Efﬁcient Spectrum Utilization of UHF Broadcast Band722137/FULLTEXT01.pdf · to reuse the spectrum unoccupied by the primary DTT network. On the other hand, the declining inﬂuence

Efficient Spectrum Utilization of UHF BroadcastBand

LEI SHI

Doctoral Thesis inInformation and Communication Technology

Stockholm, Sweden, 2014

TRITA-ICT-COS-1406ISSN 1653–6347ISRN KTH/COS-14/06-SE

KTH Communication SystemsSE-100 44 Stockholm

Sweden

Akademisk avhandling som med tillstånd av Kungl Tekniska högskolan framlägges till of-fentlig granskning för avläggande av teknologie doktorsexamen i telekommunikation tors-dagen den 12 June 2014 klockan 14.00 i Room E, Forum, Kungliga Tekniska Högskolan,Isafjordsgatan 26, Kista.

© Lei Shi, June 2014

Tryck: Universitetsservice US AB

i

Abstract

The UHF band between 470-790 MHz, currently occupied by digital ter-restrial TV (DTT) distribution in Europe, is widely regarded as a premiumspectrum band for providing mobile coverage. With the exponential increasein wireless data traffic in recent years, there has been growing interests ingaining access to this spectrum band for wireless broadband services. Thesecondary access in TV White Space is considered as one cost-effective wayto reuse the spectrum unoccupied by the primary DTT network. On the otherhand, the declining influence of DTT and the converging trends of video con-sumption on TV and mobile platforms are new incentives for the regulator toreconsider the optimal utilization of the UHF broadcast band. The proposal tore-farm the UHF band for a converged content distribution network was bornunder theses circumstances.

This thesis intends to develop a methodology for evaluating the technicalperformance of these options for utilizing UHF broadcast band and quantify-ing their gains in terms of achievable extra capacity and spectrum savings. Forthe secondary access in TV white space, our study indicates a considerable po-tential for low power secondary, which is mostly limited by the adjacent chan-nel interference generated from the densely deployed secondary devices dueto the cumulative effect of multichannel interference. On the other hand, thispotential does not translate directly into capacity for a WiFi-like secondarysystem based on CSMA/CA protocol, as the network congestion and self-interference within the secondary system has a greater impact on the networkthroughput than the primary interference constraint.

Our study on the cellular content distribution network reveals more po-tential benefits for re-farming the UHF broadcast band and reallocating it fora converged platform. This platform is based on cellular infrastructure andcan provide TV service with the same level of quality requirement as DTTby delivering the video content via either broadcast or unicast as the situa-tion dictates. We have developed a resource manage framework to minimizeits spectrum requirement for providing TV service and identified a significantamount of spectrum that can be reused by the converged platform to provideextra mobile broadband capacity in urban and sparsely populated rural areas.

Overall, we have arrived at the conclusion that the concept of cellular con-tent distribution in a re-farmed UHF band shows a more promising prospectthan the secondary access in TV white space in the long run. Nevertheless, lowpower secondary is still considered as a flexible and low-cost way to exploitthe underutilized spectrum in the short term, despite its uncertainty in futureavailability. On the other hand, the re-farming of UHF broadcast band is along and difficult regulation process with substantial opposition from the in-cumbent.The results from this study could serve as input for future regulatorydecisions on the UHF band allocation and cost-benefit analysis for deployingnew systems to access this spectrum band.

Acknowledgements

Along the journey towards making this PhD thesis during the last four anda half years, there is no lack of occasions of feeling stressed and confused.But looking back from today, I would say, without regrets, it has been a trulyunforgettable and rewarding experience, thanks to the encouragement, supportand friendship from the people in this environment. Therefore, I would liketo take the opportunity here to express my sincere gratitude to these specialpeople.

First of all, I am most grateful to my main supervisor, Prof. Jens Zander,who has offered me the extraordinary opportunity to pursue the doctoral studyhere. His critical and visionary thinking has always inspired me the most andtaught me to live with and even enjoy the uncertainties in both research andthe life ahead. I am also very thankful for his trust and support, especiallyfor my application to the EIT ICT doctoral program, which may turn out intoanother wonderful journey into the ’uncertainties’.

I am also deeply in debt to Docent Ki Won Sung, my co-advisor, withwhom I have enjoyed working closely in most of the projects and publica-tions during my PhD study. He has always been extremely patient in giv-ing me invaluable advices and feedbacks in our countless discussions. And Ihave learned a great deal from his expertise and knowledge that he has alwaysshared without any reservation.

I greatly appreciate Docent Svante Signell, for his time spent in reviewingmy Licentiate thesis. I owe special thanks to Dr. Tim Irnich, for accepting therole of opponent for my Licentiate thesis defense. I would like to thank LasseWieweg for reviewing my PhD thesis proposal and Prof. Mats Bengtssonfor his valuable comments during the evaluation of this thesis. My specialthanks to Prof. Linda Doyle for accepting to be my opponent in the Doctoraldissertation. I am also grateful to the members of the grading committee: Dr.Joachim Sachs, Prof. Tommy Svensson and Prof. Mikael Sköglund.

I consider myself lucky and privileged to have met and worked togetherwith many talented people in different research projects. I am especiallythankful to Evanny Obregon and Javier Ferrer: the time we spent workingtogether has been a fun and fruitful year at the beginning of the PhD. I wouldalso like to say thanks to all of you who made the Radio Communication

ii

iii

System group such a unique and friendly environment: Ali Özyagci, SibelTombaz, Serveh Shalmashi, Du Ho Kang, Luis Martinez, Miurel Tercero, Yan-peng Yang, Haris Celik, Amirhossein Ghanbari, Anders Laya, Ashraf Widaa,Vlad-Ioan Bratu, Dr. Sang-Wook Han, Prof. Jan Markendahl, Prof. Ben Sli-mane, Prof. Claes Beckman, Prof. Guowang Miao, Dr. Luca Stabellini, GöranAndersson, Anders Västberg and many more.

I am also grateful to Sarah Winther, Jenny Minnema, Gunilla Gabrielssonand Lissi Perseus for their excellent administrative support. My special thanksgoes to Irina Radulescu and Dr. Bogdan Timus, not only for their help evenbefore I started my PhD but also for their genuine friendship.

It is just impossible to describe how grateful I am for my parents, RudongShi and Ling Yang. Your unconditional love has always been the pillar for meto lean on. I am also extremely grateful to my wife, Molan, for coping withmy fluctuating stress level during all these years. But you have not only helpedme to get through the difficult times with your encouragements and patience,but also created the space where I can live a balanced and healthy life. Finally,I would like to dedicate this thesis to my family: without your support andunderstanding, I would never reach this far.

Contents

Preface ii

Contents iv

List of Figures vi

List of Abbreviations viii

I 1

1 Introduction 21.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Spectrum Access in UHF Band . . . . . . . . . . . . . . . . 61.3 Previous Work . . . . . . . . . . . . . . . . . . . . . . . . . . 91.4 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . 131.5 Thesis Contributions and Outline . . . . . . . . . . . . . . . . 16

2 Research Methodology 202.1 Methodology for Analyzing the Spectrum Opportunity for

Secondary Access in TV White Space . . . . . . . . . . . . . 212.2 Methodology for Evaluating the Spectrum Saving by Cellular

TV Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 26

3 Secondary Access in TV White Space 303.1 Interference Modeling . . . . . . . . . . . . . . . . . . . . . 303.2 Protection of Primary System . . . . . . . . . . . . . . . . . . 313.3 Spectrum Opportunity and Secondary Capacity . . . . . . . . 34

4 Cellular Content Distribution Network in a Re-farmed UHF Band 394.1 Feasibility Analysis . . . . . . . . . . . . . . . . . . . . . . 394.2 CellTV Deployment in Inhomogeneous Environments . . . . 41

5 Discussion 45

iv

CONTENTS v

5.1 Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455.2 Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

6 Conclusions and Future Work 496.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 496.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

Bibliography 52

II Paper Reprints 58

7 Experimental Verification of Indoor TV White Space Opportu-nity Prediction Model (P1) 60

8 A Model for Aggregate Adjacent Channel Interference in TVWhite Space (P2) 67

9 Permissible Transmit Power for Secondary User in TV WhiteSpace (P3) 74

10 Controlling Aggregate Interference under Adjacent Channel In-terference Constraint in TV White Space (P4) 81

11 Secondary Spectrum Access in TV Bands with Combined Co-Channel and Adjacent Channel Interference Constraint (P5) 89

12 On the Capacity of Wi-Fi System in TV White Space with Ag-gregate Interference Constraint (P6) 100

13 CellTV - on the Benefit of TV Distribution over Cellular Net-works A Case Study (P7) 108

14 Opportunities and challenges for TV and Mobile broadband in470-790 MHz (P8) 122

15 Spectrum Requirement for Cellular TV distribution in UHFBand from Urban to Rural Environment (P9) 141

16 Cellular TV Distribution in Heterogeneous Environments (P10) 147

List of Figures

1.1 Spectrum allocation in the UHF broadcast band [1] [2] . . . . . . 31.2 TV consumption share on different platforms (%) [3]. . . . . . . . 31.3 Wireless data traffic growth forecast (Exabytes per month) [4]. . . 41.4 Secondary Access in TV White Space. . . . . . . . . . . . . . . . 61.5 Cellular content distribution network. . . . . . . . . . . . . . . . 81.6 Summary of related literature in TV white space. . . . . . . . . . 101.7 The correlation of the investigated scenarios. . . . . . . . . . . . . 13

2.1 Overall research approach of the thesis. . . . . . . . . . . . . . . 212.2 Methodology for analyzing the spectrum opportunity for secondary

access in TV white space . . . . . . . . . . . . . . . . . . . . . . 222.3 Interference and contention among secondary users. . . . . . . . . 252.4 Methodology for evaluating the spectrum saving by cellular TV

distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.5 Illustration of the CellTV concept. . . . . . . . . . . . . . . . . . 272.6 Studied area in the Greater Stockholm region. . . . . . . . . . . . 29

3.1 Estimated spectrum availability on each adjacent channel [5]. . . . 313.2 Maximum interference power level that a certain adjacent channel

can tolerate vs. the number of WSDs simultaneously transmittingin TVWS [6] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.3 Permissible transmit power for WSD located at 50 km away fromvictim TV receiver with different q1 (q2 = q1 ≠ 0.01, ‡

S

= 8dB,‡

G

= 10dB. The notations and references are defined in [7]). . . 333.4 Permissible transmit power with different WSD densities ⁄. q1 =

0.99, qú2 = 0.95 [8]. . . . . . . . . . . . . . . . . . . . . . . . . . 35

3.5 Location probability distribution over all TV channels in all pixels[9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.6 Maximum average throughput vs. unoccupied TV channel num-ber and different TV signal P

tv

. ⁄ = 100 WSDs/km2 in rural,3000 WSDs/km2 in urban, with 2 channel aggregation [10]. . . . 37

vi

List of Figures vii

3.7 Relation between permissible power and optimal power at dif-ferent TV signal strength. ⁄ = 30 WSDs/km2 in rural, 3000WSDs/km2 in urban, 2 channel aggregation, unoccupied TV chan-nel number = 8 [10]. . . . . . . . . . . . . . . . . . . . . . . . . 38

4.1 Spectrum requirement for CellTV broadcast in a urban environ-ment [11]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.2 Spectrum requirement for hybrid CellTV unicast-broadcast in arural environment (The legends specify different receiver antennaconfigurations for unicast links) [11]. . . . . . . . . . . . . . . . . 41

4.3 Local spectrum requirement of CellTV in the Greater Stockholmregion [12]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4.4 Spectrum requirement of CellTV in heterogeneous environments[13]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.5 Optimal transmission modes for TV content delivery in heteroge-neous environments [13]. . . . . . . . . . . . . . . . . . . . . . . 44

4.6 Spectrum requirement for CellTV in the Greater Stockholm areawith different numbers of TV channels [13]. . . . . . . . . . . . . 44

5.1 Service penetrations of digital terrestrial TV in different EU mem-ber states [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

List of Abbreviations

• ACI Adjacent Channel Interference

• BW Bandwidth

• CCI Co-Channel Interference

• CSMA/CA Carrier Sense Multiple Access with Collision Avoidance

• DTT Digital Terrestrial Television

• DVB-H Digital Video Broadcasting - Handheld

• DVB-T Digital Video Broadcasting - Terrestrial

• ECC Electronic Communications Committee

• EIRP Equivalent Isotropically Radiated Power

• ETSI European Telecommunications Standards Institute

• FCC Federal Communications Commission

• HD High Definition

• ISD Inter Site Distance

• ISI Inter Symbol Interference

• ISM Industrial Scientific and Medical

• LSA Licensed Spectrum Access

• LTE Long Term Evolution

• MIMO Multiple Input and Multiple Output

• OFDMA Orthogonal Frequency-Division Multiple Access

• PU Primary User

• SD Standard Definition

• SFN Single-Frequency Network

• SINR Signal to Interference and Noise Ratio

• SINR Signal to Interference Ratio

• SU Secondary User

viii

List of Figures ix

• TVWS TV white spaces

• UHF Ultra High Frequency

• UMTS Universal Mobile Telecommunications System

• VHF Very High Frequency

• VoD Video On-Demand

• WRAN Wireless Regional Area Networks

• WRC World Radio Communication Conferences

• WSD White Space Device

• eMBMS Evolved Multimedia Broadcast and Multicast Service

Part I

1

Chapter 1

Introduction

1.1 Background

1.1.1 UHF band and digital television broadcastThe terrestrial television broadcast service initially started to operate in theVery High Frequency band (VHF). It was later expanded to the lower UltraHigh Frequency (UHF) band to fulfill the demand for additional over-the-air(OTA) television channels. The favorable propagation conditions and practicalantenna size in the UHF band make it an ideal choice for broadcast service.As more OTA television channels are introduced, the frequency band allocatedfor terrestrial television broadcast service in UHF band extends from 470 MHzto 862 MHz in Europe and from 470 to 890 MHz in the U.S.

The introduction of digital television in the 1990s brought new transmis-sion and compression technologies that enabled digital terrestrial television(DTT) networks to deliver video contents more efficiently than analog tele-vision networks. The digital television transition was completed in the U.S.in 2009, releasing the spectrum from 698-806 MHz ( the spectrum between806 and 890 MHz had already been reallocated from TV broadcast services toanalog mobile telephony in the U.S. in 1983).This process had also been com-pleted in most of European countries by 2012, releasing the spectrum from790-862 MHz. These spectrum bands released in the process of analog-to-digital transition, often called ’(the first) digital dividend’, were subsequentlyauctioned and allocated to the growing mobile broadband industry. Therefore,the spectrum left in UHF television band is only between 470-698 MHz in theU.S. and between 470-790 MHz in Europe (see Fig.1.1). 1

In addition to the release of the digital dividend, the digital switch-overhas also revitalized the terrestrial broadcasting industry and brought the most

1Although VHF band III (174-230 MHz) were still used for terrestrial TV service in some countries, it hadmostly been reallocated for digital audio broadcasting around the world, leaving the spectrum in UHF band asthe primary medium for digital terrestrial TV distribution.

