Page 1 of 24
`
`Seoseees8BatBo a
`
`
`t
`Zz
`
`O 1
`
`) a L
`
`i
`Qo
`
`RICHARD VAN NEE
`
`= e 0
`
`)
`ag
`Li
`as
`ye
`o
`
`RAMJEE PRASAD
`
`OFDM
`
`FOR WIRELESS
`
`MULTIMEDIA
`
`COMMUNICATIONS
`
`9)
`MA
`
`O b
`
`ee
`
`< O
`
`U A m
`
`a = = O U a a
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 1 of 24
`
`

`

`
`
`
`
`OFDM Wireless Multimedia
`Communications
`
`Richard van Nee
`Ramjee Prasad
`
`
`
`nunications
`
`
`
`|
`
`IM
`
`Artech House
`Boston e London
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 2 of 24
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`International Standard Book’ Number: 0-89006-530-6
`Library of Congress Catalog Card Number: 99-052312
`
`
`10987654321
`
`
`
`
`
`Tc
`
`Library of Congress Cataloging-in-Publication Data
`Nee, Richard van.
`OFDMfor wireless multimedia communications / Richard van Nee, Ramjee Prasad
`p. cm. — (Artech House universal personal communicationslibrary)
`Includes bibliographical references and index.
`ISBN 0-89006-530-6 (alk. paper)
`1. Wireless communications systems. 2. Multimedia systems. 3. Multiplexing.
`Prasad, Ramjee.II. Title III. Series
`
`I.
`
`TK5103.2.N44 2000
`621/3845—dc21
`
`99-052312
`* CIP
`
`British Library Cataloguing in Publication Data
`Nee, Richard van
`OFDMwireless multimedia communications. — (Artech House
`universal personal communicationslibrary)
`1.Wireless communications systems 2. Multimedia systems
`I. Title II. Prasad, Ramjee
`621.382
`
`ISBN 0-89006-530-6
`
`Cover design by Igor Vladman
`
`© 2000 Richatd van Nee and Ramjee Prasad
`
`
`
`
`
`All rights reserved. Printed and bound in the United States of America. No part of this book
`may be reproducedorutilized in any form or by any means,electronic or mechanical, includ-
`ing photocopying, recording, or by any information storage andretrieval system, without per-
`mission in writing from the authors.
`All terms mentioned in this book that are known to be trademarks or service marks have
`been appropriately capitalized. Artech House cannotattest to the accuracy of this informa-
`tion. Use of a term in this book should not be regarded as affecting the validity of any trade-
`mark or service mark.
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 3 of 24
`
`

`

`if
`
`
`|
`
`|
`
`Contents
`
`Preface
`Acknowledgments
`
`Chapter 1
`1.1
`Led,
`
`1.3
`
`1.4
`1.5
`1.6
`References
`Chapter 2
`21
`Ded
`2.3
`2.4
`29
`2.6
`2.7
`
`/
`Introduction
`Standardization and Frequency Bands
`Multimedia Communications
`1.2.1 The Need for High Data Rates
`1.2.2
`Services and Applications
`1.2.3 Antennas and Batteries
`1.2.4 Safety Considerations
`1.2.5 ATM-Based Wireless (Mobile) Broadband
`Multimedia Systems
`Multipath Propagation
`1.3.1 Multipath Channel Models
`1.3.2 Delay Spread Values
`Time Variation of the Channel
`History of OFDM
`Preview of the Book
`
`OFDM Basics
`Introduction
`Generation of Subcarriers using the IFFT
`Guard Time and Cyclic Extension
`Windowing
`Choice of OFDM Parameters
`OFDMSignal Processing
`Implementation Complexity of OFDM Versus
`Single Carrier Modulation
`
`vii
`
`xiii
`xvii
`
`1
`4
`7
`8
`9
`9
`10
`
`12
`15
`16
`17
`19
`20
`24
`25
`33
`33
`33
`39
`42
`46
`47
`
`48
`
`|
`
`i
`
`Page 4 of 24
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 4 of 24
`
`

