Multicarrier Communications

Lecture 1:

Introduction and OFDM Basics

Jian (Andrew) Zhang

[email protected]
Wireless Signal Processing Program
National ICT Australia

Canberra, Australia, 2007

MC 2007

c
Jian
(Andrew) Zhang

Course Description
Multicarrier systems are becoming foundation techniques in more and
more communications standards, ranging from xDSL systems over
copperline to OFDM in WLAN and next generation mobile systems.
This course covers a broad range of multicarrier systems, including
OFDM, MIMO-OFDM, generalized multicarrier systems and single
carrier block systems. Advanced receiver design will be investigated.
System design and resource optimization are studied for practical
applications.
The students will learn lots of key development and signal processing
techniques in multicarrier communications in a synthetic way, which is
not available yet in any textbook today.

MC 2007

c
Jian
(Andrew) Zhang

Course Plan
60 hours including 20 hours of lectures, 40 hours of assignments and
reading and preparation for lectures.
OFDM Systems, 6h
OFDM Basics (+2 hour)
Advanced Receiver Design (+2h)
RF Distortion and Compensation (2h)

Generalized OFDM Systems, 4h

Precoding OFDM Systems
Multicarrier+CDMA Systems
SC-FDE Systems
Asymmetric OFDM Systems

MIMO OFDM Systems, 2h

OFDMA Systems, 2h
Resource Optimization (AMC, bit loading and multiuser diversity)
2h
Practical Systems and Standards, 2h
Students Presentations, 2h
MC 2007

c
Jian
(Andrew) Zhang

Course Schedule
Course homepage:
http://users.rsise.anu.edu.au/ jian/Course mc.html
1
2
3
4
5
6
7
8
9
10

Mon, 13/08/07
Thu, 16/08/07
Mon, 20/08/07
Thu, 23/08/07
Mon, 27/08/07
Thu, 30/08/07
Mon, 17/09/07
Thu, 20/09/07
Mon, 08/10/07
Thu, 11/10/07

MC 2007

History and OFDM Basics

OFDM Receiver Design
RF Distortion and Compensation
Generalized Multicarrier Systems -1
Generalized Multicarrier Systems -2
MIMO-OFDM Systems
OFDMA Systems
Resource Optimization
Practical OFDM Systems and Standards
Students Presentation

c
Jian
(Andrew) Zhang

Textbooks

No textbook covers all the topics in the course. However, do

recommend the followings for starters and advanced readers
Juha Heiskala and John Terry, OFDM Wireless LANs: A
Theoretical and Practical Guide. (Good coverage, practical
design techniques although lots of mathematical errors)
L. Hanzo, et. al., OFDM and MC-CDMA for Broadband
Multi-user Communications, WLANs and Broadcasting. (For
advanced readers)

MC 2007

c
Jian
(Andrew) Zhang

Assessment
No formal exams. There will be three assessment tasks.
(60%) A couple of small assignments requiring answers to
exercises on Multicarrier systems, with MATLAB programming
and system simulation possibly involved.
(30%) One assignment consisting of either a current literature
review or a tutorial review of existing work related to multicarrier
communications. A 15-20 minute seminar will be arranged at the
end of the class based on the report. Students may be grouped
for this assignment depending on the number of enrolled
students.
(10%) Completing a conference paper based on small
extensions to current research. Potential topics will be given
during the lectures. This assignment is not compulsory, but will
be used as bonus for the assessment.
MC 2007

c
Jian
(Andrew) Zhang

Lecture 1

In this lecture, we will discuss

Some Background
OFDM history and applications
OFDM Principles
OFDM system structure
Spectrum shaping problem
Drawbacks of OFDM systems

MC 2007

c
Jian
(Andrew) Zhang

Challenges in Communications

Noise Problem
Intersymbol interference caused by channel dispersive
Limited Resource/System efficiency
These problems become more severe in broadband wireless
communications!

MC 2007

c
Jian
(Andrew) Zhang

Problems in Broadband Wireless Communication

Lower SNR of the received signal due to larger link loss

Intersymbol interference caused by severe/time varying channel
dispersive
Very limited radio spectrum
Interference from radio emission

MC 2007

c
Jian
(Andrew) Zhang

Wireless Channels
Two important parameters:
Coherent bandwidth [Delay spread]1 : statistical measure of
the range of frequencies over which the channel can be
consideredflat (equal gain and linear phase).
Coherent time [Doppler spread]1 : statistical measure of the
time duration over which the channel impulse response is
essentially invariant. Doppler shift fd = v cos /
Channel classification:
Large scale fading - path loss
small scale fading
Flat fading
Frequency selective fading - Multipath channels
Time varying (Doppler effect): fast/slow - Depending on the
velocity of the mobile (or the velocity of objects in the channel)
and the baseband signalling.
MC 2007

c
Jian
(Andrew) Zhang

10

Broadband Wireless Communication Solutions

For channel responses spanning tens or hundreds of symbols,

practical modulation and anti-multipath alternatives are:
1

SC modulation with receiver equalization done in the time

domain

OFDM (or frequency domain equalizer for generalized

multicarrier systems)

