You are on page 1of 63

R&I R 7/2008

Per Hjalmar Lehne, Frode Bøhagen

OFDM(A) for wireless


communication
OFDM(A) for wireless communication

R&I Research Report R 7/2008


Title OFDM(A) for wireless communication

Author(s) Per Hjalmar Lehne, Frode Bøhagen


ISBN / ISSN 82-423-0614-1 / 1500-2616
Security group OPEN
Date 2008.04.25

Abstract
This report is a tutorial on Orthogonal Frequency Division Multiplex (OFDM) and
Orthogonal Frequency Division Multiple Access (OFDMA). OFDMA is the major
transmission and access technology for future mobile broadband systems like Mobile
WiMAX and 3GPP Long Term Evolution (LTE).
One of the key features of OFDM and OFDMA is the ability to handle multipath
propagation without complex receivers. The use of simple and cost-efficient Fast Fourier
Transform (FFT) techniques makes it easily scalable with respect to bandwidth. The
main drawback is that the signal has high amplitude variability, high so-called Peak-to-
Average-Power Ratio (PAPR), which typically reduces the efficiency of the transmitter
power amplifiers. The two major standards for mobile broadband, namely Mobile WiMAX
and 3GPP LTE are very similar in the design targets and the solutions are comparable.
From a technical and performance point of view they seem to be quite equal.
Radio planning of OFDMA networks is more equal to GSM planning than WCDMA,
because intra-cell interference is basically eliminated due to the orthogonality property.
Scalable OFDMA opens up new ideas on frequency reuse, in that the reuse factor can be
between 1 and 3 (or more) on a fractional basis. Furthermore, different cell capacities
can be tuned cooperatively, e.g. by moving capacity from one cell to another dependent
on time of day.

Keywords
OFDM, OFDMA, Wireless, Mobile, 4G, WiMAX, E-UTRA, LTE

Telenor R&I R 7/2008


OFDM(A) for wireless communication

© Telenor ASA 2008.04.25

All rights reserved. No part of this publication may be reproduced or utilized in


any form or by any means, electronic or mechanical, including photocopying,
recording, or by any information storage and retrieval system, without
permission in writing from the publisher.

Telenor R&I R 7/2008


OFDM(A) for wireless communication

Preface
This report is the result of a technology study performed in 2007. The aim of
the study has been to develop the competence in Telenor R&I on the multiple
access scheme OFDMA. OFDMA is the transmission and access technology to be
used in future mobile broadband systems like Mobile WiMAX and 3GPP Long
Term Evolution.
The study was finalized with an internal workshop arranged on 23 January
2008.

Telenor R&I R 7/2008


OFDM(A) for wireless communication

Contents

1 Introduction.......................................................... 1

2 Introduction to OFDM ........................................... 2


2.1 OFDM basic concept ................................................................. 3
2.1.1 Orthogonality principle ............................................................. 4
2.1.2 Instantaneous power variations ................................................. 7
2.1.3 Wireless channel influence ...................................................... 10
2.1.4 Signal detection and demodulation ........................................... 13
2.2 Choosing the OFDM parameters ............................................... 13
2.2.1 Sub-carrier spacing and symbol time ........................................ 14
2.2.2 Number of sub-carriers........................................................... 15
2.2.3 Cyclic prefix length ................................................................ 15
2.2.4 Pulse shaping and windowing functions ..................................... 16

3 Multi-user communications – OFDMA.................. 18


3.1 Sub-carrier allocation techniques ............................................. 18
3.1.1 Consecutive (localized) frequency mapping ............................... 18
3.1.2 Distributed frequency mapping ................................................ 19
3.1.3 Channel dependent scheduling ................................................ 20
3.2 Synchronization aspects ......................................................... 20
3.3 DFT-spread OFDMA (SC-FDMA) ............................................... 21

4 Wireless standards based on OFDM(A) ............... 24


4.1 3GPP Evolved UTRA (Long Term Evolution)................................ 24
4.1.1 Radio resource definitions ....................................................... 25
4.1.2 Modulation and coding............................................................ 28
4.1.3 Multi-antenna support ............................................................ 28
4.1.4 Downlink scheduling and reference signals ................................ 29
4.1.5 Uplink scheduling and reference signals .................................... 30
4.2 Mobile WiMAX ....................................................................... 31
4.2.1 Radio resource definitions – sub-channelization.......................... 32
4.2.2 Diversity permutations ........................................................... 33
4.2.3 Contiguous permutation ......................................................... 36
4.2.4 Permutation schemes summary ............................................... 37
4.2.5 Modulation and coding............................................................ 37
4.2.6 Multi-antenna support ............................................................ 38
4.2.7 Frame Structure .................................................................... 38
4.3 E-UTRA vs. Mobile WiMAX – summary ...................................... 40
4.4 Other OFDM and OFDMA based standards ................................. 41
4.4.1 Mobile WiMAX Release 2 – IEEE 802.16m.................................. 41
4.4.2 3GPP2 Ultra Mobile Broadband (UMB) ....................................... 41
4.4.3 Wi-Fi and WLAN 802.11a-g-n .................................................. 42
4.4.4 WPAN 802.15.3a Multi Band OFDM........................................... 42
4.4.5 Digital terrestrial video broadcast – DVB-T/H ............................. 43

5 OFDMA radio planning......................................... 44


5.1 Frequency reuse .................................................................... 45
5.1.1 Frequency reuse 1 ................................................................. 46
5.1.2 Frequency reuse 3 ................................................................. 46

Telenor R&I R 7/2008


OFDM(A) for wireless communication

5.1.3 Fractional frequency reuse ...................................................... 47


5.2 OFDMA link budgets ............................................................... 48

6 Conclusions ......................................................... 50

References .................................................................. 52

Annex 1. Abbreviations ............................................... 54

Telenor R&I R 7/2008


OFDM(A) for wireless communication

1 Introduction
Orthogonal Frequency Division Multiplex (OFDM) is a multi carrier transmission
technology which was first described by R. W. Chang in 1966 at Bell Labs. The
first patent was granted in 1970 (US patent 3488445). Later, M. Fattouche and
H. Zaghloul described how the OFDM concept can be used to provide multiple
access between different transceivers (US patent 5282222, granted in 1994).
This was the first description of Orthogonal Frequency Division Multiple Access
(OFDMA).
The first OFDM based wireless standard was probably the Eureka Digital Audio
Broadcast (DAB) standard for audio broadcasting which was released in 1995.
Two years later, in 1997, the standard for terrestrial digital television, i.e. DVB-
T, was published.
OFDM and OFDMA is now becoming the major transmission and access
technology for future mobile broadband systems. Mobile WiMAX is already
available, and 3GPP has left WCDMA in favour of OFDMA for the next generation
standard Evolved UTRA, often referred to as Long Term Evolution (LTE).
This report is a tutorial on OFDM and OFDMA, and is organized in four chapters.
First, in Chapter 2 we introduce OFDM transmission, and define and discuss all
major parameters. In Chapter 3 we explain how the OFDM concept can be
augmented to comprise multiple access as OFDMA. Chapter 4 deals with the
major standards using OFDMA. The weight has been put on Mobile WiMAX and
3GPP E-UTRA, but a brief description of other OFDM(A) based standards is also
included. Finally, in Chapter 5 radio planning for an OFDMA network is
discussed and compared with current knowledge from 2G (GSM) and 3G
(WCDMA) planning. Some examples of link budget calculations are also
presented.

Telenor R&I R 7/2008 - 1


OFDM(A) for wireless communication

2 Introduction to OFDM
Orthogonal Frequency Division Multiplexing (OFDM) is a multi-carrier
modulation scheme that transmits data over a number of orthogonal sub-
carriers. A conventional transmission uses only a single carrier modulated with
all the data to be sent. OFDM breaks the data to be sent into small chunks,
allocating each sub-data stream to a sub-carrier and the data is sent in parallel
orthogonal sub-carriers. As illustrated in Figure 1, this can be compared with a
transport company utilizing several smaller trucks (multi-carrier) instead of one
large truck (single carrier).

Single Carrier

Multi Carrier

Figure 1 Single carrier vs. multi-carrier transmission


OFDM is actually a special case of Frequency Division Multiplexing (FDM). In
general, for FDM, there is no special relationship between the carrier
frequencies, f1, f2 and f3. Guard bands have to be inserted to avoid Adjacent
Channel Interference (ACI). For OFDM on the other hand, there must be a strict
relation between the frequency of the sub-carriers, i.e. fn = f1 + n⋅Δf where Δf =
1/TU and TU is the symbol time. Carriers are orthogonal to each other and can
be packed tight as shown in Figure 2.

2 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Channel
Individual channels
bandwidth Guard band

FDM

f1 f2 f3 Frequency
Channel
Individual sub-channels
bandwidth

Bandwidth Bandwidth
saving saving
OFDM

f1 f 2 f3 Frequency

Figure 2 FDM vs. OFDM


Splitting the channel into narrowband channels enables significant simplification
of equalizer design in multipath environments. Flexible bandwidths are enabled
through scalable number of sub-carriers.
Effective implementation is further possible by applying the Fast Fourier
Transform (FFT). Dividing the channel into parallel narrowband sub-channels
makes coding over the frequency band possible (COFDM). Moreover, it is
possible to exploit both time and frequency domain variations, i.e. time and
frequency domain adaptation.

2.1 OFDM basic concept


A baseband OFDM transmission model is shown in Figure 3. It basically consists
of a transmitter (modulator, multiplexer and transmitter), the wireless channel,
and a receiver (demodulator).

Telenor R&I R 7/2008 - 3


OFDM(A) for wireless communication

channel

Figure 3 Basic baseband OFDM transmission model


In Figure 3, a bank of modulators and correlators is used to describe the basic
principles of OFDM modulation and demodulation. This is not practically
feasible, and the specific choice of sub-carrier spacing being equal to the per
carrier symbol rate 1/TU makes a simple and low complexity implementation
using Fast Fourier Transform (FFT) processing possible as shown in Figure 4.
For more details on the FFT implementation we refer to [Nee00].

channel

Figure 4 Transmitter and receiver by using FFT processing


In the consecutive subsections we will discuss the most important properties of
the OFDM transmitter (Section 2.1.1), OFDM signal properties (Section 2.1.2),
the OFDM channel (Section 2.1.3), and the OFDM receiver (Section 2.1.4)

2.1.1 Orthogonality principle


The essential property of the OFDM signal is the orthogonality between the sub-
carriers. Orthogonal means “perpendicular”, or at “right angle”.
Two functions, xq(t) and xk(t), are orthogonal over an interval [a, b] if the inner
product between them is zero for all q and k, except for the case that q = k, i.e.
when xq(t) and xk(t) are the same function. Mathematically this can be written
as:
bU
⎧1, k = q
xq , xk = ∫x
a
q (t ) ⋅ x k (t ) dt =⎨
⎩0, k ≠ q
(1)

4 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

If we look at receiver branch k in Figure 3, the output of the integrator can be


expressed as follows:
TU

TU N C −1

∫ ∫ ∑
⎜ a q ⋅ e j 2πqΔft ⎟ ⋅ e − j 2πkΔft dt
1 − j 2πkΔft 1
r (t ) ⋅ e dt =
TU TU ⎜ ⎟
0 0⎝ q =0 ⎠
(2)
j 2π (q − k ) t
N C −1 TU 1
aq ⎧a , k = q
= ∑T ∫e
q =0 U 0
TU
dt = ⎨ k
⎩0 , k ≠ q

In this example we assume no noise or multipath degradation of the signal


(ideal channel). We see that the last integral satisfies the orthogonality
definition. Consequently, the harmonic exponential functions (sine wave
carriers) with frequency separation Δf = 1/TU are orthogonal.
This property gives optimum spectrum utilization and makes it possible to
separate the sub-carriers in the receiver. Figure 5 is an attempt to illustrate
how the orthogonality works in the time domain, if e.g. xq is the received sub-
carrier, and xk represents the local oscillator as shown in Figure 3. When
integrating received power over one symbol period, TU, the output of the
correlators is zero for any combination, except when k = q.
When the sub-carriers are modulated with a rectangular pulse the sub-carrier
spectrum becomes as shown in Figure 6. This implies that in the frequency
domain, the power of sub-carriers approaches zero at the centre frequency of
the neighbouring sub-carriers as shown in Figure 7. The figure shows the power
spectrum of the individual sub-carriers of an OFDM spectrum.

