3.

8 Digital Processing of Continuous-Time

Signals
– Sampling of Continuous-Time Signals
– Recovery of the Analog Signal
– Implications of the Sampling Process

3.9 Sampling of Band-pass Signals

电子科技大学 通信与信息工程学院
宽带通信网络实验室 1 2023/3/14
 Anti-
xa (t ) A/ Digital D/ Reconstruction
aliasing S/H ya (t )
D Processor A filter
filter

xa (t ) x[n] y[n]
Ideal y a (t )
Ideal Digital
inter-
sampler Processor
polation

Fig 4.1,4.2 Digital Processing of A Continuous Signal

Some concepts:
1. ADC: Analog-to-Digital Converter
2. DAC: Digital-to-Analog Converter
3. S/H : Sample-and-Hold (circuit).
4. Anti-aliasing filter.
5. Reconstruction or interpolation filter.
电子科技大学 宽带通信网络实验室 2 2023/3/14
3.8.1 Effect of Sampling in the Frequency-Domain
g a (t ) g p (t )
g a (t ) g p (t )
p (t )
(a) Ideal sampling (b)

g a (t ) g a (t )
t t
0 0
t
p(t） p(t）= d T (t )
t t
0 T 0 T
g p (t )
g p (t )
t t
0 0
电子科技大学 宽带通信网络实验室 3 2023/3/14
 • When τ« T, p(t) ~δT(t);
so we firstly discuss ideal sampling:
¥
 d（t T ) ＝ å d (t - nT )
n =-¥

The output of ideal sampling is:

(1)In time domain: ¥
g p (t ) = g a (t )d T (t ) = åg
n = -¥
a ( nT )d (t - nT )

(2)In frequency domain: ¥

1 1
G p ( j W) =
2p
Ga ( jW) * DT ( jW) =
T
å G ( j (W - k W ) )
k =-¥
a T

1 ¥
æ 2p ö
=
T
å
k =-¥
Ga ç j (W - k
è
)÷
T ø
电子科技大学 宽带通信网络实验室 4 2023/3/14
 • There are two different forms of Gp(jW):

å
¥ - jWnT
(1)G p ( jW) = n =-¥
g a (nT ) e
¥
1
(2)G p ( jW) =
T
å G ( j (W - k W ) )
k =-¥
a T

Gp(jΩ) is a periodic function of Ω consisting of a sum of shifted

and scaled replicas of Ga(jΩ) , shifted by integer multiples of ΩT
and scaled by 1/T.

电子科技大学 通信与信息工程学院
宽带通信网络实验室 5 2023/3/14
 The Discretization The periodization
In time Domain In Frequency Domain

电子科技大学 通信与信息工程学院
宽带通信网络实验室 6 2023/3/14
 p The frequency procedure of sampling
• Assume ga(t) is a band-limited signal with a CTFT
Ga(jW) as shown below:

• The spectrum P(jW) of p(t) having a sampling period T=2p/WT

is indicated below:

电子科技大学 宽带通信网络实验室 7 2023/3/14

 • Two possible spectra of Gp(jW) are shown below:

WT>2 Wm

no overlap

WT<2 Wm
overlap/
aliasing

电子科技大学 宽带通信网络实验室 8 2023/3/14

Ø Sampling Theorem

• Let ga(t) be a band-limited signal with CTFT Ga(jW)=0 for | W|> Wm ,

Then ga(t) is uniquely determined by its samples ga(nT) , -¥£n£¥ if

WT ³ 2 Wm , where WT=2p/T

• The condition WT ³ 2 Wm is often referred to as the Nyquist condition.

• The frequency WT/2 is usually referred to as the folding frequency.

电子科技大学 宽带通信网络实验室 9 2023/3/14

• The highest frequency Wm contained in ga(t) is
usually called the Nyquist frequency since it
determines the minimum sampling frequency WT
=2Wm that must be used to fully recover ga(t) from
its sampled version.
• The frequency 2Wm is called the Nyquist rate.

电子科技大学 宽带通信网络实验室 10 2023/3/14

• Oversampling - The sampling frequency is higher
than the Nyquist rate;
• Undersampling - The sampling frequency is lower
than the Nyquist rate;
• Critical sampling - The sampling frequency is
equal to the Nyquist rate;
• Note: A pure sinusoid may not be recoverable
from its critically sampled version.