MC 2007

c
Jian
(Andrew) Zhang

11

Multi-Channel Communications

Synthesis/analysis filter bank configuration;

Theoretically, any orthogonal basis can be used to realize
multi-channel communications. So why OFDM is outstanding?
MC 2007

c
Jian
(Andrew) Zhang

12

OFDM History - Major milestones

Principle for multi-channel transmission over a bandlimited
channel proposed
R. W. Chang, Synthesis of band-limited orthogonal signals for
multichannel data transmission, Bell Syst. Tech. J., vol. 45, pp.
1775-1796, Dec. 1966.
Weinstein & Ebert proposed use of FFT and guard interval in 1971
Guard Interval (GI) and raised-cosine window being utilized
Suffering from the loss of orthogonality between channels
S. Weinstein and P. Ebert, Data Transmission by Frequency-Division
Multiplexing Using the Discrete Fourier Transform, IEEE Trans.
Comm., Volume: 19, Issue: 5, Oct 1971, Pages: 628-634
Cyclic Prefix (CP) used in 1980 - Orthogonality problem being solved
1985: Cimini described use of OFDM for mobile communications
L. Cimini, Analysis and Simulation of a Digital Mobile Channel Using
Orthogonal Frequency Division Multiplexing, IEEE Trans. Comm., Vol
33, Issue: 7, Jul 1985 Pages: 665-675
MC 2007

c
Jian
(Andrew) Zhang

13

OFDM History of Applications

1987: Alard & Lasalle: COFDM for digital broadcasting
September 1988: TH-CSF LER, first experimental Digital TV link
in OFDM, Paris area
1994: US patent 5282222, Method and apparatus for multiple
access between transceivers in wireless communications using
OFDM spread spectrum
1995: ETSI Digital Audio Broadcasting standard EUreka: first
OFDM based standard
1997: ETSI DVB-T standard
1999: IEEE 802.11a wireless LAN standard (Wi-Fi)
2002: IEEE 802.11g standard for wireless LAN

MC 2007

c
Jian
(Andrew) Zhang

14

OFDM History of Applications (con.)

2004: IEEE 802.16-2004 standard for wireless MAN (WiMAX)
2004: ETSI DVB-H standard
2004: Candidate for IEEE 802.15.3a standard for wireless PAN
(MB-OFDM)
2004: Candidate for IEEE 802.11n standard for next generation
wireless LAN
2005: Candidate for 3.75G mobile cellular standards (3GPP &
3GPP2 Long Term Evolution) named High Speed OFDM Packet
Access (HSOPA)
2005: Candidate for 4G standards (CJK)
Others: optical OFDM, DSL (variant of OFDM technique)

MC 2007

c
Jian
(Andrew) Zhang

15

OFDM Principles
Multi-channel techniques: Time-, Frequency-, Code- divisions
OFDM (Orthogonal Frequency Division Multiplexing) is a special
Frequency division technique- higher spectrum efficiency

MC 2007

c
Jian
(Andrew) Zhang

16

OFDM Parameters
The whole bandwidth B is divided into N subcarriers, with
subcarrier interval f = B/N
n-th Subcarrier signal: exp(j2nft), n = 0, 1, , N 1
Windowing operator (g(t)) changes the spectrum shape of each
subcarrier
For rectangle windowing, the spectrum becomes a sinc function

Symbol duration T = 1/f = N/B

MC 2007

c
Jian
(Andrew) Zhang

17

OFDM Signal

MC 2007

c
Jian
(Andrew) Zhang

18

Mathematical Representation
Orthogonality: Set of subcarriers fn (t) = exp(j2nft),
n = 0, 1, , N 1, 0 t T is orthogonal if
Z T
 0, n 6= m
fn (t)fm (t)dt =
N, n = m
0

(1)

An OFDM symbol carriers N modulated symbols

X = (X0 , X1 , , XN1 )
The time domain signal for the OFDM symbol is
N1
1 X
Xn exp(j2nft)g(t),
x(t) =
N n=0

(2)

The bandpass representation of OFDM signals is similar to

single carrier signals
xbp (t) = Re{x(t)ej2fc t }
MC 2007

c
Jian
(Andrew) Zhang

(3)
19

Discrete Time Representation

The discrete representation can be obtained by appropriate
sampling.
As the bandwidth of an OFDM signal is B = Nf , the signal can
be completely determined by its samples if the sampling time
1
Ts = B1 = Nf
N1
1 X
xk =
Xn exp(j2nk /N),
N n=0

k = 0, 1, , N 1

(4)