2

CHAPTER 1. INTRODUCTION 3

First Digital Dividend (completed by 2012)

Second Digital Dividend (to be

confirmed in WRC-15)

490 MHz 698 MHz 790 MHz 862 MHz

DVB-T1/T2

(a) European Union

Reallocated to Analog Mobile

Telephony (1983)First Digital Dividend (Completed by 2009)

490 MHz 698 MHz 806 MHz 890 MHz

ATSC

(b) U.S.

Figure 1.1: Spectrum allocation in the UHF broadcast band [1] [2]

Figure 1.2: TV consumption share on different platforms (%) [3].

significant evolution in its quality since color television in the1950s. More TVprograms are provided and the video quality is superior to their analog coun-terparts. In particular, Digital Video Broadcasting-Terrestrial (DVB-T), theEuropean standard for DTT technology, has achieved remarkable success inthe last decades. With a share of 40% households receiving television throughDTT today and a predication of 50% by 2018, DVB-T has established itself asthe most popular platform for television reception in Europe [3] (see Fig.1.2).

The future of DTT, however, is less promising. It is facing not only in-creasing competition from other platforms, such as cable, satellite and inter-net protocol television (IPTV) via fixed broadband, but also fading dominanceof linear broadcasting in the living room. The viewers’ focus on the televi-


© 2014 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public. Page 14 of 40

Figure 10. Mobile Video Will Generate Over 69 Percent of Mobile Data Traffic by 2018

Because many Internet video applications can be categorized as cloud applications, mobile cloud traffic follows a curve similar to video. Mobile devices have memory and speed limitations that might prevent them from acting as media consumption devices, were it not for cloud applications and services. Cloud applications and services such as Netflix, YouTube, Pandora, and Spotify allow mobile users to overcome the memory capacity and processing power limitations of mobile devices. Globally, cloud applications will account for 90 percent of total mobile data traffic by 2018, compared to 82 percent at the end of 2013 (Figure 11). Mobile cloud traffic will grow 12-fold from 2013 to 2018, a compound annual growth rate of 64 percent.

Figure 1.3: Wireless data traffic growth forecast (Exabytes per month) [4].

sion has become secondary as they are increasingly engaged in other applica-tions on their mobile devices [14]. The tremendous success achieved recentlyby over-the-top (OTT) services like Netflix and BBC iPlayer manifested thegrowing popularity of video on-demand (VoD) services. They provide con-sumers with a choice of high quality contents and flexibility unmatched by theDTT service. The DTT industry has attempted to retain its competence in theVoD/nonlinear service by offering Hybrid Broadcast Broadband TV (HbbTV)service. However, as HbbTV requires broadband connections, the added valueis rather limited as compared to cable or IPTV alternatives. Moreover, the con-sumption of audio-visual content is rising rapidly on smartphones and tablets,which the DTT industry has struggled to reach without much success thusfar. Digital Video Broadcasting - Handheld (DVB-H), the digital terrestrialTV broadcasting standard for mobile reception, has failed to reach a wideaudience due to the lack of suitable devices and a viable business model formobile TV based on linear content [15].

1.1.2 Mobile Data Traffic Growth and Video DominanceThe commercial success of mobile service in the last decade has made itselfan essential part of every consumer’s daily life. With the deployment of 3Gand 4G networks around the globe and the proliferation of smartphones andtablets, mobile data traffic has been growing exponentially in recent years,leading to the projection of even an 1000-fold increase in total wireless datatraffic by 2020.

In fact, the primary drive for the data growth is audio-visual content con-sumption on mobile platforms. With a much higher bit rate than other mobilecontents and applications, mobile video traffic already constitutes more thanhalf of the global mobile traffic today. It will continue to grow at a cumulative


annual growth rate of 69% (higher than the global mobile traffic growth rate at61%) and reach almost 70% of the total share of the mobile traffic in the nextfive years (see Fig.1.3).

As suggested in many recent articles [4, 16, 17], however, such a growthrate can only be sustained if mobile operators are able to provide sufficientnetwork capacity. A possible slow-down of the growth is already evidenced inthe recent adjustment to the mobile data traffic forecast. For instance, Ciscohas predicted a 30-fold increase from 2010 to 2015 in [18]. In contrast, theirrecent publication indicates only an 11-fold increase in global mobile datatraffic between 2013 and 2018 [4].

In order to provide sufficient capacity, the operator could not rely exclu-sively on the densification of mobile network infrastructure, as it requires notonly a enormous amount of capital expenditure but also an increasing oper-ation expenditure that scales with the infrastructure and user density. Theheavy investment in next generation technology - 4G, Long Term Evolution(LTE)- could provide limited relief for the operator when the adoption of 4Gdevices eventually increases. Nevertheless, the spectral efficient improvementfrom 3G to 4G is limited as it is already approaching the Shannon bound.At the same time, mobile offload through public hotspots or residential Wi-Fi networks could be expected to account for only 50% of the total mobiledata traffic, because Wi-Fi networks are rapidly becoming congested with thegrowing traffic and interferences. Besides, more devices that have only Wi-Ficonnection today, such as tablet and laptop, would be equipped with cellularconnectivity to satisfy the increased desire for mobility.

Therefore, many studies consider that the most cost-effective way to pro-vide the extra system capacity is to obtain additional spectrum allocation. Thehigh price for spectrum licenses is in itself a clear indication of the competi-tiveness of this strategy. However, in order to obtain additional spectrum, themobile data service must compete with other services for this limited naturalresource. In this thesis, we concentrate the focus of our investigation on UHFbroadcast band for the following reasons:

• First, the deployment of DVB-T network typically consists of high powertransmitters mounted on high towers. Thus it requires fixed frequencyplanning that leaves a considerable amount of spectrum underutilized inlarge areas to avoid excessive self-interference from other transmitters.

• Second, as video service is gaining an increasingly prevailing share inthe global mobile data traffic, the consumption of television service isshifting towards cross-platform and on-demand contents. These trendsindicate a clear convergence in the future mobile broadband and broad-cast services.

• Lastly, the propagation characteristic of UHF band is ideal for mobilecoverage while the frequency range still permits antenna design withpractical handset size.


TV Coverage

TV signal

WSD

WSD

WSD

Adjacent-Channel Interference

WSD

WSD

WSD

Co-Channel Interference

WSD

Figure 1.4: Secondary Access in TV White Space.

1.2 Spectrum Access in UHF Band

As the UHF broadcast band is widely recognized as under-utilized in compar-ison with other spectrum bands occupied by mobile communication, nationalregulators in Europe and the U.S. have taken strong initiatives in facilitatingother wireless services to access this band. Different options have been dis-cussed, ranging from shared access under the regime of Licensed SpectrumAccess (LSA) to spectrum re-farming and reallocation.

1.2.1 Secondary Access in TV White SpaceA typical DTT broadcasting network consists of high-rising TV towers withhigh power transmitters. In order to minimize inter-symbol interference (ISI)or to provide regional content, the TV programs are delivered over differentfrequency bands from each TV tower or n a group of TV towers that formsa single frequency network (SFN). This combination of large coverage areaand fixed frequency planning offers a unique opportunity for secondary ac-cess in TV bands, as the primary usage can be easily identified and stored ina geo-location database. Spectrum sensing as a detection mechanism couldbe employed to complement the geolocation database for improved detectionaccuracy.

If the current situation of DTT service maintains status quo, one of thepossible option is to exploit the spectrum opportunities in the UHF band byspatially reusing the spectrum on a secondary basis. Based on the informationabout DTT coverage available in the geo-location database, the secondary sys-tem can transmit on the TV frequency channels that are locally unoccupied.This type of secondary spectrum access is termed an interweaved model. Butif we consider adjacent channel interference leakage, it could also be consid-ered as an underlay model, in which the secondary user accesses the same


spectrum utilized by the primary system. Fig.1.4 depicts the potential inter-ference link in the secondary access scenario. In both cases, the secondarysystem must ensure that the harmful interference caused to the TV receivers iswithin certain acceptable limit 2.

The sharing of the UHF Television spectrum between secondary systems,typically for mobile broad offloading or local area wireless access, and theincumbent TV broadcast service is widely known as secondary access in ’TVWhite Space’(TVWS). This concept of secondary access or licensed spectrumaccess (LSA) in TVWS was promoted by regulators in both Europe and theU.S. to satisfy the growing demand for spectrum for wireless broadband [19].

The opening of TV bands for secondary access was initiated by the Fed-eral Communications Commission (FCC) in the U.S. in 2004. Later FCCpublished a revised ruling in 2010 on secondary access in TV bands with de-tailed technical specifications and regulatory requirements [20]. OFCOM, theindependent regulator in the UK, followed the FCC’s decision and launchedthe first major pilot to test TV white space in Europe. On the European level,Electronic Communications Committee (ECC) has organized the SpectrumEngineering Group 43 (SE43) as a special task group to investigate the differ-ent approaches for secondary access. A general framework was published inthe ECC report 159 [21].

A few standards for secondary access are being developed. Most notice-ably, the IEEE working group 802.22 formed in 2004 released a standardfor Wireless Regional Area Network (WRAN) using white spaces in the TVbands in 2011. Another prominent standardization effort is the IEEE802,11.af’White-Fi’ working group. This working group sets out to define a standardfor the implementation of Wi-Fi technology in TV bands, allowing a bettercoverage and access to wider bandwidth than the traditional Wi-Fi in ISMbands [22]. The European Telecommunications Standards Institute (ETSI) hasrecently published the draft for a harmonized European standard for secondaryaccess in UHF broadcast band [23], which specifies the technical requirementsfor the devices and procedure for testing requirement compliance.

1.2.2 Second digital dividendCurrently, the 470-790MHz UHF broadcasting band is allocated for DTTbroadcasting services in Europe, but in many other regions the 700MHz bandhas already been allocated for mobile services (e.g., North African and MiddleEastern countries; the U.S. also auctioned this band in 2009). In an attemptto create an internationally harmonized frequency band in sub-1GHz for mo-bile service, the World Radio Communication Conference (WRC) in 2012

2Some studies have proposed an overlay approach, where the secondary transmission not only delivers itsown signal but also assists the primary system that operates on the same frequency by applying signal processmethod such as dirty paper coding and network coding. However, as we assume the legacy DTT system isincapable of cooperating with the secondary system, the overlay approach is not considered in our study


Cell A Unicast Coverage

Cellular BS A

Cell B Unicast Coverage

Cellular BS B

Broadcast Links

SFN

Unicast link A

Unicast link B

Broadcast Unicast A

Unicast B

Spectrum Saving

VHF/UHF band

Figure 1.5: Cellular content distribution network.

announced the decision to allocate the 700MHz band for mobile services ona co-primary basis with DTT broadcasting, making future authorization formobile use in this band easier and more attractive [24].

The decision on co-primary allocation in the 700 MHz band will be con-firmed in the WRC in 2015 with immediate effect, although the EU Mem-ber States are not obliged by the decision to make this spectrum availablefor mobile service. Recently, European Commission has issued a mandatefor the European Conference of Postal and Telecommunications Administra-tions (CEPT) to develop a set of technical conditions required for EU MemberStates to deploy mobile service in the 700 MHz band, which could be deliv-ered as early as in 2016.

Releasing the ’second digital dividend’ in the 700 MHz band would forcethe incumbent DTT network to evacuate 30% of the post-switchover spec-trum capacity for terrestrial broadcasting. It would pose a considerable chal-lenge for countries with high reliance on terrestrial reception (e.g., Italy andSpain [25]) due to the lack of spectrum for simulcasting, which is crucial forensuring the service continuity before the channel migration, network and userequipment upgrades (e.g. transmitter and receiver for DVB-T2 and wide bandaerials) are completed. Furthermore the potential gain of DVB-T2 and SFNwould be limited by the cross-border interference coordination and regionalcontent distribution in practical implementations. Consequently, DTT opera-tors might not have sufficient capacity to offer spectrum-hungry services suchas ultra-HDTV to satisfy the growing viewing demand [26] 3.


1.2.3 Mobile Broadband and Broadcast ConvergenceIn light of the converging trends of audio-visual content consumption in bothmobile broadband and TV service described in the previous section, a unifiedplatform based on single IP-access network is envisaged for providing bothbroadband and broadcast service in the UHF television band. The spectrumsharing between the distributions of traditional linear TV content and mobiledata is realized on the network/application level instead of the physical levelspectrum sharing between distinct radio access technologies. An illustrationof this concept is shown in Fig.1.5.

The key enabler technology for this concept is the Evolved MultimediaBroadcast and Multicast Service (eMBMS) introduced in 3GPP LTE (LongTerm Evolution) radio technology for point-to-multipoint or multipoint-to-multipoint service over a single frequency network (SFN) [28]. Through tighttime synchronization, the TV contents can be broadcasted over a SFN withhigh spectrum efficiency. Furthermore, additional features such as localizedcontents distribution and on-demand services are made possible by utilizingthe point-to-point communication link in cellular network, and thus consider-ably improve the flexibility of the TV service.

The discussion item 1.1 in WRC-15 agenda, ‘additional spectrum alloca-tion to the mobile service to facilitate the development of terrestrial mobilebroadband’, will exam the possibilities for further mobile allocation in the470- 694MHz, creating a converged platform to provide both TV broadcast-ing and mobile service via mobile network. The development of a convergedterrestrial broadband network is well in line with the European Commission’slong-term vision for the convergence of broadcasting and broadband services(European Commission 2013). As an all-IP network, it is expected to utilizethe spectrum in UHF band more efficiently, creating opportunities for furtherinnovation and economic growth. Although the spectrum re-farming of UHFband would be a long and difficult process due to barriers of structural natureat European Union level and resistances from established interests, it can stillbe achieved through progressive restructuring of the broadcast landscape anddevelopment of new standardization and business models. Recently, EuropeanCommission has initiated a feasibility study on the convergence of broadbandand broadcast, and the results will be available by the end of the year 2014.

