Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Content of the course
Course requirements and references
Mobile Communications
Instructor: Nguyen Le Hung
Email: nlhung@dut.udn.vn; nglhung@gmail.com
Department of Electronics & Telecommunications Engineering
Danang University of Technology, University of Danang
Mobile Communications Course Information 1
Abstract of the course
Content of the course
Course requirements and references
Abstract of the course
This undergraduate course helps students to understand
mathematical fundamentals and practical transmission
techniques in 4G mobile communications (i.e., WiMAX,
LTE).
The course lecture notes also provide some possible
research directions (in 4G mobile broadband
communications) that can be considered for ﬁnalyear
projects of undergraduate students.
Mobile Communications Course Information 2
Abstract of the course
Content of the course
Course requirements and references
Content of the course
Chapter 1: Introduction
History & development of 1G/2G/3G/4G networks.
Promises and future trends
Cellular mobile communications
Chapter 2: Mobile wireless channel models.
Path loss
Shadowing
Multipath fading channels
Chapter 3: Physicallayer transmission techniques.
Digital modulations
Performance of digital modulations over fading channels
Orthogonal frequency division multiplexing (OFDM)
fundamentals
Mobile Communications Course Information 3
Abstract of the course
Content of the course
Course requirements and references
Course requirements and references
Prerequisite: Basis knowledge of statistics, stochastic
processes and digital communications systems.
Class lecture notes: based on the following references:
Gordan L. St¨ uber, Principles of Mobile Communication,
Second Edition, 2002
A. Goldsmith, Wireless Communications, Cambridge
2005.
Recent IEEE journal and conference papers.
Course assessment:
Exercises and/or projects: 20%
Midterm exam: 30%
Final exam: 50%
Mobile Communications Course Information 4
Outline
Introduction
Cellular mobile communications
Chapter 1: Introduction to Mobile
Communications
Nguyen Le Hung
Mobile Communications Chapter 1: Introduction to Mobile Communications 1
Outline
Introduction
Cellular mobile communications
Outline of Chapter 1
1
Introduction
Development of mobile communication systems
Mobile broadband technology evolution
Promises and future trends
2
Cellular mobile communications
System model
Frequency reuse
Cellular concept
Mobile Communications Chapter 1: Introduction to Mobile Communications 2
Outline
Introduction
Cellular mobile communications
Development of mobile communication systems
Mobile broadband technology evolution
Promises and future trends
Development of mobile communications systems
time
code
frequency
code
space
FDMA (1G)
e.g., AMPS ~ 1980s
TDMA (2G)
e.g., GSM ~ 1990s
OFDM, SDMA (4G)
e.g., WiMAX, LTE
2010s
CDMA (3G)
e.g., WCDMA ~ 2000s
frequency
time
time
~ 1 Gbps (stationary),
~ 100 Mbps (mobile)
frequency
frequency
~ 14 Mbps (downlink),
~ 5.8 Mbps (uplink)
~ 50 Kbps
A new signal dimension will be exploited in 5G ?
Mobile Communications Chapter 1: Introduction to Mobile Communications 3
Outline
Introduction
Cellular mobile communications
Development of mobile communication systems
Mobile broadband technology evolution
Promises and future trends
OFDM versus FDMA
Frequency
Mobile Communications Chapter 1: Introduction to Mobile Communications 4
Outline
Introduction
Cellular mobile communications
Development of mobile communication systems
Mobile broadband technology evolution
Promises and future trends
Mobile broadband technology evolution
LTE
Advanced
802.16m
R 2.0
WCDMA
3GPP
HSDPA HSPA
Mobile
WiMAX
802.16e
R 1.0
EVDO
Rev A
EVDO
Rev B
CDMA 2000
3GPP2
LTE
802.16e
R 1.5
2005 2009 2010 2006 2007 2008 2011 2012
CDMA based OFDMA based
R99
“4G”
IMTAdvanced
Mobile Communications Chapter 1: Introduction to Mobile Communications 5
Outline
Introduction
Cellular mobile communications
Development of mobile communication systems
Mobile broadband technology evolution
Promises and future trends
Promises and future trends
multimedia services: Voice, Video distribution, Real
time videoconferencing, Data,… for both business
and residential customers:
Explosive traffic growth
Internet growth, VoIP, VideoIP, IPTV
Cell phone popularity worldwide
Ubiquitous communication for people and devices
Emerging systems opening new applications
Unified network: Single distributed network,
multiple services, packet architecture
Extracted from Digital Communication lecture notes, McGill Uni.
Mobile Communications Chapter 1: Introduction to Mobile Communications 6
Outline
Introduction
Cellular mobile communications
System model
Frequency reuse
Cellular concept
System model of cellular mobile communications
BTS
LTE/LTE Ͳ Advanced
Single Cell
MultiͲcell
approach using
game theory
Uplink (SCͲFDMA),
limited feedback design
Downlink (OFDMA)
SingleͲUser
Multiuser
Precoding
(SDMA)
MultiͲhop
Relay
BTS
BTS
BTS
BTS
InterͲcell
interference
InterͲcell
interference
SingleͲuser/MultiͲhop:
Channel Estimation,
Synchronization (CFO),
Channel Coding, ...
Network Controller
STBC with highͲspeed users
(large Doppler spread)
Cognitive radio
Space Time Block Code: STBC; PeakͲtoͲAverage Power Ratio: PAPR; VͲOFDM
UserͲcooperation
(cooperative/multiͲhop
communications)
Mobile Communications Chapter 1: Introduction to Mobile Communications 7
Outline
Introduction
Cellular mobile communications
System model
Frequency reuse
Cellular concept
Frequency reuse
The available spectrum is partitioned among the base stations
(BSs).
A given frequency band is reused at the closest possible
distance under a certain requirement of cochannel
interference.
Smaller cells have a shorter distance between reused
frequencies =⇒ an increased spectral eﬃciency.
Microcells are of great importance in improving spectral
eﬃciency.
Under frequencyreuse, users in geographically separated cells
simultaneously employ the same carrier frequency.
Mobile Communications Chapter 1: Introduction to Mobile Communications 8
Outline
Introduction
Cellular mobile communications
System model
Frequency reuse
Cellular concept
Cellular concept
The cellular layout of a conventional cellular system is quite
often described by a uniform grid of hexagonal cells or radio
coverage zones.
In practice the cells are not regular hexagons, but instead are
distorted and overlapping areas.
The hexagon is an ideal choice for representing macrocellular
coverage areas, because it closely approximates a circle and
oﬀers a wide range of tessellating reused cluster sizes.
A tessellating reuse cluster of size N can be constructed if
=
2
+ +
2
, (1)
where and are nonnegative integers and ≥ . It follows
that the allowable cluster sizes are = 1, 3, 4, 7, 9, 12, . . ..
Mobile Communications Chapter 1: Introduction to Mobile Communications 9
Outline
Introduction
Cellular mobile communications
System model
Frequency reuse
Cellular concept
Cellular concept: Multicell layout with frequencyreuse
3cell
4cell
7cell
Macrocellular deployment
with 7cell clusters
Macrocellular deployment
with 3cell clusters
Mobile Communications Chapter 1: Introduction to Mobile Communications 10
Introduction
Wireless channel modeling
Chapter 2: Wireless Channel models
Mobile Communications Chapter 2: Wireless Channel models 1
Introduction
Wireless channel modeling
Multipath wireless propagation
Path loss, shadowing and fading
Multipath wireless propagation
reflection and diffraction
Extracted from Digital Communication lecture notes, McGill Uni.
Mobile Communications Chapter 2: Wireless Channel models 2
Introduction
Wireless channel modeling
Multipath wireless propagation
Path loss, shadowing and fading
Path loss, shadowing and fading
The characteristic of (mobile) wireless channel is the variations of
the channel strength over time and frequency.
The variations can be divided into two types:
Largescale fading is yielded by:
path loss of signal as a function of distance and
shadowing by large objects such as buildings and hills.
Smallscale fading is yielded by the constructive and destructive
interference of the multiple signal paths between transmitter and
receiver.
Mobile Communications Chapter 2: Wireless Channel models 3
Introduction
Wireless channel modeling
Multipath wireless propagation
Path loss, shadowing and fading
An example of path loss, shadowing and fading
0 50 100 150 200 250 300 350
150
140
130
110
100
90
80
70
60
50
R
e
c
e
i
v
e
d
P
o
w
e
r
[
d
B
m
]
Traveled distance [m]
Pathloss
Fading +
Shadowing +
Pathloss
Shadowing +
Pathloss
Mobile Communications Chapter 2: Wireless Channel models 4
Introduction
Wireless channel modeling
Multipath wireless propagation
Path loss, shadowing and fading
An example of path loss, shadowing and fading (cont.)
0
K (dB)
P
r
P
(dB)
t
log (d)
Path Loss Alone
Shadowing and Path Loss
Multipath, Shadowing, and Path Loss
Mobile Communications Chapter 2: Wireless Channel models 5
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Path loss models
It is well known that the received signal power decays with the
square of the path length in free space.
More speciﬁcally, the received envelope power is
1
:
= 1

G

G
:
(
X
4tJ
)
2
, (1)
where:
1
is the transmitted power,
G
and G
are the transmitter and receiver antenna gains,
respectively
d is the radio path length.
Mobile Communications Chapter 2: Wireless Channel models 6
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Path loss models (cont.)
The signals in land mobile radio applications, however, do not
experience free space propagation. A more appropriate theoretical
model assumes propagation over a ﬂat reﬂecting surface (the earth).
1
:
= 41

(
`
t
4¬d
)
2
G

G
:
sin
2
(
2¬ℎ
b
ℎ
n
`
t
d
)
, (2)
where ℎ
b
and ℎ
n
are the heights of the BS and MS antennas,
respectively.
Under the condition that d ≫ℎ
b
ℎ
n
, (2) reduces to
1
:
= 1

G

G
:
(
`
t
4¬d
)
2
, (3)
where we have used the approximation sin r ≈ r for small r.
Mobile Communications Chapter 2: Wireless Channel models 7
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Path loss models (cont.)
The path loss is deﬁned by
1
¡ (J1)
= 10 log
10
(
1

G

G
:
1
:
)
= −10 log
10
{
4
(
`
t
4¬d
)
2
sin
2
(
2¬ℎ
b
ℎ
n
`
t
d
)
}
(4)
Several useful empirical models for macrocellular systems have been
obtained by curve ﬁtting experimental data.
Two of the useful models for 900 MHz cellular systems are:
Hata’s model based on Okumura’s prediction method and
Lee’s model.
Hata’s empirical model is probably the simplest to use. The
empirical data for this model was collected by Okumura in the city
of Tokyo.
Mobile Communications Chapter 2: Wireless Channel models 8
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
OkumuraHata models
With OkumuraHata’s model, the path loss between two isotropic
BS and MS antennas is
1
¡ (J1)
=
⎧
⎨
⎩
¹ +1log
10
(d) for urban area
¹ +1log
10
(d) −C for suburban area
¹ +1log
10
(d) −1 for open area
(5)
where
¹ = 69.55 + 26.16 log
10
()
t
) −13.82 log
10
(ℎ
b
) −o(ℎ
n
)
1 = 49.9 −6.55 log
10
(ℎ
b
)
C = 5.4 + 2 (log
10
()
t
,28))
2
1 = 40.94 + 4.78 (log
10
()
t
))
2
−18.33 log
10
()
t
)
Mobile Communications Chapter 2: Wireless Channel models 9
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
OkumuraHata models (cont.)
and
o
ℎ
=
⎧
⎨
⎩
[1.1 log
10
()
) −0.7] ℎ
−1.56 log
10
()
) + 0.8 for medium or small city
{
8.28 [log
10
(1.54ℎ
)]
2
−1.1 for )
≤ 200MHz
3.2 [log
10
(11.75ℎ
)]
2
−4.97 for )
≥ 400MHz
for large city
(6)
OkumuraHata’s model is expressed in terms of:
the carrier frequency: 150 ≤ )
≤ 1000(MHz),
BS antenna height: 30 ≤ ℎ
≤ 200(m),
the mobile station (MS) height: 1 ≤ ℎ
≤ 10(m),
the distance: 1 ≤ d ≤ 20(km).
Mobile Communications Chapter 2: Wireless Channel models 10
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Numerical results of OkumuraHata models
1 5 10 15 20
100
150
200
250
300
350
400
450
P
a
t
h
l
o
s
s
(
d
B
)
Distance d (km) under scale of log10
urban area
suburban area
open area
Figure 1: Path loss for ℎ
= 1.5m, ℎ
= 50m, )
= 900MHz.
Mobile Communications Chapter 2: Wireless Channel models 11
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Shadowing
A signal transmitted through a wireless channel will typically
experience random variation due to blockage from objects in the
signal path, giving rise to random variations of the received power at
a given distance.
Such variations are also caused by changes in reﬂecting surfaces and
scattering objects.
Thus, a model for the random attenuation due to these eﬀects is
also needed. Since the location, size, and dielectric properties of the
blocking objects as well as the changes in reﬂecting surfaces and
scattering objects that cause the random attenuation are generally
unknown, statistical models must be used to characterize this
attenuation.
The most common model for this additional attenuation is
lognormal shadowing.
Mobile Communications Chapter 2: Wireless Channel models 12
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Shadowing (cont.)
Empirical studies have shown that A
n
has the following lognormal
distribution:
j
·
(r) =
2
ro
·
¸
√
2¬
exp
{
−
(
10 log
10
r
2
− j
·
(dBm)
)
2o
2
·
}
j
·
(r) =
2
ro
·
¸
√
2¬
exp
{
−
(
10 log
10
r −j
·
(dBm)
)
2o
2
·
}
where:
A
and A
denote the mean envelop and mean squared levels of
received signal (where the expectation is taken over the pdf of the
received envelope).
o
stands for standard deviation.
j
(dBm)
= 30 + 10[log
10
A
2
]
j
(dBm)
= 30 + 10[log
10
A
]
Mobile Communications Chapter 2: Wireless Channel models 13
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Shadowing (cont.)
Sometimes A
n
is called the local mean because it represents the
mean envelope level where the averaging is performed over a
distance of a few wavelengths that represents a locality.
This model has been conﬁrmed empirically to accurately model the
variation in received power in both outdoor and indoor radio
propagation environments
Mobile Communications Chapter 2: Wireless Channel models 14
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Fading channel model
Two Main
Multipaths
Local
Scattering
The complex transmitted signal can be expressed by
:(t) = Re
[
r(t)c
¸2t}

]
. (7)
Over a multipath (1 physical paths) propagation channel, the
received signal can be obtained by
j
1J
(t) =
J−1
∑
l=0
c
l
(t):(t −t
l
(t)) +n(t). (8)
Mobile Communications Chapter 2: Wireless Channel models 15
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Fading channel model (cont.)
Substituting (7) into (8) yields the following
j
1J
(t) = Re
[
J−1
∑
l=0
c
l
(t)r(t −t
l
(t)) c
¸2t}
(−r
())
]
+n(t)
= Re
[(
J−1
∑
l=0
c
l
(t)r(t −t
l
(t))
)
c
¸2t}

