Wireless Information Transmission System Lab.

Introduction to Digital
Communications System

Institute of Communications Engineering

National Sun Yat-sen University

Recommended Books
Digital Communications / Fourth Edition (textbook)
-- John G. Proakis, McGraw Hill
Communication Systems / 4th Edition
-- Simon Haykin, John Wiley & Sons, Inc.
Digital Communications Fundamentals and Applications /
2nd Edition
-- Bernard Sklar, Prentice Hall
Principles of Communications / Fifth Edition
-- Rodger E. Ziemer and William H. Tranter, John Wiley &
Sons, Inc.
Modern Digital and Analog Communication Systems
-- B.P. Lathi, Holt, Rinehart and Winston, Inc.
2

Example of Communications System

Local
Loop

Switch
Transmission
Equipment
Central Office

Local
Loop

Switch
Transmission
Equipment
Central Office

Local
Loop

Switch
Transmission
Equipment
Central Office

Mobile
Switching
Center

T1/E1 Facilities
regenerator

Base
Station

A/D Conversion
(Digitization)
T1/E1 Facilities
regenerator

SONET
SDH

M
U
X

T1/E1 Facilities

A/D Conversion
(Digitization)
T1/E1 Facilities
regenerator

Mobile
Switching
Center

A/D Conversion
(Digitization)

Public Switched Telephone Network (PSTN)

3

Base
Station

Basic Digital Communication Nomenclature

Textual Message: information comprised of a sequence of
characters.
Binary Digit (Bit): the fundamental information unit for all
digital systems.
Symbol (mi where i=1,2,M): for transmission of the bit
stream; groups of k bits are combined to form new symbol
from a finite set of M such symbols; M=2k.
Digital Waveform: voltage or current waveform representing
a digital symbol.
Data Rate: Symbol transmission is associated with a symbol
duration T. Data rate R=k/T [bps].
Baud Rate: number of symbols transmitted per second [baud].
4

Nomenclature Examples

Messages, Characters, and Symbols

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Multiplexing

Interleaving

Modulation

Frequency
Spreading

Multiple
Access

TX
RF
PA

si (t )
Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Demultiplexing

Deinterleaving

Channel Bits

Optional
Essential

To Other Destinations

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Wireless Information Transmission System Lab.

Format

Institute of Communications Engineering

National Sun Yat-sen University

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Multiplexing

Interleaving

Modulation

Frequency
Spreading

Multiple
Access

si (t )

Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

TX
RF
PA

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Demultiplexing

Deinterleaving

Channel Bits

Optional
Essential

To Other Destinations

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Formatting and Baseband Transmission

10

Sampling Theorem

11

Sampling Theorem
Sampling Theorem: A bandlimited signal having no
spectral components above fm hertz can be determined
uniquely by values sampled at uniform intervals of Ts
seconds, where
1
TS
or sampling rate f S 2 f m
2 fm
In sample-and-hold operation, a switch and storage
mechanism form a sequence of samples of the
continuous input waveform. The output of the sampling
process is called pulse amplitude modulation (PAM).
12

Sampling Theorem

1
X S ( f ) = X ( f ) X ( f ) =
TS
13

X ( f nf

n =

Spectra for Various Sampling Rates

14

Natural Sampling

15

Pulse Code Modulation (PCM)

PCM is the name given to the class of baseband
signals obtained from the quantized PAM signals by
encoding each quantized sample into a digital word.
The source information is sampled and quantized to
one of L levels; then each quantized sample is digitally
encoded into an -bit (=log2L) codeword.

16

Example of Constructing PCM Sequence

17

Uniform and Non-uniform Quantization

18

Statistical Distribution of Single-Talker

Speech Amplitudes
50% of the time, speech voltage is less than RMS.
Only 15% of the time, voltage exceeds RMS.
Typical voice signal dynamic range is 40 dB.

19

Problems with Linear Quantization

Fact: Unacceptable S/N for small signals.
Solution:
Increasing quantization levels price is too high.
Applying nonlinear quantization achieved by first
distorting the original signal with a logarithmic
compression characteristic and then using a uniform
quantizer.

At the receiver, an inverse compression characteristic,

called expansion, is applied so that the overall
transmission is not distorted. The processing pair is
referred to as companding.
20

Implementation of Non-linear Quantizer

21

Companding Characteristics
In North America: -law compression:
loge [1 + ( x / xmax )]
sgn x
y = ymax
loge (1 + )
where
+ 1 for x 0
sgn x =
1 for x < 0

In Europe: A-law compression:

A( x / x max )
sgn x
y max
1 + log e A

y=
y 1 + log e [ A( x / x max )] sgn x
max
1 + log e A
22

0<

x max
x
1
<
1
A x max

Compression Characteristics
Standard values of is 255 and A is 87.6.

23

Wireless Information Transmission System Lab.

Source Coding

Institute of Communications Engineering

National Sun Yat-sen University

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Multiplexing

Interleaving

Modulation

Frequency
Spreading

Multiple
Access

si (t )

Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

TX
RF
PA

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Deinterleaving

Demultiplexing

Channel Bits

Optional
Essential

To Other Destinations

25

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Source Coding
Source coding deals with the task of forming efficient
descriptions of information sources.
For discrete sources, the ability to form reduced data
rate descriptions is related to the information content
and the statistical correlation among the source
symbols.
For analog sources, the ability to form reduced data
rate descriptions, subject to a fixed fidelity criterion I
related to the amplitude distribution and the temporal
correlation of the source waveforms.
26

Huffman Coding
The Huffman code is source code whose average word
length approaches the fundamental limit set by the
entropy of a discrete memoryless source.
The Huffman code is optimum in the sense that no other
uniquely decodable set of code-words has smaller
average code-word length for a given discrete
memoryless source.