`

`
`
`Vili
`
`References
`
`
`
`51
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Chapter 5
`95
`Coherent and Differential Detection
`
`
`Introduction
`5.1
` Coherent Detection
`5.2
` 5.2.1 Two Dimensional Channel Estimators
` 5.2.2 One Dimensional Channel Estimators
`
`103
`
`
`104
`5.2.3 Special Training Symbols
` 5.2.4 Decision Directed Channel Estimation
` Differential Detection
`
`Differential Detection in the Time Domain
`
`Sensitivity to Timing Errors
`Synchronization using the Cyclic Extension
`Synchronization using Special Training Symbols
`Optimum Timing in the Presence of Multipath
`
`53
`53
`54
`
`54
`
`55
`
`58
`
`59
`
`60
`
`62
`
`70
`
`23
`
`73
`
`74
`
`77
`
`78
`80
`86
`88
`92
`
`
`
`Chapter3
`3.1
`3,2
`
`Coding and Modulation
`Introduction
`Forward Error Correction Coding
`
`.
`
`3.2.1 Block Codes
`
`3.2.2 Convolutional Codes
`
`3.2.3 Concatenated Codes
`
`Interleaving
`
`Quadrature Amplitude Modulation
`
`Coded Modulation
`
`3.3
`
`3.4
`
`3.5
`
`References
`
`Chapter 4
`
`Synchronization
`
`Introduction
`
`Sensitivity to Phase Noise
`
`Sensitivity to Frequency Offset
`
`4.1
`
`4.2
`
`4.3
`
`4.4
`4.5
`4.6
`4.7
`References
`
`5.3.1
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 5 of 24
`
`

`

`5.3.2 Differential Detection in the Frequency Domain
`
`5.3.3 Differential Amplitude and Phase Shift Keying
`
`References
`
`Chapter 6
`6.1
`
`The Peak Power Problem
`
`Introduction
`
`6.2
`
`6.3
`
`6.4
`
`6.5
`
`Distribution of the Peak-to-Average Power Ratio
`
`Clipping and Peak Windowing
`6.3.1 Required Backoff with a Non-Ideal Power Amplifier
`6.3.2 Coding and Scrambling
`
`Peak Cancellation
`
`PAP Reduction Codes
`
`6.5.1 Generating Complementary Codes
`6.5.2 Minimum Distance of Complementary Codes
`6.5.3 Maximum Likelihood Decoding of Complementary Codes
`6.5.4 Suboptimum Decoding of Complementary Codes
`
` 1
`
`51
`
`53
`
`53
`
`54
`
`54
`
`53
`
`58
`
`59
`
`60
`
`62
`
`70
`
`73
`
`73
`
`74
`
`77
`
`78
`
`80
`
`86
`
`88
`
`92
`
`95
`
`95
`
`95
`
`96
`
`103
`
`104
`
`106
`
`107
`
`107
`
`
`
`
`112
`
`115
`
`117
`
`119
`
`119
`
`120
`
`123
`
`127
`
`130
`
`131
`
`138
`
`141
`
`144
`
`145
`
`147
`
`150
`
`150
`
`153
`
`155
`
`155
`
`156
`
`157
`
`161
`
`VL
`
`171
`
`172
`
`173
`
`175
`
`175
`
`6.5.5 Large Code Lengths
`
`6.6
`
`SYMBOLScrambling
`
`References
`
`Chapter 7
`
`Basics of CDMA
`
`71
`
`7.2
`
`73
`
`Introduction
`
`CDMA:Past, Present, and Future
`
`CDMAConcepts
`7.3.1 Pure CDMA
`
`74
`
`Basic DS-CDMA Elements
`
`-
`
`7.4.1 RAKE Receiver
`
`74.2 Power Control
`
`7.4.3 Soft Handover
`
`7.4.4
`
`Interfrequency Handover
`
`74.5 Multiuser Detection
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 6 of 24
`
`

`

`Chapter 8
`8.1
`
`8.2
`
`8.3
`
`8.4
`
`8.5
`
`8.6
`
`8.7
`
` References
`
`
`
`176
`
`179
`
`179
`
`180
`
`182
`
`182
`
`185
`
`189
`
`194
`
`194
`
`195
`
`196
`
`197
`
`197
`
`199
`
`203
`
`206
`
`208
`
`209
`
`213
`
`213
`
`213
`
`215
`
`217
`
`220
`
`220
`
`221
`
`221
`
`222
`
`Multi - Carrier CDMA
`
`Introduction
`
`Channel Model
`
`DS-CDMAand MC-CDMASystems
`
`8.3.1 DS-CDMA System
`
`8.3.2. MC-CDMA System
`
`MC-CDMaASystem Design
`BEP LOWERBound
`
`8.5.1 DS-CDMA System
`
`8.5.2 MC-CDMA System
`
`8.5.3. BEP Lower Bound Equivalence
`
`Numerical Results
`
`8.6.1 MC-CDMASystem Design
`
`8.6.2 Down - Link BEP Performance
`
`8.6.3 Up - Link BER Performance
`Conclusions
`
`Orthogonal Frequency Division Multiple Access
`Introduction
`
`Frequency Hopping OFDMA
`Differences between OFDMA and MC-CDMA
`
`OFDMA System Description
`
`9.4.1 Channel Coding
`
`9.4.2 Modulation
`
`9.4.3 Time and Frequency Synchronization
`
`9.4.4
`
`Initial Modulation Timing Synchronization
`
`9.4.5
`
`Initial Frequency Offset Synchronization
`
`Appendix 8A
`References
`
`Chapter 9
`
`9.1
`
`9.2
`
`9.3
`
`9.4
`
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 7 of 24
`
`

