chapter 5

The Discrete Fourier Transform

1. Problem P 5.1

(a) Periodic sequence: x1 (n) =

:::;

2 0; 2; 0; 2; 0; 2; 0; 2; 0; : : : . Now,

"

N 1

X1 (k) =

x1 (n) WNnk ; N = 4; W4 = e

j2=4

4
0
4
0

n =0

Hence,

X1 (0)
X1 (1)
X1 (2)
X1 (3)

=
=
=
=

2 (1) + 0 (1) + 2 (1) + 0 (1)

2 (1 ) + 0 ( j ) + 2 ( 1 ) + 0 ( j )
2 (1) + 0 ( 1) + 2 (1) + 0 ( 1)
2 (1 ) + 0 ( j ) + 2 ( 1 ) + 0 ( j )

=
=
=

M ATLAB verification:
xtilde1 = [2,0,2,0]; N = length(xtilde1);
[Xtilde1] = dft(xtilde1,N)
Xtilde1 =
4.0000
0 - 0.0000i
4.0000 + 0.0000i


(b) Periodic sequence: x2 (n) =

0 - 0.0000i

:::;

0; 0; 1; 0; 0; 0; 0; 1; 0; 0; : : : . Now,

"

N 1

X2 (k) =

x2 (n) WNnk ; N = 5; W5 = e

j2=5

= 0:3090

j0:9511

n=0

Hence,

X2 (0)
X2 (1)
X2 (2)
X2 (3)
X2 (4)

=
=
=
=
=

0 + 0 + 1 (1) + 0 + 0
0 + 0 + 1 (0:3090 j0:9511)2 + 0 + 0
0 + 0 + 1 (0:3090 j0:9511)4 + 0 + 0
0 + 0 + 1 (0:3090 j0:9511)6 + 0 + 0
0 + 0 + 1 (0:3090 j0:9511)8 + 0 + 0

0:8090 j0:5878
0:3090 + j0:9511
0:3090 j0:9511
0:8090 + j0:5878

=
=
=

M ATLAB verification:
xtilde2 = [0,0,1,0,0]; N = length(xtilde2);
[Xtilde2] = dft(xtilde2,N)
Xtilde2 =
Columns 1 through 4
1.0000
-0.8090 - 0.5878i
0.3090 + 0.9511i
Column 5
-0.8090 + 0.5878i
45

0.3090 - 0.9511i

46

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

(c) Periodic sequence: x3 (n) =

:::;

3; 3; 3; 3; 3; 3; 3; 3; : : : . Now,

"

N 1

X3 (k) =

x3 (n) WNnk ; N = 4; W4 = e

j2=4

n=0

Hence,

X3 (0)
X3 (1)
X3 (2)
X3 (3)

3 (1 )
3 (1 )
3 (1 )
3 (1 )

=
=
=

3 (1) + 3 (1) 3 (1)

3 ( j ) + 3 ( 1) 3 ( j )
3 ( 1) + 3 (1) 3 ( 1)
3 ( j ) + 3 ( 1) 3 ( j )

0
0
12
0

=
=
=

M ATLAB verification:
xtilde = [3,-3,3,-3]; N = length(xtilde);
[Xtilde] = dft(xtilde,N)
Xtilde =
0
0.0000 - 0.0000i 12.0000 + 0.0000i
(

(d) Periodic sequence: x4 (n) =

j; j;

:::;

"

j;

j; j; j;

j;

j; : : : . Now,

N 1

X4 (k) =

0.0000 - 0.0000i

x4 (n) WNnk ; N = 4; W4 = e

j2=4

j (1) + j (1) j (1) j (1)

j ( 1) + j ( j ) j ( 1) j ( j )
j (1) + j ( 1) j (1) j ( 1)
j ( 1) + j ( j ) j ( 1) j ( j )

n=0

Hence,

X4 (0)
X4 (1)
X4 (2)
X4 (3)

=
=
=
=

0
2 + j2
0
2 + j2

=
=
=

M ATLAB verification:
xtilde = [j,j,-j,-j]; N = length(xtilde4);
[Xtilde4] = dft(xtilde4,N)
Xtilde4 =
0
2.0000 + 2.0000i


(e) Periodic sequence: x5 (n) =

:::;

-2.0000 + 2.0000i

1; j; j; 1; 1; j; j; 1; j; j; : : : . Now,

"

N 1

X5 (k) =

x5 (n) WNnk ; N = 4; W4 = e

j2=4

n=0

Hence,

X5 (0)
X5 (1)
X5 (2)
X5 (3)

=
=
=
=

1 (1) + j (1) + j (1) + 1 (1)

1 (1 ) + j ( j ) + j ( 1 ) + 1 ( j )
1 (1 ) + j ( 1 ) + j (1 ) + 1 ( 1 )
1 (1 ) + j ( j ) + j ( 1 ) + 1 ( j )

=
=
=
=

2+2j
2
0
2j

M ATLAB verification:
xtilde5 = [1,j,j,1]; N5 = length(xtilde5);
[Xtilde5] = dft(xtilde5,N5)
Xtilde5 =
2.0000 + 2.0000i
2.0000 + 0.0000i
0.0000 - 0.0000i
2. Problem P 5.2

0.0000 - 2.0000i

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

47

(a) Periodic DFS sequence: X1 (k) = f5; 2 j; 3; 2 jg, N = 4. Now,

x1 (n) =

1N 1
X1 (k) WN nk ; N = 4; W4
N k
=0

=e

j2=4

Hence,
x1 (0)
x1 (1)
x1 (2)
x1 (3)

[5 (1)

[5 (1)

[5 (1)

[5 (1)

2 j (1) + 3 (1) + 2 j (1)] =4

2 j ( j) + 3 ( 1) + 2 j ( j)] =4
2 j ( 1) + 3 (1) + 2 j ( 1)] =4
2 j ( j) + 3 ( 1) + 2 j ( j)] =4

=
=
=
=

2
1:5
2
0:5

M ATLAB verification:
Xtilde1 = [5,-2*j,3,2*j]; N1 = length(Xtilde1);
[xtilde1] = real(idfs(Xtilde1,N1))
xtilde1 =
2.0000
1.5000
2.0000
-0.5000
(b) Periodic DFS sequence: X2 (k) = f4; 5; 3; 5g. Now,
x2 (n) =

1N 1
X2 (k) WNnk ; N = 4; W4 = e
N k
=0

j2=4

Hence,
x2 (0)
x2 (1)
x2 (2)
x2 (3)

[4 (1)

[4 (1)

[4 (1)

[4 (1)

5 (1) + 3 (1) 5 (1)] =4

5 ( j) + 3 ( 1) 5 ( j)] =4
5 ( 1) + 3 (1) 5 ( 1)] =4
5 ( j) + 3 ( 1) 5 ( j)] =4

0:75
0:25
4:25
0:25

=
=
=

M ATLAB verification:
Xtilde2 = [4,-5,3,-5]; N = length(Xtilde2);
[xtilde2] = real(idfs(Xtilde2,N))
xtilde2 =
-0.7500
0.2500
4.2500
0.2500
(c) Periodic DFS sequence: X3 (k) = f1; 2; 3; 4; 5g. Now,
x3 (n) =
Hence,
x3 (0)
x3 (1)
x3 (2)
x3 (3)
x3 (4)

=
=
=
=
=

1N 1
X3 (k) WNnk ; N = 5; W5 = e
N k
=0

j2=5



0
0
0
0
0
1W50 + 2W51 + 3W52 + 4W53 + 5W54  =5
1W50 + 2W52 + 3W54 + 4W56 + 5W58  =5
1W50 + 2W53 + 3W56 + 4W59 + 5W512 =5
1W50 + 2W54 + 3W58 + 4W512+ 5W5 16 =5

1W5

+ 2W5 + 3W5 + 4W5 + 5W5

= 0:3090

=
=
=
=
=

j0:9511

3
0:5000 j0:6882
0:5000 j0:1625
0:5000 + j0:1625
0:5000 + j0:6882

M ATLAB verification:
Xtilde3 = [1,2,3,4,5]; N = length(Xtilde3);
[xtilde3] = idfs(Xtilde3,N)
xtilde3 =
Columns 1 through 4
3.0000
-0.5000 - 0.6882i -0.5000 - 0.1625i
Column 5
-0.5000 + 0.6882i

-0.5000 + 0.1625i

48

S OLUTIONS M ANUAL

FOR

(e) Periodic DFS sequence: X5 (k) = f0; j; 2 j;

x5 (n) =

DSP

USING

M ATLAB

A PRIL 98

jg, N = 4. Now,

1N 1
X5 (k) WN nk ; N = 4; W4
N k
=0

=e

j2=4

Hence,
x5 (0)
x5 (1)
x5 (2)
x5 (3)

[0 (1) + j (1)

[0 (1) + j ( j)

2 j (1) j (1)] =4
2 j ( 1) j ( j)] =4
1) 2 j (1) 2 j ( 1)] =4
j) 2 j ( 1) j ( j)] =4

[0 (1) + j (

[0 (1) + j (

j0:5
0:5 + j0:5
j0:25
0:5 + j0:5

M ATLAB verification:
Xtilde5 = [0,j,-2*j,-j]; N5 = length(Xtilde5);
[xtilde5] = real(idfs(Xtilde5,N1))
xtilde1 =
0 - 0.5000i -0.5000 + 0.5000i
0.0000 - 0.2500i

0.5000 + 0.5000i

(d) Periodic DFS sequence: X (k) = f0; 0; 2; 0g. Now,

x (n) =

1N 1
X (k) WNnk ; N = 4; W4 = e
N k
=0

j2=4

Hence,
x (0)
x (1)
x (2)
x (3)

[0 (1) + 0 (1) + 2 (1) + 0 (1)] =4

[0 (1) + 0 (

=
=

j) + 2 ( 1) + 0 ( j)] =4
[0 (1) + 0 ( 1) + 2 (1) + 0 ( 1)] =4
[0 (1) + 0 ( j) + 2 ( 1) + 0 ( j)] =4

=
=

0:5
0:5
0:5
0:5

M ATLAB verification:
Xtilde4 = [00,0,2,0]; N = length(Xtilde4);
[xtilde4] = real(idfs(Xtilde4,N))
xtilde4 =
0.5000
-0.5000
0.5000
-0.5000
3. Problem P 5.3
Periodic x1 (n) with fundamental period N = 50


x1 (n) =

ne

0:3n

0;

0  n  25 
26  n  49 PERIODIC

Periodic x2 (n) with fundamental period N = 100



x2 (n) =

ne

0:3n ;

0;

0  n  25
26  n  99


PERIODIC

(a) Computation of X1 (k) using M ATLAB:

clear; close all;
% (a) DFS Xtilde1(k)
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.3a);
n1 = [0:49]; xtilde1 = [n1(1:26).*exp(-0.3*n1(1:26)),zeros(1,24)]; N1 = length(n1);
[Xtilde1] = dft(xtilde1,N1); k1 = n1;
mag_Xtilde1 = abs(Xtilde1); pha_Xtilde1 = angle(Xtilde1)*180/pi;
zei = find(mag_Xtilde1 < 1000*eps);
pha_Xtilde1(zei) = zeros(1,length(zei));

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

49

subplot(3,1,1); stem(n1,xtilde1); axis([-1,N1,min(xtilde1),max(xtilde1)]);

title(One period of the periodic sequence xtilde1(n)); ylabel(xtilde1);
ntick = [n1(1):2:n1(N1),N1];
set(gca,XTickMode,manual,XTick,ntick,FontSize,10)
subplot(3,1,2); stem(k1,mag_Xtilde1); axis([-1,N1,min(mag_Xtilde1),max(mag_Xtilde1)
title(Magnitude of Xtilde1(k)); ylabel(|Xtilde1|)
ktick = [k1(1):2:k1(N1),N1];
set(gca,XTickMode,manual,XTick,ktick,FontSize,10)
subplot(3,1,3); stem(k1,pha_Xtilde1); axis([-1,N1,-180,180]);
title(Phase of Xtilde1(k)); xlabel(k); ylabel(Angle in Deg)
ktick = [k1(1):2:k1(N1),N1];
set(gca,XTickMode,manual,XTick,ktick,FontSize,10)
set(gca,YTickMode,manual,YTick,[-180;-90;0;90;180])
Plots of x1 (n) and X1 (k) are shown in Figure 5.1.
Problem 3a Plots
One period of the periodic sequence xtilde1(n)

xtilde1

0.5

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Magnitude of Xtilde1(k)
|Xtilde1|

10
8
6
4
2
0

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Phase of Xtilde1(k)
Angle in Deg

180
90
0
90
180

8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Figure 5.1: Plots of x1 (n) and X1 (k) in Problem 5.3a

(b) Computation of X2 (k) using M ATLAB:
% (b) DFS Xtilde2(k)
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P5.3b);
n2 = [0:99]; xtilde2 = [xtilde1, zeros(1,50)]; N2 = length(n2);
[Xtilde2] = dft(xtilde2,N2); k2 = n2;
mag_Xtilde2 = abs(Xtilde2); pha_Xtilde2 = angle(Xtilde2)*180/pi;
zei = find(mag_Xtilde2 < 1000*eps);
pha_Xtilde2(zei) = zeros(1,length(zei));
subplot(3,1,1); stem(n2,xtilde2); axis([-1,N2,min(xtilde2),max(xtilde2)]);
title(One period of the periodic sequence xtilde2(n)); ylabel(xtilde2);
ntick = [n2(1):5:n2(N2),N2];
set(gca,XTickMode,manual,XTick,ntick)

50

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

subplot(3,1,2); stem(k2,mag_Xtilde2); axis([-1,N2,min(mag_Xtilde2),max(mag_Xtilde2)]);

title(Magnitude of Xtilde2(k)); ylabel(|Xtilde2|)
ktick = [k2(1):5:k2(N2),N2];
set(gca,XTickMode,manual,XTick,ktick)
subplot(3,1,3); stem(k2,pha_Xtilde2); axis([-1,N2,-180,180]);
title(Phase of Xtilde2(k)); xlabel(k); ylabel(Degrees)
ktick = [k2(1):5:k2(N2),N2];
set(gca,XTickMode,manual,XTick,ktick)
set(gca,YTickMode,manual,YTick,[-180;-90;0;90;180])
Plots of x2 (n) and X2 (k) are shown in Figure 5.2.
Problem 3b Plots
One period of the periodic sequence xtilde2(n)

xtilde2

1
0.5

|Xtilde2|

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Magnitude of Xtilde2(k)

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Phase of Xtilde2(k)

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
k

10
8
6
4
2

Degrees

180
90
0
90
180

Figure 5.2: Plots of Magnitude and Phase of X2 (k) in Problem 5.3b

(c) Changing the period from N = 50 to N = 100 resulted in a lower frequency sampling interval (higher frequency
resolution) 1 , i.e., in (3a) 1 = =25 and in (3b) 2 = =50. Hence there are more terms in the DFS expansion of
x2 (n). The shape of the DTFT begins to fill in with N = 100.
4. Problem P 5.4
New periodic sequence x3 (n) with period N = 100
x3 (n) = [x1 (n) ; x1 (n)]PERIODIC
(a) Computation of X3 (k) using M ATLAB:
clear; close all;
% (a) DFS Xtilde3(k)
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.4a);
n1 = [0:49]; xtilde1 = [n1(1:26).*exp(-0.3*n1(1:26)),zeros(1,24)]; N1 = length(n1);
n3 = [0:99]; xtilde3 = [xtilde1, xtilde1]; N3 = length(n3);
[Xtilde3] = dft(xtilde3,N3); k3 = n3;
mag_Xtilde3 = abs(Xtilde3); pha_Xtilde3 = angle(Xtilde3)*180/pi;

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

51

zei = find(mag_Xtilde3 < 0.00001);

pha_Xtilde3(zei) = zeros(1,length(zei));
subplot(3,1,1); stem(n3,xtilde3); axis([-1,N3,min(xtilde3),max(xtilde3)]);
title(One period of the periodic sequence xtilde3(n)); ylabel(xtilde3);
ntick = [n3(1):5:n3(N3),N3];
set(gca,XTickMode,manual,XTick,ntick)
subplot(3,1,2); stem(k3,mag_Xtilde3); axis([-1,N3,min(mag_Xtilde3),max(mag_Xtilde3)
title(Magnitude of Xtilde3(k)); ylabel(|Xtilde3|)
ktick = [k3(1):5:k3(N3),N3];
set(gca,XTickMode,manual,XTick,ktick)
subplot(3,1,3); stem(k3,pha_Xtilde3); axis([-1,N3,-180,180]);
title(Phase of Xtilde3(k))
xlabel(k); ylabel(Degrees)
ktick = [k3(1):5:k3(N3),N3];
set(gca,XTickMode,manual,XTick,ktick)
set(gca,YTickMode,manual,YTick,[-180;-90;0;90;180])
Plots of x3 (n) and X3 (k) are shown in Figure 5.3.
Problem 4 Plots
One period of the periodic sequence xtilde3(n)

xtilde3

1
0.5
0

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Magnitude of Xtilde3(k)

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Phase of Xtilde3(k)

10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
k

|Xtilde3|

20
15
10
5

Degrees

180
90
0
90
180

Figure 5.3: Plots of x3 (n) and X3 (k) in Problem 5.4a

(b) Comparing the magnitude plot above with that of X1 (k) in Problem (3a), we observe that these plots are essentially
similar. Plots of X3 (k) have one zero between every sample of X1 (k). (In general, for phase plots, we do get
non-zero phase values when the magnitudes are zero. Clearly these phase values have no meaning and should be
ignored. This happens because of a particular algorithm used by M ATLAB. I avoided this problem by using the
find function.) This makes sense because sequences x1 (n) and x3 (n), when viewed over < n < interval,
look exactly same. The effect of periodicity doubling is in the doubling of magnitude of each sample.
(c) We can now generalize this argument. If
8
>
<

xM (n) =

9
>
=

x1 (n) ; x1 (n) ; : : : ; x1 (n)

>
:|

{z

M times

}>
;
PERIODIC

52

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

then there will be (M 1) zeros between samples of XM (k). The magnitudes of non-zero samples of XM (k) will be
M times the magnitudes of the samples of X1 (k), i.e.,
XM (Mk)
XM (k)

=
=

M X1 (k)
0

;
;

k = 0; 1; : : : ; N 1
k 6= 0; 1; : : : ; MN

5. Problem P 5.5

X e j is a DTFT of a 10-point sequence x (n) = f2; 5; 3; 4; 2; 6; 0; 3; 3; 2g.

(a) Since y1 (n) is a 3-point IDFS of three samples of X e j on the unit circle, it can be obtained as a 3-point aliasing
operation on x (n). Thus
y1 (n) = f2 + ( 4) + 0 + 2; 5 + ( 2) + ( 3) ; 3 + 6 + ( 3)g = f0; 0; 6gperiodic
M ATLAB verification:
clear; close all;
x = [2,5,3,-4,-2,6,0,-3,-3,2]; n = 0:9;
% (a) y1(n) = 3-point IDFS{X(0),X(2*pi/3),X(4*pi/3)}
N = 3; k = 0:N-1; w = 2*pi*k/N;
[Y1] = dtft(x,n,w); y1 = real(idfs(Y1,N))
y1 =
0.0000
0.0000
6.0000

(b) Since y2 (n) is a 20-point IDFS of twenty samples of X e j on the unit circle, then from the frequency sampling
theorem there will not be any aliasing of x (n) and y2 (n) will be a zero-padded version of x (n). Thus
y2 (n) = f2; 5; 3; 4; 2; 6; 0; 3; 3; 2; 0; 0; 0; 0; 0; 0; 0; 0; 0; 0gperiodic
M ATLAB verification:
% (b) y1(n) = 20-point IDFS{X(0),X(2*pi/20),...,X(28*pi/20)}
N = 20; k = 0:N-1; w = 2*pi*k/N;
[Y2] = dtft(x,n,w); y2 = real(idfs(Y2,N))
y2 =
Columns 1 through 7
2.0000
5.0000
3.0000
-4.0000
-2.0000
6.0000
Columns 8 through 14
-3.0000
-3.0000
2.0000
0.0000
0.0000
0.0000
Columns 15 through 20
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000

0.0000
0.0000

6. Problem P 5.6

A 12-point sequence x (n) = f1; 2; 3; 4; 5; 6; 6; 5; 4; 3; 2; 1g.