电子科技大学 宽带通信网络实验室 11 2023/3/14

 Nyquist frequency Examples:

• In digital telephony, a 3.4 kHz signal bandwidth is acceptable

for telephone conversation;
• Here, a sampling rate of 8 kHz, which is greater than twice the
signal bandwidth, is used.
• In high-quality music signal processing, a bandwidth of 20
kHz has been determined to preserve the fidelity;
• Hence, in compact disc (CD) music systems, a sampling rate of
44.1 kHz, which is slightly higher than twice the signal
bandwidth, is used.

电子科技大学 宽带通信网络实验室 12 2023/3/14

The Relationship Between G(ejω) and Gp(jΩ)

• We now derive the relation between the DTFT of

g[n] and the CTFT of gp(t)
• To compare: ¥
jw
G(e ) = å g[n]e
n =-¥
-jw n

with ¥
G p (jW) = å
n =-¥
g a (nT )e-jWnT

And make use of g[n]=ga(nT), -∞<n<∞

电子科技大学 宽带通信网络实验室 13 2023/3/14

The Relationship Between G(ejω) and Gp(jΩ)
• Observation: We have
G(ejω)=Gp(jΩ)|Ω=ω/T
Or, equivalently,
Gp(jΩ)= G(ejω)|ω=ΩT
From the above observation and
1 ¥
G p ( jW) = å G a ( j (W - k WT ))
T k =-¥

电子科技大学 宽带通信网络实验室 14 2023/3/14

The Relationship Between G(ejω) and Gp(jΩ)
• We arrive at the desired result:
¥
1
G(e jw ) = å Ga ( jW - jk WT ) |W=w / T
T k =-¥
1 ¥ w
= å Ga ( j - jk WT )
T k =-¥ T

1 ¥
w 2p k
=
T
å
k =-¥
Ga ( j
T
- j
T
)

电子科技大学 宽带通信网络实验室 15 2023/3/14

The Relationship Between G(ejω) and Gp(jΩ)
• The relation derived on the previous slide can be
alternately expressed as
¥
1
G(e jWT
) = å Ga ( jW - jk WT )
T k =-¥
jw
From G(e ) = G p ( jW) |W=w / T
jw
Or from G p ( jW) = G(e ) |w =WT
It follows that G(ejω) is obtained from Gp(jΩ) by
applying the mapping Ω=ω/T
电子科技大学 宽带通信网络实验室 16 2023/3/14
• Now, the CTFT Gp(j Ω) is a periodic function of
Ω with a period ΩT = 2π/T

• Because of the mapping, the DTFT G(ejw) is a

periodic function of ω with a period 2π

电子科技大学 宽带通信网络实验室 17 2023/3/14

 Supplement: The practical sampling
• In practical condition, p(t) is a periodic rectangle
impulse with τ-width .
¥
p (t ) = å
k =-¥
Ck e jk W s t
k W st
sin( ) - j k W st
1 T /2 1 t t 2
ò ò
- jk W s t
Ck = p (t )e dt = e - jk W s t
dt =
k W st
e 2

T -T / 2 T 0 T
2
k W st
If the τ and T are fixed, then when k varies, sin(
2
)
the magnitude of Ck varies with: k W st
2
电子科技大学 宽带通信网络实验室 18 2023/3/14
Supplement: The practical sampling
• So we can see, in practical sampling.

¥
\ G p ( j W) = åC
k =-¥
k Ga ( jW - jk W s )

Ga ( jW)

W
0
G p ( jW) Wt Wt
sin( )
2 2

- Ws 0 Ws 2W s W

电子科技大学 宽带通信网络实验室 19 2023/3/14

3.8.2 Recovery of the Analog Signal

By filtering in frequency domain or interpolation in time

domain, the continuous signal can be recovered from The
Sampled Discrete signal.

• Here, the filter is called reconstruction filter or

interpolation filter or smoothing filter.
电子科技大学 宽带通信网络实验室 20 2023/3/14
• The spectra of the filter and pertinent signals are
shown below:

电子科技大学 宽带通信网络实验室 21 2023/3/14

 • If WT>2 Wm , ga(t) can be recovered exactly from gp(t)
by passing it through an ideal lowpass filter Hr(jW)
with a gain T and a cutoff frequency Wc greater than
Wm and less than WT - Wm;

• On the other hand, if WT< 2 Wm , due to the overlap of

the shifted replicas of Ga(jW) , the spectrum Gp(jW)
cannot be separated by filtering to recover Ga(jW). The
distortion caused by a part of the replicas outside the
baseband folded back or aliased into the baseband.