`

`
`
`9.4.6
`
`Synchronization Accuracy
`
`9.4.7
`
`Power Control
`
`9.4.8 Random Frequency Hopping Operation
`
`9.4.9 Dynamic Channel Allocation (Fast DCA)
`
`9.4.10 Dynamic Channel Allocation ( Simple DCA )
`
`9.4.11 Capacity of OFDMA
`
`9.5
`
`Conclusions
`
`References
`
`Chapter 10
`
`Applications of OFDM
`
`10.1
`
`10.2
`
`10.3
`
`10.4
`
`Introduction
`
`Digital Audio Broadcasting
`
`Terrestrial Digital Video Broadcasting
`
`Magic WAND
`
`10.4.1 Magic WAND Physical Layer
`
`10.4.2 Coding
`
`10.4.3 Simulated Error Probabilities
`
`10.4.4 Effects of Clipping
`
`10.4.5 Magic WAND Medium Access Control Layer
`
`10.5
`
`TEEE 802.11, HIPERLAN/2, and MMAC Wireless LAN Standards
`
`10.5.1 OFDM Parameters
`
`10.5.2 Channelization
`
`10.5.3 OFDM Signal Processing
`
`10.5.4 Training
`
`10.5.5 Differences between IEEE 802.11, HIPERLAN/2
`and MMAC
`
`10.5.6 Simulation Results
`
`References
`
`About the Authors
`
`Index
`
`222
`
`223
`
`224
`
`225
`
`227
`
`227
`
`227
`
`228
`
`229
`
`229
`
`229
`
`231
`
`233
`
`234
`
`236
`
`236
`
`237
`
`238
`
`241
`
`243
`
`244
`
`245
`
`246
`
`249
`
`250
`
`252
`
`255
`
`257
`
`176
`
`179
`
`179
`
`180
`
`182
`
`182
`
`185
`
`189
`
`194
`
`194
`
`195
`
`196
`
`197
`
`197
`
`199
`
`203
`
`206
`
`208
`
`209
`
`213
`
`213
`
`213
`
`215
`
`217
`220
`
`220
`
`221
`
`221
`
`222
`
`
`
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 8 of 24
`
`

`

`CHAPTER9
`
`
`Orthogonal Frequency Division Multiple Access
`
`
`
`
`
`
`INTRODUCTION 9.1
`
`The previous chapter described some ways in which OFDM could be used both as a
`
`modulation scheme and as part of the multiple access technique, by applying a
`
`spreading code in the frequency domain. In this chapter, a variation on this theme is
`
`described, namely orthogonal
`frequency division multiple access
`(OFDMA).
`In
`
`OFDMA,multiple access is realized by providing each user with a fraction of the
`
`available numberof subcarriers. In this way, it is equal to ordinary frequency division
`
`multiple access (FDMA); however, OFDMAavoids the relatively large guard bands
`
`that are necessary in FDMAto separate different users. An example of an OFDMA
`
`time-frequency grid is shown in Figure 9.1, where seven users a to g each use a certain
`
`fraction—which may be different for each user—of the available subcarriers. This
`
`particular example in fact is a mixture of OFDMA and Time Division Multiple Access
`
`(TDMA), because each user only transmits in one out of every four timeslots, which
`
`may contain one or several OFDM symbols.
`
`
`
`
`
`9.2
`FREQUENCY-HOPPING OFDMA
`
`In the previous example of OFDMA,every user had a fixedset of subcarriers. It is a
`relatively easy change to allow hoppingof the subcarriers per timeslot, as depicted in
`
`Figure 9.2. Allowing hopping with different hopping patterns for each user actually
`
`transforms the OFDMA system in a frequency-hopping CDMAsystem. This has the
`
`benefit of an increased frequency diversity, because each useruses all of the available
`
`bandwidth, as well as the interference averaging benefit that is common for all CDMA
`
`variants. By using forward-error correction coding over multiple hops, the system can
`
`correct for subcarriers in deep fades or subcarriers that are interfered by other users.
`
`Because the interference and fading characteristics change for every hop, the system
`
`
`
`
`Page 9 of 24
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 9 of 24
`
`