]
+n(t)
= Re
[
j(t)c
¸2t}

]
+n(t)
As a result, the received baseband signal can be determined by
j(t) =
∑
.
c
.
(t)r(t −t
.
(t)) +n
b
(t). (9)
where n
b
(t) is the receiver (thermal) noise signal.
Mobile Communications Chapter 2: Wireless Channel models 16
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Wireless channel modeling (cont.)
The next step in creating a useful channel model is to convert the
continuoustime channel to a discretetime channel.
We take the usual approach of sampling theorem.
Assuming that the input waveform is bandlimited to \, the
baseband equivalent can be represented by
r(t) =
∑
n
r
n
sinc(\t −n), (10)
where r
n
= r(n,\) and sinc(t) ≜
sin(t)
t
.
This representation follows from the sampling theorem, which says
that any waveform bandlimited to \,2 can be expanded in terms
of the orthogonal basis functions sinc(\t −n) with coeﬃcients by
samples (taken uniformly at integer multiples of 1,\)
Mobile Communications Chapter 2: Wireless Channel models 17
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Wireless channel modeling (cont.)
As a result, the baseband received signal can be determined by
j(t) =
∑
.
c
.
(t)
∑
n
r
n
sinc (\(t −t
.
(t)) −n) +n
b
(t)
=
∑
n
r
n
∑
.
c
.
(t)sinc (\(t −t
.
(t)) −n) +n
b
(t).
The sampled outputs at multiples of 1,\ is j
n
≜ j(:,\) then
j
n
=
∑
n
r
n
∑
.
c
.
(:,\)sinc (:−n −t
.
(:,\)\) +n
b
(:,\).
(11)
Mobile Communications Chapter 2: Wireless Channel models 18
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Wireless channel modeling (cont.)
Let  ≜ :−n then one can have
j
n
=
∑
l
r
n−l
∑
.
c
.
(:,\)sinc ( −t
.
(:,\)\) +n
b
(:,\)
Then, the discretetime channel model can be given by
j
n
=
∑
l
r
n−l
ℎ
l,n
+n
b
(:,\) (12)
where ℎ
l,n
=
∑
.
c
.
(:,\)sinc ( −t
.
(:,\)\)
This simple discretetime signal model is widely used in
physicallayer transmission techniques in OFDM systems (e.g., WiFi,
WiMAX, LTE)
Mobile Communications Chapter 2: Wireless Channel models 19
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Examples of transmitted baseband signal
:
01
00
10
11
I
+1 –1
–1
+1
Q
b
0
b
1
0
1
I
+1 –1
–1
+1
Q
b
0
11 10
11 11 10 11
10 10
I
+1 –1
–1
+1
Q
b
0
b
1
b
2
b
3
+3
11 01
11 00 10 00
10 01
+3
00 10
00 11 01 11
01 10
00 01
00 00 01 00
01 01
–3
–3
BPSK
QPSK
16QAM
Over multipath channels, the received signal at MS is:
j
n
=
∑
l
r
n−l
ℎ
l,n
+n
b
(:,\) (13)
It is noted that multipath fading gains ℎ
l,n
(channel impulse
response) is timevariant (depend on time index :).
Mobile Communications Chapter 2: Wireless Channel models 20
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Channel estimation in mobile communications
Source
encoder
Channel
encoder
Digital
modulation
Channel
Source
decoder
Channel
decoder
Digital
demodulation
S
h
r = Sh + n
Pilot
S
Data
S
Data
S
Pilot
S
Data
S
Data
S
Pilot
S
h h h h’ h’ h’
Mobile Communications Chapter 2: Wireless Channel models 21
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Literature Review of Channel Estimation in Wireless
Communications
Detection/decoding
in communications
3 dB f
Rx signal
vector
Tx signal
matrix
CIR
vector
Rx noise
vector
Noncoherent Coherent without using CSI
3dB performance
loss
use CSI
require Channel Estimation (CE)
vector matrix vector
(CSI)
vector
r = Sh + n
require Channel Estimation (CE)
with channel parameters as:
Deterministic unknowns Random variables
Fisher approaches:
LS ML
Bayesian approaches:
MMSE MAP LS, ML,… MMSE, MAP,…
Multipath fading channel (freq. selective) in multicarrier transmissions (e.g.,OFDM)
Timeinvariant (quasistatic) Timevariant (Timeselective)
Perfect
Synch.
Imperfect
Synch.
Channel Estimation (CE)
Blind Pilot Semiblind
Joint CE and Synch.
Semiblind
Perfect
Synch.
Imperfect
Synch.
Channel Estimation (CE)
Pilot
Joint CE and Synch.
Semiblind Pilot
Pilot design to minimize:
MSE CRLB
Pilot design to minimize:
MSE CRLB
Pilot design to minimize:
MSE BCRLB
Turbobased
Decisiondirect.
MSE CRLB MSE CRLB MSE BCRLB
Mobile Communications Chapter 2: Wireless Channel models 22
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
Timevariant path gain ℎ
,:
under mobile speed of 5 km/h
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
Time (in OFDM symbol duration)
A
b
s
o
l
u
t
e
v
a
l
u
e
o
f
a
m
p
l
i
t
u
d
e
o
f
o
n
e
p
a
t
h
g
a
i
n
h
l
Mobile user speed = 5 km/h,
f
c
= 2 GHz,
128−FFT, CP length = 10,
f
s
= 1.92 MHz,
2 time slots in LTE are considered,
Jakes model is considered.
pilot OFDM symbol
for channel estimation
Mobile Communications Chapter 2: Wireless Channel models 23
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
ℎ
,:
under mobile speed of 50 km/h
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
0.95
1
1.05
1.1
1.15
Time (in OFDM symbol duration)
A
b
s
o
l
u
t
e
v
a
l
u
e
o
f
a
m
p
l
i
t
u
d
e
o
f
o
n
e
p
a
t
h
g
a
i
n
h
l
Mobile user speed = 50 km/h,
f
c
= 2 GHz,
128−FFT, CP length = 10,
f
s
= 1.92 MHz,
2 time slots in LTE are considered,
Jakes model is considered
Data OFDM symbol
Mobile Communications Chapter 2: Wireless Channel models 24
Introduction
Wireless channel modeling
Path loss models
Shadowing
Fading channel model
ℎ
,:
under mobile speed of 300 km/h
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
0.8
0.9
1
1.1
1.2
1.3
Time (in OFDM symbol duration)
A
b
s
o
l
u
t
e
v
a
l
u
e
o
f
a
m
p
l
i
t
u
d
e
o
f
o
n
e
f
a
d
i
n
g
g
a
i
n
h
l
Mobile user speed = 300 km/h,
f
c
= 2 Ghz, 128−FFT, CP length = 10, f
s
= 1.92 Mhz,
2 time slots in LTE are considered,
Jakes model is considered.
Data OFDM symbol
Mobile Communications Chapter 2: Wireless Channel models 25
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Chapter 3: Physicallayer transmission techniques
Section 3.1: Digital modulations
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 1
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Outline of the lecture notes
1
Digital modulation techniques
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
2
Signal Space Analysis
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
3
Receiver Structure and Suﬃcient Statistics
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
4
Error Probability Analysis and the Union Bound
Error probability
The union bound on error probability
5
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 2
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Advantages over analog modulation
The advances over the last several decades in hardware and
digital signal processing have made digital transceivers much
cheaper, faster, and more powereﬃcient than analog
transceivers.
More importantly, digital modulation oﬀers a number of other
advantages over analog modulation, including:
higher data rates,
powerful error correction techniques,
resistance to channel impairments,
more eﬃcient multiple access strategies, and
better security and privacy.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 3
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Advantages over analog modulation (cont.)
Digital transmissions consist of transferring information in the
form of bits over a communications channel.
The bits are binary digits taking on the values of either 1 or 0.
These information bits are derived from the information
source, which may be a digital source or an analog source that
has been passed through an A/D converter.
Both digital and A/D converted analog sources may be
compressed to obtain the information bit sequence.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 4
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Main considerations in digital modulation techniques
Digital modulation consists of mapping the information bits
into an analog signal for transmission over the channel.
Detection consists of determining the original bit sequence
based on the signal received over the channel.
The main considerations in choosing a particular digital
modulation technique are:
high data rate
high spectral eﬃciency (minimum bandwidth occupancy)
high power eﬃciency (minimum required transmit power)
robustness to channel impairments (minimum probability of bit
error)
low power/cost implementation
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 5
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Typical types of digital modulation techniques
Often the previous ones are conﬂicting requirements, and the
choice of modulation is based on ﬁnding the technique that
achieves the best tradeoﬀ between these requirements.
There are two main categories of digital modulation:
amplitude/phase modulation
frequency modulation
Frequency modulation typically has a constant signal envelope
and is generated using nonlinear techniques, this modulation
is also called constant envelope modulation or nonlinear
modulation
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 6
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Typical types of digital modulation techniques (cont.)
Amplitude/phase modulation is also called linear modulation.
Linear modulation generally has better spectral properties
than nonlinear modulation, since nonlinear processing leads to
spectral broadening.
However, amplitude and phase modulation embeds the
information bits into the amplitude or phase of the
transmitted signal, which is more susceptible to variations
from fading and interference.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 7
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Advantages over analog modulation
Main considerations in digital modulation techniques
Typical types of digital modulation techniques
Typical types of digital modulation techniques (cont.)
In addition, amplitude and phase modulation techniques
typically require linear ampliﬁers, which are more expensive
and less power eﬃcient than the nonlinear ampliﬁers that can
be used with nonlinear modulation.
Thus, the general tradeoﬀ of linear versus nonlinear
modulation is one of better spectral eﬃciency for the former
technique and better power eﬃciency and resistance to
channel impairments for the latter technique.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 8
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Rational
Digital modulation encodes a bit stream of ﬁnite length into
one of several possible transmitted signals.
Intuitively, the receiver minimizes the probability of detection
error by decoding the received signal as the signal in the set of
possible transmitted signals that is closest to the one received.
Determining the distance between the transmitted and
received signals requires a metric for the distance between
signals.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 9
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Rational (cont.)
By representing signals as projections onto a set of basis
functions, we obtain a onetoone correspondence between the
set of transmitted signals and their vector representations.
Thus, we can analyze signals in ﬁnitedimensional vector
space instead of inﬁnitedimensional function space, using
classical notions of distance for vector spaces.
In this section we show:
how digitally modulated signals can be represented as vectors
in an appropriatelydeﬁned vector space, and
how optimal demodulation methods can be obtained from this
vector space representation.
This general analysis will then be applied to speciﬁc
modulation techniques in later sections.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 10
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Transmitted signal
Transmitter Receiver
+
n(t)
AWGN Channel
s(t)
i 1
K
m ={b ,...,b }
^
1 K
m ={b ,...,b }
^ ^
r(t)
y
s
s
1
t)
M
t)) [0, T) ≤
φ
1
t) t)} s
Figure 1: Communication system model over AWGN channel (i.e., a
special case of wireless channel).
Consider a communication system model as shown in the
above ﬁgure.
Every T seconds, the sytem sends 1 = log
2
` bits of
information through the channel for a data rate of 1 = 1,T
bits per second (bps).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 11
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Transmitted signal (cont.)
There are ` = 2
1
possible sequences of 1 bits and each bit
sequence of length 1 comprises a message
:
i
= {/
1
, ..., /
1
} ∈ , where = {:
1
, ..., :
A
} is the set of
all such messages.
The message :
i
has probability j
i
of being selected for
transmission, where
∑
A
i=1
j
i
= 1.
Suppose that message :
i
is to be transmitted over the
AWGN channel during the time interval [0, T). Since the
channel is analog, the message must be embedded into an
analog signal for channel transmission.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 12
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Transmitted signal (cont.)
Therefore, each message :
i
∈ is mapped to a unique
analog signal :
i
(t) ∈ = {:
1
(t), ..., :
A
(t)} where :
i
(t) is
deﬁned on the time interval [0, T) and has energy
1

=
∫
T
0
:
2
i
(t)dt, i = 1, ..., `. (1)
When messages :
i
are sent sequentially, the transmitted
signal becomes a sequence of the corresponding analog signals
as follows
:(t) =
∑
I
:
i
(t −/T). (2)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 13
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Transmitted and received signals
In the aforementioned model, the transmitted signal is sent
through an AWGN channel where a white Gaussian noise
process n(t) of power spectral density ·
c
,2 is added to form
the received signal
:(t) = :(t) +n(t). (3)
T 0 2T 3T 4T
s (t)
1
1
1
2
s (t−T)
s (t−2T)
s (t−3T)
s(t)
...
m
1
m
1
m
1
m
2
s
1
t)
M
t)) [0, T)
φ
1
t) t)} s
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 14
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Received signal
Given :(t), the receiver must determine the best estimate of
which :
i
(t) ∈ was transmitted during each transmission
interval [0, T).
This best estimate of :
i
(t) is mapped to a best estimate of
the message :
i
(t) ∈ and the receiver produces this best
estimate ˆ : =
{
ˆ
/
1
, ...,
ˆ
/
1
}
of the transmitted bit sequence.
The goal of the receiver design in estimating the transmitted
message is to minimize the probability of message error
1
c
=
A
∑
i=1
j ( ˆ : ∕= :
i
∣:
i
sent) j (:
i
sent) (4)
over each time interval [0, T).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 15
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Introduction
By representing the signals {:
i
(t), i = 1, ..., `} geometrically,
one can solve for the optimal receiver design in AWGN
channels based on a minimum distance criterion.
Note that, wireless channels typically have a timevarying
impulse response in addition to AWGN. We will consider the
eﬀect of an arbitrary channel impulse response on digital
modulation performance in the next sections.
The basic premise behind a geometrical representation of
signals is the notion of a basis set.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 16
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Basis function representation of signals
Speciﬁcally, using a GramSchmidt orthogonalization
procedure, it can be shown that any set of ` real energy
signals o = {:
1
(t), ..., :
A
(t)} deﬁned on [0, T) can be
represented as a linear combination of · ≤ ` real
orthogonal basis functions {c
1
(t), ..., c
A
(t)}.
We say that these basis functions span the set .
Each signal {:
i
(t) ∈ } can be represented by
:
i
(t) =
.
∑
)=1
:
i,)
c
)
(t), 0 ≤ t < T, (5)
where
:
i,)
=
∫
T
0
:
i
(t)c
)
(t)dt (6)
is a real coeﬃcient representing the projection.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 17
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Basis function representation of signals (cont.)
These basis functions have the following property
∫
T
0
c
i
(t)c
)
(t)dt =
{
1 i = ,,
0 i ∕= ,.
(7)
The basis set consists of the sine and cosine functions
c
1
(t) =
√
2
T
cos (2¬)
c
t) (8)
and
c
2
(t) =
√
2
T
sin(2¬)
c
t) . (9)
where
√
2
T
is used to obtain
∫
T
0
c
2
i
(t)dt = 1, i = 1, 2.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 18
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Basis functions in linear passband modulation techniques
With these basis functions, one only obtain an approximation
to (7), since
∫
T
0
c
2
1
(t)dt =
2
T
∫
T
0
0.5 [1 + cos (4¬)
c
t)] dt = 1+
sin (4¬)
c
t)
4¬)
c
t
(10)
The numerator in the second term of (10) is bounded by 1,
and for )
c
T ≫ 1 the denominator of this term is very large.
As a result, this second term can be neglected.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 19
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Basis functions in linear passband modulation (cont.)
With these basis functions, one can have
∫
T
0
c
1
(t)c
2
(t)dt =
2
T
∫
T
0
0.5 sin (4¬)
c
t) dt =
−cos (4¬)
c
t)
4¬)
c
t
≈ 0
(11)
where the approximation is taken as an equality as )
c
T ≫ 1.
With the basis set c
1
(t) =
√
2,T cos (2¬)
c
t) and
c
2
(t) =
√
2,T sin (2¬)
c
t), the basis function representation
(5) corresponds to the complex representation of :
i
(t) in
terms of its inphase and quadrature components with an
extra factor of
√
2,T as follows
:
i
(t) = :
i,1
√
2
T
cos (2¬)
c
t) +:
i,2
√
2
T
sin(2¬)
c
t) . (12)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 20
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Basis functions in linear passband modulation (cont.)
In practice, the basis set may include a baseband pulseshaping
ﬁlter p(t) to improve the spectral characteristics of the transmitted
signal:
:
i
(t) = :
i,1
p(t) cos (2¬)
c
t) +:
i,2
p(t) sin (2¬)
c
t) (13)
where the simplest pulse shape that satisfy (7) is the rectangular
pulse shape p(t) =
√
2,T, 0 ≤ t < T.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 21
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Deﬁnitions used in signal space representation
We denote the coeﬃcients {:
i,)
} as a vector
s
i
= [:
i,1
, ..., :
i,.
] ∈ ℛ
.
which is called the signal
constellation point corresponding to the signal :
i
(t).
The signal constellation consists of all constellation points
{s
1
, ..., s
A
}.
Given the basis functions {c
1
(t), ..., c
.
(t)} there is a
onetoone correspondence between the transmitted signal
:
i
(t) and its constellation point s
i
.
The representation of :
i
(t) in terms of its constellation point
s
i
∈ ℛ
.
is called:
its signal space representation and
the vector space containing the constellation is called the
signal space.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 22
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Deﬁnitions used in signal space representation (cont.)
A twodimensional signal space is illustrated in the below
ﬁgure, where we show s
i
∈ ℛ
2
with the ith axis of ℛ
2
corresponding to the basis function c
i
(t), i = 1, 2.
M=4, K=2
00 11
01
10
M=8, K=3
000
001
011
110
100
010
110
101
s
i1
s
i2
s
i1
s
i2
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 23
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Deﬁnitions used in signal space representation (cont.)
With this signal space representation we can analyze the
inﬁnitedimensional functions :
i
(t) as vectors s
i
in
ﬁnitedimensional vector space ℛ
2
.
This greatly simpliﬁes the analysis of the system performance
as well as the derivation of the optimal receiver design.
Signal space representations for common modulation
techniques like MPSK and MQAM are twodimensional
(corresponding to the inphase and quadrature basis
functions).
In order to analyze signals via a signal space representation,
we need to use some deﬁnitions for the vector characterization
in the vector space ℛ
.
.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 24
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Deﬁnitions used in signal space representation (cont.)
In particular, the length of a vector in ℛ
.
is deﬁned as
∥s
i
∥ =
⎷
.
∑
)=1
:
2
i,)
. (14)
The distance between two signal constellation points s
i
and s
I
is thus
∥s
i
−s
I
∥ =
⎷
.
∑
)=1
(:
i,)
−:
I,)
)
2
=
√
∫
T
0
(:
i
(t) −:
I
(t))
2
dt.
(15)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 25
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Rational
Signal and system model
Geometric representation of signals
Practical examples
Signal space representation
Deﬁnitions used in signal space representation (cont.)
Finally, the inner product ⟨:
i
(t), :
I
(t)⟩ between two real
signals :
i
(t) and :
I
(t) on the interval [0, T) is deﬁned as
⟨:
i
(t), :
I
(t)⟩ =
∫
T
0
:
i
(t):
I
(t)dt. (16)
Similarly, the inner product ⟨s
i
, s
I
⟩ between two real vectors is
⟨s
i
, s
I
⟩ = s
i
s
T
I
=
∫
T
0
:
i
(t):
I
(t)dt = ⟨:
i
(t), :
I
(t)⟩. (17)
It is noted that two signals are orthogonal if their inner
product is zero.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 26
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Receiver structure and suﬃcient statistics
Given the channel output :(t) = :
i
(t) +n(t), 0 ≤ t < T, we
now investigate the receiver structure to determine which
constellation point s
i
or, equivalently, which message :
i
, was
sent over the time interval [0, T).
A similar procedure is done for each time interval
[/T, (/ + 1)T).
We would like to convert the received signal :(t) over each
time interval into a vector, since it allows us to work in
ﬁnitedimensional vector space to estimate the transmitted
signal.
However, this conversion should not compromise the
estimation accuracy. For this conversion, consider the receiver
structure shown in the next ﬁgure.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 27
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Receiver structure and suﬃcient statistics (cont.)
) ( ) ( ) ( t n t s t r
i
³
T
dt
0
()
³
T
dt
0
()
1 1 1 ,
r n s
i
) (
1
t I
) (t
N
I
N N N i
r n s
,
Find i
i
m m ˆ
As shown in the above ﬁgure, the components of signal and
noise vectors are determined by
:
i,)
=
∫
T
0
:
i
(t)c
)
(t)dt, (18)
and
n
)
=
∫
T
0
n(t)c
)
(t)dt. (19)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 28
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Receiver structure and suﬃcient statistics (cont.)
We can rewrite :(t) as
:(t) =
.
∑
)=1
(:
i,)
+n
)
) c
)
(t) +n
·
(t) =
.
∑
)=1
:
)
c
)
(t) +n
·
(t),
(20)
where :
)
= :
i,)
+n
)
and n
·
(t) = n(t) −
∑
.
)=1
n
)
c
)
(t)
denotes the remainder noise.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 29
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection
If we can show that the optimal detection of the transmitted
signal constellation point s
i
given received signal :(t) does not
make use of the remainder noise n
·
(t), then the receiver can
make its estimate ˆ : of the transmitted message :
i
as a
function of r = (:
1
, ..., :
.
) alone.
In other words, r = (:
1
, ..., :
.
) is a suﬃcient statistic for :(t)
in the optimal detection of the transmitted messages.
Let exam the distribution of r. Since n(t) is a Gaussian
random process, if we condition on the transmitted signal
:
i
(t) then the channel output :(t) = :
i
(t) +n(t) is also a
Gaussian random process and r = [:
1
, ..., :
.
] is a Gaussian
random vector.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 30
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
Recall that :
i
= :
i,)
+n
)
. Thus, conditioned on the
transmitted constellation s
i
, we have
j
·
∣s
= 1[:
)
∣s
i
] = 1 [:
i,)
+n
)
∣:
i,)
] = :
i,)
(21)
since n(t) has zero mean, and
o
·
∣s
= 1
[
:
)
−j
·
∣s
]
2
= 1[:
i,)
+n
)
−:
i,)
∣:
i,)
]
2
= 1
[
n
2
)
]
.
(22)
With Cov [:
)
:
I
∣s
i
] = 1
[(
:
)
−j
·
)
(:
I
−j
·
) ∣s
i
]
= 1 [n
)
n
I
]
and some manipulations, one can have
1 [n
)
n
I
] =
·
0
2
∫
T
0
c
)
(t)c
I
(t)dt =
{
·
0
,2 , = /
0 , ∕= /.
. (23)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 31
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
Thus, conditioned on the transmitted constellation s
i
, the :
)
’s
are uncorrelated and, since they are Gaussian and also
independent. Moreover, 1
[
n
2
)
]
= ·
0
,2.
We have shown that, conditioned on the transmitted
constellation s
i
, :
)
is a Gaussdistributed random variable that
is independent of :
I
, / ∕= ,and has mean :
i,)
and variance
·
0
,2.
Thus, the conditional distribution of r is given by
j (r∣s
.
sent) =
Þ
∏
¸=1
j (:
¸
∣:
.
) =
1
(¬·
0
)
Þ¸2
exp
⎡
⎣
−
1
·
0
Þ
∑
¸=1
(:
¸
−:
.,¸
)
2
⎤
⎦
.
(24)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 32
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
It is also straightforward to show that 1[:
)
n
·
(t)∣s
i
] = 0 for
any t, 0 ≤ t < T. Thus, since :
)
conditioned on s
i
and n
·
(t)
are Gaussian and uncorrelated, they are independent.
Also, since the transmitted signal is independent of the noise,
:
i,)
is independent of the process n
·
(t).
We now discuss the receiver design criterion and show it is not
aﬀected by discarding n
·
(t).
The goal of the receiver design is to minimize the probability
of error in detecting the transmitted message :
i
given
received signal :(t).
To minimize 1
c
= j ( ˆ : ∕= :
i
∣:(t)) = 1 −j ( ˆ : = :
i
∣:(t)), we
maximize j ( ˆ : = :
i
∣:(t)).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 33
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
Therefore, the receiver output ˆ : given received signal :(t)
should correspond to the message :
i
that maximizes
j (:
i
sent∣:(t)).
Since there is a onetoone mapping between messages and
signal constellation points, this is equivalent to maximizing
j (:
i
sent∣:(t)).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 34
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
Recalling that :(t) is completely described by : = (:
1
, ..., :
.
)
and n
·
(t), we have
j (s
i
sent∣:(t)) = j ((:
i,1
, ..., :
i,.
) sent∣(:
1
, ..., :
.
, n
·
(t)))
=
j ((:
i,1
, ..., :
i,.
) sent, (:
1
, ..., :
.
), n
·
(t))
j ((:
1
, ..., :
.
), n
·
(t))
=
j ((:
i,1
, ..., :
i,.
) sent, (:
1
, ..., :
.
)) j (n
·
(t))
j ((:
1
, ..., :
.
)) j (n
·
(t))
= j ((:
i,1
, ..., :
i,.
) sent∣(:
1
, ..., :
.
)) . (25)
where the third equality follows from the fact that the n
·
(t) is
independent of both (:
1
, ..., :
.
) and of (:
i,1
, ..., :
i,.
).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 35
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Proofs of suﬃcient statistics for optimal detection (cont.)
This analysis shows that (:
1
, ..., :
.
) is a suﬃcient statistic for
:(t) in detecting :
i
, in the sense that the probability of error
is minimized by using only this suﬃcient statistic to estimate
the transmitted signal and discarding the remainder noise.
Since r is a suﬃcient statistic for the received signal :(t), we
call r the received vector associated with :(t).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 36
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision regions
As aforementioned, the optimal receiver minimizes error
probability by selecting the detector output ˆ : that maximizes
the probability of correct detection
1 −1
c
= j ( ˆ : sent∣r received).
In other words, given a received vector r, the optimal receiver
selects ˆ : = :
i
corresponding to the constellation s
i
that
satisﬁes
j (s
i
∣r) j (s
)
∣r) , ∀, ∕= i (26)
where j (s
i
∣r) ≜ j (s
i
sent∣r received) for the sake of
notational simplicity.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 37
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision regions(cont.)
Thus, the decision regions (7
1
, ..., 7
A
) corresponding to
(s
1
, ..., s
A
) are the subsets of the signal space ℛ
.
and
deﬁned by
7
i
= (r : j (s
i
∣r) j (s
)
∣r) , ∀, ∕= i) . (27)
Once the signal space has been partitioned by decision
regions, for a received vector r ∈ 7
i
, the optimal receiver
outputs the message estimate ˆ : = :
i
The receiver processing consists of o) computing the received
vector r from :(t), /) ﬁnding which decision region 7
i
contains r, and c) outputting the corresponding message :
i
.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 38
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
An example on decision regions
This process is illustrated in the below ﬁgure, that shows a
twodimensional signal space with four decision regions
7
1
, ..., 7
4
corresponding to four constellations s
1
, ..., s
4
.
The received vector r lies in region 7
1
, so the receiver will
output the message :
1
as the best message estimate given
received vector r.
φ
(t)
1
φ
(t)
2
s
s
s
s
1
2
3
4
1
Z Z
3
Z
2
Z
4
x
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 39
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision criterion
Using Bayes rule, one can have
j (s
i
∣r) =
j (r∣s
i
) j (s
i
)
j (r)
. (28)
To minimize error probability, the receiver output ˆ : = :
i
corresponds to the constellation point s
i
that maximizes
j (s
i
∣r), i.e., the detected transmitted constellation point ˆs
can be determined by
ˆs = arg max
s
j (r∣s
i
) j (s
i
)
j (r)
= arg max
s
j (r∣s
i
) j (s
i
) , i = 1, ..., `
(29)
where the second equality follows from the fact that j(r) is
not a function of s
i
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 40
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision criterion (cont.)
Assuming that transmitted messages :
i
are equally likely
(i.e., j(s
i
) = 1,`), (29) becomes
ˆs = arg max
s
j (r∣s
i
) , i = 1, ..., `. (30)
Let deﬁne the likelihood function associated with the receiver
as
)(s
i
) = j (r∣s
i
) . (31)
Given a received vector r, a maximum likelihood (ML)
receiver outputs ˆ : = :
i
corresponding to the constellation
point s
i
that maximizes )(s
i
).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 41
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision criterion (cont.)
Since the log function is increasing in its argument,
maximizing s
i
is equivalent to maximizing the log likelihood
function, deﬁned as )
A1
(s
i
) = log )(s
i
). Using (24) for
)
A1
(s
i
) = log )(s
i
) yields
)
1J
(s
.
) = −
1
·
0
Þ
∑
¸=1
(:
¸
−:
.,¸
)
2
+constant = −
1
·
0
∥r −s
.
∥
2
+constant.
(32)
Based on (30), the detected transmitted constellation point ˆs can
be determined by the ML criterion as
ˆs = arg min
s
∥r −s
.
∥
2
, i = 1, ..., `. (33)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 42
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General results
Proofs of suﬃcient statistics for optimal detection
Decision regions and criterion
Decision criterion (cont.)
Under the aforementioned assumption of equiprobable
(transmitted) messages :
i
, the ML structure minimizes the
probability of detection error at the receiver.
Under the aforementioned assumption of equiprobable
(transmitted) messages :
i
, the ML structure minimizes the
probability of detection error at the receiver.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 43
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
Error probability of ML detection
With j (:
i
sent) = 1,`, the error probability of the ML receiver:
s
1
s
s
s
s
s
s
s
2
3
4
5
6
7
8
0
1
Z
1
d
min
r=s +n
1
c
=
1
∑
.=1
j (r , ∈ 7
.
∣:
.
sent) j (:
.
sent)
=
1
`
1
∑
.=1
j (r , ∈ 7
.
∣:
.
sent)
= 1 −
1
`
1
∑
.=1
j (r ∈ 7
.
∣:
.
sent)
= 1 −
1
`
1
∑
.=1
∫
2
j (r∣:
.
) dr
= 1 −
1
`
1
∑
.=1
∫
2
j (r = s
.
+ n∣s
.
) dr
1
1
∑
∫
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 44
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability
As observed, (34) gives an exact solution to the error
probability but it is impossible to solve this error probability in
closedform. Therefore, the union bound on error probability
is investigated.
Let ¹
i,I
denote the event that ∥r −s
I
∥ < ∥r −s
i
∥ given that
the constellation point s
i
was sent.
If the event ¹
i,I
occurs, then the constellation point will be
decoded in error since the transmitted constellation point s
i
is
not the closest constellation point to the received vector r.
However, event ¹
i,I
does not necessarily imply that s
I
will be
decoded instead of s
i
, since there may be another
constellation point s

