EE330 Digital Signal Processing
Sampling of Continuous-Time Signals
Arbab Latif
Spring 2024
Resources:
Discrete Time Signal Processing, A. V. Oppenheim and R. W. Schaffer, 3rd Edition, 2010 (Chapter 4.1-4.3)
2 Overview:
• Periodic Sampling
• Frequency Domain Representation of Sampling
• Reconstruction of a Bandlimited Signal from Its
Samples
3 Signal Types
• Analog signals: continuous in time and amplitude
– Example: voltage, current, temperature,…
• Digital signals: discrete both in time and amplitude
– Example: attendance of this class, digitizes analog
signals,…
• Discrete-time signal: discrete in time, continuous in amplitude
– Example: hourly change of temperature
• Theory for digital signals would be too complicated
– Requires inclusion of nonlinearities into theory
• Theory is based on discrete-time continuous-amplitude
signals
– Most convenient to develop theory
– Good enough approximation to practice with some care
• In practice we mostly process digital signals on processors
– Need to take into account finite precision effects
4 Periodic/ Uniform Sampling
• Sampling is a continuous to discrete-time conversion
-3 -2 -1 0 1 2 3 4
• Most common sampling is periodic
𝑥 𝑛 = 𝑥𝑐 𝑛𝑇 −∞ < 𝑛 < ∞
• T is the sampling period in second
• fs = 1/T is the sampling frequency in Hz
• Sampling frequency in radian-per-second s=2fs rad/sec
• Use [.] for discrete-time and (.) for continuous time signals
• This is the ideal case not the practical but close enough
– In practice it is implemented with an analog-to-digital converters
– We get digital signals that are quantized in amplitude and time
5 Periodic/ Uniform Sampling
• Sampling is, in general, not reversible
• Given a sampled signal one could fit infinite continuous signals
through the samples
0.5
-0.5
-1
0 20 40 60 80 100
• Fundamental issue in digital signal processing
– If we lose information during sampling, we cannot recover it
• Under certain conditions an analog signal can be sampled without
loss so that it can be reconstructed perfectly
7
Sampling with a Periodic Impulse
Train
• Figure(a) is not a representation of any physical circuits, but it is
convenient for gaining insight in both the time and frequency
domain.
+∞
𝑠(𝑡) = 𝛿(𝑡 − 𝑛𝑇)
𝑛=−∞
(a) Overall system
(b) xs(t) for two sampling rates
9
Frequency Domain Representation
of Sampling
+∞
𝑥𝑠 (𝑡) = 𝑥𝑐 (𝑡)𝑠(𝑡) = 𝑥𝑐 (𝑡) 𝛿(𝑡 − 𝑛𝑇) (𝑀𝑜𝑑𝑢𝑙𝑎𝑡𝑖𝑜𝑛)
𝑛=−∞
+∞
𝑥𝑠 (𝑡) = 𝑥𝑐 (𝑛𝑇)𝛿(𝑡 − 𝑛𝑇) (𝑆ℎ𝑖𝑓𝑡𝑖𝑛𝑔 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑦)
𝑛=−∞
• Let us now consider the Fourier transform of xs(t):
𝐹𝑜𝑢𝑟𝑖𝑒𝑟 𝐹𝑜𝑢𝑟𝑖𝑒𝑟
• If 𝑠(𝑡) 𝑆(𝑗Ω) and 𝑥𝑐 (𝑡) 𝑋𝐶 (𝑗Ω)
∞
2𝜋
𝑆(𝑗Ω) = 𝛿(Ω − 𝑘Ω𝑠 ) where Ω𝑠 = 2𝜋/𝑇 is the sampling rate in radians/s.
𝑇
𝑘=−∞
∞
1 1
𝑋𝑠 (𝑗Ω) = 𝑋𝑐 (𝑗Ω) ∗ 𝑆(𝑗Ω) = 𝑋𝑐 𝑗(Ω − 𝑘Ω𝑠 )
2𝜋 𝑇
𝑘=−∞
10
Frequency Domain Representation
of Sampling
• By applying the continuous-time Fourier transform to
equation +∞
𝑥𝑠 (𝑡) = 𝑥𝑐 (𝑛𝑇)𝛿(𝑡 − 𝑛𝑇)
𝑛=−∞
We obtain +∞
𝑋𝑆 (𝑗Ω) = 𝑥𝑐 (𝑛𝑇)𝑒 −𝑗Ω𝑇𝑛
𝑛=−∞
+∞
𝑥[𝑛] = 𝑥𝑐 (𝑛𝑇) 𝑎𝑛𝑑 𝑋(𝑒 𝑗𝜔 ) = 𝑥[𝑛]𝑒 −𝑗𝜔𝑛
𝑛=−∞
consequently
∞
1 𝜔 2𝜋𝑘
𝑋𝑠 (𝑗Ω) = 𝑋(𝑒 𝑗𝜔 )ቚ = 𝑋(𝑒 𝑗Ω𝑇 ) ⇒ 𝑋(𝑒 𝑗𝜔 ) = 𝑋𝑐 𝑗 −
𝜔=Ω𝑇 𝑇 𝑇 𝑇
𝑘=−∞
11
Exact Recovery of Continuous-Time
from Its Samples
• (a) represents a band
limited Fourier
transform of xc(t)
Whose highest nonzero
frequency is N .
