ECE 4213/5213

Homework 3 Solution
Fall 2020 Dr. Havlicek

1(a). The following graphs were obtained by running Program P2 2 with the three specified
frequencies:

f = 0.05
Input Signal
1

0.5
Amplitude

−0.5

−1
0 20 40 60 80 100 120 140 160 180 200
Time index n
Output signal
1
Amplitude

0.5

0
0 20 40 60 80 100 120 140 160 180 200
Time index n

1
 f = 0.10
Input Signal
1

0.5

Amplitude
0

−0.5

−1
0 20 40 60 80 100 120 140 160 180 200
Time index n
Output signal
1

0.8
Amplitude

0.6

0.4

0.2
0 20 40 60 80 100 120 140 160 180 200
Time index n

f = 0.25
Input Signal
1

0.5
Amplitude

−0.5

−1
0 20 40 60 80 100 120 140 160 180 200
Time index n
Output signal
2

1.5
Amplitude

0.5

0
0 20 40 60 80 100 120 140 160 180 200
Time index n

2
1(b). The output signals depend on the frequencies of the input signal according to:
y[n] = sin2 (ω) = sin2 (2πf ).

Mathematical Verification:
Let ω = 2πf and let the input signal be x[n] = cos(ωn). Then (2.14) on page 21 of
the course Laboratory Manual becomes

y[n] = x2 [n] − x[n − 1]x[n + 1]

= cos2 (ωn) − cos[ω(n − 1)] cos[ω(n + 1)]. (1)

We have cos[ω(n − 1)] = cos(ωn − ω) and cos[ω(n + 1)] = cos(ωn + ω), so

x[n − 1]x[n + 1] = cos(ωn − ω) cos(ωn + ω). (2)

Applying the trigonometric identity cos(A + B) cos(A − B) = 21 cos(2A) + 12 cos(2B)

to (2), we obtain
1 1
x[n − 1]x[n + 1] = cos(2ωn) + cos(2ω). (3)
2 2
Upon applying the trigonometric identity cos(2A) = 2 cos2 (A) − 1 to the first term on
the right-hand side of (3), there results
1 1
x[n − 1]x[n + 1] = cos2 (ωn) − + cos(2ω). (4)
2 2
Applying the trigonometric identity cos(2A) = 1 − 2 sin2 (A) to the second term on the
right-hand side of (4) then gives us
1 1
x[n − 1]x[n + 1] = cos2 (ωn) − + − sin2 (ω)
2 2
= cos2 (ωn) − sin2 (ω). (5)

Substituting (5) into (1), we then obtain

y[n] = cos2 (ωn) − cos2 (ωn) + sin2 (ω) = sin2 (ω), (6)

which verifies the desired result.

3
2(a). Listing of the modified program to make the input a cosine plus constant:

% Program Q2_6
% Modified version of P2_2 to make the input a cosine plus a constant.
clf;
n = 0:200;
K = 0.5;
f = 0.1;
arg = 2*pi*f*n;
x = cos(arg) + K;
% Compute the output signal
x1 = [x 0 0]; % x1[n] = x[n+1]
x2 = [0 x 0]; % x2[n] = x[n]
x3 = [0 0 x]; % x3[n] = x[n-1]
y = x2.*x2-x1.*x3;
y = y(2:202);
% Plot the input and output signals
subplot(2,1,1)
plot(n, x)
xlabel(’Time index n’);ylabel(’Amplitude’);
title(’Input Signal’)
subplot(2,1,2)
plot(n,y)
xlabel(’Time index n’);ylabel(’Amplitude’);
title(’Output signal’);

4
Graphs obtained by running the modified program:

f = 0.1, K = 0.5
Input Signal
1.5

1
Amplitude

0.5

−0.5
0 20 40 60 80 100 120 140 160 180 200
Time index n
Output signal
3

2
Amplitude

0
0 20 40 60 80 100 120 140 160 180 200
Time index n

2(b). From Eq. (3) on page 366 of the paper by Restrepo and Quiroga, we obtained

y[n] = T K[cos(ω0 n) + K] = sin2 (ω0 ) + 2K[1 − cos(ω0 )] cos(ω0 n). (7)

I will select choice (ii) and verify the correctness of (7) by starting with Eq. (2.14) on page
21 of the course Laboratory Manual.

Let x[n] = cos(ω0 n) + K. Eq. (2.14) of the course Laboratory Manual says that

y[n] = x2 [n] − x[n − 1]x[n + 1]. (8)

We have that
x2 [n] = cos2 (ω0 n) + 2K cos(ω0 n) + K 2 . (9)
Working on the second term on the right-hand side of (8), we obtain

x[n − 1] = cos[ω0 (n − 1)] + K = cos(ω0 n − ω0 ) + K

5
and
x[n + 1] = cos[ω0 (n + 1)] + K = cos(ω0 n + ω0 ) + K,
so

x[n − 1]x[n + 1] = cos(ω0 n − ω0 ) cos(ω0 n + ω0 ) + K[cos(ω0 n − ω0 ) + cos(ω0 n + ω0 )] + K 2 . (10)

Now, it was already shown in (2)-(5) above that

cos(ω0 n − ω0 ) cos(ω0 n + ω0 ) = cos2 (ω0 n) − sin2 (ω0 ). (11)

Substituting this into (10) and applying the trigonometric identity cos(A+B)+cos(A−B) =
2 cos(A) cos(B) to the term in square brackets, we get

x[n − 1]x[n + 1] = cos2 (ω0 n) − sin2 (ω0 ) + 2K cos(ω0 n) cos(ω0 ) + K 2 . (12)

Substituting (9) and (12) into (8) then produces the final result

y[n] = sin2 (ω0 ) + 2K cos(ω0 n) − 2K cos(ω0 n) cos(ω0 )

= sin2 (ω0 ) + 2K[1 − cos(ω0 )] cos(ω0 n), (13)

thereby verifying the correctness of (7).

