Professional Documents
Culture Documents
Amirpasha Shirazinia
Digital Communication
Lecture 1
1 / 15
Todays Lecture
Course outline
Aims of the course
Introduction to digital communication
Digital Communication
Lecture 1
2 / 15
Course Outline
Part I January-March
12 + 1 Lectures
10 Tutorials
3 Hand-in assignments
Part II March-June
18 Lectures
11 Tutorials
1 Recap lecture of part I
Bonus question
Written exam
Digital Communication
Lecture 1
3 / 15
Teachers
Digital Communication
Lecture 1
4 / 15
Course Material
Main textbooks
Digital Communications by Bernard Sklar (BS)
ISBN 0-13-084788-7
Wireless Communications by Andrea Goldsmith (AG)
ISBN 978-0-521-83716-3
Additional materials
News
Lecture notes, exercises, old exams
Digital Communication
Lecture 1
5 / 15
Language
Digital Communication
Lecture 1
6 / 15
Teaching style
Based on
Lecture slides: Definition, summary, important results, ...
Lecture notes (written on the board): derivations, comments on
slides, ...
Some teaching notes during tutorials.
Lecture slides and lecture notes can be downloaded at the course page on
studentportalen.
Digital Communication
Lecture 1
7 / 15
Pre-requisite Courses
Digital Communication
Lecture 1
8 / 15
Outline Part I
Introduction (lec. 1)
Formatting: Sampling, quantization & baseband transmission (lec. 2)
Receiver structure (lec. 3-4)
Digital modulation schemes (lec. 5-7)
Channel coding (lec. 8-11)
Channel equalization (lec. 12)
Repetition and summary (lec. 13 in period 2)
Digital Communication
Lecture 1
9 / 15
Assessment
Hand-in assignments
3 assignments
Work in groups of 2-3
Do not miss the deadlines
Feel free to ask for help ,
Please read Rules Requirements on Studentportalen
Exam
Covers all material
Focuses on parts I and II equally
Digital Communication
Lecture 1
10 / 15
Aims Part I
We aim to teach/learn
How information is transferred from source to receiver through a digital
communication system
How to assess the quality of the received information
Some important modulation schemes
The most important features of error correcting codes
Digital Communication
Lecture 1
11 / 15
Introduction
In this course, we are focusing on
how the information is communicated digitally,
how good the information is received.
Why communication?
Humans need ...
Why digital representation/signaling?
Digital Communication
Lecture 1
12 / 15
Wired/Wireless Communication
Wired communication:
ADSL: copper twisted wire
Optical fiber: (almost) attenuation-free communication with high
bit-rate
Wireless communication:
Radio signal: e.g., satellite, mobile system, radar
Infrared signals: TV remote control
Digital Communication
Lecture 1
13 / 15
c
Encode
Format
Analog
info
Information bits
Format
Coded bits
Decode
Waveforms
Channel
Demodulate
c
s(t)
Modulate
r(t)
Receiver (RX)
Digital Communication
Lecture 1
14 / 15
Random Signals
Digital Communication
Lecture 1
15 / 15
Digital Communication
Lecture 2
1/9
Bonus Question
Digital Communication
Lecture 2
2/9
Todays Lecture
Sampling
Quantization
Baseband modulation (coding)
Digital Communication
Lecture 2
3/9
Introduction
Transmitter (TX)
bit
b
c
Encode
Format
Analog
info
Information bits
Format
s(t)
Modulate
Coded bits
Decode
Waveforms
Demodulate
c
Channel
r(t)
Receiver (RX)
Digital Communication
Lecture 2
4/9
Sampling
Definition: is a discretizing process of a continuous signal in time.
and the signal is sampled evenly at a rate of fs = T1s > 2fmax , then
S(t) can be exactly recovered from its samples, i.e., Si s.
Nyquist frequency: 2fmax .
Amirpasha Shirazinia (UU)
Digital Communication
Lecture 2
5/9
Sampling Methods
fs < 2fmax .
