1

ELEC 6601
Name ID Project
Syeda Shawlin Monjur 40257281 01
2

1.Title: Analysis of Fourier Transform Properties for a Random Complex-

Valued Signal

Objective:
This project demonstrates two key properties of the Fourier Transform (FT)
for a random complex-valued signal x[n]x[n]x[n]. The main goals are:

1. To verify that the Fourier Transform of the conjugate-symmetric part of

the signal corresponds to the real part of the original Fourier Transform.

2. To verify that the Fourier Transform of the conjugate and time-reversed

part of the signal corresponds to the conjugate of the original Fourier
Transform.

Problem Statement

Given a random complex-valued signal x[n], where the real and imaginary
parts are independently drawn from a uniform distribution over the range
[0,1], we define:

 X(e^jω) as the Fourier Transform of x[n],

 xe[n] as the conjugate-symmetric part of the signal, and

 x∗[−n] as the conjugate and time-reversed part of the signal.

The objective is to:

1. Show that the Fourier Transform of xe[n], denoted as Xe(e^jω), is equal

to the real part of X(e^jω), i.e.,

F(xe[n]) = Re{X(e^jω)}.

2. Demonstrate that the Fourier Transform of x∗[−n], denoted as

X∗(−e^jω), is equal to the conjugate of X(e^jω)

F(x∗[−n]) = X∗(e^jω).
3

Methodology

1. Signal Generation: A random complex-valued signal x[n] of length

N=1024 was generated, where the real and imaginary parts were
independently drawn from a uniform distribution over [0,1].

2. Fourier Transform: The Fourier Transform X(e^jω) of the signal x[n]

was computed using the Fast Fourier Transform (FFT).

3. Part I: Conjugate-Symmetric Part: The conjugate-symmetric part

xe[n] was defined as:

xe[n]=1/2(x[n]+x∗[−n]).

Its Fourier Transform Xe(e^jω) was computed, and its magnitude was
compared to the real part of the original Fourier Transform Re{X(e^jω)}.

4. Part II: Conjugate and Time-Reversed Part: The conjugate and

time-reversed signal x∗[−n] was constructed, and its Fourier Transform
X∗(−e^jω) was computed. The result was compared with the
conjugate of the original Fourier Transform X∗(e^jω).

Results

Figure 1: Part I – Fourier Transform Comparison of Xe(e^jω) and

Re{X(ejω)}.
4

Figure 2: Part II – Fourier Transform Comparison of X∗(−e^jω) and

X∗(e^jω).

1. Part I: The Fourier Transform of the conjugate-symmetric part xe[n]

shown in the plot (see Figure 1), the magnitude of ∣Xe(e^jω)∣ (blue line)
was compared to the real part of the original Fourier Transform. As

aligns closely with the real part of X(e^jω) (red dashed line),
confirming that:

F(xe[n]) = Re{X(e^jω)}.

This verifies the desired property for the conjugate-symmetric part.

2. Part II: The Fourier Transform of the conjugate and time-reversed

signal x∗[−n] was compared to the conjugate of the original Fourier

∣X∗(−ejω)∣ (blue line) matches ∣X∗(ejω)∣ (red dashed line). This

Transform. The plot (see Figure 2) shows that the magnitude of

confirms that:

F(x∗[−n]) = X∗(e^jω).

Therefore, the Fourier Transform of the conjugate and time-reversed

signal correctly corresponds to the conjugate of the original Fourier
Transform.

Conclusion
5

Both parts of the problem have been successfully demonstrated. The key
properties of the Fourier Transform for a random complex-valued signal
were verified as follows:

1. The Fourier Transform of the conjugate-symmetric part of the signal

equals the real part of the original Fourier Transform.

2. The Fourier Transform of the conjugate and time-reversed part of the

signal equals the conjugate of the original Fourier Transform.

