Efficient Spectrum Utilization of UHF


Doctoral Thesis in and Technology Stockholm, Sweden, 2014 TRITA-ICT-COS-1406 KTH Communication Systems ISSN 1653–6347 SE-100 44 Stockholm ISRN KTH/COS-14/06-SE 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


The UHF band between 470-790 MHz, currently occupied by digital ter- restrial TV (DTT) distribution in Europe, is widely regarded as a premium spectrum band for providing mobile coverage. With the exponential increase in data traffic in recent years, there has been growing interests in gaining access to this spectrum band for services. The secondary access in TV White Space is considered as one cost-effective way to reuse the spectrum unoccupied by the primary DTT network. On the other hand, the declining influence of DTT and the converging trends of con- sumption on TV and mobile platforms are new incentives for the regulator to reconsider the optimal utilization of the UHF broadcast band. The proposal to re-farm the UHF band for a converged content distribution network was born under theses circumstances. This thesis intends to develop a methodology for evaluating the technical performance of these options for utilizing UHF broadcast band and quantify- ing their gains in terms of achievable extra capacity and spectrum savings. For the 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 due to the cumulative effect of multichannel interference. On the other hand, this potential does not translate directly into capacity for a WiFi-like secondary system based on CSMA/CA protocol, as the network congestion and self- interference within the secondary system has a greater impact on the network throughput 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 for a converged platform. This platform is based on cellular infrastructure and can provide TV service with the same level of quality requirement as DTT by delivering the video content via either broadcast or unicast as the situa- tion dictates. We have developed a resource manage framework to minimize its spectrum requirement for providing TV service and identified a significant amount of spectrum that can be reused by the converged platform to provide extra 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 prospect than the secondary access in TV white space in the long run. Nevertheless, low power secondary is still considered as a flexible and low-cost way to exploit the underutilized spectrum in the short term, despite its uncertainty in future availability. On the other hand, the re-farming of UHF broadcast band is a long and difficult regulation process with substantial opposition from the in- cumbent.The results from this study could serve as input for future regulatory decisions on the UHF band allocation and cost-benefit analysis for deploying new systems to access this spectrum band. Acknowledgements

Along the journey towards making this PhD thesis during the last four and a 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 truly unforgettable and rewarding experience, thanks to the encouragement, support and friendship from the people in this environment. Therefore, I would like to take the opportunity here to express my sincere gratitude to these special people. 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 study here. His critical and visionary thinking has always inspired me the most and taught me to live with and even enjoy the uncertainties in both research and the life ahead. I am also very thankful for his trust and support, especially for my application to the EIT ICT doctoral program, which may turn out into another wonderful journey into the ’uncertainties’. I am also deeply in debt to Docent Ki Won Sung, my co-advisor, with whom 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 I have learned a great deal from his expertise and knowledge that he has always shared without any reservation. I greatly appreciate Docent Svante Signell, for his time spent in reviewing my Licentiate thesis. I owe special thanks to Dr. Tim Irnich, for accepting the role of opponent for my Licentiate thesis defense. I would like to thank Lasse Wieweg for reviewing my PhD thesis proposal and Prof. Mats Bengtsson for his valuable comments during the evaluation of this thesis. My special thanks to Prof. Linda Doyle for accepting to be my opponent in the Doctoral dissertation. 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 together with many talented people in different research projects. I am especially thankful to Evanny Obregon and Javier Ferrer: the time we spent working together has been a fun and fruitful year at the beginning of the PhD. I would also like to say thanks to all of you who made the Communication

ii iii

System group such a unique and friendly environment: Ali Özyagci, Sibel Tombaz, 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, Göran Andersson, Anders Västberg and many more. I am also grateful to Sarah Winther, Jenny Minnema, Gunilla Gabrielsson and Lissi Perseus for their excellent administrative support. My special thanks goes to Irina Radulescu and Dr. Bogdan Timus, not only for their help even before I started my PhD but also for their genuine friendship. It is just impossible to describe how grateful I am for my parents, Rudong Shi and Ling Yang. Your unconditional love has always been the pillar for me to lean on. I am also extremely grateful to my wife, Molan, for coping with my fluctuating stress level during all these years. But you have not only helped me 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 and understanding, I would never reach this far. Contents

Preface ii

Contents iv

List of Figures vi

List of Abbreviations viii


1 Introduction 2 1.1 Background ...... 2 1.2 Spectrum Access in UHF Band ...... 6 1.3 Previous Work ...... 9 1.4 Problem Formulation ...... 13 1.5 Thesis Contributions and Outline ...... 16

2 Research Methodology 20 2.1 Methodology for Analyzing the Spectrum Opportunity for Secondary Access in TV White Space ...... 21 2.2 Methodology for Evaluating the Spectrum Saving by Cellular TV Distribution ...... 26

3 Secondary Access in TV White Space 30 3.1 Interference Modeling ...... 30 3.2 Protection of Primary System ...... 31 3.3 Spectrum Opportunity and Secondary Capacity ...... 34

4 Cellular Content Distribution Network in a Re-farmed UHF Band 39 4.1 Feasibility Analysis ...... 39 4.2 CellTV Deployment in Inhomogeneous Environments . . . . 41

5 Discussion 45


5.1 Benefits ...... 45 5.2 Challenges ...... 46 5.3 Summary ...... 48

6 Conclusions and Future Work 49 6.1 Concluding Remarks ...... 49 6.2 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 TV White Space (P2) 67

9 Permissible Transmit Power for Secondary User in TV White Space (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 in 470-790 MHz (P8) 122

15 Spectrum Requirement for Cellular TV distribution in UHF Band 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] ...... 3 1.2 TV consumption share on different platforms (%) [3]...... 3 1.3 Wireless data traffic growth forecast (Exabytes per month) [4]. . . 4 1.4 Secondary Access in TV White Space...... 6 1.5 Cellular content distribution network...... 8 1.6 Summary of related literature in TV white space...... 10 1.7 The correlation of the investigated scenarios...... 13

2.1 Overall research approach of the thesis...... 21 2.2 Methodology for analyzing the spectrum opportunity for secondary access in TV white space ...... 22 2.3 Interference and contention among secondary users...... 25 2.4 Methodology for evaluating the spectrum saving by cellular TV distribution...... 26 2.5 Illustration of the CellTV concept...... 27 2.6 Studied area in the Greater Stockholm region...... 29