Telenor R&I R 7/2008 - 5


OFDM(A) for wireless communication

xq, q=6 xq . xq*


1.5 1.5

1 1

0.5 0.5

0 0

-0.5 -0.5

-1 -1

-1.5 -1.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

xk, k=7 xq . xk*


1.5 1.5

1 1

0.5 0.5

0 0

-0.5 -0.5

-1 -1

-1.5 -1.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Figure 5 Orthogonality principle in the time domain. Leftmost graphs show the
input signal shapes over one symbol period. The rightmost graphs show the
integrand in the case of equal signal (upper) and orthogonal signal (lower). It is
easily seen that the total area under the curve when q = k is positive, while it
sums up to zero over one symbol period when the frequencies are different and
spaced 1/TU apart

Frequency domain

Time domain

Figure 6 Time and frequency domain representation of the baseband signal

6 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 7 Power spectrum of orthogonal sub-carriers


A real OFDM power spectrum will not look like the one in Figure 7, because a
power based measurement cannot distinguish the separate sub-carriers. Figure
8 shows a field measurement of a mobile WiMAX spectrum with a bandwidth of
5 MHz. The signal consists of 360 active sub-carriers. The FFT-size is 512. The
amplitude variations are due to frequency selective fading caused by multipath
transmissions. This is treated in section 2.1.3.

Figure 8 Field measurement of an OFDM spectrum (WiMAX) in a LOS


environment

2.1.2 Instantaneous power variations


An OFDM signal consists of a number of independently modulated symbols.

Telenor R&I R 7/2008 - 7


OFDM(A) for wireless communication

N c −1
x(t) = ∑a
k =0
k ⋅ e ( j2 πkΔf t ) (3)

Adding carriers with different frequencies and modulation gives large amplitude
variations. This is the Peak Power Problem of OFDM. It can be shown that the
maximum value of the Peak to Average Power Ratio (PAPR) is the same as the
number of sub-carriers:
PAPRmax = N C (4)

Figure 9 shows a simulated waveform of an OFDM-signal with 8 sub-carriers


and BPSK modulation. The expanded part of the graph shows that the
amplitude variations are large and that the maximum value of the sum can be
as high as 8.

Figure 9 Amplitude variations in an OFDM signal. Example is 8 sub-carriers


and BPSK modulation
A large PAPR is negative for the power amplifier efficiency since a backoff is
needed to avoid amplitude distortion and the following harmonic frequencies.
Figure 10 shows a typical input-output characteristic of a power amplifier. In
order to avoid or limit signal distortion input signals must be kept below the
non-linear area. The consequence is that the amplifier is not fully utilized.

8 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

AM/AM characteristic

POUT

OBO

IBO

Average Peak
PIN

Figure 10 Typical input-output characteristics of a power amplifier showing the


relation between output backoff (OBO) and input backoff (IBO)
Different measures to avoid large PAPRs are used and are classified in three
major techniques:
• Signal distortion techniques
o Clipping (signal distortion, out of band radiation, window)
o Peak windowing (reduce the out of band radiation)
o Peak cancellation (linear, can avoid out of band radiation)
• Coding techniques
o Special FEC codes that excludes OFDM symbols with a large PAPR
(decreasing the desired PAPR this way may decrease achievable
coding gain).
o Tone reservation is another coding technique where a subset of
OFDM sub-carriers are not used for data transmission but instead
are modulated to suppress the largest peaks of the overall OFDM
signal.
o Pre-filtering or pre-coding, linear processing applied to the OFDM-
symbols before OFDM modulation.
• Scrambling techniques
o Different scrambling sequences are applied, and the one resulting
in the smallest PAPR is chosen

Telenor R&I R 7/2008 - 9


OFDM(A) for wireless communication

PAPR reduction techniques applied to different wireless standards must partly


be described in the standards, as e.g. coding and scrambling. Signal distortion
techniques however can be a pure implementation aspect, but the use of such
usually means a reduced BER performance in the receiver.
No special technique seems to be specified in WiMAX, but in E-UTRA the uplink
transmission is based on a linear pre-coding technique called DFT-spread
OFDM, which effectively reduces the PAPR with 2-3 dB [Dahl07] [Myu06b]. This
is detailed a bit more in section 3.3.

2.1.3 Wireless channel influence


A wireless channel introduces impairments to the signal, mainly due to
multipath transmission as shown in Figure 11.

Diffracted and Refracted Path


[α k ,τ k ]

[α 0 ,τ 0 ]
LOS Path
[α1 ,τ 1 ]

Reflected Path
K −1
h (t ) = ∑ α δ (t − τ
k =0
k k)

Figure 11 The multipath channel


Figure 12 (left) shows a measured channel impulse response, both the direct
path signal and the reflected (echoed) signals. In addition, Figure 12 (right)
illustrates a simplification of the impulse response.

Amplitude
[α]

τ τ τ2 Time
[τ]

Figure 12 Multipath channel impulse response. Left: measured. Right:


simplified two ray model

10 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

The consequence of multipath propagation is that time dispersion is introduced


in the signal. Figure 13 illustrates the effect in the time domain and how a cyclic
prefix (CP) may be used to mitigate the effect. Time-adjacent symbols start to
overlap and inter-symbol-interference (ISI) is introduced (Figure 13a). In
Figure 13b the symbol is prolonged by adding a guard time between the
symbols, but just adding an “empty” guard time destroys the orthogonality and
introduces inter-carrier interference (ICI). In order to avoid this and thus to
maintain the orthogonality, the prefix is made cyclic by taking a copy of the last
part of the symbol and put it in front (Figure 13c). The prefix time must be
longer than the longest excess delay which can occur in the channel.

a) Multipath introduces inter-symbol-interference (ISI)

b) Prefix is added to avoid ISI

TC T

c) The prefix is made cyclic to avoid inter-carrier-interference (ICI)


(maintain orthogonality)

Figure 13 Time dispersion due to multipath propagation and the method of


cyclic prefix insertion
Due to the large bandwidth of an OFDM signal, the multipath effect is frequency
dependent and results in frequency selective fading as shown in Figure 14.

Telenor R&I R 7/2008 - 11


OFDM(A) for wireless communication

Figure 14 Frequency selective fading of an OFDM signal


Figure 15 shows a measured OFDM spectrum (mobile WiMAX, 5 MHz) in an
NLOS scenario. A strong multipath component is causing severe frequency
selective fading.

Figure 15 Measured OFDM-spectrum in an NLOS scenario showing how


multipath reflections cause severe frequency selective fading
Coding should be performed over several uncorrelated carriers to utilize the
frequency diversity (frequency interleaving). It is equivalent to coding in time
domain to achieve time diversity (time interleaving). Often we will see the term
Coded OFDM (COFDM) used in this respect. Especially the digital terrestrial
broadcast standards DVB-T and DVB-H use the term COFDM.
In reality, any OFDM and OFDMA based standard uses frequency domain coding
to exploit the frequency diversity.

12 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

2.1.4 Signal detection and demodulation


Time and frequency properties of the OFDM signal are tightly coupled, and both
must be properly recovered in the receiver.
When performing timing recovery, the sum of the timing error (Δτ and the
maximum excess delay (τmax), should be within the prefix time. Then, the only
effect is a phase rotation that increases with increasing distance from the
carrier (centre frequency). However, a carrier frequency synchronization error
will introduce phase rotation, amplitude reduction and ICI. Frequency offsets of
up to 2% of Δf are negligible and even offsets of 5-10% can be tolerated in
many situations.
To help the receiver in performing these tasks, pilot or reference symbols are
inserted in the OFDM signal. These are known signal sequences spread out in
time and frequency which the receiver can use to recover both time and
frequency references. Additionally, some of these are usually tailored to enable
reliable channel estimations. The simplest way is to reserve a number of sub-
carriers to carry reference signals as shown in Figure 16. In this case, the sub-
carriers are often called pilot signals. More common is to reserve time/
frequency symbols as shown in Figure 17. In this case, they are not transmitted
continuously, neither in time nor frequency.

Pilot carriers /reference signals


Data carriers

Frequency/subcarrier

Figure 16 Pilot carriers in an OFDM signal

Figure 17 Time-frequency grid with reference symbols

2.2 Choosing the OFDM parameters


We have now defined and described the important properties of the OFDM-
signal, and we shall now discuss the criteria for designing a system based on
OFDM.
Some of the key parameters for the OFDM signal are:

Telenor R&I R 7/2008 - 13


OFDM(A) for wireless communication

• Sub carrier spacing (Δf) and Symbol time (TU)


• Number of sub-carriers(NC)
• The cyclic prefix length (TCP)
• Pulse shape, or windowing function.

2.2.1 Sub-carrier spacing and symbol time


As we have seen, symbol time and sub-carrier spacing are inverse entities. Two
factors determine the selection of the OFDM sub-carrier spacing, Δf:
• It is preferable to have as small SC spacing as possible, since this gives
a long symbol period and consequently the relative cyclic prefix
overhead will be minimized.
• Too small SC spacing increases the sensitivity to Doppler spread and
different kinds of frequency inaccuracies.
Doppler spread is typical for mobile systems and the amount is directly limited
by the relative velocity between the mobile and the base station. Frequency
variations due to Doppler spread leads to losing the orthogonality in the
receiver and ICI occurs. Dependent on the targeted mobility for the system and
the allowed amount of ICI, the sub-carrier distance can be selected. According
to [Dahl07] a normalized Doppler spread (fDoppler/Δf) of 10 % leads to a Signal
to Interference Ratio (SIR) of 18 dB.
If we want to design a system in the 2 GHz band (λ = 15 cm) supporting
terminal mobility up to v = 200 km/h (= 55.5 m/s), we get a maximum
Doppler frequency of fDoppler = v/λ = 370 Hz. If we require an SIR of 30 dB, we
can allow approximately 2.5 % normalized Doppler spread. This gives us a sub-
carrier spacing of 14.8 kHz, which is quite close to the choice made in E-UTRA.
This is a very simple example, and in a real design situation there are other
factors which must be taken into account as well. Some of the effects of varying
the sub-carrier spacing are shown in Figure 18.

14 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 18 Effects of changing the sub-carrier spacing

2.2.2 Number of sub-carriers


When the sub-carrier spacing is determined, the available bandwidth
determines the number of sub-carriers, NC, to be used. The basic bandwidth of
an OFDM signal is NC⋅Δf. However, the frequency spectrum of a basic OFDM
signal falls off very slowly outside the basic bandwidth, and lower and upper
guard bands must be inserted. By applying pulse shaping or windowing, the
out-of-band emissions can be reduced. In practice, typically 10 % guard band is
needed [Dahl07], implying that of a spectrum allocation of 5 MHz, 4.5 MHz is
usable. For the E-UTRA with 15 kHz sub-carrier spacing this means that
approximately 300 sub-carriers can be exploited in a 5 MHz channel. For WiMAX
with 10.94 kHz sub-carrier spacing, 400 sub-carriers can be used.