(a) DFT X (k):
clear; close all;
%
xn = [1,2,3,4,5,6,6,5,4,3,2,1]; N = length(xn); % given signal x(n)
Xk = dft(xn,N); k = 0:N-1;
% DFT of x(n)
mag_Xk = abs(Xk); pha_Xk = angle(Xk)*180/pi;
% Mag and Phase of X(k)
zei = find(mag_Xk < 0.00001);
% Set phase values to
pha_Xk(zei) = zeros(1,length(zei));
% zero when mag is zero
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,HS020500);
subplot(2,1,1); stem(k,mag_Xk); axis([0,N,0,40])

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

53

set(gca,XTickMode,manual,XTick,[0:1:N]);
set(gca,YTickMode,manual,YTick,[0;20;40]);
xlabel(k); ylabel(magnitude); title(Magnitude plots of DFT and DTFT)
hold on
subplot(2,1,2); stem(k,pha_Xk); axis([0,N,-180,180])
set(gca,XTickMode,manual,XTick,[0:1:N]);
set(gca,YTickMode,manual,YTick,[-180;-90;0;90;180]);
xlabel(k); ylabel(Degrees); title(Phase plots of DFT and DTFT)
hold on
The stem plot of X (k) is shown in 5.4.
Plots in Problem P2.6
Magnitude plots of DFT and DTFT

magnitude

40

20

10

11

12

10

11

12

Phase plots of DFT and DTFT

180

Degrees

90
0
90
180

6
k

Figure 5.4: Plots of DTFT and DFT of signal in Problem 5.6

(b) DTFT X e j :
[X,w] = freqz(xn,1,1000,whole);
mag_X = abs(X); pha_X = angle(X)*180/pi;
Dw = (2*pi)/N;
subplot(2,1,1); plot(w/Dw,mag_X); grid
hold off
subplot(2,1,2); plot(w/Dw,pha_X); grid
hold off

% DTFT of xn
% mag and phase of DTFT
% frequency resolution

The continuous plot of X e j is also shown in Figure 5.4.

(c) Clearly, the DFT in part (6a) is the sampled version of X e j .

(d) It is possible to reconstruct the DTFT from the DFT if length of the DFT is larger than or equal to the length of
sequence x (n). We can reconstruct using the complex interpolation formula
X e j

N 1

k=0

X (k)

2k
N


;

where () = e

j(N 1)=2 sin (N =2)

N sin (=2)

54

S OLUTIONS M ANUAL

For N = 12, we have

X e j

FOR

DSP

USING

11

X (k) e

M ATLAB

j (5:5)

k=0

A PRIL 98

sin (6)
12 sin (=2)

7. Problem P 5.7
This problem is done using M ATLAB.
(a) x1 (n) = 2 cos (0:2n) [u (n)

u (n

10)].

% (a) x1(n) = 2*cos(0.2*pi*n)*[u(n)-u(n-10)]

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.7a);
n1 = [0:9]; x1 = 2*cos(0.2*pi*n1); N1 = length(n1); N = 100;
% Length of DFT
[X1] = dft([x1, zeros(1,N-N1)],N);
mag_X1 = abs(X1(1:N/2+1)); w = (0:N/2)*2*pi/N;
subplot(2,1,1); stem(n1,x1); axis([-1,N1,min(x1),max(x1)]);
title(Sequence x1(n)); ylabel(x1); ntick = [n1(1):5:n1(N1),N1];
set(gca,XTickMode,manual,XTick,ntick); xlabel(n);
subplot(2,1,2); plot(w/pi,mag_X1); axis([0,1,0,max(mag_X1)]);
title(Magnitude of DTFT X1(omega)); ylabel(|X1|);
xlabel(frequency in pi units); suptitle(Problem P 5.7a Plots);
The plots are shown in Figure 5.5.
Problem P 5.7a Plots
Sequence x1(n)
2

x1

1
0
1
2

10

n
Magnitude of DTFT X1(omega)
10

|X1|

8
6
4
2
0
0

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units

0.7

0.8

0.9

Figure 5.5: Sequence and its DTFT magnitude plots in Problem P5.7a
(b) x2 (n) = sin (0:45n) sin (0:55n) ; 0  n  50.
% (b) x2(n) = sin(0.45*pi*n)*sin(0.55*pi*n), 0 <= n <= 50
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P5.7b);
n2 = [0:50]; x2 = sin(0.45*pi*n2).*sin(0.55*pi*n2); N2 = length(n2); N = 300;
[X2] = dft([x2, zeros(1,N-N2)],N);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

55

mag_X2 = abs(X2(1:N/2+1)); w = (0:N/2)*2*pi/N;

subplot(2,1,1); stem(n2,x2); axis([-1,N2,min(x2),max(x2)]);
title(Sequence x2(n)); ylabel(x2); ntick = [n2(1):10:n2(N2)];
set(gca,XTickMode,manual,XTick,ntick); xlabel(n);
subplot(2,1,2); plot(w/pi,mag_X2); axis([0,1,0,max(mag_X2)]);
title(Magnitude of DTFT X2(omega)); ylabel(|X2|);
xlabel(frequency in pi units); suptitle(Problem P 5.7b Plots);
The plots are shown in Figure 5.6.
Problem P 5.7b Plots
Sequence x2(n)

x2

0.5
0
0.5
1

10

20

30

40

50

n
Magnitude of DTFT X2(omega)
25

|X2|

20
15
10
5
0
0

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units

0.7

0.8

0.9

Figure 5.6: Sequence and its DTFT magnitude plots in Problem P5.7b
(c) x3 (n) = 3 (2)n ;

10  n  10. This problem is done using M ATLAB.

% (c) x3(n) = 3*(2)n*[u(n+10)-u(n-11)]

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.7c);
n3 = [-10:10]; x3 = 3*(2).n3; N3 = length(n3); N = 100;
% Length of DFT
[X3] = dft([x3, zeros(1,N-N3)],N);
mag_X3 = abs(X3(1:N/2+1)); w = (0:N/2)*2*pi/N;
subplot(2,1,1); stem(n3,x3); axis([-11,11,min(x3),max(x3)]);
title(Sequence x3(n)); ylabel(x3); ntick = [n3(1):5:n3(N3),N3];
set(gca,XTickMode,manual,XTick,ntick); xlabel(n);
subplot(2,1,2); plot(w/pi,mag_X3); axis([0,1,0,max(mag_X3)]);
title(Magnitude of DTFT X3(omega)); ylabel(|X3|);
xlabel(frequency in pi units); suptitle(Problem P 5.7c Plots);
The plots are shown in Figure 5.7.
(e) x5 (n) = 5 0:9e j=4

n

u (n).

% (e) x5(n) = 5*(0.9*exp(j*pi/4)).n*u(n)

Hf_1 = figure(Units,normalized,position,[.1,.1,.8,.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.7e);

56

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Problem P 5.7c Plots

Sequence x3(n)
3000
2500
x3

2000
1500
1000
500
10

0
n
Magnitude of DTFT X3(omega)

10

6000

|X3|

4000

2000

0
0

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units

0.7

0.8

0.9

Figure 5.7: Plots in Problem P5.7c

n5 = [0:99]; x5 = 5*(0.9*exp(j*pi/4)).n5;
N5 = length(n5);
% Length of DFT
[X5] = dft(x5,N5); x5 = real(x5);
mag_X5 = abs(X5(1:N5/2+1)); w = (0:N5/2)*2*pi/N5;
subplot(2,1,1); stem(n5,x5); %axis([-11,11,min(x5),max(x5)]);
title(Sequence real(x5(n))); ylabel(real(x5)); ntick = [n5(1):10:n5(N5),N5];
set(gca,XTickMode,manual,XTick,ntick); xlabel(n);
subplot(2,1,2); plot(w/pi,mag_X5); axis([0,1,0,max(mag_X5)]);
title(Magnitude of DTFT X5(omega)); ylabel(|X5|);
xlabel(frequency in pi units); suptitle(Problem P 5.7e Plots);
The plots are shown in Figure 5.8.
8. Problem P 5.8
The impulse response h (n) is causal and real-valued.


(a) It is known that Re H e j



Re H e

k
= 5k=0 (0:5) cos k.

(
= Re

Consider
)

h (k) e

jk

k =0

h (k) Re e

k =0

jk

h (k) cos k

k=0

Comparing with the given expression, we obtain



h (n) =


(b) It is known that Im H e j

5
= `=0 2` sin `

and

(0:5)

0;
R

0k5
else

e j d = 0. From the second condition

H e j d = h (0) = 0

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

57

Problem P 5.7e Plots

Sequence real(x5(n))
6

real(x5)

4
2
0
2
4
0

10

20

30

40

50
60
70
n
Magnitude of DTFT X5(omega)

0.1

0.2

0.3

0.4
0.5
0.6
frequency in pi units

80

90

100

0.8

0.9

|X5|

40
30
20
10
0
0

0.7

Figure 5.8: DTFT plot in Problem P5.7e.

Consider

Im H e

h (`) e

= Im

`=

h (`) Im e

`=

Comparing with the given expression, we obtain

h (n ) =

j`

j`

h ( ) sin
`

`=

2`; 0  `  5
0;
else

9. Problem P 5.9
An N-point sequence x (n):
N 1

N 1

(a) The N-point DFT of x (n): X (k) = x (m) WNmk . The N-point DFT of X (k): y (n) = X (k) WNkn . Hence,
N 1

y (n)

m=0

N 1

x (m) WNmk

N 1

WNkn =

N 1

x (m) WNmkWNkn 0  n  N
;

m=0
m=0
k =0
N 1
N 1

(m+n)k
x ( m)
WN
=
x ( m)
N (m + n
r=
m=0
m=0
k=0

N
x ( n + rN ) = Nx (( n))N ; 0 n N 1
r=
k=0
N 1

k=0

rN ) ; 0  n  N

 

This means that y (n) is a circularly folded and amplified (by N) version of x (n). Continuing further, if we take
two more DFTs of x (n) then
x ( n)

N-point
DFT

N-point
DFT

N-point
DFT

N-point
DFT

! N 2 x (n)

Therefore, if a given DFT function is working correctly then four successive applications of this function on any
arbitrary signal will produce the same signal (multiplied by N 2 ). This approach can be used to verify a DFT
function.

58

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

(b) M ATLAB function for circular folding:

function x2 = circfold(x1,N)
% Circular folding with respect to N
% ---------------------------------% function x2 = circfold(x1,N)
%
x2(n) = x1((-n) mod N)
%
x2 = real(dft(dft(x1,N),N))/N;
(c) M ATLAB verification:
x = [1,2,3,4,5,6,6,5,4,3,2,1], N = length(x);
x =
1
2
3
4
5
6
6
5
y = circfold(x,N)
y =
Columns 1 through 7
1.0000
1.0000
2.0000
3.0000
4.0000
Columns 8 through 12
6.0000
5.0000
4.0000
3.0000
2.0000

5.0000

6.0000

Clearly, the circular folding of the sequence is evident.

10. Problem P 5.10. Circular even-odd decomposition of a complex sequence.
xec (n) ,

1

[x (n) + x (( n))N ]
2
1
xoc (n) , [x (n) + x (( n))N ]
2

(a) Using the DFT properties of conjugation and circular folding, we obtain
DFT [xec (n)] =
=

1
fDFT [x (n)] + DFT [x (( n))N ]g
2

1
X (k) + X  (( k))N ; where X (k) = DFT [x (( n))N ]
2
1

= fX (k) + X (k)g = Re [X (k)] = Re [X (( k))N ]
2

similarly, we can show that

DFT [xoc (n)] = j Im [X (k)] = j Im [X (( k))N ]
(b) The modified circevod function:
function [xec, xoc] = circevod(x)
% Complex-valued signal decomposition into circular-even and circular-odd parts
% ---------------------------------------------------------------------------% [xec, xoc] = circecod(x)
%
N = length(x); n = 0:(N-1);
xec = 0.5*(x + conj(x(mod(-n,N)+1)));
xoc = 0.5*(x - conj(x(mod(-n,N)+1)));
(c) M ATLAB verification:
% (c) Matlab Verification
n = 0:19; x = (0.9*exp(j*pi/3)).n; N = length(x);
[xec, xoc] = circevod(x);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

59

X = dft(x,N); Xec = dft(xec,N); Xoc = dft(xoc,N);

Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P5.10);
subplot(2,2,1); stem(n,real(X)); axis([-0.5,20.5,-1,7]);
title(Real{DFT[x(n)]}); ylabel(Re{X(k)});
subplot(2,2,3); stem(n,real(Xec)); axis([-0.5,20.5,-1,7]);
title(DFT[xec(n)]); ylabel(Xec(k)); xlabel(k);
subplot(2,2,2); stem(n,imag(X)); axis([-0.5,20.5,-5,5]);
title(Imag{DFT[x(n)]}); ylabel(Im{X(k)});
subplot(2,2,4); stem(n,imag(Xoc)); axis([-0.5,20.5,-5,5]);
title(DFT[xoc(n)]); ylabel(Xoc(k)); xlabel(k);
suptitle(Problem P 5.10 Plots);
The plots are shown in Figure 5.9.
Problem P 5.10 Plots
Real{DFT[x(n)]}

Imag{DFT[x(n)]}
5

Im{X(k)}

Re{X(k)}

6
4
2

0
0

10

15

20

DFT[xec(n)]

10

15

20

15

20

DFT[xoc(n)]
5

Xoc(k)

Xec(k)

0
0

10
k

15

20

10
k

Figure 5.9: Plots in Problem P5.10

11. Problem P 5.11

Since x (n) is a real-valued sequence, its DFT is conjugate symmetric. Thus
X (k) = f0:25; 0:125

j0:3; 0; 0:125

j0:06; 0:5; 0:125 + j0:06; 0; 0:125 + j0:3g:

From this we can compute the 8-point sequence x (n). M ATLAB script:
clear, close all;
N = 8; X = [0.25,0.125-j*0.3,0,0.125-j*0.06,0.5];
X = [X,conj(X(4:-1:2))]
% Reconstruct the entire X(k)
X =
Columns 1 through 4
0.2500
0.1250 - 0.3000i
0
0.1250 - 0.0600i
Columns 5 through 8

60

S OLUTIONS M ANUAL

FOR

DSP

0.5000
0.1250 + 0.0600i
x = real(idft(X,N))
x =
Columns 1 through 7
0.1562
0.0324
0.1538
0.0324
Column 8
-0.0949

USING

M ATLAB

A PRIL 98

0.1250 + 0.3000i

0.0312

-0.0949

0.0337

(a) Consider x1 (n) = x ((2 n))8 = x (( [n 2]))8 . It is obtained by first circular folding and then circular shifting by
2 of x (n). Hence using properties of the DFT
X1 (k) = DFT [x ((2

n))8 ] = W8

2k

((

k))8 :

M ATLAB script:
% (a) x1(n) = x((2-n))_8; Circ folding followed
N = 8; WN = exp(-j*2*pi/N); k = 0:N-1; m = 2;
X1 = circfold(X,N)
X1 =
Columns 1 through 4
0.2500 + 0.0000i
0.1250 + 0.3000i
0.0000
Columns 5 through 8
0.5000 + 0.0000i
0.1250 - 0.0600i
0.0000
X1 = (WN.(m*k)).*X1
X1 =
Columns 1 through 4
0.2500 + 0.0000i
0.3000 - 0.1250i
0.0000
Columns 5 through 8
0.5000 + 0.0000i -0.0600 - 0.1250i
0.0000
% Matlab verification
x1 = circfold(x,N); x1 = cirshftt(x1,m,N);
X12 = dft(x1,N)
X12 =
Columns 1 through 4
0.2500 + 0.0000i
0.3000 - 0.1250i
0.0000
Columns 5 through 8
0.5000 + 0.0000i -0.0600 - 0.1250i
0.0000
difference = max(abs(X1-X12))
difference =
4.7692e-016
(b) Let X

by circ shifting by 2

- 0.0000i

0.1250 + 0.0600i

- 0.0000i

0.1250 - 0.3000i

+ 0.0000i

-0.0600 + 0.1250i

+ 0.0000i

0.3000 + 0.1250i

+ 0.0000i

-0.0600 + 0.1250i

- 0.0000i

0.3000 + 0.1250i

mk be the 10-point DFT of x n . Then

10

( )

( )

5k
DFT [x ((n + 5))10 ] = W10
X

mk

10

( )

Note that x ((n + 5))10 is 10-point circularly shifted sequence of x (n), shifted to the left by 5 samples. M ATLAB
script:
% (b) x2(n) = x((n+5))_{10}
N = 10; WN = exp(-j*2*pi/N); k = 0:N-1; m = -5;
X2 = (WN.(m*k)).*dft([x,0,0],N)
X2 =
Columns 1 through 4
0.2500
-0.2916 + 0.2848i
0.0176 - 0.0487i
Columns 5 through 8

-0.0863 - 0.1368i

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

0.0108 + 0.0047i -0.5000 + 0.0000i

Columns 9 through 10
0.0176 + 0.0487i -0.2916 - 0.2848i
% Matlab verification
x2 = cirshftt(x,m,N); X22 = dft(x2,N)
X22 =
Columns 1 through 4
0.2500
-0.2916 + 0.2848i
Columns 5 through 8
0.0108 + 0.0047i -0.5000 - 0.0000i
Columns 9 through 10
0.0176 + 0.0487i -0.2916 - 0.2848i
difference = max(abs(X2-X22))
difference =
8.2523e-016

USING

M ATLAB

61

0.0108 - 0.0047i

-0.0863 + 0.1368i

0.0176 - 0.0487i

-0.0863 - 0.1368i

0.0108 - 0.0047i

-0.0863 + 0.1368i

(c) Consider x3 (n) = x2 (n) = x (n) x (n). Hence the DFT of x3 (n) is given by
X3 (k) =

1
X (k) N X (k) ;
N

N = 8:

M ATLAB script:
% (c) x3(n) = x(n)*x(n);
N = 8;
X3 = circonvf(X,X,N)/N;
% Matlab verification
x3 = x.*x;
X32 = dft(x3,N);
difference = max(abs(X3-X32))
difference =
4.5103e-017
(d) Let X

mk be the 10-point DFT of x n . Then

10

( )

( )

DFT x (n) 8 x (( n))8


= X (k) X ((

k))8

Note that x (( n))8 and X (( k))8 are the 8-point circular foldings of their respective sequences. M ATLAB script:
clear, close all;
N = 8; X = [0.25,0.125-j*0.3,0,0.125-j*0.06,0.5]; k = 0:N-1; n = k;
X = [X,conj(X(4:-1:2))]
% Reconstruct the entire X(k)
X =
Columns 1 through 4
0.2500
0.1250 - 0.3000i
0
0.1250 - 0.0600i
Columns 5 through 8
0.5000
0.1250 + 0.0600i
0
0.1250 + 0.3000i
x = real(idft(X,N))
x =
Columns 1 through 7
0.1562
0.0324
0.1538
0.0324
0.0312
-0.0949
0.0337
Column 8
-0.0949
%
% (d) x4(n) = x(n)(8)x((-n))_8
X0 = X(mod(-k,N)+1);
% DFT of x((-n))_8

62

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

X4 = X .* X0
% DFT of x(n) (8) x((-n))_8
X4 =
Columns 1 through 7
0.0625
0.1056
0
0.0192
0.2500
0.0192
Column 8
0.1056
% Verification
x4 = circonvt(x,x(mod(-n,N)+1),N); X44 = dft(x4,N);
difference = max(abs(X4-X44))
difference =
2.1330e-016

