Sampling and aliasing

Signals and systems

PhD. Alberth Ronal Tamo Calla

Signals and Systems Topics
Domain Topic Discrete Time Continuous Time

Time Signals ✔ ✔
Systems
Convolution
Frequency Fourier series ✔ ✔
Fourier transforms
Frequency response
Generalized z / Laplace Transforms
Frequency
Transfer Functions
System Stability
Mixed Signal Sampling ➔

5-2
Aliasing
• Definition for “alias” [dictionary.com]
“a false name used to conceal one's identity; an assumed name”
• Is x2[n]=cos(2.4pn) an alias for x1[n]=cos(0.4pn)?
x2[n] = cos(2.4pn) = cos(0.4pn + 2pn) = cos(0.4pn) = x1[n]

f2 = 1.2 Hz

Sample at
fs = 1 Hz

f1 = 0.2 Hz

5-3
Aliasing
• MATLAB code from previous slide
%% Part 1: Define Signals %% Part 2: Generate Plots
wHat = 0.4*pi; figure;
nmax = 12; plot(t, x2cont, 'm-', 'LineWidth', 1);
n = 0:nmax;
x1 = cos(wHat*n); figure;
x2 = cos(2.4*pi*n); plot(t, x2cont, 'm-', 'LineWidth', 1);
hold;
fs = 1; stem(n, x1, 'Linewidth', 2, 'MarkerEdgeColor', 'black');
f1 = 0.2; stem(n, x2, 'Linewidth', 2, 'MarkerEdgeColor', 'black');
w1Hat = 2*pi*f1/fs;
period = round(fs/f1); figure;
f2 = 1.2; plot(t, x2cont, 'm-', 'LineWidth', 1);
w2Hat = 2*pi*f2/fs; hold;
Ts = 1/fs; stem(n, x1, 'Linewidth', 2, 'MarkerEdgeColor', 'black');
tmax = (nmax/period)*(1/f1); stem(n, x2, 'Linewidth', 2, 'MarkerEdgeColor', 'black');
t = 0 : (Ts/100) : tmax; plot(t, x1cont, 'b-', 'LineWidth', 2);
x1cont = cos(2*pi*f1*t);
x2cont = cos(2*pi*f2*t);
5-4
Aliasing
• Continuous-time sinusoid y[n] = y(Tsn)
= A cos(2p(f0 + lfs)Tsn + f)
x(t) = A cos(2p f0 t + f) = A cos(2pf0Tsn + 2plfsTsn + f)
= A cos(2pf0Tsn + 2pln + f)
• Sample at Ts = 1/fs = A cos(2pf0Tsn + f)
x[n] = x(Tsn) = A cos(2p f0 Ts n + f) = x[n]
Here, fsTs = 1
• Keeping the sampling period same,
sample Since l is an integer,
y(t) = A cos(2p (f0 + l fs) t + f) cos(x + 2 p l) = cos(x)

where l is an integer • y[n] indistinguishable from x[n]

5-5
Another Kind of Aliasing (Folding)
• Continuous-time sinusoid y[n] = y(Tsn)
= A cos(2p(-f0 + lfs)Tsn + f)
x(t) = A cos(-2p f0 t + f) = A cos(-2pf0Tsn + 2plfsTsn + f)
= A cos(-2pf0Tsn + 2pln + f)
• Sample at Ts = 1/fs = A cos(-2pf0Tsn + f)
x[n] = x(Tsn) = A cos(-2p f0 Ts n + f) = x[n]
Here, fsTs = 1
• Keeping the sampling period
same, sample Since l is an integer,
y(t) = A cos(2p (-f0 + l fs) t + f) cos(x + 2 p l) = cos(x)

where l is an integer • y[n] indistinguishable from x[n]

5-6
Spectrum of a Discrete-Time Signal
• Spectrum of a sinusoidal signal ½

( )
cos ŵ0n = 0.5 e( - jŵ0n
+e
jŵ0n
) ŵ
Frequency components at-ŵ 0 andŵ 0 -ŵ0 ŵ 0
• Example: Two of the aliases for x1[n] = cos(0.4pn)
x2[n] = cos(2.4pn) = cos(0.4pn + 2pn) = cos(0.4pn) = x1[n]
x3[n] = cos(-1.6pn) = cos(0.4pn − 2pn) = cos(0.4pn) = x1[n]
Only need to add principal frequency components for synthesis
½

ŵ
-2.4p -1.6p -0.4p 0.4p 1.6p 2.4p

Aliases Principal Components Aliases 5-7

Ideal Reconstruction
• Sampling theorem
Possible to reconstruct continuous-time signal from samples
Does not say how to do it
• Reconstruction: Ideal case
Need a formula for signal of one or more sinusoids
y(t) = y[n] n= f t for - ¥ < n < ¥ Practical?
s

Consider y[n] = A cos(2 p f0 n Ts + f) where fs > 2 f0

y(t) = y[n] n= f t = Acos(2p f0t + f ) for -¥ < t < ¥
s
• Reconstruction: Practical case
Interpolation fills in values between samples
There are other approaches to “undo” sampling process 5-8
Discrete-to-Continuous Conversion
• Input: sequence of samples y[n] %% y[n] is sinusoid w/
%% period of 8 samples
• Output: continuous-time signal y(t) via interpolation f0 = 0.125;
(“connecting the dots”) fs = 1;
wHat = 2*pi*f0/fs;
• Example: Cosine of frequency f0 n = 0 : 8;
yofn = cos(wHat*n);
From sampling theorem, fs > 2 f0 or f0 < ½ fs
y[n] = A cos(2p f0 Ts n +f) y(t) = A cos(2p f0 t +f)
fs = 20*f0;
~ Ts = 1/fs;
t = 0 : Ts : (1/f0);
yoft = cos(2*pi*f0*t);

figure;
stem(n, yofn);
figure;
stem(n, yofn);
hold
plot(t, yoft);

5-9
 Sampling– SPFirst Sec. 4-1.5

Interpolation From Tables

• Using mathematical tables of x f(x)
numeric values of functions to
0 0.0
compute a value of the function
1 1.0
• Estimate f(1.5) from table 2 4.0
Zero-order hold: take value to be f(1)
to make f(1.5) = 1.0 (“stairsteps”) 3 9.0
Linear interpolation: average values of 9
~
f ( x)
nearest two neighbors to get f(1.5) = 2.5
Curve fitting: fit four points in table to
4
polynomal a0 + a1 x + a2 x2 + a3 x3
which gives f(1.5) = x2 = 2.25 1 x
0 1 2 3

5-10