3.1 Estimated spectrum availability on each adjacent channel [5]. . . . 31 3.2 Maximum interference power level that a certain adjacent channel can tolerate vs. the number of WSDs simultaneously transmitting in TVWS [6] ...... 32 3.3 Permissible transmit power for WSD located at 50 km away from victim TV receiver with different q (q = q 0.01, ‡ =8dB, 1 2 1 ≠ S ‡G = 10dB. The notations and references are defined in [7]). . . 33 3.4 Permissible transmit power with different WSD densities ⁄. q1 = 0.99, q2ú =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- 2 ber and different TV signal Ptv. ⁄ = 100 WSDs/km 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, 3000 WSDs/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 a rural environment (The legends specify different receiver configurations for unicast links) [11]...... 41 4.3 Local spectrum requirement of CellTV in the Greater Stockholm region [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 area with 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 • CCI Co-Channel Interference • CSMA/CA Carrier Sense Multiple Access with Collision Avoidance • DTT Digital Terrestrial • DVB-H - Handheld • DVB-T Digital Video Broadcasting - Terrestrial • ECC Electronic Committee • EIRP Equivalent Isotropically Radiated Power • ETSI European Standards Institute • FCC Federal Communications Commission • HD High Definition • ISD Inter Site Distance • ISI Inter Interference • ISM Industrial Scientific and Medical • LSA Licensed Spectrum Access • LTE Long Term Evolution • MIMO Multiple Input and Multiple Output • OFDMA Orthogonal -Division Multiple Access • PU Primary User • SD Standard Definition • SFN Single-Frequency Network • SINR Signal to Interference and Ratio • SINR Signal to Interference Ratio • SU Secondary User

viii List of Figures ix

• TVWS TV • UHF Ultra • UMTS Universal Mobile Telecommunications System • VHF • VoD Video On-Demand • WRAN Wireless Regional Area Networks • WRC World Radio Communication Conferences • WSD White Space Device • eMBMS Evolved Multimedia Broadcast and Service Part I

1 Chapter 1


1.1 Background

1.1.1 UHF band and broadcast The broadcast service initially started to operate in the Very High Frequency band (VHF). It was later expanded to the lower (UHF) band to fulfill the demand for additional over-the-air (OTA) television channels. The favorable propagation conditions and practical antenna size in the UHF band make it an ideal choice for broadcast service. As more OTA television channels are introduced, the frequency band allocated for terrestrial television broadcast service in UHF band extends from 470 MHz to 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 between 806 and 890 MHz had already been reallocated from TV broadcast services to analog in the U.S. in 1983).This process had also been com- pleted in most of European countries by 2012, releasing the spectrum from 790-862 MHz. These spectrum bands released in the process of analog-to- digital transition, often called ’(the first) digital dividend’, were subsequently auctioned and allocated to the growing mobile broadband . Therefore, the spectrum left in UHF television band is only between 470-698 MHz in the U.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-over has 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 had mostly been reallocated for broadcasting around the world, leaving the spectrum in UHF band as the primary medium for digital terrestrial TV distribution.


Second Digital First Digital Dividend DVB-T1/T2 Dividend (to be (completed by 2012) confirmed in WRC-15)

490 MHz 698 MHz 790 MHz 862 MHz (a) European Union

Reallocated to First Digital Dividend ATSC Analog Mobile (Completed by 2009) Telephony (1983)

490 MHz 698 MHz 806 MHz 890 MHz (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 in the1950s. More TV programs are provided and the video quality is superior to their analog coun- terparts. In particular, Digital Video Broadcasting-Terrestrial (DVB-T), the European standard for DTT technology, has achieved remarkable success in the last decades. With a share of 40% households receiving television through DTT today and a predication of 50% by 2018, DVB-T has established itself as the 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, and inter- net protocol television (IPTV) via fixed broadband, but also dominance of linear broadcasting in the living room. The viewers’ focus on the televi-

CHAPTERFigure 1. 10. INTRODUCTION Mobile Video Will Generate Over 69 Percent of Mobile Data Traffic by 2018 4

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

Because many video applications can be categorized as cloud applications, mobile cloud traffic follows siona hascurvebecome similar to video. secondary Mobile devices as theyhave memory are increasingly and limitations engaged that might in prevent other them applica- from acting as consumption devices, were it not for cloud applications and services. Cloud applications and services such tionsas onNetflix, their YouTube, mobile Pandora, devices and Spotify [14]. allow The mobile tremendous users to overcome success the memory achieved capacity and recently processing by over-the-toppower limitations of (OTT) mobile devices. services Globally, like cloud Netflix applications and will BBC account iPlayer for 90 percent manifested of total mobile the data growingtraffic by popularity 2018, compared of to video82 percent on-demand at the end of 2013 (VoD) (Figure services.11). Mobile cloud They traffic provide will grow 12 con--fold from sumers2013 withto 2018, a a choice compound of annual high growth quality rate of contents 64 percent. and flexibility unmatched by the DTT service. The DTT industry has attempted to retain its competence in the VoD/nonlinear service by offering Hybrid Broadcast Broadband TV (HbbTV) service. However, as HbbTV requires broadband connections, the added value is rather limited as compared to cable or IPTV alternatives. Moreover, the con- sumption of audio-visual content is rising rapidly on and tablets, which the DTT industry has struggled to reach without much success thus far. Digital Video Broadcasting - Handheld (DVB-H), the digital terrestrial TV broadcasting standard for mobile reception, has failed to reach a wide audience due to the lack of suitable devices and a viable model for mobile TV based on linear content [15].

1.1.2 Mobile Data Traffic Growth and Video Dominance

© 2014 Cisco and/orThe its affiliates commercial. All rights reserved. success This document of is mobile Cisco Public. service in the last decade has made itselfPage 14 of 40 an essential part of every consumer’s daily life. With the deployment of 3G and 4G networks around the globe and the proliferation of smartphones and tablets, mobile data traffic has been growing exponentially in recent years, leading to the projection of even an 1000-fold increase in total wireless data traffic 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 mobile contents and applications, mobile video traffic already constitutes more than half of the global mobile traffic today. It will continue to grow at a cumulative CHAPTER 1. INTRODUCTION 5

