ADVANCED DIGITAL

COMMUNICATIONS
Lecture #3: Sampling, Quantization
and Encoding
Introduction to Digital
Communications
 Advantages
 Immunity to noise and distortion compared
to analog systems
 Digital circuits are more reliable and cheaper
 Different kinds of signals e.g. voice, media,
data etc are treated identically since a bit is
still a bit
 Error detection and correction (channel
coding) available
 Better suited for signal processing functions
against interference, jamming, privacy and
encryption
 Disadvantages
 Uses more bandwidth
Introduction to Digital
Communications
Introduction to Digital
Communications
Formatting
 Transmit and Receive Formatting
 Transition from information source  digital
symbols  information sink

5
Formatting
 Transforming an analog signal into a
digital signal compatible with digital
communication system
 Sampling
 Quantization
 Symbol to bit mapper (PCM)
SYMBOL TO
BIT MAPPER
SAMPLING QUANTIZATION
101110….
Sampling
 Sampling is the processes of converting continuous-
time analog signal, x(t), into a discrete-time signal by
taking the “samples” at discrete-time intervals
 Sampling analog signals makes them discrete in time
but still continuous valued
 If done properly (Nyquist theorem is satisfied),
sampling does not introduce distortion
 Sampled values:
 The value of the function at the sampling points
 Sampling interval:
 The time that separates sampling points (interval b/w
samples), Ts
 If the signal is slowly varying, then fewer samples per
second will be required than if the waveform is rapidly
varying
 So, the optimum sampling rate depends on the
maximum frequency component present in the signal
Sampling
 Sampling Rate (or sampling frequency fs):
 The rate at which the signal is sampled, expressed
as the number of samples per second (reciprocal of
the sampling interval), 1/Ts = fs
 Nyquist Sampling Theorem (or Nyquist
Criterion):
 If the sampling is performed at a proper rate, no
info is lost about the original signal and it can be
properly reconstructed later on
 Statement:

“If a signal is sampled at a rate at least, but not

exactly equal to twice the max frequency
component of the waveform, then the waveform
can be exactly reconstructed from the samples
without any distortion”
f s 2 f max
Sampling
 3 types of sampling
 Impulse Sampling (a.k.a. Ideal Sampling)
 Natural Sampling
 Sample and Hold (Simplest and Most
Popular)
Ideal Sampling

x(t )


x (t )    (t  nTs )
n 

xs (t )  x(t ).x (t )  x(t )   (t  nTs )
n 

  x(nTs ) (t  nTs )
n 
Ideal Sampling

X(f )


1
X ( f ) 
Ts
  ( f  nf )
n 
s

X s ( f ) X ( f ) * X ( f )

1
X ( f ) *
Ts
 ( f  nf s )
n 

1

Ts
 X(f  nf s )
n 
Convolution Examples
Convolution Examples

13
Convolution Examples

14
Convolution Examples

15
Natural Sampling

x(t )

 
x p (t ) P (t ) *   t  nT    Pt  nT 
n  n 

where P(t) is a rectangle pulse with width T


xs (t )  x(t )  Pt  nT 
n 
Natural Sampling

X(f )


1 k  k
X p( f )   P    f  
k  T T   T
1 1

T T


1 k  k

X s ( f ) X s ( f ) 
k  T
P   
T  
 f  
T
1 1 

1 k  k
  P  X  f  
T T

k  T T   T
 Sample and Hold

p (t )

p (t )

x (t )


1
xs (t )  p (t ) * [ x(t ).x (t )] X s ( f ) P( f )
Ts
 X ( f  nf )
n 
s

 
1
 p (t ) * [ x(t )   (t  nTs )] Tssinc( fTs ).
Ts
 X ( f  nf )
s
n 
n 
Sample and Hold
Sampling Theorem
 A bandlimited signal having no spectral
fm
components above hertz can be
determined uniquely by values sampled
at uniform intervals
T 
1 of
s
2 fm

 or in terms of the sampling rate, called

f s 2 f m
the Nyquist criterion

 Nyquist rate f s 2 f m
 Recovering the Analog
Signal
 One way of recovering the original signal from sampled signal
Xs(f) is to pass it through a Low Pass Filter (LPF) as shown below