Other types of sampling methods: natural sampling, sample and hold,
...
Digital Communication
Lecture 2
6/9
Quantization
technique)
Definition: A mapping from a value in a continuous set into a digit in
a discrete set.
In a general view, it can be divided into two types:
Uniform quantization
Non-uniform quantization
Digital Communication
Lecture 2
7/9
Digital Communication
Lecture 2
8/9
Baseband Modulation/Coding
information
Baseband modulation/coding: In practice, we present the binary
Digital Communication
Lecture 2
9/9
Digital Communication
Lecture 3
1/9
Bonus Question
Digital Communication
Lecture 3
2/9
Todays Lecture
Digital Communication
Lecture 3
3/9
Introduction
Transmitter (TX)
bit
b
c
Encode
Format
Analog
info
Information bits
Format
s(t)
Modulate
Coded bits
Decode
Waveforms
Demodulate
c
Channel
r(t)
Receiver (RX)
Digital Communication
Lecture 3
4/9
Noise
electrons.
Noise Statistics: We normally model the thermal noise by a random
1
2 2
n2
e 22 .
N0
2
= 2 , i.e., the
Digital Communication
Lecture 3
5/9
AWGN
0.8
2
= 0.25
0.7
2 = 1
0.6
0.5
0.4
0.3
0.2
0.1
0
4
Digital Communication
Lecture 3
6/9
Important condition:
j (t)k (t)dt =
0
Kj j = k
0 otherwise
(1)
j, k = 1, . . . , N , 0 t T .
Digital Communication
Lecture 3
7/9
Digital Communication
Lecture 3
8/9
Digital Communication
Lecture 3
9/9
Digital Communication
Lecture 4
1/9
Bonus Question
Digital Communication
Lecture 4
2/9
Todays Lecture
Detection
Analysis of matched filter
Analysis of error probability
Digital Communication
Lecture 4
3/9
Introduction
Receiver structure for baseband modulation
n(t)
r(t)
s(t)
Matched filter
z(t)
h(t)
Detector
0 or 1?
s(t)
waveforms (signaling).
Digital Communication
Lecture 4
4/9
(S/N ), at time t = T
n(t)
r(t)
s(t)
Matched filter
h(t)
z(t)
Digital Communication
Lecture 4
5/9
Error Probability
si (t)
r(t)
s2 (t) = a2 .
Digital Communication
Lecture 4
6/9
Error Probability
Conditional probability of z(t) given that waveforms s1 (t) and st (t)
0.14
0.12
f (z |s 1 )
f (z |s 2 )
0.1
0.08
0.06
0.04
0.02
0
8
a2
a1
Digital Communication
Lecture 4
7/9
PB .
... Error probability becomes
PB = Q
RT
where = E1b 0 s1 (t)s2 (t)dt =
space representation.)
Eb (1 )
,
N0
1
Eb s1
Digital Communication
Lecture 4
8/9
Digital Communication
Lecture 4
9/9
January 9, 2015
Digital Communication
Lecture 5
1 / 10
Bonus Question
Digital Communication
Lecture 5
2 / 10
Todays Lecture
c
Encode
Format
Analog
info
Information bits
Format
s(t)
Modulate
Coded bits
Decode
Channel
Demodulate
c
Waveforms
r(t)
Receiver (RX)
Digital Communication
Lecture 5
3 / 10
Principles
probability of error.