电子科技大学 宽带通信网络实验室 22 2023/3/14

 ^g (t )
• We now derive the expression for the output a of the
ideal lowpass reconstruction filter Hr(jW) as a function of
the samples g[n].
• The impulse response hr(t) of the ideal lowpass
reconstruction filter is obtained by taking the inverse
CTFT of Hr(jW)
ì T, W £ Wc
H r ( jW) = í
î 0, W > Wc

电子科技大学 宽带通信网络实验室 23 2023/3/14

 • Thus, the impulse response is given by
h r (t ) = 1 ¥ H ( jW) e jWt dW = T Wc e jWt dW
ò
2 p -¥ r ò
2 p -Wc
sin(Wct )
= , -¥ £t £ ¥
WT t / 2
• The input to the lowpass filter is the impulse
train gp(t):
¥
g p (t ) = ån=-¥ g[n] d(t - nT )

电子科技大学 宽带通信网络实验室 24 2023/3/14

 ^ (t ) of the ideal lowpass
• Therefore, the output g a
filter is given by:
¥
gˆ a (t ) = h r(t )* gp(t ) = å
n =-¥
g[ n]h r(t - nT )

Applying hr(t)=sin(Wct)/(WTt/2), and assuming:

Wc= WT/2= p/T , we get:
¥
sin[p (t - nT ) / T ]
g a (t ) = å g[n]
^
n = -¥ p (t - nT ) / T

电子科技大学 宽带通信网络实验室 25 2023/3/14

 h(t - nT ) • g a (nT )

g^a (t ) is called
Poisson sum
formula or
g a (t ) interpolation
formula.

电子科技大学 宽带通信网络实验室 26 2023/3/14

 3.8.3 Implications of the Sampling Process

• Example: Consider the three continuous time

sinusoidal signals:
g1 (t ) = cos( 6 pt )
g 2 (t ) = cos(14 pt )
g 3 (t ) = cos( 26pt )
• Their corresponding CTFTs are:
G1 ( jW) = p[d(W - 6p) + d(W + 6p)]
G2 ( jW) = p[d(W - 14 p) + d(W + 14 p)]
G3 ( jW) = p[d(W - 26p) + d(W + 26p)]
电子科技大学 宽带通信网络实验室 27 2023/3/14
 • These three transforms are plotted below

电子科技大学 宽带通信网络实验室 28 2023/3/14

 • These continuous-time signals sampled at a rate of
T = 0.1 sec, i.e., with a sampling frequency WT =20p
rad/sec
• The sampling process generates the continuous-time
impulse trains, g1p(t), g2p(t) , and g3p(t)
• Their CTFTs are given by:
¥
Gp ( jW) = 10åk =-¥ G ( j (W - kWT ) ), 1£  £ 3

电子科技大学 宽带通信网络实验室 29 2023/3/14

 • Plots of the 3 CTFTs are shown below:

{Wk } = {±W0 ± k WT } = {±6p ± 20p ´ k}

电子科技大学 宽带通信网络实验室 30 2023/3/14
• These figures also indicate by dotted lines the frequency
response of an ideal lowpass filter with a cutoff at
Wc=WT/2=10p and a gain T=0.1.
• In fact, the 3 discrete-time sinusoidal signals are:
g 1[ n ] = g 1 ( nT ) = cos( 6p n / 10 ) = cos( 0 .6p n )
g 2 [ n ] = g 2 ( nT ) = cos(14p n / 10 ) = cos(1 .4p n )
g 3 [ n ] = g 3 ( nT ) = cos( 26p n / 10 ) = cos( 2 .6p n )
And g1[ n ] = g 2 [ n ] = g 3 [ n ]
电子科技大学 宽带通信网络实验室 31 2023/3/14
 • If reconstruction filter
ì 0.1, (6 - D)p £ W £ (6 + D)p
H r ( jW) = í
î 0, otherwise,
^
g a (t ) = cos(6p t )
• If reconstruction filter
ì 0.1, (34 - D)p £ W £ (34 + D)p
H r ( jW) = í
î 0, otherwise,
^
g a (t ) = cos(34p t )
电子科技大学 通信与信息工程学院
宽带通信网络实验室 32 2023/3/14
 3.9 Sampling of Band-pass Signals
• The conditions developed earlier for the unique
representation of a continuous-time signal by the
discrete-time signal obtained by uniform
sampling assumed that the continuous-time
signal is bandlimited in the frequency range
from DC to Wm --- lowpass signal.
• There are continuous-time signals which are
bandlimited to a higher frequency range WL£ |W|
£ WH with WL >0 --- bandpass signals.