 If fs > 2B then we recover x(t) exactly

 Else we run into some problems and signal is
not fully recovered
Aliasing
 Undersampling and Aliasing
 If the waveform is undersampled (i.e. fs < 2B)
then there will be spectral overlap in the sampled
signal
Aliasing effect
 The signal at the output of the filter will be
different from the original signal spectrum

This is the outcome of aliasing!

 This implies that whenever the sampling
condition is not met, an irreversible overlap
of the spectral replicas is produced
Aliasing effect

24
Aliasing effect
 This could be due to:
1. x(t) containing higher frequency than were
expected
2. An error in calculating the sampling rate
 Under normal conditions, undersampling of
signals causing aliasing is not recommended
Solution : Anti-Aliasing
Analog Filter
 All physically realizable signals are not
completely bandlimited
 If there is a significant amount of energy in
frequencies above half the sampling
frequency (fs/2), aliasing will occur
 Aliasing can be prevented by first passing the
analog signal through an anti-aliasing filter
(also called a prefilter) before sampling is
performed
 The anti-aliasing filter is simply a LPF with
cutoff frequency equal to half the sample rate
Solution : Anti-Aliasing
Analog Filter
 Aliasing is prevented by forcing the
bandwidth of the sampled signal to satisfy
the requirement of the Sampling Theorem
 Quantization
 Amplitude quantizing: Mapping samples of a
continuous amplitude waveform to a finite set of
amplitudes. Out

In
Quantized

where q=(Vp-(-
values

Vp)) / L
= 2Vp / L
= Vpp / L
 Uniform Quantization
 A quantizer with equal quantization level is a Uniform
Quantizer
 Each sample is approximated within a quantile interval
 Uniform quantizers are optimal when the input
distribution is uniform
 i.e. when all values within the range are equally
likely

 q
Most ADC’s are implemented using uniform q
quantizers
 e
 Error of a uniform quantizer is bounded2by 2
Uniform Quantization
 If we assume that the quantization error, e, the mean-
squared value (noise variance) of the quantization
error is given by:
q / 2
MSE  2
 e 2 p (e)de
 q/2
2
1
q / 2 q
 e 2 de 
 q/2 q 12

 where p(e) = 1/q is the (uniform) probability density

function of the quantization error. The variance, σ2,
corresponds to the average quantization noise power.
Uniform Quantization
 The peak power of the analog signal (normalized
to 1Ω) can be expressed as
2 2
2  V pp   Lq  L2 q 2
Vp     
 2   2  4

 where L is the number of quantization levels.

 The ratio of peak signal power to average
quantization noise power (S/N)q , assuming that
there are no errors due to lSI or channel noise:
S L2 q 2 / 4
   2 3L2
 N  q q / 12
Types of Quantizers

Uniform Non-uniform
Types of Quantizers
Non-uniform Quantization
 Nonuniform quantizers have unequally
spaced levels
 The spacing can be chosen to optimize the Signal-to-

Noise Ratio for a particular type of signal

 It is characterized by:
 Variable step size
 Quantizer size depend on signal size
Non-uniform Quantization
 Many signals such as speech have a
nonuniform distribution
 Basic principle is to use more levels at
regions with large probability density
function (pdf)
 Concentrate quantization levels in areas of
largest pdf
 Or use fine quantization (small step size)
for weak signals and coarse quantization
(large step size) for strong signals
Non-uniform Quantization
 In speech, weak signals are more frequent
than strong ones.
 Using equal step sizes (uniform quantizer) S
 