annual growth rate of 69% (higher than the global mobile traffic growth rate at 61%) and reach almost 70% of the total share of the mobile traffic in the next five years (see Fig.1.3). As suggested in many recent articles [4, 16, 17], however, such a growth rate can only be sustained if mobile operators are able to provide sufficient network capacity. A possible slow-down of the growth is already evidenced in the recent adjustment to the mobile data traffic forecast. For instance, Cisco has predicted a 30-fold increase from 2010 to 2015 in [18]. In contrast, their recent publication indicates only an 11-fold increase in global mobile data traffic between 2013 and 2018 [4]. In order to provide sufficient capacity, the could not rely exclu- sively on the densification of mobile network infrastructure, as it requires not only a enormous amount of capital expenditure but also an increasing oper- ation expenditure that scales with the infrastructure and user density. The heavy investment in next generation technology - 4G, Long Term Evolution (LTE)- could provide limited relief for the operator when the adoption of 4G devices eventually increases. Nevertheless, the spectral efficient improvement from 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 mobile data traffic, because Wi-Fi networks are rapidly becoming congested with the growing traffic and interferences. Besides, more devices that have only Wi-Fi connection today, such as tablet and laptop, would be equipped with cellular connectivity 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. The high price for spectrum licenses is in itself a clear indication of the competi- tiveness of this strategy. However, in order to obtain additional spectrum, the mobile data service must compete with other services for this limited natural resource. In this thesis, we concentrate the focus of our investigation on UHF broadcast band for the following reasons: • First, the deployment of DVB-T network typically consists of high power mounted on high towers. Thus it requires fixed frequency planning that leaves a considerable amount of spectrum underutilized in large areas to avoid excessive self-interference from other transmitters. • Second, as video service is gaining an increasingly prevailing share in the global mobile data traffic, the consumption of television service is shifting towards cross-platform and on-demand contents. These trends indicate a clear convergence in the future mobile broadband and broad- cast services. • Lastly, the propagation characteristic of UHF band is ideal for mobile coverage while the frequency range still permits antenna design with practical handset size. CHAPTER 1. INTRODUCTION 6

TV signal


Adjacent-Channel Co-Channel Interference Interference WSD WSD


TV Coverage 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, national regulators in Europe and the U.S. have taken strong initiatives in facilitating other wireless services to access this band. Different options have been dis- cussed, ranging from shared access under the regime of Licensed Spectrum Access (LSA) to spectrum re-farming and reallocation.

1.2.1 Secondary Access in TV White Space A typical DTT broadcasting network consists of high-rising TV towers with high power transmitters. In order to minimize inter-symbol interference (ISI) or to provide regional content, the TV programs are delivered over different frequency bands from each TV tower or n a group of TV towers that forms a single frequency network (SFN). This combination of large coverage area and fixed frequency planning offers a unique opportunity for secondary ac- cess in TV bands, as the primary usage can be easily identified and stored in a geo-location database. Spectrum sensing as a detection mechanism could be employed to complement the geolocation database for improved detection accuracy. If the current situation of DTT service maintains status quo, one of the possible option is to exploit the spectrum opportunities in the UHF band by spatially reusing the spectrum on a secondary basis. Based on the information about 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. But if we consider adjacent channel interference leakage, it could also be consid- ered as an underlay model, in which the secondary user accesses the same CHAPTER 1. INTRODUCTION 7

spectrum utilized by the primary system. Fig.1.4 depicts the potential inter- ference link in the secondary access scenario. In both cases, the secondary system must ensure that the harmful interference caused to the TV receivers is within 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 the incumbent TV broadcast service is widely known as secondary access in ’TV White Space’(TVWS). This concept of secondary access or licensed spectrum access (LSA) in TVWS was promoted by regulators in both Europe and the U.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 FCC published a revised ruling in 2010 on secondary access in TV bands with de- tailed technical specifications and regulatory requirements [20]. , the independent regulator in the UK, followed the FCC’s decision and launched the first major pilot to test TV white space in Europe. On the European level, Electronic Communications Committee (ECC) has organized the Spectrum Group 43 (SE43) as a special task group to investigate the differ- ent approaches for secondary access. A general framework was published in the 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 standard for Wireless Regional Area Network (WRAN) using white spaces in the TV bands in 2011. Another prominent standardization effort is the IEEE802,11.af ’White-Fi’ working group. This working group sets out to define a standard for the implementation of Wi-Fi technology in TV bands, allowing a better coverage and access to wider bandwidth than the traditional Wi-Fi in ISM bands [22]. The European Telecommunications Standards Institute (ETSI) has recently published the draft for a harmonized European standard for secondary access in UHF broadcast band [23], which specifies the technical requirements for the devices and procedure for testing requirement compliance.

1.2.2 Second digital dividend Currently, the 470-790MHz UHF broadcasting band is allocated for DTT broadcasting services in Europe, but in many other regions the 700MHz band has already been allocated for mobile services (e.g., North African and Middle Eastern countries; the U.S. also auctioned this band in 2009). In an attempt to 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 its own signal but also assists the primary system that operates on the same frequency by applying signal process method such as dirty paper coding and network coding. However, as we assume the legacy DTT system is incapable of cooperating with the secondary system, the overlay approach is not considered in our study CHAPTER 1. INTRODUCTION 8


Unicast link B Unicast link A Broadcast Links

Cellular BS B

Cell B Unicast Coverage

Cellular BS A

Unicast Unicast Spectrum Broadcast A B Saving Cell A Unicast Coverage

VHF/UHF band Figure 1.5: Cellular content distribution network.

announced the decision to allocate the 700MHz band for mobile services on a co-primary basis with DTT broadcasting, making future authorization for mobile 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 available for mobile service. Recently, European Commission has issued a mandate for the European Conference of Postal and Telecommunications Administra- tions (CEPT) to develop a set of technical conditions required for EU Member States 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 force the 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 and Spain [25]) due to the lack of spectrum for simulcasting, which is crucial for ensuring the service continuity before the channel migration, network and user equipment upgrades (e.g. and receiver for DVB-T2 and wide band aerials) are completed. Furthermore the potential gain of DVB-T2 and SFN would be limited by the cross-border interference coordination and regional content distribution in practical implementations. Consequently, DTT opera- tors might not have sufficient capacity to offer spectrum-hungry services such as ultra-HDTV to satisfy the growing viewing demand [26] 3. CHAPTER 1. INTRODUCTION 9