with ∥r −s

∥ < ∥r −s
I
∥ < ∥r −s
i
∥.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 45
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability (cont.)
The constellation point is decoded correctly if
∥r −s
i
∥ < ∥r −s
I
∥ ∀/ ∕= i. Therefore,
1
c
(:
i
sent) = j
⎛
⎜
⎜
⎝
A
∪
I=1
I∕=i
¹
i,I
⎞
⎟
⎟
⎠
≤
A
∑
I=1
I∕=i
j (¹
i,I
) . (35)
where the inequality follows from the union bound on
probability.
More speciﬁcally, j (¹
i,I
) can be determined by
j (¹
.,
) = j (∥s

−r∥ < ∥s
.
−r∥ ∣:
.
sent)
= j (∥s

−(s
.
+ n)∥ < ∥s
.
−(s
.
+ n)∥)
= j (∥n + s
.
−s

∥ < ∥n∥)
= j
(
2 ∥n∥ ∥s
.
−s

∥ cos c < −∥s
.
−s

∥
2
)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 46
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability (cont.)
Since ⟨n, s
i
−s
I
⟩ = ∥n∥ ∥s
i
−s
I
∥ cos c, one will have
n = ∥n∥ cos c =
⟨n,s
−s
⟩
∥s
−s
∥
is a Gaussian random variable with
zeromean and variance ·
0
,2
As a result, j (¹
i,I
) can be simpliﬁed to
j (¹
i,I
) = j
(
n < −
∥s
i
−s
I
∥
2
)
= j
(
n
d
i,I
2
)
=
∫
∞
o
,
¸2
1
√
¬·
0
exp
(
−n
2
·
0
)
dn = Q
(
d
i,I
√
2·
0
)
.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 47
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability (cont.)
Substituting (36) into (35), one can have
1
c
(:
i
sent) ≤
A
∑
I=1
I∕=i
Q
(
d
i,I
√
2·
0
)
. (36)
where the Q function, Q(.), is deﬁned as the probability that
a Gaussian random variable r with zeromean and variance of
1 is bigger than ., i.e.,
Q(.) = j (r .) =
∫
∞
:
1
√
2¬
exp
(
−r
2
2
)
dr. (37)
Summing (36) over all possible messages yields the union bound
1
c
=
1
∑
.=1
j (:
.
) 1
c
(:
.
sent) ≤
1
`
1
∑
.=1
1
∑
=1
∕=.
Q
(
d
.,
√
2·
0
)
. (38)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 48
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability (cont.)
Note that the Q function cannot be solved in closedform. It
can be obtained from the complementary error function as
Q(.) =
1
2
erfc
(
:
√
2
)
.
One can upper bound Q(.) with the closedform expression
Q(.) ≤
1
.
√
2¬
exp
(
−.
2
,2
)
. (39)
and this bound is quite tight for . ≫ 0.
Let deﬁne the minimum distance of the constellation as
d
nia
= min
i,I
d
i,I
, one can simplify (41) with looser bound
1
c
≤ (` −1)Q
(
d
nia
√
2·
0
)
. (40)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 49
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
The union bound on error probability (cont.)
Using (39) for the Q function yields a closedform bound
1
c
≤
` −1
√
¬
exp
(
−d
2
nia
4·
0
)
. (41)
Note that for binary modulation (` = 2), there is only one
way to make an error and d
nia
is the distance between the
two signal constellation points, so the bound is exact
1
o
= Q
(
d
nia
√
2·
0
)
. (42)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 50
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
Bit error rate
Recall that 1
c
is the probability of a symbol (message) error:
1
c
= j ( ˆ : ∕= :
i
∣:
i
sent) where :
i
corresponds to a message
with op
2
` bits.
However, system designers are typically more interested in the
bit error probability, also called the bit error rate (BER), than
in the symbol error probability, since bit errors drive the
performance of higher layer networking protocols and
endtoend performance.
Thus, we would like to design the mapping of ` possible bit
sequences to a message :
i
, i = 1, ..., ` so that a symbol
error associated with an adjacent decision region, which is the
most likely way to make an error, corresponds to only one bit
error.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 51
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
Error probability
The union bound on error probability
Bit error rate (cont.)
With such a mapping, assuming that mistaking a signal
constellation for a constellation other than its nearest
neighbors has a very low probability, we can make the
approximation
1
o
≈
1
c
op
2
`
. (43)
The most common form of mapping with the property is
called Gray coding, which will be discussed in more detail.
Signal space concepts are applicable to any modulation where
bits are encoded as one of several possible analog signals,
including the amplitude, phase, and frequency modulations as
discussed later.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 52
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
General principles
The basic principle of passband digital modulation is to
encode an information bit stream into a carrier signal which is
then transmitted over a communications channel.
Demodulation is the process of extracting this information bit
stream from the received signal. Corruption of the transmitted
signal by the channel can lead to bit errors in the
demodulation process.
The goal of modulation is to send bits at a high data rate
while minimizing the probability of data corruption.
In general, modulated carrier signals encode information in the
amplitude c(t), frequency )(t), or phase 0(t) of a carrier
signal.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 53
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
General principles (cont.)
Thus, the modulated signal can be represented as
:(t) = c(t) cos [2¬ ()
t
+)(t)) +0(t) +c
0
] = c(t) cos [2¬)
t
t +c(t) +c
0
]
where c(t) = 2¬)(t)t +0(t) and c
0
is the phase oﬀset of the
carrier. This representation combines frequency and phase
modulation into angle modulation.
One can rewrite the righthand side of (44) in terms of its inphase
and quadrature components as:
:(t) = c(t) cos c(t) cos [2¬)
t
t] −c(t) sin c(t) sin [2¬)
t
t]
= :
J
(t) cos [2¬)
t
t] −:
Q
(t) sin [2¬)
t
t] (44)
where :
J
(t) = c(t) cos c(t) is called the inphase component of :(t)
and :
Q
(t) = c(t) sin c(t) is called its quadrature component.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 54
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
General principles (cont.)
We can write :(t) in its complex baseband representation as
:(t) = Re
[
n(t)c
)2¬)

]
(45)
where n(t) = :
1
(t) +,:
Q
(t).
This representation is useful since receivers typically process
the inphase and quadrature signal components separately.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 55
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation
In amplitude and phase modulation the information bit stream
is encoded in the amplitude and/or phase of the transmitted
signal.
Speciﬁcally, over a time interval of T

, 1 = log
2
` bits are
encoded into the amplitude and/or phase of the transmitted
signal :(t), 0 ≤ t < T

.
The transmitted signal over this period
:(t) = :
1
(t) cos [2¬)
c
t] −:
Q
(t) sin [2¬)
c
t] can be written in
terms of its signal space representation as
:(t) = :
i,1
c
1
(t) +:
i,2
c
2
(t) (46)
where basis functions c
1
(t) = p(t) cos (2¬)
c
t +c
0
) and
c
2
(t) = −p(t) sin (2¬)
c
t +c
0
), where p(t) is a shaping pulse.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 56
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation (cont.)
To send the ith message over the time interval [/T, (/ + 1)T),
we set :
1
(t) = :
i,1
p(t) and :
Q
(t) = :
i,2
p(t). These inphase
and quadrature signal components are baseband signals with
spectral characteristics determined by the pulse shape p(t).
In particular, their bandwidth 1 equals the bandwidth of g(t),
and the transmitted signal :(t) is a passband signal with
center frequency fc and passband bandwidth 21.
In practice we take 1 = 1
j
,T

where 1
j
depends on the
pulse shape: for rectangular pulses 1
j
= .5 and for raised
cosine pulses .5 ≤ 1
j
≤ 1.
Thus, for rectangular pulses the bandwidth of p(t) is .5,T

and the bandwidth of :(t) is 1,T

.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 57
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation (cont.)
Since the pulse shape p(t) is ﬁxed, the signal constellation for
amplitude and phase modulation is deﬁned based on the
constellation point: (:
i,1
, :
i,2
) ∈ ℝ
2
, i = 1, ..., `.
The complex baseband representation of :(t) is
:(t) = Re
[
r(t)c
)c
0
c
)2¬)

]
(47)
where:
r(t) = :
J
(t) +,:
Q
(t) = (:
.,1
+,:
.,2
) p(t).
The constellation point s
.
= (:
.,1
, :
.,2
) is called the symbol
associated with the log
2
` bits and
T
s
is called the symbol time and the bit rate for this
modulation is 1 bits per symbol or 1 = log
2
`,T
s
bits per
second.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 58
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation (cont.)
There are three main types of amplitude/phase modulation:
Pulse Amplitude Modulation (MPAM): information encoded in
amplitude only.
Phase Shift Keying (MPSK): information encoded in phase
only.
Quadrature Amplitude Modulation (MQAM): information
encoded in both amplitude and phase.
The number of bits per symbol 1 = log
2
`, signal
constellation (:
i,1
, :
i,2
) ∈ ℝ
2
, i = 1, ..., `, and the choice of
shaping pulse p(t) determines the digital modulation design.
The pulse shape p(t) is designed to improve spectral eﬃciency
and combat intersymbolinterference (ISI).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 59
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation (cont.)
Amplitude and phase modulation over a given symbol period
can be generated using the modulator structure shown in the
next Figure.
Note that the basis functions in this ﬁgure have an arbitrary
phase c
0
associated with the transmit oscillator.
Demodulation over each symbol period is performed using the
demodulation structure of Figure xx +1, which is equivalent
to the structure of for c
1
(t) = p(t) cos (2¬)
c
t +c) and
c
2
(t) = −p(t) sin (2¬)
c
t +c).
Typically the receiver includes some additional circuitry for
carrier phase recovery that matches the carrier phase c at the
receiver to the carrier phase c
0
at the transmitter, which is
called coherent detection.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 60
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Amplitude and phase modulation (cont.)
If c −c
0
= △c ∕= 0 then the inphase branch will have an
unwanted term associated with the quadrature branch and
vice versa, i.e., :
1
= :
i,1
cos (△c) +:
i,2
sin (△c) +n
1
and
:
2
= :
i,1
sin(△c) +:
i,2
cos (△c) +n
2
can result in signiﬁcant
performance degradation.
The receiver structure also assumes that the sampling function
every T

seconds is synchronized to the start of the symbol
period, which is called synchronization or timing recovery.
Receiver synchronization and carrier phase recovery are
complex receiver operations that can be highly challenging in
wireless environments.
We will assume perfect carrier recovery in our discussion of
MPAM, MPSK and MQAM, and therefore set c = c
0
= 0 for
their analysis.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 61
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (MPAM)
We will start by looking at the simplest form of linear
modulation, onedimensional MPAM, which has no
quadrature component (:
i,2
= 0).
For MPAM all of the information is encoded into the signal
amplitude ¹
i
. The transmitted signal over one symbol time is
given by
:
i
(t) = Re
[
¹
i
p(t)c
)2¬)