12. Problem P 5.12 (Two real DFTs using one complex DFT)
x (n) = xR (n) + jxI (n) ; xR (n) and xI (n) are real sequences
Then
xR (n)

jxI (n)

(a) Consider
XR (k)

def
=

1
fx (n) + x (n)g ;
2
1
fx (n) x (n)g
2

and

1
DFT [xR (n)] = fDFT [x (n)] + DFT [x (n)]g
2
1
def
f
X (k) + X  (( k))N g = Xec
2

Similarly
jXI (k)

def
=

1
DFT [ jxI (n)] = fDFT [x (n)] DFT [x (n)]g
2
1
def
f
X (k) X  (( k))N g = Xoc
2

9
>
=
>
;

) XI = Xjoc =

(b) M ATLAB function real2dft:

function [X1,X2] = real2dft(x1,x2,N)
% DFTs of two real sequences
% [X1,X2] = real2dft(x1,x2,N)
% X1 = N-point DFT of x1
% X2 = N-point DFT of x2
% x1 = real-valued sequence of length <= N
% x2 = real-valued sequence of length <= N
%
N = length of DFT
%
% Check for length of x1 and x2
if length(x1) > N
error(*** N must be >= the length of x1 ***)
end
if length(x2) > N
error(*** N must be >= the length of x2 ***)
end
N1 = length(x1); x1 = [x1 zeros(1,N-N1)];
N2 = length(x2); x2 = [x2 zeros(1,N-N2)];
x = x1 + j*x2;
X = dft(x,N);
[X1, X2] = circevod(X); X2 = X2/j;

jXoc

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

63

We will also need the circevod function for complex sequences (see Problem P5.10). This can be obtained from
the one given in the text by two simple changes.
function [xec, xoc] = circevod(x)
% Complex signal decomposition into circular-even and circular-odd parts
% ---------------------------------------------------------------------% [xec, xoc] = circecod(x)
%
N = length(x); n = 0:(N-1);
xec = 0.5*(x + conj(x(mod(-n,N)+1)));
xoc = 0.5*(x - conj(x(mod(-n,N)+1)));
(c) M ATLAB verification:
clear; close all;
N = 64; n = 0:N-1; x1 = cos(0.25*pi*n); x2 = sin(0.75*pi*n);
[X1,X2] = real2dft(x1,x2,N);
X11 = dft(x1,N); X21 = dft(x2,N);
difference = max(abs(X1-X11))
difference =
1.4918e-013
difference = max(abs(X2-X21))
difference =
1.4914e-013
13. Problem P 5.13
Circular shifting:
(a) The M ATLAB routine cirshftf.m to implement circular shift is written using the frequency-domain property

y (n) = x ((n

m))N

= IDFT

X (k) WNmk

This routine will be used in the next problem to generate a circulant matrix and has the following features. If m is a
scaler then y (n) is circularly shifted sequence (or array). If m is a vector then y (n) is a matrix, each row of which
is a circular shift in x (n) corresponding to entries in the vector m.
function y = cirshftf(x,m,N)
% Circular shift of m samples wrt size N in sequence x: (freq domain)
% ------------------------------------------------------------------% function y=cirshift(x,m,N)
%
y : output sequence containing the circular shift
%
x : input sequence of length <= N
%
m : sample shift
%
N : size of circular buffer
%
% Method: y(n) = idft(dft(x(n))*WN(mk))
%
% If m is a scalar then y is a sequence (row vector)
% If m is a vector then y is a matrix, each row is a circular shift
% in x corresponding to entries in vecor m
% M and x should not be matrices
%
% Check whether m is scalar, vector, or matrix
[Rm,Cm] = size(m);
if Rm > Cm
m = m; % make sure that m is a row vector

64

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

end
[Rm,Cm] = size(m);
if Rm > 1
error(*** m must be a vector ***) % stop if m is a matrix
end
% Check whether x is scalar, vector, or matrix
[Rx,Cx] = size(x);
if Rx > Cx
x = x; % make sure that x is a row vector
end
[Rx,Cx] = size(x);
if Rx > 1
error(*** x must be a vector ***) % stop if x is a matrix
end
% Check for length of x
if length(x) > N
error(N must be >= the length of x)
end
x=[x zeros(1,N-length(x))];
X=dft(x,N);
X=ones(Cm,1)*X;
WN=exp(-2*j*pi/N);
k=[0:1:N-1];
Y=(WN.(m * k)).*X;
y=real(conj(dfs(conj(Y),N)))/N;
(b) M ATLAB verification:
n = [0:1:10]; x = 11*ones(1,length(n))-n;
y = cirshftf(x,10,15)
y =
Columns 1 through 7
6.0000
5.0000
4.0000
3.0000
Columns 8 through 14
0.0000
0.0000
0.0000
11.0000
Column 15
7.0000

2.0000

1.0000

0.0000

10.0000

9.0000

8.0000

14. Problem P 5.14

Parsevals relation for the DFT:
N 1

jx (n)j

N 1

n=0

N 1

x (n) x (n) =
n =0
n=0
(
)
N 1
1N 1
nk

X (k)
x (n) WN
=
N k=0
n =0

1N 1
X (k) WN nk x (n)
N k
=0

1N 1
X (k )
N k
=0

N 1

N 1

jx (n)j2 = N X (k) X  (k) = N jX (k)j2

n =0

k=0

M ATLAB verification:
x = [1,2,3,4,5,6,6,5,4,3,2,1]; N = length(x);
% power of x(n) in the time-domain
power_x = sum(x.*conj(x))

k=0

n=0

Therefore,
N 1

)

N 1

x (n) WNnk

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

65

power_x =
182
X = dft(x,N);
power_X = (1/N)*sum(X.*conj(X))
power_X =
182.0000
15. Problem P 5.15
M ATLAB function circonvf:
function y = circonvf(x1,x2,N)
%
%function y=circonvf(x1,x2,N)
%
% N-point circular convolution between x1 and x2: (freq domain)
% ------------------------------------------------------------%
y : output sequence containing the circular convolution
%
x1 : input sequence of length N1 <= N
%
x2 : input sequence of length N2 <= N
%
N : size of circular buffer
%
% Method: y(n) = idft(dft(x1)*dft(x2))
% Check for length of x1
if length(x1) > N
error(N must be >= the length of x1)
end
% Check for length of x2
if length(x2) > N
error(N must be >= the length of x2)
end
x1=[x1 zeros(1,N-length(x1))];
x2=[x2 zeros(1,N-length(x2))];
X1=fft(x1); X2=fft(x2);
y=real(ifft(X1.*X2));
16. Problem P 5.16. Circular convolution using circulant matrix operation.

x1 (n) = f1; 2; 2g ; x2 (n) = f1; 2; 3; 4g ; x3 (n) , x1 (n) 4 x2 (n)

(a) Using the results from Example 5.13, we can express the above signals as
2

1
6 2 7
7
x1 = 6
4 2 5;
0

1
6 2
X2 = 6
4 3
4

4
1
2
3

3
4
1
2

2
3 7
7
4 5
1

The matrix X2 has the property that its every row or column can be obtained from the previous row or column using
circular shift. Such a matrix is called a circulant matrix. It is completely described by the first column or the row.
The following M ATLAB function circulnt uses the mod function to generate a circulant matrix.

66

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

function C = circulnt(h,N)
% Circulant matrix generation using vector data values
% ---------------------------------------------------% function C = circulnt(h,N)
%
%
C : Circulant matrix
%
h : input sequence of length <= N
%
N : size of the circular buffer
%
Method: C = h((n-m) mod N); n : col vec, m : row vec
Mh
h
C
m
nm
C(:)

=
=
=
=
=
=

length(h); h = reshape(h,Mh,1);
[h; zeros(N-Mh,1)];
zeros(N,N);
0:N-1; n=m;
mod((n*ones(1,N)-ones(N,1)*m),N);
h(nm+1);

(b) Circular convolution:

1
6 2
x 3 = X2 x 1 = 6
4 3
4

4
1
2
3

3
4
1
2

%
%
%
%
%
%

reshape h into column vector

zero-pad h
establish size of C
indices n and m
(n-m) mod N in matrix form
h((n-m) mod N)

32

2
1
6 2 7 6
3 7
76
7=6
4 54 2 5 4
1
0

15
12 7
7
9 5
14

M ATLAB script:
% Chapter 5: Problem P5.16: Circular convolution using Circulant matrix
clear, close all;
N = 4; x1 = [1,2,2,0]; x2 = [1,2,3,4];
% (a) Circulant matrix
X2 = circulnt(x2,N)
X2 =
1
4
3
2
1
4
3
2
1
4
3
2

2
3
4
1

% (b) Circular Convolution

x3 = X2*x1; x3 = x3
x3 =
15
12
9
14
17. Problem P 5.17.
M ATLAB function circulnt:
function C = circulnt(x,N)
% Circulant matrix generation using vector data values
% ---------------------------------------------------% function C = circulnt(h,N)
%
%
C : Circulant matrix
%
x : input sequence of length <= N
%
N : size of the circular buffer
%
Method: C = h((n-m) mod N);
Mx = length(x);

% length of x

A PRIL 98

S OLUTIONS M ANUAL

x
C
m
x
C

=
=
=
=
=

FOR

[x, zeros(N-Mx,1)];
zeros(N,N);
0:N-1;
circfold(x,N);
cirshift(x,m,N);

%
%
%
%
%

DSP

USING

M ATLAB

67

zero-pad x
establish size of C
indices n and m
Circular folding
Circular shifting

M ATLAB verification on sequences in Problem 5.16 (or Example 5.13):

clear, close all;
N = 4; x1 = [1,2,2,0]; x2 = [1,2,3,4];
% (a) Circulant matrix
X2 = circulnt(x2,N)
X2 =
1.0000
4.0000
3.0000
2.0000
2.0000
1.0000
4.0000
3.0000
3.0000
2.0000
1.0000
4.0000
4.0000
3.0000
2.0000
1.0000
% (b) Circular Convolution
x3 = X2*x1; x3 = x3
x3 =
15.0000
12.0000
9.0000

14.0000

18. Problem P 5.18

(a) Circular convolution:
x1 (n) = f1; 1; 1; 1g ; x2 (n) = cos (n=4) RN (n) ; N = 8;

x 3 (n ) = x 1 (n ) N x 2 (n )

Circular convolutions using M ATLAB:

clear, close all;
%(a) x1(n) = [1,1,1,1]; x2(n) = cos(pi*n/4); N = 8;
N = 8; n = 0:N-1;
x1 = [1,1,1,1,0,0,0,0]; x2 = cos(pi*n/4);
x3 = circonvt(x1,x2,N)
x3 =
Columns 1 through 7
1.0000
2.4142
2.4142
1.0000
-1.0000
Column 8
-1.0000
(d) Circular convolution:
x1 (n) = nRN ; x2 (n) = (N

n) RN (n) ; N = 10;

-2.4142

-2.4142

x3 (n) = x1 (n) N x2 (n)

Circular convolutions using M ATLAB:

N = 10; n = [0:N-1]; x1 = n; x2 = (N-n);
x3 = real(idft(dft(x1,N).*dft(x2,N),N))
x3 =
Columns 1 through 7
285.0000 250.0000 225.0000 210.0000
Columns 8 through 10
250.0000 285.0000 330.0000

205.0000

210.0000

225.0000

68

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

19. Problem P 5.19

(b) Error between Linear and Circular Convolutions:
x1 (n) = cos (2n=N ) R16 (n) ; x2 (n) = sin (2n=N ) R16 (n) ;

N = 32

M ATLAB script:
% (b) x1(n) = cos(2*pi*n/N)*R16, x2(n) = cos(2*pi*n/N)*R16; N = 32;
N = 32; n = 0:15;
x1 = cos(2*pi*n/N); x2 = sin(2*pi*n/N);
x3 = circonvf(x1,x2,N)
x3 =
Columns 1 through 7
0.0000
0.1951
0.5740
1.1111
1.7678
2.4944
3.2336
Columns 8 through 14
3.9231
4.5000
4.9039
5.0813
4.9888
4.5962
3.8890
Columns 15 through 21
2.8701
1.5607
0.0000
-1.3656
-2.4874
-3.3334
-3.8891
Columns 22 through 28
-4.1573
-4.1575
-3.9231
-3.5000
-2.9424
-2.3097
-1.6629
Columns 29 through 32
-1.0607
-0.5556
-0.1913
0.0000
x4 = conv(x1,x2), x4 = [x4, zeros(1,33)]; N4 = length(x4);
x4 =
Columns 1 through 7
0
0.1951
0.5740
1.1111
1.7678
2.4944
3.2336
Columns 8 through 14
3.9231
4.5000
4.9039
5.0813
4.9888
4.5962
3.8890
Columns 15 through 21
2.8701
1.5607
0.0000
-1.3656
-2.4874
-3.3334
-3.8891
Columns 22 through 28
-4.1573
-4.1575
-3.9231
-3.5000
-2.9424
-2.3097
-1.6629
Columns 29 through 31
-1.0607
-0.5556
-0.1913
e = round(x3 - x4(1:N))
e =
Columns 1 through 12
0
0
0
0
0
0
0
0
0
0
0
0
Columns 13 through 24
0
0
0
0
0
0
0
0
0
0
0
0
Columns 25 through 32
0
0
0
0
0
0
0
0
x4(N+1:N4)
ans =
Columns 1 through 12
0
0
0
0
0
0
0
0
0
0
0
0
Columns 13 through 24
0
0
0
0
0
0
0
0
0
0
0
0
Columns 25 through 32
0
0
0
0
0
0
0
0
(e) Error between Linear and Circular Convolutions:
x1 (n) = f1; 1; 1; 1g ; x2 (n) = f1; 0; 1; 0g ;
M ATLAB script:

N =5

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

69

% (e) x1(n) = {1,-1,1,-1}, x2(n) = {1,0,-1,0}; N = 5

x1 = [1,-1,1,-1]; x2 = [1,0,-1,0]; N = 5;
x3 = circonvf(x1,x2,N)
x3 =
2.0000
-1.0000
0.0000
0.0000
-1.0000
x4 = conv(x1,x2), N4 = length(x4);
x4 =
1
-1
0
0
-1
1
0
e = x3 - x4(1:N)
e =
1.0000
0.0000
0.0000
0.0000
0.0000
x4(N+1:N4)
ans =
1
0
20. Problem P 5.21.

x1 (n) = f2; 1; 1; 2g ; x2 (n) = f1; 1; 1; 1g

(a) Circular convolution x1 (n) N x2 (n) for N = 4; 7; 8. M ATLAB script:

clear, close all;
x1 = [2,1,1,2]; x2 = [1,-1,1,-1];
% (a) Circular Convolution for N = 4,7,8
y_4 = round(circonv(x1,x2,4))
y_4 =
0
0
0
0
y_7 = round(circonv(x1,x2,7))
y_7 =
2
-1
2
0
-2
1
-2
y_8 = round(circonv(x1,x2,8))
y_8 =
2
-1
2
0
-2
1
-2

(b) Linear convolution x1 (n)  x2 (n). M ATLAB script:

% (b) linear Convolution
y_l = conv(x1,x2)
y_l =
2
-1
2
0

-2

-2

(c) The minimum value of N should be 7.

(d) Since x1 (n) and x2 (n) are 4-point sequences, the minimum length should be 4 + 4

1 = 7.

Chapter 6

Digital Filter Structures

1. Problem P 6.1
A causal LTI system is described by
5

y ( n) =

k=0

and is driven by the input

 k

1
2

x (n

k) +
`=

 k

1
3

y (n

`)

x (n) = u (n) ; 0  n  100

The M ATLAB script to determine various realizations is:

clear; close all;
%% Direct Forms %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Direct Form Coefficients ***)
*** Direct Form Coefficients ***
k = 0:5; b = (0.5).k
b =
1.0000
0.5000
0.2500
0.1250
0.0625
0.0312
l = 1:5; a = [1, -(1/3).l]
a =
1.0000
-0.3333
-0.1111
-0.0370
-0.0123
-0.0041
% Output Response
[x,n] = stepseq(0,0,100);
y_DF = filter(b,a,x);
%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
1
B =
1.0000
0.5000
0.2500
1.0000
-0.5000
0.2500
1.0000
0.5000
0
A =
1.0000
0.4522
0.0745
1.0000
-0.1303
0.0843
1.0000
-0.6553
0
% Output Response
70

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

71

y_CF = casfiltr(b0,B,A,x);
%% Parallel Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Parallel Form Coefficients ***)
*** Parallel Form Coefficients ***
[C,B,A] = dir2par(b,a)
C =
-7.5938
B =
3.4526
0.8052
3.3193
-0.0367
1.8219
0
A =
1.0000
0.4522
0.0745
1.0000
-0.1303
0.0843
1.0000
-0.6553
0
% Output Response
y_PF = parfiltr(C,B,A,x);
%% Filter response plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P6.1);
subplot(3,1,1); stem(n,y_DF); axis([-1,101,0,5])
set(gca,XTickMode,manual,XTick,[0:25:100],fontsize,10);
set(gca,YTickMode,manual,YTick,[0;4]);
ylabel(y_DF); title(Direct Form Filter Response)
subplot(3,1,2); stem(n,y_CF); axis([-1,101,0,5])
set(gca,XTickMode,manual,XTick,[0:25:100],fontsize,10);
set(gca,YTickMode,manual,YTick,[0;4]);
ylabel(y_DF); title(Cascade Form Filter Response)
subplot(3,1,3); stem(n,y_PF); axis([-1,101,0,5])
set(gca,XTickMode,manual,XTick,[0:25:100],fontsize,10);
set(gca,YTickMode,manual,YTick,[0;4]);
xlabel(n); ylabel(y_DF); title(Parallel Form Filter Response)
The block diagrams are shown in Figure 6.1 and the output responses are shown in Figure 6.2.
2. Problem P 6.2
The IIR filter is decribed by the system function


H (z) = 2

1 + 0z 1 + z 2
1 0:8z 1 + 0:64z


2

2
1

z 1
0:75z


1

1 + 2z 1 + z 2
1 + 0:81z 2

The given structure is a parallel form. The M ATLAB script to determine various realizations is:
clear; close all;
%% Cascade Form Coefficients %%%%%%%%%%%%%%%%%%%%%%%%%%
b0 = [2]; B = [1,0,1; 2,-1,0; 1,2,1];
A = [1,-0.8,0.64; 1,-0.75,0; 1,0,0.81];
%% Direct Forms %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Direct Form Coefficients ***)

72

S OLUTIONS M ANUAL

x(n)

FOR

z
z

-1

USING

y(n)

1
-1

DSP

0.1111

-1

0.1111

-1

0.0370

z-1

0.0370

z-1 0.125

z 0.0625

0.0123

0.0123

z 0.0625

z-1 0.03125

0.0041

z-1

0.0041

z-1 0.03125

-1

x(n)

1
z

1
z

0.5

y(n)

-1

0.1303

z-1
-0.0745

0.25

b. Direct Form II
1

-1

-0.4522

0.5

-1

-1

a. Direct Form I
1

y(n)

1
z

z-1 0.125

x(n)
0.3333

0.25

0.3333

A PRIL 98

-1

0.5

-1

M ATLAB

-1

-0.5

0.6553

0.5

z-1
0.25

-0.0843

0.25

c. Cascade Form
-7.5938
3.4526

x(n)
-0.4522

-0.0745

z-1

0.1303

-1

-1

-0.0843

0.6553

-1

y(n)

-1

0.8052

3.3193
-0.0367

1.8219

d. Parallel Form

Figure 6.1: Block Diagrams in Problem P 6.1

*** Direct Form Coefficients ***

[b,a] = cas2dir(b0,B,A); b = b(1:length(b)-1), a = a(1:length(a)-1)
b =
4
6
4
4
0
-2
a =
1.0000
-1.5500
2.0500
-1.7355
1.0044
-0.3888
%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
4
B =
1.0000
0.0000
1.0000
1.0000
2.0000
1.0000
1.0000
-0.5000
0
A =
1.0000
0.0000
0.8100
1.0000
-0.8000
0.6400