The MATLAB simulations and the corresponding plots confirm these

relationships, and the results are consistent with the theoretical
expectations.
6

Appendix

N = 1024;
n = 0:N-1;

real_part = rand(1, N);

imag_part = rand(1, N);
x = real_part + 1i * imag_part;

X = fft(x);

x_e = 0.5 * (x + conj(flip(x)));

X_e = fft(x_e);

real_X = real(X);

figure;
subplot(2,1,1);
freqs = (0:N-1) * (2*pi/N);
plot(freqs, abs(X_e), 'b', 'DisplayName', '|X_e(e^{j\omega})|');
hold on;
plot(freqs, abs(real_X), 'r--', 'DisplayName', 'Re{X(e^{j\omega})}');
title('Part I: Fourier Transform Comparison');
xlabel('\omega');
ylabel('Magnitude');
legend;
grid on;

x_conj_time_rev = conj(flip(x));

X_conj_time_rev = fft(x_conj_time_rev);

X_conj = conj(X);

subplot(2,1,2);
plot(freqs, abs(X_conj_time_rev), 'b', 'DisplayName', '|X^*(-e^{j\omega})|');
hold on;
plot(freqs, abs(X_conj), 'r--', 'DisplayName', '|X^*(e^{j\omega})|');
title('Part II: Fourier Transform Comparison');
xlabel('\omega');
ylabel('Magnitude');
legend;
grid on;

sgtitle('Demonstration of Fourier Transform Properties');

7

2.Title: Verifying Fourier Transform Properties

Introduction:

In this problem, we are tasked with verifying four properties of the

Discrete-Time Fourier Transform (DTFT) for a random real-valued signal
x[n]. These properties are:

1. The real part of X(e^jω) is even: XR(e^jω) = XR(e^−jω).

2. The imaginary part of X(e^jω) is odd: XI(e^jω)=−XI(e^−jω).

3. The magnitude of X(e^jω) is even: ∣X(e^jω)∣=∣X(e^−jω)∣.

4. The phase of X(e^jω) is odd: ∠X(e^jω)=−∠X(e^−jω.

The random signal x[n] is generated, and the DTFT is computed using
MATLAB to demonstrate these properties.

Methodology:

1. Signal Generation: A random real-valued signal x[n] of length N=40

is generated using the rand function.

2. DTFT Computation: The Fourier Transform X(e^jω) is computed

using a direct summation approach, where:

for a set of frequency values ω.

3. Property Verification:

 The real part of X(e^jω) , XR(e^jω)), is compared with its circularly

shifted version to check whether it is even.
8

 The imaginary part, XI(e^jω), is checked for oddness by comparing it

with its negated circular shift.

 The magnitude ∣X(e^jω)∣ is verified for evenness.

 The phase ∠X(e^jω) is checked for oddness by comparing it with its

negated circular shift.

Results:

Real and Imaginary Parts: The real part of X(e^jω) was verified to
be even, as demonstrated in Figure 1. The imaginary part was
confirmed to be odd by comparing the original with its negated
version.

Figure :1 Real and Imaginary part

9

Figure 2: Magnitude and Phase part.

Magnitude: The magnitude ∣X(e^jω) as shown in the upper subplot of

Figure 2, is even. This is evident from the comparison of ∣X(e^jω)∣ with
∣X(e^−jω)∣ which match closely, confirming the symmetry.

∠X(e^jω)\ is odd, as seen by the close match between the phase and
Phase: The lower subplot of Figure 2 demonstrates that the phase

its negated circular shift.

10

Conclusion:

The MATLAB implementation successfully verified the following

properties of the DTFT for the real-valued signal x[n]:

The real part is even.

The imaginary part is odd.

The magnitude is even.

The phase is odd.

These results align with the theoretical expectations for the Fourier
transform of a real-valued signal, confirming the correctness of the
analysis.