27

Huffman Encoding Algorithm

1. The source symbols are listed in order of decreasing
probability. The two source symbols of lowest
probability are assigned a 0 and a 1.
2. These two source symbols are regarded as being
combined into a new source symbol with probability
equal to the sum of the two original probabilities. The
probability of the new symbol is placed in the list in
accordance with its value.
3. The procedure is repeated until we are left with a final
list of source statistics of only two for which a 0 and a 1
are assigned.
4. The code for each (original) source symbol is found by
working backward and tracing the sequence of 0s and 1s
assigned to that symbol as well as its successors.
28

Example of Huffman Coding

Symbol
S0
S1
S2
S3
S4
Symbol
S0
S1
S2
S3
S4

Probability
0.4
0.2
0.2
0.1
0.1

Code Word
00
10
11
010
011

Stage 1

Stage 2

Stage 3

Stage 4

0.4
0.2
0.2
0.1 0
0.1
1

0.4
0.2
0.2 0
0.2 1

0.4
0.4 0
0.2
1

0.6 0
0.4
1

29

Properties of Huffman Code

Huffman encoding process is not unique.
Code words for different Huffman encoding process
can have different lengths. However, the average
code-word length is the same.
When a combined symbol is moved as high as
possible, the resulting Huffman code has a
significantly smaller variance than when it is moved
as low as possible.
Huffman code is a prefix code.
A prefix code is defined as a code in which no code-word
is the prefix of any other code-word.
30

Bit Compression Technologies for Voice

Differential PCM (DPCM)
Adaptive DPCM

Delta Modulation (DM)

Adaptive DM (ADM)

.
.
.
Speech Encoding

31

Differential PCM (DPCM)

32

Delta Modulation (DM)

Delta modulation is a one-bit DPCM.
Advantage: bit compression.
Disadvantage: slope overload.

33

Speech Coding Objective

Reduce the number of bits needed to be transmitted,
therefore lowering the bandwidth required.

34

Speech Properties
Voiced Sound
Arises in generation of vowels and latter portion of some consonants.
Displays long-term repetitive pattern corresponding to the duration of a
pitch interval
Pulse-like waveform.

Unvoiced Sound
Arises in pronunciation of certain consonants such as s, f, p, j,
x, , etc.
Noise-like waveform.

35

Categories of Speech Encoding

Waveform Encoding
Treats voice as analog signal and does not use properties of
speech:

Source Model Coding or Vocoding

Treats properties of speech to preserve word information

Hybrid or parametric methods

Combines waveform and vocoding
36

Linear Predictive Coder (LPC)

37

Multi-Pulse Linear Predictive Coder

(MP-LPC)

38

Regular Pulse Excited Long Term Prediction

Coder (RPE-LPT)

39

Code-Excited Linear Predictive (CELP)

40

Speech Coder Complexity

41

Speech Processing for GSM

Composition of the 13 kbps signal:

36 bits for filter parameters every 20 ms.
9 bits for LTP every 5 ms.
47 bits for RPE every 5 ms.

Thus, in a 20 ms (2080-bit block, or 260 sample) interval,

we need a total of
36+9*20/5+47*20/5=260 bits.
Data Rate = 260/(20 ms) = 13 kbps.
42

Speech Processing for IS-54

Composition of the 7.95 kbps signal:

43 bits for filter parameters every 20 ms.
7 bits for LTP every 5 ms.
88 bits for codebook every 20 ms.

Thus, in a 20 ms (2080-bit block, or 260 samples) interval, we

need a total of:
43+7*20/5+88=159 bits.
Data Rate = 159/(20ms) = 7.95 kbps.
43

Wireless Information Transmission System Lab.

Channel Coding

Institute of Communications Engineering

National Sun Yat-sen University

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Multiplexing

Interleaving

Modulation

Frequency
Spreading

Multiple
Access

si (t )

Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

TX
RF
PA

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Deinterleaving

Demultiplexing

Channel Bits

Optional
Essential

To Other Destinations

45

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Channel Coding
Error detecting coding: Capability of detecting errors so
that re-transmission or dropping can be done.
Cyclic Redundancy Code (CRC)

Error Correcting Coding: Capability of detecting and

correcting errors.
Block Codes: BCH codes, RS codes, etc.
Convolutional codes.
Turbo codes.

46

Linear Block Codes

Encoder transforms block of k successive binary digits
into longer block of n (n>k) binary digits.
Called an (n,k) code.
Redundancy = n-k; Code Rate = k/n;
There are 2k possible messages.
There are 2k possible code words corresponding to the
messages.
Code word (or code vector) is an n-tuple from the space
Vn of all n-tuple.
Storing the 2k code vector in a dictionary is prohibitive
for large k.
47

Vector Spaces
The set of all binary n-tuples, Vn, is called a vector
space over GF (2).
GF: Galois Field.
Two operations are defined:
Addition: V + U = V1 + U 1 + V2 + U 2 + ... + Vn + U n
Scalar Multiplication: aV = aV1 + aV2 + ... + aVn

Example: Vector Space V4

0000 0001 0010 0011 0100 0101 0110 0111
1000 1001 1010 1011 1100 1101 1110 1111
(0101)+(1110)=(0+1, 1+1, 0+1, 1+0)=(1, 0, 1, 1)
1(1010)=(11, 10, 11, 10)=(1, 0, 1, 0)
48

Subspaces
A subset S of Vn is a subspace if
The all-zero vector is in S
The sum of any two vectors in S is also in S.