]
= ¹
i
p(t) cos(2¬)
c
t), 0 ≤ t ≤ T

≫ 1,)
c
,
(48)
where ¹
i
= (2i −1 −`)d, i = 1, 2, ..., ` deﬁnes the signal
constellation, parameterized by the distance d which is
typically a function of the signal energy, and p(t) is the
shaping pulse.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 62
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
π
2
Shaping
Filter
g(t)
Shaping
Filter
g(t)
s(t)
In−Phase branch
Quadrature Branch
i1
i2
s
s
i1
s g(t)
s g(t)
i2
c
−sin(2 f t+ )
cos(2 f t+ )
c
π φ
0
cos(2 f t+ )
c
π φ
0
π φ
0
Figure 2: Amplitude and phase modulator.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 63
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
The minimum distance between constellation points is
d
nia
= :in
i,)
∣ ¹
i
−¹
)
∣= 2d. The amplitude of the
transmitted signal takes on M diﬀerent values, which implies
that each pulse conveys op
2
` = 1 bits per symbol time T

.
Over each symbol period the MPAM signal associated with
the ith constellation has energy
1

=
∫
T
0
:
2
i
(t)dt =
∫
T
0
¹
2
i
p
2
(t) cos
2
(2¬)
c
t)dt = ¹
2
i
. (49)
It is noted that the energy is not the same for each signal
:
i
(t), i = 1, ..., `.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 64
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
Assuming equally likely symbols, the average energy is
1

=
1
`
A
∑
i=1
¹
2
i
. (50)
i
1
^
m=m
Find i: x Z
i
i
T
T
In−Phase branch
π/2
g(T−t)
g(T−t)
cos (2 f t+ ) φ
i1 1
2 i2 2
Quadrature branch
s
s
π
c
−sin (2 f t+ )
c
π φ
r(t)=s (t)+n(t)
r =s +n
r =s +n
Figure 3: Amplitude and phase demodulator.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 65
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
The constellation mapping is usually done by Gray encoding,
where the messages associated with signal amplitudes that are
adjacent to each other diﬀer by one bit value, as illustrated in
the below ﬁgure.
With this encoding method, if noise causes the demodulation
process to mistake one symbol for an adjacent one (the most
likely type of error), this results in only a single bit error in the
sequence of K bits. Gray codes can be designed for MPSK and
square MQAM constellations, but not rectangular MQAM.
M=4, K=2
00 01 11 10
M=8, K=3
000 001 011 010 110 111 101 100
2d
2d
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 66
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
An example: For p(t) =
√
2,T

, 0 ≤ t < T

a rectangular
pulse shape, ﬁnd the average energy of 4PAM modulation.
Solution: For 4PAM, the ¹
i
values are ¹
i
= {3d, −d, d, 3d}.
Hence, the average is
1

=
d
2
4
(9 + 1 + 1 + 9) = 5d
2
. (51)
The decision regions 7
i
, i = 1, ..., ` associated with the
pulse amplitude ¹
i
= (2i −1 −`)d for ` = 4 and ` = 8
as shown in the next ﬁgure. Mathematically, for any `, these
decision regions are deﬁned by
7
i
=
⎧
⎨
⎩
(−∞, ¹
i
+d) i = 1,
[¹
i
−d, ¹
i
+d) 2 ≤ i ≤ ` −1
[¹
i
−d, ∞) i = `
(52)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 67
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
One can see that MPAM has only a single basis function
c
1
(t) = p(t) cos(2¬)
c
t).
Thus, the coherent demodulator for MPAM reduces to the
demodulator shown in the next ﬁgure, where the
multithreshold device maps r to a decision region 7
i
and
outputs the corresponding bit sequence
ˆ : = :
i
= {/
1
, ..., /
1
}.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 68
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Pulse amplitude modulation (cont.)
2d
2d
A
1
A A A
2 3 4
Z
1
Z Z
2
Z
3 4
A A A A
Z Z Z Z Z
1
Z Z
A
1
A
2 3
Z
2 3 4 5 6 7 8
A A
8 7 6 5 4
Figure 5: Decision Regions for MPAM.
i
s (t)+n(t)
s
x
g (T −t) 0
X
cos(2 f t)
c
π
^
s
T
Multithreshold Device
2d
4d
−2d
−4d
−(M−2)d
(M−2)d
m=m =b b ...b
i 1 2 K
}
Z
i
Figure 6: Coherent Demodulator for MPAM.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 69
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying (MPSK)
For MPSK modulation, all of the information is encoded in
the phase of the transmitted signal.
Thus, the transmitted signal over one symbol time is given by
:
i
(t) = Re
[
¹p(t)c
)2¬(i−1)¸A
c
)2¬)

]
, 0 ≤ t ≤ T

(53)
= ¹p(t) cos
[
2¬)
c
t +
2¬(i −1)
`
]
= ¹p(t) cos
[
2¬(i −1)
`
]
cos(2¬)
c
t)
− ¹p(t) sin
[
2¬(i −1)
`
]
sin(2¬)
c
t).
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 70
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying (cont.)
The constellation points or symbols (:
i,1
, :
i,2
) are given by
:
i,1
= ¹cos
[
2¬(i−1)
A
]
and :
i,2
= ¹sin
[
2¬(i−1)
A
]
for
i = 1, ..., `. 0
i
=
2¬(i−1)
A
, i = 1, 2, ..., ` = 2
1
are the
diﬀerent phases in the signal constellation points that convey
the information bits.
The minimum distance between constellation points is
d
nia
= 2¹sin(¬,`), where ¹ is typically a function of the
signal energy.
2PSK is often referred to as binary PSK or BPSK, while 4PSK
is often called quadrature phase shift keying (QPSK), and is
the same as MQAM with ` = 4 which is deﬁned below.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 71
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying (cont.)
All possible transmitted signals :
i
(t) have equal energy:
1

=
∫
T
0
:
2
i
(t)dt = ¹
2
. (54)
Note that for p(t) =
√
2,T

, 0 ≤ t ≤ T

, i.e., a rectangular
pulse, this signal has constant envelope, unlike the other
amplitude modulation techniques MPAM and MQAM.
However, rectangular pulses are spectrallyineﬃcient, and
more eﬃcient pulse shapes make MPSK nonconstant
envelope.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 72
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying (cont.)
Analogous to MPAM, MPSK constellation mapping is usually
done by Gray encoding, where the messages associated with
signal phases that are adjacent to each other diﬀer by one bit
value, as illustrated in the below Figure.
With this encoding method, mistaking a symbol for an
adjacent one causes only a single bit error.
M=4, K=2
00 11
01
10
M=8, K=3
000
001
011
110
100
010
110
101
s
i1
s
i2
s
i1
s
i2
Figure 7: Gray Encoding for MPSK.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 73
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying: Decision region
The decision regions 7
i
, i = 1, ..., `, associated with MPSK
for ` = 8 are shown in the next ﬁgure.
If we represent r = :c
)0
∈ ℝ
2
in polar coordinates then these
decision regions for any ` are deﬁned by
7
i
=
{
:c
)0
: 2¬(i −.5),` ≤ 0 ≤ 2¬(i +.5),`
}
. (55)
For the special case of BPSK, the decision regions simplify to
7
1
= (r : r 0) and 7
2
= (r : r ≤ 0).
Moreover BPSK has only a single basis function
c
1
(t) = p(t) cos(2¬)
c
t) and, since there is only a single bit
transmitted per symbol time T

, the bit duration T
o
= T

.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 74
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying: Decision region (cont.)
Thus, the coherent demodulator for BPSK reduces to the
demodulator shown in the next ﬁgure, where the threshold
device maps r to the positive or negative half of the real line,
and outputs the corresponding bit value.
We have assumed in this ﬁgure that the message
corresponding to a bit value of 1, :
1
= 1, is mapped to
constellation point :
1
= ¹ and the message corresponding to
a bit value of 0, :
2
= 0, is mapped to the constellation point
:
2
= −¹.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 75
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Phase shift keying: Decision region (cont.)
Z
1
Z
2
Z
3
Z
4
Z
1
Z
2
Z
Z
Z
Z
Z
Z
3
4
5
6
7
8
Figure 8: Decision Regions for MPSK.
i
s (t)+n(t)
g (T −t)
b 0 X
cos(2 f t)
c
π
^
T
b
Threshold Device
2
m=1 or 0
m=1
^
^
m=0
}
}
r
1
Z :r>0
Z :r<0
Figure 9: Coherent Demodulator for BPSK.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 76
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Quadrature amplitude modulation (MQAM)
For MQAM, the information bits are encoded in both the
amplitude and phase of the transmitted signal.
Thus, whereas both MPAM and MPSK have one degree of
freedom in which to encode the information bits (amplitude or
phase), MQAM has two degrees of freedom.
As a result, MQAM is more spectrallyeﬃcient than MPAM
and MPSK, in that it can encode the most number of bits per
symbol for a given average energy.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 77
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Quadrature amplitude modulation (cont.)
The transmitted signal is given by
:
i
(t) = Re
[
¹
i
c
)0
p(t)c
)2¬)

]
(56)
= ¹
i
cos(0
i
)p(t) cos(2¬)
c
t) −¹
i
sin(0
i
)p(t) sin(2¬)
c
t).
where 0 ≤ t ≤ T

.
The energy in :
i
(t) is
1

=
∫
T
0
:
2
i
(t)dt = ¹
2
i
. (57)
that is the same as for MPAM.
The distance between any pair of symbols in the signal
constellation is
d
i,I
= ∥s
i
−s
I
∥ =
√
(:
i,1
−:
I,1
)
2
+ (:
i,2
−:
I,2
)
2
. (58)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 78
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Quadrature amplitude modulation (cont.)
For square signal constellations, where :
i,1
and :
i,2
take
values on (2i −1 −1)d, i = 1, 2, ..., 1 = 2

, the minimum
distance between signal points reduces to d
nia
= 2d, the
same as for MPAM.
In fact, MQAM with square constellations of size 1
2
is
equivalent to MPAM modulation with constellations of size 1
on each of the inphase and quadrature signal components.
Common square constellations are 4QAM and 16QAM, which
are shown in the below ﬁgure.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 79
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Quadrature amplitude modulation: constellation and
decision regions
4−QAM
16−QAM
Figure 10: 4QAM and 16QAM
Constellations.
Z
1
Z
2
Z
3
Z
4
Z
5
Z
6
Z
7
Z
8
Z
9
Z
10
Z
11
Z
12
Z
13
Z
14
Z
15
Z
16
Figure 11: Decision Regions for
MQAM withM = 16.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 80
Digital modulation techniques
Signal Space Analysis
Receiver Structure and Suﬃcient Statistics
Error Probability Analysis and the Union Bound
Passband modulation
General principles
Amplitude and phase modulation
Pulse amplitude modulation (MPAM)
Phase shift keying (MPSK)
Quadrature amplitude modulation (MQAM)
Quadrature amplitude modulation: constellation and
decision regions
These square constellations have ` = 2
2
= 1
2
constellation
points, which are used to send 2 bits/symbol, or  bits per
dimension.
It can be shown that the average power of a square signal
constellation with  bits per dimension, o

, is proportional to
4

,3, and it follows that the average power for one more bit
per dimension o
+1
≈ 4o

.
Thus, for square constellations it takes approximately 6 dB
more power to send an additional 1 bit/dimension or 2
bits/symbol while maintaining the same minimum distance
between constellation points.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.1: Digital modulations 81
Introduction
AWGN channels
Fading Channels
Chapter 3: Physicallayer transmission techniques
Section 3.2: Performance analysis over fading channels
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 1
Introduction
AWGN channels
Fading Channels
Outline of the lecture notes
1
Introduction
2
AWGN channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
3
Fading Channels
Introduction
Outage probability
Average probability of error
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 2
Introduction
AWGN channels
Fading Channels
Introduction
We now consider the performance of the digital modulation
techniques discussed in the previous chapter when used over AWGN
channels and channels with ﬂatfading.
There are two performance criteria of interest: the probability of
error, deﬁned relative to either symbol or bit errors, and the outage
probability, deﬁned as the probability that the instantaneous
signaltonoise ratio falls below a given threshold.
Wireless channels may also exhibit frequency selective fading and
Doppler shift. Frequencyselective fading gives rise to intersymbol
interference (ISI), which causes an irreducible error ﬂoor in the
received signal.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 3
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
SignaltoNoise power ratio and bit/symbol energy
In this section we deﬁne the signaltonoise power ratio (SNR) and
its relation to energyperbit (1
b
) and energypersymbol (1
s
).
We then examine the error probability on AWGN channels for
diﬀerent modulation techniques as parameterized by these energy
metrics. Our analysis uses the signal space concepts of previous
section.
In an AWGN channel, the modulated signal :(t) = Re
[
n(t)c
¸2t}