Appendix

N = 40;
n = 0:N-1;

x = rand(1, N);

frequencies = (0:N-1) * (2 * pi / N);

X = zeros(1, N);
for k = 1:N
X(k) = sum(x .* exp(-1j * frequencies(k) * n));
end

XR = real(X);
XI = imag(X);

is_even_real = all(abs(XR - circshift(XR, N/2)) < 1e-10);

is_odd_imaginary = all(abs(XI + circshift(XI, N/2)) < 1e-10);

magnitude_X = abs(X);
is_even_magnitude = all(abs(magnitude_X - circshift(magnitude_X, N/2)) < 1e-10);

phase_X = angle(X);
is_odd_phase = all(abs(phase_X + circshift(phase_X, N/2)) < 1e-10);

fprintf('Property I (Real part is even): %s\n', mat2str(is_even_real));

fprintf('Property II (Imaginary part is odd): %s\n', mat2str(is_odd_imaginary));
fprintf('Property III (Magnitude is even): %s\n', mat2str(is_even_magnitude));
fprintf('Property IV (Phase is odd): %s\n', mat2str(is_odd_phase));

figure (1);
11

subplot(2, 1, 1);
plot(frequencies, XR, 'b', 'DisplayName', 'Real Part');
hold on;
plot(frequencies, circshift(XR, N/2), 'r--', 'DisplayName', 'Circshift Real Part');
title('Real Part of X(e^{j\omega})');
xlabel('\omega');
ylabel('Magnitude');
legend;
grid on;

subplot(2, 1, 2);
plot(frequencies, XI, 'b', 'DisplayName', 'Imaginary Part');
hold on;
plot(frequencies, -circshift(XI, N/2), 'r--', 'DisplayName', 'Negated Circshift Imaginary Part');
title('Imaginary Part of X(e^{j\omega})');
xlabel('\omega');
ylabel('Magnitude');
legend;
grid on;

figure (2);
subplot(2, 1, 1);
plot(frequencies, magnitude_X, 'b', 'DisplayName', '|X(e^{j\omega})|');
hold on;
plot(frequencies, circshift(magnitude_X, N/2), 'r--', 'DisplayName', '|X(e^{-j\omega})|');
title('Magnitude of X(e^{j\omega})');
xlabel('\omega');
ylabel('Magnitude');
legend;
grid on;

subplot(2, 1, 2);
plot(frequencies, phase_X, 'b', 'DisplayName', 'Phase of X(e^{j\omega})');
hold on;
plot(frequencies, -circshift(phase_X, N/2), 'r--', 'DisplayName', 'Negated Circshift Phase');
title('Phase of X(e^{j\omega})');
xlabel('\omega');
ylabel('Radians');
legend;
grid on;
12

3.Title: Convolution and Fourier Transform Analysis of Discrete-

Time Signals

Introduction

In this problem, we are tasked with analyzing two discrete-time signals

x[n]=u[n+1]−u[n−1] and h[n]=(0.4)^nu[n], where u[n] is the unit step
function. We aim to:

1. Plot the signals x[n] and h[n] the time domain.

2. Perform the convolution of x[n] and h[n] and plot the resulting signal
y[n]=x[n]∗h[n].

3. Verify that the Fourier Transform of the convolved signal Y(e^jω) is

equal to the product of the Fourier Transforms of x[n] and h[n],i.e.,
Y(e^jω)=X(e^jω)H(e^jω).

Methodology

We used MATLAB to define the two signals, compute their convolution, and
calculate their Fourier Transforms. The steps are as follows:

1. Define the range of n for the discrete signals.

2. Define x[n] as the difference of two unit step functions, resulting in a

signal that is non-zero at n=−1and n=0.

3. Define h[n] as an exponentially decaying signal starting from n=0.

4. Compute the convolution y[n] using the MATLAB conv function.

5. Use the FFT (Fast Fourier Transform) to compute the Fourier Transforms
X(e^jω), H(e^jω), and Y(e^jω), and plot their magnitudes for
verification.
13

Results

1. Time-Domain Plots of x[n]xand h[n]:

 Signal x[n]:

o The signal is non-zero at n=−1and n=0, resembling a short

impulse response. The values are 1 at these points and 0
elsewhere.