Example of S: V 0 = 0000
V 1 = 0101

V 2 = 1010
V 3 = 1111

49

Reducing Encoding Complexity

Key feature of linear block codes: the 2k code vectors
form a k-dimensional subspace of all n-tuples.
Example: k = 3, 2k = 8, n = 6, ( 6 , 3 ) code
Message
000
100
010
110
001
101
011
111

Code Word
000000
110100
011010
101110
101001
011101
110011
000111
50

A 3 - dimensiona l subspace of

the vector space of all 6 - tuples.

Reducing Encoding Complexity

It is possible to find a set of k linearly independent n tuples v1 , v 2 , ..., v k such that each n-tuple of the suspace
is a linear combination of v1 , v 2 , ..., v k .

Code word u = m1 v1 + m2 v 2 + ... + mk v k

where mi = 0 or 1
i = 1,..., k

51

Generator Matrix
v1n
v1 v11 v12

v v
v
v
2n
G = 2 = 21 22
= k n Generator Matrix

v
v
v
v
k k1 k 2
kn

The 2k code vectors can be described by a set of k linearly

independent code vectors.
Let m=[m1, m2, , mk] be a message.
Code word corresponding to message m is obtained by:
v1
v
u = mG = [m1 m2
mk ] 2


v k
52

Generator Matrix
Storage is greatly reduced.
The encoder needs to store the k rows of G instead of
the 2k code vectors of the code.
For example:
v1 1 1 0 1 0 0
Let G = v 2 = 0 1 1 0 1 0 and m = [1 1 0 ]
v 3 1 0 1 0 0 1
Then
v1 = 1 v1 + 1 v 2 + 0 v 3
u = [1 1 0 ] v 2 = 1 [110100 ] + 1 [ 011010 ] + 0 [101001]
v 3 = [1 0 1 1 1 0] Code Vector for m = [110 ]
53

Systematic Code

54

Parity Check Matrix

For each generator matrix G, there exists a parity check matrix H
such that the rows of G are orthogonal to the rows of H. (uh=0)

h1 h11
h h
2
21

H=
=

h( n k ) h( n k )1
u = u1 , u2 , , un
uH T = u1hi1 + u2 hi 2 +

h12
h22
h( n k )2

h1n
h2 n

h( n k ) n

+ un hin = 0

where i = 1, 2, , n k
U is a code word generated by matrix G if and only if uHT=0
55

Parity Check Matrix and Syndrome

In a systematic code with G=[Pkxr Ikxk]
H=[Irxr PTrxk]
r

Received
Vector

Code
Vector

Error
Vector

Syndrome of r used for error detection and correction

s = rH T
= 0
Syndrome s
0

If r is a code vector
Otherwise
56

Example of Syndrome Test

1 1 0

G = 0 1 1
1 0 1

1 0 0

0 1 0
0 0 1

Ik

H = [ I n k PT ]
1 0 0 1 0 1
H = 0 1 0 1 1 0
0 0 1 0 1 1

The 6-tuple 1 0 1 1 1 0 is the code vector corresponding to the

1 0 0
message 1 1 0.
0

0
T
s = u H = [1 0 1 1 1 0]
1
0

0
1
= [ 0 0 0]
0
1

0 1

1
0
1
1

Compute the syndrome for the non-code-vector 0 0 1 1 1 0

s = [ 0 0 1 1 1 0] H T = [1 0 0]
57

Weight and Distance of Binary Vectors

Hamming Weight of a Vector:
w(v) = Number of non-zero bits in the vector.

Hamming Distance between 2 vectors:

d(u,v) = Number of bits in which they differ.
For example: u=10010110001
v=11001010101
d(u,v) = 5.

d(u,v) =w(u+v)
The Hamming Distance between 2 vectors is equal to the
Hamming Weight of their vector sum.
58

Minimum Distance of a Linear Code

The set of all code vectors of a linear code form a
subspace of the n-tuple space.
If u and v are 2 code vectors, then u+v must also be a
code vector.
Therefore, the distance d(u,v) between 2 code vectors
equals the weight of a third code vector.
d(u,v) =w(u+v)=w(w)

Thus, the minimum distance of a linear code equals

the minimum weight of its code vectors.
A code with minimum distance dmin can be shown to
correct (dmin-1)/2 erroneous bits and detect (dmin-1)
erroneous bits.
59

Example of Minimum Distance

dmin=3
60

Example of Error Correction and Detection

Capability
v

d min (u , v ) = 7
t max

d min 1
=
:
Error
Correcting
Strength

mmax = d min 1 : Error Detecting Strength

61

Convolutional Code Structure

1
1

k bits

n-1

Output

62

Convoltuional Code
Convolutional codes
k = number of bits shifted into the encoder at one time
k=1 is usually used!!

n = number of encoder output bits corresponding to the k

information bits
r = k/n = code rate
K = constraint length, encoder memory

Each encoded bit is a function of the present input bits

and their past ones.