2.2.3 Cyclic prefix length


The cyclic prefix (CP) length, Tcp, should cover the maximum length of the time
dispersion. Increasing Tcp without decreasing Δf implies increased overhead in
power and bandwidth.
Too short CP gives ISI. On the other hand, increasing the relative length of the
CP leads to increased power loss. Thus, choosing Tcp is a trade-off between
power loss and time dispersion.
The maximum time dispersion experienced depends on the radio channel and
its multipath properties as discussed earlier in section 2.1.3. If for example we
target our system for ranges up to 10 km, an excess delay of reflected and
diffracted signals corresponding to Δs = 2 km is easily encountered. The excess
delay is then Δt = Δs/c = 6.7 μs, which could be used as the value for the CP.

Telenor R&I R 7/2008 - 15


OFDM(A) for wireless communication

In a real channel, the power delay profile (PDP) often follows an exponential
decay function, and our design criteria will be the amount of remaining energy
in the tail of the PDP which is tolerated to interfere with the next symbol (see
Figure 19).

Received instantanous power

Remaining power
leaking into next
symbol period

Excess delay
Covered by the CP

Figure 19 Channel power delay profile (PDP) and cyclic prefix (CP)
dimensioning
The PDP is dependent on the environment, thus the system must be designed
bearing this in mind. Multipath channel measurements on 900 MHz and 1.7 GHz
done by Telenor R&I in 1990 and 1991 [Løv92] [Ræk95] showed that for urban
areas (downtown Oslo) the delay spread (DS) in 90 % of the cases were below
0.7 μs, which should be well below the chosen CP lengths of E-UTRA and Mobile
WiMAX (approximately 5 and 10 μs, resp). However for different rural
environments, the DS values could be between 5 and 20 μs in 90 % of the
cases. These kinds of environments may be more challenging.

2.2.4 Pulse shaping and windowing functions


As mentioned when discussing the number of sub-carriers to use, the spectrum
of a basic OFDM signal falls off very slowly outside the basic bandwidth.
Especially it falls off much slower than for WCDMA. The reason is the use of
rectangular pulse shaping which leads to high side lobe level. An example is
shown in Figure 20 for 16, 64 and 256 sub-carriers. For a larger number of sub-
carriers the spectrum falls off more rapidly because side lobes are closer
together.
Pulse-shaping or time-domain windowing is then usually employed to suppress
most of the out-of-band emissions. A common technique is to apply a raised
cosine window in time domain. The roll-off factor β is defined as the portion of
the total symbol time Ts = TU + TCP where the roll-off is taking place, as shown
in Figure 21. In [Nee00] it is shown that a roll-off factor β = 2.5 % already
makes a large improvement in the out of band emissions. Larger roll-off of
course improves the spectrum further, however at the cost of smaller delay
spread tolerance since the usable part of the symbol gets shorter. The effective
guard time is reduced by βTs.

16 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Choice of windowing will be a trade-off between spectral properties and reduced


tolerance against delay spread.

Figure 20 Power spectral density without windowing for 16, 64 and 256 sub-
carriers [Nee00]

Ts = TU + TCP
TCP TU

Figure 21 OFDM cyclic extension and windowing


It varies whether windowing is actually used in different systems. In e.g.
Evolved UTRA and Mobile WiMAX no filtering is specified, while in the 3GPP2
standardised Ultra Mobile Broadband (UMB) a raised cosine filter is defined with
a roll-off factor between 2.4 and 2.9 %, dependent on choice of CP length.

Telenor R&I R 7/2008 - 17


OFDM(A) for wireless communication

3 Multi-user communications – OFDMA


Multi user communications require multiple access techniques. Generally, any of
the familiar techniques (TDMA and FDMA) can be employed regardless of the
OFDM-nature of the signal. However, the OFDM properties can also be used for
multiple access, i.e. Orthogonal Frequency Division Multiple Access (OFDMA).
OFDM can be used as a multiple access scheme by allowing simultaneous
frequency-separated transmissions to/from multiple mobile terminals as shown
in Figure 22.

Figure 22 OFDMA principles. Upper: Consecutive channel multiplexing. Lower:


Distributed channel multiplexing

3.1 Sub-carrier allocation techniques


The allocation of sub-carriers to the different users can be done in basically two
different ways:
• Consecutive (or localized) frequency mapping
• Distributed frequency mapping

3.1.1 Consecutive (localized) frequency mapping


Localized frequency mapping means that all sub-carriers belonging to the link
between one user terminal and the base station are allocated in one contiguous
block. An example is shown in Figure 23. The major drawback with this method
is that it is sensitive to frequency selective fading, see e.g. Figure 15. Whole
blocks can be erased.

18 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

A variant of localized frequency mapping is shown in Figure 24. In this case, the
sub-carriers are localized in equal size blocks, and resources allocated to one
user consist of one or more block. In this way, the system becomes more
robust to frequency selective fading.

Figure 23 Consecutive frequency mapping

Figure 24 Blockwise consecutive frequency mapping


For satisfactory performance, localized frequency mapping should be combined
with channel dependent scheduling, which is briefly described in section 3.1.3.
In E-UTRA an arbitrary number of sub-blocks of 12 sub-carriers can be
allocated on the downlink and scheduled differently every 0.5 ms (see section
4.1.1). In Mobile WiMAX, “Band AMC” is a localized frequency mapping
technique (see section 4.2.3).

3.1.2 Distributed frequency mapping


At the opposite end of the scale is the fully distributed mapping in which single
sub-carriers belonging to one link are spread across the whole OFDM bandwidth
as shown in Figure 25. It fully exploits the channel’s frequency diversity but
puts stronger demands on frequency synchronization.

Figure 25 Distributed frequency mapping


Obviously, this method is much more robust to frequency selective fading. A
method for re-scheduling resources in the time domain must also be employed.
Distributed mapping is employed in Mobile WiMAX as “FUSC” and “PUSC” (see
section 4.2.2). Clusters of 14 sub-carriers are re-scheduled every second OFDM
symbol, approximately 0.2 ms. The clusters are spread over the whole OFDM
bandwidth and is commonly called diversity sub-channelization. Although it may

Telenor R&I R 7/2008 - 19


OFDM(A) for wireless communication

seem like a blockwise method as described in the previous section, a detailed


examination reveals that this is a true distributed technique.

3.1.3 Channel dependent scheduling


Handling the frequency selective fading is important in OFDM-based systems in
order to give satisfactory performance. In OFDMA we have the possibility of
using channel estimates to optimize the sub-carrier allocation based on these
estimates.
Figure 26 shows how the time-frequency fading is different for two different
users due to their different locations [Dahl07]. By obtaining channel estimates
often enough, the time-frequency blocks can be re-scheduled to maximize the
performance. The example shows how it can be used in E-UTRA.

Figure 26 Channel dependent scheduling [Dahl07]

3.2 Synchronization aspects


In downlink (DL) all control is located in the base station, including full control
of time- and frequency synchronization.
In the uplink (UL) in principle the same techniques can be used as for DL,
however since the terminals generally have no knowledge of each other, some
demands are imposed. The ICI must be avoided or kept at a minimum, and
there are two effects which cause deterioration of this. Effect number one is
when the orthogonality between sub-carriers belonging to different users is
reduced due to imperfect frequency synchronization between the terminals;
hence good frequency synchronization is important. Additionally, the timing
must be good enough so that ISI is kept within the CP. The second effect is
when different mobile terminals are received with significantly different power
at the base station. Since perfect orthogonality is practically impossible, strong
received sub-carriers may interfere unacceptably on weaker received ones.
Consequently a power control on the mobile stations is important.

20 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

3.3 DFT-spread OFDMA (SC-FDMA)


In section 2.1.2, we showed that OFDM-signals suffer from a large peak-to-
average-power ratio (PAPR) when combining several narrowband carriers. This
imposes demands on the power amplifiers of the transmitter to be linear and be
operating with a large back-off to avoid harmonic and inter-modulation
distortion. In a base station, this is not a major problem, however in the
terminal this will lead to higher power consumption and shorter battery life as
well as a more expensive unit.
This is why an alternative for the uplink multiple access scheme has been
suggested and eventually adopted for 3GPP E-UTRA.
The UL multiple access scheme for 3GPP E-UTRA is called Single-Carrier
Frequency Division Multiple Access (SC-FDMA), a variant of DFT-spread OFDMA.
It is very similar to traditional OFDMA as used in the WiMAX uplink and is based
on the same building blocks. The difference is that the resulting signal applied
to the transmitter resembles a single-carrier behaviour and consequently a
lower PAPR. A short description is given here, while the details are given in the
next chapter. The description is mostly based on two papers by Myung et al
[Myu06] [Myu07] and by the recently published book by Dahlman et al
[Dahl07]. Figure 27 shows the signal chain from transmitter to receiver for both
conventional OFDMA and DFT-spread OFDMA.
NC NC
N N

SC de-mapping
NC-point IDFT

+CP, D/A+RF

NC-point DFT
RF+A/D, -CP
SC mapping

Channel

OFDMA
NC NC
N N
SC de-mapping
NC-point IDFT

+CP, D/A+RF

NC-point DFT
RF+A/D, -CP
SC mapping
N-point

N-point
IDFT
DFT

Channel

DFT-spread OFDMA

Figure 27 Transmitter and receiver structure of conventional OFDMA and DFT-


spread OFDMA
From the figure we see that in DFT-spread OFDMA, the block of input symbols,
{xn}, on the transmitter is fed through an N-point DFT before the sub-carrier
mapping is performed (N is the number of allocated SCs to the specific user).
This implies that all (time) symbols are spread on all allocated sub-carriers and
all sub-carriers are modulated with a weighted sum of all symbols. The sub-
carrier mapping can be both localized and distributed as explained above. When
applying the usual Inverse NC-point DFT after the sub-carrier mapping, new
time-domain symbols are generated (NC is the total number of SCs in the OFDM
channel). After adding CP and possible windowing a serial sequence of symbols

Telenor R&I R 7/2008 - 21


OFDM(A) for wireless communication

is modulated and transmitted, instead of the usual parallel OFDM-scheme. On


the receiver, an extra N-point IDFT is applied to recreate the original symbols.
The resulting signal can be said to have a “single-carrier” property, which is
intuitively “easy” to understand when using localized SC mapping. When
distributed SC mapping is used, this is not so obvious. It is shown in [Myu06b]
that the reduced PAPR holds for both cases.
Simulations done on different variants of SC-FDMA have been done in [Myu06b]
and show that the PAPR is reduced by approximately 2-3 dB for localized
mapping compared to conventional OFDMA (see Figure 28)

Figure 28 Complementary Cumulative Distribution Functions (CCDF) for the


PAPR of Interleaved (distributed) FDMA, Localized FDMA and OFDMA. 256 sub-
carriers total (NC), 64 allocated to user (N). Pulse shaping is with raised cosine
roll-off of 50 %. Left: QPSK, Right: 16-QAM [Myu06b]
However, this may only be the upside of SC-FDMA. It is also shown in [Alam07]
that the cost associated is 2 to 3 dB performance loss in fading channels in the
receiver, see Figure 29. Figure 30 also shows that when performing the DFT-
spreading, the PAPR is moved to the frequency domain leading to larger
instantaneous out-of band emissions.