1.3 Previous Work

1.3.1 Secondary Access in TV White SpaceIn recent research, considerable efforts have been dedicated to the primary sig-nal detection problem, aiming at finding the ‘spectral holes’ via spectrum sens-

3The Swedish government has recently announced the decision to reallocate the spectrum band between694-790 MHz from DTT broadcasting to MBB from the year 2017 [27].


Single User Multiple User

Co-Channel Interference

Adj-Channel Interference

ECC 159Reference Geometry:Deterministic Model

Pessimistic Estimation

ECC 159: Reference Geometry andFixed Interference Margin

ECC 159Analytical Estimation for Permissible SU Power

Underestimate Inteference

Pessimistic Estimation

Coordinated Aggregate Interference Control for

Cellular-like SU Not Applicable to

Short Range SU

Figure 1.6: Summary of related literature in TV white space.

ing [29]. A few recent EU projects, such as, QUASAR, QoSMOS, FARAMIRand CogEU have extend the research focus from purely discovering the spec-trum opportunities to the exploitation of the secondary access opportunity andhave promoted the use of geo-location database approach [30]. In particular,the QUASAR project have put specificial emphasis on the evaluations of ac-tual TV white space availability and the utility for secondary access in Europebacked by empirical spectrum models and economical analysis [31].

To model the effect of secondary interference on TV reception, measure-ment data for TV to TV interference [32] and the results from the study onLong Term Evolution (LTE) deployment in the 800 MHz band [33] have beenwidely used as the reference data for any hypothetical secondary access sce-nario. Other recent measurements provide results for the effect of interferencefrom IEEE 802.22 transmitters [34] or WiMAX transmitters operating on asingle TV channel, either on the same or adjacent to the receiving TV chan-nel.

Utilizing these measurements results, the regulators have developed method-ologies for estimating the maximum permissible transmit power or exclusiondistance for secondary device [21] [20]. These frameworks relies on the use ofgeo-location database and computational intensive Monte-Carlo simulation toidentify spectrum opportunity for secondary access in TV band. An analyti-cal framework has also been developed in [21] to facilitate large scale analysis.However, this frameworks is primarily designed for single secondary user onlycase and, as we have identified, it overestimates the permissible power levelby almost 10 dB in certain situations.

Several recent studies have attempted to quantify the spectrum opportu-nity by applied these frameworks in their analysis. The authors in [35] hadinvestigated the TVWS usage case in the U.S. based on FCC’s framework


with simplified models for propagation and secondary system deployment. Ina similar fashion, [31] presents the estimated TVWS availability in Europeancountries. A comparison of the two frameworks proposed by FCC and ECCis described in [36] via a case study of Finland.

In most of these works, the availability of TVWS has been quantified andrepresented in terms of the varying number of available TV channels for asingle secondary user at different locations. The scalability of the secondarysystem, therefore, was not taken into account of the analysis, as the assessmentmethodologies lack the proper model for the aggregate interference from mul-tiple secondary users to the primary system. Although ECC’s approach canbe extended to multiple secondary users case by applying an additional in-terference margin to the secondary transmit power, it does not guarantee anyefficient spectrum usage. Besides, the deterministic model for adjacent chan-nel interference (ACI) in the regulatory framework [37] is rather simplistic andwould lead to pessimistic conclusions as in [38]. Thus, the previous analysiscan only indicate the potential of TVWS to a limited extent.

On the other hand, one common conclusion we can draw from these stud-ies is that the secondary system with high power and tall tower is not a fa-vorable candidate for secondary access in TV bands. Because an oversizedsecondary cell can not efficiently utilize the highly localized spectrum reuseopportunity. Combining the analysis from both technical and business per-spective, [39] suggests that the commercial sweet point for secondary accessin TV bands would be the deployment of short range system, e.g., ’Wi-Fi’ or’Femto’ like systems, which coincides with the forecast that 70% of the wire-less traffic would be generated indoor by 2015 [40]. With the developmentof device-to-device communication and the proliferation of smart phones andtablet devices, we can foresee the scenario with portable/mobile secondarydevices densely deployed in both indoor and outdoor environments. Someof these devices would inevitably be located close to the TV receivers, caus-ing interferences from different adjacent TV channels used by the secondarydevices. However, the few studies on aggregate secondary interference inTVWS [41] [42] [43]have exclusively focused on the co-channel interferencesoriginated from typical ’cellular-like’ secondary system deployed outside theTV coverage area.

The architecture and capacity of a Wi-Fi like secondary system in TVWSwas studied in [44] and [45]. Results of both papers demonstrate the advantageof TVWS in coverage. However, these studies neglect the impact of adjacentchannel interference from secondary users to the TV receiver. Therefore, theperformance of the secondary Wi-Fi system may have been overestimated.

1.3.2 Cellular TV DistributionThe concept of distributing TV contents using a cellular structure was firstmentioned in [46] for coverage extension using relays. This idea has been


gaining increasing popularity since the introduction of MBMS the in 3G net-work. Nevertheless, most of the recent studies have been focused on mobileTV service provision. The system architecture of MBMS in 3G networks isdescribed in [47]. A hybrid broadcast-unicast deployment for handling the"long tail" contents is investigated in [48]. In [49], the authors perform a de-tailed traffic analysis for this hybrid solution. [50] discusses the mobile TV ser-vice provision in 3G networks from the implementation and cost aspects. [51]advances the study on the convergence of mobile TV service and mobilebroadband (MBB) network from 3G to 4G networks. A general roadmap andanalytical models for assessing the network performance of mobile video ser-vice in 4G networks are presented in [52]. The performance is evaluated interms of coverage and throughput for different deployment options using ad-vanced features introduced in the LTE system.

However, the demand and quality requirement for DTT service differs sig-nificantly from the mobile TV service. The current DVB-T system offers highdefinition TV programs that requires a data rate over 7 Mbps, whereas the datarate of a typical mobile TV transmission is in the range of hundreds of kbps.Furthermore, the location probability, the measure for DTT coverage qualityis commonly defined at a level of 95% at the coverage edge, which is muchhigher than the 70% location probability requirement of a mobile TV service .This strict coverage requirement for the DTT service is a formidable challengefor any attempt to replace it with mobile networks. Considerable bandwidthwould be needed to deliver the same high quality TV programs to the fixedreceivers even at the edge of the coverage, even though fixed TV receivers canrely on more advanced antenna configurations with higher performance thanmobile receivers. Consequently, the existing results on mobile TV cannot bedirectly applied to the study on cellular TV distribution system as we haveenvisaged.

Using LTE technology to provide over-the-air TV service has been pro-posed in [53], which relies on a ’tower-overlay’ architecture. It assumes thatthe DTT network would employ a modified LTE standard for broadcasting TVcontent to both mobile and fixed reception. Recent studies have considered us-ing not only LTE technology but also cellular infrastructure for providing TVservices. In [54], the authors have investigated the spectrum requirement fordelivering today’s over-the-air TV service over cellular networks. The cal-culations are focused on densely populated urban areas in different cities inthe U.S.. Since the typical inter-site-distance (ISD) of cellular networks ofthe study area is smaller than 2 km, it ensures good performance of the SFNsbased on the eMBMS technology. However, as shown in [47], the spectral ef-ficiency of the SFN rapidly degrades as the propagation delay increases withlarger ISDs, results in a considerably higher spectrum requirement for broad-casting in rural areas than in urban areas. Therefore, it is not evident thatreplacing DTT service with mobile networks is feasible based on the resultsfrom urban scenarios alone.


Converged Cellular TV

Network

DTT Dominance:Short Range

Secondary Access

Limited Main-Stream Content

In-Home Viewing

On-the-Move Viewing

Diversified On-Demand Content

Status QuoMobile TV

Broadcasting

Home Hybrid TV Ubiquitous On-Demand TV

TV

Consu

mption

Trend

s

Figure 1.7: The correlation of the investigated scenarios.

One of the key strengths for cellular TV distribution is its capability todynamically allocate different content into either broadcast or unicast service,and form SFN to improve the efficiency for broadcast distribution. Whilethe resource allocation between broadcast and unicast has been investigatedin [48], [49] and [55], these analysis are limited to either an isolated cell ora homogeneous environment. Studies on SFN deployment have consideredinhomogeneous geographic settings [56], but the service demand for broad-casting is assumed to be constant. Thus, it is unclear what effect the variationsin both service demand and cell density would have on the overall spectrumrequirement for cellular TV distribution in an inhomogeneous environment.

1.4 Problem Formulation

As discussed above, both secondary access in TVWS and a converged cellu-lar TV distribution system are possible ways to improve the utilization effi-ciency of UHF band and provide extra wireless network capacity to satisfy thegrowing mobile data traffic driven by audio-visual consumptions on mobileplatforms. As illustrated in Fig.1.7, secondary access in TVWS is a flexi-ble solution targeted at a short-term scenario in which the DTT service andspectrum allocation in UHF band would largely remain unchanged, whereasa converged content distribution network is a visionary approach fitted fora long-term time frame in which the UHF band would be progressively re-farmed and completely reallocated from DTT to the converged platform. The


second digital dividend should not be viewed as an isolated piecemeal action,but rather as an initial step for the comprehensive and coherent inventory pro-cess in the UHF band of bridging the two scenarios.

However, these novel concepts for spectrum access in UHF band still facenumerous challenges to materialize in large scale implementations. On onehand, the mobile industry in most countries remains skeptical about the com-mercial viability of secondary system in TVWS, as concrete results on the sys-tem scalability and convincing business models are still lacking. The prospectof TVWS is further clouded by the regulatory decisions on reallocate the 700MHz band in the future. On the other hand, while this reallocation creates aunique opportunity to initiate progress in re-farming the UHF broadcast spec-trum for a converged multimedia platform, quantitative evidence for its ben-efits remains to be established to justify any investment and regulation plansfor development and deployment. Therefore, the objective of this disserta-tion is to develop technical frameworks for evaluating the improvements inthe utilization of UHF broadcast band, so that we can provide answers to thefollowing research question:

• How much addition capacity can be provided for wireless broadbandservice by allowing it to access the UHF broadcast band?

The premises of this study is that the audio-visual service provided by DTTbroadcast should be maintained or replaced by another system with equal oreven improved quality of service. The technical results provide direct inputto the ongoing technical standardization process. The implications of theseresults for the evolving broadcast-broadband ecosystem could be taken intoconsideration for future regulatory decisions on the UHF band allocations.

1.4.1 Secondary Access in TV White SpaceThe study on the TVWS case focuses on the scalability issue of secondarysystem and provides technical evidence to illustrate the real opportunity inTVWS under aggregate interference constraints. The investigation focused onlow power WiFi-like secondary system, because its small footprint is consid-ered to be more suitable for exploiting the local spectrum opportunity than ahigh power cellular-like secondary system. In particular, this thesis aims atanswering the following two research questions on TVWS:

• Is there a regulation policy that ensures reliable primary protectionand allows secondary access in TVWS to be a commercial success?The major concern for secondary access in TVWS is the protection ofthe primary service against harmful secondary interference. Althoughthe regulation framework for controlling the secondary user transmissionpower has already been established, it relies on computation-intensiveMonte-Carlo simulation. Given the size of typical TV coverage, an ana-lytical approach is needed for quick large scale analysis. Besides, due to


the passive nature of TV receivers, it is impossible to verify the exact lo-cation of any TV receiver, which has considerable impact on the adjacentchannel interference when the secondary user is located close to the TVreceiver. Thus, one of the challenges is how to model this uncertainty inthe interference geometry and incorporate it into the overall framework.

• Is TVWS suitable for massive use by short range secondary systems?The performance of the secondary system is not only limited by the pri-mary interference constraint but also affected by the coordination withinthe secondary system or different secondary systems. Therefore, in or-der to determine the scalability of the secondary system, the analysismust take into account the aggregate interference not only from multiplesecondary users to primary user but also within the secondary system.

1.4.2 Cellular TV distributionThe study on the cellular content distribution network in UHF band exploitsthe possibility of completely replacing the terrestrial TV broadcast service byreusing existing cellular infrastructure operating in the UHF broadcast band.The analysis performed in this dissertation emphasizes the spectrum require-ment for such a system to replace the terrestrial TV broadcast service. Inparticular, this thesis aims at answering the following two research questions:

• Is it feasible to provide terrestrial TV service in UHF broadcast bandusing existing cellular network infrastructure?The major difference between the study on cellular TV distribution andthose on typical mobile TV broadcast services is that here we assumeCellTV must replace DTT network and provide service to fixed TV re-ceptions, and thus must satisfy the considerably higher requirement onthe quality of service. Besides, a cellular TV distribution network mayutilize unicast in areas where broadcast (over SFN) is not efficient fordelivery of low-demand content, but it must ensure low blocking proba-bility for the unicast links.

• Can we deploy the cellular TV distribution network in an inhomoge-neous environment and make it adapt to the future changes in audio-visual content consumption trends?The resource allocation for the cellular TV distribution network is pri-marily dependent on the distribution of viewing demand and the avail-ability of cellular network infrastructure. Providing a seamless coverageover a large area with inhomogeneous network density and user demandrequires a smooth transition between different transmission modes forcontent delivery, i.e., broadcast over SFN and unicast. The choice oftransmission modes should be included in an optimization resource al-location framework to minimize the overall spectrum requirement. Inaddition, while today’s terrestrial TV broadcast service is still primarily


linear broadcasting, we expect the cellular replacement would be ableto deliver more on-demand/unicast service to conform with the shiftingtrends in TV consumption.

1.5 Thesis Contributions and Outline

To provide answers to these research questions, we have divided the thesisinto the following two main areas: the interference modeling and performanceevaluation of secondary access in TV band, and the investigation on the spec-trum requirement for cellular content distribution system.

1.5.1 Secondary Access in TV White SpaceInterference Modeling

In Paper 1, we have investigated the effect of secondary interference to TV re-ception on adjacent channels. The main contribution includes the experimen-tal verification of adjacent channel interference on TV reception. This modelis further extended in Paper 2 to take into account the accumulative effect ofmultichannel interferences. The proposed model is again verified through ex-tensive laboratory experiments, whose results show good agreement with thepredication of the model.

• Paper 1. E. Obregon, L. Shi, J. Ferrer and J. Zander, “ExperimentalVerification of Indoor TV White Space Opportunity Prediction Model,”in 5th International Conference Cognitive Radio Oriented Wireless Net-works and Communications (CROWNCOM 10), Cannes, France, June,2010.

• Paper 2. E. Obregon, L. Shi, J. Ferrer and J. Zander, “A Model forAggregate Adjacent Channel Interference in TV White Space,” in IEEE73rd Vehicular Technology Conference, (VTC 2011-Spring), Budapest,Hungary, May, 2011.