`

`
`
`
`
`
`
`
`
`214
`
`
`performance dependson the average received signal power and interference, rather than
`on the worst case fading and interference power.
`
`Figure 9.1 Example of the time-frequency grid with seven OFDMAusers,a to g, whichall have a fixed
`set of subcarriers every four timeslots.
`
`Time
`
`Time
`
`Frequency
`Frequency
`
`
`
`
`
`seepPELOTON
`
`Figure 9.2 Exampleofthe time-frequency grid with three hopping users,a, b and c, which all have one
`hop every four timeslots.
`
`A major advantage of frequency-hopping CDMA systemsover direct-sequence
`or multicarrier CDMA ‘systems is that it
`is relatively easy to eliminate intracell . _
`interference by using orthogonal hopping patterns within a cell. An example of such an
`orthogonal hoppingset is depicted in Figure 9.3. For N subcarriers;it is always possible
`
`to construct N orthogonal-hopping patterns. Some useful construction rules for
`
`generating hopping patterns can be foundin [1].
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 10 of 24
`
`

`

`
`
`
`
`i
`
`\
`i
`
`DIFFERENCES BETWEEN OFDMA AND MC-CDMA
`9.3.
`_ The main differences between OFDMAand the MC-CDMAtechniques discussed in
`Chapter 8 is that users within the same cell use a distinct set of subcarriers, while in
`MC-CDMA,all users use all subcarriers simultaneously. To distinguish different users,
`orthogonal or near-orthogonal spreading codes are used in MC-CDMA.Because of
`code distortion by multipath fading channels, however, MC-CDMA loses_its
`orthogonality in the uplink even for a single cell. This makes rather complicated
`equalization techniques necessary, which introduce a loss in SNR performance and
`diminish the complexity advantage of OFDM over single-carrier techniques. OFDMA
`does not have this disadvantage, because in a single cell, all users have different
`subcarriers,
`thereby eliminating the possibility of
`intersymbol or
`intercarrier
`interference. Hence, OFDMA doesnotsuffer from intracell interference, provided that
`the effects of frequency and timing offsets between users are kept at a sufficiently low
`level. This is a major advantage of OFDMA compared with MC-CDMA and DS-
`CDMA, because in those systems,
`intracell
`interference is the main source of
`
`215
`
`
`ar and interference, rather than
`
`Figure 9.3 Example of six orthogonal hoppingpatterns with six different hopping frequencies.
`
`. users, a to g, whichall have a fixed
`
`Frequency
` interference. A typical ratio of intracell and intercell interference is 0.55 [2]. Because
`
`isers, a, b and c, whichall have one
`
`the system capacity is inversely proportional to the total amount of interference power,
`a capacity gain of 2.8 can be achieved by eliminating all intracell interference,so thatis
`the maximum capacity gain of OFDMAover DS-CDMA and MC-CDMA networks.
`The main advantages of using CDMAin general or MC-CDMAin particular is
`interference averaging. In CDMA,theinterference consists of a much larger numberof
`k
`- Systems over direct-sequence
`interfering signals than the interference in a non-CDMA system.Each interfering signal
`y easy to eliminate intracell .
`is subject to independentfading, caused by shadowing and multipath effects. For both
`a cell. Anexample of such an
`the CDMA and the non-CDMAsystem, an outage occurs whenthe total interference
`ibearriers; it is always possible
`power(after despreading for the CDMA system) exceeds some maximum value. In a
`iseful construction rules for
`=
`non-CDMAsystem,the interference usually consists of a single or a few cochannel
`;
`interferers. Because of fading, the interference poweris fluctuating over a large range,
`ey* ootrere
`yty
`
`
`
`
`
`
`il
`
`
`Page 11 of 24
`
`|
`
`i
`i
`|
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 11 of 24
`
`