This is exactly the N-point discrete Fourier transform (DFT) of

the input data X

MC 2007

c
Jian
(Andrew) Zhang

20

OFDM over Selective Fading Channels

In a fading multipath environment, inter-carrier interference (ICI)
and inter-symbol interference (ISI) will be introduced.
To solve this problem, every OFDM block is extended by a guard
interval TG , TG > Td where Td is the maximum multipath delay
spread.
The most common guarding interval is cyclic prefix (CP),
zero-padding (ZP) is also applicable.
This guarding interval recovers the orthogonality of subcarriers,
and thus removes ICI and ISI

MC 2007

c
Jian
(Andrew) Zhang

21

Channel Viewpoint

MC 2007

c
Jian
(Andrew) Zhang

22

Interpretation from Convolution Theorem

Convolution Theorem: Convolution in the time domain is

multiplication in the Fourier domain for continuous systems. But
with the Discrete Fourier samples, multiplication in the frequency
domain is cyclic convolution in the time domain.
Equalization in frequency domain is simple given a linear
channel
By adding a CP or ZP, weve made linear convolution look cyclic.
Referring to a Discrete Signal Processing textbook, we know CP
and ZP correspond to overlap-and-save and overlap-and-add
algorithms, respectively.

MC 2007

c
Jian
(Andrew) Zhang

23

Realization of Modern OFDM Transmitter

Time domain sample of OFDM signal is approximately Gaussian

distributed.
Excise:P
Derive the form of the analogue OFDM signal
N1
x(t) = i=0
Xi ej2ift .
(Hint: Recall the sampling theory and the reconstruction technology)

MC 2007

c
Jian
(Andrew) Zhang

24

Signal Model in Matrix form

Received signal over frequency selective channels

y = HNN x

h0
hN1
h1
h0

..
..
= .
.

hN2 hN3
hN1 hN2

h2
h1
h3
h2

..
.. FH x
..

. .
.

h0 hN1
h1
h0

where hi = 0, for i [L, N 1] given a channel delay spread L.

MC 2007

c
Jian
(Andrew) Zhang

25

Parameter Configuration

Subcarrier Interval f : f fCB to ensure each subcarrier sees

a flat-fading channel, where fCB 1/d is the coherent
bandwidth
Number of subcarriers N: Firstly, N = B/f > B/fCB ; Secondly,
larger N leads to higher spectrum efficiency.
For a system with fixed bandwidth B, the OFDM symbol period
T = 1/f = N/B;
System overhead (or spectrum efficiency) is defined as
TG
1
T +TG = 1+N/(BTG ) ;
For a fixed delay spread, the larger N is, the lower the system
overhead (or the higher the spectrum efficiency).

MC 2007

c
Jian
(Andrew) Zhang

26

Spectrum Shaping/Pulse Shaping

Problems:
The spectrum of a rectangular window is a sinc function which is
infinite.
The sidelobe of OFDM signals should decay fast to improve
spectrum efficiency and avoid interference to adjoint bands.

Solutions:
1
2

MC 2007

Use guarding tones (Zero tones) at the edge of the band.

For a fixed bandwidth B, the value of N (number of subcarriers)
affects the sidelobes. Sidelobes decay faster for a larger N.
Use better windowing functions (Pulse shaping). - Tradeoff
between tails in time and frequency domain.

c
Jian
(Andrew) Zhang

27

Spectrum Shaping/Pulse Shaping (con.)

Power spectrum density of OFDM signals when a raised cosine
function is used. Left: N = 96, right: N = 1536

MC 2007

c
Jian
(Andrew) Zhang

28

Modulations

Coherent modulations: xPSK, xQAM, etc. (a coherent detector is

used)
- Require channel estimation
Differential modulations over either time samples (Doppler
problem) or subcarriers (Tone correlation and Phase rotation
problems)
No need for additional pilots, training data etc for channel
estimation
3dB performance loss
reduced data rate

MC 2007

c
Jian
(Andrew) Zhang

29

OFDM System Structure

MC 2007

c
Jian
(Andrew) Zhang

30

Drawbacks of OFDM Systems

High sensitivity to carrier frequency offsets (CFO)

High peak-to-average power ratios (PAPR) problem
Limited Frequency Diversity

MC 2007

c
Jian
(Andrew) Zhang

31

CFO Sensitivity
Received signal with a frequency offset fo
rk =

N1
X

Hm Xm ej2k(m+)/N + nk , k [0, N 1]

(5)

m=0

where  = fo /f is the normalized CFO.