63

Generator Sequence
u

r0

r1

r2

g 0(1) = 1, g1(1) = 0, g 2(1) = 1, and g 3(1) = 1 .

Generator Sequence: g(1)=(1 0 1 1)
u

r0

r1

r2

r3

g 0( 2 ) = 1, g1( 2 ) = 1, g 2( 2 ) = 1, g 3( 2 ) = 0, and g 4( 2 ) = 1 .
Generator Sequence: g(2)=(1 1 1 0 1)
64

Convolutional Codes
An Example (rate=1/2 with K=2)

x1

G1(x)=1+x2
G2(x)=1+x1+x2

x2

0(00)

00
Present

Next

00

00

00

00

10

11

01

00

11

01

10

00

10

01

01

10

11

10

11

01

10

11

11

01

1(11)

0(11)

Output

0(01)
01

10
1(00)

0(10)

1(10)
11

1(01)

State Diagram
65

Trellis Diagram Representation

10

0(
11
)

01

0(0
1

0(0
1

0(0
1

)
0(0
1

10

01

0(00)

0)
1(0

0)

0)
1(0

1 (0

10

01

00

01

0(00)

00

0(
11
)

0(
11
)

0(
11
)

01

00 0(00)
)
1 (1 1

00 0(00)
)
1 (1 1

)
1 (1 1

00 0(00)

0(
11
)

0(00)
)
1 (1 1

00

0(0
1

0(00)
)
1 (1 1

10

10
)
0(10

)
0(10

)
0(10

)
10
1(

)
10
1(

)
10
1(

)
10
1(

0(10

00

11 1(01) 11 1(01) 11 1(01) 11

Trellis termination: K tail bits with value 0 are usually added to the end of the code.
66

00

Encoding Process

01

0)

0)
)

01

1 (0

1 (0

0)

0(0
1

0(0
1

0(00)

00

10

10

10
)
0(10

)
0(10

)
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

)
10
1(

11 1(01) 11 1(01) 11 1(01) 11

67

0
11

00

0(00)

0(
11
)

0(
11
)

0(
11
)

01

1 (0

10

00 0(00)

0(
11
)

00 0(00)

0
10

)
1 (1 1

01

10

1
01

)
1 (1 1

)
1 (1 1

00 0(00)

1
10

0(0
1

1
00

0 (0
1)

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

0
01

0(
11
)

Input: 1
Output: 11

01

00

Viterbi Decoding Algorithm

Maximum Likelihood (ML) decoding rule
received sequence r

ML

detected sequence d

min(d,r) !!

Viterbi Decoding Algorithm

An efficient search algorithm
Performing ML decoding rule.
Reducing the computational complexity.

68

Viterbi Decoding Algorithm

Basic concept
Generate the code trellis at the decoder
The decoder penetrates through the code trellis level by level in
search for the transmitted code sequence
At each level of the trellis, the decoder computes and
compares the metrics of all the partial paths entering a node
The decoder stores the partial path with the larger metric and
eliminates all the other partial paths. The stored partial path is
called the survivor.

69

Viterbi Decoding Process

1 (0

0(0
1

0)

0)

0)
)

01

10

10

10
)
0(10

)
0(10

)
0(10

)
0(10

)
10
1(

)
10
1(

)
10
1(

11 1(01) 11 1(01) 11 1(01) 11

70

11
11

00

0(00)

0(
11
)

01

1 (0

1 (0
0(0
1

0(
11
)

01

0(00)

00

)
10
1(

10

00 0(00)

0(
11
)

0(
11
)

0(
11
)

00 0(00)

10
11

)
1 (1 1

01

10

01
01

)
1 (1 1

00 0(00)

10
10

0(0
1

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(0
1

Output: 11
Receive: 11

01

00

Viterbi Decoding Process

0(0
1

0(0
1

0)

10

10

10
)
0(10

)
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

2
11 1(01) 11 1(01) 11 1(01) 11

71

11
11

00

0(00)

0(
11
)

01

1 (0

0)

0(0
1

0(00)

00

0(
11
)

01

1 (0

0)

10
)
10
1(

01

1 (0

10

00 0(00)

0(
11
)

0(
11
)

0(
11
)

01

00 0(00)

10
11

)
1 (1 1

01
01

)
1 (1 1

00 0(00)

10
10

0(0
1

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

01

00

Viterbi Decoding Process

0(10

)
)

10
)
10
1(

10
)
10
1(

)
10
1(

0(0
1

0(0
1

0(0
1

)
0(0
1

11 1(01) 11 1(01) 11 1(01) 11

72

11
11

00

0(00)

0(
11
)

01

0)

0(00)

00

1 (0

0)

0)

10

01

1 (0

1 (0

)
10
1(

10
0(10

0(
11
)

0(
11
)

0(
11
)

01

00 0(00)

0(
11
)

10
11

)
1 (1 1

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(10

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

01

00

Viterbi Decoding Process

10
11

)
)
10
1(

)
10
1(

)
10
1(

0(10

10
)

10

0(0
1

)
0(0
1

)
0(0
1

0(0
1

10

0)

)
10
1(

10
0(10

01

1 (0

0)

0)
)