0
16 QAM 1/2, Red: OFDMA, Blue:IFDMA, FFT size:1024, M=128
10

IFDMA
-1
PER

10

3 dB loss

OFDMA

-2
10
4 6 8 10 12 14 16 18 20 22 24
av. SNR per subcarrier(dB)

Figure 29 Signal-to-noise ratio (SNR) degradation for IFDMA (DFT-spread


OFDMA with distributed mapping) compared to OFDMA [Alam07]

22 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Inst. PSD (4 symbols), N=1024, M=128


10
SC-FDMA
OFDMA
0

-10

-20

-30

-40

-50

-60
-2000 -1500 -1000 -500 0 500 1000 1500 2000
subcarrier

Figure 30 Frequency spectrum of SC-FDMA and OFDMA showing increased


out-of-band emissions due to higher PAPR in the frequency domain [Alam07]

Telenor R&I R 7/2008 - 23


OFDM(A) for wireless communication

4 Wireless standards based on OFDM(A)


Several wireless and wired standards for both fixed and mobile communications
employ OFDM and OFDMA-techniques. Two of them will be specifically treated
in this report targeting the performance and radio planning aspects. These are
the Evolved UTRA (E-UTRA) standard from 3GPP, commonly known as Long
Term Evolution (LTE). The other is the IEEE 802.16e, mostly confined to the
profiles defined by the WiMAX Forum.
Some other standards using OFDM or OFDMA are:
• Ultra Mobile Broadband (UMB) also called EV-DO Rev. C, which is the
“LTE of 3GPP2”, i.e. the evolvement of the cdma2000 standard.
• WLAN IEEE 802.11a, IEEE 802.11g, and IEEE 802.11n
• WPANs – Ultra wideband based on Multiband-OFDM
• Terrestrial Digital Broadcast standards: DVB-T and DVB-H
E-UTRA and WiMAX are explained briefly from a bottom-up perspective, i.e.
emphasis is put on the OFDM and OFDMA aspects defining the physical layer
parameters like:
• Sub-carrier spacing and symbol period
• Number of sub-carriers
• Cyclic prefix length
• Multi-user scheduling aspects
• Pilot and reference signals

4.1 3GPP Evolved UTRA (Long Term Evolution)


3GPP Evolved UTRA (E-UTRA), usually named Long Term Evolution (LTE) is
formally still in its specification phase. It is targeted for finalization before the
end of 2007, so we should regard the physical layer specification to be final. It
is the “4G” technology from 3GPP. A new series of 3GPP specifications is
created, series 36, which covers all aspects of E-UTRA.
In this context we will describe the physical layer of the E-UTRA bottom-up and
concentrate on the OFDMA and SC-FDMA specific aspects. This means that
coding and protocol aspects will be touched on very lightly. We will also use the
term E-UTRA instead of LTE since this is the formal name used by 3GPP in the
standards documents.
The expected performance for E-UTRA is high with data rates above 100 Mb/s
on the downlink and 50 Mb/s on the uplink in a 20 MHz bandwidth using 2x2
MIMO. In September 2007, Nokia Siemens Networks (NSN) conducted a trial in
Berlin assisted by the Heinrich Hertz Institute (HHI) demonstrating peak data
rates of 160 Mb/s in a 20 MHz bandwidth in the 2.6 GHz frequency band using
2x2 MIMO1.

1
Press release from NSN at:
http://www.nokiasiemensnetworks.com/global/Press/Press+releases/news-
archive/Up_to_ten_times_faster_mobile_broadband_data_rates_a_step_closer_to_reality.htm

24 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

The multiple access scheme is based on OFDMA for the downlink and SC-FDMA
for the uplink. Cyclic prefix (CP) is used in both directions to protect against
ISI. Both frequency division duplex (FDD) and time division duplex (TDD) are
supported in order to support transmission both in paired and unpaired
spectrum. The bandwidth is scalable from 1.4 to 20 MHz.
The sub-carrier (SC) distance is chosen to be 15 kHz, defining the symbol
length to be 66.67 μs. It is designed to be narrower than the channel coherence
bandwidth so that the fading of each sub-carrier becomes approximately flat,
i.e. frequency non-selective. In this way the frequency domain equalization in
the receiver can consist of only 1 tap per SC. At the same time it needs to be
large enough to minimize ICI due to Doppler effects and phase noise in the
transmitter and receiver.
An additional motivation for this choice was to simplify the implementation of
WCDMA/HSPA/E-UTRA multi-mode terminals [Dahl07]. The sampling rate (fs =
Δf ⋅ NFFT, see Table 2) will be a multiple of the WCDMA/HSPA chip rate of 3.84
MHz.

4.1.1 Radio resource definitions


The basic concept is shown in Figure 31. The time and frequency domain is
organised in a grid of physical resource blocks spanning a number of sub-
carriers and time slots. The resource block is the smallest unit to which user
traffic is allocated, i.e. two users cannot share one resource block.

Figure 31 E-UTRA concept [Ekst06]


In the time-domain, the resource block is spanning one slot of 0.5 ms duration.
The slot consists of 7 or 6 consecutive OFDMA (downlink) or SC-FDMA (uplink)
symbols as shown in Figure 34. In the frequency domain it spans 180 kHz,
consisting of 12 consecutive sub-carriers. SC-FDMA symbols are often called
blocks, since they do not correspond to single input symbols.
The full number of sub-carriers during one slot is termed resource grid, and the
smallest unit, one symbol length on one sub-carrier, is called a resource
element. See Figure 32. In case of multi-antenna transmission, one resource
grid is defined per antenna port. Consequently there are 12⋅7 = 84 resource
elements per resource blocks (type 1 radio frame and normal CP, see Figure 33
and Table 1).

Telenor R&I R 7/2008 - 25


OFDM(A) for wireless communication

DL
N symb

Frequency
Resource
block

Resource element:
N scRB

One time-frequency symbol

Time

l=0 l = N symb
DL
−1

Figure 32 E-UTRA resource grid defining the physical resource block and
resource elements (Downlink example) [TS 36.211]
Further, two slots form one subframe of 1 ms duration, and 10 subframes (or
20 slots) form one radio frame of 10 ms. This is called a type 1 frame structure
defined for both FDD and TDD operation and is shown in Figure 33. An
alternative frame structure called “type 2” is also defined for TDD operation.
This is defined to optimize co-existence with legacy 1.28 Mchip/s UTRA TDD
systems. It is not treated further here.
Likewise, for the slot structure, two main types exist depending on the choice of
length of the CP. The “mainstream” choice will probably be the normal CP of
5.21/4.69 μs giving 7 symbols in a slot as shown in Figure 34.
In the standard documents, the time unit is often given as an integer number of
Ts = 1/(Δf⋅2048) ≈ 32.55 ns. Looking at Table 2, we see that this is the period
of the sampling frequency of the 20 MHz bandwidth transmission.

Figure 33 Type 1 radio frame of E-UTRA [TS 36.211]

26 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Slot: 0.5 ms

CP0 Symb#0 CP1 Symb#1 CP2 Symb#2 CP3 Symb#3 CP4 Symb#4 CP5 Symb#5 CP6 Symb#6

Figure 34 Slot structure with 7 OFDM/SC-FDMA symbols and short CP


For FDD, 10 subframes are available for downlink transmission and 10
subframes for uplink transmission in each 10 ms interval. For TDD a subframe
(two slots) is either allocated to downlink or uplink transmission. Subframes 0
and 5 are always allocated for downlink transmission.
The normal CP provides for a time delay caused by multipath of up to 1.5 km
and will be more than adequate for most coverage scenarios. However, an
extended CP of 16.67 μs is defined to cater for longer delays, especially for
broadcast scenarios receiving from multiple base stations in a single frequency
network (SFN). Use of extended CP then provides 6 symbols per slot.
Additionally it is possible on the downlink to combine the extended CP with half
inter-carrier distance (7.5 kHz) to increase the robustness against long delays
and multipath even more. The alternatives for the slot structure are listed in
Table 1.

Table 1 Values for slot lengths and cyclic prefix for E-UTRA
Number of w/normal CP 7
symbols per slot w/extended CP 6
Cyclic prefix Normal TCP0 = 5.21 μs (for 1 symbol per slot)
duration TCP1-6 = 4.69 μs (for 6 symbols per slot)
Extended TCP-e = 16.67 μs (for all 6 symbols in a slot)

Unless otherwise noted, we shall focus this description on the use of type 1
radio frames and slot structures with normal CP.
E-UTRA builds on the concept of scalable OFDMA (S-OFDMA), i.e. several
bandwidths are supported. The same flexibility is provided in the SC-FDMA
uplink. The parameters common for downlink and uplink are given in Table 2.

Telenor R&I R 7/2008 - 27


OFDM(A) for wireless communication

Table 2 Supported bandwidths and common DL and UL parameters for E-UTRA


[TR 36.104]
System channel 1.4 1.6 3 3.2 5 10 15 20
bandwidth [MHz] (TDD (TDD
only)* only)*

Slot duration [ms] 0.5

Sub-carrier frequency 15 (7.5)


spacing, Δf [kHz]

Useful symbol time, TU 66.67 (133.33)


[μs]

Cyclic prefix/guard Normal CP: 5.21 / 4.69


time, TCP [μs] Extended CP: 16.67

OFDMA symbol Normal CP: 71.88 / 71.36


duration, Extended CP: 83.33
Tsym = TU + TCP [μs]

Guard time overhead, Normal CP: 6.67


TCP/(TCP+TU) [%] Extended CP: 20.0

Resource block BW 180 kHz / 12 sub-carriers

Sampling frequency, 1.92 1.92 3.84 3.84 7.68 15.36 23.04 30.72
(15 000⋅NFFT [MHz])

FFT size, (NFFT) 128 128 256 256 512 1024 1536 2048

Occupied sub-carriers 72 84 180 192 300 600 900 1200

Resource mapping Blockwise contiguous

Duplex methods FDD and TDD

Modulation schemes QPSK, 16-QAM, 64-QAM; adaptive

Coding schemes 1/3 rate “tail-biting convolutional code”


1/3 rate Turbo code

* Included for spectrum compatibility with Low Chip Rate (LCR) TDD
[TR 25.913].

4.1.2 Modulation and coding


Adaptive modulation and coding is chosen and a number of modulation and
coding schemes (MCS) are defined. Supported modulation formats are QPSK,
16-QAM and 64-QAM, which are adaptively selected based on traffic types,
available bandwidths and radio channel quality.
Channel coding is not done on the physical layer but on the so-called transport
channels, and two main channel coding schemes are applied [TS 36.212], a so-
called “tail-biting convolutional code” with rate 1/3 and a Turbo code with rate
1/3. For control channels a 1/16 rate block code and a 1/3 rate repetition code
are also used. A table of combined modulation schemes and coding schemes is
not yet compiled.

4.1.3 Multi-antenna support


E-UTRA supports the use of multiple antennas at both the base station and the
terminal, a necessity to reach the performance goals. Different ways of
exploiting the multi-antenna possibilities are [Dahl07]:

28 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

• Receive diversity, common in cellular systems for many years on the


uplink to suppress fading. In E-UTRA also available on the downlink.
• Beamforming on the base station to improve received SNR/SIR
• Space-Time Coding (STC) such as Alamouti coding to improve spatial
diversity and reduce fading margin
• Spatial multiplexing (SM) or MIMO resulting in increased data rate
provided the channel is sufficiently “spread”.
Up to four antennas are supported on each side, while 2x2 is the so-called
“baseline” configuration. The different multi-antenna techniques are beneficial
in different scenarios and conditions.