After transferring to frequency domain
R` =

N1
X
k=0

rk ej2`k /N = X` H`

sin()
ej(N1)/N + I` + Z`
N sin(/N)

P
sin()
j(N1)/N ej(l`)/N is
where I` = N1
l=0,l6=` Xl Hl N sin((l`+/N)) e
the ICI term.
Different to single carrier systems, CFO causes ICI in OFDM
systems.
MC 2007

c
Jian
(Andrew) Zhang

32

CFO Sensitivity (cont.)

MC 2007

c
Jian
(Andrew) Zhang

33

CFO Sensitivity - SNR Degradation

For relatively small frequency errors, the degradation in dB can be
approximated by
10
SNRloss =
(Ts )2 Es /N0 dB
(6)
3 ln 10

MC 2007

c
Jian
(Andrew) Zhang

34

PAPR Problem

OFDM suffers from high PAPR defined as,

=

maxt[0,T ] |x(t)|2
maxt[0,T ] |x(t)|2
=
Pav
E{|x(t)|2 }

(7)

Large PAPR implies larger dynamic range of signals and causes

problems in power amplifier, ADC and so on.
PAPR can be characterized by CCDF.

MC 2007

c
Jian
(Andrew) Zhang

35

CCDF of PAPR
Complementary cumulative distribution function (CCDF): Probability
that the PAPR of an OFDM symbol exceeds a given threshold

MC 2007

c
Jian
(Andrew) Zhang

36

Limited Frequency Diversity

OFDM converts frequency selective channels to flat fading

channels, however, (rayleigh) fading is introduced to each
subcarrier.
Multipath energy is collected, however, not in a constructive way.
Each subcarrier only sees a portion of the whole channel energy.
Maximal degree of diversity equals to the number of multipath
signals given independent distributed multipaths. According to
the Parsavels theorem, it is equivalent to assign one symbol to
all subcarriers so that all energy can be collected.

MC 2007

c
Jian
(Andrew) Zhang

37

Diversity Improvement

Exploiting the frequency diversity between different subcarriers by

Cross-subcarrier Modulation
Frequency/Time spreading
Precoding (To be discussed later)
Interleaving and Coding across multiple subcarriers

MC 2007

c
Jian
(Andrew) Zhang

38

Diversity Improvement - Cross-subcarrier Modulation

Example: Dual Carrier Modulation (DCM, proposed for Multiband
OFDM systems)
Basic QPSK approach
Map 2 interleaved bits onto a QPSK constellation and then map
symbol onto the appropriate IFFT tone.
When there is a deep fade on the tone, the system has to rely
solely on strength of error correction code to recover lost
information.

Basic idea behind DCM:

Map 4 interleaved bits onto two 16-point symbols using two fixed
but different mappings. This yields a 16-QAM-like constellation.
Map the resulting two 16-point symbols onto two different IFFT
tones separated by 50 tones.
Even if there is a deep fade on one of the two tones, the 4 bits of
information can be recovered using simple (ML) detection
scheme.
MC 2007

c
Jian
(Andrew) Zhang

39

Diversity Improvement - Time/Frequency Spreading

Time or Frequency spreading is achieved by putting the same

symbol over multiple (consecutive) OFDM symbols or multiple
(largely spaced) subcarriers;
Spreading code can also be used to spread the symbol first,
which leads to the concepts Multicarrier CDMA etc.
Diversity is achieved at the cost of reduced data rate.

MC 2007

c
Jian
(Andrew) Zhang

40

Coded OFDM
FEC coding in OFDM has two roles: Achieving 1) frequency diversity
and 2). Error correction
Convolution coding is the most widely used code.
hard vs soft decoding
Interleaving is important for improving the decoding performance
Avoid multiple adjoint fading subcarriers
In a fading channel, errors tend to bunch together and most of
the errors will occur around the fade. Interleaver spreads out the
errors so that the code can handle them.
Diversity effect obtained by coding is actually realized only when
interleaving is applied
Adjoint subcarriers are (highly) correlated
Achieves temporal diversity when interleaving over multiple OFDM
symbols

Question: Order of the diversity achievable by coding and

interleaving?
MC 2007

c
Jian
(Andrew) Zhang

41

Homework

Derive the OFDM signal models with Cyclic Prefix and Zero padding,
by following the overlap-and-save and overlap-and-sum processes
respectively. Discuss the advantages and disadvantages of the two
systems.

MC 2007

c
Jian
(Andrew) Zhang

42

What will be covered in Lecture 2

Receiver Design for OFDM Systems

Packet Detection
(Advanced) Synchronization Algorithms
(Advanced) CFO Estimation and Compensation Algorithms
(Advanced) Channel Estimation and Tracking Algorithms
(Advanced) Equalization Algorithms

MC 2007

c
Jian
(Andrew) Zhang

43