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

73

Problem P 6.1 Plots

Direct Form Filter Response

y_DF

25

50
Cascade Form Filter Response

75

100

25

50
Parallel Form Filter Response

75

100

25

50
n

75

100

y_CF

y_PF

Figure 6.2: Output Responses in Problem P 6.1

1.0000

-0.7500

%% Parallel Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

disp(*** Parallel Form Coefficients ***)
*** Parallel Form Coefficients ***
[C,B,A] = dir2par(b,a)
C =
5.1440
B =
2.0137
0.1106
-10.8732
15.0013
7.7155
0
A =
1.0000
0.0000
0.8100
1.0000
-0.8000
0.6400
1.0000
-0.7500
0
The block diagrams are shown in Figure 6.3.
3. Problem P 6.3
The IIR filter is decribed by the system function
H (z) =

14:75 12:9z
3
1 76 z 1 + 32
z

1
2


+

24:5 + 26:82z
1 z 1 + 0:5z

1
2

The given structure is a parallel form. The M ATLAB script to determine various realizations is:
clear; close all;
%% Parallel Form Coefficients %%%%%%%%%%%%%%%%%%%%%%%%%%

74

S OLUTIONS M ANUAL

x(n)

DSP

FOR

y(n)

4
z-1
z

-1

-1

z-1
z

-1

USING

6
4

2.05

-1

-1

1.7355

1.0044

z-1

0.3888

x(n)

1.55

-1

0.81

z-1

1.7355

-1

1.0044

z-1

0.3888

-1

b. Direct Form II
1

0.8

z-1
0

z-1
-1

2.05

z-1
0

y(n)

a. Direct Form I
1

A PRIL 98

x(n)

z-1

1.55

M ATLAB

0.64

z-1
2

z-1

0.75

0.5

c. Cascade Form
5.144
2.0137

x(n)
0
0.81

-1

y(n)

0.1106

z-1
10.8732

0.8
0.64

-1

-1

-1

15.0013

7.7155
0.75

d. Parallel Form

Figure 6.3: Block Diagrams in Problem P 6.2

C = []; B = [-14.75, -12.9; 24.5, 26.82];

A = [1, -7/8, 3/32; 1, -1, 1/2];
%% Direct Forms %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Direct Form Coefficients ***)
*** Direct Form Coefficients ***
[b,a] = par2dir(C,B,A)
b =
9.7500
7.2325 -15.6456
-3.9356
a =
1.0000
-1.8750
1.4688
-0.5312
0.0469
%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
9.7500
B =

y(n)

A PRIL 98

S OLUTIONS M ANUAL

1.0000
1.0000

1.8251
-1.0834

0.3726
0

1.0000
1.0000

-1.0000
-0.8750

0.5000
0.0938

FOR

DSP

USING

M ATLAB

75

A =

%% Parallel Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

disp(*** Parallel Form Coefficients ***)
*** Parallel Form Coefficients ***
[C,B,A] = dir2par(b,a)
C =
[]
B =
24.5000
26.8200
-14.7500 -12.9000
A =
1.0000
-1.0000
0.5000
1.0000
-0.8750
0.0938
The block diagrams are shown in Figure 6.4.
x(n)

y(n)

9.75
z-1 7.2325

1.8750

z-1
-15.6456

-1.4688

z-1

-3.9356

z-1
z-1
z

-0.0469

y(n)

9.75

z-1

0.5312

a. Direct Form I

x(n)
1.8750

z-1 7.2325

-1.4688

z-1
-15.6456
z-1

0.5312

-1

-0.0469

-3.9356

-1

b. Direct Form II
1

9.75

x(n)

1
y(n)

z-1
1.0

1.8251

0.8750

0.3726

-0.0938

z-1
-0.5

z-1
-1.0834
z-1
0

c. Cascade Form
-14.75
x(n)

y(n)
7/8
-3/32

-1

-12.9

z-1

24.5
1
-0.5

-1

26.82

z-1

d. Parallel Form

Figure 6.4: Block Diagrams in Problem P 6.3

76

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

4. Problem P 6.5
A linear shift-invariant system has the system function
H (z) =

0:5 1 + z
1

3 1
7
+ 8z 2
2z

13 3
16 z

1 6
1 4
6z

11 5
7
+ 16 z 6
32 z

The given block diagram structure is a cascade form.

(a) Conversion to the cascade form: The M ATLAB script is:
clear; close all;
%% Direct Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
b0 = 0.5; broots = -ones(1,6); b = b0*poly(broots);
a = [1,-3/2,7/8,-13/16,-1/8,-11/32,7/16];
%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
0.5000
B =
1.0000
2.0067
1.0067
1.0000
2.0000
1.0000
1.0000
1.9933
0.9934
A =
1.0000
1.0000
0.5000
1.0000
-0.5000
1.0000
1.0000
-2.0000
0.8750
The block diagram structure is shown in Figure 6.5.
x(n)

0.5

-1

-1

-1

-0.5

-1

0.5

-1

-1

z
-0.875

y(n)

-1

-1

-1

-1

Figure 6.5: Cascade Form Structure in Problem P 6.5

(b) The above solution is not unique because the numerator and the denominator factors can be grouped differently.
5. Problem P 6.6
A linear shift-invariant system has the system function
H (z) =

5 + 11:2z

+ 5:44z 2
1 + 0:8z 1

0:384z
0:512z

The given structure is a parallel form.

(a) Conversion to the parallel form: The M ATLAB script is

3
3

2:3552z
0:4096z

4
4

1:2288z

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

77

clear; close all;

%% Direct Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
b = [5, 11.2, 5.44, -0.384, -2.3552, -1.2288];
a = [1, 0.8, 0, -0.512, -0.4096];
%% Parallel Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Parallel Form Coefficients ***)
*** Parallel Form Coefficients ***
[C,B,A] = dir2par(b,a)
C =
2.0000
3.0000
B =
1.0000
2.0000
2.0000
3.0000
A =
1.0000
0.8000
0.6400
1.0000
0.0000
-0.6400
The block diagram structure is shown in Figure 6.6.

z-1
2

x(n)

y(n)

z-1
-0.8

z-1
-0.64
2

z-1
0

z-1
0.64
Figure 6.6: Parallel Form Structure in Problem P6.6
(b) The above solution is unique because the numerator residues and the corresponding denominator poles must be
grouped into a second order section.
6. Problem P 6.7
A linear shift-invariant system has the system function
H (z) =

0:5 1 + z
1

3 1
7
+ 8z 2
2z

13 3
16 z

1 6
1 4
8z

11 5
7
+ 16 z 6
32 z

78

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

The new structure is a parallel-of-cascades form.

(a) Conversion to the parallel-of-cascades form: The M ATLAB script is
clear; close all;
%% Direct Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
b = 0.5*poly([-1,-1,-1,-1,-1,-1]);
a = [1, -3/2, 7/8, -13/16, -1/8, -11/32, 7/16];
%% Parallel Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Parallel Form Coefficients ***)
*** Parallel Form Coefficients ***
[C,B,A] = dir2par(b,a)
C =
1.1429
B =
-0.0210
-0.0335
0.9252
-3.6034
-1.5470
9.9973
A =
1.0000
1.0000
0.5000
1.0000
-0.5000
1.0000
1.0000
-2.0000
0.8750
%% Parallel of Cascade Forms: %%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Parallel of Cascade Forms Coefficients ***)
*** Parallel of Cascade Forms Coefficients ***
B = B(1:2,:), A = A(1:2,:)
B =
-0.0210
-0.0335
0.9252
-3.6034
A =
1.0000
1.0000
0.5000
1.0000
-0.5000
1.0000
[b1,a1] = par2dir(C,B,A)
b1 =
Columns 1 through 4
2.0470
-2.1298
-2.0023
Column 5
0.5714 + 0.0000i
a1 =
Columns 1 through 4
1.0000
0.5000
1.0000
Column 5
0.5000 + 0.0000i
[b0,B1,A1] = dir2cas(b1,a1)
b0 =
2.0470
B1 =
1.0000
0.9985
0.5050
1.0000
-2.0389
0.5527
A1 =
1.0000
1.0000
0.5000
1.0000
-0.5000
1.0000
The block diagram structure is shown in Figure 6.7.

-0.9780

0.7500

A PRIL 98

S OLUTIONS M ANUAL

x(n)

FOR

DSP

USING

2.0470

M ATLAB

z-1
0.9985

0.5

z-1
2.0359

0.5

z-1
0.5050

z-1
0.5527

79

y(n)

1.5470
z-1
9.9973

2.0
0.875

z-1

Figure 6.7: Parallel of Cascades Structure in Problem P6.6

(b) The above solution is unique because the numerator residues and the corresponding denominator poles must be
grouped into a second order section.
7. Problem P 6.8
A causal FIR filter is described by
10

y (n) =

k=0

 j5 kj

1
2

x (n

k)

The M ATLAB script to determine various realizations is:

clear; close all;
%% Direct Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Direct Form Coefficients ***)
*** Direct Form Coefficients ***
k = 0:10; b = (0.5).(abs(5-k)), a = 1;
b =
Columns 1 through 7
0.0312
0.0625
0.1250
0.2500
0.5000
1.0000
Columns 8 through 11
0.2500
0.1250
0.0625
0.0312
%% Linear-Phase Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Linear-Phase Form Coefficients ***)
*** Linear-Phase Form Coefficients ***
M = length(b); L = (M-1)/2;
B = b(1:1:L+1)
B =
0.0312
0.0625
0.1250
0.2500
0.5000
1.0000
%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
0.0312
B =

0.5000

80

S OLUTIONS M ANUAL

1.0000
1.0000
1.0000
1.0000
1.0000

1.7495
0.5046
-0.5506
-2.0806
2.3770

FOR

DSP

USING

M ATLAB

A PRIL 98

3.4671
0.2884
0.2646
3.7789
1.0000

A =
1
1
1
1
1

0
0
0
0
0

0
0
0
0
0

The block diagrams are shown in Figure 6.8.

0.0312

0.25

z-1

0.0625

z-1

0.1250

z-1

0.25

z-1

0.1250

0.0312

z-1

0.0625

z-1

x(n)

y(n)
a. Direct Form

z-1

x(n)

z-1
z-1
z-1

0.0312

-1

z
0.0625

-1

0.5

0.25

1
y(n)

b. Linear-Phase Form

x(n)

1
-1

z-1
1.7495

3.4671

-1

-1

0.5046

-0.5506

-2.0806

-1

1
-1

-1

z-1

-1

2.3770
-1

0.2884

0.2646

3.7789

1.0

c. Cascade Form

Figure 6.8: Block Diagrams in Problem P 6.8

8. Problem P 6.9
The FIR filter is decribed by the system function
10

H (z) =

(2z)

k=0

The M ATLAB script to determine various realizations is:

clear; close all;
%% Direct Form %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(*** Direct Form Coefficients ***)

y(n)

A PRIL 98

S OLUTIONS M ANUAL

*** Direct Form Coefficients ***

k = 0:10; b = (0.5).k, a = 1;
b =
Columns 1 through 7
1.0000
0.5000
0.2500
Columns 8 through 11
0.0078
0.0039
0.0020

FOR

DSP

0.1250

USING

M ATLAB

0.0625

0.0312

81

0.0156

0.0010

%% Cascade Form: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

disp(*** Cascade Form Coefficients ***)
*** Cascade Form Coefficients ***
[b0,B,A] = dir2cas(b,a)
b0 =
1
B =
1.0000
0.9595
0.2500
1.0000
0.6549
0.2500
1.0000
0.1423
0.2500
1.0000
-0.4154
0.2500
1.0000
-0.8413
0.2500
A =
1
0
0
1
0
0
1
0
0
1
0
0
1
0
0
The block diagrams are shown in Figure 6.9.
z-1

0.00098

0.00391

0.125

z-1

0.00195

z-1

0.00781

z-1

0.25

z-1

0.5

z-1

x(n)

y(n)
a. Direct Form

x(n)

1
-1

z-1
0.9595

0.25

-1

1
-1

0.6549

0.1423

-0.4154

-0.8413

-1

-1

-1

0.25

-1

0.25

-1

0.25

b. Cascade Form

Figure 6.9: Block diagrams in Problem P 6.9

-1

0.25

y(n)

Chapter 7

FIR Filter Design

1. Problem P 7.2
Amplitude response function for Type-II LP FIR filter
(a) Type-II ) symmetric h(n) andMeven, i.e.,
h(n) = h(M

n); 0  n  M

1;

1
2

is not an integer

Consider,
H e

M 1

jn

h (n ) e

M
2

n=0

jn

h (n ) e

M 1

n=0

h (n) e

jn

Change of variable in the second sum above:

n!M
Hence,
H e j

n)

M
2

n2=0 h (n) e

! M2

jn +
M 1
2
n=0
M
2 1
n=0
M
2 1
n=0

1; M

1 ! 0; and h (n) ! h (n)

2
h (n) e
nn
=0

j( M2 1 )

j( M2 1 )

h(n) e+ j(

j( M2 1 )

2h(n) cos

h (n ) e

j(M 1 n)

jn+ j( M2 1 )

M 1
2

n)

 M 1

j(M 1 n)+ j( M2 1 )

+e

+e

j( M2 1 n)



Change of variable:
n!n)n=0!n=

M
2
and



cos

 

Hence,
H e
Define b(n) = 2h

H e

M
2

=e

j( M2 1 )

M
2

2h

n=1

! cos

1!n=1

M
M
; n =
2
2

M
2

 

n


1
2



 

n cos n

1
2



n . Then,

=e

j( M2 1 )

M
2

b (n) cos

 

n =1

82

1
2



M
2

) Hr () = b (n) cos
n=1

 

1
2



A PRIL 98

S OLUTIONS M ANUAL


(b) Now cos n

i.e.,

1
2



DSP

FOR

M ATLAB

USING

83

can be written as a linear combination of higher harmonics in cos multiplied by cos



1
 2 
3
cos
 2 
5
cos
2

etc. Note that

the lowest harmonic frequency is zero and the highest harmonic frequency is (n
cos n 12 expansion. Hence,
 

1
2

b (n) cos n

Hr () =

n=1



M
2

= cos
2

b (n) cos (n)

where b (n) are related to b (n) through the above trignometric identities.
2. (Problem P 7.3) Type-III ) anti-symmetric h(n) and Modd, i.e.,
h(n) = h(M

n); 0  n  M

1; h

= 0;

1
2

is an integer

(a) Consider,
H e j

M 1

jn

h (n) e

M 3
2

n=0

h (n ) e

M 1

jn

n=0

h (n) e

jn

M +1
2

Change of variable in the second sum above:

n!M

n)

M+1
2

! M2

1 ! 0; and h (n) !

h (n)

Hence,
H e

M 3
2

h (n) e

M 3
2

h (n) e

jn

n=0

j ( M2 1 )

j(M 1 n)

n=0

M 3
2

h (n )

jn+ j( M2 1 )

j(M 1 n)+ j( M2 1 )

n=0

j ( M2 1 )

M 3
2

h (n )

e+ j(

M 1
2

n)

j( M2 1 n)

n=0

j ( M2 1 )

je

M 3
2

2h(n) sin



M
2

n!n)n=0!n=


and
sin

1
2

H e

je

j( M2 1 )

1
2

 

Hence,
j

M 1
2

n=1

 

n=0

Change of variable:

cos

M
2

fcos 0g
2

cos f2 cos 1g
2

cos fcos 0 cos + cos2g

2

cos

2h

n=

3
2

!n=1

! sin (n)
M

1
2

n sin (n)

1) in the

84

S OLUTIONS M ANUAL

Define c(n) = 2h

M 1
2

FOR

DSP

USING

M ATLAB

A PRIL 98

n . Then,

H e

je

M 1
2

M 1
2

n =1

n=1

j( M2 1 )

c (n) sin (n) ) Hr () = c (n) sin (n)

(b) Now sin (n) can be written as a linear combination of higher harmonics in cos multiplied by sin , i.e.,
sin ()
sin (2)
sin (3)

=
=
=

sin fcos0g
sin f2 cosg
sin fcos0 + 2 cos2g

etc. Note that the lowest harmonic frequency is zero and the highest harmonic frequency is (n
expansion. Hence,
Hr () =

M 1
2

M 3
2

n=1

c (n) sin (n) = sin c (n) cos (n)

where c (n) are related to c (n) through the above trignometric identities.
Problem P 7.5
M ATLAB function for amplitude response:
function [Hr,w,P,L] = ampl_res(h);
%
% function [Hr,w,P,L] = ampl_res(h)
%
% Computes Amplitude response Hr(w) and its polynomial P of order L,
%
given a linear-phase FIR filter impulse response h.
% The type of filter is determined automatically by the subroutine.
%
% Hr = Amplitude Response
% w = frequencies between [0 pi] over which Hr is computed
% P = Polynomial coefficients
% L = Order of P
% h = Linear Phase filter impulse response
%
M = length(h);
L = floor(M/2);
if fix(abs(h(1:1:L))*1010) = fix(abs(h(M:-1:M-L+1))*1010)
error(Not a linear-phase impulse response)
end
if 2*L = M
if fix(h(1:1:L)*1010) == fix(h(M:-1:M-L+1)*1010)
disp(*** Type-1 Linear-Phase Filter ***)
[Hr,w,P,L] = hr_type1(h);
elseif fix(h(1:1:L)*1010) == -fix(h(M:-1:M-L+1)*1010)
disp(*** Type-3 Linear-Phase Filter ***)
h(L+1) = 0;
[Hr,w,P,L] = hr_type3(h);
end
else
if fix(h(1:1:L)*1010) == fix(h(M:-1:M-L+1)*1010)
disp(*** Type-2 Linear-Phase Filter ***)

1) in the sin n

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

85

[Hr,w,P,L] = hr_type2(h);
elseif fix(h(1:1:L)*1010) == -fix(h(M:-1:M-L+1)*1010)
disp(*** Type-4 Linear-Phase Filter ***)
[Hr,w,P,L] = hr_type4(h);
end
end
M ATLAB verification:
clear; close all;
%% Matlab verification
h1 = [1 2 3 2 1]; [Hr1,w,P1,L1] = ampl_res(h1);
*** Type-1 Linear-Phase Filter ***
P1, L1,
P1 =
3
4
2
L1 =
2
%
h2 = [1 2 2 1]; [Hr2,w,P2,L2] = ampl_res(h2);
*** Type-2 Linear-Phase Filter ***
P2, L2,
P2 =
4
2
L2 =
2
%
h3 = [1 2 0 -2 -1]; [Hr3,w,P3,L3] = ampl_res(h3);
*** Type-3 Linear-Phase Filter ***
P3, L3,
P3 =
0
4
2
L3 =
2
%
h4 = [1 2 -2 -1]; [Hr4,w,P4,L4] = ampl_res(h4);
*** Type-4 Linear-Phase Filter ***
P4, L4,
P4 =
4
2
L4 =
2
%
%% Amplitude response plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.5);
subplot(2,2,1);plot(w/pi,Hr1);title(Type-1 FIR Filter);
ylabel(Hr(w));
set(gca,XTickMode,manual,XTick,[0:0.2:1],fontsize,10);
subplot(2,2,2);plot(w/pi,Hr2);title(Type-2 FIR Filter);
ylabel(Hr(w));
set(gca,XTickMode,manual,XTick,[0:0.2:1],fontsize,10);

86

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

subplot(2,2,3);plot(w/pi,Hr3);title(Type-3 FIR Filter);

ylabel(Hr(w)); xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[0:0.2:1],fontsize,10);
subplot(2,2,4);plot(w/pi,Hr4);title(Type-4 FIR Filter);
ylabel(Hr(w)); xlabel(frequency in pi units);
set(gca,XTickMode,manual,XTick,[0:0.2:1],fontsize,10);
% Super Title
suptitle(Problem P7.5);
The plots are shown in Figure 7.1.
Problem P7.5
Type1 FIR Filter