4.1.4 Downlink scheduling and reference signals


The downlink modulation and multiple access is S-OFDMA as described in
section 2.2 with the parameters given above in Table 1 and Table 2.
Resources to different users on the downlink are allocated in the
aforementioned resource blocks. This is the smallest allocation unit. A
scheduled terminal can be assigned an arbitrary combination of 180 kHz wide
resource blocks. Scheduling takes part in every subframe (1 ms) and can be
channel dependent, i.e. be based on channel quality estimates which the
terminal reports back to the base station. The details are currently not
completely described in [TS 36.211].
Reference and pilot channels/signals are inserted in the OFDM grid in order to
improve the time and frequency synchronization in the receiver. As described in
section 2.1.4, this is important to avoid signal degradation. It is also used for
measuring downlink channel quality which is reported back to the base station
for e.g. scheduling purposes as described in the previous section.
In the time domain, cell-specific reference signals are transmitted in all
downlink subframes (2 slots) supporting unicast transmission. Reference signals
are not transmitted in each OFDM-symbol, but are mapped in the time-
frequency domain to the resource elements as shown in Figure 35. This is the
mapping for a single antenna transmission, however multi-antenna mappings
are similar, but reference signals are positioned differently and non-overlapping
(see [TS 36.211]). These same resource elements are also blocked for
transmission on the other antennas. The presence of different reference signal
mappings actually define the antenna ports.
Consequently, there are 4 reference signal resource elements in each resource
block of 84 resource elements on a single antenna. We see that they are
regularly positioned each 7th OFDM-symbol and each 6th sub-carrier.

Telenor R&I R 7/2008 - 29


OFDM(A) for wireless communication

Frequency (subcarrier)

180 KHz (12 SC)

Slot = 0.5 ms
Subframe = 1 ms

Figure 35 Mapping of downlink reference signals for a single antenna


transmitter [TS 36.211]
The reference signal structure described here gives the same positions of
reference symbols in consecutive subframes, but according to [Dahl07] a
reference symbol frequency hopping may be applied. This is at time of writing
not included in [TS 36.211].

4.1.5 Uplink scheduling and reference signals


The uplink modulation and multiple access technique chosen is the SC-FDMA
technique described in section 2.2.1. It has many similarities with OFDMA, and
uses the same parameters (Table 1 and Table 2) as the downlink.
Also in the uplink, resources are assigned to the resource blocks. Different from
the downlink, only a contiguous frequency region can be assigned to the
terminals. This is a consequence of the use of the “single-carrier” transmission
on the uplink. The scheduling decisions are also here taken each subframe (1
ms), however, an alternative may be inter-slot frequency hopping, implying
that the physical resources used in the two slots of a subframe do not occupy
the same set of SCs [Dahl07]. It is not however described in the standard yet
[TS 36.211].
Reference signals are defined also in the uplink for the purpose of demodulation
reference and sounding reference (channel estimation). There is one
demodulation reference signal per slot. The principle is different from that on
the downlink in that it is not possible to frequency multiplex the reference
signals with the data transmission. Instead, the uplink reference signals are
time-multiplexed with the uplink data and transmitted within the fourth block
(“symbol”) of each uplink slot as shown in Figure 36. Details are still open in
[TS 36.211].

30 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 36 Uplink reference signals are inserted within the fourth block of each
uplink slot [Dahl07]

4.2 Mobile WiMAX2


Mobile WiMAX is based on the air interface of IEEE 802.16e-2005 and utilizes
the concept of scalable OFDMA (S-OFDMA). S-OFDMA supports a wide range of
bandwidths to flexibly address the needs in various frequency spectrums. The
WiMAX Forum has selected a subset of the 802.16e standard as mandatory in
order to ensure equipment interoperability and provides test and conformance
specifications. WiMAX Release 1 has been ready since 2006 and covers
“profiles” for the system bandwidths 5, 7, 8.75 and 10 MHz and frequency
bands 2.3, 2.5, 3.3 and 3.5 GHz [WiMAX06].
Possibly important for future deployment of WiMAX technology is the fact that
ITU-R at the Radio Assembly (RA) in October 2007 approved Mobile WiMAX to
be included into the IMT-2000 family under the name “OFDMA TDD WMAN”3 .
This means that the existing 3G frequency bands generally are open for the
deployment of Mobile WiMAX as a direct competitor to UTRA/WCDMA provided
that the national regulators’ license policies are technology neutral.
Mobile uses OFDMA as the multiple access scheme both in uplink and downlink.
Cyclic prefix is used in both directions to protect against ISI. Bandwidth
scalability from 1.25 to 20 MHz is supported by adjusting the FFT size while
maintaining fixed sub-carrier frequency spacing. The sub-carrier spacing is
chosen to be 10.94 kHz, which defines the useful symbol length to be 91.4 μs.
Since the resource unit sub-carrier bandwidth and symbol duration are fixed,
the impact to higher layers is minimal when scaling the bandwidth. The OFDMA
parameters applied by mobile WiMAX are summarized in Table 3.

2
The presentation of mobile WiMAX is based on [WiMAX06], [Nua07], and [And07].
3
ITU-R press release from RA-07 at: http://www.itu.int/newsroom/press_releases/2007/30.html

Telenor R&I R 7/2008 - 31


OFDM(A) for wireless communication

Table 3 Mobile WiMAX OFDMA parameters [WiMAX06]


System channel bandwidth [MHz] 1.25* 5 10 20* 7 8.75
Sub-carrier frequency spacing Δf 10.94 7.81 9.77
[kHz]
Useful symbol time, TU = 1/Δf [μs] 91.4 128.0 102.4
Cyclic prefix/guard time, TCP = TU /8 11.4 16.0 12.8
[μs]
OFDMA symbol duration, Tsym = TU + 102.9 144.0 115.2
TCP [μs]
Guard time overhead, 11.1
TCP/(TCP+TU) [%]
Sampling Frequency, fs [MHz] 1.4 5.6 11.2 22.4 8.0 10.0
FFT Size, NFFT 128 512 1024 2048 1024 1024
Occupied sub-carriers (DL-PUSC/UL- 360/ 720/
PUSC) 272 560
Resource mapping Distributed or contiguous
Duplex method TDD only
Modulation schemes QPSK, 16-QAM, 64-QAM; adaptive
Coding schemes 1/2, 2/3, 3/4, 5/6 rate convolutional code
1/2, 2/3, 3/4, 5/6 rate convolutional Turbo code
x2, x4, x6 repetition code
* Not in Mobile WiMAX release 1

In Mobile WiMAX the frequency domain consists of three types of sub-carriers


(Figure 37):
• Data sub-carriers for data transmission
• Pilot sub-carriers for estimation and synchronization purposes
• Null sub-carriers; used for guard bands and DC carriers

Figure 37 Frequency domain view of OFDMA sub-carriers

4.2.1 Radio resource definitions – sub-channelization


Data and pilot sub-carriers are grouped into subsets of sub-carriers called sub-
channels. WiMAX supports sub-channelization in both DL and UL. The smallest
time-frequency resource unit that is allocated to transmit one data block in
mobile WiMAX is referred to as a slot. The number of slots applied for a
transmission depends on the size of the data block. Furthermore, the number of
frequency-time symbols (combination of one sub-carrier and one symbol
period) used for data in a slot is 48, while the number of frequency-time
symbols used for pilot symbols varies between different permutation schemes.
There are two types of sub-carrier permutations for sub-channelization;
diversity and contiguous. In general, diversity sub-carrier permutations perform
well in mobile applications since the channel quality becomes difficult to track,

32 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

while contiguous sub-carrier permutations are well suited for fixed, portable, or
low mobility environments. These options enable the system designer to trade-
off mobility for throughput.

4.2.2 Diversity permutations


The diversity permutation draws sub-carriers pseudo-randomly to form a sub-
channel. It provides frequency diversity and inter-cell interference averaging.
The diversity permutations include DL fully used sub-carrier (FUSC), DL
partially used sub-carrier (PUSC), and UL PUSC.
For DL FUSC, all data sub-carriers are applied when creating the various sub-
channels. Each sub-channel is made up of 48 data sub-carriers that are
distributed evenly throughout the entire frequency band as shown in Figure 38.

Figure 38 FUSC sub-carrier permutation scheme


The pilot sub-carriers are allocated first, and then the remaining sub-carriers
are mapped to the different sub-channels. For some of the pilot symbols the
frequency position is fixed (constant set pilot sub-carrier) while some are
changing positions from symbol interval to symbol interval (variable set pilot
sub-carrier).
With DL PUSC, the available sub-carriers are grouped into clusters containing
14 contiguous sub-carriers over two symbol intervals, with pilot and data
allocations in each. A re-arranging scheme is used to form 6 groups of clusters
such that each group is made up of clusters that are distributed throughout the
sub-carrier space. A new permutation is performed within each group to form
the sub-channels, where one sub-channel is made up of 28 sub-carriers. The
permutation procedure is illustrated in Figure 39.

Telenor R&I R 7/2008 - 33


OFDM(A) for wireless communication

Figure 39 DL PUSC sub-carrier permutation scheme


Even if this may look like a blockwise distribution, the detailed example in
Figure 40 shows that this is a truly distributed technique.
The motivation behind DL PUSC is to make it possible to obtain diversity gain
over the whole bandwidth while keeping the orthogonality for closely separated
cells. This is done by dividing the groups of clusters between cells that are
closely spaced and thus potentially sources to interference. If all six groups are
used for each cell, the PUSC scheme will be similar to the DL FUSC scheme.

34 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Frequency

P P
Cluster: 14 SC x 2 symbols
P P

30 clusters/420 SCs Physical mapping

Cluster renumbering

Major group: 10 clusters/120 data SCs


Logical mapping

Sub-carrier mapping

Logical sub-channel/24 data SCs from a group

DL-PUSC, NFFT = 512

Figure 40 Detailed example of the DL PUSC permutation technique for a FFT length of 512 (360 active sub-carriers)

Telenor R&I R 7/2008 - 35


OFDM(A) for wireless communication

Analogous to the cluster structure for DL, a tile structure is defined for the UL
PUSC. The available sub-carrier space is split into tiles (four sub-carriers over 3
symbol periods), which is pseudo randomly gathered into 6 groups. As for DL
PUSC, a new permutation is performed inside each of the groups to form the
sub-channels. The sub-channel comprises 24 sub-carriers. The procedure is
illustrated in Figure 41.

Figure 41 UL PUSC sub-carrier permutation scheme


We should also mention that there exists a DL tile usage of sub-carriers (TUSC)
permutation. This is similar to the UL PUSC, and the motivation is to have a
symmetric permutation on UL and DL for closed loop advanced antenna
systems (AAS) (see Section 4.2.6). Consequently, a symmetric TDD link does
not need to have an explicit feedback link between BS and MS to exchange
channel state information due to reciprocity. Note that DL TUSC is not
mandatory in the current WiMAX profiles.

4.2.3 Contiguous permutation


The contiguous permutation groups a block of contiguous sub-carriers to form a
sub-channel. The contiguous permutations include DL adaptive modulation and
coding (AMC) and UL AMC, and have the same structure. Nine contiguous sub-
carriers are gathered in a bin, with 8 sub-carriers assigned for data and one
assigned for a pilot as illustrated in Figure 42.

36 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 42 Band AMC sub-carrier permutation scheme


A sub-channel in AMC is defined as a collection of bins of the type (N x M = 6),
where N is the number of contiguous bins and M is the number of contiguous
symbols. Consequently, the allowed combinations are: (6 bins, 1 symbol), (3
bins, 2 symbols), (2 bins, 3 symbols), and (1 bin, 6 symbols). AMC permutation
enables multi-user diversity by choosing the sub-channel with the best
frequency response for each user.

4.2.4 Permutation schemes summary


The mandatory profiles in the current WiMAX profiles are DL FUSC, DL PUSC,
DL AMC, UL PUSC and UL AMC. In Table 4 we have summarized the mobile
WiMAX permutation schemes presented and the corresponding slot definitions.