`

`
`
`SadntsonNDmEENEanTEaPeEPmEEEPrEsenseenaenetmenmnmtnDanpemcreneenperemnrsnenraramereremeemeeemermnememens
`
`216
`
`so a large fading margin has to be taken into account, which reduces system capacity.
`In a CDMA system,the interference is a sum of a large numberofinterfering signals.
`Because all these signals fade independently, the fluctuation in the total interference
`poweris muchless than the powerfluctuation of a single interfering signal. Hence,in a
`CDMAsystem the fading margin can be significantly smaller than the margin for a
`non-CDMAsystem. This improvement in margin largely determines the capacity gain
`of a CDMAsystem.
`In OFDMA,interference averaging is obtained by having different hopping
`patterns within each cell. The hopping sequences are constructed in such a way that two
`users in different cells interfere with each other only during a small fraction of all hops.
`In a heavily loaded system, many hops will interfere, but the interference will be
`different for each hop. Hence, by forward-error correction across several hops, the
`OFDMA performance will be limited by the average amountof interference rather than
`the worst case interference. An additional advantage of OFDMA over DS-CDMA and
`MC-CDMAis that there are some relatively simple ways to reduce the amount of
`intercell interference. For instance, the receiver can estimate the signal quality of each
`hop and use this information to give heavily interfered hops a lower weight in the
`decodingprocess.
`
`
`SneeeeeaneEneee
`
`Onorton OESEEEAESTERETERENit“yeeaAEfHaeTer
`
`eae
`
`1
`
`|
`
`i
`
`_—
`
`Another important feature of CDMAis the possibility to perform soft handover
`by transmitting two signals from different base stations simultaneously on the same
`channel to one mobile terminal. Combining the signals from different base stations
`gives a diversity gain that significantly reduces the fading margin, because the
`probability that two base stations are in a fade is much smaller than the probability that
`one base station is in a fade. Less fading means that less power has to be transmitted,
`and henceless interference is generated, which gives an improvementin the capacity of
`the system. A nice feature of CDMAsoft handoveris that it has no impact on the
`complexity of the mobile terminal; as far as the mobile terminal is concerned, the
`overlapping signals of different base stations have the same effect as overlapping
`signals caused by multipath propagation.
`
`For OFDMAsystems, two basic soft handover methodsexist, applicable to both
`the uplink and the downlink.
`(base station-to-mobile, mobile-to-base station). A
`requirement for both methodsis that the transmissions from andto the basestations are
`synchronized such that the delay differences at the two base stations are well within the
`guard time of the OFDM symbols.
`The first technique is to use the same set of subcarriers and the same hopping
`sequence in two cells to connect to two base stations. Hence, in the downlink the
`mobile receives a sum of two signals with identical data content. The mobile is not able
`to distinguish between the two basestations; the effect of soft handover is similar to
`that of adding extra multipath components, increasing the diversity gain. This type of
`soft handoveris similar to soft handover in DS-CDMAnetworks.
`
`The second way for soft handoff is to use different sets of subcarriers in two
`cells. In contrast to the first method, in the downlink the mobile has to distinguish now
`
`Page 12 of 24
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 12 of 24
`
`

`

`between the two base stations. It has to demodulate the signals from the two base
`stations separately, after which they can be combined, preferably by using maximal
`ratio combining. This type of soft handoveris similar to the one that could be used in a
`non-CDMAnetwork.
`
`
`
`
`Advantages of the second method over the first—in the downlink—are an
`increased SNR gain because of receiver diversity, and more freedom for the base
`stations to allocate available subcarriers. In the first method, base stations are forced to
`use the same subcarriers. A main advantage of the first method is its simpler
`implementation; no additional hardware is needed, only some extra protocol features to
`connect to two base stations simultaneously. The second method does require extra
`hardware, because it has to demodulate an extra set of subcarriers. Further, it has to
`perform extra processing for the maximal ratio combining of the signals from the
`
`different basestations.
`
`
`
`
`
`
`
`OFDMA SYSTEM DESCRIPTION
`9.4
`
`As an example of an OFDMAsystem,this section gives a description of a system that
`wasproposed for the European UMTS[3, 4]. Table 9.1 summarizes the parameters and
`key technical characteristics of this OFDMAair interface.
`
`
`
`
`Table 9.1
`Parameters of the Proposed OFDMAsystem
`
`Parameter
`
`4.1666 kHz
`288.46ps
`:
`
`bandslot of 100 kHz
`
`time slots and 1 bandslot
`
`s|6|b
`
`andslot and 1 time slot (= 1 symbol)
`
`Figure 9.4 illustrates the time-frequency grid of the OFDMA system. The
`resources (time and frequency) are allocated based on the type of services and
`operational environment. The numberoftimeslots and bandslots per user is variable to
`realize variable data rates. The smallest data rate is obtained for one bandslot of 24
`
`
`subcarriers per timeslot of 288.46 us.
`
`
`
`‘ent sets of subcarriers in two
`mobile has to distinguish now
`
`Page 13 of 24
`
`
`
`
`217
`
`
`
`
`
`
`iich reduces system capacity,
`aumberof interfering signals.
`ition in the total interference
`interfering signal. Hence, in a
`maller than the margin for a
`determines the capacity gain
`
`by having different hopping
`tructed in such a waythat two
`ig a small fraction of all hops.
`but the interference will be
`tion across several hops, the
`unt of interference rather than
`JFDMA over DS-CDMA and
`ays to reduce the amount of
`late the signal quality of each
`. hops a lower weight in the
`
`ility to perform soft handover
`. simultaneously on the same
`. from different base stations
`fading margin, because the
`aaller than the probability that
`; powerhasto be transmitted,
`mprovementin the capacity of
`that it has no impact on the
`le terminal is concerned, the
`> same effect as overlapping
`
`tthods exist, applicable to both
`, mobile-to-base station). A
`ym andto the base stations are
`ise stations are well within the
`
`carriers and the same hopping
`Hence, in the downlink the
`‘ontent. The mobile is not able
`of soft handover is similar to —
`\e diversity gain. This type of
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 13 of 24
`
`