 N q
gives low for weak S signals

 and high
N
for strong signals. q
Non-uniform Quantization
 Companding is a method of reducing the
number of bits required in ADC while
achieving an equivalent dynamic range or
Signal-to-Quantization-Noise Ratio (SQNR)
 In order to improve the resolution of weak
signals within a converter, and hence
enhance the SQNR, the weak signals need
to be enlarged, or the quantization step
size decreased, but only for the weak
signals
 But strong signals can potentially be reduced
without significantly degrading the SQNR or
alternatively increasing quantization step
size
Nonuniform Quantization
 The compression process at the
transmitter must be matched with an
equivalent expansion process at the
receiver
Non-uniform Quantization
 After compression, input to the
quantizer will have a more uniform
distribution after sampling
Companding
 At the receiver, the signal is expanded
by an inverse operation
 The process of COMpressing and
exPANDING the signal is called
companding
 Companding is a technique used to
reduce the number of bits required in
ADC or DAC while achieving comparable
SQNR
 Companding

compressing + expanding companding

y C (x) x̂
x(t ) y (t ) yˆ (t ) xˆ (t )

x ŷ
Compress Qauntize Expand
Transmitter Channel Receiver
Companding
 Basically, companding introduces a
nonlinearity into the signal
 This maps a non-uniform distribution into
something that more closely resembles a
uniform distribution
 A standard ADC with uniform spacing between
levels can be used after the compandor (or
compander)
 The companding operation is inverted at the
receiver
 There are in fact two standard logarithm
based companding techniques
 US standard called µ-law companding
 European standard called A-law companding
 -Law Companding
Standard
  x 
log e  1     
 x 
  max  
y  ymax sgn( x)
log e 1   
 1 x 0
where sgn x  
 x and y represent the input and output voltages  1 x0
  is a constant number determined by experiment

 In the U.S., telephone lines uses companding with  = 255

 Samples 4 kHz speech waveform at 8,000 sample/sec

 Encodes each sample with 8 bits, L = 256 quantizer levels
 Hence data rate R = 64 kbit/sec
  = 0 corresponds to uniform quantization
 -Law Companding
Characteristics
A-Law Companding
Standard
  x 
 A 

 y x
 max  sgn( x) x 1
0 
 1  log e A
max
xmax A
y 
  x 
1  log e  A 
  xmax  1 x
 ymax sgn( x)  1
 1  log e A A xmax
 where
 x and y represent the input and output voltages
 A = 87.6
 A is a constant number determined by
experiment
A -Law Companding
Characteristics
Pulse Code Modulation
(PCM)
 Pulse Code Modulation refers to a digital
baseband signal that is generated
directly from the quantizer output
 Sometimes the term PCM is used
interchangeably with quantization
Pulse Code Modulation
(PCM)
Encoding (PCM)
 A uniform linear quantizer is called Pulse
Code Modulation (PCM).
 Pulse code modulation (PCM): Encoding the
quantized signals into a digital word (PCM
word or codeword).
 Each quantized sample is digitally encoded
into an l bits codeword where L in the
number of quantization levels and
 Encoding
Example
amplitude
x(t)
111 3.1867

110 2.2762 Quant. levels

101 1.3657

100 0.4552

011 -0.4552 boundaries

010 -1.3657

001 -2.2762 x(nTs): sampled values

xq(nTs): quantized values
000 -3.1867
Ts: sampling time
PCM t
codeword 110 110 111 110 100 010 011 100 100 011 PCM sequence
Encoder Design
 Algorithm
 Vin value lies
between two
comparators
 ∆V = Vref/2N
 N = Encoder
output bits
 Comparators
≥2N-1
 Encoder Design Example
Vin = 5.5V, Vref= 8V

Vin lies in between Vcomp5 & Vcomp6 0

Vcomp5 = Vref*5/8 = 5V
0
Vcomp6 = Vref*6/8 = 6V
1
Comparator 1 - 5 => output 1 1
Comparator 6 - 7 => output 0
1
Encoder Octal Input = sum(0011111)
=5
1
Encoder Binary Output = 1 0 1 1
5.5V