Table 4 Mobile WiMAX permutation schemes


Permutation scheme Slot definition
DL FUSC 1 sub-channel x 1 OFDMA symbol
DL PUSC 1 sub-channel x 2 OFDMA symbols
UL PUSC and DL TUSC 1 sub-channel x 3 OFDMA symbols
DL AMC and UL AMC 1 sub-channel x (1, 2 or 3) OFDMA symbols

4.2.5 Modulation and coding


WiMAX supports adaptive modulation and coding on all permutation schemes,
not only the one named “AMC”. Supported modulation formats are QPSK,
16.QAM and 64-QAM.
Channel coding supported are Convolutional Codes (CC) and Convolutional
Turbo Codes (CTC) with variable code rate and repetition coding. Optionally,
Block Turbo Codes and Low Density Parity Check Code (LPDC) are supported.
Code rates are 1/2, 2/3, 3/4 or 5/6. For more details, refer to [WiMAX06] which
also shows a table of supported physical layer bit rates.

Telenor R&I R 7/2008 - 37


OFDM(A) for wireless communication

4.2.6 Multi-antenna support


OFDMA allows smart antenna operations to be performed on frequency flat sub-
carriers, i.e. complex equalizers are not required to compensate for frequency
selective fading. OFDMA is therefore very well-suited to support smart antenna
technologies. Mobile WiMAX supports a full range of smart antenna technologies
to enhance system performance. The smart antenna technologies supported
include:
• Beamforming: With beamforming, the system uses multiple antennas to
transmit or receive weighted signals to improve coverage and capacity
of the system and reduce outage probability.
• Space-Time Coding (STC): Transmit diversity such as Alamouti coding is
supported to provide spatial diversity and to reduce fade margin.
• Spatial Multiplexing (SM): Spatial multiplexing is supported to take
advantage of higher peak rates and increased throughput. With spatial
multiplexing, multiple streams are transmitted over multiple antennas.
The receiver must also have multiple antennas which are used to
separate the different streams to achieve higher throughput compared
to single antenna systems. In UL, each user has only one transmit
antenna, i.e. two users can transmit collaboratively in the same slot as if
two streams are spatially multiplexed from two antennas of the same
user. This is called UL collaborative SM.
The supported features in the Mobile WiMAX performance profile are listed in
the following Table 5.

Table 5 Advanced antenna options for Mobile WiMAX


Link Beamforming Space Time Spatial
Coding Multiplexing
DL Nt ≥ 2, Nr ≥ 1 Nt = 2, Nr ≥ 1 Nt = 2, Nr ≥ 2
UL Nt ≥ 1, Nr ≥ 2 N/A Nt = 1, Nr ≥ 2

4.2.7 Frame Structure


The 802.16e physical layer specification supports TDD and Full and Half-Duplex
FDD operation; however the initial release of mobile WiMAX certification profiles
will only include TDD. With ongoing releases, FDD profiles will be considered by
the WiMAX Forum to address specific market opportunities where local
spectrum regulatory requirements either prohibit TDD or are more suitable for
FDD deployments.
The OFDMA frame structure for TDD operation is shown in Figure 43, where the
following fields are defined:
• Preamble: The preamble, used for synchronization, is the first OFDM
symbol of the frame.
• Frame Control Header (FCH): The FCH follows the preamble. It provides
the frame configuration information such as MAP message length, coding
scheme and usable sub-channels.
• DL-MAP and UL-MAP: The DL-MAP and UL-MAP provide sub-channel
allocation and other control information for the DL and UL sub-frames
respectively.

38 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

• UL Ranging: The UL ranging sub-channel is allocated for mobile stations


(MSs) to perform closed-loop time, frequency, and power adjustment as
well as bandwidth requests.
• UL CQICH: The UL CQICH channel is allocated for the MS to feedback
channel state information.
• UL ACK: The UL ACK is allocated for the MS to feedback DL HARQ
acknowledge.
• DL/UL Bursts: The DL and UL bursts are regions in the time-frequency
domain where the same burst profile is used, i.e. combination of coding
and modulation.

Figure 43 WiMAX TDD OFDMA Frame Structure


Both the UL and DL frame can be split into different zones, and different
permutation schemes can be applied in the different zones. This makes the
system very flexible with respect to different kinds of MSs with different design
complexity. The principle is illustrated in Figure 44.
(FCH and MAP)
Preamble

TUSC
PUSC

PUSC

PUSC
FUSC

AMC

AMC

DL

- Must appear in every frame

- May appear in every frame

Figure 44 Multi-zone frame structure

Telenor R&I R 7/2008 - 39


OFDM(A) for wireless communication

The figure shows that it is only the permutation scheme of the FCH and MAP
information that are predefined; the rest of the frame is flexible.

4.3 E-UTRA vs. Mobile WiMAX – summary


From the details in sections 4.1and 4.2, we can summarize a few of the
essential parameters from the OFDMA physical layers of the E-UTRA and WiMAX
standards.

Table 6 Comparison of basic OFDMA parameters of E-UTRA and Mobile WiMAX


E-UTRA Mobile WiMAX
Release 1*
Supported bandwidths [MHz] 1.4, 1.6, 3, 3.2, 5, 10, 5, 7, 8.75, 10
15, 20
Sub-carrier frequency spacing, 15, 7.5 (opt.) 10.94 (7.81, 9.77)
Δf [kHz]
Useful symbol time [μs] 66.67, 133.33 91.4 (128.0, 102.4)
Cyclic prefix / guard time Norm: 5.21/4.69; 11.4 (16.0, 12.8)
[ms] Ext: 16.67
Guard time overhead, Norm: 6.67; 11.1
TCP/(TCP+TU) [%] Ext: 20.0
Sub-channelizations Distributed, Diversity, Contiguous
Contiguous RBs
FDD or TDD operation FDD and TDD TDD
*
Numbers in brackets are valid for the 7 and 8.75 MHz bandwidths,
respectively.

It is important to note that while Mobile WiMAX has been standardized for a
while, the E-UTRA is still in writing. Consequently, it may seem that E-UTRA has
more options. This may not be the practical situation when equipment starts to
be available. There is a high probability that E-UTRA implementations will be
confined to e.g. a subset of the supported bandwidths, at least from the
beginning. It is also expected that since the basis for Mobile WiMAX, namely the
IEEE 802.16e-2005 standard, contains a wider range of options, more of these
will be part of future releases from the WiMAX Forum.
The OFDM SC spacing for E-UTRA and Mobile WiMAX are 15 kHz and 10.94 kHz,
respectively. From the short discussion in section 2.2.1, we saw that a smaller
SC spacing makes the system more vulnerable to frequency errors, either
caused by mobile radio propagation Doppler effects or by other sources. The
difference is actually quite large and implies that Mobile WiMAX probably is less
robust to high velocity mobility than E-UTRA, but we talk velocities way beyond
100 km/h in any case. This tendency is even strengthened if bandwidths of 7
MHz or 8.75 MHz are chosen, because the sub-carrier distance is even smaller.
E-UTRA and Mobile WiMAX seem to have different approaches to choosing the
appropriate guard time. While Mobile WiMAX has chosen a single value to cover
the targeted scenarios, E-UTRA provides two choices:
• E-UTRA normal CP is 5.21/4.69 μs, covering an excess delay of up to 1.5
km
• Mobile WiMAX CP is 11.4 μs, covering an excess delay of up to 3.4 km
• E-UTRA extended CP is 16.67 μs, covering an excess delay of up to 5 km

40 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

The overhead also varies accordingly, and it seems that 3GPP tries to target E-
UTRA for slightly better resource utilization in the normal case, at the expense
of less robustness for long delays, which might occur in rural scenarios. The
option of an extended CP covers this in addition to the broadcast SFN scenario.
Mobile WiMAX, on the other hand, compromises and has chosen a value in
between, basically covering all practical scenarios.
Differences in time and frequency domain are shown in Figure 45.

E-UTRA

Δf = 15 kHz TCP ≈ 5 μs TU = 66.67 μs

Mobile WiMAX

Δf = 10.94 kHz TCP = 11.4 μs TU = 91.4 μs

Frequency Time

Figure 45 Sub-carrier distance, useful symbol time and cyclic prefix for E-
UTRA and Mobile WiMAX
In the end, the differences on the physical layer of E-UTRA and Mobile WiMAX
are small, and it is not possible to point out a “winner” based on this.

4.4 Other OFDM and OFDMA based standards


In this section is given a brief overview of other OFDM or OFDMA based wireless
standards for communication or broadcast.

4.4.1 Mobile WiMAX Release 2 – IEEE 802.16m


This is the next generation WiMAX standard which targets performances above
WiMAX R1 and E-UTRA [Alam07]. The standardisation work started in 2007.
The bandwidth support is increased upwards and includes 5, 10, 20 and 40
MHz. The requirements for peak data rates are above 350 Mb/s downlink using
4x4 MIMO and above 200 Mb/s uplink using 2x2 MIMO. The average throughout
per sector shall be more than 40 Mb/s downlink and 12 Mb/s uplink. Terminal
mobility up to 350 km/h is required.

4.4.2 3GPP2 Ultra Mobile Broadband (UMB)


The 3GPP2 is standardising the wireless technologies based on the cdmaOne
and cdma2000 standards. Their counterpart to E-UTRA is called Ultra Mobile

Telenor R&I R 7/2008 - 41


OFDM(A) for wireless communication

Broadband (UMB), or cdma2000 EV-DO Rev. C. This is also based on scalable


OFDMA. The main parameters are [UMB]:
• Bandwidth support: 1.25, 2.5, 5, 10, 20 MHz
• Number of sub-carriers (FFT size): 128, 256, 512, 1024, 2048
• Sub-carrier spacing/useful symbol duration: 9.6 kHz / 104.17 μs
• Cyclic prefix (CP) length: Choice between 6.51, 13.02, 19.53, 26.04 μs
• Windowing: raised cosine, guard interval: 3.26 μs
• Modulation: Adaptive, QPSK, 8-PSK, 16-QAM, 64-QAM, hierarchical
modulation
The standard was released in September 20074.

4.4.3 Wi-Fi and WLAN 802.11a-g-n


The WLAN standards IEEE 802.11a, g and n are all OFDM based, however a
different multiple access technique is used. This is called Carrier Sense Multiple
Access (CSMA), basically a refined ALOHA technique. The channel bandwidth
(.11a, .11g) is 22 MHz. 802.11n can support so-called double bandwidth of 40
MHz. The number of sub-carriers is 52 with a spacing of 312.5 kHz. This gives a
useful symbol length of 3.2 μs. The CP is 0.8 μs. Adaptive modulation is
supported between BPSK, QPSK, 16-QAM and 64-QAM [802.11a]. Compared to
WiMAX and E-UTRA there are short symbols and CP, meaning that the design is
targeting short ranges.

4.4.4 WPAN 802.15.3a Multi Band OFDM


The IEEE 802.15.3a is a short range standard employing Ultra Wideband (UWB)
technology. The UWB band is defined from 3.1 to 10.6 GHz. This standard has
been released via Ecma International, originally a manufacturers association
called “European Computer Manufacturers Association”5. Later they turned into
a standardisation group called “Ecma International - European association for
standardizing information and communication systems”.
The physical layer of the IEEE 802.15.3a standard is based on Multi Band OFDM
(MB-OFDM) [ECMA-368]. The whole UWB band is divided into 14 bands of 528
MHz each. These are organized as 4 groups with 3 bands each and one group
with the two upper bands as shown in Figure 46. Each band uses 100 sub-
carriers and 12 pilots. Within each band group coded data are spread using
time-frequency codes (TFC) giving support to 7 logical channels per band group
of the first four groups and 2 for the last group, in total 30 channels. The
standard supports data rates from 53.3 to 480 Mb/s.