1.2.3 Mobile Broadband and Broadcast Convergence In light of the converging trends of audio-visual content consumption in both mobile broadband and TV service described in the previous section, a unified platform based on single IP-access network is envisaged for providing both broadband and broadcast service in the UHF television band. The spectrum sharing between the distributions of traditional linear TV content and mobile data is realized on the network/application level instead of the physical level spectrum sharing between distinct radio access technologies. An illustration of this concept is shown in Fig.1.5. The key enabler technology for this concept is the Evolved Multimedia Broadcast and Multicast Service (eMBMS) introduced in 3GPP LTE (Long Term Evolution) radio technology for point-to-multipoint or multipoint-to- multipoint service over a single frequency network (SFN) [28]. Through tight time synchronization, the TV contents can be broadcasted over a SFN with high spectrum efficiency. Furthermore, additional features such as localized contents distribution and on-demand services are made possible by utilizing the point-to-point communication link in , 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 mobile broadband’, will exam the possibilities for further mobile allocation in the 470- 694MHz, creating a converged platform to provide both TV broadcast- ing and mobile service via mobile network. The development of a converged terrestrial broadband network is well in line with the European Commission’s long-term vision for the convergence of broadcasting and broadband services (European Commission 2013). As an all-IP network, it is expected to utilize the spectrum in UHF band more efficiently, creating opportunities for further and economic growth. Although the spectrum re-farming of UHF band would be a long and difficult process due to barriers of structural nature at European Union level and resistances from established interests, it can still be achieved through progressive restructuring of the broadcast landscape and development of new standardization and business models. Recently, European Commission has initiated a feasibility study on the convergence of broadband and 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 Space In 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 between 694-790 MHz from DTT broadcasting to MBB from the year 2017 [27]. CHAPTER 1. INTRODUCTION 10

Adj-Channel Interference

ECC 159 ECC 159: Reference Geometry: Reference Geometry and Deterministic Model Fixed Interference Margin Pessimistic Estimation Pessimistic Estimation Single User Multiple User ECC 159 Coordinated Aggregate Analytical Estimation for Interference Control for Permissible SU Power Cellular-like SU

Not Applicable to Underestimate Inteference Short Range SU

Co-Channel Interference Figure 1.6: Summary of related literature in TV white space.

ing [29]. A few recent EU projects, such as, QUASAR, QoSMOS, FARAMIR and CogEU have extend the research focus from purely discovering the spec- trum opportunities to the exploitation of the secondary access opportunity and have 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 Europe backed 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 on Long Term Evolution (LTE) deployment in the 800 MHz band [33] have been widely used as the reference data for any hypothetical secondary access sce- nario. Other recent measurements provide results for the effect of interference from IEEE 802.22 transmitters [34] or WiMAX transmitters operating on a single 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 exclusion distance for secondary device [21] [20]. These frameworks relies on the use of geo-location database and computational intensive Monte-Carlo simulation to identify 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 only case and, as we have identified, it overestimates the permissible power level by 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] had investigated the TVWS usage case in the U.S. based on FCC’s framework CHAPTER 1. INTRODUCTION 11

with simplified models for propagation and secondary system deployment. In a similar , [31] presents the estimated TVWS availability in European countries. A comparison of the two frameworks proposed by FCC and ECC is described in [36] via a case study of Finland. In most of these works, the availability of TVWS has been quantified and represented in terms of the varying number of available TV channels for a single secondary user at different locations. The scalability of the secondary system, therefore, was not taken into account of the analysis, as the assessment methodologies lack the proper model for the aggregate interference from mul- tiple secondary users to the primary system. Although ECC’s approach can be extended to multiple secondary users case by applying an additional in- terference margin to the secondary transmit power, it does not guarantee any efficient spectrum usage. Besides, the deterministic model for adjacent chan- nel interference (ACI) in the regulatory framework [37] is rather simplistic and would lead to pessimistic conclusions as in [38]. Thus, the previous analysis can 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 oversized secondary cell can not efficiently utilize the highly localized spectrum reuse opportunity. Combining the analysis from both technical and business per- spective, [39] suggests that the commercial sweet point for secondary access in 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 development of device-to-device communication and the proliferation of smart phones and tablet devices, we can foresee the scenario with portable/mobile secondary devices densely deployed in both indoor and outdoor environments. Some of these devices would inevitably be located close to the TV receivers, caus- ing interferences from different adjacent TV channels used by the secondary devices. However, the few studies on aggregate secondary interference in TVWS [41] [42] [43]have exclusively focused on the co-channel interferences originated from typical ’cellular-like’ secondary system deployed outside the TV coverage area. The architecture and capacity of a Wi-Fi like secondary system in TVWS was studied in [44] and [45]. Results of both papers demonstrate the advantage of TVWS in coverage. However, these studies neglect the impact of adjacent channel interference from secondary users to the TV receiver. Therefore, the performance of the secondary Wi-Fi system may have been overestimated.

1.3.2 Cellular TV Distribution The concept of distributing TV contents using a cellular structure was first mentioned in [46] for coverage extension using relays. This idea has been CHAPTER 1. INTRODUCTION 12

gaining increasing popularity since the introduction of MBMS the in 3G net- work. Nevertheless, most of the recent studies have been focused on mobile TV service provision. The system architecture of MBMS in 3G networks is described 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 mobile broadband (MBB) network from 3G to 4G networks. A general roadmap and analytical models for assessing the network performance of mobile video ser- vice in 4G networks are presented in [52]. The performance is evaluated in terms 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 high definition TV programs that requires a data rate over 7 Mbps, whereas the data rate of a typical mobile TV transmission is in the range of hundreds of kbps. Furthermore, the location probability, the measure for DTT coverage quality is commonly defined at a level of 95% at the coverage edge, which is much higher than the 70% location probability requirement of a mobile TV service . This strict coverage requirement for the DTT service is a formidable challenge for any attempt to replace it with mobile networks. Considerable bandwidth would be needed to deliver the same high quality TV programs to the fixed receivers even at the edge of the coverage, even though fixed TV receivers can rely on more advanced antenna configurations with higher performance than mobile receivers. Consequently, the existing results on mobile TV cannot be directly applied to the study on cellular TV distribution system as we have envisaged. 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 that the DTT network would employ a modified LTE standard for broadcasting TV content to both mobile and fixed reception. Recent studies have considered us- ing not only LTE technology but also cellular infrastructure for providing TV services. In [54], the authors have investigated the spectrum requirement for delivering today’s over-the-air TV service over cellular networks. The cal- culations are focused on densely populated urban areas in different cities in the U.S.. Since the typical inter-site-distance (ISD) of cellular networks of the study area is smaller than 2 km, it ensures good performance of the SFNs based on the eMBMS technology. However, as shown in [47], the spectral ef- ficiency of the SFN rapidly degrades as the propagation delay increases with larger ISDs, results in a considerably higher spectrum requirement for broad- casting in rural areas than in urban areas. Therefore, it is not evident that replacing DTT service with mobile networks is feasible based on the results from urban scenarios alone. CHAPTER 1. INTRODUCTION 13

Diversified On- Demand Content

Ubiquitous On- Home Hybrid TV Demand TV

Converged Cellular TV Network

In-Home TV On-the-Move Viewing Viewing

ConsumptionTrends DTT Dominance: Short Range Secondary Access

Mobile TV Status Quo Broadcasting

Limited Main- Stream Content Figure 1.7: The correlation of the investigated scenarios.