Both papers were written in collaboration with Evanny Obregon and JavierFerrer. The author of this thesis is mainly responsible for the designing of themeasurement campaign and result analysis in Paper 1. In Paper 2, the authorof this thesis has proposed the analytical model for multichannel interferencebased on initial observations from test experiments and performed the mea-surement with the other co-authors. Professor Jens Zander provided directionsand valuable insights in all papers.

Primary Protection

Paper 3 identifies one mathematical flaw of the original analytical approachin ECC report 159 and proposes a new analytical method for estimating the


permissible transmit power for a single secondary user. This contribution hasbeen accepted by SE43 working group and included in the later release ofthe ECC regulatory framework. Paper 4 presents a stochastic model based onthe Poisson point process for estimating the aggregate interference from ran-domly deployed secondary users transmitting on different adjacent TV chan-nels. These contributions have been previously published in the followingpapers:

• Paper 3. L. Shi, K.W. Sung and J. Zander,“Permissible Transmit Powerfor Secondary User in TV White Space,” in 7th International Confer-ence Cognitive Radio Oriented Wireless Networks and Communications(CROWNCOM 12), Stockholm, Sweden, June, 2012.

• Paper 4. L. Shi, K.W. Sung and J. Zander, “Controlling Aggregate In-terference under Adjacent Channel Interference Constraint in TV WhiteSpace,”in 7th International Conference Cognitive Radio Oriented Wire-less Networks and Communications, 2012, (CROWNCOM 12), Stock-holm, Sweden, June, 2012.

The author of this thesis proposed the original problem formulation and pro-gramed the simulation for numerical analysis. The analytical framework isdeveloped as a result of the research discussion between the author of this the-sis and Dr. Ki Won Sung. These papers were jointly edited by the author ofthis thesis, Dr. Ki Won Sung and Professor Jens Zander.

Evaluating Spectrum Opportunity

Paper 5 combines CCI and ACI into a single aggregate interference con-straint by incorporating the accumulative effect model and the stochastic ap-proach for estimating ACI. An optimal resource allocation framework for asecondary system under the interference constraint is formulated to maximizethe spectrum opportunity for secondary access. Paper 6 builds upon the pri-mary protection framework developed in earlier publications and introducesthe CSMA/CA protocol to coordinate the media access among the secondaryusers.The system throughput of a ’WiFi-like’ secondary system under primaryprotection constraint is compared to a typical WiFi system operating in theISM band.

• Paper 5. L. Shi, K.W. Sung and J. Zander, “Secondary Spectrum Ac-cess in TV Bands with Combined Co-Channel and Adjacent ChannelInterference Constraint”, in IEEE International Symposium on DynamicSpectrum Access Network (DySPAN 2012), Seattle, USA, October, 2012.

• Paper 6. Y.P. Yang, L. Shi, and J. Zander, “On the Capacity of Wi-Fi Sys-tem in TV White Space with Aggregate Interference Constraint”in 8thInternational Conference Cognitive Radio Oriented Wireless Networksand Communications (CROWNCOM 13), Washington, D.C., U.S., July,2013.


The author of this thesis proposed the original problem formulation and actedas leading author of these papers. In paper 6, Yan Peng Yang developed thesimulation framework and obtained the numerical results. Professor Jens Zan-der and Dr. Ki Won Sung provided directions and valuable insights in allpapers.

1.5.2 Cellular content distribution systemFeasibility of CellTV concept

Paper 7 describes the network model for a cellular content distribution systemand the envisaged scenario for a converged platform operating in a re-farmedUHF band. The technical feasibility is evaluated through a numerical studyon Swedish environments with moderate viewing demand changes in the year2020. Paper 8 investigates the feasibility of the CellTV concept from the reg-ulatory and business perspective, presenting the major challenges and benefitsidentified for CellTV.

• Paper 7. L. Shi, E. Obregon, K.W. Sung, J. Zander and J. Boström,“CellTV - on the Benefit of TV Distribution over Cellular Networks ACase Study”, IEEE Transaction on Broadcasting, March. 2014.

• Paper 8. L. Shi, K.W. Sung and J. Zander, “Opportunities and Challengesfor TV and Mobile Broadband in 470-790 MHz” 24th European Re-gional International Telecommunication Society Conference , Florence,Italy, Oct. 2013.

The original problem formulation of Paper 7 is a result of joint discussionbetween Dr. Ki Won Sung, Evanny Obregon and Jan Bostrom. Jan Bostromfrom PTS has provided valuable input to the scenario modeling. The author ofthis thesis and Evanny Obregon have performed the numerical analysis. Theauthor of this thesis has acted as leading author of these papers. Professor JensZander and Dr. Ki Won Sung have provided valuable insights and helped inediting of all papers.

Minimizing CellTV Spectrum Requirement

Paper 9 presents an initial resource allocation framework for the cellular TVdistribution network. It optimizes the spectrum resource allocation betweenunicast and broadcast to provide a seamless coverage from urban to rural areawith minimum spectrum requirement. Paper 10 extends the analysis froma Stockholm case study to more general situations and identifies the impactof heterogeneous environments and the evolving viewing habits on the localspectrum requirement for CellTV.

• Paper 9. L. Shi, K.W. Sung and J. Zander, “Spectrum Requirement forCellular TV distribution in UHF Band from Urban to Rural Environ-


ment” in IEEE International Symposium on Dynamic Spectrum AccessNetworks (DySPAN 2014), McLean, Virginia, USA, Apr. 2014.

• Paper 10. L. Shi, K.W. Sung and J. Zander, “Cellular TV Distribution inHeterogeneous Environments” to be submitted. 2014.

The author of this thesis proposed the original problem formulation and actedas leading author of these papers. Professor Jens Zander and Dr. Ki Won Sungprovided directions and valuable insights in all papers.

1.5.3 Thesis OutlineThis composite thesis is divided into two parts: an overview of the thesis andverbatim copies of the papers included in this thesis. The first part containssix chapters: Chapter 2 describes the research approach and the main contri-bution of the thesis. Chapter 3 and Chapter 4 provide brief summaries of thekey results from the studies on secondary access in TV white space and cellu-lar content distribution system, respectively. Chapter 5 presents a qualitativediscussion on the benefits and challenges of the investigated technologies forspectrum access. Finally, Chapter 6 draws the main conclusions and presentsthe future work. The second part contains the reprint of eight published confer-ence papers, one published journal article and one submitted journal article.

Chapter 2

Research Methodology

This thesis intends to answer the research question posed in the previouschapter through a holistic approach with a focus on technical analysis. Theoverall problem is divided into two scenarios with complementary time scaleand service requirements. The corresponding technical solutions, namely, thesecondary access in TV white space with little change in DTT service andspectrum allocation, and cellular TV distribution in a harmonized UHF band,are proposed for technical assessments. The studies are performed withinthe scope of each scenario aiming at developing evaluation methodologies toquantify the spectrum and capacity gains achievable with these solutions.

The analysis emphasizes the interdependency between the technical andregulatory domain, which has direct influence over the investigation scenarioand the solution space. The technical solutions proposed in the thesis employand build on existing regulatory frameworks whenever applicable, so that themethodology developed in the thesis could be easily incorporated into the fu-ture development of the regulatory framework. In the secondary access in theTVWS scenario, our focus is on the development of a scalable framework forcontrolling the aggregate secondary interference from short-range system op-erating on multiple TV channels; in cellular TV distribution, the focus is ondeveloping an evaluation methodology in conjunction with the resource man-agement for the cellular content distribution network.

The thesis further extends the discussion to the implications for the regu-lation and business domains based on the insights gained from the technicalassessment. Major challenges for the deployment of the technical solutionsare examined from the regulatory and business perspectives. The potentialbenefits of each solution are compared with a qualitative analysis. Fig. 2.1depicts an overview of our research methodology.

In this thesis, we have applied different research approaches to provide acomprehensive analysis on different aspects of our problem. The choice of aspecific approach is based on the size of the particular problem and the level

20

CHAPTER 2. RESEARCH METHODOLOGY 21

Scenarios

TV Service Requirement

Utilization of UHF Band

Near Future Long-Term Future

Secondary Access in

TVWSConverged

CellTV platorm

Qualitative Comparison

Spectrum Opportunity Assessment

Spectrum Saving

Evaluation

Regulation Policy

Secondary Interference Management

SFN/Unicast Resource

Management

User DemandDistribution

Figure 2.1: Overall research approach of the thesis.

of details of the results we are interested in. In certain cases, compromises aremade between approximation accuracy and computational complexity, but theeffects of simplified assumption are always verified against detailed simula-tion. In order to draw more direct implications for real world scenarios, weperform case studies based on real world statistics whenever applicable. Thefollowing sections describe the research methodologies for secondary accessin TV white space and cellular TV distribution.

2.1 Methodology for Analyzing the Spectrum Opportunityfor Secondary Access in TV White Space

To evaluate the capacity secondary access in TV white space, we have fol-lowed the methodology as illustrated in Fig. 2.2. Theoretical models are pro-posed to characterize the effect of secondary interference on TV receptionand verified through a measurement campaign [5, 6]. Then we developed ananalytical framework for controlling secondary transmission and ensuring pri-mary protection based on the existing regulatory framework [7, 8]. Applyingthis framework to spectrum opportunity assessment, we have obtained the per-missible secondary transmit power and the maximum density of the secondary


Scenario TV Coverage Short Range SU

Interference Modeling

Cumulative Multi-ch

Interference

Regulatory Framework

Secondary Media Access

Control

Secondary Access in TVWS

Secondary System Capacity

Spectrum Opportunity

Primary Protection Framework

Single Adj-Ch Interference

Single SU on Co-Ch

Multi SUs on Adj-Chs

Figure 2.2: Methodology for analyzing the spectrum opportunity for secondary access inTV white space

users allowed to access the spectrum [9]. Finally, the capacity of secondarysystem is evaluated by incorporating a CSMA/CA-based media access controlprotocol to coordinate the channel access among secondary users [10].

2.1.1 Scenario DescriptionIn the studied scenario for secondary access in TV white space, we assumethat the primary TV service would remain the same as today in terms of con-sumption pattern and channel/frequency allocation. Its quality of service isdetermined by the coverage quality, which is defined by the location probabil-ity for TV reception:

q1 = Pr{Ptv

Ø P min

tv

+ Itv

}, (2.1)

where Ptv

is the received TV signal power, and P min

tv

is the sensitivity levelof the TV receiver. I

tv

is the TV-to-TV self-interference. Since both the use-ful TV signal and interferences are subject to fading, the location probabilityis less than one. In most cases, the requirement is defined by 95% locationprobability. A typical TV transmitter has a coverage area of an 80 to 100 kmradius.

The secondary system considered in the thesis is short-range WiFi-likesystem deployed inside TV coverage area but operating on the locally unoc-cupied TV channels. Although the interference modeling and analysis are ap-plicable to cellular-like high power systems, the concern of adjacent channelinterference is more relevant for short-range systems deployed in high densityand probably in close proximity to the TV receivers.


2.1.2 Interference ModelingSince TV receivers are not designed to tolerate interference, most of the TVreceiver filter has poor adjacent channel selectivity. Secondary interferencecoming from an unoccupied channel adjacent to the receiving TV channelcould damage the TV reception. The effect of the adjacent channel interfer-ence (ACI) is represented by the protection ratio, “

k

, defined as follows:

SN

IN+k

= “k

. (2.2)

where SN

is the received TV signal power in channel N , and IN+k

is themaximum received secondary interference power on channel N + k beyondwhich it could cause visible distortion in the video quality. The value of “

k

varies slightly with the quality of different TV receivers, but it follows a gen-eral pattern that complies with previous measurement results

When the interference arrives on not only one but multiple different chan-nels, we notice that the damaging effect on TV reception is cumulative. There-fore, we proposed a theoretical model to describe this effect:

SN

Ømÿ

k=1I

N+k

“k

. (2.3)

This proposed mathematical model for the cumulative effect of multichannelsecondary interferences on the primary TV receiver is verified through mea-surements conducted in both laboratory and real environments with commer-cially available product (e.g. TV receiver and antennas). The measurement inthe laboratory is primarily for calibration purpose and to verify whether thesignal leakage into the cable would cause any harmful impact on the TV re-ception. To identify the maximum tolerable secondary interference, we keepincreasing the transmit power of the secondary device until visual distortionon the TV quality is observed. The measurement is repeated for different TVsignal power levels and adjacent channels on which the secondary devices aretransmitting [5, 6].

2.1.3 Primary Protection FrameworkTo ensure the protection of the primary system, we have first developed an an-alytical approach for estimating the maximum permissible transmit power ofa single secondary user. This approach is based on the regulatory frameworkdeveloped in [21]:

q2 = Pr{Ptv

Ø P min

tv

+ Itv

+ “su

(�f) GPsu

} Ø q1 ≠ �q, (2.4)

where Psu

is the secondary transmit power and “su

(�f) is the same protec-tion as defined in (2.2) with a frequency offset of �f between the interference


and TV signal. G is the coupling gain between the SU and the TV receiver,which can be estimated by the geo-location database. �q denotes the maxi-mum allowed decrement in location probability caused by the secondary in-terference.

Based on this formulation for primary protection, we have developed ananalytical approach to obtain the permissible secondary transmit power di-rectly from (2.4). The key here is to distinguish the case when TV receptionis actually violated by secondary interference and not due to the TV self-interference or shadow fading on the TV signal itself. Then the probabilis-tic constraint could be transformed into a linear equation in the logarithmicaldomain by approximating the shadow fading components by log-normal ap-proximation [7].

For the multiple secondary user case, we focus on developing a statisticalapproach to estimate aggregate adjacent channel interference from multiplerandomly deployed secondary users to complement the deterministic modelproposed by the regulator. The random placement of the secondary user ismodeled by the Poisson spatial process. Further, the fact that 95% of theadjacent interference is generated within a small distance of 500 meters fromthe TV receiver is used as the basis for our assumption that the secondary userscontributing to the aggregated interference would have similar transmit powerdue to geo-location database resolution limitation. Thus we can approximatethe aggregate secondary interference by a sum of lognormal random variablesusing a cumulant matching method. The permissible transmit power for thisset of secondary users can thus be obtained in logarithmic form in a similarfashion to the single user case. The accuracy of these assumptions is verifiedagainst extensive Monte-Carlo simulations [8].

2.1.4 Maximizing Spectrum OpportunityWe develop a complete analytical framework for multi-channel multi-user sec-ondary interference control by incorporating our model of adjacent channelinterference with other existing works on co-channel interference. With thecombined interference constraint, an optimization problem is formulated tomaximize the number of simultaneously active secondary users permitted toaccess the TV white space.