01

1 (0

1 (0

0(
11
)

0(
11
)

0(
11
)

0(
11
)

01

11 1(01) 11 1(01) 11 1(01) 11

73

0(00)

00

11
11

00

0(00)

0(
11
)

)
1 (1 1

00 0(00)

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(10

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

01

00

Viterbi Decoding Process

10
11

0(
11
)

)
0(0
1

0(0
1

0(0
1

0)

0(0
1

01

1 (0

0)

0)

1 (0

1 (0
)

01

3
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

10

10

10

0(10

)
10
1(

10

11 1(01) 11 1(01) 11 1(01) 11

74

0(00)

00

11
11

00

0(00)

0(
11
)

0(
11
)

0(
11
)

01

)
1 (1 1

00 0(00)

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(
11
)

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

01

00

Viterbi Decoding Process

10
11

0(
11
)

)
0(0
1

0(0
1

0(0
1

0)

0(0
1

3
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

3
)

1
)

2
)

10

0(10

)
10
1(

10

11 1(01) 11 1(01) 11 1(01) 11

75

0(00)

10

00

01

10

11
11

01

1 (0

0)

0)

1 (0

1 (0
)

01

0(00)

00

0(
11
)

0(
11
)

0(
11
)

01

)
1 (1 1

00 0(00)

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(
11
)

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

00

Viterbi Decoding Process

10
11

0(
11
)

)
0(0
1

0(0
1

0(0
1

0)

0(0
1

3
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

3
)

1
)

2
)

10

0(10

)
10
1(

10

11 1(01) 11 1(01) 11 1(01) 11

76

0(00)

10

00

01

10

11
11

01

1 (0

0)

0)

1 (0

1 (0
)

01

0(00)

00

0(
11
)

0(
11
)

0(
11
)

01

)
1 (1 1

00 0(00)

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(
11
)

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Output: 11
Receive: 11

00

Viterbi Decoding Process

10
11

0(
11
)

)
0(0
1

0(0
1

0(0
1

0)

0(0
1

3
0(10

0(10

)
10
1(

)
10
1(

)
10
1(

3
)

1
)

2
)

10

0(10

)
10
1(

10

11 1(01) 11 1(01) 11 1(01) 11

77

0(00)

10

00

01

10

11
11

01

1 (0

0)

0)

1 (0

1 (0
)

01

0(00)

00

0(
11
)

0(
11
)

0(
11
)

01

)
1 (1 1

00 0(00)

)
1 (1 1

00 0(00)

01
01

00 0(00)

01

10

10
10

0(
11
)

00
00

)
1 (1 1

0(00)
)
1 (1 1

00

0 (0
1)

0(00)
)
1 (1 1

00

01
11

0(10

Decision:11
Receive: 11

00

Channel Coding in GSM

78

Channel Coding in IS-54/136

79

Turbo Codes Basic Concepts

Turbo coding uses parallel concatenation of two
recursive systematic convolutional codes joined through
an interleaver.
Information bits are encoded block by block.
Turbo codes uses iterative decoding techniques.
Soft-output decoder is necessary for iterative decoding.
Turbo codes can approach to Shannon limit.

80

Turbo Codes Encoder - An Example

X(t)
Y(t)
X(t)

Interleaver

Y(t)

X'(t)

When the switch is placed on the low position, the tail bits are feedback
and the trellis will be terminated.
81

Turbo Codes Encoding Example

A systematic convolutional encoder with memory 2
The dotted line is for termination code
Test sequence: 1011

X0
X1
1101

82

Turbo Codes Encoding Example

X0=1
X1=1
1101

00
01

11

10
11

83

Turbo Codes Encoding Example

X0=0
X1=1
110

00
01
10

11

01

11

84

Turbo Codes Encoding Example

X0=1
X1=0
11

00
01
10
11

11

01
10

85

Turbo Codes Encoding Example

X0=1
X1=0
1

00
01
10
11

11

01
10

10

86

Turbo Codes Encoding Example

X0=0
X1=1
1

00
01
10
11

11
01
01
10

10

87

Turbo Codes Encoding Example

X0=1
X1=1
0

00
01
10
11

11
11
01
01
10

10

88

Turbo Codes Encoding Example

X0=0
X1=0
0

00
01
10
11

11
11
01
01
10

10

89

00

Turbo Codes Encoding Example

X0

1101

X1
D

Interleaver

1011

(X0)
X2
D

Output sequence: X0, X1, X2, X0, X1, X2, X0, X1, X2,...

90

Turbo Codes Encoding Example

The second encoder input is the interleaved
data
1 0

1101

1011

1 1
00

00
01

11

11
10

00

10

10
11
91

00

CRC in WCDMA
gCRC24(D) = D 24 + D 23 + D 6 + D 5 + D + 1;
gCRC16(D) = D 16 + D 12 + D 5 + 1;
gCRC12(D) = D 12 + D 11 + D 3 + D 2 + D + 1;
gCRC8(D) = D 8 + D 7 + D 4 + D 3 + D + 1.