• (b) represents a
periodic impulse train
with S frequency.
• (c) shows the output of
impulse modulator in
the case
S − N N S 2 N
12
Exact Recovery of Continuous-Time
from Its Samples
• In this case X C ( j )
don’t overlap
• therefore xc(t) can be
recovered from xs(t)
with an ideal low pass
filter H r ( j ) with gain
T and cutoff frequency
N C S − N
• It means X r ( j ) = X C ( j )
=
13 Aliasing Distortion
• (a) represents a band
limited Fourier
transform of xc(t)
Whose highest nonzero
frequency is N .
• (b) represents a
periodic impulse train
with S frequency.
• (c) shows the output of
impulse modulator in
the case
S − N N S 2 N
14 Aliasing Distortion
• In this case the copies of X C ( j ) overlap and is not longer
recoverable by lowpass filtering therefore the reconstructed signal
is related to original continuous-time signal through a distortion
referred to as aliasing distortion.
15
Example: The effect of aliasing in the
sampling of cosine signal
• Suppose x c (t ) = cos(0t )
16 Nyquist Sampling Theorem
• Sampling theorem describes precisely how much information is
retained when a function is sampled, or whether a band-limited
function can be exactly reconstructed from its samples.
• Sampling Theorem: Suppose that 𝑥𝑐 (𝑡) 𝑋𝐶 (𝑗Ω) is band-limited
to a frequency interval −Ω𝑁 , Ω𝑁 , i.e., 𝑋𝐶 (𝑗Ω)
𝑋𝐶 (𝑗Ω) = 0 for Ω ≥ Ω𝑁
Ω
−Ω𝑁 0 Ω𝑁
Then xc(t) can be exactly reconstructed from equidistant samples
𝑥[𝑛] = 𝑥𝑐 (𝑛𝑇𝑠 ) = 𝑥𝑐 (2𝜋𝑛/Ω𝑠 ) Ω𝑠 > 2Ω𝑁
where 𝑇𝑠 = 2𝜋/Ω𝑠 is the sampling period, 𝑓𝑠 = 1/𝑇𝑠 is the sampling
frequency (samples/second), Ω𝑠 = 2𝜋/𝑇𝑠 is for radians/second.
17 Oversampled
• Suppose that x c (t ) X C ( j ) is band-limited:
X C ( )
A
0
− N N
• Then if T S is sufficiently small, X (e j
) appears as:
A X (e j )
Ts
− N T S N T S
− 2 −
0
2
• Condition: 2 − N T S N T S or N T S or S 2 N
18 Critically Sampled
Critically sampled: N T S = or S = 2 N
A X (e j )
Ts
− 2 − 0 2
According to the Sampling Theorem, in general the signal cannot be
reconstructed from samples at the rate T S = / N .
This is because of errors will occur if X c ( N ) 0 , the folded
frequencies will add at = .
Consider the case: x c (t ) = A sin( N t ) Aj ( − N ) − ( + N )
and note that for T S = / N .
x (nT s ) = A sin(c nT s ) = A sin(n ) = 0 (for all n )
19 Undersampled (aliased)
If sampling theorem condition is not satisfied Ω𝑁 𝑇𝑆 > 𝜋 or Ω𝑆 < 2Ω𝑁
𝐴
𝑋(𝑒 𝑗𝜔 )
𝑇𝑠
0
−2𝜋 −𝜋 𝜋 2𝜋
• The frequencies are folded - summed. This changes the shape of the
spectrum. There is no process whereby the added frequencies can be
discriminated - so the process is not reversible.
• Thus, the original (continuous) signal cannot be reconstructed exactly.
Information is lost, and false (alias) information is created.
• If a signal is not strictly band-limited, sampling can still be done at twice the
effective band-limited.