The original formulation has a probabilistic non-convex constraint. Toobtain the numerical solution efficiently, we first apply the Fenton-Wickinsonmethod [57] to approximate the aggregate interference with a single log-normalvariable and transform the constraint into a deterministic one. Then we isolatethe nonlinear term and approximate it with a linear constant. Therefore, thereformulated optimization problem can be easily solved by a linear program-ing algorithm in MATLAB. Again, the approximation shows a good match tothe simulation results with less than 1 dB difference in the worst case [9].


TX A in CH 1

SU to PU

Interference

Secondary Uplink

TX C in CH 1

TX D in CH 2

SU to SU

interference

TX B in CH 1

Contention domain of TX A

Figure 2.3: Interference and contention among secondary users.

2.1.5 Evaluating Secondary System CapacityTo evaluate the capacity of the secondary system, we assume the WiFi-likesecondary users follow the CSMA/CA protocol to coordinate the channel con-tention among themselves (see Fig.2.3). Its capacity is assessed in terms of theaverage throughput estimated as:

TP

avg

=

1

T

1

Ntot

Tÿ

t=1

Ntÿ

n=1TP

t

n

, (2.5)

where T is the total simulation time slots, Ntot

is the total number of deployedsecondary users, N

t

is the number of active secondary users transmitting with-out collision at time slot t. TP

t

n

is the throughput of the nth active link ob-tained by the Shannon formula:

TP

t

n

= BWn

log2(1 +

P y

su,i

gn,n

N0 + “pu

Ipu

+

qmœAy

nP y

su,i

gm,n

), (2.6)

where gn,n

denotes the path loss between the nth secondary transmitter and itsreceiver. “

pu

Ipu

represents the interference from the primary system and N0is the noise. As presented in [45], we define Ay

n

as the set of secondary usersoperating on channel y outside the contention domain of the nth transmittingsecondary user.

The transmit power of the secondary user is optimized so that the aver-age throughput is maximized, but it must be below the permissible transmit


CellTV Concept

'Worst-Case' Scenarios

Network Model

View Demand Model

Optimized Resource

Management

Inhomogeneous Environment

Feasibility Analysis

Quantify Spectrum Saving

Regulation Policy

Case Study: Stockholm Area

Figure 2.4: Methodology for evaluating the spectrum saving by cellular TV distribution.

power obtained by following the previously defined procedures for primaryprotection. All users are assumed to have infinite buffer size, but due to colli-sion avoidance, the number of simultaneously active users is less than the totalnumber of secondary users. This fact is modeled by a thinning process and isaccounted for in the primary protection calculation [10].

2.2 Methodology for Evaluating the Spectrum Saving byCellular TV Distribution

We have followed the methodology depicted in Fig.2.4 for the study on a cellu-lar content distribution network. The concept of ’CellTV’ is proposed and de-fined by its network architecture and service provision. The initial feasibilityanalysis is performed in cooperation with the Swedish regulator, PTS, to ver-ify if CellTV could provide the same level of service as DTT in the re-farmedUHF broadcast band in Sweden. Having observed that certain spectrum sav-ing could be achieved by CellTV but with different transmission modes ineach scenario, a resource management framework for CellTV is developedto optimize the network configuration and thus maximize the spectrum sav-ing in an inhomogenous environments. The flexibility for CellTV to adapt tothe varying local environment and viewing patterns is demonstrated through acase study of the greater Stockholm region with diverse morphologies.


Urban Suburban Rural

Cellular sites

SFN1

SFN2

Unicast

MBB

MBB

MBBEx Area

Ex Area

Ex Area

UH

F BC

Ban

d

Figure 2.5: Illustration of the CellTV concept.

2.2.1 CellTV ConceptCellTV denotes the concept of using cellular infrastructure and the UHF broad-cast band to distribute audio-visual content and replace the traditional terres-trial TV broadcasting services. The network is based on mobile infrastruc-ture with Multimedia Broadcast Multicast Services (MBMS) capability. TheTV content can be distributed either via a unicast data link or broadcast overSFNs. The transmission modes depend on the local infrastructure availabilityand can be dynamically switched according to the viewing demand. Othermobile traffic can also share the same cellular infrastructure and be deliveredover any unused frequency/time resource block (see Fig.2.5).

One of the key challenges for CellTV is to fulfill the strict requirementfor TV service coverage. In particular, the location probability for coveragerequirement, as defined in (2.1), must be higher than 95% at the coverageboundary. It is equivalent to 99% service availability within the whole TVcoverage area. In addition, due to the fluctuations in the traffic load for on-demand unicast service, there is a risk of temporary blocking which is notdefined in the traditional TV broadcast service. A strict requirement for tem-poral availability should be enforced on the CellTV system.

The TV service quality is also defined by its content. A TV channel couldbe either an HD program or an SD program. Each has a different minimumdata rate requirement. A TV channel could also contain either linear or non-linear content. While a linear channel could be distributed via both broadcastand unicast, a non-linear channel could only be distributed via a unicast linkas a video on-demand service. The number of TV channels also varies atdifferent locations. Certain TV channels are available only in local areas orcontain regional contents. The CellTV system must distribute all these contentto fulfill the viewing demand.

2.2.2 Feasibility AnalysisAs an initial study, we investigate the feasibility of the CellTV concept byperforming numerical analysis for the case of Sweden in the year 2020. TheTV viewing habits are assumed to remain largely unchanged, but the number


of TV channels, especially those contains HD content, is assumed to increase.The system concept is deemed feasible only if the TV service could be

adequately provided by the existing cellular infrastructure operating in the re-farmed 470-790 MHz band. In the case that less than 320 MHz spectrumwould be required, any spectrum remaining is considered as spectrum saving.The spectrum requirement is determined by the estimated spectrum efficiencyof CellV, which is defined by a simple modification of the Shannon formula.

ESE(bits/s/Hz) = —eff log2[(1 + ›eff SINR)]. (2.7)

Here, the parameter —eff

is determined by the Adjacent Channel LeakagesRatio (ACLR) requirements and protocol overheads. ›

eff

corresponds to theSINR which is mainly affected by the modulation, coding and MIMO modes.The SINR calculation is different for unicast and broadcast links. For broad-cast over SFN, the SINR includes the gain from constructive interference ar-rived within the cyclic prefix in an OFDMA-based SFN. The bandwidth re-quired for broadcasting can be directly obtained by dividing the required daterate of the TV programs with the effective spectrum efficiency [11].

For unicast, the SINR calculation is similar to a typical cellular systemwith certain network load, which depends on the total bandwidth available forunicast. The bandwidth for unicast also affects the blocking probability. Thus,a multi-Erlang model is developed to classify users with a different bandwidthrequirement, and the Kaufman-Roberts recursion [58] is used to derive thetotal bandwidth that satisfies the blocking probability requirement.

Sweden is chosen for our case study because it has excellent coveragein both cellular and DTT networks. Besides, the DTT service penetration inSweden is moderate, close to the average penetration level in Europe. We haveapplied the statistical data of population density distribution and cellular basestation location in our analysis. The dense urban area in Stockholm centerand sparsely populated area in northern Sweden are chosen for the feasibilitystudy as they are considered to be the most spectrum-demanding scenarios dueto high viewer density and low cellular infrastructure availability, respectively[59, 60] (see Fig. 2.6).

2.2.3 Resource Management for CellTVThe local infrastructure density and population density in each cell would af-fect the spectral requirement for each transmission mode. In an inhomogenousenvironment, the resources allocations in neighboring areas may be differentand could also affect each other’s spectral efficiency. Consequently, the overallspectrum requirement of the whole network is dependent on the local environ-ment at each base station. Thus, we have developed a framework to optimizethe resource management for CellTV and minimize its overall spectrum re-quirement.


80 km

Figure 2.6: Studied area in the Greater Stockholm region.

The optimization framework determines the transmission mode, the op-erating frequency for each TV channel at each base station. The associationbetween the transmitters and the SFNs is also determined by the optimiza-tion framework. All TV channels must be provided in all the cells, either byunicast or broadcast, fulling the coverage and blocking requirement [12, 13].

In contrast to the feasibility analysis on isolated homogenous environ-ments, here we apply the resource management framework to the whole areawithin an 80 km radius from Stockholm city center (as marked in Fig. 2.6) toquantify the spectrum saving achievable by CellTV. The selected area coversdistinctive morphologies and has a good coverage of both DTT and mobileservices. We divide this circular area into multiple rings. The transition fromdense urban to rural area is characterized by the gradual changes in base sta-tion density, and population density in each ring. We further assume differentpropagation conditions, transmitter and receiver types1 and over-the-air TVservice penetration ratios associated with different morphologies.

1For MIMO receivers, a spacing of more than 2-3 meters between each antenna element is required to avoidsignificant spatial correlation in rural areas with line-of-sight [61].

Chapter 3

Secondary Access in TV WhiteSpace

3.1 Interference Modeling

3.1.1 Adjacent Channel InterferenceSince TV receivers are not designed to tolerate interference, they are particu-larly vulnerable to adjacent channel interferences from secondary users, whichcould be deployed in close proximity to the TV receiver. A simulation studyin [62] has shown the effect of ACI on the spectrum opportunity of TV whitespace. The measurement results obtained in [5] (e.g. Fig.3.1) exhibit goodmatch with the simulation results, although the model is slightly pessimisticand should server as a lower-bound estimate.

Overall, the effect of adjacent channel interference leakage from an indoorsecondary device into the roof-top antenna or cable is almost negligible. Onthe other hand, a set-top TV antenna could be very sensitive to secondary in-terference from devices nearby and thus would severely limit the number ofavailable TV channels accessible for secondary devices. The effect of inter-ference on the first two adjacent channel is the most significant. For some TVreceivers, the interference received on channels around the N+9 channel couldbe equally damaging to the TV reception.

3.1.2 Cumulative Effect of Multichannel InterferenceTo model the cumulative effect, we first define an equivalent CCI, ˆI that wouldresult in the same effect as the aggregate interference experienced on multiplechannels. The equivalent CCI is approximated by the weighted sum of allACIs

ˆI =

mÿ

k=1I

k

“k

“0, (3.1)

30

CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 31

N+1 N+2 N+3 N+4 N+5 N+6 N+7 N+8 N+9 N+10N+11N+12N+13N+14N+150

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Adjacent Channel Index

Est

ima

ted

Usa

ge

Pro

ba

bili

ty f

or

WS

D t

ran

smis

sio

n

Set−top Antenna (Apartment in the city center)

Simulation ResultsMeasurement Data

Figure 3.1: Estimated spectrum availability on each adjacent channel [5].

where “0 is the minimum SIR ratio for successful TV reception, and “k

isthe minimum protection ratio of the k

th

neighboring channel. Ik

is the ACIreceived on the k

th

neighboring channel.As expected, the weighted sum power is always equal to or less than the

desired TV signal level; thus, we can conclude that the proposed model is ver-ified. Fig. 3.2 illustrates how the maximum permissible received interferencein each adjacent channel decreases as the number of simultaneously accessedchannels increases. The implication of the accumulative effect of multi chan-nel interference depends not only on the number of empty TV channels butalso on the amount of active secondary users transmitting on all these chan-nels [6].

3.2 Protection of Primary System

3.2.1 Single Secondary UserThe permissible transmit power for a single secondary user with known me-dian pathloss to the TV receiver can be obtained by solving (2.4) analytically.The original approach proposed by the ECC report [21] ignores the fact thatthe TV signal outage could also be caused by self-interference and shadowfading on the received signal. Consequently, this approach overestimates thepermissible power by almost 10 dB and could lead to significant violation to


1 2 3 5−60

−58

−56

−54

−52

−50

−48

−46

−44

−42

−40

Number of simultaneously used adjacent channels (1 WSD per adjacent channel )

Maxi

mum

tole

rable

inte

rfere

nce

pow

er

(dB

m)

TV signal level =−78dBm, N=Channel 27 (522MHz)

Ch. N−1 (Measurements)Ch. N+1 (Measurements)Ch. N+1 (Theoretical)Ch. N−1 (Theoretical)

Figure 3.2: Maximum interference power level that a certain adjacent channel can toleratevs. the number of WSDs simultaneously transmitting in TVWS [6]

the TV reception, which can clearly be seen in Fig.3.3 [7].To avoid this overestimation of spectrum opportunity, we propose to sep-

arate the effect of TV signal outage and secondary interference violation byconditional probability. Letting Z denote P

tv

≠ (P min

tv

+ Itv

), (2.4) can beshortened as

q2 = Pr

;P

su

Æ 1

“su

(�f)GZ

<. (3.2)

Note that Z Æ 0 represents the case when the TV signal is in outage due toself-interference or shadow fading in the TV signal, thus,

Pr {Z Æ 0} = 1 ≠ q1, (3.3)

and (3.2) is equivalent to the following:

q2 = Pr{Z < 0} · Pr

;P ú

su

Æ 1

“su

(�f)GZ|Z < 0

<

+ Pr{Z Ø 0} · Pr

;P ú

su

Æ 1

“su

(�f)GZ|Z Ø 0

<

=q1 · Pr

;P ú

su

Æ 1

“su

(�f)GZ|Z Ø 0

<.

(3.4)

Then Z Õ= Z|

z>0 can be safely approximated as a lognormal random vari-able and be expressed in dB. Finally, the permissible transmit power can be


0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 142

44

46

48

50

52

54

56

58

60

q1

Pe

rmis

sib

le S

U t

ran

smit

po

we

r (d

Bm

)

SE43 Approach

Proposed Approach

Monte Carlo Simulation

Lower Bound [5]

Figure 3.3: Permissible transmit power for WSD located at 50 km away from victim TVreceiver with different q1 (q2 = q1 ≠ 0.01, ‡

S

= 8dB, ‡G

= 10dB. The notations andreferences are defined in [7]).

expressed as follows:

Psu(dBm) Æ m

Z

Õ(dBm) ≠ mG(dB) ≠ “

su

(�f)(dB)

≠Ô

2erfc≠1[2(1 ≠ q2

q1)]

Ò‡2

Z

Õ(dB) + ‡2G(dB).

(3.5)

With the new approach, q2 is explicitly bounded by q1, suggesting that any ad-ditional secondary interference will only deteriorate the TV reception quality.In contrast, with the original approach introduced in [21], q2 can be potentiallyset to any value even higher than q1 despite its logical contradiction.

As can be seen in Fig 3.3, the results from the proposed approach arein good agreement with the Monte Carlo simulation results. The actual TVviolation is always kept below the target level.