]
has receiver noise n(t) added to it prior to reception. The noise n(t)
is a white Gaussian random process with zeromean and power
spectral density ·
0
,2.
The received signal is thus :(t) = :(t) +n(t).
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 4
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
SignaltoNoise power ratio and bit/symbol energy (cont.)
We deﬁne the received signaltonoise power ratio (SNR) as the
ratio of the received signal power 1
:
to the power of the noise
within the bandwidth of the transmitted signal :(t).
The received power 1
:
is determined by the transmitted power and
the path loss and multipath fading.
The noise power is determined by the bandwidth of the transmitted
signal and the spectral properties of n(t). Speciﬁcally, if the
bandwidth of the complex envelope n(t) of :(t) is 1 then the
bandwidth of the transmitted signal :(t) is 21.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 5
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
SignaltoNoise ratio and bit/symbol energy (cont.)
Since the noise n(t) has uniform power spectral density ·
0
,2, the
total noise power within the bandwidth 21 is
1
n
= ·
0
,2 ×21 = ·
0
1. So, the received SNR is given by
o·1 =
1
:
·
0
1
. (1)
In systems with interference, we often use the received
signaltointerferenceplusnoise power ratio (SINR) in place of SNR
for calculating error probability. If the interference statistics
approximate those of Gaussian noise then this is a reasonable
approximation.
The received SINR is given by
o·1 =
1
:
·
0
1 +1
J
. (2)
where 1
J
is the average power of the interference.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 6
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
SignaltoNoise ratio and bit/symbol energy (cont.)
The SNR is often expressed in terms of the signal energy per bit 1
b
or per symbol 1
s
as
o·1 =
1
:
·
0
1
=
1
s
·
0
1T
s
=
1
b
·
0
1T
b
. (3)
where T
s
and T
b
are the symbol and bit durations, respectively. For
binary modulation (e.g., BPSK), T
s
= T
b
and 1
s
= 1
b
.
For data shaping pulses with T
s
= 1,1 (e.g., raised cosine pulses
with a = 1), one will have SNR = 1
s
,·
0
for multilevel signaling
and SNR = 1
b
,·
0
for binary signaling. For general pulses,
T
s
= /,1 for some constant /, we have / ×SNR = 1
s
,·
0
.
The quantities ¸
s
= 1
s
,·
0
and ¸
b
= 1
b
,·
0
are sometimes called
the SNR per symbol and the SNR per bit, respectively.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 7
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
SignaltoNoise ratio and bit/symbol energy (cont.)
For performance speciﬁcation, we are interested in the bit error
probability 1
b
as a function of ¸
b
.
However, for Marray signaling (e.g., MPAM and MPSK), the bit
error probability depends on both the symbol error probability and
the mapping of bits to symbols. Thus, we typically compute the
symbol error probability 1
s
as a function of ¸
s
based on the signal
space concepts of previous section and then obtain 1
b
as a function
of ¸
b
using an exact or approximate conversion.
The approximate conversion typically assumes that the symbol
energy is divided equally among all bits, and that Gray encoding is
used so that at reasonable SNRs, one symbol error corresponds to
exactly one bit error.
These assumptions for Marray signaling lead to the approximations:
¸
b
≈
¸
s
log
2
`
and 1
b
≈
1
s
log
2
`
. (4)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 8
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
Error probability for BPSK and QPSK
Consider BPSK modulation with coherent detection and perfect
recovery of the carrier frequency and phase. With binary modulation
each symbol corresponds to one bit, so the symbol and bit error rates
are the same. The transmitted signal is :
1
(t) = ¹p(t)co:(2¬)
t
t) to
send a 0 bit and :
2
(t) = −¹p(t)co:(2¬)
t
t) to send a 1 bit. Note
that for binary modulation where ` = 2, there is only one way to
make an error and d
n.n
is the distance between the two signal
constellation points, so the probability of error is also the bound:
1
b
= Q
(
d
n.n
√
2·
0
)
. (5)
In previous chapter, we have d
n.n
=∥ s
1
−s
2
∥=∥ ¹−(−¹) ∥= 2¹.
The energyperbit can be determined by
1
b
=
∫
T
0
:
2
1
(t)dt =
∫
T
0
:
2
2
(t)dt =
∫
T
0
¹
2
p
2
(t) cos
2
(2¬)
t
t)dt = ¹
2
.
(6)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 9
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
Error probability for BPSK and QPSK (cont.)
Thus, the signal constellation for BPSK in terms of energyperbit is
given by s
0
=
√
1
b
and s
1
= −
√
1
b
. This yields the minimum
distance d
n.n
= 2¹ = 2
√
1
b
. Substituting this into (5) yields
1
b
= Q
(
2
√
1
b
√
2·
0
)
= Q
(
√
21
b
·
0
)
= Q
(
√
2¸
b
)
. (7)
QPSK modulation consists of BPSK modulation on both the
inphase and quadrature components of the signal. With perfect
phase and carrier recovery, the received signal components
corresponding to each of these branches are orthogonal. Therefore,
the bit error probability on each branch is the same as for
BPSK:1
b
= Q
(
√
2¸
b
)
.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 10
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
Error probability for BPSK and QPSK (cont.)
The symbol error probability equals the probability that either
branch has a bit error:
1
s
= 1 −
[
1 −Q
(
√
2¸
b
)]
2
. (8)
Example: Find the bit error probability 1
b
and symbol error
probability 1
s
of QPSK assuming ¸
b
= 7 dB. Solution: We have
¸
b
= 10
7¸10
= 5.012, then 1
b
= Q
(
√
2¸
b
)
= 7.726 ×10
−4
and
1
s
= 1 −
[
1 −Q
(
√
2¸
b
)]
2
= 1.545 ×10
−3
.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 11
Introduction
AWGN channels
Fading Channels
SignaltoNoise power ratio and bit/symbol energy
Error probability for BPSK and QPSK
Approximate symbol and bit error probabilities for typical modulations
Approximate symbol and bit error probabilities for typical
modulations
Many of the approximations or exact values for 1
s
derived above for
coherent modulation are in the following form:
1
s
(¸
s
) ≈ c
1
Q
(
√
a
1
¸
s
)
(9)
where c
1
and a
1
depend on the type of approximation and the
modulation type. In the below table, we summarize the speciﬁc values of
c
1
and a
1
for common 1
s
expressions for PSK, QAM, and FSK
modulations based on the derivations in the prior sections.
Modulation P
s
(γ
s
) P
b
(γ
b
)
BFSK: P
b
= Q
√
γ
b
BPSK: P
b
= Q
√
2γ
b
QPSK,4QAM: P
s
≈ 2 Q
√
γ
s
P
b
≈ Q
√
2γ
b
MPAM: P
s
≈
2(M−1)
M
Q
6γ
s
M
2
−1
P
b
≈
2(M−1)
M log
2
M
Q
6γ
b
log
2
M
(M
2
−1)
MPSK: P
s
≈ 2Q
√
2γ
s
sin(π/M)
P
b
≈
2
log
2
M
Q
2γ
b
log
2
M sin(π/M)
Rectangular MQAM: P
s
≈
4(
√
M−1)
√
M
Q
3γ
s
M−1
P
b
≈
4(
√
M−1)
√
M log
2
M
Q
3γ
b
log
2
M
(M−1)
Nonrectangular MQAM: P
s
≈ 4Q
3γ
s
M−1
P
b
≈
4
log
2
M
Q
3γ
b
log
2
M
(M−1)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 12
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Introduction
In AWGN the probability of symbol error depends on the received
SNR ¸
s
. In a fading channel, the received signal power varies
randomly over distance or time due to shadowing and/or multipath
fading. Thus, in fading ¸
s
is a random variables with distribution
j
;
(¸), and therefore 1
s
(¸
s
) is also random.
The performance metric when ¸
s
is random depends on the rate of
change of the fading. There are three diﬀerent performance criteria
that can be used to characterize the random variable 1
s
:
The outage probability,
, deﬁned as the probability that
falls
below a given value corresponding to the maximum allowable
.
The average error probability,
, averaged over the distribution of
.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 13
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Introduction (cont.)
If the power of the received signal (with fading) is changing slowly
(slowfading), then a deep fade will aﬀect many simultaneous
symbols. Thus, fading may lead to large error bursts, which cannot
be corrected for with coding of reasonable complexity. Therefore,
these error bursts can seriously degrade endtoend performance.
In this case acceptable performance cannot be guaranteed over all
time or, equivalently, throughout a cell, without drastically
increasing transmit power. Under these circumstances, an outage
probability is speciﬁed so that the channel is deemed unusable for
some fraction of time or space.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 14
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Outage probability
The outage probability relative to 1
ou
is deﬁned as
1
ou
= j (¸
s
< ¸
0
) =
∫
;
0
0
j
;
(¸)d¸. (10)
where ¸
0
typically speciﬁes the minimum SNR required for
acceptable performance. For example, if we consider digitized voice,
1
b
= 10
−3
is an acceptable error rate since it generally cannot be
detected by the human ear. Thus, for a BPSK signal in Rayleigh
fading, ¸
b
< 7 dB would be declared an outage, so we set ¸
0
= 7 dB.
In Rayleigh fading with j
;
(¸) =
1
;
c
−;
¸;
, one will have
1
ou
=
∫
;
0
0
1
¸
s
c
−;
¸;
d¸ = 1 −c
−;
0
¸;
. (11)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 15
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Outage probability (cont.)
Example: Determine the required ¸
b
for BPSK modulation in slow
Rayleigh fading such as 50% of the time (or in space),
1
b
(¸
b
) < 10
−4
.
Solution: For BPSK modulation in AWGN, the target BER is
obtained at 8.5 dB (i.e., for 1
b
(¸
b
) = Q(
√
2¸
b
), one have
1
b
(
10
0.85
)
= 10
−4
). Thus, ¸
0
= 8.5 dB, since we want
1
ou
= j(¸
b
< ¸
0
), we have
¸
b
=
¸
0
−ln(1 −1
ou
)
=
10
.85
−ln(1 −.05)
= 21.4 dB. (12)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 16
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error
The average probability of error is used as a performance metric
when ¸
s
is roughly constant over a symbol time. Then the averaged
probability of error is computed by integrating the error probability
in AWGN over the fading distribution:
1
s
=
∫
∞
0
1
s
(¸)j
;
(¸)d¸. (13)
where 1
s
(¸) is the probability of symbol error in AWGN channels
with SNR ¸, which can be approximated by the expressions in the
aforementioned table.
For a given distribution of the fading amplitude : (i.e., Rayleigh,
Rician, lognormal, etc.), we compute j
;
(¸) by making the change
of variable
j
;
(¸)d¸ = j(:)d:. (14)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 17
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error (cont.)
For instance, in Rayleigh fading (refer to Zheng’s paper on Modiﬁed
Jake channel model), the received signal amplitude r has the
Rayleigh distribution
j(:) =
:
o
2
c
−:
2
¸(2c
2
)
, : ≥ 0. (15)
The SNR per symbol for a given amplitude : is
¸ =
:
2
T
s
2o
2
n
. (16)
where o
2
n
= ·
0
,2 is the PSD of the noise in the inphase and
quadrature branches.
Diﬀerentiating both sides of this expression yields
d¸ =
:T
s
o
2
n
d:. (17)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 18
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error (cont.)
Substituting (16) and (17) into (15) and then (14) yields
j
;
(¸) =
o
2
n
o
2
T
s
c
−;c
2
¸(c
2
T
)
. (18)
Since the average SNR per symbol ¸
s
is just o
2
T
s
,o
2
n
, one can
rewrite (18) as
j
;
(¸) =
1
¸
s
c
;¸;
, (19)
which is exponential. For binary signaling, this reduces to
j
;
(¸) =
1
¸
b
c
;¸;
. (20)
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 19
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error (cont.)
Integrating the error probability of BPSK in AWGN over the
distribution (20) yields the following average probability of error for
BPSK in Rayleigh fading:
1
b
=
1
2
(
1 −
√
¸
b
1 +¸
b
)
≈
1
4¸
b
. (21)
where the approximation holds for large ¸
b
.
If we use the general approximation 1
s
≈ c
1
Q
(√
a
1
¸
s
)
then the
average probability of symbol error in Rayleigh fading can be
approximated as
1
s
≈
∫
∞
0
c
1
Q
(
√
a
1
¸
)
1
¸
s
c
−;¸;
d¸
s
=
c
n
2
(
1 −
√
.5a
1
¸
s
1 + .5a
1
¸
s
)
≈
c
1
2a
1
¸
s
.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 20
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error: Numerical results of BPSK
0 5 10 15 20
10
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10
0
γ
b
(dB)
P
b
AWGN
Rayleigh fading
Figure 1: Average probability of bit
error
for BPSK in Rayleigh Fading
and AWGN.
0 10 20 30 40
10
−9
10
−8
10
−7
10
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10
0
Average SNR γ
b
(dB)
A
v
e
r
a
g
e
B
i
t
E
r
r
o
r
P
r
o
b
a
b
i
l
i
t
y
m = 0.5
m = 1
(Rayleigh)
m = 1.5
m = 2
m = 2.5
m = 3
m = 4
m = 5
m = ∞
(No fading)
Figure 2: Average
for BPSK in
Nakagami Fading.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 21
Introduction
AWGN channels
Fading Channels
Introduction
Outage probability
Average probability of error
Average probability of error: Numerical results of MQAM
0 5 10 15 20 25 30 35 40
10
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10
0
γ
b
(dB)
P
b
M = 4
M = 16
M = 64
Rayleigh fading
AWGN
Figure 3: Average
for MQAM in Rayleigh Fading and AWGN.
Mobile communications  Chapter 3: Physicallayer transmission techniques Section 3.2: Performance analysis over fading channels 22
Introduction
Linear Block Codes
Chapter 3: Physicallayer transmission techniques
Section 3.4: Channel Coding
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 1
Introduction
Linear Block Codes
Purpose of using channel coding
Coding allows bit errors (introduced by transmission of a modulated
signal through a wireless channel) to be either (i) detected or (ii)
corrected by a decoder in the receiver..
Coding can be considered as the embedding of signal constellation
points in a higher dimensional signaling space than is needed for
communications.
By going to a higher dimensional space, the distance between points
can be increased, which provides for better error correction and
detection.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 2
Introduction
Linear Block Codes
Overview of code design
The main reason to apply error correction coding in a wireless
system is to reduce the probability of bit or block error.
The bit error probability 1
b
for a coded system is the probability
that a bit is decoded in error.
The block error probability 1
bl
, also called the packet error rate, is
the probability that one or bits in a block of coded bits is decoded in
error.
Block error probability is useful for packet data systems where bits
are encoded and transmitted in blocks.
The amount of error reduction provided by a given code is typically
characterized by its coding gain in AWGN and its diversity gain in
fading.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 3
Introduction
Linear Block Codes
BER performance of channel coding
Uncoded Coded
10
−6
10
10
−4
−2
C
g1
C
g2
P
b
SNR (dB)
4 5 6 7 8 9 10 11
10
−7
10
−6
10
−5
10
−4
10
−3
10
−2
Eb/No (dB)
D
e
c
o
d
e
d
B
E
R
Uncoded
Hamming (7,4) t=1
Hamming (15,11) t=1
Hamming (31,26) t=1
Extended Golay (24,12) t=3
BCH (127,36) t=15
BCH (127,64) t=10
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 4
Introduction
Linear Block Codes
BER performance of turbo coding
4 5 6 7 8 9 10 11
10
−7
10
−6
10
−5
10
−4
10
−3
10
−2
Eb/No (dB)
D
e
c
o
d
e
d
B
E
R
Uncoded
Hamming (7,4) t=1
Hamming (15,11) t=1
Hamming (31,26) t=1
Extended Golay (24,12) t=3
BCH (127,36) t=15
BCH (127,64) t=10
0.5 1 1.5 2
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
E
b
/N
o
in dB
B
E
R
1 iteration
2 iterations
3 iterations
6 iterations
10 iterations
18 iterations
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 5
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Introduction
Linear block codes are simple codes that are basically an extension
of singlebit parity check codes for error detection.
A singlebit parity check code is one of the most common forms of
detecting transmission errors.
This code uses one extra bit in a block of n data bits to indicate
whether the number of 1s in a block is odd or even.
Linear block codes extend this notion by using a larger number of
parity bits to either detect more than one error or correct for one or
more errors.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 6
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Binary linear block codes
A binary block code generates a block of n coded bits from /
information bits, called an (n, /) binary block code.
The coded bits are also called codeword symbols. The n codeword
symbols can take on 2
n
possible values corresponding to all possible
combinations of the n binary bits.
We select 2

codewords from these 2
n
possibilities to form the code,
such that each / bit information block is uniquely mapped to one of
these 2

codewords.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 7
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Binary linear block codes (cont.)
The rate of the code is 1
t
= /,n information bits per codeword
symbol.
If we assume that codeword symbols are transmitted across the
channel at a rate of 1
s
symbols/second, then the information rate
associated with an (n, /) block code is 1
b
= 1
t
1
s
=

n
1
s
bits/second. Thus, we see that block coding reduces the data rate
compared to what we obtain with uncoded modulation by the code
rate 1
t
.
A block code is called a linear code when the mapping of the /
information bits to the n codeword symbols is a linear mapping.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 8
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
The vector space of binary tuples
The set of all binary ntuples 1
n
is a vector space over the binary
ﬁeld, which consists of the two elements 0 and 1.
All ﬁelds have two operations, addition and multiplication: for the
binary ﬁeld these operations correspond to binary addition (modulo
2 addition) and standard multiplication.
A subset o of 1
n
is called a subspace if it satisﬁes the following
conditions:
The allzero vector is in
The set is closed under addition, such that if
∈ and
∈ ,
then
+
∈ .
An (n, /) block code is linear if the 2

lengthn codewords of the
code form a subspace of 1
n
. Thus, if C
.
and C
¸
are two codewords
in an (n, /) linear block code, then C
.
+C
¸
must form another
codeword of the code.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 9
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
The vector space of binary tuples: an example
Example: The vector space 1
3
consists of all binary tuples of length
3: 1
3
= {[000], [001], [010], [011], [100], [101], [110], [111]}.
Note that 1
3
is a subspace of itself, since it contains the all zero
vector and is closed under addition. Determine which of the
following subsets of B3 form a subspace:
1
= {[000], [001], [100], [101]}
2
= {[000], [100], [110], [111]}
3
= {[001], [100], [101]}
Solution: It is easily veriﬁed that ¹
1
is a subspace, since it contains
the allzero vector and the sum of any two tuples in ¹
1
is also in ¹
1
.
¹
2
is not a subspace since it is not closed under addition, as
110 + 111 = 001 ⊈ ¹
2
.
¹
3
is not a subspace since it is not closed under addition
(001 + 001 = 000 ⊈ ¹
3
) and it does not contain the all zero vector.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 10
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Hamming distance
Intuitively, the greater the distance between codewords in a given
code, the less chance that errors introduced by the channel will cause
a transmitted codeword to be decoded as a diﬀerent codeword.
We deﬁne the Hamming distance between two codewords C
.
and
C
¸
, denoted as d(C
.
, C
¸
) or d
.¸
, as the number of elements in
which they diﬀer:
d
.¸
=
n
∑
l=1
C
.
() +C
¸
(), (1)
where C
.
() denotes the th bit in C
.
.
For example, if C
.
= [00101] and C
¸
= [10011] then d
.¸
= 3.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 11
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
The weight of a given code
We deﬁne the weight of a given codeword C
.
as the number of 1s in
the codeword, so C
.
= [00101] has weight 2.
The weight of a given codeword C
.
is just its Hamming distance d
0.
with the all zero codeword C
0
= [00 . . . 0] or, equivalently, the sum
of its elements:
n(C
.
) =
n
∑
l=1
C
.
(). (2)
Since 0 + 0 = 1 + 1 = 0, the Hamming distance between C
.
and C
¸
is equal to the weight of C
.
+C
¸
.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 12
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
The weight of a given code (cont.)
Since the Hamming distance between any two codewords equals the
weight of their sum, we can determine the minimum distance
between all codewords in a code by just looking at the minimum
distance between all codewords and the all zero codeword.
Thus, we deﬁne the minimum distance of a code as
d
min
= min
.,.∕=0
d
0.
, (3)
which implicitly deﬁnes C
0
as the allzero codeword. It is noted that
the minimum distance of a linear block code is a critical parameter
in determining its probability of error.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 13
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Generator matrix
The generator matrix is a compact description of how codewords are
generated from information bits in a linear block code.
The design goal in linear block codes is to ﬁnd generator matrices
such that their corresponding codes are easy to encode and decode
yet have powerful error correction/detection capabilities.
Consider an (n, /) code with / information bits denoted as
U
.
= [n
.1
, . . . , n
.
] (4)
that are encoded into the codeword
C
.
= [c
.1
, . . . , c
.n
] (5)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 14
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Generator matrix (cont.)
We represent the encoding operation as a set of n equations:
c
.¸
= n
.1
p
1¸
+ n
.2
p
2¸
+ . . . + n
.
p
¸
, , = 1, . . . , n, (6)
where p
.¸
is binary (0 or 1) and binary (standard) multiplication is
used. One can write these n equations in matrix form as
C
.
= U
.
G, (7)
where the / ×n generator matrix G for the code is deﬁned as
G =
⎡
⎢
⎢
⎢
⎣
p
11
p
12
. . . p
1n
p
21
p
22
. . . p
2n
.
.
.
.
.
.
.
.
.
.
.
.
p
1
p
2
. . . p
n
⎤
⎥
⎥
⎥
⎦
. (8)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 15
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Generator matrix (cont.)
If we denote the th row of G as g
l
= [p
l1
, . . . , p
ln
] then we can
write any codeword C
.
as linear combinations of these row vectors
as follows:
C
.
= n
.1
g
1
+ n
.2
g
2
+ . . . + n
.
g

. (9)
Since a linear (n, /) block code is a subspace of dimension / in the
larger ndimensional space, the / row vectors {g
l
}

l=1
of G must be
linearly independent, so that they span the /dimensional subspace
associated with the 2

codewords.
Thus, G has rank /.
Since the set of basis vectors for this subspace is not unique, the
generator matrix is also not unique.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 16
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Generator matrix (cont.)
A systematic linear block code is described by a generator matrix of
the form
G = [I

∣P] =
⎡
⎢
⎢
⎢
⎣
1 0 . . . 0 j
11
j
12
. . . j
1(n−)
0 1 . . . 0 j
21
j
22
. . . j
2(n−)
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 . . . 1 j
1
j
2
. . . j
(n−)
⎤
⎥
⎥
⎥
⎦
,
where I

is a / ×/ identity matrix and P is a / ×(n −/) matrix
that determines the redundant, or parity, bits to be used for error
correction or detection.
The codeword output from a systematic encoder is of the form
C
.
= U
.
G = U
.
[I

∣P] = [n
.1
, . . . , n
.
, j
1
, . . . , j
n−
] , (10)
where
j
¸
= n
.1
j
1¸
+ . . . + n
.
j
¸
, , = 1, . . . , n −/. (11)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 17
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
An example: Problem statement
Systematic linear block codes are typically implemented with n −/
modulo2 adders tied to the appropriate stages of a shift register.
The resultant parity bits are appended to the end of the information
bits to form the codeword.
Find the corresponding implementation for generating a (7, 4) binary
code with the generator matrix
G =
⎡
⎢
⎢
⎣
1 0 0 0 1 1 0
0 1 0 0 1 0 1
0 0 1 0 0 0 1
0 0 0 1 0 1 0
⎤
⎥
⎥
⎦
,
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 18
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
An example: Solutions
The matrix G is already in systematic form with
P =
⎡
⎢
⎢
⎣
1 1 0
1 0 1
0 0 1
0 1 0
⎤
⎥
⎥
⎦
,
From (11), the ﬁrst parity bit in the codeword is
j
1
= n
.1
j
11
+ n
.2
j
21
+ n
.3
j
31
+ n
.4
j
41
= n
.1
+ n
.2
+ 0 + 0.
Similarly, j
2
= n
.1
+ n
.4
, j
3
= n
.2
+ n
.3
.
The codeword output is [n
.1
, n
.2
, n
.3
, n
.4
, j
1
, j
2
, j
3
]
+ + +
u
i1
u
u u
i2
i3 i4
p
1 2
p
3
p
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 19
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Parity Check Matrix and Syndrome Testing
The parity check matrix is used to decode linear block codes with
generator matrix G. The parity check matrix H corresponding to a
generator matrix G = [I