Type2 FIR Filter

10

Hr(w)

Hr(w)

6
4

2
0

2
0

0.2

0.4

0.6

0.8

0.2

Type3 FIR Filter

5

Hr(w)

Hr(w)

0.6

0.8

0.8

Type4 FIR Filter

3
2

3
2
1

1
0

0.4

0.2

0.4

0.6

0.8

frequency in pi units

0.2

0.4

0.6

frequency in pi units

Figure 7.1: Amplitude Response Plots in Problem P 7.5

Problem P 7.6
The filter H (z) has the following four zeros
z1 = re j ;

1
z2 = e j ;
r

z2 = re

1
z2 = e
r

The system function can be written as

H (z)

=
=
=
=

1 z1z 1
1 z2 z 1 1
z3z 1 1 z4
z 1

1 j 1
1
j 1 1
j z 1 1
j z 1
r e z 1 re
re
1 re z



1 (2r cos) z 1 +
r2 z 2 1
2r 1 cos
z 1 + r 2 z 2


1 2 cos r + r 1 z 1 + r2 + r 2 + 4 cos2 z 2 2 cos r + r

Hence the impulse response of the filter is



h (n) = 1; 2 cos r + r

"

r2 + r

2
+ 4 cos

2 cos r + r


;

+z 4

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

87

which is a finite-duration symmetric impulse response. This implies that the filter is a linear-phase FIR filter.
Problem P 7.7
The bandstop filter specifications are:
lower passband edge:
lower stopband edge:
upper stopband edge:
upper passband edge:
passband ripple:
stopband attenuation:

0:3
0:4
0:6
0:7
0:5 dB
40 dB

Hanning window design using M ATLAB:

clear; close all;
%% Specifications:
wp1 = 0.3*pi; % lower passband edge
ws1 = 0.4*pi; % lower stopband edge
ws2 = 0.6*pi; % upper stopband edge
wp2 = 0.7*pi; % upper passband edge
Rp = 0.5;
% passband ripple
As = 40;
% stopband attenuation
%
% Select the min(delta1,delta2) since delta1=delta2 in windodow design
[delta1,delta2] = db2delta(Rp,As);
if (delta1 < delta2)
delta2 = delta1; disp(Delta1 is smaller than delta2)
[Rp,As] = delta2db(delta1,delta2)
end
%
tr_width = min((ws1-wp1),(wp2-ws2));
M = ceil(6.2*pi/tr_width); M = 2*floor(M/2)+1, % choose odd M
M =
63
n = 0:M-1;
w_han = (hanning(M));
wc1 = (ws1+wp1)/2; wc2 = (ws2+wp2)/2;
hd = ideal_lp(pi,M)+ideal_lp(wc1,M)-ideal_lp(wc2,M);
h = hd .* w_han;
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db((ws1/delta_w)+1:(ws2/delta_w)+1))),
% Actual Attn
Asd =
43
Rpd = -min(db(1:(wp1/delta_w)+1)),
% Actual passband ripple
Rpd =
0.0884
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.7);
subplot(2,2,1); stem(n,hd); title(Ideal Impulse Response);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(hd(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)

88

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

subplot(2,2,2); stem(n,w_han); title(Hanning Window);

axis([-1,M,-0.1,1.1]); xlabel(n); ylabel(w_ham(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
set(gca,YTickMode,manual,YTick,[0;1],fontsize,10)
subplot(2,2,3); stem(n,h); title(Actual Impulse Response);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,4); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,[0;0.3;0.4;0.6;0.7;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0 ;0.3;0.4;0.6;0.7; 1 ],...
fontsize,10)
set(gca,YTickMode,manual,YTick,[-40;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 40; 0 ]);grid
% Super Title
suptitle(Problem P7.7: Bandstop Filter);
The filter design plots are shown in Figure 7.2.
Problem P7.7: Bandstop Filter
Ideal Impulse Response

Hanning Window

0.8
1

w_ham(n)

hd(n)

0.6
0.4
0.2
0
0

0.2
0

62

62

Actual Impulse Response

Magnitude Response in dB

0.8

Decibels

h(n)

0.6
0.4
0.2

40

0
0.2
0

62

0.3 0.4

0.6 0.7

frequency in pi units

Figure 7.2: Bandstop filter design using Hanning Window in Problem P 7.7
Problem P 7.8
The bandpass filter specifications are:
lower stopband edge:
lower passband edge:
upper passband edge:
upper stopband edge:
passband ripple:
stopband attenuation:

0:3
0:4
0:5
0:6
0:5 dB
50 dB

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

89

Hamming window design using M ATLAB:

clear; close all;
%% Specifications:
ws1 = 0.3*pi; % lower stopband edge
wp1 = 0.4*pi; % lower passband edge
wp2 = 0.5*pi; % upper passband edge
ws2 = 0.6*pi; % upper stopband edge
Rp = 0.5;
% passband ripple
As = 50;
% stopband attenuation
%
% Select the min(delta1,delta2) since delta1=delta2 in windodow design
[delta1,delta2] = db2delta(Rp,As);
if (delta1 < delta2)
delta2 = delta1; disp(Delta1 is smaller than delta2)
[Rp,As] = delta2db(delta1,delta2)
end
%
tr_width = min((wp1-ws1),(ws2-wp2));
M = ceil(6.6*pi/tr_width); M = 2*floor(M/2)+1, % choose odd M
M =
67
n = 0:M-1;
w_ham = (hamming(M));
wc1 = (ws1+wp1)/2; wc2 = (ws2+wp2)/2;
hd = ideal_lp(wc2,M)-ideal_lp(wc1,M);
h = hd .* w_ham;
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)+1]))),
% Actual Attn
Asd =
51
Rpd = -min(db(ceil(wp1/delta_w)+1:floor(wp2/delta_w)+1)), % Actual passband ripple
Rpd =
0.0488
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.8);
subplot(2,2,1); stem(n,hd); title(Ideal Impulse Response: Bandpass);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(hd(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,2); stem(n,w_ham); title(Hamming Window);
axis([-1,M,-0.1,1.1]); xlabel(n); ylabel(w_ham(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
set(gca,YTickMode,manual,YTick,[0;1],fontsize,10)
subplot(2,2,3); stem(n,h); title(Actual Impulse Response: Bandpass);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,4); plot(w/pi,db); title(Magnitude Response in dB);

90

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)

set(gca,XTickMode,manual,XTick,[0;0.3;0.4;0.5;0.6;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0 ;0.3;0.4;0.5;0.6; 1 ],...
fontsize,10)
set(gca,YTickMode,manual,YTick,[-50;0])
set(gca,YTickLabelMode,manual,YTickLabels,[-50; 0 ]);grid
The filter design plots are shown in Figure 7.3.
Homework4 : Problem 3
Ideal Impulse Response: Bandpass

Hamming Window

0.3
1

w_ham(n)

0.2

hd(n)

0.1
0
0.1
0.2

0
0

66

66

Actual Impulse Response: Bandpass

Magnitude Response in dB

0.3

0.2

Decibels

h(n)

0.1
0
0.1

50

0.2
0

66

0.3 0.4 0.5 0.6

frequency in pi units

Figure 7.3: Bandpass filter design using Hamming Window in Problem P 7.8
3. (Problem P 7.9) The highpass filter specifications are:
stopband edge:
passband edge:
passband ripple:
stopband attenuation:

0:4
0:6
0:5 dB
60 dB

Kaiser window design using M ATLAB:

clear; close all;
%% Specifications:
ws = 0.4*pi; % stopband edge
wp = 0.6*pi; % passband edge
Rp = 0.5;
% passband ripple
As = 60;
% stopband attenuation
%
% Select the min(delta1,delta2) since delta1=delta2 in windodow design
[delta1,delta2] = db2delta(Rp,As);
if (delta1 < delta2)
delta2 = delta1; disp(Delta1 is smaller than delta2)

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

91

[Rp,As] = delta2db(delta1,delta2)
end
%
tr_width = wp-ws; M = ceil((As-7.95)/(14.36*tr_width/(2*pi))+1)+1;
M = 2*floor(M/2)+3, % choose odd M, Increased order by 2 to get Asd>60
M =
41
n = [0:1:M-1]; beta = 0.1102*(As-8.7);
w_kai = (kaiser(M,beta));
% Kaiser Window
wc = (ws+wp)/2; hd = ideal_lp(pi,M)-ideal_lp(wc,M); % Ideal HP Filter
h = hd .* w_kai;
% Window design
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = -floor(max(db(1:1:(ws/delta_w)+1))),
% Actual Attn
Asd =
61
Rpd = -min(db((wp/delta_w)+1:1:501)),
% Actual passband ripple
Rpd =
0.0148
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.9);
subplot(2,2,1); stem(n,hd); title(Ideal Impulse Response);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(hd(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,2); stem(n,w_kai); title(Kaiser Window);
axis([-1,M,-0.1,1.1]); xlabel(n); ylabel(w_kai(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
set(gca,YTickMode,manual,YTick,[0;1],fontsize,10)
subplot(2,2,3); stem(n,h); title(Actual Impulse Response);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,4); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,[0;0.4;0.6;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0 ;0.4;0.6; 1 ],...
fontsize,10)
set(gca,YTickMode,manual,YTick,[-60;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 60; 0 ]);grid
% Super Title
suptitle(Problem P7.9: Highpass Filter Design);

The filter design plots are shown in Figure 7.4.

Problem P 7.10

92

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Problem P7.9: Highpass Filter Design

Ideal Impulse Response

Kaiser Window

0.6
1

w_kai(n)

hd(n)

0.4
0.2
0
0.2
0

0.4
0

40

40

Actual Impulse Response

Magnitude Response in dB

0.6

Decibels

h(n)

0.4
0.2
0

60

0.2
0.4
0

40

0.4

0.6

frequency in pi units

Figure 7.4: Highpass Filter design using Kaiser window in Problem P 7.9.
The bandpass filter specifications are:
lower stopband edge:
lower passband edge:
upper passband edge:
upper stopband edge:
passband tolerance:
stopband tolerance:

0:25
0:35
0:65
0:75
0:05
0:01

Kaiser window design using M ATLAB:

clear; close all;
%% Specifications:
ws1 = 0.25*pi; % lower stopband edge
wp1 = 0.35*pi; % lower passband edge
wp2 = 0.65*pi; % upper passband edge
ws2 = 0.75*pi; % upper stopband edge
delta1 = 0.05; % passband ripple magnitude
delta2 = 0.01; % stopband attenuation magnitude
%
%% Determination of Rp and As in dB
[Rp,As] = delta2db(delta1,delta2)
Rp =
0.8693
As =
40.4238
%
%% Determination of Window Parameters
tr_width = min((wp1-ws1),(ws2-wp2));
M = ceil((As-7.95)/(14.36*tr_width/(2*pi))+1)+1;

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

93

M = 2*floor(M/2)+1, % Odd filter length

M =
49
n=[0:1:M-1];
if As >= 50
beta = 0.1102*(As-8.7);
elseif (As < 50) & (As > 21)
beta = 0.5842*(As-21)(0.4) + 0.07886*(As-21);
else
error(As must be greater than 21)
end
w_kai = (kaiser(M,beta));
wc1 = (ws1+wp1)/2; wc2 = (ws2+wp2)/2;
hd = ideal_lp(wc2,M)-ideal_lp(wc1,M);
h = hd .* w_kai;
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)+1]))),
% Actual Attn
Asd =
42
Rpd = -min(db(ceil(wp1/delta_w)+1:floor(wp2/delta_w)+1)), % Actual passband ripple
Rpd =
0.1097
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.10);
subplot(2,2,1); stem(n,hd); title(Ideal Impulse Response: Bandpass);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(hd(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,2); stem(n,w_kai); title(Kaiser Window);
axis([-1,M,-0.1,1.1]); xlabel(n); ylabel(w_ham(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
set(gca,YTickMode,manual,YTick,[0;1],fontsize,10)
subplot(2,2,3); stem(n,h); title(Actual Impulse Response: Bandpass);
axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,4); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,[0;0.25;0.35;0.65;0.75;1])
set(gca,XTickLabelMode,manual,XTickLabels,...
[ 0 ;0.25;0.35;0.65;0.75; 1 ], fontsize,10)
set(gca,YTickMode,manual,YTick,[-40;0])
set(gca,YTickLabelMode,manual,YTickLabels,[-40; 0 ]);grid
The filter design plots are shown in Figure 7.5.
4. (Problem P 7.11b) The kai hpf function:
function [h,M] = kai_hpf(ws,wp,As);
% [h,M] = kai_hpf(ws,wp,As);

94

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Homework4 : Problem 4
Ideal Impulse Response: Bandpass

Kaiser Window
1

w_ham(n)

0.4

hd(n)

0.2
0
0.2

0
0.4

48

48

Actual Impulse Response: Bandpass

Magnitude Response in dB
0

0.4

Decibels

h(n)

0.2
0

40

0.2
0.4

48

0.250.35

0.650.75

frequency in pi units

Figure 7.5: Bandpass filter design using Kaiser Window in Problem P 7.10

%
%
%
%
%
%
%

HighPass FIR filter design using Kaiser window

h
M
ws
wp
As

=
=
=
=
=

Impulse response of the designed filter

Length of h which is an odd number
Stopband edge in radians (0 < wp < ws < pi)
Passband edge in radians (0 < wp < ws < pi)
Stopband attenuation in dB (As > 0)

if ws <= 0
error(Stopband edge must be larger than 0)
end
if wp >= pi
error(Passband edge must be smaller than pi)
end
if wp <= ws
error(Passband edge must be larger than Stopband edge)
end
% Select the min(delta1,delta2) since delta1=delta2 in windodow design
tr_width = wp-ws; M = ceil((As-7.95)/(14.36*tr_width/(2*pi))+1)+1;
M = 2*floor(M/2)+1;
% choose odd M
if M > 255
error(M is larger than 255)
end
n = [0:1:M-1]; beta = 0.1102*(As-8.7);
w_kai = (kaiser(M,beta));
% Kaiser Window
wc = (ws+wp)/2; hd = ideal_lp(pi,M)-ideal_lp(wc,M); % Ideal HP Filter
h = hd .* w_kai;
% Window design

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

95

M ATLAB verification:
clear; close all;
%% Specifications:
ws = 0.4*pi; % stopband edge
wp = 0.6*pi; % passband edge
As = 60;
% stopband attenuation
%
[h,M] = kai_hpf(ws,wp,As); n = 0:M-1;
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = -floor(max(db(1:1:(ws/delta_w)+1))),
% Actual Attn
Asd =
60
Rpd = -min(db((wp/delta_w)+1:1:501)),
% Actual passband ripple
Rpd =
0.0147
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.11b);
subplot(2,1,1); stem(n,h); title(Actual Impulse Response);
axis([-1,M,min(h)-0.1,max(h)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,1,2); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,[0;0.4;0.6;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0 ;0.4;0.6; 1 ],...
fontsize,10)
set(gca,YTickMode,manual,YTick,[-60;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 60; 0 ]);grid
% Super Title
suptitle(Problem P7.11b: Highpass Filter Design);
The filter design plots are shown in Figure 7.6.
Problem P 7.12
The staircase filter specifications are:
Band-1:
Band-2:
Band-3:

0
0:4
0:8









0:3
0:7

Ideal Gain
Ideal Gain
Ideal Gain

=
=
=

1
0:5
0

Blackman window design using M ATLAB:

clear; close all;
%% Specifications:
w1L = 0.0*pi; w1U =
w2L = 0.4*pi; w2U =
w3L = 0.8*pi; w3U =
%
%% Determination of

0.3*pi; delta1 = 0.010; % Band-1 Specs

0.7*pi; delta2 = 0.005; % Band-2 Specs
1.0*pi; delta3 = 0.001; % Band-3 Specs
Rp and As in dB

1
2
3

=
=
=

0:010
0:005
0:001

96

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Problem P7.11b: Highpass Filter Design

Actual Impulse Response
0.6
0.4

h(n)

0.2
0
0.2
0.4
0

38

n
Magnitude Response in dB

Decibels

60

0.4

0.6

frequency in pi units

Figure 7.6: Highpass filter design using the ntexttt fkai hpfg function in Problem 7.11b.

As = -20*log10(delta3)
As =
60.0000
%
%% Determination of Window Parameters
tr_width = min((w2L-w1U),(w3L-w2U));
M = ceil(11*pi/tr_width); M = 2*floor(M/2)+1, % choose odd M
M =
111
n=[0:1:M-1];
w_bla = (blackman(M));
wc1 = (w1U+w2L)/2; wc2 = (w2U+w3L)/2;
hd = 0.5*ideal_lp(wc1,M) + 0.5*ideal_lp(wc2,M);
h = hd .* w_bla;
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([floor(w3L/delta_w)+1:501]))),
% Actual Attn
Asd =
79
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.12);
subplot(2,2,1); stem(n,w_bla); title(Blackman Window);
axis([-1,M,-0.1,1.1]); xlabel(n); ylabel(w_ham(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
set(gca,YTickMode,manual,YTick,[0;1],fontsize,10)

A PRIL 98

S OLUTIONS M ANUAL

DSP

FOR

USING

M ATLAB

97

subplot(2,2,2); stem(n,h); title(Actual Impulse Response);

axis([-1,M,min(hd)-0.1,max(hd)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;M-1],fontsize,10)
subplot(2,2,3); plot(w/pi,mag); title(Magnitude Response);
axis([0,1,0,1]); xlabel(frequency in pi units); ylabel(|H|)
set(gca,XTickMode,manual,XTick,[0;0.3;0.4;0.7;0.8;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0;0.3;0.4;0.7;0.8;
fontsize,10)
set(gca,YTickMode,manual,YTick,[0;0.5;1])
set(gca,YTickLabelMode,manual,YTickLabels,[ 0 ;0.5; 1 ]);grid

1],

subplot(2,2,4); plot(w/pi,db); title(Magnitude Response in dB);

axis([0,1,-As-30,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,[0;0.3;0.4;0.7;0.8;1])
set(gca,XTickLabelMode,manual,XTickLabels,[ 0;0.3;0.4;0.7;0.8;
fontsize,10)
set(gca,YTickMode,manual,YTick,[-60;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 60; 0 ]);grid

1],

% Super Title
suptitle(Problem P7.12: Staircase Filter);
The filter design plots are shown in Figure 7.7.
Problem P7.12: Staircase Filter
Blackman Window

Actual Impulse Response

0.6

h(n)

w_ham(n)

0.4
0.2
0
0
0

110

110

Magnitude Response

Magnitude Response in dB
0

Decibels

|H|

0.5

0.3 0.4

0.7 0.8

frequency in pi units

60

0.3 0.4

0.7 0.8

frequency in pi units

Figure 7.7: Staircase filter design using Blackman Window in Problem P 7.12
Problem P 7.18
Bandpass filter design using Parks-McClellan Algorithm

98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

(a) Since we are given M = 25, As = 50 dB, and the ideal cutoff frequencies, c1 = =3 and c2 = 2=3, we will
have to determine the band edge frequencies, s1 , p1 , p2 , and S2 , to implement the Parks-McClellan algorithm.
Using the formula due to Kaiser we can determine the transition width as follows
M

1'

As 13
14:36 (=2)

) ' 2 14 A36s (M13 1)

:

Since no additional information is given, we will assume that the transition bandwidths are equal and that the
tolerances are also equal in each band, i.e.,
s 1
and