92

Channel Coding Adopted in WCDMA

Type of TrCH

Coding scheme

Coding rate

Convolutional
coding

1/2

BCH
PCH
RACH

1/3, 1/2
CPCH, DCH, DSCH,
FACH

Turbo coding
No coding
93

1/3

Convolutional Coding in WCDMA

Input

D
Output 0
G0 = 561 (octal)
Output 1
G1 = 753 (octal)

(a) Rate 1/2 convolutional coder

Input

D
Output 0
G0 = 557 (octal)
Output 1
G1 = 663 (octal)
Output 2
G2 = 711 (octal)

(b) Rate 1/3 convolutional coder

94

Turbo Coder in WCDMA

xk
1st constituent encoder

zk

xk
Input

Output

Input

Turbo code
internal interleaver
Output

xk

2nd constituent encoder

zk

xk

95

Wireless Information Transmission System Lab.

Interleaving

Institute of Communications Engineering

National Sun Yat-sen University

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Multiplexing

Interleaving

Modulation

Frequency
Spreading

Multiple
Access

si (t )

Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

TX
RF
PA

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Deinterleaving

Demultiplexing

Channel Bits

Optional
Essential

To Other Destinations

97

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Bursty Error in Fading Channel

98

Interleaving Mechanism (1/2)

x

Bit
Interleaver

y
x

j x n-bit
Shift registers

Write Clock

Read Clock

Bit Stream before entering bit interleaver:

x=(a11 a12 a1n a21 a22 a2n aj1 aj2 ajn)
99

Interleaving Mechanism (2/2)

Conceptually, the WRITE clock places the bit stream
x by the row while the REA clock takes the bit stream
y by the column:
a11
a
21
.

.
.

a j1

a12
a 22
.
.
.
a j2

.
.
.
.
.
.

.
.
.
.
.
.

. a1n
. a 2 n
.
.

.
.
.
.

. a jn

Bit stream at the output of the bit interleaver:

y = (a11

a21 ... a j1

a12

a22 ... a j 2 ... a1n

100

a2 n ... a jn )

Burst Error Protection with Interleaver

101

Wireless Information Transmission System Lab.

Modulation

Institute of Communications Engineering

National Sun Yat-sen University

Typical Digital Communications System

From Other Sources
Information Bits

Source Bits

Source
Encoding

Format

Encryption

Channel Bits
Channel
Encoding

Interleaving

Multiplexing

Modulation

Frequency
Spreading

Multiple
Access

si (t )

Digital
Input

C
H
A
N
N
E
L

mi

Bit
Stream

Synchronization

Digital
Waveform

Digital
Output

m i

Format

TX
RF
PA

si (t )
Source
Decoding

Information Sink

Decryption

Source Bits

Channel
Decoding

Deinterleaving

Demultiplexing

Channel Bits

Optional
Essential

To Other Destinations

103

Demodulation

Frequency
Despreading

Multiple
Access

RX
RF
IF

Modulation
Digital Modulation: digital symbols are transformed into
waveforms that are compatible with the characteristics of the
channel.
In baseband modulation, these waveforms are pulses.
In bandpass modulation, the desired information signal
modulates a sinusoid called a carrier. For radio transmission,
the carrier is converted in an electromagnetic (EM) wave.
Why modulation?
Antenna size should be comparable with wave length
baseband transmission is not possible.
Modulation may be used to separate the different signals
using a single channel.
104

PCM Waveform Representations

105

PCM Waveform Representations

PCM waveform is also called line codes.
Digital baseband signals often use line codes to provide
particular spectral characteristics of a pulse train.
NRZ-L.
NRZ-M.
NRZ-S.
Unipolar-RZ.
Polar-RZ.
Bi--L.
Bi--M.

Bi--S.
Dicode-NRZ.
Dicode-RZ.
Delay Mode.
4B3T.
Multi-level.
etc.
106

PCM Waveform : NRZ-L

1

+E
0
-E

NRZ Level (or NRZ Change)

One is represented by one level.
Zero is represented by the other level.

107

PCM Waveform : NRZ-M

1

+E
0
-E

NRZ Mark (Differential Encoding)

One is represented by a change in level.
Zero is represented by a no change in level.

108

PCM Waveform : NRZ-S

1

+E
0
-E

NRZ Space (Differential Encoding)

One is represented by a no change in level.
Zero is represented by a change in level.

109

PCM Waveform : Unipolar-RZ

1

+E
0
-E

Unipolar - RZ
One is represented by a half-bit width pulse.
Zero is represented by a no pulse condition.

110

PCM Waveform : Polar-RZ

1

+E
0
-E

Polar - RZ
One and Zero are represented by opposite
level polar pulses that are one half-bit in width.

111

PCM Waveform : Bi--L

1

+E
0
-E

Bi--L (Biphase Level or Split Phase Manchester

11 + 180o)
One is represented by a 10.
Zero is represented by a 01.
112

PCM Waveform : Bi--M

1

+E
0
-E

Bi--M ( Biphase Mark or Manchester 1)

A transition occurs at the beginning of every bit period.
One is represented by a second transition one half bit
period later.
Zero is represented by no second transition.
113

PCM Waveform : Bi--S

1

+E
0
-E

Bi--S ( Biphase Space)

A transition occurs at the beginning of every bit period.
One is represented by no second transition.
Zero is represented by a second transition one-half bit
period later.
114

PCM Waveform : Dicode - NRZ

1

+E
0
-E

Dicode Non-Return-to-Zero
A One to Zero or Zero to One changes polarity.
Otherwise, a Zero is sent.