∣P] is deﬁned as
H =
[
P
T
∣I
n−
]
. (12)
It is easily veriﬁed that GH
T
= 0
,n−
, where 0
,n−
denotes an
allzero / ×(n −/) matrix. Thus,
C
.
H
T
= U
.
GH
T
= 0
n−
, (13)
where 0
n−
denotes the allzero row vector of length n −/.
Multiplication of any valid codeword with the parity check matrix
results in an allzero vector. This property is used to determine
whether the received vector is a valid codeword or has been
corrupted, based on the notion of syndrome testing, as deﬁned now.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 20
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Parity Check Matrix and Syndrome Testing (cont.)
Let R be the received codeword resulting from transmission of
codeword C.
In the absence of channel errors, R = C.
If the transmission is corrupted, one or more of the codeword
symbols in R will diﬀer from those in C. Therefore, the received
codeword can be written as
R = C+e, (14)
where e = [c
1
, c
2
, . . . , c
n
] is the error vector indicating which
codeword symbols were corrupted by the channel.
We deﬁne the syndrome of R as
S = RH
T
. (15)
If R is a valid codeword (i.e.,R = C
.
for some i), then
S = C
.
H
T
= 0
n−
.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 21
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Parity Check Matrix and Syndrome Testing (cont.)
Thus, the syndrome equals the allzero vector if the transmitted
codeword is not corrupted, or is corrupted in a manner such that
the received codeword is a valid codeword in the code that is
diﬀerent from the transmitted codeword.
If the received codeword R contains detectable errors, then
S

∕= 0
n−
.
If the received codeword contains correctable errors, then the
syndrome identiﬁes the error pattern corrupting the transmitted
codeword, and these errors can then be corrected.
Note that the syndrome is a function only of the error pattern e and
not the transmitted codeword C, since
S = RH
T
= (C+e)H
T
= CH
T
+eH
T
= 0
n−
+eH
T
. (16)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 22
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Parity Check Matrix and Syndrome Testing (cont.)
Since S = eH
T
corresponds to n−/ equations in n unknowns, there
are 2

possible error patterns that can produce a given syndrome S.
However, since the probability of bit error is typically small and
independent for each bit, the most likely error pattern is the one
with minimal weight, corresponding to the least number of errors
introduced in the channel.
Thus, if an error pattern ˆe is the most likely error associated with a
given syndrome S, the transmitted codeword is typically decoded as
ˆ
C = R+ˆe = C+e +ˆe. (17)
When the most likely error pattern does occur, i.e., ˆe = e, then
ˆ
C = C, i.e., the corrupted codeword is correctly decoded. The
decoding process and associated error probability will be covered in
later Sections.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 23
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Parity Check Matrix and Syndrome Testing (cont.)
Let C
u
denote a codeword in a given (n, /) code with minimum
weight (excluding the allzero codeword).
Then C
u
H
T
= 0
n−
is just the sum of d
min
columns of H
T
, since
d
min
equals the number of 1s (the weight) in the minimum weight
codeword of the code.
Since the rank of H
T
is at most n −/, this implies that the
minimum distance of an (n, /) block code is upper bounded by
d
min
≤ n −/ + 1. (18)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 24
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Hard Decision Decoding (HDD)
The probability of error for linear block codes depends on whether
the decoder uses soft decisions or hard decisions.
In hard decision decoding (HDD) each coded bit is demodulated as
a 0 or 1, i.e. the demodulator detects each coded bit (symbol)
individually.
For example, in BPSK, the received symbol is decoded as a 1 if it is
closer to
√
1
b
and as 0 if it closer to −
√
1
b
.
Hard decision decoding uses minimumdistance decoding based on
Hamming distance.
In minimumdistance decoding the n bits corresponding to a
codeword are ﬁrst demodulated, and the demodulator output is
passed to the decoder.
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 25
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Hard Decision Decoding (cont.)
The decoder compares this received codeword to the 2/ possible
codewords comprising the code, and decides in favor of the
codeword that is closest in Hamming distance (diﬀers in the least
number of bits) to the received codeword.
Mathematically, for a received codeword R the decoder uses the
formula:
choose C
¸
subject to d (C
¸
, R) ≤ d (C
.
, R) , ∀i ∕= ,. (19)
If there is more than one codeword with the same minimum distance
to R, one of these is chosen at random by the decoder.
Maximumlikelihood decoding picks the transmitted codeword that
has the highest probability of having produced the received
codeword, i.e. given the received codeword R, the
maximumlikelihood decoder chooses the codeword C
¸
as
C
¸
= arg max
.=1,...,2
j(R∣C
.
). (20)
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 26
Introduction
Linear Block Codes
Introduction
Binary linear block codes
Generator matrix
Parity Check Matrix and Syndrome Testing
Hard Decision Decoding (HDD)
Hard Decision Decoding (cont.)
Since the most probable error event in an AWGN channel is the
event with the minimum number of errors needed to produce the
received codeword, the minimumdistance criterion (19) and the
maximumlikelihood criterion (8.33) are equivalent.
Once the maximumlikelihood codeword C
.
is determined, it is
decoded to the / bits that produce codeword C
.
.
C
1
C
C
C
2
3
4
min
d
t
Mobile communicationsChapter 3: Physicallayer transmissions Section 3.4: Channel Coding 27
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Chapter 3: Physicallayer transmission techniques
Section 3.4: Orthogonal Frequency Division Multiplexing
(OFDM) Fundamentals
Instructor: Nguyen Le Hung
Email: nlhung@dut.udn.vn; nnguyenlehung@yahoo.com
Department of Electronics & Telecommunications Engineering
Danang University of Technology, University of Danang
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 1
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
1
Introduction
Development of mobile communications systems  Revisit
Mobile broadband technology evolution  Revisit
2
Fundamentals of OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
3
Possible research problems in OFDM transmissions
Carrier frequency oﬀset (CFO) estimation
Doubly selective channel estimation
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 2
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Development of mobile communications systems  Revisit
Mobile broadband technology evolution  Revisit
Development of mobile communications systems  Revisit
time
code
frequency
code
space
FDMA (1G)
e.g., AMPS ~ 1980s
TDMA (2G)
e.g., GSM ~ 1990s
OFDM, SDMA (4G)
e.g., WiMAX, LTE
2010s
CDMA (3G)
e.g., WCDMA ~ 2000s
frequency
time
time
~ 1 Gbps (stationary),
~ 100 Mbps (mobile)
frequency
frequency
~ 14 Mbps (downlink),
~ 5.8 Mbps (uplink)
~ 50 Kbps
A new signal dimension will be exploited in 5G ?
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 3
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Development of mobile communications systems  Revisit
Mobile broadband technology evolution  Revisit
Development of mobile communications systems  Revisit
time
code
frequency
code
space
FDMA (1G)
e.g., AMPS ~ 1980s
TDMA (2G)
e.g., GSM ~ 1990s
OFDM, SDMA (4G)
e.g., WiMAX, LTE
2010s
CDMA (3G)
e.g., WCDMA ~ 2000s
frequency
time
time
~ 1 Gbps (stationary),
~ 100 Mbps (mobile)
frequency
frequency
~ 14 Mbps (downlink),
~ 5.8 Mbps (uplink)
~ 50 Kbps
A new signal dimension will be exploited in 5G ?
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 3
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Development of mobile communications systems  Revisit
Mobile broadband technology evolution  Revisit
Mobile broadband technology evolution  Revisit
LTE
Advanced
802.16m
R 2.0
WCDMA
3GPP
HSDPA HSPA
Mobile
WiMAX
802.16e
R 1.0
EVDO
Rev A
EVDO
Rev B
CDMA 2000
3GPP2
LTE
802.16e
R 1.5
2005 2009 2010 2006 2007 2008 2011 2012
CDMA based OFDMA based
R99
“4G”
IMTAdvanced
The last decade has witnessed numerous intensive studies in employing
orthogonal frequency division multiplexing (OFDM) for the emerging
broadband communications systems (e.g., WiFi, WiMAX, LTE) to exploit
its high spectral eﬃciency and robustness against multipath (frequency
selective) fading channels.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 4
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Functional blocks in OFDM transmitter and receiver
Burst structure
S/P
IFFT
Insert CP
DAC RF
Clk
Osc
RF
LO
Pilotinsertion
Conv.
Encoder
P/S
c
i
MQAM
mapping
Pilot OFDM
symbols
Data OFDM
symbols
Information
bits,u
i
RF
LO
Clk
Osc
MQAM
Demapper
RF ADC
S/P decoder
1
FFT
CP
Information
bits,u
i
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 5
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Transmitted OFDM signals
Consider a coded OFDM system using `ary quadrature amplitude
modulation (MQAM).
The encoded bit stream is bitinterleaved. Then, the resulting
sequence of interleaved bits is organized as a sequence of Qbit
tuples {d
,n
} where Q = log
2
` and
d
,n
= [d
,n,0
, ..., d
,n,Q−1
]
T
.
The sequence is further mapped to a complexvalued symbol
A
,n
∈ where is the `ary modulation signaling alphabet, and
: and / denote the indices of OFDM symbol and subcarrier,
respectively.
m
X
, 0
m N
X
, 1
1 , 0 m
X
1 , 1 m N
X
m k
X
,
1 , m k
X
T
Q m k m k m k
d d ] ,..., [
1 , , 0 , , ,
d
The mth OFDM symbol
The kth subcarrier
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 6
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Transmitted OFDM signals
Consider a coded OFDM system using `ary quadrature amplitude
modulation (MQAM).
The encoded bit stream is bitinterleaved. Then, the resulting
sequence of interleaved bits is organized as a sequence of Qbit
tuples {d
,n
} where Q = log
2
` and
d
,n
= [d
,n,0
, ..., d
,n,Q−1
]
T
.
The sequence is further mapped to a complexvalued symbol
A
,n
∈ where is the `ary modulation signaling alphabet, and
: and / denote the indices of OFDM symbol and subcarrier,
respectively.
m
X
, 0
m N
X
, 1
1 , 0 m
X
1 , 1 m N
X
m k
X
,
1 , m k
X
T
Q m k m k m k
d d ] ,..., [
1 , , 0 , , ,
d
The mth OFDM symbol
The kth subcarrier
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 6
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Transmitted OFDM signals (cont.)
Each OFDM symbol consists of · information bearing subcarriers
A
,n
, / = 0, ..., · −1, where · is the size of the fast Fourier transform
(FFT) and inverseFFT (IFFT) used in the multicarrier transmission.
m n
x
,
QAM symbols
m k
X
,
CP samples
IFFT
P/S
DAC RF
Burst structure
S/P
IFFT Insert CP
DAC RF
Clk
Osc
RF
LO
Pilotinsertion
Conv.
Encoder
P/S
c
i
MQAM
mapping
Information
bits,u
i
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 7
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Transmitted OFDM signals (cont.)
Copy and paste the last N
g
samples
CP portion CP portion
m N N
g
x
,
….
m N
x
, 1 m
x
, 0 m
x
, 1 m
x
, 2
….
m N N
g
x
, 1 m N N
g
x
,
….
m N
x
, 1
one OFDM symbol of samples
After inverse FFT (IFFT) and cyclic preﬁx (CP) insertion, the
transmitted baseband signal of the :th OFDM symbol can be written as
r
n,n
=
1
√
·
Þ−1
∑
=0
A
,n
exp
(
,2¬/n
·
)
, (1)
where n ∈ {−·
¸
, ..., 0, ..., · −1}, ·
¸
denotes the CP length.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 8
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Transmitted OFDM signals (cont.)
Copy and paste the last N
g
samples
CP portion CP portion
m N N
g
x
,
….
m N
x
, 1 m
x
, 0 m
x
, 1 m
x
, 2
….
m N N
g
x
, 1 m N N
g
x
,
….
m N
x
, 1
one OFDM symbol of samples
After inverse FFT (IFFT) and cyclic preﬁx (CP) insertion, the
transmitted baseband signal of the :th OFDM symbol can be written as
r
n,n
=
1
√
·
Þ−1
∑
=0
A
,n
exp
(
,2¬/n
·
)
, (1)
where n ∈ {−·
¸
, ..., 0, ..., · −1}, ·
¸
denotes the CP length.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 8
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels
Two Main
Multipaths
Local
Scattering
j
n
=
∑
l
r
n−l
ℎ
l,n
+n
b
(:,\) (2)
The th (timevariant) channel tap gain that includes the eﬀect of
transmitreceive ﬁlters and timevariant channel propagation is denoted
by ℎ
l,n,n
where n and : stand for the indices of the timedomain sample
and OFDM symbol, respectively.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 9
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels (cont.)
In the timevariant multipath channels, a number of basis expansion
models (BEMs) can be employed for reducing the number of channel
parameters while capturing the timevariation of the channels.
Using BEMs, ℎ
l,n,n
(after CP removal) can be approximated by
ℎ
l,n,n
=
Q
∑
g=1
/
n+Þ
¸
+nÞ
s
,g
c
g,l
,  ∈ {0, ..., 1 −1}, (3)
where ·
s
= · +·
¸
denotes the OFDM symbol length after CP
insertion, n = 0, ..., · −1, : = 0, ..., ` −1 and ` is the number
of both data and pilot OFDM symbols in a burst. /
n+Þ
¸
+nÞ
s
,g
stand for the ¡th basis function values of the used BEM.
Copy and paste the last N
g
samples
CP portion CP portion
m N N
g
x
,
….
m N
x
, 1 m
x
, 0 m
x
, 1 m
x
, 2
….
m N N
g
x
, 1 m N N
g
x
,
….
m N
x
, 1
one OFDM symbol of samples
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 10
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels (cont.)
In the timevariant multipath channels, a number of basis expansion
models (BEMs) can be employed for reducing the number of channel
parameters while capturing the timevariation of the channels.
Using BEMs, ℎ
l,n,n
(after CP removal) can be approximated by
ℎ
l,n,n
=
Q
∑
g=1
/
n+Þ
¸
+nÞ
s
,g
c
g,l
,  ∈ {0, ..., 1 −1}, (3)
where ·
s
= · +·
¸
denotes the OFDM symbol length after CP
insertion, n = 0, ..., · −1, : = 0, ..., ` −1 and ` is the number
of both data and pilot OFDM symbols in a burst. /
n+Þ
¸
+nÞ
s
,g
stand for the ¡th basis function values of the used BEM.
Copy and paste the last N
g
samples
CP portion CP portion
m N N
g
x
,
….
m N
x
, 1 m
x
, 0 m
x
, 1 m
x
, 2
….
m N N
g
x
, 1 m N N
g
x
,
….
m N
x
, 1
one OFDM symbol of samples
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 10
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels (cont.)
It is noted that the use of BEMs oﬀers a signiﬁcant dimension
reduction in the timevariant channel representation, i.e., Q ≪ ·`.
For instance, in the current LTE system settings, three LTE time
slots contain 21 OFDM symbols with 128FFT (the smallest FFT
size used in the LTE settings) and the resulting number of the
timevariant channel parameters corresponding to one channel tap
gain ℎ
l,n,n
in (3) will be ·` = 128 ×21 = 2688.
By using the discrete prolate spheroidal (DPS)BEM as shown in (3),
the number of the basis functions Q can vary from 3 (for low user
speeds, e.g., about 10km/h) to 5 (for moderate user speeds, e.g.,
about 100km/h) under the required MSE of the DPSBEMbased
channel approximation below 10
−10
as shown in Fig. 1.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 11
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels (cont.)
It is noted that the use of BEMs oﬀers a signiﬁcant dimension
reduction in the timevariant channel representation, i.e., Q ≪ ·`.
For instance, in the current LTE system settings, three LTE time
slots contain 21 OFDM symbols with 128FFT (the smallest FFT
size used in the LTE settings) and the resulting number of the
timevariant channel parameters corresponding to one channel tap
gain ℎ
l,n,n
in (3) will be ·` = 128 ×21 = 2688.
By using the discrete prolate spheroidal (DPS)BEM as shown in (3),
the number of the basis functions Q can vary from 3 (for low user
speeds, e.g., about 10km/h) to 5 (for moderate user speeds, e.g.,
about 100km/h) under the required MSE of the DPSBEMbased
channel approximation below 10
−10
as shown in Fig. 1.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 11
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Timevariant multipath channels (cont.)
It is noted that the use of BEMs oﬀers a signiﬁcant dimension
reduction in the timevariant channel representation, i.e., Q ≪ ·`.
For instance, in the current LTE system settings, three LTE time
slots contain 21 OFDM symbols with 128FFT (the smallest FFT
size used in the LTE settings) and the resulting number of the
timevariant channel parameters corresponding to one channel tap
gain ℎ
l,n,n
in (3) will be ·` = 128 ×21 = 2688.
By using the discrete prolate spheroidal (DPS)BEM as shown in (3),
the number of the basis functions Q can vary from 3 (for low user
speeds, e.g., about 10km/h) to 5 (for moderate user speeds, e.g.,
about 100km/h) under the required MSE of the DPSBEMbased
channel approximation below 10
−10
as shown in Fig. 1.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 11
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Numerical results of BEMbased channel ﬁtting
1 2 3 4 5 6 7 8 9 10 11 12 13 14
10
−30
10
−25
10
−20
10
−15
10
−10
10
−5
10
0
Number of used basis functions
M
S
E
o
f
D
P
S
−
B
E
M
−
b
a
s
e
d
c
h
a
n
n
e
l
f
i
t
t
i
n
g
1 km/h
5 km/h
10 km/h
20 km/h
30 km/h
40 km/h
50 km/h
60 km/h
70 km/h
80 km/h
90 km/h
100 km/h
120 km/h
140 km/h
160 km/h
180 km/h
200 km/h
Figure 1: Normalized MSE of DPSBEMbased approximation (ﬁtting) of
timevarying channels generated by Jakes model under diﬀerent mobile speeds.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 12
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals
Over the above doublyselective channels, after CP removal, the nth
received sample in the :th OFDM symbol can be represented by
j
n,n
=
J−1
∑
l=0
ℎ
l,n,n
r
n−l,n
+.
n,n
, (4)
where n = 0, ..., · −1 and .
n,n
is the additive white Gaussian noise
(AWGN) with variance ·
o
.
m N N
g
x
,
….
m N
x
, 1 m
x
, 0 m
x
, 1 m
x
, 2
….
m N N
g
x
, 1 m N N
g
x
,
….
m N
x
, 1
one OFDM symbol of samples
CP portion
m N
g
h
, 0 ,
….
m
h
, 0 , 1 m
h
, 0 , 0
Channel impulse response
m m m N m m N N m N m
x h x h x h y
g g
, 0 , 0 , 0 , 1 , 0 , 1 , , 0 , , 0
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 13
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals in the frequency domain
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+j
,n
+7
,n
, (5)
where H
,n
=
1
Þ
∑
Þ−1
n=0
∑
J−1
l=0
ℎ
l,n,n
c
−¸2tl¸Þ
,
7
,n
=
1
√
Þ
∑
Þ−1
n=0
.
n,n
c
(
−¸
2¬!r
.
)
and j
,n
=
1
Þ
∑
Þ−1
.=0,.∕ =
A
.,n
∑
Þ−1
n=0
(
∑
J−1
l=0
ℎ
l,n,n
c
−¸2t.l¸Þ
)
c
¸2tn(.−)¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l,n,n
r
n−l,n
+.
n,n
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 14
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals in the frequency domain
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+j
,n
+7
,n
, (5)
where H
,n
=
1
Þ
∑
Þ−1
n=0
∑
J−1
l=0
ℎ
l,n,n
c
−¸2tl¸Þ
,
7
,n
=
1
√
Þ
∑
Þ−1
n=0
.
n,n
c
(
−¸
2¬!r
.
)
and j
,n
=
1
Þ
∑
Þ−1
.=0,.∕ =
A
.,n
∑
Þ−1
n=0
(
∑
J−1
l=0
ℎ
l,n,n
c
−¸2t.l¸Þ
)
c
¸2tn(.−)¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l,n,n
r
n−l,n
+.
n,n
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 14
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals in the frequency domain
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+j
,n
+7
,n
, (5)
where H
,n
=
1
Þ
∑
Þ−1
n=0
∑
J−1
l=0
ℎ
l,n,n
c
−¸2tl¸Þ
,
7
,n
=
1
√
Þ
∑
Þ−1
n=0
.
n,n
c
(
−¸
2¬!r
.
)
and j
,n
=
1
Þ
∑
Þ−1
.=0,.∕ =
A
.,n
∑
Þ−1
n=0
(
∑
J−1
l=0
ℎ
l,n,n
c
−¸2t.l¸Þ
)
c
¸2tn(.−)¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l,n,n
r
n−l,n
+.
n,n
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 14
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals over blockfading channel
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+ 0 +7
,n
, (6)
where H
,n
=
∑
J−1
l=0
ℎ
l
c
−¸2tl¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l
r
n−l,n
+ .
n,n
.
Burst structure
S/P
IFFT Insert CP
DAC RF
Clk
Osc
RF
LO
Pilotinsertion
Conv.
Encoder
P/S
c
i
MQAM
mapping
Information
bits,u
i
RF
LO
Clk
Osc
MQAM
Demapper
RF ADC S/P decoder
1
FFT CP
Information
bits,u
i
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 15
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals over blockfading channel
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+ 0 +7
,n
, (6)
where H
,n
=
∑
J−1
l=0
ℎ
l
c
−¸2tl¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l
r
n−l,n
+ .
n,n
.
Burst structure
S/P
IFFT Insert CP
DAC RF
Clk
Osc
RF
LO
Pilotinsertion
Conv.
Encoder
P/S
c
i
MQAM
mapping
Information
bits,u
i
RF
LO
Clk
Osc
MQAM
Demapper
RF ADC S/P decoder
1
FFT CP
Information
bits,u
i
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 15
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Received OFDM signals over blockfading channel
At the receiver, after performing CP removal and FFT, the received
samples in the frequency domain can be determined by
Y
,n
=
1
√
·
Þ−1
∑
n=0
j
n,n
c
−¸2tn¸Þ
= H
,n
A
,n
+ 0 +7
,n
, (6)
where H
,n
=
∑
J−1
l=0
ℎ
l
c
−¸2tl¸Þ
.
It is noted that the received samples in the time domain are:
j
n,n
=
J−1
∑
l=0
ℎ
l
r
n−l,n
+ .
n,n
.
Burst structure
S/P
IFFT Insert CP
DAC RF
Clk
Osc
RF
LO
Pilotinsertion
Conv.
Encoder
P/S
c
i
MQAM
mapping
Information
bits,u
i
RF
LO
Clk
Osc
MQAM
Demapper
RF ADC S/P decoder
1
FFT CP
Information
bits,u
i
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 15
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
MQAM symbol detection in the frequency domain
Under maximum likelihood (ML) detection, the transmitted QAM
symbols can be recovered in the frequency domain by
ˆ
A
,n
= arg min
·
!¸r
∈
∣Y
,n
−H
,n
A
,n
∣
2
. (7)
Based on the detected QAM symbols
ˆ
A
,n
, the transmitted data
bits can be recovered accordingly by channel decoders.
It is noted that the ML symbol detection process (7) needs to know
channel frequency response (CFR) H
,n
before providing a detected
version of the transmitted symbol
ˆ
A
,n
Possible research problem
In OFDMbased transmissions (e.g., WiFi, WiMAX, LTE systems) over
timevariant multipath channels, the problem of estimating channel
responses (i.e., H
,n
) is of importance in research and industry as well.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 16
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
MQAM symbol detection in the frequency domain
Under maximum likelihood (ML) detection, the transmitted QAM
symbols can be recovered in the frequency domain by
ˆ
A
,n
= arg min
·
!¸r
∈
∣Y
,n
−H
,n
A
,n
∣
2
. (7)
Based on the detected QAM symbols
ˆ
A
,n
, the transmitted data
bits can be recovered accordingly by channel decoders.
It is noted that the ML symbol detection process (7) needs to know
channel frequency response (CFR) H
,n
before providing a detected
version of the transmitted symbol
ˆ
A
,n
Possible research problem
In OFDMbased transmissions (e.g., WiFi, WiMAX, LTE systems) over
timevariant multipath channels, the problem of estimating channel
responses (i.e., H
,n
) is of importance in research and industry as well.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 16
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Functional blocks in OFDM transmitter and receiver
Transmitted OFDM signals
Timevariant multipath channels
Received OFDM signals
MQAM symbol detection in the frequency domain
Downlink multiuser MIMOOFDM transmissions in
WiMAX systems (IEEE 802.16e)
M
U