3.2.2 Multiple Secondary UserCombining the model for the cumulative effect of multichannel interferencein (2.3) and the primary protection requirement defined in (2.4), we obtain thefollowing constraint for a multiple secondary user transmitting on unoccupied


TV channels:

qi,x

2 Ø Pr

Y]

[P i,x

tv

≠ P min

tv

≠ “tv

Ii,x

tv

Øÿ

yœX

≠i

P i,yúsu

“su

(�fx≠y

)

N

iÁÿ

n=1G

n

Z^

\ ,

(3.6)where N i

Á

denotes the total number of active WSDs transmitting on all adja-cent channels inside the dominant interference region centered at pixel i. X≠

i

defines the set of channels that are not occupied in pixel i.Furthermore, recall that the WSDs are deployed following the Poisson

distribution. We can use lognormal distribution to approximate the aggregateinterference from the dominant interference region. There are other candidatedistributions for the interference approximation, but lognormal distributionexhibits a good match of the upper tail of the actual interference distributionwhile being analytically tractable. Denote Gi

a

=

qN

iÁ

n=1 Gn

. The parametersof its lognormal approximation, µ‚

G

ia

and ‡‚G

ia

can be obtained by cumulantmatching [8]:

Ÿ1(Gi

a

) = expËµ‚

G

ia

+ ‡2‚G

ia

/2

È, (3.7)

Ÿ2(Gi

a

) =

Ëexp(‡2

‚G

ia

) ≠ 1

Èexp

12µ‚

G

ia

+ ‡2‚G

ia

2. (3.8)

The cumulant Ÿm

(Gi

a

) is given by

Ÿm

(Gi

a

) = 2fi⁄iÂµm

(Gf

)Âµm

(G◊

)

⁄R‘

d0

gm

r,su

(r)rdr. (3.9)

Here ⁄ is the secondary user density and Âµm

is the mth raw cumulant. R‘

isthe radius of the dominant interference region and d0 is the minimal physicaldistance between the secondary user and the TV receiver. Then we can fol-low the methodology described in the previous section for single secondaryuser to convert (3.6) into the logarithmic domain and obtain the permissiblesecondary transmit power for each available TV channel. As shown in Fig3.4, the results from the proposed approach closely match the Monte Carlosimulation and adapt to the variation of WSD density. In comparison, thedeterministic reference geometry approach proposed in the regulatory frame-work [21] is overly pessimistic and underestimates the spectrum opportunityfor adjacent channel operation [8].

3.3 Spectrum Opportunity and Secondary Capacity

To evaluate the spectrum opportunity, we have developed a complete analyt-ical framework for multi-channel secondary interference control by incorpo-rating our model for adjacent channel interference with existing works on co-channel interference. With the combined interference constraint, an optimiza-


0 200 400 600 800 1000 1200 1400 1600−30

−20

−10

0

10

20

30

40

50

SU density per km2

Pe

rmis

sib

le S

U t

ran

smit

po

we

r (d

Bm

)

Ref Geo @ N+1Proposed Approach @ N+1Simulation @ N+1Ref Geo @ N+5Proposed Approach @ N+5Simulation @ N+5

Figure 3.4: Permissible transmit power with different WSD densities ⁄. q1 = 0.99, qú2 =

0.95 [8].

tion problem is formulated to maximize the number of admitted secondaryusers as follows:

maxx

ki

mÿ

i=1

ÿ

¸œLi

x¸

i

s.t. Pr

;N

tv

+ Ik

eff,i Æ Sk

i

“0

<Ø qú

2 ,

0 Æÿ

¸œLi

x¸

i

Æ ci

,

0 Æ x¸

i

Æ cÕi

,

’fk

œ Fi

, ’ai

œ A.

(3.10)

Here, Ik

eff,i is the aggregate interference from both co-channel and adjacentchannels. x¸

i

is the number of WSDs allowed to transmit on channel f¸

inpixel a

i

. Li

is the set of unoccupied TV channels in pixel ai

, and m is thetotal number of pixels of the studied area. N

tv

is the noise floor and Sk

i

is thereceived TV signal power. c

i

defines the total number of WSDs in pixel ai

,while cÕ

i

set the limit on the maximum number of users that can be admitted oneach TV channel. Finally, F

i

represent the set of broadcasting TV channels inpixel a

i

and A the set of pixels of the studied area.


0.9 0.92 0.94 0.96 0.98 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Location Probability

F(x

)

Empirical CDF

q2 with contr. on both ACI and CCI

q2 with contr. on only ACI

q2 with contr. on only CCI

q1

Figure 3.5: Location probability distribution over all TV channels in all pixels [9].

By applying Fenton-Wickson lognormal approximation and the relaxationof the constraint as proposed in [42], the original optimization problem can besimplified into a linear optimization problem and solved by standard numericalsolvers. The numerical results indicate that a large number of low-power sec-ondary devices can be active simultaneously without violating TV receptionand the adjacent channel interference has the dominant impact on TV recep-tion (as illustrated in Fig. 3.5). A high-power secondary system, on the otherhand, can only transmit at limited locations and causes mostly co-channel in-terference to the TV receivers. Identifying these limiting factors that dominatethe TVWS availability could significantly simplify the regulation process withlittle risk for compromising the protection of TV reception and efficiency ofTVWS utilization [9].

To evaluate the capacity of a low-power secondary system, we includethe self-interference and channel contention among secondary users in the as-sessment. We assume the secondary system are multiple pairs of randomlydeployed secondary access points and user equipment. Each user will try toaccess any unoccupied TV channels assisted by a geolocation database, buthas to compete for those channels with other secondary users following theCSMA/CA protocol. As a worst-case scenario, we assume each user has fullload and infinite buffer size.

Instead of simply using the maximum permissible power, an optimal trans-


4 8 12 16 20 24 28 3250

75

100

125

150

175

200

Unoccupied TV channel number

Maxi

mum

ave

rage thro

ughput per

use

r (M

bps)

TVWS in Ptv=−65dbm Rural area

TVWS in Ptv=−71dbm Rural area

TVWS in Ptv=−65dbm Urban area

TVWS in Ptv=−71dbm Urban area

ISM band in Rural area

ISM band in Urban area

Figure 3.6: Maximum average throughput vs. unoccupied TV channel number and differ-ent TV signal P

tv

. ⁄ = 100 WSDs/km2 in rural, 3000 WSDs/km2 in urban, with 2 channelaggregation [10].

mit power P opt

su

is chosen by an exhaustive search in the simulation such thatthe average throughput is maximized:

P opt

su

= arg max

Psu<P

úsu

TP

avg

(Psu

), (3.11)

where P úsu

denotes the WSD maximum permissible transmit power due toeither physical limitation or primary protection constraint.

From our numerical results presented in Fig. 3.6, we conclude that low-power indoor secondary access in TV white space performs worse than a tradi-tional WiFi system in the ISM band even with more unoccupied TV channels,if the TV signal is low. On the other hand, if the TV signal is good or it is de-ployed in the urban environment, the secondary system could achieve a higherthroughput given a larger number of unoccupied TV channels. This could beexplained by the fact that the secondary interference and network congestion,rather than the primary interference constraint, is the major limiting factorfor secondary capacity except in rural areas with low TV signal quality. Thiseffect is illustrated by Fig. 3.7, which shows that the optimal secondary trans-mit power is considerably lower than the permissible level according to theprimary protection constraint [10].


−71 −70 −69 −68 −67 −66 −650

5

10

15

20

25

30

TV signal strength (dbm)

Seco

ndary

use

r tr

ansm

it pow

er

(dbm

)

Max permissible in Rural area

Max permissible in Urban area

Optimal value in Rural area

Optimal value in Urban area

Figure 3.7: Relation between permissible power and optimal power at different TV signalstrength. ⁄ = 30 WSDs/km2 in rural, 3000 WSDs/km2 in urban, 2 channel aggregation,unoccupied TV channel number = 8 [10].

Chapter 4

Cellular Content DistributionNetwork in a Re-farmed UHFBand

4.1 Feasibility Analysis

In the feasibility analysis, we have assumed two fixed TV channel allocationschemes. One is to broadcast all TV channels over SFN regardless of channelpopularity. The other is to broadcast only the four most popular TV channelsover SFN and unicast the rest of the less popular TV channels. Therefore, thekey is to calculate the spectral efficiency, which is a function of the signal-to-interference-and-noise-ratio (SINR) as defined in (2.7).

The SINR for broadcast over SFN in cell 0 is calculated as follows

SINR

broad

=

qm

i=0Ê(·i)P̄

qiqm

i=1(1≠Ê(·i))P̄

qi+ N0

, (4.1)

where ·i

is the propagation delay, ¯P is average power associated with basestation i, and q

i

represents the propagation loss to the base station i whichaccounts for distance-based path loss and shadowing. The total number ofcells in the MBSFN area is given by m. For a given · , the weight function ofthe constructive portion of a received SFN signal is given by [63] [64]:

Ê(·) =

Y____]

____[

0, · < ≠Tu

;1 +

·

Tu, ≠T

u

Æ · < 0;1, 0 Æ · < T

CP

;1≠(·≠TCP )

Tu, T

CP

Æ · < TCP

+ Tu

;0, otherwise,

(4.2)

where Tu

is the length of the useful signal frame and TCP

is the length of thecyclic prefix [52].

39

CHAPTER 4. CELLULAR CONTENT DISTRIBUTION NETWORK IN ARE-FARMED UHF BAND 40

0 500 1000 1500

50

100

150

200

250

300

350

Inter Site Distance (m)

Re

qu

ire

d B

an

dw

idth

(M

Hz)

Urban Broadcast 4*1 legacy Rx antenna

Urban Broadcast 4*4 diversity

Urban Broadcast 8*8 diversity

BW required by current DVB−T deployment: 320 MHz

Figure 4.1: Spectrum requirement for CellTV broadcast in a urban environment [11].

The unicast SINR of an arbitrary viewer at location ri

is expressed as

SINRuni

(ri

, X) =

¯P/q0(ri

)

qm

Õ

l=1 Xl

¯P/ql

(ri

) + N0(4.3)

where X is the interference collision vector conditioned on the network loadx and mÕ is the number of interfering base stations (sites allocated with thesame spectrum for unicast). The network load x in the system is obtained bysolving the fixed point equation

x = min

C(flHD BRHD +flSD BRSD) ·

⁄R

0

2r dr

R2 qX

[Pr(X|x

)BWuni

ESE(r, X)]

, 1

D.

(4.4)Here, ESE(r, x) is spectral efficiency averaged over shadow fading. fl isthe traffic intensity of unicast TV viewers in the cell and BR is the bit raterequirement of the TV channel. Note that the network load x is thus dependenton the total bandwidth available for unicast BW

uni

and the traffic intensity ofunicast TV viewers in a cell watching either HD or SD programs (flHD andflSD).

Based on our analysis in a typical Swedish urban and sparsely populatedarea, we concluded that it is possible to deliver the TV content using cellularinfrastructure with UHF television band. In dense urban areas, considerable


4 6 8 10 12 14 160

20

40

60

80

100

120

140

160

180

Inter Site Distance (Km)

Re

qu

ire

d B

an

dw

idth

(M

Hz)

Rural Unicast with 4x1 legacy Rx antenna, 0.1% blocking

Rural Unicast with 4x1 legacy Rx antenna, 5% blocking

Rural Unicast with 4x4 MIMO, 0.1% blocking

Rural Unicast with 4x4 MIMO, 5% blocking

Figure 4.2: Spectrum requirement for hybrid CellTV unicast-broadcast in a rural environ-ment (The legends specify different receiver antenna configurations for unicast links) [11].

spectrum saving of up to 250 MHz can be achieved by broadcasting all TVchannels over SFN (see Fig.4.1), thanks to the densely deployed base stations.In sparsely populated areas, broadcasting over SFN is no longer as efficient asin the urban area due to the diminishing SFN gain in areas with large inter-sitedistances. Therefore, only a few most popular TV channels are broadcasted.Unicasting the rest of the channels to the few viewers achieves the best resultwith around 200-150 MHz spectrum saving (see Fig.4.2). More spectrumsaving is possible if an advanced MIMO receiver antenna is deployed [11].

4.2 CellTV Deployment in Inhomogeneous Environments

The previous study has led us to the conclusion that the most efficient dis-tribution method is broadcast in urban and unicast in rural areas. However,we have treated these two environments as isolated entities. In reality, how-ever, the transition from one environment to another is gradual and continuous.There is no pre-defined boundary for the transmission mode to switch frombroadcast to unicast. Instead, the decision on the resources allocation betweenunicast and broadcast should be dependent on the local viewing demand andcellular infrastructure density. Furthermore, the resource allocation decisionin a certain area will affect its neighbors’ spectrum efficiency due to potentialinterference and/or SFN gains. Consequently, it will affect the spectrum effi-


0 10 20 30 40 50 60 70 800

50

100

150

200

250

300

350

400

450

Sp

ec

tru

m r

eq

uir

em

en

t p

er

ce

ll (

MH

z)

Distance from urban center (km)

Total local spectrum requirement

Spectrum occupied by assisting cells

Niche CH SFN1

POP CH SFN1

Niche CH Unicast

POP CH SFN2

RuralSuburbanUrban

Figure 4.3: Local spectrum requirement of CellTV in the Greater Stockholm region [12].

ciency of the whole network. The discrepancy between the resource allocationin different locations will inevitably reduce the overall spectrum efficiency ofthe CellTV network. Thus, we have developed a framework that optimizes thefrequency allocation, the transmission method of each TV channel, and theassociation of base stations to different SFNs. As a result the overall spectrumrequirement of the CellTV system is minimized under the constraint that allTV channels must be delivered either by broadcasting with good coverage orby unicast with low blocking probability [12].

For the quantitative analysis, we have used the population distribution dataand cellular base station location information in the area within an 80 km ra-dius from Stockholm city center. An interesting observation is that the pop-ulation per base station is in fact lowest in urban areas because of the highdensity of base stations. Similar trends have been identified in other Swedishcities as well.