115

PCM Waveform : Dicode - RZ

1

+E
0
-E

Dicode Return-to-Zero
A One to Zero or Zero to One transition produces
a half duration polarity change.
Otherwise, a Zero is sent.
116

PCM Waveform : Delay Mode

1

+E
0
-E

Dicode Non-Return-to-Zero
A One is represented by a transition at the midpoint of
the bit interval.
A Zero is represented by a no transition unless it is
followed by another zero. In this case, a transition is
placed at the end of bit period of the first zero.
117

PCM Waveform : 4B3T

O --

118

PCM Waveform : 4B3T

Ternary words in the middle column are balanced in
their DC content.
Code words from the first and third columns are selected
alternately to maintain DC balance.
If more positive pulses than negative pulses have been
transmitted, column 1 is selected.
Notice that the all-zeros code word is not used.

119

PCM Waveform : Multilevel Transmission

Multilevel transmission with 3 bits per signal interval.

120

Criteria for Selecting PCM Waveform

DC component: eliminating the dc energy from the
signals power spectrum.
Self-Clocking: Symbol or bit synchronization is
required for any digital communication system.
Error detection: some schemes provide error detection
without introducing additional error-detection bits.
Bandwidth compression: some schemes increase
bandwidth utilization by allowing a reduction in
required bandwidth for a given data rate.
Noise immunity.
Cost and complexity.
121

Spectral Densities of Various PCM Waveforms

122

Linear Modulation Techniques

Digital modulation techniques may be broadly classified as linear
and nonlinear.
In linear modulation techniques, the amplitude of the transmitted
signal, s(t), varies linearly with the modulating digital signal, m(t).
Linear modulation techniques are bandwidth efficient, though
they must be transmitted using linear RF amplifiers which have
poor power efficiency.
Using power efficient nonlinear amplifiers leads to the
regeneration of filtered sidelobes which can cause severe adjacent
channel interference, and results in the loss of all the spectral
efficiency gained by linear modulation.
Clever ways have been developed to get around these difficulties:
QPSK, OQPSK, /4-QPSK.
123

Digital Modulations

Basic digital modulated signal:

v(t) = A(t) cos (t + )
Where A(t) = Amplitude; = Frequency; = Phase;
124

Basic Digital Modulations

125

Extended Modulated Signals M-FSK

Example: 16-FSK

Every 4 bits is encoded as: A cos( j t )

Gray Coding.

126

j = 1,2,,16

Extended Modulated Signals M-PSK

Example: 16-PSK

Every 4 bits is encoded as: A sin( t + j )

Gray Coding.

Dotted lines are decision boundaries.

127

j = 1, 2, ,16

Extended Modulated Signals 16-QAM

Every 4 bits is represented by one point in the signal constellation.
Every point has its unique amplitude and phase.

128

Binary Phase Shift Keying (BPSK)

In BPSK, the phase of a constant amplitude carrier signal is
switched between two values according to the two possible
signals m1 and m2 corresponding to binary 1 and 0. Normally,
the two phases are separated by 180o.
2 Eb
sBPSK ( t ) = m ( t )
cos ( 2 f c t + c )
0 t Tb
Tb
= Re { g BPSK ( t ) exp ( j 2 f c t )}

sin fTb
2 Eb
jc
g BPSK ( t ) =
m ( t ) e Pg BPSK (t ) ( f ) = 2 Eb

Tb
fT

2
2

Eb sin ( f f c ) Tb sin ( f f c ) Tb

+
PBPSK ( f ) =

2 ( f f c ) Tb ( f f c ) Tb

129

Power Spectral Density (PSD) of a BPSK

Signal.

130

BPSK Receiver
BPSK uses coherent or synchronous demodulation,
which requires that information about the phase and
frequency of the carrier be available at the receiver.
If a low level pilot carrier signal is transmitted along
with the BPSK signal, then the carrier phase and
frequency may be recovered at the receiver using a
phase locked loop (PLL).
If no pilot carrier is transmitted, a Costas loop or
squaring loop may be used to synthesize the carrier
phase and frequency from the received BPSK signal.
131

BPSK Receiver with Carrier Recovery

Circuits

132

Operations of BPSK Receiver with Carrier

Recovery Circuits
The received signal is squared to generate a DC signal and an
amplitude varying sinusoid at twice the carrier frequency.
The DC signal is filtered out using a bandpass filter with center
frequency tuned to 2fc.
A frequency divider is used to recreate the waveform
cos(2fct+).
The output of the multiplier is applied to an integrate and dump
circuit which forms the low pass filter segment of a BPSK
detector.
If the transmitter and receiver pulse shapes are matched, then the
detection will be optimum.
A bit synchronizer is used to facilitate sampling of the integrator
output precisely at the end of each bit period.
133

Differential Phase Shift Keying (DPSK)

Differential PSK is a noncoherent form of phase shift keying
which avoids the need for a coherent reference signal at the
receiver.

d k = mk d k 1

134

Block Diagram of DPSK Receiver

135

Quadrature Phase Shift Keying (QPSK)

136

Spectrum of QPSK Signals

sin ( f f ) T
c
s
PQPSK ( f ) = Eb
( f f c ) T

137

sin ( f f c ) Ts
+
( f fc ) T

Block Diagram of a QPSK Transmitter

138

Block Diagram of a QPSK Receiver

139

Offset QPSK (OQPSK)

For QPSK, the occasional phase shift of radians can cause the
signal envelope to pass through zero for just an instant.
The amplification of the zero-crossings brings back the filtered
sidelobes since the fidelity of the signal at small voltage levels is
lost in transmission.
To prevent the regeneration of sidelobes and spectral widening, it
is imperative that QPSK signals that use pulse shaping be
amplified only using linear amplifiers, which are less efficient.
A modified form of QPSK, called offset QPSK (OQPSK) or
staggered QPSK is less susceptible to these deleterious effects
and supports more efficient amplification.
OQPSK ensures there are fewer baseband signal transitions.
Spectrum of an OQPSK signal is identical to that of QPSK.
140

Offset QPSK (OQPSK)

The time offset waveforms that are applied to the in-phase and
quadrature arms of an OQPSK modulator. Notice that a halfsymbol offset is used.