`

`Page 14 of 24
`
`e Use of frequency-hopping OFDMA for interference averaging and frequency
`diversity;
`e Time-division duplex MAC with dynamic channel allocation used for unpaired
`spectrum allocations, asymmetrical services, and unlicensed usage;
`e Straightforward and efficient high bit rate support by allocating more subcarriers
`and/or timeslots;
`Small guard band requirements at approximately 100 kHz;
`e
`e No frequency planning option available; effective re-use factorof1;
`e GSM backwards compatibility; and
`e Minimum bandwidth requirements for system deployment only 1.6 MHz (orless)
`and deploymentpossible in steps of 100kHz.
`
`218
`
`
`Subcarrier
`Spacing =
`4.166kHz
`
`1 Bandslot =
`100kHz
`
`|||
`
`I|
`
`Frequency
`
`
`
`
`
`
`Timeslot
`288,46's
`
`Time
`
`Figure 9.4 Time-frequency grid.
`
`The following summary shows some advantages of the proposed OFDMA
`
`system:
`
`Figure 9.5 shows the TDMA framestructure. Each frame is of length 4.615 ms,
`which is divided into 4 subframes of length 1.1534 ms. A sub-frame contains 4 time-
`slots of duration 288.46 ps. The timeslot contains a guard period, power control
`information and data. Every OFDM symbolis mappedonto onetime-slot. The structure
`of an OFDMAsymbolis depicted in Figure 9.6.
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 14 of 24
`
`

`

`a
`
`219
`
`¢ 4.615ms frame
`
`
` « 4 frames (18.46ms) Interleave _»
`
`a—
`4 frames (18.46ms) Interleave
`>
`TT
`1.1534mswring
`
`SSS
`
`
`
`
`
`
`
`288.46[us] time
`
` Guard period
`
`s of the proposed OFDMA
`
`Time Alignment
`
`‘e averaging and frequency
`
`allocation used for unpaired
`
`7 allocating more subcarriers
`
`Figure 9.5 Frame (TDMA)Structure.
`
`Modulation Period (288.46ys)
`
`
`Effective Modulation Period (240ys)
`
`
`
`
`Pre-Guard Time ( ... ys)
`Ramp Time (10ps)
`
`Ramp Time (10us)
`Post-Guard Time (... ys)
`
`
`
`ment only 1.6 MHz (orless)
`
`Figure 9.6 OFDM modulationburst.
`
`i frame is of length 4.615 ms,
`\ sub-frame contains 4 time-
`suard period, power control
`o one time-slot. The structure
`
`The whole system frequency band is divided into small blocks (bandslots) with
`a fixed number of subcarriers. To maintain compatibility with GSM, a .100-kHz
`bandslot is chosen that consists of 24 subcarriers. Therefore, the subcarrier spacing is
`100/24 = 4.167 kHz. Figure 9.7 shows the OFDMAfrequencystructure.
`In each bandslot,
`the two subcarriers at the edge of the bandslot are left
`unmodulated to relax receiver blocking requirements. In addition, the interference of
`two adjacent blocks of subcarriers
`is
`reduced, which may occur when their
`orthogonality is compromised becauseof nonlinear PA effects. Adjacent bandslots can
`be concatenated to allow transmission of widebandservices.
`
`
`
`
`
`
`
`Page 15 of 24
`
`
`
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 15 of 24
`
`