= c1

; p
1
2

= c1 +

; p
2
2

= c2

; s
2
2

= c2 +

1 = 2 = 3

Using this we will run the remez algorithm and check for the stopband attenuation of 50 dB. If the actual attenuation is less than (more than) 50 then we will increase (decrease) until the attenuation condition is met. In the
following M ATLAB script, the conditionw were met at the above computed value.
clear; close all;
%% Specifications
wc1 = pi/3;
% lower cutoff frequency
wc2 = 2*pi/3; % upper cuoff frequency
As = 50;
% stopband attenuation
M = 25;
% filter length
%
% (a) Design
tr_width = 2*pi*(As-13)/(14.36*(M-1)), % transition width in radians
tr_width =
0.6746
ws1 = wc1-tr_width/2; wp1 = wc1+tr_width/2;
wp2 = wc2-tr_width/2; ws2 = wc2+tr_width/2;
f = [0,ws1/pi,wp1/pi,wp2/pi,ws2/pi,1];
m = [0,0,1,1,0,0];
n = 0:M-1;
h = remez(M-1,f,m);
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)+1]))), % Actual Attn
Asd =
50
Rpd = -min(db(ceil(wp1/delta_w)+1:floor(wp2/delta_w)+1)), % Actial ripple
Rpd =
0.0518
%
%% Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.18);
subplot(2,1,1); stem(n,h); title(Impulse Response: Bandpass);
axis([-1,M,min(h)-0.1,max(h)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;12;24],fontsize,10)
subplot(2,1,2); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-80,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,f,fontsize,10)

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

99

set(gca,YTickMode,manual,YTick,[-50;0])
set(gca,YTickLabelMode,manual,YTickLabels,[-50; 0 ],fontsize,10);grid
The impulse response plot is shown in Figure 7.8.
Homework4 : Problem 5a
Impulse Response: Bandpass
0.4

h(n)

0.2

0
0.2

12

24

n
Magnitude Response in dB

Decibels

50

0.226

0.4407

0.5593

0.774

frequency in pi units

Figure 7.8: Impulse Response Plot in Problem P 7.18

(b) Amplitude Response:
% (b) Amplitude Response Plot
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P7.18b);
[Hr,w,a,L] = Hr_type1(h);
plot(w/pi,Hr); title(Amplitude Response in Problem 5b)
xlabel(frequency in pi units); ylabel(Hr(w))
axis([0,1,-0.1,1.1])
set(gca,XTickMode,manual,XTick,f)
The amplitude response plot is shown in Figure 7.9.
Problem P 7.19
Bandstop filter design of Problem P7.7 using Parks-McClellan Algorithm
(a) M ATLAB script:
clear; close all;
%% Specifications
wp1 = 0.3*pi; % lower passband edge
ws1 = 0.4*pi; % lower stopband edge
ws2 = 0.6*pi; % upper stopband edge
wp2 = 0.7*pi; % upper passband edge
Rp = 0.5;
% passband ripple

100

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Amplitude Response in Problem 5b

0.8

Hr(w)

0.6

0.4

0.2

0
0

0.226

0.4407 0.5593
frequency in pi units

0.774

Figure 7.9: Amplitude Response Plot in Problem P 7.18

As = 40;
% stopband attenuation
%
% (a) Design
delta1 = (10(Rp/20)-1)/(10(Rp/20)+1);
delta2 = (1+delta1)*(10(-As/20));
weights = [delta2/delta1, 1, delta2/delta1];
delta_f =min((wp2-ws2)/(2*pi), (ws1-wp1)/(2*pi));
M = ceil((-20*log10(sqrt(delta1*delta2))-13)/(14.6*delta_f)+1);
M = 2*floor(M/2)+1
M =
33
f = [0, wp1/pi, ws1/pi, ws2/pi, wp2/pi, 1];
m = [1 1 0 0 1 1];
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,[1]);
delta_w = pi/500;
Asd = floor(-max(db([floor(ws1/delta_w)+1:floor(ws2/delta_w)]))), % Actual Attn
Asd =
40
M = M+2
M =
35
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,[1]);
delta_w = pi/500;
Asd = floor(-max(db([floor(ws1/delta_w)+1:floor(ws2/delta_w)]))), % Actual Attn
Asd =
40

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

101

n = 0:M-1;
%
%%Filter Response Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.19a);
subplot(2,1,1); stem(n,h); title(Impulse Response: Bandpass);
axis([-1,M,min(h)-0.1,max(h)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;17;34],fontsize,10)
subplot(2,1,2); plot(w/pi,db); title(Magnitude Response in dB);
axis([0,1,-60,5]); xlabel(frequency in pi units); ylabel(Decibels)
set(gca,XTickMode,manual,XTick,f,fontsize,10)
set(gca,YTickMode,manual,YTick,[-40;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 40; 0 ],fontsize,10);grid
% Super Title
suptitle(Problem P7.19a: Bandstop Filter);
The filter response plots are shown in Figure 7.10.
Problem P7.19a: Bandstop Filter Design
Impulse Response: Bandpass
0.6

h(n)

0.4
0.2
0
0.2
0

17

34

n
Magnitude Response in dB

Decibels

40

0.3

0.4

0.6

0.7

frequency in pi units

Figure 7.10: Bandstop Filter Response Plots in Problem P7.19a

(b) Amplitude Response:
% (b) Amplitude Response Plot
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P7.19b);
[Hr,w,a,L] = Hr_type1(h);
plot(w/pi,Hr); title(Amplitude Response in Problem P7.19b)
xlabel(frequency in pi units); ylabel(Hr(w))

102

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

axis([0,1,-0.1,1.1])
set(gca,XTickMode,manual,XTick,f)
The amplitude response plot is shown in Figure 7.11.
Amplitude Response in Problem P7.19b

0.8

Hr(w)

0.6

0.4

0.2

0
0

0.3

0.4
0.6
frequency in pi units

0.7

Figure 7.11: Amplitude Response Plot in Problem P 7.19b

Problem P 7.21 Multiband filter design using Parks-McClellan Algorithm

The filter specifications are:
Lower band-edge
Upper band-edge
Ideal value
Tolerance

Band-1
0
0:4
0:4
0:05

Band-2 Band-3
0 :8
0:5
0:7
1
0:95
0:025
0:05
0:025

M ATLAB Script:
clear; close all;
%% Specifications
f = [0,0.4,0.5,0.7,0.8,1];
m = [0.4,0.4,0.95,0.95,0.025,0.025];
delta1 = 0.05; delta2 = 0.05; delta3 = 0.025;
weights = [delta3/delta2, delta3/delta2, delta3/delta3];
As = -20*log10(0.05)
As =
26.0206
%
%% Design
delta_f = 0.05; % Transition width in cycles per sample
M = ceil((-20*log10(sqrt(delta2*delta3))-13)/(14.6*delta_f)+1)
M =

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

103

23
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([(0.8*pi/delta_w)+1:501]))), % Actual Attn
Asd =
24
M = M+1
M =
24
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,1);
Asd = floor(-max(db([(0.8*pi/delta_w)+1:501]))), % Actual Attn
Asd =
25
M = M+1
M =
25
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,1);
Asd = floor(-max(db([(0.8*pi/delta_w)+1:501]))), % Actual Attn
Asd =
25
M = M+1
M =
26
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,1);
Asd = floor(-max(db([(0.8*pi/delta_w)+1:501]))), % Actual Attn
Asd =
25
M = M+1
M =
27
h = remez(M-1,f,m,weights);
[db,mag,pha,grd,w] = freqz_m(h,1);
Asd = floor(-max(db([(0.8*pi/delta_w)+1:501]))), % Actual Attn
Asd =
26
n = 0:M-1;
%
%% Impulse Response Plot
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.21a);
stem(n,h); title(Impulse Response Plot in Problem 6);
axis([-1,M,min(h)-0.1,max(h)+0.1]); xlabel(n); ylabel(h(n))
set(gca,XTickMode,manual,XTick,[0;13;26],fontsize,12)
%
%% Amplitude Response Plot
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P7.21b);
[Hr,w,a,L] = Hr_type1(h);

104

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

plot(w/pi,Hr); title(Amplitude Response in Problem 6)

xlabel(frequency in pi units); ylabel(Hr(w))
axis([0,1,0,1])
set(gca,XTickMode,manual,XTick,f)
set(gca,YTickMode,manual,YTick,[0;0.05;0.35;0.45;0.9;1]); grid

Impulse Response Plot in Problem 6

0.5

0.4

0.3

h(n)

0.2

0.1

0.1

0.2
0

13
n

26

Figure 7.12: Impulse Response Plot in Problem P 7.21

The impulse response is shown in Figure 7.12 and the amplitude response plot is shown in Figure 7.13.
5. (Problem P 7.20) Design of a 25-tap differentiator with unit slope using Parks-McClellan algorithm
(a) M ATLAB script:
clear; close all;
% Specifications
M = 25; w1 = 0.1*pi; w2 = 0.9*pi; % slope = 1 sam/cycle
%
% (a) Design
f = [w1/pi w2/pi]; m = [w1/(2*pi) w2/(2*pi)];
h = remez(M-1,f,m,differentiator);
[db,mag,pha,grd,w] = freqz_m(h,1);
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.20a);
subplot(2,1,1);stem([0:1:M-1],h);title(Impulse Response); axis([-1,25,-0.2,.2]);
xlabel(n,fontsize,10); ylabel(h(n),fontsize,10);
set(gca,XTickMode,manual,XTick,[0;12;24],fontsize,10);
subplot(2,1,2);plot(w/(2*pi),mag);title(Magnitude Response);grid;
axis([0,0.5,0,0.5]);
xlabel(Normalized frequency in cycles/sam,fontsize,10)
set(gca,XTickMode,manual,XTick,[0;w1/(2*pi);w2/(2*pi);0.5],fontsize,10);
set(gca,YTickMode,manual,YTick,[0;0.05;0.45;0.5],fontsize,10);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

105

Amplitude Response in Problem 6

1

Hr(w)

0.9

0.45
0.35

0.05
0
0

0.4
0.5
frequency in pi units

0.7

0.8

Figure 7.13: Amplitude Response Plot in Problem P 7.21

The differentiator design plots are shown in Figure 7.14 from which we observe that this filter provides a slope
equal to 1 sam/cycle or =2 sam/rad.

(b) Output when x (n) = 3 sin (0:25n) ; 0  n  100: First determine the sign of h (n) and then appropriately convolve
with x (n).
% (b) Differentiator verification
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P7.20b);
[Hr,w,P,L] = ampl_res(h); subplot; plot(w/(2*pi), Hr);
*** Type-3 Linear-Phase Filter ***
title(Amplitude Response); grid; axis([0,0.5,-0.5,0]);
set(gca,XTickMode,manual,XTick,[0;w1/(2*pi);w2/(2*pi);0.5],fontsize,10);
set(gca,YTickMode,manual,YTick,[-0.5;-0.45;-0.05;0],fontsize,10);
The amplitude response plot is shown in Figure 7.15. The sign of h (n) from Figure 7.15 is negative, hence we will
convolve x (n) with h (n) and then compare the input and output in the steady-state (i.e. when n > 25) with output
shifted by 12 to the left to account for the phase delay of the differentiator.
Hf_3 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_3,NumberTitle,off,Name,P7.20c);
n=[0:1:100]; x = 3*sin(0.25*pi*n); y = conv(x,-h);
m = [41:1:81];
plot(m,x(41:1:81),m,y(41+12:1:81+12));grid % add 12 sample delay to y
xlabel(n); title(Input-Output Sequences); axis([40,82,-4,4]);
set(gca,XTickMode,manual,XTick,[41;81],fontsize,10);
set(gca,YTickMode,manual,YTick,[-3;0;3],fontsize,10);
The input-output plots are shown in Figure 7.16. Since the slope is =2 sam/rad, the gain at = 0:25 is equal to
0:125. Therefore, the output (when properly shifted) should be
y (n) = 3 (0:125) cos (0:25n) = 0:375 cos (0:25n)

106

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Impulse Response
0.2

h(n)

0.1

0.1

0.2

12
n

24

Magnitude Response
0.5
0.45

0.05
0

0.05

0.45

0.5

Normalized frequency in cycles/sam

Figure 7.14: Digital differentiator design using PM algorithm in Problem P 7.20a.

From the above figure we can verify that y (n)(the lower curve) is indeed a cosine waveform with amplitude  0:4.
Problem P 7.22
Narrow bandpass filter design using Parks-McClellan Algorithm
We want to design a 50th-order narrowband bandpass filter to filter out noise component with center frequency of c =
=2, bandwidth of 0:02, and stopband attenuation of 30 dB.
(a) In this design we already know the order of the filter. The only parameters that we dont know are the stopband
cutoff frequencies s1 and s2 . Let the transition bandwidth be and let the passband be symmetrical with respect
to the center frequency c . Then
p1

= c

0:01; p2

= c + 0:01;

s 1

= p1

; and s2

= p2 +

We will aslo assume that the tolerances in each band are equal. Now we will begin with initial value for = 0:2
and run the remez algorithm to obtain the actual stopband attenuation. If it is smaller (larger) than the given 30 dB
then we will increase (decrease) then iterate to obtain the desired solution. The desired solution was found for
= 0:5. M ATLAB Script:
clear; close all;
%% Specifications
N = 50;
% Order of the filter
w0 = 0.5*pi;
% Center frequency
Bandwidth = 0.02*pi;
% Bandwidth
%
%
Deltaw = Transition bandwidth (iteration variable)
%
wp1 = w0-Bandwidth/2; wp2 = w0+Bandwidth/2;
% (a) Design

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

107

Amplitude Response
0

0.05

0.45

0.5

0.05

0.45

0.5

Figure 7.15: Amplitude response in Problem P 7.20b.

Deltaw = 0.02*pi;
% Initial guess
ws1=wp1-Deltaw; ws2=wp2+Deltaw;
F=[0, ws1, wp1, wp2, ws2, pi]/pi;
m=[0,0,1,1,0,0];
h=remez(50,F,m);
[db,mag,pha,grd,w]=freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)]))), % Actual Attn
Asd =
13
% Next iteration
Deltaw = Deltaw+0.01*pi;
ws1=wp1-Deltaw; ws2=wp2+Deltaw;
F=[0, ws1, wp1, wp2, ws2, pi]/pi;
h=remez(50,F,m);
[db,mag,pha,grd,w]=freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)]))), % Actual Attn
Asd =
20
% Next iteration
Deltaw = Deltaw+0.01*pi;
ws1=wp1-Deltaw; ws2=wp2+Deltaw;
F=[0, ws1, wp1, wp2, ws2, pi]/pi;
h=remez(50,F,m);
[db,mag,pha,grd,w]=freqz_m(h,1);

108

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

InputOutput Sequences

41

81

Figure 7.16: Digital differentiator operation in Problem P 7.20b.

delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)]))), % Actual Attn
Asd =
26
% Next iteration
Deltaw = Deltaw+0.01*pi;
ws1=wp1-Deltaw; ws2=wp2+Deltaw;
F=[0, ws1, wp1, wp2, ws2, pi]/pi;
h=remez(50,F,m);
[db,mag,pha,grd,w]=freqz_m(h,1);
delta_w = pi/500;
Asd = floor(-max(db([1:floor(ws1/delta_w)]))), % Actual Attn
Asd =
30
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P7.22a);
plot(w/pi,db); axis([0,1,-50,0]); title(Log-Magnitude Response in P7.22a);
xlabel(frequency in pi units); ylabel(DECIBELS)
set(gca,XTickMode,manual,XTick,[0;ws1/pi;ws2/pi;1],fontsize,10)
set(gca,YTickMode,manual,YTick,[-30;0])
set(gca,YTickLabelMode,manual,YTickLabels,[ 30; 0 ],fontsize,10);grid
The log-magnitude response is shown in Figure 7.17.
(b) The time-domain response of the filter. M ATLAB script:
% (b) Time-domain Response
n = [0:1:200]; x = 2*cos(pi*n/2)+randn(1,201); y = filter(h,1,x);
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

109

LogMagnitude Response in P7.22a

DECIBELS

30

0.44

0.56

frequency in pi units

Figure 7.17: Log-Magnitude Response Plot in Problem P 7.22a

set(Hf_2,NumberTitle,off,Name,P7.22b);
subplot(211);stem(n(101:201),x(101:201));title(Input sequence x(n))
subplot(212);stem(n(101:201),y(101:201));title(Output sequence y(n))
The time-domain response is shown in Figure 7.18.

110

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Input sequence x(n)

4
2
0
2
4
100

110

120

130

140

150

160

170

180

190

200

170

180

190

200

Output sequence y(n)

4
2
0
2
4
100

110

120

130

140

150

160

Figure 7.18: The Time-domain Response in Problem P7.22b

Chapter 8

IIR Filter Design

1. Problem P 8.1
Analog Butterworth lowpass filter design: p = 30 rad/s, R p = 1 dB, s = 40 rad/s, As = 30 dB.
M ATLAB Script:
clear, close all;
% Filter Specifications
Wp = 30; Ws = 40; Rp = 1; As = 30;
% Filter Design
[b,a] = afd_butt(Wp,Ws,Rp,As); format short e
*** Butterworth Filter Order = 15
% Cascade Structure
[C,B,A] = sdir2cas(b,a)
C =
2.8199e+022
B =
0
0
1
A =
1.0000e+000 6.1393e+001 9.8484e+002
1.0000e+000 5.7338e+001 9.8484e+002
1.0000e+000 5.0777e+001 9.8484e+002
1.0000e+000 4.1997e+001 9.8484e+002
1.0000e+000 3.1382e+001 9.8484e+002
1.0000e+000 1.9395e+001 9.8484e+002
1.0000e+000 6.5606e+000 9.8484e+002
0 1.0000e+000 3.1382e+001
% Frequency Response
Wmax = 100; [db,mag,pha,w] = freqs_m(b,a,Wmax); pha = unwrap(pha);
% Impulse Response
%
The impulse response of the designed filter when computed by Matlab is numerically
%
unstable due to large coefficient values. Hence we compute the impulse response
%
of the filter with Wp/10 and Ws/10 band edges to keep coefficient values small.
%
The actual impulse response is time-scaled and amplitude scaled version of the
%
computed impulse response.
[b,a] = afd_butt(Wp/10,Ws/10,Rp,As); [ha,x,t] = impulse(b,a);
*** Butterworth Filter Order = 15
t = t/10; ha = ha/10;
%
111

112

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

% Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.1);
%
subplot(2,2,1);plot(w,mag); axis([0,Wmax,0,1.1]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(Magnitude,fontsize,10); title (Magnitude Response,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
magRp = round(10(-Rp/20)*100)/100;
set(gca,YTickMode,manual,Ytick,[0;magRp;1],fontsize,10);grid
%
subplot(2,2,2);plot(w,db); axis([0,Wmax,-100,0]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(Decibels,fontsize,10); title (Log-Magnitude Response,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
set(gca,YTickMode,manual,Ytick,[-100;-As;0],fontsize,10);grid
AS = [ ,num2str(As)];
set(gca,YTickLabelMode,manual,YTickLabels,[100;AS; 0]);
%
minpha = floor(min(pha/pi)); maxpha = ceil(max(pha/pi));
subplot(2,2,3);plot(w,pha/pi); axis([0,Wmax,minpha,maxpha]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(Phase in pi units,fontsize,10); title (Phase Response,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
phaWp = (round(pha(Wp/Wmax*500+1)/pi*100))/100;
phaWs = (round(pha(Ws/Wmax*500+1)/pi*100))/100;
set(gca,YTickMode,manual,Ytick,[phaWs;phaWp;0],fontsize,10); grid
%
subplot(2,2,4); plot(t,ha); title (Impulse Response,fontsize,10);
xlabel(t (sec), fontsize,10); ylabel(ha(t),fontsize,10);
%
suptitle(Analog Butterworth Lowpass Filter Design Plots in P 8.1)
The system function is given by
Ha (s) = 2:8199  1022


1
s2 + 61:393s
  + 984:84



1
s + 31:382


1
1

2 + 57:338s + 984:84
2 + 50:777s + 984:84
s
s



1
1

2 + 41:997s + 984:84
2 + 31:382s + 984:84
s
s



1
1
s2 + 19:395s + 984:84
s2 + 6:5606s + 984:84

The filter design plots are given in Figure 8.1.