4
Press release from 3GPP2 at:
http://www.3gpp2.org/Public_html/News/Release_UMBSpecification24SEP2007.pdf
5
See: http://www.ecma-international.org/memento/history.htm

42 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 46 Organisation of the UWB band for MB-OFDM [ECMA-368]


The OFDM parameters employed:
• Number of sub-carriers (per band): 100 (FFT: 128)
• Sub-carrier spacing / useful symbol length: 4.125 MHz / 242.42 ns
• Guard interval (zero-padded suffix): 70.08 ns (37 samples)
Note that this standard does not use a cyclic prefix; the guard interval is only
zero padding. From the values it is obvious that this is an extreme short range
system.

4.4.5 Digital terrestrial video broadcast – DVB-T/H


Broadcast is a one-way service, and consequently no multiple access technique
is employed. The “new” digital standards, DVB-T and DVB-H for terrestrial and
mobile television are based on OFDM with the following parameters
[EN 300 744]:
• Channel bandwidths: 5, 6, 7 and 8 MHz
• Number of sub-carriers (including pilots):
2K mode: 1705 (FFT: 2048)
4K mode: 3409 (FFT: 4096) – only DVB-H
8K mode: 6817 (FFT: 8192)
• Sub-carrier spacing (8 MHz channel): 4.464, 2.232, 1.116 kHz
• Useful symbol length (8 MHz channel): 224, 448, 896 μs
• Guard intervals can be chosen as fraction of the useful symbol between
1/32, 1/16, 1/8 and 1/4, leading to CP lengths ranging from 7 to 224 μs
dependent on chosen mode.
• Modulation: QPSK, 16-QAM, 64-QAM, hierarchical modulation
DVB-T is currently deployed to replace analogue TV transmissions throughout
Europe. Mobile TV based on DVB-H is available in a few countries, i.a. Italy and
Finland.

Telenor R&I R 7/2008 - 43


OFDM(A) for wireless communication

5 OFDMA radio planning


Since Mobile WiMAX is a more mature standard than E-UTRA, i.e. Mobile WiMAX
products are already hitting the market; we have found most information
concerning radio planning of OFDMA systems from the WiMAX literature.
However, we do not see any reasons that radio planning should differ much
between Mobile WiMAX and E-UTRA.
In general, a radio planning procedure may have the following steps [Hag07]:
1. Find the prerequisites
Typical parameters are bandwidth, frequency, transmitter power,
etc.
2. Define cell edge quality
Typical parameters are UL/DL bit rate and coverage probability.
3. Calculate coverage
The coverage is found based on the link budget.
Result: Max path loss and cell range
4. Calculate capacity
Result: Throughput distribution for users or the cell
The procedure is illustrated in Figure 47.

Define edge Input parameters:


• power
quality • antennas
• etc …

If coverage or Results:
capacity do Coverage • Path loss
not meet • Cell range
requirements

Results:
• Cell capacity
Capacity • Throughput
• Distributions

Done

Figure 47 Radio planning procedure


When comparing OFDMA planning to GSM and UMTS planning, it is our opinion
that it has most resemblance to GSM. The main reason for this conclusion is
that for OFDMA systems it is common to assume that the intra-cell interference
is negligible due to the orthogonal nature of the sub-carriers. For UMTS, on the
other hand, allowing intra-cell interference is an integral part of WCDMA
operation (especially on the uplink).
However, to obtain high efficiency, aggressive frequency reuse is seen as the
most interesting way to deploy new OFDMA systems. Consequently, the inter-

44 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

cell interference for the upcoming OFDMA systems is expected to be much


larger than for traditional GSM system. In the next section we will look at the
frequency reuse patterns suggested for OFDMA systems and the corresponding
inter-cell interference situations that arise. After that we will look at some link-
budgets suggested by Nokia Siemens and Ericsson.

5.1 Frequency reuse


Traditional reuse patterns for conventional cellular deployments used cell
frequency reuse factors as high as seven to mitigate inter-cellular co-channel
interference (CCI). These deployments assured a minimal spatial separation of
5:1 between the interfering signal and the desired signal, but only 1/7 of the
frequency resources can be utilized at each BS. These high frequency reuse
factors can of course be employed for OFDMA systems as well.
As mentioned above, for the OFDMA systems, adjacent channel interference
(ACI) is controlled by the orthogonal nature of the sub-carriers. However, due
to the frequency reuse in the cellular system, CCI is present. In Figure 48 the
SNIR cumulative distribution function (CDF) is plotted for different frequency
reuse factors in a regular hexagonal 3 sector deployment for fully loaded cells
with optimal cell selection.

Figure 48 SNIR (C/I) for different reuse factors [Lam07]


The figure clearly illustrates how the SNIR increases with increasing reuse
factor.
With technologies such as WCDMA and OFDMA, aggressive frequency reuse
schemes can be employed to improve overall spectrum efficiency. The
frequency reuse configurations seen as the most interesting for a multi-cellular
deployment of the OFDMA systems are BSs with 3 sectors, and a sector reuse
of one or three.

Telenor R&I R 7/2008 - 45


OFDM(A) for wireless communication

5.1.1 Frequency reuse 1


With a frequency reuse of 1, also referred to as universal frequency reuse, the
same channel is deployed in each of the 3 BS sectors as shown in Figure 49.

Figure 49 Scenario with reuse factor 1 [WiMAX07]


This approach has the advantage of using the least amount of spectrum and
may in many cases represent the only deployment reuse alternative due to
limited spectrum availability. It has the highest spectral efficiency and gives the
best aggregated performance.
With reuse 1, two different inter-cell interference schemes can be employed to
mitigate CCI at the sector boundaries and at the cell-edge, i.e. avoidance and
randomization. Inter-cell interference randomization is achieved by either
spreading the transmitted information over a larger part of the available
bandwidth (scrambling in the frequency domain), or by applying frequency
hopping [Bachl07]. The objective is to make the interference experienced by
the users random; hence all users share the burden of having a strong co-
channel interferer.

5.1.2 Frequency reuse 3


An alternative to universal frequency reuse (reuse 1) is a reuse of 3, where
each of the three sectors is assigned a unique channel. Thus, assuming the
same channel bandwidth, a 3-sector base station deployment requires three
times as much spectrum as reuse 1. Reuse 3 eliminates CCI at the sector
boundaries and significantly decreases CCI between neighbour cells. This
happens due to the increased spatial separation for channels operating at the
same frequency, provided that the cell sector boundaries are properly aligned.
A reuse of 3 enables greater use of all of the sub-carriers, thus increasing the
spectral efficiency of each channel. The scenario is illustrated in Figure 50.

46 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

Figure 50 Scenario with reuse factor 3 [WiMAX07]


Although the improvements in channel spectral efficiency within each cell with
reuse 3 can be significant, the overall spectral efficiency will always be lower
when the added spectrum requirements are taken into account.

5.1.3 Fractional frequency reuse


Inter-cell interference avoidance is usually based on fractional frequency reuse.
The key features of this scheme are summarized below.
• Users in a particular sector are designated into multiple classes, e.g.,
centre cell users (those that are closest to the base station) and cell
edge users. The user geometry is dynamically updated.
• Different bandwidth allocation patterns are assigned for different user
classes. A centre cell user can be allocated the entire spectrum, whereas
only a fraction of the entire spectrum is used when transmitting to users
at the edge of the cell.
• Transmissions across BSs and sectors are coordinated so that maximal
interference avoidance is achieved.
As a result some channel capacity is sacrificed since some sub-carriers will not
be fully utilized throughout the entire cell. Therefore, strictly speaking, the
reuse factor for such a scheme is larger than 1, and may be a fractional number
[Jia07]. The fractional frequency reuse scheme is illustrated in Figure 51, where
three frequency groups are utilized (f1, f2, and f3).

Telenor R&I R 7/2008 - 47


OFDM(A) for wireless communication

Figure 51 Fractional frequency reuse [Dahl07]


Fractional frequency reuse can also be accommodated by time-coordination as
shown in Figure 52.

DL subframe UL subframe

Pre- Centre cell Whole cell Centre cell Whole cell


amble with FRF = 1 with FRF = 3 with FRF = 1 with FRF = 3

Figure 52 Time coordinated fractional frequency reuse in WiMAX TDD


Another variant of fractional frequency reuse is called Inter-cell interference
coordination and implies power coordination. The transmission power of parts of
the spectrum is reduced in one cell and the interference seen in the neighbour-
ing cells in this part of the spectrum will be reduced. An example is shown
Figure 53 [Dahl07].

Figure 53 Example of inter-cell interference coordination, where parts of the


spectrum is restricted in terms of transmission power [Dahl07]

5.2 OFDMA link budgets


In the link budget the interference is accounted for by introducing an
interference margin. By searching relevant references, we found that the
interference margin applied in the different link budgets varies quite a lot.
WiMAX Forum states that the interference margin should be 2 dB for DL and 3
dB for UL assuming a frequency reuse of 1 (fractionally frequency reuse), while
the interference margin can be reduced to 0.2 dB for a frequency reuse of 3
[WiMAX06]. Furthermore, we included an example of a link budget for E-UTRA,

48 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

from Nokia Siemens (Figure 54). In this example the interference margin is
proposed at 2 dB for the UL and 3 dB for the DL. Other examples show
variations from 1 – 6 dB [Lam07] [Hag07].
To summarize, the interference margin is dependent on the frequency reuse
strategy employed, and further investigations should be performed before we
can give an accurate recommendation on the interference margin to be utilized
in the radio planning.

Uplink Downlink
Data rate kbps 64,0 Data rate Mbps 1,0

Transmitter - UE Transmitter - Node B


a Max. TX power dBm 24,0 a HS-DSCH power dBm 46,0
b TX antenna gain dBi 0,0 b TX antenna gain dBi 18,0
c Body loss dB 0,0 c Cable loss dB 2,0
d EIRP dBm 24,0 =a + b - c d EIRP dBm 62,0 =a + b - c

Receiver - Node B Receiver - UE


e Node B noise figure dB 2,0 e UE noise figure dB 7,0
f Thermal noise dBm -118,4 = k(Boltzmann) x T(290K) x B(360kHz) f Thermal noise dBm -104,5 = k(Boltzmann) x T(290K) x B(9MHz)
g Receiver noise floor dBm -116,4 =e + f g Receiver noise floor dBm -97,5 =e + f
h SINR dB -7,0 From simulation (Nokia) h SINR dB -10,0 From simulation (Nokia)
i Receiver sensitivity dBm -123,4 =g + h i Receiver sensitivity dBm -107,5 =g + h
j Interference margin dB 2,0 j Interference margin dB 3,0
k Cable loss dB 2,0 k Control channel overhedB 1,0
l RX antenna gain dBi 18,0 l RX antenna gain dBi 0,0
m MHA gain dB 2,0 m Body loss dB 0,0

Maximum path loss dB 163,4 =d - i - j + k + l - m Maximum path loss dB 165,5 =d - i - j + k + l - m

Figure 54 E-UTRA link budget (outdoor) Nokia Siemens [Holma07]