Digital Communication
Lecture 5
4 / 10
Modulation Types
Digital Communication
Lecture 5
5 / 10
Modulation Types
s(t) = A(t) cos 2 fc + f (t) t + (t) +
...
n
o
= u(t)ej2fc t , u(t) = sI (t) + jsQ (t) = (si1 + jsi2 )g(t)
Digital Communication
Lecture 5
5 / 10
InPhase branch
Shaping
Filter
g(t)
s i1g(t)
cos(2f c t+ 0)
cos(2f c t+ 0)
s(t)
2
sin(2 f c t+ 0)
s i2
Shaping
Filter
g(t)
s i2g(t)
Quadrature Branch
Digital Communication
Lecture 5
6 / 10
g(Tt)
cos (2f c t+ )
^
m=m
i
x(t)=s i(t)+n(t)
Find i: x Z i
/2
sin (2f c t+)
Ts
g(Tt)
x2 =s i2+n2
Quadrature branch
Digital Communication
Lecture 5
7 / 10
01
11
10
2d
M=8, K=3
000
001
011
010
110
111
101
100
2d
Digital Communication
Lecture 5
8 / 10
transmitted signal
Representation: si (t) = u(t)ej2fc t , where
u(t) = Ag(t)e2(i1)/M , 0 t Ts .
s i2
M=4, K=2
M=8, K=3
s i2
011
01
010
11
00
si1
001
110
000
110
10
si1
100
101
Digital Communication
Lecture 5
9 / 10
4QAM
16QAM
Digital Communication
Lecture 5
10 / 10
February 6, 2015
Digital Communication
Lecture 6
1/9
Bonus Question
Digital Communication
Lecture 6
2/9
Todays Lecture
Frequency modulation
Performance analysis of modulation techniques
Digital Communication
Lecture 6
3/9
Frequency Modulation
For frequency modulation: the information bits are encoded into the
basis functions:
f = min |fi fj |
i,j
1
2Ts
1
Ts
if i = j
if i 6= j
Digital Communication
Lecture 6
4/9
si (t) = A cos 2 fc + i fc t + i
where i = (2i 1 M ) for i = 1, 2, . . . , M .
Digital Communication
Lecture 6
5/9
Frequency Demodulators
Digital Communication
Lecture 6
6/9
Digital Communication
Lecture 6
7/9
Digital Communication
Lecture 6
8/9
Pb = Q
QPSK:
Pb = Q
p
p
p
2Eb /N0 = Q( 2b )
p
2Eb /N0 = Q( 2b )
Ps = 1 (1 Pb )2
Digital Communication
Lecture 6
9/9
Digital Communication
Lecture 7
1/7
Bonus Question
Digital Communication
Lecture 7
2/7
Todays Lecture
Digital Communication
Lecture 7
3/7
3
s
Rectangular M-QAM: Ps(MQAM) 2( M 1) Q( M1
)
M
q
3
s
Non-rectangular M-QAM: Ps(MQAM) 4Q( M1
)
1
P
ms
m+1 M1
Coherent M-FSK: Ps(MF SK) M
(1)
exp
m=1
m
m+1
m+1
Digital Communication
Lecture 7
4/7
Modulation
BFSK:
BPSK:
QPSK,4QAM:
Ps (s )
Ps 2 Q s
2(M 1)
Q
M
MPAM:
Ps
MPSK:
Ps 2Q
Rectangular MQAM:
Nonrectangular MQAM:
Ps
6 s
M 2 1
2s sin(/M )
q
3 s
2( M 1)
Q
M
1
M
Ps 4Q
q
3 s
M 1
Pb
2(M 1)
M log2 M Q
q
6 b log2 M
(M 2 1)
2
2b log2 M sin(/M )
log2 M Q
q
3 b log2 M
Q
Pb 2(M M1)
(M 1)
log2 M
Pb
Pb (b )
Pb = Q b
Pb = Q 2b
Pb Q 2b
Pb
4
log2 M Q
q
3 b log2 M
(M 1)
Table 6.1: Approximate Symbol and Bit Error Probabilities for Coherent Modulations
Digital Communication
Lecture 7
5/7
10
10
10
10
10
10
BFSK
BPSK
8PAM
12
8PSK
10
8QAM
14
10
16
10
10
15
Digital Communication
Lecture 7
6/7
Spectral Efficiency
Digital Communication
Lecture 7
7/7
Digital Communication
Lecture 8
1/8
Todays Lecture
Digital Communication
Lecture 8
2/8
Transmitter (TX)
bit
b
Format
Analog
info
Encode
Information bits
Format
Coded bits
Waveforms
Channel
Demodulate
Decode
c
s(t)
Modulate
r(t)
Receiver (RX)
Digital Communication
Lecture 8
3/8
Pros?