2. Problem P 8.2
Analog Elliptic lowpass filter design: p = 10 rad/s, R p = 1 dB, s = 15 rad/s, As = 40 dB.
M ATLAB Script:
clear; close all;
% Filter Specifications
Wp = 10; Ws = 15; Rp = 1; As = 40;
% Filter Design

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

113

Analog Butterworth Lowpass Filter Design Plots in P 8.1

Magnitude Response

LogMagnitude Response
0

1
0.89
Decibels

Magnitude

30

30 40
Analog frequency in rad/sec

100

100

30 40
Analog frequency in rad/sec

Phase Response

Impulse Response

0.1

0.05
ha(t)

Phase in pi units

100

3.47

4.89

30 40
Analog frequency in rad/sec

100

0.05

0.5

1
t (sec)

1.5

Figure 8.1: Analog Butterworth Lowpass Filter Plots in Problem P 8.1

[b,a] = afd_elip(Wp,Ws,Rp,As); format short e;

*** Elliptic Filter Order = 5
% Rational Function Form
a0 = a(1); b = b/a0, a = a/a0
b =
4.6978e-001
0 2.2007e+002
0 2.2985e+004
a =
1.0000e+000 9.2339e+000 1.8471e+002 1.1292e+003 7.8813e+003 2.2985e+004
% Frequency Response
Wmax = 30; [db,mag,pha,w] = freqs_m(b,a,Wmax); pha = unwrap(pha);
% Impulse Response
[ha,x,t] = impulse(b,a);
%
% Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.2);
%
subplot(2,2,1);plot(w,mag); axis([0,Wmax,0,1]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(Magnitude,fontsize,10); title (Magnitude Response,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
magRp = round(10(-Rp/20)*100)/100;
set(gca,YTickMode,manual,Ytick,[0;magRp;1],fontsize,10);grid
%
subplot(2,2,2);plot(w,db); axis([0,Wmax,-100,0]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(log-Magnitude in dB,fontsize,10);

114

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

title (Log-Magnitude Response,fontsize,10);

set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
set(gca,YTickMode,manual,Ytick,[-100;-As;0],fontsize,10);grid
AS = [ ,num2str(As)];
set(gca,YTickLabelMode,manual,YTickLabels,[100;AS; 0],fontsize,10);
%
minpha = floor(min(pha/pi)); maxpha = ceil(max(pha/pi));
subplot(2,2,3);plot(w,pha/pi); axis([0,Wmax,minpha,maxpha]);
xlabel(Analog frequency in rad/sec,fontsize,10);
ylabel(Phase in pi units,fontsize,10);
title (Phase Response,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;Wp;Ws;Wmax],fontsize,10);
phaWp = (round(pha(Wp/Wmax*500+1)/pi*100))/100;
phaWs = (round(pha(Ws/Wmax*500+1)/pi*100))/100;
set(gca,YTickMode,manual,Ytick,[phaWs;phaWp;0],fontsize,10); grid
%
subplot(2,2,4); plot(t,ha); title (Impulse Response,fontsize,10);
xlabel(t (sec), fontsize,10); ylabel(ha(t),fontsize,10);
%
suptitle(Analog Elliptic Lowpass Filter Design Plots in P 8.2)
The system function is given by
Ha (s) =

4
2
:46978s + 220:07s + 2298:5
5
4
3
s + 9:23s + 184:71s + 1129:2s2 + 7881:3s + 22985

The filter design plots are given in Figure 8.2.

Analog Elliptic Lowpass Filter Design Plots in P 8.2
Magnitude Response

LogMagnitude Response
0

Magnitude

logMagnitude in dB

1
0.89

10
15
Analog frequency in rad/sec

30

40

100

10
15
Analog frequency in rad/sec

Phase Response

Impulse Response

3
2

1.26

ha(t)

Phase in pi units

30

1.64

1
0
1

10
15
Analog frequency in rad/sec

30

t (sec)

Figure 8.2: Analog Elliptic Lowpass Filter Design Plots in P 8.2

3. Problem P 8.3

10

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

115

The filter passband must include the 100 Hz component while the stopband must include the 130 Hz component. To
obtain a minimum-order filter, the transition band must be as large as possible. This means that the passaband cutoff must
be at 100 Hz while the stopband cutoff must be at 130 Hz. Hence the analog Chebyshev-I lowpass filter specifications
are: p = 2 (100) rad/s, R p = 2 dB, s = 2 (130) rad/s, As = 50 dB.
M ATLAB Script:

clear, close all;

% Filter Specifications
Fp = 100; Fs = 130; Rp = 2; As = 50;
Wp = 2*pi*Fp; Ws = 2*pi*Fs;
% Filter Design
[b,a] = afd_chb1(Wp,Ws,Rp,As); format short e
*** Chebyshev-1 Filter Order = 9
% Cascade Structure
[C,B,A] = sdir2cas(b,a)
C =
7.7954e+022
B =
0
0
1
A =
1.0000e+000 1.4245e+002 5.1926e+004
1.0000e+000 1.1612e+002 1.6886e+005
1.0000e+000 7.5794e+001 3.0183e+005
1.0000e+000 2.6323e+001 3.8862e+005
0 1.0000e+000 7.5794e+001
% Frequency Response
Fmax = 200; Wmax = 2*pi*Fmax; [db,mag,pha,w] = freqs_m(b,a,Wmax); pha = unwrap(pha);
%
% Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.3);
%
subplot(2,1,1);plot(w/(2*pi),mag); axis([0,Fmax,0,1.1]); set(gca,fontsize,10);
xlabel(Analog frequency in Hz); ylabel(Magnitude);
title (Magnitude Response,fontsize,12);
set(gca,XTickMode,manual,Xtick,[0;Fp;Fs;Fmax]);
magRp = round(10(-Rp/20)*100)/100;
set(gca,YTickMode,manual,Ytick,[0;magRp;1]);grid
%
subplot(2,1,2);plot(w/(2*pi),db); axis([0,Fmax,-100,0]); set(gca,fontsize,10);
xlabel(Analog frequency in Hz); ylabel(Decibels);
title (Log-Magnitude Response,fontsize,12);
set(gca,XTickMode,manual,Xtick,[0;Fp;Fs;Fmax]);
set(gca,YTickMode,manual,Ytick,[-100;-As;0]);grid
AS = [ ,num2str(As)];
set(gca,YTickLabelMode,manual,YTickLabels,[100;AS; 0]);
%
suptitle(Analog Chebyshev-I Lowpass Filter Design Plots in P 8.3)

116

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

The system function is given by

Ha (s) = 7:7954  1022


1
2
s + 142:
45s
 + 51926

s2 + 75:794s + 301830



1
2
s + 116:12s
 + 168860 
1
1
2
s + 26:323s + 388620
s + 75:794

The magnitude response plots are given in Figure 8.3.

Analog ChebyshevI Lowpass Filter Design Plots in P 8.3
Magnitude Response
1

Magnitude

0.79

100
130
Analog frequency in Hz

200

LogMagnitude Response

Decibels

50

100

100
130
Analog frequency in Hz

200

Figure 8.3: Analog Chebyshev-I Lowpass Filter Plots in Problem P 8.3

4. Problem P 8.4
Analog Chebyshev-II lowpass filter design: p = 2 (250) rad/s, R p = 0:5 dB, s = 2 (300) rad/s, As = 45 dB.
M ATLAB Script:
clear; close all;
% Filter Specifications
Fp = 250; Fs = 300; Rp = 0.5; As = 45;
Wp = 2*pi*Fp; Ws = 2*pi*Fs;
% Filter Design
[b,a] = afd_chb2(Wp,Ws,Rp,As); format short e;
*** Chebyshev-2 Filter Order = 12
% Rational Function Form
a0 = a(1); b = b/a0, a = a/a0
b =
Columns 1 through 6
5.6234e-003
0 1.4386e+006
Columns 7 through 12
9.0401e+020
0 6.1946e+027
Column 13

5.9633e+013

1.9564e+034

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

117

2.3171e+040
a =
Columns 1 through 6
1.0000e+000 1.5853e+004 1.2566e+008 6.5800e+011 2.5371e+015 7.5982e+018
Columns 7 through 12
1.8208e+022 3.5290e+025 5.5639e+028 6.9823e+031 6.9204e+034 4.7962e+037
Column 13
2.3171e+040
% Frequency Response
Fmax = 500; Wmax = 2*pi*Fmax; [db,mag,pha,w] = freqs_m(b,a,Wmax); pha = unwrap(pha);
% Impulse Response
%
The impulse response of the designed filter when computed by Matlab is numerically
%
unstable due to large coefficient values. Hence we will compute the impulse respon
%
of the filter with Wp/1000 and Ws/1000 band edges to keep coefficient values small
%
The actual impulse response is time-scaled and amplitude scaled version of the
%
computed impulse response.
[b,a] = afd_chb2(Wp/1000,Ws/1000,Rp,As); [ha,x,t] = impulse(b,a);
*** Chebyshev-2 Filter Order = 12
t = t/1000; ha = ha/1000;
%
% Plots
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.4);
%
subplot(2,2,1);plot(w/(2*pi),mag); axis([0,Fmax,0,1]); set(gca,fontsize,10);
xlabel(Analog frequency in Hz); ylabel(Magnitude);
title (Magnitude Response,fontsize,12);
set(gca,XTickMode,manual,Xtick,[0;Fp;Fs;Fmax]);
magRp = round(10(-Rp/20)*100)/100;
set(gca,YTickMode,manual,Ytick,[0;magRp;1]);grid
%
subplot(2,2,2);plot(w/(2*pi),db); axis([0,Fmax,-100,0]); set(gca,fontsize,10);
xlabel(Analog frequency in Hz); ylabel(log-Magnitude in dB);
title (Log-Magnitude Response,fontsize,12);
set(gca,XTickMode,manual,Xtick,[0;Fp;Fs;Fmax]);
set(gca,YTickMode,manual,Ytick,[-100;-As;0]);grid
AS = [ ,num2str(As)];
set(gca,YTickLabelMode,manual,YTickLabels,[100;AS; 0]);
%
minpha = floor(min(pha/pi)); maxpha = ceil(max(pha/pi));
subplot(2,2,3);plot(w/(2*pi),pha/pi); axis([0,Fmax,minpha,maxpha]); set(gca,fontsize
xlabel(Analog frequency in Hz); ylabel(Phase in pi units);
title (Phase Response,fontsize,12);
set(gca,XTickMode,manual,Xtick,[0;Fp;Fs;Fmax]);
phaWp = (round(pha(Wp/Wmax*500+1)/pi*100))/100;
phaWs = (round(pha(Ws/Wmax*500+1)/pi*100))/100;
set(gca,YTickMode,manual,Ytick,[phaWs;phaWp;0]); grid
%
subplot(2,2,4); plot(t,ha); set(gca,fontsize,10);
title (Impulse Response,fontsize,12); xlabel(t (sec)); ylabel(ha(t));
%
suptitle(Analog Chebyshev-II Lowpass Filter Design Plots in P 8.4)

118

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

The filter design plots are given in Figure 8.4.

Digital Butterworth Filter Design Plots in P 8.7
LogMagnitude Response

Decibel

50
0

0.4

0.6

Frequency in Hz

Impulse Response

ha(t)

0.1

0.1

10

20

30

40

50
60
time in seconds

70

80

90

100

Figure 8.4: Analog Chebyshev-II Lowpass Filter Plots in Problem P 8.4

5. Problem P 8.5
M ATLAB function afd.m:
function [b,a] = afd(type,Fp,Fs,Rp,As)
%
% function [b,a] = afd(type,Fp,Fs,Rp,As)
%
Designs analog lowpass filters
% type = butter or cheby1 or cheby2 or ellip
%
Fp = passband cutoff in Hz
%
Fs = stopband cutoff in Hz
%
Rp = passband ripple in dB
%
As = stopband attenuation in dB
type = lower([type, ]); type = type(1:6);
twopi = 2*pi;
if type == butter
[b,a] = afd_butt(twopi*Fp,twopi*Fs,Rp,As);
elseif type == cheby1
[b,a] = afd_chb1(twopi*Fp,twopi*Fs,Rp,As);
elseif type == cheby2
[b,a] = afd_chb2(twopi*Fp,twopi*Fs,Rp,As);
elseif type == ellip
[b,a] = afd_elip(twopi*Fp,twopi*Fs,Rp,As);
else
error(Specify the correct type)
end
6. Problem P 8.6

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

Digital Chebyshev-1 Lowpass Filter Design using Impulse Invariance. M ATLAB script:
clear; close all; Twopi = 2*pi;
%% Analog Filter Specifications
Fsam = 8000;
Fp = 1500;
Rp = 3;
Fs = 2000;
As = 40;
%
%% Digital Filter Specifications
wp = Twopi*Fp/Fsam;
Rp = 3;
ws = Twopi*Fs/Fsam;
As = 40;
(a) Part (a): T

= 1.

%
%
%
%
%

Sampling
Passband
Passband
Stopband
Stopband

Rate in sam/sec
edge in Hz
Ripple in dB
edge in Hz
attenuation in dB

%
%
%
%

Passband
Passband
Stopband
Stopband

edge in rad/sam
Ripple in dB
edge in rad/sam
attenuation in dB

M ATLAB script:

%% (a) Impulse Invariance Digital Design using T = 1

T = 1;
OmegaP = wp/T;
% Analog Prototype Passband edge
OmegaS = ws/T;
% Analog Prototype Stopband edge
[cs,ds] = afd_chb1(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Chebyshev-1 Filter Order =
[b,a] = imp_invr(cs,ds,T);
[C,B,A] = dir2par(b,a),
C =
[]
B =
-0.0561
0.0558
0.1763
-0.1529
-0.2787
0.2359
0.1586
0
A =
1.0000
-0.7767
0.9358
1.0000
-1.0919
0.8304
1.0000
-1.5217
0.7645
1.0000
-0.8616
0

7
% II Transformation
% Parallel form

The filter design plots are shown in Figure 8.5.

(b) Part (b): T

= 1=8000.

M ATLAB script:

%% (b) Impulse Invariance Digital Design using T = 1/8000

T = 1/8000;
OmegaP = wp/T;
% Analog Prototype Passband edge
OmegaS = ws/T;
% Analog Prototype Stopband edge
[cs,ds] = afd_chb1(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Chebyshev-1 Filter Order =
[b,a] = imp_invr(cs,ds,T);
[C,B,A] = dir2par(b,a),
C =
[]
B =
1.0e+003 *

7
% II Transformation
% Parallel form

119

120

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Filter Design Plots in P 8.6a

LogMagnitude Response

Decibel

40

60

1500

2000
Frequency in Hz

4000

Impulse Response
0.3

h(n)

0.2
0.1
0
0.1
0.2

10

20

30

40

50
n

60

70

Figure 8.5: Impulse Invariance Design Method with T

-0.4487
1.4102
-2.2299
1.2684

0.4460
-1.2232
1.8869
0

1.0000
1.0000
1.0000
1.0000

-0.7767
-1.0919
-1.5217
-0.8616

80

=1

90

100

in Problem P 8.6a

A =
0.9358
0.8304
0.7645
0

The filter design plots are shown in Figure 8.6.

(c) Comparison: The designed system function as well as the impulse response in part 6b are similar to those in part 6a
except for an overall gain due to Fs = 1=T = 8000. This problem can be avoided if in the impulse invariance design
method we set
h (n) = T
ha (nT )
7. Problem P 8.7
Digital Butterworth Lowpass Filter Design using Impulse Invariance. MATLAB script:
clear; close all;
%% Digital Filter Specifications
wp = 0.4*pi;
% Passband edge in rad/sam
Rp = 0.5;
% Passband Ripple in dB
ws = 0.6*pi;
% Stopband edge in rad/sam
As = 50;
% Stopband attenuation in dB
%% Impulse Invariance Digital Design using T = 2
T = 2;
OmegaP = wp/T;
% Analog Prototype Passband edge
OmegaS = ws/T;
% Analog Prototype Stopband edge

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

121

Filter Design Plots in P 8.6b

LogMagnitude Response

Decibel

40

60
0

1500
2000
Frequency in Hz

4000

Impulse Response
3000

h(n)

2000
1000
0
1000
2000

10

20

30

40

50
n

60

Figure 8.6: Impulse Invariance Design Method with T

70

80

= 1=8000 in

90

100

Problem P 8.6b

[cs,ds] = afd_butt(OmegaP,OmegaS,Rp,As); % Analog Prototype Design

*** Butterworth Filter Order = 17
[b,a] = imp_invr(cs,ds,T);
% II Transformation: rational form
% Plots of Log-magnitude Response and Impulse Response
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.7);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-60,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;
% Impulse response of the prototype analog filter
Nmax = 50; t = 0:T/10:T*Nmax; [ha,x,t] = impulse(cs,ds,t);
subplot(2,1,2); plot(t,ha); axis([0,T*Nmax,-0.1,0.2]); set(gca,fontsize,10);
xlabel(time in seconds,fontsize,10); ylabel(ha(t),fontsize,10);
title(Impulse Response,fontsize,12); hold on;
% Impulse response of the digital filter
[x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
stem(n*T,h); hold off;
suptitle(Digital Butterworth Filter Design Plots in P 8.7)
The filter design plots are shown in Figure 8.7.
Comparison: From Figure 8.7 we observe that the impulse response h (n) of the digital filter is a sampled version of the
impulse response ha (t ) of the analog proptotype filter as expected.

122

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Digital Butterworth Filter Design Plots in P 8.7

LogMagnitude Response

Decibel

50
0

0.4

0.6

Frequency in Hz

Impulse Response

ha(t)

0.1

0.1

10

20

30

40

50
60
time in seconds

70

Figure 8.7: Impulse Invariance Design Method with T

80

=2

90

100

in Problem P 8.7

8. M ATLAB function dlpfd ii.m:

function [b,a] = dlpfd_ii(type,wp,ws,Rp,As,T)
%
% function [b,a] = dlpfd_ii(type,wp,ws,Rp,As,T)
%
Designs digital lowpass filters using impulse invariance mapping
% type = butter or cheby1
%
wp = passband cutoff in radians
%
ws = stopband cutoff in radians
%
Rp = passband ripple in dB
%
As = stopband attenuation in dB
%
T = sampling interval
if (type == cheby2)|(type == ellip )
error(Specify the correct type as butter or cheby1)
end
Fs = 1/T; twopi = 2*pi; K = Fs/twopi;
% Analog Prototype Specifications: Inverse mapping for frequencies
Fp = wp*K;
% Prototype Passband freq in Hz
Fs = ws*K;
% Prototype Stopband freq in Hz
ep = sqrt(10(Rp/10)-1);
% Passband Ripple parameter
Ripple = sqrt(1/(1+ep*ep));
% Passband Ripple
Attn = 1/(10(As/20));
% Stopband Attenuation
% Analog Butterworth Prototype Filter Calculation:
[cs,ds] = afd(type,Fp,Fs,Rp,As);
% Impulse Invariance transformation:
[b,a] = imp_invr(cs,ds,T);
[C,B,A] = dir2par(b,a)

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

123

M ATLAB verification using Problem P8.7:

clear; close all; format short e
%% Problem P8.7 : Butterworth Design
%% Digital Filter Specifications
wp = 0.4*pi;
% Passband
Rp = 0.5;
% Passband
ws = 0.6*pi;
% Stopband
As = 50;
% Stopband
%
% Impulse Invariance Digital Design using T
T = 2;
[b,a] = dlpfd_ii(butter,wp,ws,Rp,As,T);

edge in rad/sam
Ripple in dB
edge in rad/sam
attenuation in dB
= 2

*** Butterworth Filter Order = 17

C =
[]
B =
3.3654e+000 -1.2937e+000
-8.3434e+000 -6.9148e+000
2.9916e-002 2.7095e-001
-5.2499e+001 2.6839e+001
1.2550e+002 4.3610e+000
1.7953e+002 -1.0414e+002
-5.1360e+002 1.0421e+002
-1.3402e+002 8.1785e+001
4.0004e+002
0
A =
1.0000e+000 -3.9001e-001 4.8110e-001
1.0000e+000 -4.0277e-001 3.0369e-001
1.0000e+000 -4.1965e-001 7.8138e-001
1.0000e+000 -4.3154e-001 1.9964e-001
1.0000e+000 -4.6254e-001 1.3864e-001
1.0000e+000 -4.8933e-001 1.0298e-001
1.0000e+000 -5.0923e-001 8.2651e-002
1.0000e+000 -5.2131e-001 7.2211e-002
1.0000e+000 -2.6267e-001
0
%
% Plots of Log-magnitude Response and Impulse Response
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.8b);
% Log-Magnitude response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-60,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in pi units); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;
% Magnitude response
subplot(2,1,2); plot(w/pi,mag); axis([0,1,0,0.55]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[0;0.5]);grid

124

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

set(gca,YTickLabelMode,manual,YtickLabels,[0.0;0.5]);
xlabel(Frequency in pi units); ylabel(|H|);
title(Magnitude Response,fontsize,12); axis;
suptitle(Digital Butterworth Filter Design Plots in P 8.8)
The filter design plots are given in Figure 8.8.
Digital Butterworth Filter Design Plots in P 8.8
LogMagnitude Response

Decibel

50
0

0.4
0.6
Frequency in pi units

Magnitude Response

|H|

0.5

0.0

0.4
0.6
Frequency in pi units

Figure 8.8: Digital filter design plots in Problem P8.8.