141

/4-DQPSK

142

Generic /4-DQPSK Transmitter

143

/4-DQPSK Baseband Differential

Detector

144

Detection of Binary Signals in Gaussian

Noise

145

Digital Demodulation Techniques

Coherent detection: Exact replicas of the possible arriving
signals are available at the receiver. This means that the
receiver has exact knowledge of the carrier waves phase
reference, in which case we say the receiver is phase-locked to
the transmitter. Coherent detection is performed by crosscorrelating the received signal with each one of the replicas,
and then making a decision based on comparisons with preselected thresholds.
Non-coherent detection: Knowledge of the carrier waves
phase is not required. The complexity of the receiver is
thereby reduced but at the expense of an inferior error
performance, compared to a coherent system.
146

Correlation Demodulator

147

Matched Filter Demodulator

148

Inter-Symbol Interference (ISI)

149

Inter Symbol Interference (ISI)

Inter-Symbol Interference (ISI) arises because of
imperfections in the overall frequency response of the
system. When a short pulse of duration Tb seconds is
transmitted through a band-limited system, the
frequency components constituting the input pulse
are differentially attenuated and differentially delayed
by the system. Consequently, the pulse appearing at
the output of the system is dispersed over an interval
longer than Tb seconds, thereby resulting in intersymbol interference.
Even in the absence of noise, imperfect filtering and
system bandwidth constraints lead to ISI.
150

Nyquist Channels for Zero ISI

The Nyquist channel is not physically realizable since it
dictates a rectangular bandwidth characteristic and an infinite
time delay.
Detection process would be very sensitive to small timing
errors.
Solution: Raised Cosine Filter.

151

Raised Cosine Filter

1
2 f + W 2W0
H ( f ) = cos (
)
4 W W0

0
1
W0 =
2T
Excess Bandwidth : W W0
W W0
Roll - Off Factor : r =
W0
152

for f < 2W0 W

for 2W0 W < f < W
for f > W

Raised Cosine Filter Characteristics

153

Raised Cosine Filter Characteristics

154

Equalization
In practical systems, the frequency response of the
channel is not known to allow for a receiver design that
will compensate for the ISI.
The filter for handling ISI at the receiver contains
various parameters that are adjusted with the channel
characteristics.
The process of correcting the channel-induced distortion
is called equalization.

155

Equalization

156

Introduction to RAKE Receiver

Multiple versions of the transmitted signal are seen at
the receiver through the propagation channels.
Very low correlation between successive chips is in
CDMA spreading codes.
If these multi-path components are delayed in time
by more than a chip duration, they appear like
uncorrelated noise at a CDMA receiver.
Combine
Coherently

Equalization is
NOT necessary
157

Introduction to RAKE Receiver

To utilize the advantages of diversity techniques,
channel parameters are necessary to be estimated.
Arrival time of each path, Amplitude, and Phase.

Maximal Ratio Combiner (MRC): The combiner that

achieves the best performance is one in which each
output is multiplied by the corresponding complexvalued (conjugate) channel gain. The effect of this
multiplication is to compensate for the phase shift in the
channel and to weight the signal by a factor that is
proportional to the signal strength.
158

Maximum Ratio Combining (MRC)

MRC: Gi=Aie-jqi
G1

G2

Coherent Combining

GL

Channel Estimation
Best Performance
Receiver

159

Maximum Ratio Combining (MRC)

L

Received Envelope:rL = Gl rl
l =1

Total Noise Power: = Gl n2,l

2
n

l =1
L

2
L

r
=
SNR: SNRL =
2
2 n

G r
l =1
L

2 Gl n2,l
2

l =1

Since

G r
l =1

rl
= Gl n ,l

l =1
n ,l
L

160

Maximum Ratio Combining (MRC)

2

Chebychev's Inequality : Gl rl Gl n ,l
l =1

SNRL

1
l =1

2
n ,l

l =1

l =1

l =1

rl

n ,l

rl

n ,l

L
rl
1
= 2 = SNRl
2 l =1 n ,l l =1
L

2
G

l n ,l
2

l =1

With equality hold : Gl n ,l = k

rl*

n ,l

Output SNR = Sum of SNRs from all branches @ Gl rl*

161

Example of RAKE Receiver Structure

162

Advantages of RAKE Receiver

Consider a receiver with only one finger:
Once the output of a single correlator is corrupted by
fading, large bit error is expected.

Consider a RAKE receiver

If the output of a single correlator is corrupted by fading,
the others may NOT be.
Diversity is provided by combining the outputs
Overcome fading
Improve CDMA reception

163