One of the key strengths for cellular TV distribution is its capability to dynamically allocate different content into either broadcast or unicast service, and form SFN to improve the efficiency for broadcast distribution. While the resource allocation between broadcast and unicast has been investigated in [48], [49] and [55], these analysis are limited to either an isolated cell or a homogeneous environment. Studies on SFN deployment have considered inhomogeneous geographic settings [56], but the service demand for broad- casting is assumed to be constant. Thus, it is unclear what effect the variations in both service demand and cell density would have on the overall spectrum requirement 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 capacity to satisfy the growing mobile data traffic driven by audio-visual consumptions on mobile platforms. 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 and spectrum allocation in UHF band would largely remain unchanged, whereas a converged content distribution network is a visionary approach fitted for a long-term time frame in which the UHF band would be progressively re- farmed and completely reallocated from DTT to the converged platform. The CHAPTER 1. INTRODUCTION 14

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 face numerous challenges to materialize in large scale implementations. On one hand, 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 prospect of TVWS is further clouded by the regulatory decisions on reallocate the 700 MHz band in the future. On the other hand, while this reallocation creates a unique 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 plans for development and deployment. Therefore, the objective of this disserta- tion is to develop technical frameworks for evaluating the improvements in the utilization of UHF broadcast band, so that we can provide answers to the following research question: • How much addition capacity can be provided for wireless broadband service by allowing it to access the UHF broadcast band? The premises of this study is that the audio-visual service provided by DTT broadcast should be maintained or replaced by another system with equal or even improved quality of service. The technical results provide direct input to the ongoing technical standardization process. The implications of these results for the evolving broadcast-broadband ecosystem could be taken into consideration for future regulatory decisions on the UHF band allocations.

1.4.1 Secondary Access in TV White Space The study on the TVWS case focuses on the scalability issue of secondary system and provides technical evidence to illustrate the real opportunity in TVWS under aggregate interference constraints. The investigation focused on low power WiFi-like secondary system, because its small footprint is consid- ered to be more suitable for exploiting the local spectrum opportunity than a high power cellular-like secondary system. In particular, this thesis aims at answering the following two research questions on TVWS: • Is there a regulation policy that ensures reliable primary protection and allows secondary access in TVWS to be a commercial success? The major concern for secondary access in TVWS is the protection of the primary service against harmful secondary interference. Although the regulation framework for controlling the secondary user transmission power has already been established, it relies on computation-intensive Monte-Carlo simulation. Given the size of typical TV coverage, an ana- lytical approach is needed for quick large scale analysis. Besides, due to CHAPTER 1. INTRODUCTION 15

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 adjacent channel interference when the secondary user is located close to the TV receiver. Thus, one of the challenges is how to model this uncertainty in the 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 within the secondary system or different secondary systems. Therefore, in or- der to determine the scalability of the secondary system, the analysis must take into account the aggregate interference not only from multiple secondary users to primary user but also within the secondary system.

1.4.2 Cellular TV distribution The study on the cellular content distribution network in UHF band exploits the possibility of completely replacing the terrestrial TV broadcast service by reusing 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. In particular, this thesis aims at answering the following two research questions: • Is it feasible to provide terrestrial TV service in UHF broadcast band using existing cellular network infrastructure? The major difference between the study on cellular TV distribution and those on typical mobile TV broadcast services is that here we assume CellTV must replace DTT network and provide service to fixed TV re- ceptions, and thus must satisfy the considerably higher requirement on the quality of service. Besides, a cellular TV distribution network may utilize unicast in areas where broadcast (over SFN) is not efficient for delivery 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 coverage over a large area with inhomogeneous network density and user demand requires a smooth transition between different transmission modes for content delivery, i.e., broadcast over SFN and unicast. The choice of transmission modes should be included in an optimization resource al- location framework to minimize the overall spectrum requirement. In addition, while today’s terrestrial TV broadcast service is still primarily CHAPTER 1. INTRODUCTION 16

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

1.5 Thesis Contributions and Outline

To provide answers to these research questions, we have divided the thesis into the following two main areas: the interference modeling and performance evaluation 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 Space Interference 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 model is further extended in Paper 2 to take into account the accumulative effect of multichannel interferences. The proposed model is again verified through ex- tensive laboratory experiments, whose results show good agreement with the predication of the model. • Paper 1. E. Obregon, L. Shi, J. Ferrer and J. Zander, “Experimental Verification 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 for Aggregate Adjacent Channel Interference in TV White Space,” in IEEE 73rd Vehicular Technology Conference, (VTC 2011-Spring), Budapest, Hungary, May, 2011. Both papers were written in collaboration with Evanny Obregon and Javier Ferrer. The author of this thesis is mainly responsible for the designing of the measurement campaign and result analysis in Paper 1. In Paper 2, the author of this thesis has proposed the analytical model for multichannel interference based on initial observations from test experiments and performed the mea- surement with the other co-authors. Professor Jens Zander provided directions and valuable insights in all papers.

Primary Protection Paper 3 identifies one mathematical flaw of the original analytical approach in ECC report 159 and proposes a new analytical method for estimating the CHAPTER 1. INTRODUCTION 17

permissible transmit power for a single secondary user. This contribution has been accepted by SE43 working group and included in the later release of the ECC regulatory framework. Paper 4 presents a stochastic model based on the 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 following papers: • Paper 3. L. Shi, K.W. Sung and J. Zander,“Permissible Transmit Power for 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 White Space,”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 is developed 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 of this 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 a secondary system under the interference constraint is formulated to maximize the spectrum opportunity for secondary access. Paper 6 builds upon the pri- mary protection framework developed in earlier publications and introduces the CSMA/CA protocol to coordinate the media access among the secondary users.The system throughput of a ’WiFi-like’ secondary system under primary protection constraint is compared to a typical WiFi system operating in the ISM band. • Paper 5. L. Shi, K.W. Sung and J. Zander, “Secondary Spectrum Ac- cess in TV Bands with Combined Co-Channel and Adjacent Channel Interference Constraint”, in IEEE International Symposium on Dynamic Spectrum 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 8th International Conference Cognitive Radio Oriented Wireless Networks and Communications (CROWNCOM 13), Washington, D.C., U.S., July, 2013. CHAPTER 1. INTRODUCTION 18