9. Problem P 8.11
Digital Butterworth Lowpass Filter Design using Bilinear transformation. M ATLAB script:
clear; close all;
%% Digital Filter Specifications
wp = 0.4*pi;
% Passband edge in rad/sam
Rp = 0.5;
% Passband Ripple in dB
ws = 0.6*pi;
% Stopband edge in rad/sam
As = 50;
% Stopband attenuation in dB
(a) Part(a): T

= 2.

M ATLAB script:

T = 2;
OmegaP = (2/T)*tan(wp/2);
% Analog Prototype Passband edge
OmegaS = (2/T)*tan(ws/2);
% Analog Prototype Stopband edge
[cs,ds] = afd_butt(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Butterworth Filter Order = 11
[b,a] = bilinear(cs,ds,1/T)
% Bilinear Transformation
b =
Columns 1 through 7
0.0004
0.0048
0.0238
0.0715
0.1429
0.2001

0.2001

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

125

Columns 8 through 12
0.1429
0.0715
0.0238
0.0048
0.0004
a =
Columns 1 through 7
1.0000
-1.5495
2.5107
-2.1798
1.7043
-0.8997
0.4005
Columns 8 through 12
-0.1258
0.0309
-0.0050
0.0005
0.0000
%
% Plots of Log-magnitude Response and Impulse Response
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.11a);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-60,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;
% Impulse response of the prototype analog filter
Nmax = 50; t = 0:T/10:T*Nmax; [ha,x,t] = impulse(cs,ds,t);
subplot(2,1,2); plot(t,ha); axis([0,T*Nmax,-0.3,0.4]); set(gca,fontsize,10);
xlabel(time in seconds,fontsize,10); ylabel(ha(t),fontsize,10);
title(Impulse Response,fontsize,12); hold on;
% Impulse response of the digital filter
[x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
stem(n*T,h); hold off;
suptitle(Digital Butterworth Filter Design Plots in P 8.11a);
The filter design plots are shown in Figure 8.9.
Comparison: If we compare filter orders from two methods then bilinear transformation gives the lower order than
the impulse invariance method. This implies that the bilinear transformation design method is a better one in all
aspects. If we compare the impulse responses then we observe from Figure 8.9 that the digital impulse response is
not a sampled version of the analog impulse response as was the case in Figure 8.7.
(b) Part (b): Use of the butter function. M ATLAB script:
[N,wn] = buttord(wp/pi,ws/pi,Rp,As);
[b,a] = butter(N,wn)
b =
Columns 1 through 7
0.0005
0.0054
0.0270
0.0810
0.1619
0.2267
0.2267
Columns 8 through 12
0.1619
0.0810
0.0270
0.0054
0.0005
a =
Columns 1 through 7
1.0000
-1.4131
2.3371
-1.9279
1.5223
-0.7770
0.3477
Columns 8 through 12
-0.1066
0.0262
-0.0042
0.0004
0.0000
%
% Plots of Log-magnitude Response and Impulse Response
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P8.11b);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);

126

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Digital Butterworth Filter Design Plots in P 8.11a

LogMagnitude Response

Decibel

50
0

0.4

0.6

Frequency in Hz

Impulse Response

ha(t)

0.2

0.2
0

10

20

30

40

50
60
time in seconds

70

Figure 8.9: Bilinear Transformation Design Method with T

80

=2

90

100

in Problem P 8.11a

subplot(2,1,1); plot(w/pi,db); axis([0,1,-60,0]); set(gca,fontsize,10);

set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;
% Impulse response of the digital filter
Nmax = 50; [x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
subplot(2,1,2); stem(n,h); axis([0,Nmax,-0.3,0.4]); set(gca,fontsize,10);
xlabel(n,fontsize,10); ylabel(h(n),fontsize,10);
title(Impulse Response,fontsize,12);
suptitle(Digital Butterworth Filter Design Plots in P 8.11b);
The filter design plots are shown in Figure 8.10.
Comparison: If we compare the plots of filter responses in part 9a with those in part 9b, then we observe that the
butter function satisfies stopband specifications exactly at s . Otherwise the both designs are essentially similar.
10. Problem P 8.13
Digital Chebyshev-1 Lowpass Filter Design using Bilinear transformation. M ATLAB script:
clear; close all; Twopi = 2*pi;
%% Analog Filter Specifications
Fsam = 8000;
Fp = 1500;
Rp = 3;
Fs = 2000;
As = 40;
%

%
%
%
%
%

Sampling
Passband
Passband
Stopband
Stopband

Rate in sam/sec
edge in Hz
Ripple in dB
edge in Hz
attenuation in dB

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

127

Digital Butterworth Filter Design Plots in P 8.11b

LogMagnitude Response

Decibel

50
0

0.4

0.6

Frequency in Hz

Impulse Response

h(n)

0.2

0.2
0

10

15

20

25
n

30

35

40

45

50

Figure 8.10: Butterworth filter design using the butter function in Problem P 8.11b
%%
wp
Rp
ws
As

Digital Filter Specifications

= Twopi*Fp/Fsam;
= 3;
= Twopi*Fs/Fsam;
= 40;

(a) Part(a): T

= 1.

%
%
%
%

Passband
Passband
Stopband
Stopband

edge in rad/sam
Ripple in dB
edge in rad/sam
attenuation in dB

M ATLAB script:

%% (a) Bilinear Transformation Digital Design using T = 1

T = 1;
OmegaP = (2/T)*tan(wp/2);
% Analog Prototype Passband edge
OmegaS = (2/T)*tan(ws/2);
% Analog Prototype Stopband edge
[cs,ds] = afd_chb1(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Chebyshev-1 Filter Order =
[b,a] = bilinear(cs,ds,1/T);
[C,B,A] = dir2cas(b,a),
C =
0.0011
B =
1.0000
2.0126
1.0127
1.0000
1.9874
0.9875
1.0000
2.0001
0.9999
A =
1.0000
-0.7766
0.9308
1.0000
-1.1177
0.7966
1.0000
-1.5612
0.6901

The filter design plots are shown in Figure 8.11.

(b) Part(b): T

= 1=8000.

M ATLAB script:

% Bilinear Transformation
% Cascade form

128

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Filter Design Plots in P 8.13a

LogMagnitude Response

Decibel

40

60

1500

2000
Frequency in Hz

4000

Impulse Response
0.3

h(n)

0.2
0.1
0
0.1
0.2

10

20

30

40

50
n

60

70

Figure 8.11: Bilinear Transformation Design Method with T

80

=1

90

100

in Problem P 8.13a

%% (b) Impulse Invariance Digital Design using T = 1/8000

T = 1/8000;
OmegaP = (2/T)*tan(wp/2);
% Analog Prototype Passband edge
OmegaS = (2/T)*tan(ws/2);
% Analog Prototype Stopband edge
[cs,ds] = afd_chb1(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Chebyshev-1 Filter Order =
[b,a] = bilinear(cs,ds,1/T);
[C,B,A] = dir2cas(b,a),
C =
0.0011
B =
1.0000
2.0265
1.0267
1.0000
1.9998
1.0001
1.0000
1.9737
0.9739
A =
1.0000
-0.7766
0.9308
1.0000
-1.1177
0.7966
1.0000
-1.5612
0.6901

6
% II Transformation
% Cascade form

The filter design plots are shown in Figure 8.12.

(c) Comparison: If we compare the designed system function as well as the plots of system responses in part 10a and
in part 10a, then we observe that these are exactly same. If we compare the impulse invariance design in Problem
6 with this one then we note that the order of the impulse invariance designed filter is one higher. This implies that
the bilinear transformation design method is a better one in all aspects.
11. Digital lowpass filter design using elliptic prototype.
M ATLAB script using the bilinear function:

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

129

Filter Design Plots in P 8.13b

LogMagnitude Response

Decibel

40

60

1500

2000
Frequency in Hz

4000

Impulse Response
0.3

h(n)

0.2
0.1
0
0.1
0.2

10

20

30

40

50
n

60

70

Figure 8.12: Bilinear Transformation Design Method with T

80

90

= 1=8000 in

100

Problem P 8.13b

clear; close all;

%% Digital Filter Specifications
wp = 0.4*pi;
% Passband edge in rad/sam
Rp = 1;
% Passband Ripple in dB
ws = 0.6*pi;
% Stopband edge in rad/sam
As = 60;
% Stopband attenuation in dB
%
%% (a) Bilinear Transformation Digital Design using T = 2
T = 2;
OmegaP = (2/T)*tan(wp/2);
% Analog Prototype Passband edge
OmegaS = (2/T)*tan(ws/2);
% Analog Prototype Stopband edge
[cs,ds] = afd_elip(OmegaP,OmegaS,Rp,As); % Analog Prototype Design
*** Elliptic Filter Order = 5
[b,a] = bilinear(cs,ds,1/T);
% Bilinear Transformation
%
% Plots of Log-magnitude Response and Impulse Response
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.14a);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-80,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;

130

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

% Impulse response of the prototype analog filter

Nmax = 50; t = 0:T/10:T*Nmax; [ha,x,t] = impulse(cs,ds,t);
subplot(2,1,2); plot(t,ha); axis([0,T*Nmax,-0.3,0.4]); set(gca,fontsize,10);
xlabel(time in seconds,fontsize,10); ylabel(ha(t),fontsize,10);
title(Impulse Response,fontsize,12); hold on;
% Impulse response of the digital filter
[x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
stem(n*T,h); hold off;
suptitle(Digital Elliptic Filter Design Plots in P 8.14a);
The filter design plots are shown in Figure 8.13.
Digital Elliptic Filter Design Plots in P 8.14a
LogMagnitude Response

Decibel

60
80

0.4

0.6

Frequency in Hz

Impulse Response

ha(t)

0.2

0.2
0

10

20

30

40

50
60
time in seconds

70

80

90

100

Figure 8.13: Digital elliptic lowpass filter design using the bilinear function in Problem P8.14a.
M ATLAB script using the ellip function:
%% (b) Use of the Ellip function
[N,wn] = ellipord(wp/pi,ws/pi,Rp,As);
[b,a] = ellip(N,Rp,As,wn);
%
% Plots of Log-magnitude Response and Impulse Response
Hf_2 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_2,NumberTitle,off,Name,P8.14b);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-80,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;wp/pi;ws/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

131

title(Log-Magnitude Response,fontsize,12); axis;

% Impulse response of the digital filter
Nmax = 50; [x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
subplot(2,1,2); stem(n,h); axis([0,Nmax,-0.3,0.4]); set(gca,fontsize,10);
xlabel(n,fontsize,10); ylabel(h(n),fontsize,10);
title(Impulse Response,fontsize,12);
suptitle(Digital Elliptic Filter Design Plots in P 8.14b);
The filter design plots are shown in Figure 8.14. From these two figures we observe that both functions give the same
design in which the digital filter impulse response is not a sampled version of the corresponding analog filter impulse
response.
Digital Elliptic Filter Design Plots in P 8.14b
LogMagnitude Response

Decibel

60
80

0.4

0.6

Frequency in Hz

Impulse Response

h(n)

0.2

0.2
0

10

15

20

25
n

30

35

40

45

50

Figure 8.14: Digital elliptic lowpass filter design using the ellip function in Problem P8.14b.
12. Digital elliptic highpass filter design using bilinear mapping.
M ATLAB function dhpfd bl.m:
function [b,a] = dhpfd_bl(type,wp,ws,Rp,As)
%
% function [b,a] = dhpfd_bl(type,wp,ws,Rp,As,T)
%
Designs digital highpass filters using bilinear
%
b = Numerator polynomial of the highpass filter
%
a = Denominator polynomial of the highpass filter
% type = butter or cheby1 or cheby2 or ellip
%
wp = passband cutoff in radians
%
ws = stopband cutoff in radians (ws < wp)
%
Rp = passband ripple in dB
%
As = stopband attenuation in dB
% Determine the digital lowpass cutoff frequecies:
wplp = 0.2*pi;

132

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

alpha = -(cos((wplp+wp)/2))/(cos((wplp-wp)/2));
wslp = angle(-(exp(-j*ws)+alpha)/(1+alpha*exp(-j*ws)));
%
% Compute Analog lowpass Prototype Specifications:
T = 1; Fs = 1/T;
OmegaP = (2/T)*tan(wplp/2);
OmegaS = (2/T)*tan(wslp/2);
% Design Analog Chebyshev Prototype Lowpass Filter:
type = lower([type, ]); type = type(1:6);
if type == butter
[cs,ds] = afd_butt(OmegaP,OmegaS,Rp,As);
elseif type == cheby1
[cs,ds] = afd_chb1(OmegaP,OmegaS,Rp,As);
elseif type == cheby2
[cs,ds] = afd_chb2(OmegaP,OmegaS,Rp,As);
elseif type == ellip
[cs,ds] = afd_elip(OmegaP,OmegaS,Rp,As);
else
error(Specify the correct type)
end
% Perform Bilinear transformation to obtain digital lowpass
[blp,alp] = bilinear(cs,ds,Fs);
% Transform digital lowpass into highpass filter
Nz = -[alpha,1]; Dz = [1,alpha];
[b,a] = zmapping(blp,alp,Nz,Dz);
(a) Design using the dhpfd bl function:
clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.17)
%% Digital Filter Specifications
type = ellip;
% Elliptic
ws = 0.4*pi;
% Passband
As = 60;
% Stopband
wp = 0.6*pi;
% Stopband
Rp = 1;
% Passband
%
%% (a) Use of the dhpfd_bl function
[b,a] = dhpfd_bl(type,wp,ws,Rp,As)

design
edge in rad/sam
attenuation in dB
edge in rad/sam
Ripple in dB

*** Elliptic Filter Order = 5

b =
0.0208
-0.0543
0.0862
-0.0862
0.0543
-0.0208
a =
1.0000
2.1266
2.9241
2.3756
1.2130
0.3123
% Plot of Log-magnitude Response
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,1); plot(w/pi,db); axis([0,1,-80,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;ws/pi;wp/pi;1]);

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

133

set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Design using the dhpfd_bl function,fontsize,12); axis;
The filter frequency response plot is shown in the top row of Figure 8.15.
(b) Design using the ellip function:
%% (b) Use of the Ellip function
[N,wn] = ellipord(wp/pi,ws/pi,Rp,As);
[b,a] = ellip(N,Rp,As,wn,high)
b =
0.0208
-0.0543
0.0862
-0.0862
0.0543
-0.0208
a =
1.0000
2.1266
2.9241
2.3756
1.2130
0.3123
%
% Plot of Log-magnitude Response
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,2); plot(w/pi,db); axis([0,1,-80,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0;ws/pi;wp/pi;1]);
set(gca,YTickMode,manual,Ytick,[-80;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[80;AS; 0]);
xlabel(Frequency in Hz); ylabel(Decibel);
title(Design using the ellip function,fontsize,12); axis;
suptitle(Digital Elliptic Filter Design Plots in P 8.17);
The filter frequency response plot is shown in the bottom row of Figure 8.15. Both M ATLAB scripts and the Figure
8.15 indicate that we designed essentially the same filter.
13. Digital Chebyshev-2 bandpass filter design using bilinear transformation. M ATLAB script:
clear; close all;
Hf_1 = figure(Units,normalized,position,[0.1,0.1,0.8,0.8],color,[0,0,0]);
set(Hf_1,NumberTitle,off,Name,P8.19)
%% Digital Filter Specifications
ws1 = 0.3*pi;
% Lower passband edge in rad/sam
ws2 = 0.6*pi;
% upper passband edge in rad/sam
As = 50;
% Stopband attenuation in dB
wp1 = 0.4*pi;
% Lower passband edge in rad/sam
wp2 = 0.5 *pi;
% Lower passband edge in rad/sam
Rp = 0.5;
% Passband Ripple in dB
%
%% Use of the cheby2 function
ws = [ws1,ws2]/pi; wp = [wp1,wp2]/pi;
[N,wn] = cheb2ord(wp,ws,Rp,As)
N =
5
wn =
0.3390
0.5661
[b,a] = cheby2(N,As,wn)
b =
Columns 1 through 7

134

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

A PRIL 98

Digital Elliptic Filter Design Plots in P 8.17

Design using the dhpfd_bl function

Decibel

60
80

0.4

0.6

Frequency in Hz

Design using the ellip function

Decibel

60
80

0.4

0.6

Frequency in Hz

Figure 8.15: Digital elliptic highpass filter design plots in Problem 8.17.
0.0050
-0.0050
0.0087
-0.0061
0.0060
0.0000
-0.0060
Columns 8 through 11
0.0061
-0.0087
0.0050
-0.0050
a =
Columns 1 through 7
1.0000
-1.3820
4.4930
-4.3737
7.4582
-5.1221
5.7817
Columns 8 through 11
-2.6221
2.0882
-0.4936
0.2764
%
% Plot of the filter Responses
% Impulse response
Nmax = 50; [x,n] = impseq(0,0,Nmax); h = filter(b,a,x);
subplot(2,1,1); stem(n,h); axis([0,Nmax,-0.15,0.15]); set(gca,fontsize,10);
xlabel(n,fontsize,10); ylabel(h(n),fontsize,10);
title(Impulse Response,fontsize,12);
% Frequency response
[db,mag,pha,grd,w] = freqz_m(b,a);
subplot(2,1,2); plot(w/pi,db); axis([0,1,-70,0]); set(gca,fontsize,10);
set(gca,XTickMode,manual,Xtick,[0,ws,wp,1]);
set(gca,YTickMode,manual,Ytick,[-70;-As;0]);grid
AS = [num2str(As)];
set(gca,YTickLabelMode,manual,YtickLabels,[70;AS; 0]);
xlabel(Frequency in pi units); ylabel(Decibel);
title(Log-Magnitude Response,fontsize,12); axis;
suptitle(Digital Chebyshev-2 Bandpass Filter Design Plots in P 8.19);
The filter impulse and log-magnitude response plots are shown in Figure 8.16.

A PRIL 98

S OLUTIONS M ANUAL

FOR

DSP

USING

M ATLAB

135

Digital Chebyshev2 Bandpass Filter Design Plots in P 8.19

Impulse Response
0.15
0.1

h(n)

0.05
0
0.05
0.1
0

10

15

20

25
n

30

35

40

45

50

LogMagnitude Response

Decibel

50

70

0.3

0.4
0.5
0.6
Frequency in pi units

Figure 8.16: Digital Chebyshev-2 bandpass filter design plots in Problem P8.19.