电子科技大学 宽带通信网络实验室 33 2023/3/14

• To prevent aliasing a bandpass signal can of course be
sampled at a rate greater than twice the highest
frequency, i.e. by ensuring WT ³2 WH
• However, due to the bandpass spectrum of the
continuous-time signal, the spectrum of the discrete-
time signal obtained by sampling will have spectral
gaps with no signal components present in these gaps.
• Moreover, if WH is very large, the sampling rate also
has to be very large which may not be practical in
some situations.
电子科技大学 宽带通信网络实验室 34 2023/3/14
• Let DW= WH - WL define the bandwidth of the
bandpass signal.
• Assume: WH = M(DW)
• We choose the sampling frequency WT to satisfy the
condition:
WT = 2(DW) = 2WH/M
• WT is smaller than 2WH , the Nyquist rate

undersampling
电子科技大学 宽带通信网络实验室 35 2023/3/14
 ¥
G p ( jW ) = 1
T å Ga ( j (W - k WT ))
k = -¥
= 1
å¥ Ga ( jW - j 2k ( DW ) )
T k = -¥
• As before, Gp(jW) consists of a sum of Ga(jW) and
replicas of Gp(jW) shifted by integer multiples of twice
the bandwidth Δω and scaled by 1/T;
• The amount of shift for each value of k ensures that
there will be no overlap between all shifted replicas.
no aliasing
电子科技大学 宽带通信网络实验室 36 2023/3/14
 • Figures below illustrate the idea:
Ga ( jW)

W
- WH - WL 0 WL WH

(M=3)
G p ( jW) no aliasing

W
- WH - WL 0 WL WH
电子科技大学 通信与信息工程学院
宽带通信网络实验室 37 2023/3/14
 • Figures below illustrate the idea:
Ga ( jW)

- WH WL WH
W
- WL 0

(M=4) G p ( jW) no aliasing

W
- WH - WL 0 WL WH

电子科技大学 通信与信息工程学院
宽带通信网络实验室 38 2023/3/14
 Discuss:
(1)If ga(t) is bandlimited with its frequency band in (fL,
fH) .
(2)If using sampling rate: f = 2( f + f ) 4 f
s
L H
= 0

2n + 1 2n + 1

Where, n is a maximum integer which meets

fs≥2(fH-fL)=2B (n = 0,1,2……)
Then, the sampled signal ga(nTs) can be used to
represent original signal ga(t).

电子科技大学 宽带通信网络实验室 39 2023/3/14

Note:
• Before band-sampling, the signal in only one
frequency-band can be permitted.
• For lowest sampling rate, that is fs=2B, the signal
center frequency f0 should be:

2n + 1
f0 = B or f L + f H = (2 n + 1) B
2

电子科技大学 宽带通信网络实验室 40 2023/3/14

• The band-sampling result is moving the signal in the
frequency band of (nB, (n+1)B)(n=0,1,2,…) to the
frequency band of (0, B).
• When n is odd number, the relationship between them is
reversal; while n is even, the relationship is opposite.

电子科技大学 宽带通信网络实验室 41 2023/3/14

• As can be seen, ga(t) can be recovered from gp(t) by
passing it through an ideal bandpass filter with a
passband given by WL£ |W| £ WH and a gain of T.
• Note: Any of the replicas in the lower frequency bands
can be retained by passing through bandpass filters
with passbands
WL- k(DW) £ |W| £ WH - k(DW) , 1 £ k £ M-1
providing a translation to lower frequency ranges.

电子科技大学 宽带通信网络实验室 42 2023/3/14

Homework
• 3.11,3.17, 3.18(a)(c)(e), 3.25, 3.58(a)(c)(e), 3.62,
3.63, 3.66(a)(b)
• M3.1, M3.2

电子科技大学 宽带通信网络实验室 43 2023/3/14