M
I
M
O
e
n
c
o
d
e
r
Base station
User 4
FFT Decoder
I
F
F
T
User 3
FFT Decoder
User 2
FFT Decoder
User 1
FFT Decoder
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 17
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Carrier frequency oﬀset (CFO) estimation
Doubly selective channel estimation
Carrier frequency oﬀset (CFO) estimation
In the presence of CFO, after CP removal, the nth received sample
in the :th OFDM symbol can be represented by
j
n,n
= c
¸
2¬z
.
(n+Þ
¸
+nÞ
s
)
J−1
∑
l=0
ℎ
l,n,n
r
n−l,n
+.
n,n
, (8)
where:
= 0, ..., −1
Δ and = Δ denote the absolute and normalized CFOs,
respectively.
is the sampling period of the system.
As observed in (8), the presence of CFO introduces a timedomain
phase rotation that will translate into ICI in the frequency domain.
In addition, the timevariation of the multipath channels also induces
ICI in the frequency domain.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 18
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Carrier frequency oﬀset (CFO) estimation
Doubly selective channel estimation
Carrier frequency oﬀset (CFO) estimation (cont.)
Consequently, the presence of both CFO and doubly selective
channels would incur a signiﬁcant ICI power at OFDM receivers,
giving rise to a considerable irreducible error ﬂoor in the receiver
performance.
For CFO compensation and reliable coherent data
detection/decoding, the CFO and CIR estimates are indispensible at
OFDM receivers.
To obtain the estimates, CFO and channel estimates can be
obtained either jointly or separately by existing estimation
techniques (e.g., RLS, ML, MAP,...).
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 19
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Carrier frequency oﬀset (CFO) estimation
Doubly selective channel estimation
Doubly selective channel estimation: Introduction
The last decade has witnessed numerous intensive studies in
employing OFDM for broadband communication systems to exploit
its high spectral eﬃciency and robustness against multipath
(frequencyselective) fading channels.
In the current literature, most of these studies have assumed
frequencyselective channels to be timeinvariant (i.e., quasistatic or
blockfading) within a transmission burst.
This channel assumption can be used in a wireless system with
stationary and/or lowspeed users.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 20
Outline
Introduction
Fundamentals of OFDM transmissions
Possible research problems in OFDM transmissions
Carrier frequency oﬀset (CFO) estimation
Doubly selective channel estimation
Doubly selective channel estimation: Introduction (cont.)
In a wireless network with rapidly moving nodes (e.g., users in cars
and trains in 4GLTE systems), the resultant timeselectivity of the
channel impulse response (CIR) introduces a large number of
channel parameters (much greater than that of
quasistatic/blockfading channels).
In addition, the timevariation of the channel leads to a loss of
subcarrier orthogonality, resulting in intercarrier interference (ICI) in
OFDM receivers.
Under such a scenario, the assumption of quasistatic fading
channels becomes inappropriate.
As a result, time and frequencyselective (doubly selective) channels
should be considered in the wireless system investigation and
analysis.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.4: OFDM fundamentals 21
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Chapter 3: Physicallayer transmission techniques
Section 3.5: Diversity techniques
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 1
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
1
Introduction
2
Independent Fading Paths
Space diversity
Frequency diversity
Time diversity
3
Receiver diversity techniques
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
4
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 2
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Introduction
As observed in Section 3.2, Rayleigh fading induces a very large
power penalty on the performance of modulation over wireless
channels.
One of the most powerful techniques to mitigate the eﬀects of fading
is to use diversitycombining of independently fading signal paths.
Diversitycombining exploits the fact that independent signal paths
have a low probability of experiencing deep fades simultaneously.
These independent paths are combined in some ways such that the
fading of the resultant signal is reduced.
Diversity techniques that mitigate the eﬀect of multipath fading are
called microdiversity.
Diversity to mitigate the eﬀects of shadowing from buildings and
objects is called macrodiversity. Macrodiversity is generally
implemented by combining signals received by several base stations
or access points.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 3
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Space diversity
Frequency diversity
Time diversity
Space diversity
There are many ways of achieving independent fading paths in a
wireless system.
One method is to use multiple transmit or receive antennas, also
called an antenna array, where the elements of the array are
separated in distance. This type of diversity is referred to as space
diversity.
Note that with receiver space diversity, independent fading paths are
generated without an increase in transmit signal power or bandwidth.
Coherent combining of the diversity signals leads to an increase in
SNR at the receiver over the SNR that would be obtained with just
a single receive antenna.
Space diversity also requires that the separation between antennas is
large enough so that the fading amplitudes corresponding to each
antenna are approximately independent.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 4
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Space diversity
Frequency diversity
Time diversity
Frequency diversity
Frequency diversity is achieved by transmitting the same narrowband
signal at diﬀerent carrier frequencies.
This technique requires additional transmit power to send the signal
over multiple frequency bands.
Spread spectrum techniques are sometimes described as providing
frequency diversity since the channel gain varies across the
bandwidth of the transmitted signal.
However, this is not equivalent to sending the same information
signal over indepedently fading paths.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 5
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Space diversity
Frequency diversity
Time diversity
Time diversity
Time diversity is achieved by transmitting the same signal at
diﬀerent times.
Time diversity does not require increased transmit power, but it does
decrease the data rate since data is repeated in the diversity time
slots rather than sending new data in these time slots.
Time diversity can also be achieved through coding and interleaving.
Time diversity cannot be used for stationary wireless applications,
since fading gains are highly correlated over time.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 6
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Maximal Ratio Combining (MRC)
1
1 1
T j
e a h
2
2 2
T j
e a h
M
j
M M
e a h
T
x
X
X
X
1
1
T j
e g
¦
2
2
T j
e g
M
j
M
e g
T
De
mod
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 7
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Maximal Ratio Combining (cont.)
In receiver diversity the independent fading paths associated with
multiple receive antennas are combined to obtain a resultant signal
that is then passed through a standard demodulator.
Under the use of ` receive antennas over ﬂatfading (single
channeltap, i.e., 1 = 1) channels, the received signals are
j
.
= ℎ
.
r + n
.
, i = 1, ..., ` (1)
where ℎ
.
= ℎ
.,1
+ ,ℎ
.,J
= o
.
c
¸0
and n
.
∼ (0, ·
0
).
Weight each branch with p
.
c
−¸0
: Cophasing. If not cophasing,
then what happens ?
Combine signals from these ` receive antennas, one have
j =
1
∑
.=1
p
.
c
−¸0
j
.
=
(
1
∑
.=1
p
.
o
.
)
r +
1
∑
.=1
p
.
c
−¸0
n
.
. (2)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 8
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Maximal Ratio Combining (cont.)
After combining the signals, the resultant SNR is
SNR =
(
∑
1
.=1
p
.
o
.
)
2
·
0
∑
1
.=1
p
2
.
, (3)
One needs to ﬁnd {p
.
}
1
.=1
to maximize SNR ?.
The solution to the simple optimization problem can be obtained by
taking partial derivatives of (3) or using the Swartz inequality. In
particular, the solution is
p
:
= o
.
,
√
·
0
(4)
and the resultant combined SNR ¸
Σ
is
¸
Σ
=
∑
=1
o
2
Þ
0
=
∑
1
.=1
¸
.
. (5)
The ¸
Σ
increases linearly with the number of diversity branches `.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 9
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Maximal Ratio Combining: An example of 2 Rxantennas
1
1 1
T j
e a h
2
2 2
T j
e a h
Channel
estimator
*
1
h
*
1
h
X X
*
Channel
estimator
*
2
h
2
h
Maximum
likelihood
detector
1 1 1
n x h y
1
n
x
2
n
2 2 2
n x h y
Interference
+ noise
Interference
+ noise
y
Anten1
Anten2
xˆ
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 10
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
MRC: Probability of error in symbol detection
The detection performance of a diversity system, whether it uses
space diversity or another form of diversity, in terms of probability of
symbol error 1
c
for demodulation in AWGN with SNR ¸
Σ
can be
determined by
1
c
=
∫
∞
0
1
c
(¸)j
;
Σ
(¸)d¸. (6)
We can obtain a simple upper bound on the average probability of
error by applying the Chernoﬀ bound Q(r) ≤ c
−i
2
¸2
to the Q
function.
Recall that for static channel gains with MRC, we can approximate
the probability of error as
1
c
= c
1
Q
(
√
a
1
¸
Σ
)
≤ c
1
c
−o
;
Σ
¸2
= c
1
c
−o
(;
1
+...+;
)¸2
.
(7)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 11
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
MRC: Probability of error in symbol detection (cont.)
Integrating over the chisquared distribution for ¸
Σ
yields
1
c
≤ c
1
1
∏
.=1
1
1 + a
1
¸
.
,2
. (8)
In the limit of high SNR and assuming that the ¸
.
’s are identically
distributed with ¸
.
= ¸, one will have
1
c
≈ c
1
(
a
1
¸
2
)
−1
. (9)
The distribution of the combined SNR j
;
Σ
(¸) leads to a decrease in
1
c
due to diversity combining.
The resultant performance advantage is called the diversity gain.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 12
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Diversity order
For some diversity systems, their averaged probability of error can be
expressed in the form
1
c
= c¸
−1
, (10)
where c is a constant depending on the speciﬁc modulation and
coding, ¸ is the averaged received SNR per branch and ` is called
the diversity order of the system.
The diversity order indicates how the slope of the average probability
of error as a function of averaged SNR changes with diversity.
Recall that a general approximation for average error probability in
Rayleigh fading with no diversity is 1
c
= c
1
,(2a
1
¸). This
expression has a diversity order of one, consistent with a single
receive antenna.
The maximum diversity order of a system with ` antennas is `,
and when the diversity order equals ` the system is said to achieve
full diversity order.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 13
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Diversity order: Numerical results of MRC
0 5 10 15 20 25 30
10
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10
0
γ
b
(dB)
P
b
M = 1
M = 2
M = 4
M = 8
M = 10
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 14
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
EqualGain Combining (EGC)
1
1 1
T j
e a h
2
2 2
T j
e a h
M
j
M M
e a h
T
x
X
X
X
1
T j
e
¦
2
T j
e
M
j
e
T
De
mod
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 15
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
EqualGain Combining (cont.)
MRC requires knowledge of the timevarying SNR on each branch,
which can be very diﬃcult to measure.
A simpler technique is equalgain combining, which cophases the
signals on each branch and then combines them with equal
weighting, i.e., p
.
= c
−¸0
.
The SNR of the combiner output, assuming the same noise PSD ·
0
in each branch, is then given by
¸
Σ
=
1
·
0
`
(
1
∑
.=1
∣ℎ
.
∣
)
2
, (11)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 16
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Selection combining (SC)
1
1 1
T j
e a h
2
2 2
T j
e a h
M
j
M M
e a h
T
x
De
mod
Measure
SNR
Measure
SNR
Measure
SNR
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 17
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Selection combining (cont.)
In selection combining (SC), the combiner outputs the signal on
the branch with the highest SNR.
Since only one branch is used at a time, SC often requires just one
receiver that is switched into the active antenna branch.
A dedicated receiver on each antenna branch may be needed for
systems that transmit continuously in order to simultaneously and
continuously monitor SNR on each branch.
Since only one branch output is used, cophasing of multiple
branches is not required.
As a result, this technique can be used with either coherent or
diﬀerential modulation.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 18
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Selection combining (cont.)
Under SC implementation, the pdf of ¸
Σ
is
j
;
Σ
(¸) =
`
¸
(
1 −c
−;¸;
)
1−1
c
−;¸;
(12)
As a result, the averaged output SNR of the combiner in Rayleigh
fading is
¸
Σ
= ¸
∑
1
.=1
1
.
. (13)
The average SNR gain increases with M, but not linearly.
The biggest gain is obtained by going from no diversity (` = 1) to
twobranch diversity (` = 2).
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 19
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
SC: averaged probability of error in BPSK detection
0 5 10 15 20 25 30
10
−6
10
−5
10
−4
10
−3
10
−2
10
−1
10
0
γ
b
(dB)
P
b
M = 1
M = 2
M = 4
M = 8
M = 10
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 20
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Threshold Combining (TC)
1
1 1
T j
e a h
2
2 2
T j
e a h
M
j
M M
e a h
T
x
De
mod
Compare
SNR
T
J
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 21
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Maximal Ratio Combining (MRC)
EqualGain Combining (EGC)
Selection combining (SC)
Threshold Combining (TC)
Threshold Combining (cont.)
SC for wireless systems transmitting continuously may require a
dedicated receiver on each branch to continuously monitor branch
SNR.
A simpler type of combining, called threshold combining, avoids
the need for a dedicated receiver on each branch by scanning each
of the branches in sequential order and outputting the ﬁrst signal
with SNR above a given threshold ¸
T
.
As in SC, since only one branch output is used at a time, cophasing
is not required.
Thus, this technique can be used with either coherent or diﬀerential
(noncoherent) modulation.
There are several criteria the combiner can use to decide which
branch to switch to.
The simplest criterion is to switch randomly to another branch.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 22
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Transmitter Diversity: Introduction
In transmit diversity, there are multiple transmit antennas with the
transmit power divided among these antennas.
Transmit diversity is desirable in cellular systems where more space,
power, and processing capability is available on the transmit side
rather than the receive side.
Transmit diversity design depends on whether or not the complex
channel gain is known at the transmitter or not.
When this gain is known, the system is very similar to receiver
diversity.
However, without this channel knowledge, transmit diversity gain
requires a combination of space and time diversity via a novel
technique called the Alamouti scheme.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 23
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Channel Known at Transmitter: Transmission model
1
1 1
T j
e a h
2
2 2
T j
e a h
M
j
M M
e a h
T
Demod
Channel
estimator
x
2
2
T j
e g
X
X
X
1
1
T j
e g
M
j
M
e g
T
Modulation
Coded bits
Base Station
Limited feedback link of channel state information (CSI)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 24
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Channel known at transmitter: detailed implementations
Consider a transmit diversity system with ` transmit antennas and
one receive antenna.
Assume the path gain associated with the ith transmit antenna
given by ℎ
.
= o
.
c
¸0
is known at the transmitter via limited feedback
links from mobile terminals.
This is referred to as having channel side information (CSI) at the
transmitter or CSIT.