Equations (7) and (13) provide the answer to how the output signal y[n] depends on the
DC value K. Compared to the case where x[n] = cos(ω0 n), there is an extra term added
to y[n]. This extra term is a cosine having the same frequency as the input and having an
amplitude of 2K[1 − cos(ω0 )]. This term adds a ripple to y[n] which may be plainly seen in
the graph on the preceding page. The amplitude of the ripple is proportional to K.

6
3(a). I selected the paper:
P. Maragos, J.F. Kaiser, and T.F. Quatieri, “On separating amplitude from frequency
modulations using energy operators,” in Proc. IEEE Int’l. Conf. Acoust., Speech, Signal
Proc., San Francisco, CA, Mar. 23-26, 1992, vol. II, pp. 1-4.

Another good one is:

P. Maragos, J.F. Kaiser, and T.F. Quatieri, “Energy separation in signal modulations
with applications to speech analysis,” IEEE Trans. Signal Process., vol. 41, no. 10,
Oct. 1993, pp. 3024-3051.

3(b). Listing of Matlab program to implement DESA-1:

% Program DESA1
% Implement the DESA-1 discrete energy separation algorithm
clear;
clf;
% Generate the input signal
n = 0:250;
A = 5.0;
f = 0.1;
w = 2*pi*f;
x = A*cos(w*n);

x2 = [0 x 0]; % zero pad x[n]

y = [0 x 0] - [0 0 x]; % y[n] = x[n] - x[n-1]
z = [x 0 0] - [0 x 0]; % z[n] = x[n+1] - x[n]

% TK operator on y[n]
ynp1 = [y 0 0]; % y[n+1]
yn = [0 y 0]; % y[n]
ynm1 = [0 0 y]; % y[n-1]
Psi_y = yn.*yn-ynm1.*ynp1;

% TK operator on z[n]
znp1 = [z 0 0]; % z[n+1]
zn = [0 z 0]; % z[n]
znm1 = [0 0 z]; % z[n-1]
Psi_z = zn.*zn-znm1.*znp1;

% TK operator on x[n]
x2np1 = [x2 0 0]; % x2[n+1]

7
x2n = [0 x2 0]; % x2[n]
x2nm1 = [0 0 x2]; % x2[n-1]
Psi_x2 = x2n.*x2n-x2nm1.*x2np1;

% Resize TK outputs to original size of x[n]

Psi_y = Psi_y(3:length(Psi_y)-2);
Psi_z = Psi_z(3:length(Psi_z)-2);
Psi_x2 = Psi_x2(3:length(Psi_x2)-2);

% Estimate the IA and IF

IFhat = acos(1 - (Psi_y + Psi_z)./(4*Psi_x2))/(2*pi);
IAhat = sqrt(Psi_x2./(1 - (1 - (Psi_y + Psi_z)./(4*Psi_x2)).^2));

% Plot the input and output signals

figure(1);
plot(n,x);
axis([0 250 -6 6]);
xlabel(’Time index n’);ylabel(’Amplitude’);
title(’Input Signal’);

figure(2);
plot(n,IAhat);
xlabel(’Time index n’);ylabel(’IA a[n]’);
title(’Instantaneous Amplitude (IA)’);

figure(3);
plot(n,IFhat);
xlabel(’Time index n’);ylabel(’IF \omega[n] / (2\pi)’);
title(’Instantaneous Frequency (IF)’);

8
Plots obtained by running the program with the input signal x[n] = 5 cos[2π(0.1)n]:
Input Signal
6

Amplitude
0

−2

−4

−6
0 50 100 150 200 250
Time index n
Instantaneous Amplitude (IA)
6.5

5.5

5
a[n]
IA

4.5

3.5

3
0 50 100 150 200 250
Time index n
Instantaneous Frequency (IF)
0.17

0.16

0.15

0.14
ω[n] / (2π)

0.13
IF

0.12

0.11

0.1

0.09
0 50 100 150 200 250
Time index n

9
3(c). Plots obtained by running the program on an AM-FM chirp signal:
Input Signal
6

Amplitude
0

−2

−4

−6
0 50 100 150 200 250
Time index n
Instantaneous Amplitude (IA)
6

4
a[n]

3
IA

0
0 50 100 150 200 250
Time index n
Instantaneous Frequency (IF)
0.5

0.45

0.4

0.35
ω[n] / (2π)

0.3

0.25
IF

0.2

0.15

0.1

0.05

0
0 50 100 150 200 250
Time index n

10