Telenor R&I R 7/2008 - 49


OFDM(A) for wireless communication

6 Conclusions
In this report we have explained how transmission and multiple access based
on OFDM and OFDMA is designed, with special attention to the Mobile WiMAX
standard and the Evolved UTRA (also known as LTE) standard, both which are
based on Scalable OFDMA.
One major advantage of OFDM is that it can efficiently handle multipath
transmission without complex receivers. A simple one-tap equalized is enough.
This also opens for efficient and simple support of multiple antennas. The use of
Fast Fourier Transform (FFT) techniques is “simple” and cost-efficient. Changing
the FFT size also makes it easily scalable with respect to bandwidth. On the
downside, the OFDM-signal possesses the property of having a high so-called
Peak-to Average Power Ratio (PAPR), which increases with the number of sub-
carriers used, however this is also present in WCDMA on a smaller scale. This is
unfavourable because to ensure linear behaviour a large backoff is needed in
the amplifier. On the base station transmitter, this may not be a major
problem, but on the user terminal transmitter, this implies higher cost and
possibly higher power consumption and shorter battery life.
OFDM used as a multiple access technique, commonly known as OFDMA,
benefits from the scalability in assigning resources to different users in a
flexible manner whether FDD or TDD operation is assumed. Since OFDM utilizes
time- and frequency diversity, channel dependent scheduling is possible.
Main advantages of OFDMA:
• Scalability
• Effective implementation using FFT (favourable for MIMO
implementation)
• Opportunistic frequency scheduling
Main disadvantage of OFDMA
• Larger PAPR
The two major standards for mobile broadband are Mobile WiMAX, based on
IEEE 802.16e, and the Long Term Evolution (LTE) standard from 3GPP, formally
named Evolved UTRA (E-UTRA). There are no major differences between these
two standards on the physical layer. Both are designed for the same range of
bandwidths. The basic design is fairly similar, however details may differ. Based
on slightly different choices of values for the OFDM-parameters we might say
that Mobile WiMAX may have a better performance in “heavy” multipath, since
the symbols and cyclic prefix are longer than for E-UTRA. On the other hand,
this gives smaller sub-carrier distance which puts higher demands on frequency
accuracies and smaller tolerance for Doppler spread in the channel. Mobile
WiMAX uses more resources for pilots or reference signals which at least in
theory should give more robust performance in difficult radio conditions. On the
other hand, it consumes more radio resources and power than E-UTRA, which
can result in loss in efficiency.
Conclusively on Mobile WiMAX vs. E-UTRA, there is no clear winner when it
comes to technical performance. E-UTRA has however the advantage of being
designed for smoother co-existence with and migration from WCDMA/HSPA.
Radio planning OFDMA networks is more equal to GSM planning than WCDMA
planning. In an interference-limited scenario, intra-cell interference is basically

50 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

eliminated due to the fact that users are orthogonal to each other in either
time- or frequency domain. This is similar to GSM, where there is orthogonality
in the time-domain within the cell. It is different from WCDMA, where intra-cell
interference on the uplink is a major factor when planning for high capacity.
Main interference is from neighbour cells and the problem is to estimate
reasonable interference margins.
A few examples of link budgets for Mobile WiMAX and E-UTRA show that there
are some differences in the interference margin estimates.
Additionally, Scalable OFDMA opens a few new ideas on frequency reuse
methods. In WCDMA, a frequency reuse factor of 1 (same carrier in all cells) is
often used, but frequency planning with e.g. 3 frequencies are also possible. In
OFDMA it is possible to adjust the frequency reuse factor between 1 and e.g. 3
using fractional frequency reuse which opens up the possibility for dynamic
frequency reuse and also dynamic capacity planning. Dividing each cell in an
inner part using all frequencies (reuse 1) and outer part using 1/3 of the
frequencies (reuse 3) makes it possible to tune the different cell capacities
cooperatively, e.g. by moving capacity from one cell to another dependent on
the time of day.

Telenor R&I R 7/2008 - 51


OFDM(A) for wireless communication

References
[802.11a] IEEE 802 LAN/MAN Standards Committee. Part 11: Wireless
LAN Medium Access Control (MAC) and Physical Layer (PHY)
specifications. High-speed Physical Layer in the 5 GHz Band.
IEEE Std 802.11a-1999(R2003). IEEE, NJ, USA, June 2003.

[Alam07] Alamouti S M. Mobile WiMAX: Vision & Evolution. IEEE Mobile


WiMAX Symposium. Orlando, FL USA, 27-28 March 2007 [online
8 Feb 2008] URL: http://www.ieee-
mobilewimax.org/downloads/Alamouti.pps

[And07] Andrews J G, Ghosh A, and Muhamed R, Fundamentals of


WiMAX - Understanding Broadband Wireless Networking, ISBN:
0132225522, Prentice Hall, 2007.

[Bachl07] Bachl R, Gunreben P, Das S, Tatesh S. The Long Term Evolution


Towards a New 3GPP Air Interface Standard. Bell Labs Technical
Journal, 11(4), p25-51. 2007.

[Dahl07] Dahlman E, Parkvall S, Sköld J, Beming P. 3G Evolution. HSPA


and LTE for Mobile Broadband. Academic Press/Elsevier. Oxford,
UK. 2007. ISBN: 9780123725332.

[ECMA-368] Ecma International. High Rate Ultra Wideband PHY and MAC
Standard. ECMA-368, 1st edition. December 2005

[Ekst06] Ekström H, et al. Technical Solutions for the 3G Long-Term


Evolution. IEEE Communications Magazine. 44(3). P38-45.
March 2006.

[EN 300 744] ETSI. Digital Video Broadcasting (DVB); Framing structure,
channel coding and modulation for digital terrestrial television.
ETSI EN 300 744, V1.5.1, Nov 2004.

[Hag07] Hagström U, LTE Overview, Ericsson presentation at Telenor,


2007-09-12.

[Holma07] Holma H and Toskala A, WCDMA for UMTS – HSPA evolution


and LTE, 4th edition, ISBN: 9780470319338, John Wiley &
Sons, 2007.

[Lam07] Lamminmäki J, Mobile WiMAX radio network - Planning and


Dimensioning overview, Nokia presentation at Telenor, 2007-
09-14.

[Løv92] Løvnes G, Paulsen S E, Rækken R H. UHF radio channel


characteristics. Part two: Wideband propagation measurements
in large cells. Norwegian Telecom Research (Telenor R&I).
Research report TF R 19/92. Kjeller, Norway. 1992.

[Myu06] Myung H G, Lim J, Goodman D J. Single Carrier FDMA for Uplink


Wireless Transmission. IEEE Vehicular Technology Magazine,
3(1), p30-38. Sept. 2006.

52 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

[Myu06b] Myung H G, Lim J, Goodman D J. Peak-to Average Power Ratio


of Single Carrier FDMA Signals with Pulse Shaping. 17th IEEE
International Symposium on Personal, Indoor and Mobile Radio
Communications 2006 (PIMRC06). Helsinki, Finland, September
2006.

[Myu07] Myung H G. Introduction to Single Carrier FDMA. 15th European


Signal Processing Conference (EUSIPCO) 2007, Poznan, Poland,
Sep. 2007.

[Nee00] Van Nee R, Prasad R. OFDM for Wireless Multimedia


Communications. Artech House, Boston, MA, USA. 2000. ISBN:
0-89006-530-6.

[Nua07] Nuaymi L, WiMAX - Technology for Broadband Wireless Access,


ISBN: 9780470028087, John Wiley & Sons, 2007.

[Ræk95] Rækken R H, Løvnes G. Multipath propagation and its influence


on digital mobile communication systems. Telektronikk, 91(4),
p109-126. 1995.

[TR 25.814] 3rd Generation Partnership Project. Physical layer aspects for
evolved Universal Terrestrial Radio Access (UTRA) (Release 7).
3GPP TR 25.814, V7.1.0. Sept 2006.

[TR 25.913] 3rd Generation Partnership Project. Requirements for Evolved


UTRA (E-UTRA) and Evolved UTRAN (E-UTRAN) (Release 7).
3GPP TR 25.913, V7.3.0. March 2006.

[TS 36.104] 3rd Generation Partnership Project. Evolved Universal Terrestrial


Radio Access (E-UTRA); Base Station (BS) radio transmission
and reception (Release 8). 3GPP TS 36.104, V8.0.0. Dec 2007

[TS 36.211] 3rd Generation Partnership Project. Evolved Universal Terrestrial


Radio Access (E-UTRA); Physical channels and modulation
(Release 8). 3GPP TS 36.211, V8.1.0. Nov 2007.

[TS 36.212] 3rd Generation Partnership Project. Evolved Universal Terrestrial


Radio Access (E-UTRA); Multiplexing and channel coding
(Release 8). 3GPP TS 36.212, V8.1.0. Nov 2007.

[UMB] 3rd Generation Partnership Project 2. Physical Layer for Ultra


Mobile Broadband (UMB) Air Interface Specification. 3GPP2
C.S0084-001-0, Version 2.0. August 2007

[WiMAX06] WiMAX Forum, Mobile WiMAX - Part I: A Technical Overview and


Performance Evaluation, August 2006, available at:
http://www.wimaxforum.org/technology/downloads/

[WiMAX07] WiMAX Forum, A Comparative Analysis of Mobile WiMAX™


Deployment Alternatives in the Access Network, May 2007,
available at:
http://www.wimaxforum.org/technology/downloads/

Telenor R&I R 7/2008 - 53


OFDM(A) for wireless communication

Annex 1. Abbreviations
3GPP 3rd Generation Partnership Project
3GPP2 3rd Generation Partnership Project 2
16-QAM 16-state Quadrature Amplitude Modulation
64-QAM 64-state Quadrature Amplitude Modulation
AAS Advanced Antenna System
ACI Adjacent Channel Interference
AMC Adaptive Modulation and Coding
BER Bit Error Rate
BPSK Binary Phase Shift Keying
CCDF Complementary Cumulative Distribution Function
CCI Co-Channel Interference
CDF Cumulative Distribution Function
COFDM Coded OFDM
CP Cyclic Prefix
CSMA Carrier Sense Multiple Access
DL Downlink
DS Delay Spread
DVB-H Digital Video Broadcast Handheld
DVB-T Digital Video Broadcast Terrestrial
E-UTRA Evolved UMTS Terrestrial Radio Access
FDD Frequency Division Duplex
FDM Frequency Division Multiplex
FDMA Frequency Division Multiple Access
FEC Forward Error Correction
FFT Fast Fourier Transform
FRF Frequency Reuse Factor
FUSC Full Usage of Sub-Carriers
HSPA High Speed Packet Access
ICI Inter Carrier Interference
IEEE Institute of Electrical and Electronics Engineers
ISI Inter Symbol Interference
ITU-R International Telecommunication Union – Radio communications sector
LCR Low Chip Rate
LTE Long Term Evolution
MB-OFDM Multi-Band Orthogonal Frequency Division Multiplex
MCS Modulation and Coding Schemes
MS Mobile Station
OFDM Orthogonal Frequency Division Multiplexing
OFDMA Orthogonal Frequency Division Multiple Access
PAPR Peak to Average Power Ratio
PDP Power Delay Profile
PUSC Partial Usage of Sub-Carriers
QPSK Quadrature Phase Shift Keying
SC Sub-Carrier
SC-FDMA Single Carrier FDMA
SFN Single Frequency Network
SIR Signal to Interference Ratio
SM Spatial Multiplexing
SNIR Signal to Noise and Interference Ratio
S-OFDMA Scalable OFDMA
STC Space Time Coding

54 - Telenor R&I R 7/2008


OFDM(A) for wireless communication

TDD Time Division Duplex


TDMA Time Division Multiple Access
TFC Time Frequency Code
TUSC Tile Usage of Sub-Carriers
UL Uplink
UMB Ultra Mobile Broadband
UTRA Universal Terrestrial Radio Access
UWB Ultra Wideband
WCDMA Wideband Code Division Multiple Access
WiMAX Worldwide Interoperability for Microwave Access
WLAN Wireless Local Area Network
WMAN Wireless Metropolitan Area Network
WPAN Wireless Personal Area Network

Telenor R&I R 7/2008 - 55