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

1.5.2 Cellular content distribution system Feasibility of CellTV concept Paper 7 describes the network model for a cellular content distribution system and the envisaged scenario for a converged platform operating in a re-farmed UHF band. The technical feasibility is evaluated through a numerical study on Swedish environments with moderate viewing demand changes in the year 2020. Paper 8 investigates the feasibility of the CellTV concept from the reg- ulatory and business perspective, presenting the major challenges and benefits identified 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 A Case Study”, IEEE Transaction on Broadcasting, March. 2014. • Paper 8. L. Shi, K.W. Sung and J. Zander, “Opportunities and Challenges for TV and Mobile Broadband in 470-790 MHz” 24th European Re- gional International Society Conference , Florence, Italy, Oct. 2013. The original problem formulation of Paper 7 is a result of joint discussion between Dr. Ki Won Sung, Evanny Obregon and Jan Bostrom. Jan Bostrom from PTS has provided valuable input to the scenario modeling. The author of this thesis and Evanny Obregon have performed the numerical analysis. The author of this thesis has acted as leading author of these papers. Professor Jens Zander and Dr. Ki Won Sung have provided valuable insights and helped in editing of all papers.

Minimizing CellTV Spectrum Requirement Paper 9 presents an initial resource allocation framework for the cellular TV distribution network. It optimizes the spectrum resource allocation between unicast and broadcast to provide a seamless coverage from urban to rural area with minimum spectrum requirement. Paper 10 extends the analysis from a Stockholm case study to more general situations and identifies the impact of heterogeneous environments and the evolving viewing habits on the local spectrum requirement for CellTV. • Paper 9. L. Shi, K.W. Sung and J. Zander, “Spectrum Requirement for Cellular TV distribution in UHF Band from Urban to Rural Environ- CHAPTER 1. INTRODUCTION 19

ment” in IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN 2014), McLean, Virginia, USA, Apr. 2014. • Paper 10. L. Shi, K.W. Sung and J. Zander, “Cellular TV Distribution in Heterogeneous Environments” to be submitted. 2014. The author of this thesis proposed the original problem formulation and acted as leading author of these papers. Professor Jens Zander and Dr. Ki Won Sung provided directions and valuable insights in all papers.

1.5.3 Thesis Outline This composite thesis is divided into two parts: an overview of the thesis and verbatim copies of the papers included in this thesis. The first part contains six chapters: Chapter 2 describes the research approach and the main contri- bution of the thesis. Chapter 3 and Chapter 4 provide brief summaries of the key results from the studies on secondary access in TV white space and cellu- lar content distribution system, respectively. Chapter 5 presents a qualitative discussion on the benefits and challenges of the investigated technologies for spectrum access. Finally, Chapter 6 draws the main conclusions and presents the 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 previous chapter through a holistic approach with a focus on technical analysis. The overall problem is divided into two scenarios with complementary time scale and service requirements. The corresponding technical solutions, namely, the secondary access in TV white space with little change in DTT service and spectrum allocation, and cellular TV distribution in a harmonized UHF band, are proposed for technical assessments. The studies are performed within the scope of each scenario aiming at developing evaluation methodologies to quantify the spectrum and capacity gains achievable with these solutions. The analysis emphasizes the interdependency between the technical and regulatory domain, which has direct influence over the investigation scenario and the solution space. The technical solutions proposed in the thesis employ and build on existing regulatory frameworks whenever applicable, so that the methodology developed in the thesis could be easily incorporated into the fu- ture development of the regulatory framework. In the secondary access in the TVWS scenario, our focus is on the development of a scalable framework for controlling the aggregate secondary interference from short-range system op- erating on multiple TV channels; in cellular TV distribution, the focus is on developing 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 technical assessment. Major challenges for the deployment of the technical solutions are examined from the regulatory and business perspectives. The potential benefits of each solution are compared with a qualitative analysis. Fig. 2.1 depicts an overview of our research methodology. In this thesis, we have applied different research approaches to provide a comprehensive analysis on different aspects of our problem. The choice of a specific approach is based on the size of the particular problem and the level


Utilization of UHF Band

Long-Term Near Future Scenarios Future Secondary TV Service Converged Access in Requirement CellTV platorm TVWS User Demand Distribution

Secondary SFN/Unicast Regulation Interference Resource Policy Management Management

Spectrum Spectrum Qualitative Opportunity Saving Comparison Assessment Evaluation

Figure 2.1: Overall research approach of the thesis.

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

2.1 Methodology for Analyzing the Spectrum Opportunity for 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 reception and verified through a measurement campaign [5, 6]. Then we developed an analytical framework for controlling secondary transmission and ensuring pri- mary protection based on the existing regulatory framework [7, 8]. Applying this framework to spectrum opportunity assessment, we have obtained the per- missible secondary transmit power and the maximum density of the secondary CHAPTER 2. RESEARCH METHODOLOGY 22

Secondary Scenario TV Coverage Short Range SU Media Access Control

Cumulative Secondary Interference Single Adj-Ch Spectrum Secondary Multi-ch Access in TVWS Modeling Interference Opportunity System Capacity Interference

Primary Single SU on Multi SUs on Protection Co-Ch Adj-Chs Framework

Regulatory Framework

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

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

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

q =Pr P P min + I , (2.1) 1 { tv Ø tv tv} min where Ptv is the received TV signal power, and Ptv is the sensitivity level of the TV receiver. Itv is the TV-to-TV self-interference. Since both the use- ful TV signal and interferences are subject to fading, the location probability is less than one. In most cases, the requirement is defined by 95% location probability. A typical TV transmitter has a coverage area of an 80 to 100 km radius. The secondary system considered in the thesis is short-range WiFi-like system 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 channel interference is more relevant for short-range systems deployed in high density and probably in close proximity to the TV receivers. CHAPTER 2. RESEARCH METHODOLOGY 23

2.1.2 Interference Modeling Since TV receivers are not designed to tolerate interference, most of the TV receiver filter has poor adjacent channel selectivity. Secondary interference coming from an unoccupied channel adjacent to the receiving TV channel could damage the TV reception. The effect of the adjacent channel interfer- ence (ACI) is represented by the protection ratio, “k, defined as follows:

SN = “k. (2.2) IN+k

where SN is the received TV signal power in channel N, and IN+k is the maximum received secondary interference power on channel N + k beyond which it could cause visible 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:

m S I “ . (2.3) N Ø N+k k kÿ=1 This proposed mathematical model for the cumulative effect of multichannel secondary 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 in the laboratory is primarily for calibration purpose and to verify whether the signal leakage into the cable would cause any harmful impact on the TV re- ception. To identify the maximum tolerable secondary interference, we keep increasing the transmit power of the secondary device until visual distortion on the TV quality is observed. The measurement is repeated for different TV signal power levels and adjacent channels on which the secondary devices are transmitting [5, 6].