 Signal h[n]:

o The signal h[n]=(0.4)^nu[n] is an exponentially decaying

sequence that starts at n=0. For positive n, the signal decays
rapidly, while it remains zero for n<0.
14

Figure: 1

2. Convolution of x[n] and h[n]:

The result of the convolution y[n]=x[n]∗h[n] is a delayed version of h[n],

starting at n=−1, and following the same exponential decay. The width of
the response increases slightly due to the convolution with x[n], but the
overall shape is similar to h[n].

Figure: 2

3. Fourier Transform Analysis:

We computed the Discrete-Time Fourier Transform (DTFT) of each signal

using the FFT, and plotted the magnitudes:

 Magnitude of X(e^jω):
15

o The Fourier Transform of x[n], which is a short signal in time,

results in a broad spectrum with significant frequency content
spread across the range.

 Magnitude of H(e^jω):

o The Fourier Transform of the exponentially decaying signal h[n]

shows a smooth curve, with higher values at lower frequencies,
and a gradual decrease at higher frequencies.

 Magnitude of Y(e^jω)=X(e^jω)H(e^jω):

o The product of the two transforms, Y(e^jω), combines the

features of X(e^jω) and H(e^jω), resulting in a magnitude plot
that decays smoothly but includes fluctuations inherited from
X(e^jω).

The plots of ∣Y(e^jω)∣ verify that the convolution in the time domain
corresponds to the multiplication of the Fourier Transforms in the
frequency domain.
16

Figure: 3

Conclusion

In this analysis, we verified the key property of convolution: that the

Fourier Transform of the convolved signal is equal to the product of the
individual Fourier Transforms. The signals x[n] and h[n] were analyzed
both in the time and frequency domains, and the results were consistent
with the expected theoretical behavior.

This exercise confirms that convolution in the time domain corresponds to

multiplication in the frequency domain, a fundamental property in signal
processing. The MATLAB simulations successfully demonstrated this
relationship, and the results align with the theoretical expectations.

Appendix
n = -10:20;

u = @(n) double(n >= 0);

x_n = u(n+1) - u(n-1);

h_n = (0.4).^n .* u(n);

figure;
subplot(2,1,1);
stem(n, x_n, 'filled');
title('Signal x[n]');
xlabel('n');
ylabel('x[n]');
grid on;

subplot(2,1,2);
stem(n, h_n, 'filled');
title('Signal h[n]');
xlabel('n');
ylabel('h[n]');
grid on;

y_n = conv(x_n, h_n);

n_conv = (min(n) + min(n)):(max(n) + max(n));

figure;
stem(n_conv, y_n, 'filled');
title('Convolution of x[n] and h[n], y[n] = x[n] * h[n]');
xlabel('n');
ylabel('y[n]');
grid on;
17

N = 512; % FFT length

X = fft(x_n, N);
H = fft(h_n, N);
Y = X .* H;

omega = linspace(-pi, pi, N);

figure;
subplot(3,1,1);
plot(omega, abs(fftshift(X)));
title('Magnitude of X(e^{j\omega})');
xlabel('\omega');
ylabel('|X(e^{j\omega})|');
grid on;

subplot(3,1,2);
plot(omega, abs(fftshift(H)));
title('Magnitude of H(e^{j\omega})');
xlabel('\omega');
ylabel('|H(e^{j\omega})|');
grid on;

subplot(3,1,3);
plot(omega, abs(fftshift(Y)));
title('Magnitude of Y(e^{j\omega}) = X(e^{j\omega}) H(e^{j\omega})');
xlabel('\omega');
ylabel('|Y(e^{j\omega})|');
grid on;