Protect information bits
Decrease BER coding gain
How?
By adding redundancy (more bits) to information bits
Cons?
Rate penalty (reduce data rate)
Bandwidth expansion
Delay
Digital Communication
Lecture 8
4/8
Digital Communication
Lecture 8
5/8
Some Definitions
codewords
Code Weight: Number of 1s in a codeword
Minimum distance: Hamming distance between a codeword and the
all-zero codeword.
Digital Communication
Lecture 8
6/8
(n > k).
For an information sequence U and codeword C, encoding is
Digital Communication
Lecture 8
7/8
(n > k).
For an information sequence U and codeword C, encoding is
corrupt by noise.
minimum distance of linear binary block codes (n, k):
dmin n k + 1.
Digital Communication
Lecture 8
7/8
Less complexity.
Digital Communication
Lecture 8
8/8
Digital Communication
Lecture 9
1/7
Bonus Question
Digital Communication
Lecture 9
2/7
Todays Lecture
Digital Communication
Lecture 9
3/7
Channel Decoding
either 0 or 1.
Soft-decision decoding (SDD): The distance between received bit
Digital Communication
Lecture 9
4/7
Digital Communication
Lecture 9
5/7
Digital Communication
Lecture 9
5/7
decoded in error.
Upper-bound
n
X
n j
Pe
p (1 p)nj ,
j
j=t+1
Pe
dX
min
j=t+1
dmin j
p (1 p)nj .
j
1
k Pe
Digital Communication
Lecture 9
6/7
Pe
2
X
2
X
i=2
p
Q( 2Rc b wi )
i=2
p
Q( 2Rc b dmin )
Digital Communication
Lecture 9
7/7
Digital Communication
Lecture 10
1 / 11
Bonus Question
Digital Communication
Lecture 10
2 / 11
Todays Lecture
Digital Communication
Lecture 10
3 / 11
Convolutional Codes
Codes with memory.
A Convolutional coded symbol is generated by passing information
...
k bits
1
...
Stage 1
...
...
Stage 2
...
Stage K
Digital Communication
Lecture 10
4 / 11
Encoder Characterization
Digital Communication
Lecture 10
5 / 11
Trellis diagram
n = 3, k = 1, K = 3
S=S2S3
t0
C1 C2 C
3
Encoder
Output
00
000
t1
000
t2
S1
S2
S3
Stage 1
Stage 2
Stage 3
01
011
100
010
011
100
010
t5
000
111
011
t 43
000
111
111
111
111
t3
000
100
010
010
10
001
001
101
11
001
101
101
110
110
101
110
S1=0
S1=1
Digital Communication
Lecture 10
6 / 11
Digital Communication
Lecture 10
7 / 11
n
X
j=1
P
and check i Bi for all paths and choose the maximum. Equivalently,
find the minimum distance codeword by comparing the distance of all
codewords with the received codeword.
Digital Communication
Lecture 10
7 / 11
n
X
j=1
P
and check i Bi for all paths and choose the maximum. Equivalently,
find the minimum distance codeword by comparing the distance of all
codewords with the received codeword.
Soft-decision: At branch i of the trellis diagram, check the metric
i =
n
X
j=1
and check
Rij (2Cij 1)
Digital Communication
Lecture 10
7 / 11
n
X
j=1
P
and check i Bi for all paths and choose the maximum. Equivalently,
find the minimum distance codeword by comparing the distance of all
codewords with the received codeword.