By applying this optimization framework to a continuous area, we havefound that, while the overall spectrum requirement for CellTV is 185 MHzand there are positive spectrum savings in both urban and rural areas, thereis a local spectrum shortage in suburban areas where the transmission modesfor most of the TV channels switch from broadcast to unicast. The spectrumshortage is caused by the fact that, adjacent to the border of an SFN for broad-casting, a certain number of tiers of base stations are forbidden to reuse thesame spectrum band as the one occupied by the SFN to ensure that an accept-


1000 2000 3000 4000 5000 6000 700010

0

101

102

103

104

ISD (m)

Po

pu

lati

on

De

ns

ity

pe

r K

m2

BW Req = 200 MHzBW Req = 160 MHzBW Req = 120 MHzBW Req = 80 MHzBW Req = 40 MHzBW Req = 20 MHzStockholm Statistics

Figure 4.4: Spectrum requirement of CellTV in heterogeneous environments [13].

able spectral efficiency is achieved at the SFN border.The spectrum saving in rural areas is limited due to the relatively the higher

number of households per BS and a higher reliance on over-the-air TV recep-tion. Reducing the inter-site interference through advanced multi-cell coordi-nation could potentially alleviate the “spectrum bottleneck” and make unicastin rural areas more efficient [12],

To gain a deeper insight into the impact of the heterogeneous environmenton the spectrum requirement of CellTV, we extend the case study to a moregeneralized environment. As illustrated in Fig.4.4, the spectrum requirementis more sensitive to the inter-site distance when the base station density ishigh. In contrast, in rural areas with large inter-site distance, the spectrumrequirement becomes more sensitive to the population density.

These trends could be explained by optimal transmission modes in a het-erogeneous environment as depicted in Fig.4.5. In urban areas, most of the TVchannels are likely to be broadcasted and thus the inter-site distance has a sig-nificant impact on the spectral efficiency of CellTV. On the other hand, onlythe most popular TV channels may still be broadcasted and the rest shouldbe transmitted via unicast links. Thus, the spectrum requirement is mainlydependent on the population density [13].


1000 2000 3000 4000 5000 6000 700010

0

101

102

103

104

ISD (m)

Po

pu

lati

on

De

ns

ity

pe

r K

m2

Switching condition between unicast and SFN (MCC = 6dB)

Unicast all channelsBroadcast Popular CHs onlyBroadcast all CHsStockholm Statisitcs

Figure 4.5: Optimal transmission modes for TV content delivery in heterogeneous envi-ronments [13].

60 70 80 90 100 110 120 130150

200

250

300

350

400

Sp

ec

tru

m r

eq

uir

em

en

t (M

Hz)

Number of TV channels

New linear TV channels

New VoD channels

DVB−T2 (3 SFNs 256QAM 5/6 coding)

SpectrumSavings

Figure 4.6: Spectrum requirement for CellTV in the Greater Stockholm area with differentnumbers of TV channels [13].

Chapter 5

Discussion

This chapter presents the comparative analysis of the two alternatives fromthe perspectives of regulatory/business challenges and the effect of evolvingtrends in multimedia consumption. We will discuss the implication of thosechallenges and development on the roles of terrestrial TV and mobile broad-band.

5.1 Benefits

Based on previous technical analysis, here we present a qualitative comparisonof the potential benefits of each framework on the following aspects:

• CapacityThe major limitation of TV white space is its unsuitability for wide-rangesecondary access. Thus, the extra capacity for mobile broadband pro-vided by secondary access is less than what can be achieved by CellTV.Short-range secondary access would be able to achieve a higher capacity,but the spectrum opportunity is usually lower in dense urban areas dueto aggregated interference constraint and a higher number of occupiedTV channels. On the other hand, the spectrum saving by CellTV is thehighest in urban areas where the value of the extra capacity is also higherdue to a greater demand for mobile broadband. Thus, CellTV has a bet-ter capability to accommodate more data traffic for mobile broadbandand also more TV contents with better quality for TV service in UHFtelevision band.

• FlexibilityThe spectrum opportunity for secondary access shrinks as more spec-trum is occupied by an increasing number of TV channels or TV chan-nels with improved quality. In contrast, the spectrum saving of CellTV isless sensitive to the increase in the number of TV channels due to its uni-cast capability, although it still requires more bandwidth to support the

45

CHAPTER 5. DISCUSSION 46

improved quality or a larger audience base. Both systems can flexiblyadapt to the local environments by adjusting their transmission configu-rations to maximize the spectrum opportunity or spectrum saving. Nev-ertheless, as the spectrum sharing between TV and mobile broadbandservices in CellTV is achieved on the network layer, it can better adaptto future changes in demand and regulatory landscapes as compared tothe physical layer sharing between secondary system and DTT networkin TV white space.

• Societal benefitsIn the TV white space approach, the integrity of the TV broadcastingservice is maintained by the implementation of strict primary protectionrequirements. Although secondary access in TV white space is not suit-able for cellular-type system, it can still improve the wireless broadbandcoverage in rural and suburban area by exploiting the excellent propaga-tion conditions in the UHF band and providing indoor-to-outdoor radiocoverage. In the CellTV system, certain business models and regulationincentives must be developed to ensure that the mobile operators providefree-to-air access for the public TV channels. Cellular infrastructure isregarded as less reliable than the DTT network in emergency or disas-trous situations. But this concern is outweighed by the improvement inmobile broadband service that helps to reach the EU Digital Agenda’starget for near-universal broadband coverage.

• End-user experienceIn general, the wireless broadband service provided by short-range sec-ondary system in TV white space is similar to that of a typical WiFi sys-tem, but with a better coverage quality. On the other hand, the CellTVsystem not only enables cross-platform content accessibility but also in-teractivity for non-linear services. Wide-area broadband coverage alsoensures its service continuity.

5.2 Challenges

The challenges are discussed from the following perspectives:• Technology

After years of intense research and development, the technology for sec-ondary access in TV white space based on geo-location database is ap-proaching maturity. Several prototype systems are being tested in theU.S. and the UK. In contrast, the cellular content distribution systemis still in the conceptual development stage, although its core technicalcomponent, i.e., LTE with eMBMS, has already been deployed by a fewmobile operator on a limited scale and can be easily adjusted accordingto the specific system requirements. On the user device side, no commer-cial equipment for TV white space exists yet and it is unlikely to become


5

21

9

89

60

88

48

30 38 40

BE DK DE ES FR IT FI SE UK EU27

DTT Penetration (%)

Figure 5.1: Service penetrations of digital terrestrial TV in different EU member states [3].

available in the near future due to the lack of interest from the wirelessindustry. On the other hand, handsets with LTE eMBMS capability isexpected to reach the market by the year 2014. The more challengingissue for CellTV is the development of the MIMO antenna that can beeasily deployed in the UHF band.

• BusinessSince secondary access in TV white space is unsuitable for providingcellular-like broadband coverage and the short-range secondary systemlacks a decisive advantage over traditional Wi-Fi in the ISM band, thewireless industry has so far exhibited limited enthusiasm in its commer-cialization. For CellTV, mobile operators are facing the same challengethat they experienced in previous attempts to provide mobile TV ser-vice, i.e., to develop a viable business model for delivering TV servicesover cellular network. In particular, to provide those ’free-to-air’ publicchannels, it may require cooperation among multiple operators to utilizethe spectrum efficiently and thus complicate the control of their own sub-scribers. Nevertheless, the growing trends of on-demand viewing and theimproved capability for end-user device caching and pre-fetching may bethe key differentiations from previous business models centered arounda linear broadcasting service. Besides the possibility of extra revenuesources, a re-farmed UHF band is essential for providing additional mo-bile broadband capacity and it is a strong incentive for mobile operatorsto invest in the network infrastructure deployment and user equipmentupgrades necessary for the implementation of the CellTV system.

• RegulationDespite the strong initiative from regulators in both Europe and the U.S.


to develop secondary systems in TV white space, the decision to reallo-cate more spectrum in the post-digital-switchover UHF band to mobileservice has reinforced the reluctancy of the wireless industry to investin this technology and hindered its commercialization process. The dis-cussion on the possibility of a re-farmed UHF band is still at a veryearly stage. An official investigation led by the European Commissionhas just started at the beginning of 2014 and will be finished by the endof the year. Even if the results were to appear positive towards the re-farming option, the implementation of this re-farm process would en-counter strong resistance from the DTT broadcaster sector, escalating thetechnical issues to a political one. The fact that the dependence on theDTT service varies significantly across European countries would onlyaggravates the problems in creating a harmonized decision (see Fig.5.1).

5.3 Summary

Overall, the technology and regulation framework for secondary access in TVwhite space is more mature than CellTV. But due to the limited benefits of de-ploying short-range secondary system and the uncertainty in the future spec-trum allocation in the UHF band, it lacks the strong support of the wirelessindustry and has an unpromising prospect for any large-scale commercial de-ployment. In contrast, a converged content distribution platform could bringnumerous benefits for both the mobile industry and the end-user as discussedabove. Although it is still a considerable challenge to develop a sound busi-ness strategy for providing CellTV service, the real barriers to the harmonizedre-farming process may come from the political domain, i.e., the strong oppo-sition from the incumbents and the country differentiation within the EuropeanUnion [65].

Chapter 6

Conclusions and Future Work

6.1 Concluding Remarks

The UHF band between 470-790 MHz, currently occupied by digital terres-trial TV distribution in Europe, is widely regarded as a premium spectrumband for providing mobile coverage. The efficient utilization of this spec-trum is of great importance for the future success of the wireless industry.The digital terrestrial TV network has been the most popular platform for TVconsumption and has enjoyed exclusive access to this spectrum in the pastdecades. However, the consumption of linear mainstream TV content broad-casted by the DTT service is gradually giving way to on-demand service withdiversified contents. The increasing popularity of ’over-the-top’ service, i.e.delivering the audio-visual content over the internet to various platforms in-cluding mobile devices, is challenging the dominance of traditional linear TVin the living room. At the same time, the growing video consumption on mo-bile platforms via wireless broadband is creating a significant challenge formobile operators to provide sufficient capacity with the existing infrastructureand spectrum.

In light of these developing trends, a series of recent regulatory decisionhas opened up the possibilities for wireless broadband service to access theUHF broadcast band. This dissertation has studied two possible approachesfor reusing the spectrum by the wireless broadband service with distinctivetimeframes and scopes. The short-term option, with the assumption of limitedchanges in TV consumption patterns in the near future, is spectrum-sharingbetween legacy TV broadcast networks and secondary systems for wirelessbroadband services. The long term option instead envisages a converged all-IPplatform based on cellular networks operating in a re-farmed UHF broadcastband to provide both TV and mobile services.

For secondary access in TV white space, we have identified that the ex-isting regulation policy could result in not only unreliable primary protec-

49

CHAPTER 6. CONCLUSIONS AND FUTURE WORK 50

tion but also underestimation of the TV white space availability. A statisticalmodel and analytical algorithm is proposed to augment the existing regula-tion framework in estimation accuracy and computation efficiency. Applyingthese models, we have demonstrated that there is a considerable potential forlow-power secondary system. Due to the cumulative effect of multi-channelinterference, the adjacent channel interference is the dominant factor that lim-its the spectrum opportunity for secondary devices deployed with high density.On the other hand, a densely deployed secondary system based on CSMA/CAprotocol is able to achieve a throughput comparable to that of a traditionalWi-Fi of similar bandwidth in the ISM band. In fact, our analysis shows thatthe capacity of the secondary system is mostly limited by self-interferenceand network congestion within the secondary system rather than by primaryprotection constraint.

Being a novel concept, the converged content distribution platform basedon cellular network operating in a re-farmed UHF band is only beginningto emerge as the regulation environment and video consumption trends haveevolved in recent years. The feasibility of a cellular content distribution net-work has been investigated from technical, regulatory and business perspec-tives. With the advancement of cellular technology, the cellular TV distribu-tion is shown to provide a improved quality of service and flexibility for TVviewers. At the same time, it is able to use the provide extra capacity for othermobile services in the UHF broadcast band.

We have addressed the deployment issue of CellTV in an inhomogeneousenvironment by developing a resource management framework. In a casestudy of the Swedish environment, we have identified a significant amountof spectrum saving in dense urban areas by employing broadcast over singlefrequency network consisting of closely deployed base stations. Considerablespectrum saving is also observed in rural and sparsely populated area as thelimited number of viewers per cell ensures the efficiency of unicast for theniche TV channels in an on-demand fashion. Finally, we have demonstratedthe flexibility advantage of the CellTV system which enables it to adapt to thefuture changes in TV consumption trends more efficiently.

Overall, we conclude that substantial additional capacity could be gainedby allowing other wireless broadband systems to access the UHF broadcastband. Nevertheless, both of the investigated approaches for spectrum accessare facing great opportunities and challenges. The technical and regulationframework is maturing gradually for secondary access in TV white space usinggeo-location databases. However, the strong initiative taken by the regulatorhas thus far failed to stimulate enough interest from the industry, due to theuncertainty in future spectrum allocation in the UHF band and the fact that TVwhite space is less suited for cellular-type secondary systems. Thus we canforesee a limited commercial success of secondary access in TV white spacein the near future, mostly in the form of short-range system offloading and asa testbed for novel spectrum sharing mechanisms.

CHAPTER 6. CONCLUSIONS AND FUTURE WORK 51

On the other hand, CellTV has great flexibility to adapt to future changesand can potentially provide access for mobile broadband to a considerableamount of spectrum in the UHF band. These are strong incentives for mobileoperators to promote the converged platform as a long-term strategy. But thespectrum re-farming would be a long and difficult process, which is furthercomplicated by the strong opposition from the incumbent and the heteroge-neous situation among the EU member states. More quantitative evidence onthe benefits of CellTV is needed to influence future regulatory decisions andpersuade the mobile operator to invest in the necessary infrastructure and userdevice upgrades.

6.2 Future Work

This dissertation has contributed to the increasingly thorough study on thetechnical aspects of secondary access in TV white space, although there isstill a considerable room for improvement in secondary access control and thecoexistence of multiple secondary users. As demonstrated by recent researchtrends, the study on how to realize secondary access is shifting its focus fromtechnical analysis and implementation to refining the spectrum-sharing con-cept and improving the regulatory framework to facilitate secondary access.

The study on the converged content distribution system in a re-farmedUHF band has only taken its initial step. So far we have only assumed that itwould rely entirely on cellular infrastructure. The possibility to combine DTTinfrastructure and cellular ones has not been fully exploited yet. The enhance-ment in the computation and storage capability of user devices also deservesfurther attention, as device caching and pre-fetching add another dimension tothe content distribution system.

Bibliography

[1] Electronic Communications Committee (ECC) within the EuropeanConference of Postal and Telecommunications Administrations (CEPT),“The European Table of Frequency Allocations and Applications inthe Frequency Range 8.3 kHz to 3000 GHz ,” Ocotober 2013.[Online]. Available: http://www.erodocdb.dk/docs/doc98/official/pdf/ERCRep025.pdf

[2] Federal Communications Commission, “FCC Online Table of FrequencyAllocations,” April 2014. [Online]. Available: http://transition.fcc.gov/oet/spectrum/table/fcctable.pdf

[3] L. Backlund, “The importance of UHF Spectrum forDTT and European Consumers,” June 2013. [Online]. Avail-able: http://www.broadcast-networks.eu/wp-content/uploads/2013/07/BNE-European-Spectrum-Summit-2013-D.pdf

[4] Cisco, “ Cisco Visual Networking Index: Global Mobile DataTraffic Forecast Update, 2013Ð2018,” February 2014. [On-line]. Available: http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/

[5] E. Obregon, L. Shi, J. Ferrer, and J. Zander, “Experimental Verificationof Indoor TV White Space Opportunity Prediction Model,” in 5th Inter-national Conference Cognitive Radio Oriented Wireless Net- works andCommunications, June 2010.