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

q =Pr P P min + I + “ (f) GP q q, (2.4) 2 { tv Ø tv tv su su}Ø 1 ≠

where Psu is the secondary transmit power and “su(f) is the same protec- tion as defined in (2.2) with a of f between the interference CHAPTER 2. RESEARCH METHODOLOGY 24

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 an analytical approach to obtain the permissible secondary transmit power di- rectly from (2.4). The key here is to distinguish the case when TV reception is 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 logarithmical domain by approximating the shadow fading components by log-normal ap- proximation [7]. For the multiple secondary user case, we focus on developing a statistical approach to estimate aggregate adjacent channel interference from multiple randomly deployed secondary users to complement the deterministic model proposed by the regulator. The random placement of the secondary user is modeled by the Poisson spatial process. Further, the fact that 95% of the adjacent interference is generated within a small distance of 500 meters from the TV receiver is used as the basis for our assumption that the secondary users contributing to the aggregated interference would have similar transmit power due to geo-location database resolution limitation. Thus we can approximate the aggregate secondary interference by a sum of lognormal random variables using a cumulant matching method. The permissible transmit power for this set of secondary users can thus be obtained in logarithmic form in a similar fashion to the single user case. The accuracy of these assumptions is verified against extensive Monte-Carlo simulations [8].

2.1.4 Maximizing Spectrum Opportunity We develop a complete analytical framework for multi-channel multi-user sec- ondary interference control by incorporating our model of adjacent channel interference with other existing works on co-channel interference. With the combined interference constraint, an optimization problem is formulated to maximize the number of simultaneously active secondary users permitted to access the TV white space. The original formulation has a probabilistic non-convex constraint. To obtain the numerical solution efficiently, we first apply the Fenton-Wickinson method [57] to approximate the aggregate interference with a single log-normal variable and transform the constraint into a deterministic one. Then we isolate the nonlinear term and approximate it with a linear constant. Therefore, the reformulated optimization problem can be easily solved by a linear program- ing algorithm in MATLAB. Again, the approximation shows a good match to the simulation results with less than 1 dB difference in the worst case [9]. CHAPTER 2. RESEARCH METHODOLOGY 25

TX D in CH 2

PU to ce en SU fer er Int TX A in CH 1 Sec ondary Uplin SU k to ce SU ren rfe inte

TX B in CH 1

TX C in CH 1 Contention domain of TX A

Figure 2.3: Interference and contention among secondary users.

2.1.5 Evaluating Secondary System Capacity To evaluate the capacity of the secondary system, we assume the WiFi-like secondary 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 the average throughput estimated as:

T N 1 1 t TP = TPt , (2.5) avg T N n tot t n ÿ=1 ÿ=1 where T is the total simulation time slots, Ntot is the total number of deployed secondary users, Nt is the number of active secondary users transmitting with- t out collision at time slot t. TPn is the throughput of the nth active link ob- tained by the Shannon formula: y t Psu,ign,n TPn = BWnlog2(1 + y ), (2.6) N0 + “puIpu + m Ay Psu,igm,n œ n where gn,n denotes the between the nqth secondary transmitter and its receiver. “puIpu represents the interference from the primary system and N0 y is the noise. As presented in [45], we define An as the set of secondary users operating on channel y outside the contention domain of the nth transmitting secondary 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 CHAPTER 2. RESEARCH METHODOLOGY 26

'Worst-Case' Case Study: Scenarios Stockholm Area

Optimized Network Model Resource Management

Feasibility Quantify CellTV Concept Analysis Spectrum Saving

View Demand Inhomogeneous Model Environment

Regulation Policy

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

power obtained by following the previously defined procedures for primary protection. 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 total number of secondary users. This fact is modeled by a thinning process and is accounted for in the primary protection calculation [10].

2.2 Methodology for Evaluating the Spectrum Saving by Cellular 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 feasibility analysis 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-farmed UHF broadcast band in Sweden. Having observed that certain spectrum sav- ing could be achieved by CellTV but with different transmission modes in each scenario, a resource management framework for CellTV is developed to optimize the network configuration and thus maximize the spectrum sav- ing in an inhomogenous environments. The flexibility for CellTV to adapt to the varying local environment and viewing patterns is demonstrated through a case study of the greater Stockholm region with diverse morphologies. CHAPTER 2. RESEARCH METHODOLOGY 27

Urban Suburban Rural

Cellular sites

Ex SFN1 Unicast Area MBB

MBB Ex MBB Area SFN2 Ex Area UHF UHF BC Band

Figure 2.5: Illustration of the CellTV concept.

2.2.1 CellTV Concept CellTV 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. The TV content can be distributed either via a unicast data link or broadcast over SFNs. The transmission modes depend on the local infrastructure availability and can be dynamically switched according to the viewing demand. Other mobile traffic can also share the same cellular infrastructure and be delivered over any unused frequency/time resource block (see Fig.2.5). One of the key challenges for CellTV is to fulfill the strict requirement for TV service coverage. In particular, the location probability for coverage requirement, as defined in (2.1), must be higher than 95% at the coverage boundary. It is equivalent to 99% service availability within the whole TV coverage 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 not defined 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 could be either an HD program or an SD program. Each has a different minimum data rate requirement. A TV channel could also contain either linear or non- linear content. While a linear channel could be distributed via both broadcast and unicast, a non-linear channel could only be distributed via a unicast link as a video on-demand service. The number of TV channels also varies at different locations. Certain TV channels are available only in local areas or contain regional contents. The CellTV system must distribute all these content to fulfill the viewing demand.