Let r denote the transmitted signal with total energy per symbol 1
s
This signal is multiplied by a complex gain p
.
c
−¸0
, 0 ≤ p
.
≤ 1 and
sent through the ith transmit antenna.
Due to the average total energy constraint 1
s
, the weights p
.
c
−¸0
must satisfy
∑
1
.=1
p
2
.
= 1
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 25
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Channel known at transmitter (cont.)
The weighted signals transmitted over all antennas are added via
signal superposition at the receive antenna, which leads to a
received signal given by
j =
1
∑
.=1
p
.
o
.
r + n, n ∼ (0, ·
0
) . (14)
One can obtain the weights p
.
that achieve the maximum SNR:
p
.
=
o
√
∑
=1
o
2
, (15)
and the resultant SNR is
¸
Σ
=
1
s
·
0
1
∑
.=1
o
2
.
=
1
∑
.=1
¸
.
, (16)
where
=
2
/
0
equal to the branch SNR between the th transmit
antenna and the receive antenna.
Thus, we see that transmit diversity when the channel gains are known at
the transmitter is very similar to receiver diversity with MRC: the received
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 26
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Channel unknown at transmitter: the Alamouti scheme
We now consider the same model as in the previous subsection but
assume that the transmitter no longer knows the channel gains
ℎ
.
= o
.
c
¸0
, so there is no CSIT.
In this case it is not obvious how to obtain diversity gain. Consider,
for example, a naive strategy whereby for a twoantenna system we
divide the transmit energy equally between the two antennas.
Thus, the transmit signal on antenna i will be r
.
=
√
.5r where r is
the transmit signal with energy per symbol 1
s
.
Assume two antennas have complex Gaussian channel gains
{
ℎ
.
= o
.
c
¸0
}
2
.=1
with zeromean and unit variant (·
0
= 1).
The received signal is
j =
√
.5(ℎ
1
+ ℎ
2
)r + n. (17)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 27
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme (cont.)
Note that ℎ
1
+ ℎ
2
is the sum of two complex Gaussian random
variables, and is thus a complex Gaussian as well with mean equal to
the sum of means (zero) and variance equal to the sum of variances.
Thus
√
.5(ℎ
1
+ ℎ
2
) is a complex Gaussian random variable with
zeromean and unitvariance (1), so the received signal has the
same distribution as if we had just used one antenna with the
full energy per symbol.
In other words, we have obtained no performance advantage from
the two antennas, since we could not divide our energy intelligently
between them or obtain coherent combining through cophasing.
Transmit diversity gain can be obtained even in the absence of
channel information with an appropriate scheme to exploit the
antennas.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 28
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme (cont.)
A particularly simple and prevalent scheme for this diversity that
combines both space and time diversity was developed by Alamouti.
Alamouti’s scheme is designed for a digital communication system
with twoantenna transmit diversity.
The scheme works over two symbol periods where it is assumed
that the channel gain is constant over this time duration.
Over the ﬁrst symbol period two diﬀerent symbols :
1
and :
2
each
with energy 1
s
,2 are transmitted simultaneously from antennas 1
and 2, respectively.
Over the next symbol period, symbol −:
∗
2
is transmitted from
antenna 1 and symbol :
∗
1
is transmitted from antenna 2, each with
symbol energy 1
s
,2.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 29
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme (cont.)
Assume complex channel gains
{
ℎ
.
= o
.
c
¸0
}
2
.=1
between the ith
transmit antenna and the receive antenna.
The received symbol over the ﬁrst symbol period is
j
1
= ℎ
1
:
1
+ ℎ
2
:
2
+ n
1
, (18)
and the received symbol over the second symbol period is
j
2
= −ℎ
1
:
∗
2
+ ℎ
2
:
∗
1
+ n
2
, (19)
where {n
.
}
2
.=1
is the AWGN noise sample at the receiver associated
with the ith symbol transmission. We assume the noise sample has
zeromean and power of ·
0
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 30
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme (cont.)
The receiver uses these sequentially received symbols to form the
vector y = [j
1
j
∗
2
]
T
given by
y =
[
ℎ
1
ℎ
2
ℎ
∗
2
−ℎ
∗
1
] [
:
1
:
2
]
+
[
n
1
n
∗
2
]
= H
.
s + n, (20)
where H
.
=
[
ℎ
1
ℎ
2
ℎ
∗
2
−ℎ
∗
1
]
, s = [:
1
:
2
]
T
and n = [n
1
n
2
]
T
.
Let us deﬁne the new vector z = H
J
.
y. The structure of H
.
implies that
H
J
.
H
.
=
(
∣ℎ
1
∣
2
+ ∣ℎ
2
∣
2
)
I
2
(21)
is diagonal and thus
z = [.
1
.
2
]
T
=
(
∣ℎ
1
∣
2
+∣ℎ
2
∣
2
)
I
2
s + ˜ n, (22)
where ˜ n = H
J
.
n is a complex Gaussian noise vector with mean zero
and covariance matrix 1
(
˜ n˜ n
J
)
=
(
∣ℎ
1
∣
2
+∣ℎ
2
∣
2
)
I
2
·
0
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 31
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme (cont.)
The diagonal nature of z eﬀectively decouples the two symbol
transmissions, so that each component of z corresponds to one of
the transmitted symbols:
.
.
=
(
∣ℎ
1
∣
2
+∣ℎ
2
∣
2
)
:
.
+ ˜ n
.
, i = 1, 2. (23)
The received SNR thus corresponds to the SNR for .
.
given by
¸
.
=
(
∣ℎ
1
∣
2
+∣ℎ
2
∣
2
)
1
s
2·
0
, (24)
where the factor of 2 comes from the fact that :
.
is transmitted
using half the total symbol energy 1
s
.
The received SNR is thus equal to the sum of SNRs on each branch,
identical to the case of transmit diversity with MRC assuming that
the channel gains are known at the transmitter.
Thus, the Alamouti scheme achieves a diversity order of 2, the
maximum possible for a twoantenna transmit system, despite the
fact that channel knowledge is not available at the transmitter.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 32
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme: An example of 2 Txantennas
1
1 1
T j
e a h
2
2 2
T j
e a h
Channel
estimator
Combiner
Maximum likelihood detector
*
2
1
s
s
1
y
Antenna 1
Antenna 2
1
ˆ s
*
1
2
s
s
2
1
n
n
Interference
+ noise
1
h
2
h
2
y
2
ˆ s
1
h
2
h
Rxantenna
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 33
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
The Alamouti scheme: BER results of BPSK
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 34
Outline
Introduction
Independent Fading Paths
Receiver diversity techniques
Transmitter Diversity
Channel Known at Transmitter
Channel Unknown at Transmitter
Possible problems to be considered in theses
In Alamouti’s scheme, wireless channels are assumed to be ﬂat and
blockfading.
Doubly selective channels can be considered in Alamouti’s scheme
by using OFDM and BEMs.
The study results can be employed in LTE downlink transmissions
with mobile terminals.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.5: Diversity techniques 35
Outline
Introduction
Precoding
Scheduling (user selection)
Chapter 3: Physicallayer transmission techniques
Section 3.6: Space Division Multiple Access (SDMA)
Instructor: Nguyen Le Hung
Email: nlhung@dut.udn.vn; nnguyenlehung@yahoo.com
Department of Electronics & Telecommunications Engineering
Danang University of Technology, University of Danang
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 1
Outline
Introduction
Precoding
Scheduling (user selection)
1
Introduction
SDMA and OFDM
Multiuser transmission
2
Precoding
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
3
Scheduling (user selection)
Exhaustive selection
Greedy selection
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 2
Outline
Introduction
Precoding
Scheduling (user selection)
SDMA and OFDM
Multiuser transmission
SDMA with OFDM
The integration of multiantenna and OFDM techniques has
provided remarkable diversity and capacity gains in broadband
wireless communications.
In multiuser (MU) transmissions, the use of multiantenna array at
the base station (BS) enables simultaneous transmission of multiple
data streams to multiple users by exploiting spatial separations
among users.
A
B
S
/
e
N
B
A
M
S
/
U
E
(
a
)
IFFT
SUMIMO
precoder
A
B
S
/
e
N
B
A
M
S
/
U
E
1
(
b
)
IFFT
MUMIMO
precoder
A
M
S
/
U
E
2
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 3
Outline
Introduction
Precoding
Scheduling (user selection)
SDMA and OFDM
Multiuser transmission
A simple example of multiuser (MU) transmission
1 , 1
h
2 , 1
h
M
h
, 1
Base Station
1
s
Modulation
Coded bits
of user 1
2
s
Modulation
Coded bits
of user 2
1 , 2
h
2 , 2
h
M
h
, 2
Antenna 1
Antenna M
Demod
Channel
estimator
User 2
Demod
Channel
estimator
User 1
1
y
2
y
j
1
= :
1
1
∑
n=1
ℎ
1,n
+:
2
1
∑
n=1
ℎ
1,n
+.
1
, and j
2
= :
2
1
∑
n=1
ℎ
2,n
+:
1
1
∑
n=1
ℎ
2,n
+.
2
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 4
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Precoding classiﬁcation
In the socalled space division multiple access (SDMA), multiuser
diversity is the primary factor that increases signiﬁcantly the system
sumrate (throughput).
As a result, an appropriate multiuser encoding technique (at the BS)
is indispensable to attain the considerable sumrate gain in SDMA.
It is wellknown that dirty paper coding (DPC) is an optimal
multiuser encoding strategy that achieves the capacity limit of MU
broadcast (BC) channels but at the cost of extremely high
computation burden as the number of users is large.
Recent studies have introduced several suboptimal multiuser
encoding techniques with lower complexity (relative to DPC) that
can be categorized into:
nonlinear precoding such as: vector perturbation, Tomlinson
Harashima techniques
linear precoding such as: minimum mean squared error (MMSE),
zeroforcing.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 5
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Multiuser transmission techniques
Broadband communications
LTE (4G) system
Broadband communications
(high data rate and reliability)
Diversity
Time Freq.
Signal
Space
Multi
user
Space
Multipath channel
Modeling
CSI feedback
Analog Digital
Vector
quantization
g
Quasistatic Timevariant
BEMs AR
LBG
Grassmannian
Random
Scheduling Precoding
Exhaustive
search
Greed or iterative
search
Linear
methods
Nonlinear
methods
Codebook
based ones
MMSE BD DPC THP PU
2
RC
Random
user selection
VP
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 6
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
An example of linear precoding
1 , 1
h
2 , 1
h
M
h
, 1
Base Station
Feedback link of
channel state information (CSI)
1
s
X
X
X
1 , 1
w
Modulation
Coded bits
of user 1
2 , 1
w
M
w
, 1
2
s
X
X
X
1 , 2
w
Modulation
Coded bits
of user 2
2 , 2
w
M
w
, 2
1 , 2
h
2 , 2
h
M
h
, 2
Antenna 1
Antenna M
Demod
Channel
estimator
User 2
Demod
Channel
estimator
User 1
1
y
2
y
¸
1
= s
1
f
∑
r=1
r
1¸r
ℎ
1¸r
+s
2
f
∑
r=1
r
2¸r
ℎ
1¸r
+z
1
¸ and ¸
2
= s
2
f
∑
r=1
r
2¸r
ℎ
2¸r
+s
1
f
∑
r=1
r
1¸r
ℎ
2¸r
+z
2
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 7
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Interuser interference
The received signals at usern can be determined by
j
u
= :
u
1
∑
n=1
n
u,n
ℎ
u,n
+:
u
′
1
∑
n=1
n
u
′
,n
ℎ
u,n
+.
u
, n, n
′
∈ {1, 2},
(1)
where :
u
′
∑
1
n=1
n
u
′
,n
ℎ
u,n
is called as interuser interference
that would signiﬁcantly degrade the performance of the system.
Precoding design is to ﬁnd the weighting coeﬃcients {n
u,n
}
2
u=1
that satisfy the following condition
1
∑
n=1
n
u
′
,n
ℎ
u,n
= 0 with n, n
′
∈ {1, 2} (2)
to eliminate the interuser interference :
′
u
∑
1
n=1
n
u
′
,n
ℎ
u,n
.
The above technique is called as zeroforcing (ZF) precoding.
The problem of ﬁnding the weighting coeﬃcients {n
u,n
}
2
u=1
can be
easily solved by expressing received signals in a vector form.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 8
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Zero forcing (ZF) precoding formulation
In the presence of two users, the previous equations become
[
j
1
j
2
]
=
[
ℎ
1,1
. . . ℎ
1,1
ℎ
2,1
. . . ℎ
2,1
]
⎡
⎢
⎣
n
1,1
n
2,1
.
.
.
.
.
.
n
1,1
n
2,1
⎤
⎥
⎦
[
:
1
:
2
]
+
[
.
1
.
2
]
.
In the presence of l users, the received signal can be expressed by:
y = HWs +z, (3)
where y =
⎡
⎢
⎣
j
1
.
.
.
j
⎤
⎥
⎦
, H =
⎡
⎢
⎣
ℎ
1,1
. . . ℎ
1,
.
.
.
.
.
.
.
.
.
ℎ
,1
. . . ℎ
,
⎤
⎥
⎦
, s =
⎡
⎢
⎣
·
1
.
.
.
·
⎤
⎥
⎦
W =
⎡
⎢
⎣
n
1,1
. . . n
,1
.
.
. . . .
.
.
.
n
1,
. . . n
,
⎤
⎥
⎦
= [w
1
, . . . , w
] with
w
= [n
,1
, . . . , n
,
]
, and z = [.
1
, . . . , .
]
.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 9
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Zeroforcing precoding formulation (cont.)
To eliminate interuser interference, precoding matrix W can be
determined by
W = H
J
(
HH
J
)
−1
≜ H
†
(4)
so that
y = HWs +z = s +z. (5)
With precoding, the received signal can be written by
y = Hx +z, (6)
where x = [r
1
, . . . , r
1
]
T
= Ws are the transmitted signals in a
vector form at ` antennas in the base station.
Under the power constraint of 1
max
at the BS, one has
[
1
∑
n=1
∣r
n
∣
2
]
=
[
∥ x ∥
2
]
≤ 1
max
, (7)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 10
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Power allocation in ZF precoding
The power constraint (7) is equivalent to
I
∑
u=1
`
u
1
u
≤ 1
max
. (8)
where `
u
=
[
(
HH
J
)
−1
]
u,u
and :
u
=
√
1
u
:
u
After ZF precoding, the received signals at l users are given by
y =
⎡
⎢
⎣
j
1
.
.
.
j
I
⎤
⎥
⎦
=
⎡
⎢
⎣
√
1
1
:
1
.
.
.
√
1
I
:
I
⎤
⎥
⎦
+
⎡
⎢
⎣
.
1
.
.
.
.
I
⎤
⎥
⎦
(9)
Hence, the resultant sumrate of the multiuser system is
C = max
1
r
:
∑
!
r=1
X
r
1
r
≤1
max
I
∑
u=1
log
2
(1 +1
u
) (bps/Hz) (10)
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 11
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Power allocation in ZF precoding (cont.)
The optimal power allocation [1
u
, n ∈ {1, ..., l}] in (10) can be
easily determined by the following waterﬁlling process
1
u
= (j,`
u
− 1)
+
(11)
where r
+
denotes max(r, 0), and the water level j is chosen to
satisfy
I
∑
u=1
(j −`
u
)
+
= 1
max
. (12)
Given a set of selected users Ω = {1, ..., l}, the above precoding
process attempts to eliminate the interuser interference and
maximize the system sumrate.
The problem of how to perform user selection (ﬁnding the set
Ω = {1, ..., l}) with a reasonable complexity for maximizing the
system sumrate will be addressed in the next section.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 12
Outline
Introduction
Precoding
Scheduling (user selection)
Precoding classiﬁcation
An example of linear precoding
Power allocation in ZF precoding
Possible research problems
Precoding in LTE downlink transmissions
Data bits
of user 1
Channel
encoder
Interleaver
Layer
mapper
MQAM
mapper
MQAM
mapper
Precoding
OFDMA
modulator
OFDMA
modulator
Precoding matrix
generator
Recovered data bits Channel decoder
Channel
Estimator
OFDMA
Demodulator
BER evaluator
of user 1
OFDMA
Demodulator
Channel State
Information (CSI)
MIMO
demapper
Limited feedback
link
User 1
Base Station (BS)
Data bits
of user N
Channel
encoder
Interleaver
Layer
mapper
MQAM
mapper
MQAM
mapper
W
X
Y= W*X
BER evaluator
of user N
Multipath fading
channel
User N
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 13
Outline
Introduction
Precoding
Scheduling (user selection)
Exhaustive selection
Greedy selection
Exhaustive selection
Given a precoding technique, scheduling (user selection) is to ﬁnd a
set of users among all active users to maximize the system sumrate.
Obviously, the simple optimal method for user selection is exhaustive
search but its complexity is impractically high as the number of users
is large.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 14
Outline
Introduction
Precoding
Scheduling (user selection)
Exhaustive selection
Greedy selection
Greedy selection
Greedy user selection algorithm
1
Initialization: Θ
0
= {1, 2, ..., ·
I
} is the set of all available users’
indices
Ω
0
= {∅} is the set of selected users initially assigned to a null set.
j = 0 stands for the number of selected users, initially set to zero.
C
0
= 0 is the system sumrate of selected users, initially set to zero.
2
Repetition: Assuming that selecting user n in the set Θ
q
maximizes the resulting sumrate of the system called C
max
.
j = j + 1
If C
max
< C
−1
or j .
or j .
go to Step 3 otherwise do:
=
max
Ω
= Ω
−1
∪
{} (select one more user)
Θ
= Θ
−1
∖{} (ignore user in later consideration)
Go to Step 2.
3
Stop the user selection process and compute the ZF weighting
vectors based on the composite channel matrix of selected users.
Mobile Communications  Chapter 3: Physicallayer transmissions Section 3.6: Space Division Multiple Access (SDMA) 15
This action might not be possible to undo. Are you sure you want to continue?
Use one of your book credits to continue reading from where you left off, or restart the preview.