[6] ——, “A Model for Aggregate Adjacent Channel Interference in TVWhite Space,” in IEEE 73rd Vehicular Technology Conference, May2011.

[7] L. Shi, K. W. Sung, and J. Zander, “Permissible Transmit Power forSecondary User in TV White Space,,” in 7th International ConferenceCognitive Radio Oriented Wireless Networks and Communications, June2012.

[8] ——, “Controlling Aggregate Interference under Adjacent Channel In-terference Constraint in TV White Space,” in 7th International Confer-

52

BIBLIOGRAPHY 53

ence Cognitive Radio Oriented Wireless Networks and Communications,June 2012.

[9] ——, “Secondary Spectrum Access in TV Bands with Combined Co-Channel and Adjacent Channel Interference Constraint,” in IEEE Inter-national Symposium on Dynamic Spectrum Access Networks , October2012.

[10] Y. P. Yang, L. Shi, and J. Zander, “On the Capacity of Wi-Fi Systemin TV White Space with Aggregate Interference Constraint,” in IEEEInternational Symposium on Dynamic Spectrum Access Networks , July2013.

[11] L. Shi, E. Obregon, K. W. Sung, J. Zander, and J. Boström, “CellTV - onthe Benefit of TV Distribution over Cellular Networks A Case Study,”IEEE Transaction on Broadcasting, vol. 60, no. 1, pp. 73–84, March2014.

[12] L. Shi, K. W. Sung, and J. Zander, “Spectrum Requirement for CellularTV distribution in UHF Band from Urban to Rural Environment,” inIEEE International Symposium on Dynamic Spectrum Access Networks(DySPAN 2014), April 2014.

[13] ——, “Cellular TV Distribution in Heterogeneous Environments,” sub-mitted for publication, 2014.

[14] Ericsson, “ 2013 TV and Media report,” June 2013. [On-line]. Available: http://www.ericsson.com/res/docs/2013/consumerlab/tv-and-media-consumerlab2013.pdf

[15] G. Winslow, “ Mobilizing for Mobile DTV,” April 2012. [Online]. Avail-able: http://www.broadcastingcable.com/article/483472-Mobilizing_for_Mobile_DTV.php

[16] J. Zander and P. Mahonen, “Riding the Data Tsunami in the Cloud:Myths and Challenges in Future Wireless Access,” IEEE Communica-tions Magazine, vol. 51, no. 3, pp. 145–151, March 2013.

[17] R. Schoenen and H. Yanikomeroglu, “User-in-the-Loop: Spatial andTemporal Demand Shaping for Sustainable Wireless Networks,” IEEECommunications Magazine, vol. 52, no. 2, pp. 196–203, February 2014.

[18] Cisco, “Cisco Visual Networking Index: Global Mobile DataTraffic Forecast Update, 2010-2015,” 2010. [Online]. Available:http://www.cisco.com/en/US/netsol/ns827/.

[19] M. Nekovee, “Quantifying the Availability of TV White Spaces for Cog-nitive Radio Operation in the UK,” in IEEE International Conference onCommunicatgoions Workshops, June 2009.

BIBLIOGRAPHY 54

[20] Federal Communications Commission, “In the Matter of UnlicensedOperation in the TV Broadcast Bands: Second Memorandum Opinionand Order,” September 2010. [Online]. Available: http://www.fcc.gov

[21] Electronic Communications Committee, “Technical and OperationalRequirements for the Possible Operation of Cognitive Radio Systems inthe White Spaces of the Frequency Band 470-790MHz,” ECC Report159, January 2011. [Online]. Available: http://www.ero.dk

[22] C. Ghosh, S. Roy, and D. Cavalcanti, “Coexistence Challenges for Het-erogeneous Cognitive Wireless Networks in TV White Spaces,” IEEETransaction on Wireless Communications, vol. 18, no. 4, pp. 22 –31,August 2011.

[23] European Telecommunications Standards Institute, “ White SpaceDevices (WSD); Wireless Access Systems Operating in the 470 MHzto 790 MHz Frequency Band; Harmonized EN Covering the EssentialRequirements of Article 3.2 of the R&TTE Directive ,” February 2014.[Online]. Available: {http://www.etsi.org/deliver/}

[24] OFCOM, “ Future Use of the 700MHz Band,” April 2013. [Online].Available: http://stakeholders.ofcom.org.uk/consultations/700mhz-cfi/

[25] C. C. March, “Broadcasters and Radio Spectrum: the Emergence of aEuropean Dgital Dividend in the United Kingdom and Spain.” in 22ndEuropean Regional Conference of the International TelecommunicationsSociety, 2011.

[26] European Broadcasting Union, “Benefits and Limitations of SingleFrequency Networks (SFN) for DTT,” June 2013. [Online]. Available:tech.ebu.ch/docs/techreports/tr016.pdf

[27] Dagens Nyheter, “ Digital-TV Fär Nu Lämna Frekvenser tillMobilerna,” February 2014. [Online]. Available: http://www.dn.se/debatt/digital-tv-far-nu-lamna-frekvenser-till-mobilerna/

[28] 3GPP, “ Introduction of the Multimedia Broadcasting Multicast Service(MBMS) in the Radio Access Network (Rel.7 and 8),” December 2007.[Online]. Available: http://www.qtc.jp/3GPP/Specs/25346-750.pdf

[29] T. Yucek and H. Arslan, “A Survey of Spectrum Sensing Algorithmsfor Cognitive Radio Applications,” IEEE Communications Surveys andTutorials, vol. 11, no. 1, pp. 116 –130, 2009.

[30] R. Murty, R. Chandra, T. Moscibroda, and P. Bahl, “SenseLess: ADatabase-Driven White Spaces Network,” IEEE Transactions on MobileComputing, vol. 11, no. 2, pp. 189 –203, February 2012.

BIBLIOGRAPHY 55

[31] J. van de Beek, J. Riihijarvi, A. Achtzehn, and P. Mahonen, “TV WhiteSpace in Europe,” IEEE Transactions on Mobile Computing, vol. 11,no. 2, pp. 178 –188, February 2012.

[32] D. Prendergast, Y. Wu, and G. Gagnon, “The Effect of UnlicensedCognitive Device Operation on Digital Television Performance in theVHF/UHF Band,” in IEEE Symposium on Radio and Wireless, January2008, pp. 499 –502.

[33] Electronic Communications Committee, “Measurements on the Perfor-mance of DVB-T Receivers in the Presence of Interference from theMobile Service (Especially from LTE),” ECC Report 148, June 2010.[Online]. Available: http://www.ero.dk

[34] B. Randhawa, “Conducted Measurements to Quantify DVBTInterference into DTT Receivers,” 2009. [Online]. Avail-able: http://stakeholders.ofcom.org.uk/binaries/consultations/ddr/statement/ERA2.pdf

[35] K. Harrison, S. Mishra, and A. Sahai, “How Much White-Space Ca-pacity Is There?” in IEEE Symposium on New Frontiers in DynamicSpectrum, April 2010.

[36] R. Januarytti, J. Kerttula, K. Koufos, and K. Ruttik, “Aggregate Inter-ference with FCC and ECC White Space Usage Rules: Case Study inFinland,” in IEEE Symposium on New Frontiers in Dynamic Spectrum,May 2011.

[37] European Broadcasting Union, “The Cumulative Effect of MultipleWSD Interference,” September 2011. [Online]. Available: http://www.cept.org/meeting-documents

[38] H. Karimi, “Geolocation Databases for White Space Devices in the UHFTV Bands: Specification of Maximum Permitted Emission Levels,” inNew Frontiers in Dynamic Spectrum Access Networks (DySPAN), 2011IEEE Symposium on, May 2011, pp. 443 –454.

[39] EU FP7 Project QUASAR , “Model Integration and Spectrum Assess-ment Methodology,” March 2011.

[40] K. W. Sung, L. Shi, and J. Zander, “Coexistence of lte femtocells withgsm cellular network,” in Personal Indoor and Mobile Radio Communi-cations (PIMRC), 2010 IEEE 21st International Symposium on, Septem-ber 2010, pp. 1556 –1560.

[41] J. K. Yngve Selen, “Optimizing Power limits for White Space Devicesunder a Probability Constraint on Aggregated Interference,” in IEEE In-ternational Symposium on Dynamic Spectrum Access Networks, October2012.

BIBLIOGRAPHY 56

[42] K. Ruttik, K. Koufos, and R. Jäntti, “Modeling of the secondary sys-tem’s generated interference and studying of its impact on the secondarysystem design ,” Radio Engineering, vol. 18, no. 7, pp. 1271–1278, De-cember 2010.

[43] K. Koufos, K. Ruttik, and R. Januarytti, “Controlling the Interferencefrom Multiple Secondary Systems at the TV Cell Border,” in IEEE 22ndInternational Symposium on Personal Indoor and Mobile Radio Com-munications (PIMRC), September 2011.

[44] S. Kawade and M. Nekovee, “Broadband Wireless Delivery Using anInside-Out TV White Space Network Architecture,” in IEEE GlobalTelecommunications Conference, 2011.

[45] L. Simic, M. Petrova, and P. Mahonen, “Wi-Fi, but Not on Steroids: Per-formance Analysis of a Wi-Fi-Like Network Operating in TVWS UnderRealistic Conditions,” in IEEE International Conference on Communi-cations, 2012.

[46] S. A. Lery, W. H. Paik, and R. M. Rast, “Extending HDTV CoverageUsing Low Power Repeaters: Cellular Approach,” IEEE Transactionson Broadcasting, vol. 38, no. 3, pp. 145–150, December 1992.

[47] F. Hartung, U. Horn, J. Huschke, M. Kampmann, T. Lohmar, andM. Lundevall, “Delivery of Broadcast Services in 3G Networks,” IEEETransactions on Broadcasting, vol. 53, no. 1, pp. 188–199, 2007.

[48] D. Catrein, J. Huschke, and U. Horn, “Analytic Evaluation of a HybridBroadcast-Unicast TV Offering,” in IEEE Vehicular Technology Confer-ence, May 2008.

[49] T. Lohmar and U. Horn, “Hybrid Broadcast-Unicast Distribution of Mo-bile TV over 3G Networks,” in Local Computer Networks, Proceedings2006 31st IEEE Conference on, 2006, pp. 850–851.

[50] D. Gomez-Barquero, N. Cardona, A. Bria, and J. Zander, “AffordableMobile TV Services in Hybrid Cellular and DVB-H Systems,” IEEE Net-work, vol. 21, no. 2, pp. 34–40, 2007.

[51] J. Huschke, “Facilitating Convergence between Broadcasting and Mo-bile Services Using LTE Networks,” in Telecom World (ITU WT) Tech-nical Symposium at ITU, 2011, pp. 123–128.

[52] L. Rong, S. Elayoubi, and O. Haddada, “Performance Evaluation of Cel-lular Networks Offering TV Services,” IEEE Transactions on VehicularTechnology, vol. 60, no. 2, pp. 644–655, 2011.

[53] U. H. Reimers, “ Quo Vadis - Germany as a DTT Case Study,” February2014. [Online]. Available: https://tech.ebu.ch/docs/techreview/trev_2013-Q1_DTT_Reimers.pdf

BIBLIOGRAPHY 57

[54] J. Huschke, J. Sachs, K. Balachandran, and J. Karlsson, “Spectrum Re-quirements for TV Broadcast Services Using Cellular Transmitters,” inIEEE Symposium on New Frontiers in Dynamic Spectrum Access Net-works (DySPAN) , May 2011.

[55] V. Kamble, S. Kalyanasundaram, V. Ramachandran, and R. Agrawal,“Efficient Resource Allocation Strategies for Multicast/Broadcast Ser-vices in 3GPP Long Term Evolution Single Frequency Networks,” inIEEE Wireless Communications and Networking Conference, 2009.

[56] A. Ligeti, “Single Frequency Network Planning,” Ph.D. dissertation,KTH, Signals, Sensors and Systems, 1999.

[57] L. F. Fenton, “The Sum of Log-Normal Probability Distributions in Scat-ter Transmission Systems,” IRE Transactions on Communications Sys-tems, vol. 8, no. 1, pp. 57–67, 1960.

[58] J. W. Roberts, “A Service System with Heterogeneous User Require-ments,” Performance of Data Communications Systems and Their Appli-cations, vol. 38, no. 3, pp. 423–431, December 1981.

[59] Post och telestyrelsen, “Data on Current GSM and UMTS NetworkInfrastructure in Sweden,” June 2012. [Online]. Available: http://www.pts.se/

[60] ——, “ Data on Population Density in Sweden,” June 2012. [Online].Available: http://www.pts.se/

[61] J. Y. R. Parviainen, J. P. Nuutinen, P. Talmola, J. Henriksson, H. Himma-nen, R. Ekman, and E. Huuhka, “Experimental Investigations on MIMORadio Channel Characteristics on UHF Band,” in IEEE InternationalConference on Communications, June 2009.

[62] E. Obregon. and J. Zander, “Short Range White Space Utilization inBroadcast Systems for Indoor Environments,” in IEEE Symposium onNew Frontiers in Dynamic Spectrum, April 2010.

[63] P. Mogensen, N. Wei, I. Z. Kovács, F. Frederiksen, A. Pokhariya, K. I.Pedersen, T. Kolding, K. Hugl, and M. Kuusela, “LTE Capacity Com-pared to the Shannon Bound,” in IEEE Conference on Vehicular Tech-nology, June 2007.

[64] R. Rebhan and J. Zander, “On the Outage Probability in Single Fre-quency Networks for Digital Broadcasting,” IEEE Transactions onBroadcasting, vol. 39, no. 4, pp. 395–401, December 1993.

[65] L. Shi, K. W. Sung, and J. Zander, “Opportunities and Challenges for TVand Mobile Broadband in 470-790 MHz,” in 24th European RegionalInternational Telecommunication Society Conference , October 2013.

Documents

Efﬁcient Spectrum Utilization of UHF Broadcast Band722137/FULLTEXT01.pdf · to reuse the spectrum unoccupied by the primary DTT network. On the other hand, the declining inﬂuence