Soft-decision: At branch i of the trellis diagram, check the metric
i =
n
X
j=1
and check
Rij (2Cij 1)
Digital Communication
Lecture 10
7 / 11
Digital Communication
Lecture 10
8 / 11
Digital Communication
Lecture 10
9 / 11
Digital Communication
Lecture 10
10 / 11
Digital Communication
Lecture 10
11 / 11
Digital Communication
Lecture 11
1 / 12
Bonus Question
Digital Communication
Lecture 11
2 / 12
Todays Lecture
Digital Communication
Lecture 11
3 / 12
Trellis Diagram
n = 3, k = 1, K = 3
S=S2S3
t0
C1 C2 C
3
Encoder
Output
00
000
t1
000
t2
S1
S2
S3
Stage 1
Stage 2
Stage 3
01
011
100
010
011
100
010
t5
000
111
011
t 43
000
111
111
111
111
t3
000
100
010
010
10
001
001
101
11
001
101
101
110
110
101
110
S1=0
S1=1
Digital Communication
Lecture 11
4 / 12
State Diagram
The right one is a modified version of the left one by splitting the
000
011
Digital Communication
Lecture 11
5 / 12
State Diagram
The right one is a modified version of the left one by splitting the
000
011
T (D) = Xe /Xa =
d=df
ad D d .
Digital Communication
Lecture 11
5 / 12
Digital Communication
Lecture 11
6 / 12
T (D, N, J) = Xe /Xa =
d=df
P P
m
Dd J m N l ,
T (D, N, J) =
J 3 N D6
= J 3 N D6 + J 4 N 2 D8 + . . .
1 JN D2 L(1 + J)
can be written
P in terms of d. In a compact way
T (D, N ) = d ad N f (d) D d .
Amirpasha Shirazinia (UU)
Digital Communication
Lecture 11
7 / 12
decoder.
In general, decoding error probability Pe is upper-bounded as
Pe
ad P2 (d)
d=df
transfer function)
P2 (d): pairwise error probability of a path with Hamming distance of d
Digital Communication
Lecture 11
8 / 12
modulation
P2 (d) = Q
2Ec d
N0
p
= Q( 2b Rc d) exp(b Rc d)
P2 (d) =
d
X
j=(d+1)/2
d j
p (1 p)dj
j
Digital Communication
Lecture 11
9 / 12
f (d)ad P2 (d),
d=df
Digital Communication
Lecture 11
10 / 12
Interleaving
the decoder.
Interleaving is achieved by spreading the coded symbols in different
Digital Communication
Lecture 11
11 / 12
Interleaving: Example
Consider a code with t = 1 ability of correction and 3 coded bits.
A bursterror
error of of
length
3 cannot3
becan
corrected.
A burst
length
not be corrected.
A1 A2 A3 B1 B2 B3 C1 C2 C3
2 errors
A1 B1 C1 A2 B2 C2 A3 B3 C3
Interleaver
Deinterleaver
A1 B1 C1 A2 B2 C2 A3 B3 C3
A1 A2 A3 B1 B2 B3 C1 C2 C3
1 errors
1 errors
1 errors
ECE 6640
Digital Communications I: Modulation and Coding Course, Period 3 2006, Sorour Falahati, Lecture 13
Digital Communication
13
Lecture 11
12 / 12
March 9, 2015
Digital Communication
Lecture 12
1/7
Bonus Question
Digital Communication
Lecture 12
2/7
Todays Lecture
Equalization
Digital Communication
Lecture 12
3/7
received signal
Digital Communication
Lecture 12
4/7
Equalizer
demodulation
Digital Communication
Lecture 12
5/7
Equalizer
demodulation
Categories:
Analog equalizer
Digital equalizer (more common)
1 Linear equalizer: ZF, MMSE
2 Non-linear equalizer: ML (using Viterbi)
Digital Communication
Lecture 12
5/7
Linear Equalizers
We have
Heq (z) =
L
X
wi z i
i=0
E[(dk dk )2 ], k = 0, . . . , L
Digital Communication
Lecture 12
6/7
Digital Communication
Lecture 12
7/7