`

`
`
`1
`
`|I
`
`\
`|
`
`System Band
`
`(4.17kHze24sub-carriers)
`
`(4.17kHz*24sub-carriers)
`
`uard carrier! Guard carrier
`
`100kHz
`Guard band
`Guard band
`|
`>
`1 >
`
`
`
` NAN
`
`MAAKA+>
`
`OFDM sub-carrier spacing = 4.17kHz
`
`Figure 9.7 OFDMAfrequencystructure.
`
`| j
`
`
`
`
`Seeeeee
`
`
`
`9.4.1 Channel Coding
`
`Convolutional encoding and soft-decision Viterbi decoding is used for the basic data
`transmission. The objective of this coding is to achieve good quality in the tough
`mobile radio channel. A constraint length of seven is used together with variable
`coderates in the range of 1/4 to 3/4. To achieve very low bit-error rates (e.g. 10°) for
`video encoding or data transmission, a concatenated coding scheme is used with an
`inner convolutional code and an outer Reed-Solomon code.
`
`9.4.2 Modulation
`
`The modulation schemes of the OFDMA proposal are QPSK and 8-PSK with
`differential encoding in the frequency domain. An optional coherent mode with 16-
`QAM is available, which uses pilot subcarriers to obtain a channel estimate at the
`receiver. For differential encoding, each bandslot contains one known reference
`subcarrier value, as depicted in Figure 9.8.
`
`
`
`
`seeeeenieeietennettmnrenmememeenecenesneneeeeeeetintata
`fpsentangleereinestereee
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 16 of 24
`
`

`

`
`
`
` Differential Encoding
`Reference Subcarrier
`
`
`
`
`
`Figure 9.8 Reference subcarrierallocation.
`
`
`
`9.4.3 Time and Frequency Synchronization
`
`
` Synchronization is an essential issue for the OFDMAsystem. The following aspects are
`considered for uplink and downlink:
`initial modulation timing synchronization,
`
`
`modulation timing tracking, initial frequency offset synchronization, and frequency
`tracking.
`
`
`
` 9.4.4
`Initial Modulation Timing Synchronization
`
`Initial timing synchronization is required to adjust the mobile station’s internal timing
`
`to the base station’s timeframe. After switching on, the mobile station monitors the
`initial acquisition channel (LACH)andthe broadcast channel (BCCH).
`After the mobile has detected the base station’s timing, it sends a random access
`channel (RACH) packet to the base station. The base station measures the time offset
`for the received RACH packet and sends back the necessary timing advance to the
`mobile (similar to GSM). In the frame structure of the OFDMA system,reserved slots
`for reception of RACH packetsexist.
`Because of the time and frequency structure of the OFDMAsystem,the timing
`trackingis less critical compared with other OFDM systems whereusersare interleaved
`in the frequency domain. The base station can measure the position of the received
`OFDMburst within the allocated slot for each mobile station individually and send the
`according timing alignment information back to the mobile station. In addition, timing
`information can be refined and tracked after the transformation in the subcarrier
`domain, where a time shift is observed as a phaserotation.
`
`2 Bandslot = 200kHz
`
`
`
`
`
`
`
`
`
`
`
`1g is used for the basic data
`> good quality in the tough
`used together with variable
`bit-error rates (e.g. 10°) for
`ing scheme is used with an
`
`re QPSK and 8-PSK with
`nal coherent mode with 16-
`n a channel estimate at the
`tains one known reference
`
`
`
`
`
`
`
`
`
`
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 17 of 24
`
`

`

`222
`
`
`In the mobile station, the timing information is obtained and adjusted by the
`above-mentioned correlation algorithm. Accurate timing information is required to
`determine the position of the useful data samples within each burst so the FFT window
`can be placed correctly. The guard samples relax the requirement for accurate timing
`because the position of the FFT window can beshifted within the guard time without
`performance degradation. Additional timing offset correction can be performed to cope
`with the FFT window misplacements.
`
`:
`
`;
`
`:
`
`:
`;
`
`;
`
`;
`
`;
`
`;
`
`;
`;
`
`
`
`In the downlink, only an IACH is multiplexed to allow fast and precise initial
`timing and frequency synchronization. In the actual communication mode, timing and
`frequency tracking can be performed using a correlation-based synchronization
`algorithm. In the uplink, the rough timing offset is detected by the base station by
`measuring the arrival time of the RACHburst. This gives an initial time-advance value
`that is reported back to the mobile. During communication, the arrival time of the burst
`is detected by the base station using the proposed tracking algorithm (same as in the
`downlink) or a tracking algorithm in the frequency (subcarrier) domain, based on the
`detected constellation rotation.
`
`9.4.5
`
`Initial Frequency Offset Synchronization
`
`After initial timing synchronization of the mobile station, the frequency offset can be
`measured by phase comparison of the (ideally) equal time samples within each burst.
`Equal samples are placed in the guard interval of the OFDM burst. A phaserotation
`indicates a frequency offset. Using this technique, a frequency error up to half the
`subcarrier spacing can be detected. The initial offset, however, can be larger, so it has
`to be detected using the specially designed symbols in the IACH channel.
`
`9.4.6 Synchronization Accuracy
`
`The proposed synchronization, acquisition, and tracking algorithm is independent of the
`modulation scheme (coherent or noncoherent). For coherent 16-QAM reception, further
`processing in the frequency (subcarrier domain)
`is possible to improve the
`performance. Frequency domain time tracking (or combined time-domain/frequency-
`domaintracking algorithms) can be based on observing phase shifts of the knownpilots
`within the time-frequency gridon the subcarrier domain.
`
`Both algorithms can also be combined. The alignment values are calculated
`regularly and reported to the mobile station. Accuracy requirementsare relaxed because
`the design of the burst allows some overlapping arrival (another advancing feature of
`the raised cosine pulse-shaping besides the reduction of out-of-band emission). In
`addition, the guard time helps to compensate timing misalignments. The OFDMAburst
`design provides a guardinterval at the front and an additional guard interval at the back
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 18 of 24
`
`

