Digital Signal Processing Reference
In-Depth Information
Check the demodulated output on the scope and plot the error signal
between the recovered baseband output and the input modulating signal.
Determine the percentage mean-squared error (mse) between the two signals.
6.3.3
Filtering of Noisy Audio Signals
Exercise 4: Filtering of one-dimensional time signals mixed with random
noise
Communication systems face the common problem of noise.
3
As shown in
Figure 6.7, the simplest form of noise is additive noise
n
(
t
), which adds on
to the transmitted signal
s
(
t
). Several methods have been developed to tackle
the problem of noise removal from the corrupted signal
y
(
t
)
= s
(
t
)
+ n
(
t
). The
commonly used methods include autocorrelation and filtering.
a.
Create a new model file either in MATLAB or Simulink, as shown in
Figure 6.7. Generate an analog sinusoidal signal at a frequency of
3 KHz and amplitude of 5 volts. Verify the sinusoidal output,
s
(
t
), on
both the oscilloscope and the FFT analyzer, if using Simulink. If using
MATLAB, use the
fft
command to generate the output spectrum,
according to the procedure for periodic signals detailed in Section 3.1.3.
b.
Generate a uniform random noise signal,
n
(
t
), with a signal-to-noise
voltage ratio (SNR) of 30 dB. Check the output on the oscilloscope.
c.
Combine the signal
s
(
t
) and the noise
n
(
t
) and check the noisy output
on the oscilloscope and the FFT analyzer.
In this experiment, two types of noise-removal filters will be designed and
tested.
•
Digital bandpass filter
Design a digital bandpass filter with a center frequency of 3 KHz,
and suitable bandwidth to filter out the sinusoidal signal
s
(
t
) from
Noisy signal
y(t)
Input
signal
s(t)
Sample
and Hold
circuit
Digital
bandpass
filter
+
Filtered
output
Uniform
random noise
n(t)
FIGURE 6.7
Model for filtering of noisy sinusoidal signal.