`

`223
`
`aSess —_wN
`
`btained and adjusted by the
`information is required to
`ich: burst so the FFT window
`iirement for accurate timing
`ithin the guard time without
`on can be performed to cope
`
`, the frequency offset can be
`e samples within each burst.
`*DM burst. A phase rotation
`quency error up to half the
`‘ever, can be larger, so it has
`
`of the OFDM symbol—see Figure 9.9—which provides robustness against a timing
`inaccuracy of +10 us.
`
`
`Ramp Time (10us)
`Pre-Guard Time (28ys)
`
`
`Post-Guard Time (28ps)
`Ramp Time (10ps)
`
`Optimum receiver FFT window
`
`No degradation
`
`‘Delayed’ receiver FFT window
`
`No degradation
`
`‘Early’ receiver FFT window
`'Too Early' receiver FFT window
`
`No degradation
`Small degradation
`
`Figure 9.9 OFMDAburst and synchronization requirements.
`
`
`enoattaaGtAAEETEERIEktnyEAI,ceAARIESSoSTARA
`
`
`Modulation Period (288.46ps) Effective Modulation Period (240s)
`
`9.4.7 Power Control
`
`Powercontrol in the uplink removes the unevennessof received signal strength at the
`base station side and decreases the total power to the minimum level required to
`support the specified quality of service (e.g. BER). The accuracyis less critical than
`CDMA because with OFDMA,orthogonality is always provided within one cell. A
`precise power control, however, not only improves the transmission performance ‘but
`also minimizes the interference to othercells and therefore increases overall capacity.
`The OFDMAconcept uses both closed-loop and open-loop power control.
`Based on quality parameters, measured ona slot-by-slot basis, the power is adjusted in
`the mobile as well as in the basestation transmitter. Each receiver measures the quality
`of the received burst (C/I ratio) and transmits in the next burst a request to the opposite
`transmitter to increase, keep, or decrease the powerlevel in steps of 1 dB. For the
`fastest power control mode, one subcarrier is dedicated to carry power control
`information, and the poweris then adjusted on a frame-by-framebasis (each 1.152 ms).
`Figure 9.10 depicts the power control operation.
`
`gorithm is independent of the
`at 16-QAM reception, further
`possible to improve the
`ined time-domain/frequency-
`ase shifts of the known pilots
`
`allow fast and precise initial
`nunication mode, timing and
`lation-based synchronization
`scted by the base station by
`an initial time-advance value
`1, the arrival time of the burst
`ng algorithm (same as in the
`carrier) domain, based on the
`
`mment values are calculated
`jirements are relaxed because
`‘another advancing feature of
`of out-of-band emission). In
`ignments. The OFDMA burst
`mal guard interval at the back
`
`
`Page 19 of 24
`
`
`
`EX. 1017
`Ericsson v. Intellectual Ventures
`Page 19 of 24
`
`

`

`224
`
`
`
`
`Received
`Bits
`
`;
`Received
`Signal
`
`Transmit channel Power Control Data
`
`Transmit
`Modulator
`Sigg!
`
`
`|,Transmit Data_| Channel Transmit
`Decoder
`Bits
`
`
`
`
`
`
`Received Data
`
`
`(soft decision)
`Channel
`
`Decoder
`
`Demodulator
`
`
`Quality
`Receive channel
`Power Control Data
`Detect
`
`
`
`
`
`s=
`
`
`
`Power Control Step = -1, 0, +4 [dB]
`Power Control Period = 1.2 [msec/contrl]
`
`Figure 9.10 Operation of power control in a mobilestation.
`
`9.4.8 Random Frequency-Hopping Operation
`
`Frequency hopping is very effective to achieve frequency diversity and interference
`diversity. Frequency diversity is useful to average the frequency-selective channel
`properties (fading dips). Interference diversity is one of the important techniques used
`in the OFDMAproposal and has been shown to improve capacity in slow frequency-
`hopping TDMAsystems[8].
`
`The random hopping pattern is designed to be orthogonal within one cell (no
`collision