2.2.2 Feasibility Analysis As an initial study, we investigate the feasibility of the CellTV concept by performing numerical analysis for the case of Sweden in the year 2020. The TV viewing habits are assumed to remain largely unchanged, but the number CHAPTER 2. RESEARCH METHODOLOGY 28

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 spectrum would be required, any spectrum remaining is considered as spectrum saving. The spectrum requirement is determined by the estimated spectrum efficiency of 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 Leakages Ratio (ACLR) requirements and protocol overheads. ›eff corresponds to the SINR which is mainly affected by the , 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 date rate of the TV programs with the effective spectrum efficiency [11]. For unicast, the SINR calculation is similar to a typical cellular system with certain network load, which depends on the total bandwidth available for unicast. The bandwidth for unicast also affects the blocking probability. Thus, a multi-Erlang model is developed to classify users with a different bandwidth requirement, and the Kaufman-Roberts recursion [58] is used to derive the total bandwidth that satisfies the blocking probability requirement. Sweden is chosen for our case study because it has excellent coverage in both cellular and DTT networks. Besides, the DTT service penetration in Sweden is moderate, close to the average penetration level in Europe. We have applied the statistical data of population density distribution and cellular base station location in our analysis. The dense urban area in Stockholm center and sparsely populated area in northern Sweden are chosen for the feasibility study as they are considered to be the most spectrum-demanding scenarios due to high viewer density and low cellular infrastructure availability, respectively [59, 60] (see Fig. 2.6).

2.2.3 Resource Management for CellTV The local infrastructure density and population density in each cell would af- fect the spectral requirement for each transmission mode. In an inhomogenous environment, the resources allocations in neighboring areas may be different and could also affect each other’s spectral efficiency. Consequently, the overall spectrum requirement of the whole network is dependent on the local environ- ment at each base station. Thus, we have developed a framework to optimize the resource management for CellTV and minimize its overall spectrum re- quirement. CHAPTER 2. RESEARCH METHODOLOGY 29

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 association between the transmitters and the SFNs is also determined by the optimiza- tion framework. All TV channels must be provided in all the cells, either by unicast 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 area within an 80 km radius from Stockholm city center (as marked in Fig. 2.6) to quantify the spectrum saving achievable by CellTV. The selected area covers distinctive morphologies and has a good coverage of both DTT and mobile services. We divide this circular area into multiple rings. The transition from dense urban to rural area is characterized by the gradual changes in base sta- tion density, and population density in each ring. We further assume different propagation conditions, transmitter and receiver types1 and over-the-air TV service penetration ratios associated with different morphologies.

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

Secondary Access in TV White Space

3.1 Interference Modeling

3.1.1 Adjacent Channel Interference Since TV receivers are not designed to tolerate interference, they are particu- larly vulnerable to adjacent channel interferences from secondary users, which could be deployed in close proximity to the TV receiver. A simulation study in [62] has shown the effect of ACI on the spectrum opportunity of TV white space. The measurement results obtained in [5] (e.g. Fig.3.1) exhibit good match with the simulation results, although the model is slightly pessimistic and should server as a lower-bound estimate. Overall, the effect of adjacent channel interference leakage from an indoor secondary device into the roof-top antenna or cable is almost negligible. On the 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 of available TV channels accessible for secondary devices. The effect of inter- ference on the first two adjacent channel is the most significant. For some TV receivers, the interference received on channels around the N+9 channel could be equally damaging to the TV reception.

3.1.2 Cumulative Effect of Multichannel Interference To model the cumulative effect, we first define an equivalent CCI, Iˆthat would result in the same effect as the aggregate interference experienced on multiple channels. The equivalent CCI is approximated by the weighted sum of all ACIs m “k Iˆ = Ik , (3.1) “0 kÿ=1 30 CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 31

Set−top Antenna (Apartment in the city center) 1 Simulation Results 0.9 Measurement Data








0.1 Estimated Usage Probability for WSD transmission

0 N+1 N+2 N+3 N+4 N+5 N+6 N+7 N+8 N+9 N+10N+11N+12N+13N+14N+15 Adjacent Channel Index

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

where “0 is the minimum SIR ratio for successful TV reception, and “k is the minimum protection ratio of the kth neighboring channel. Ik is the ACI received on the kth 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 interference in each adjacent channel decreases as the number of simultaneously accessed channels increases. The implication of the accumulative effect of multi chan- nel interference depends not only on the number of empty TV channels but also 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 User The 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 that the TV signal outage could also be caused by self-interference and shadow fading on the received signal. Consequently, this approach overestimates the permissible power by almost 10 dB and could lead to significant violation to CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 32

TV signal level =−78dBm, N=Channel 27 (522MHz) −40 Ch. N−1 (Measurements) −42 Ch. N+1 (Measurements) Ch. N+1 (Theoretical) −44 Ch. N−1 (Theoretical)







Maximum tolerable interference power (dBm) −58

−60 1 2 3 5 Number of simultaneously used adjacent channels (1 WSD per adjacent channel ) Figure 3.2: Maximum interference power level that a certain adjacent channel can tolerate vs. 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 by conditional probability. Letting Z denote P (P min + I ), (2.4) can be tv ≠ tv tv shortened as 1 q =Pr P Z . (3.2) 2 su Æ “ (f)G ; su < Note that Z 0 represents the case when the TV signal is in outage due to Æ self-interference or shadow fading in the TV signal, thus, Pr Z 0 =1 q , (3.3) { Æ } ≠ 1 and (3.2) is equivalent to the following: 1 q =PrZ<0 Pr P ú Z Z<0 2 { }· su Æ “ (f)G | ; su < 1 +PrZ 0 Pr P ú Z Z 0 (3.4) { Ø }· su Æ “ (f)G | Ø ; su < 1 =q Pr P ú Z Z 0 . 1 · su Æ “ (f)G | Ø ; su < Then ZÕ = Z can be safely approximated as a lognormal random vari- |z>0 able and be expressed in dB. Finally, the permissible transmit power can be CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 33




54 SE43 Approach Proposed Approach 52 Monte Carlo Simulation Lower Bound [5] 50


46 Permissible SU transmit power (dBm)


42 0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1 q 1

Figure 3.3: Permissible transmit power for WSD located at 50 km away from victim TV receiver with different q (q = q 0.01, ‡ =8dB, ‡ = 10dB. The notations and 1 2 1 ≠ S G references are defined in [7]).

expressed as follows:

P m m “ (f) su(dBm) Æ ZÕ(dBm) ≠ G(dB) ≠ su (dB) 1 q2 2 2 (3.5) Ô2erfc≠ [2(1 )] ‡ + ‡ . ≠ ≠ q1 ZÕ(dB) G(dB) Ò 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 potentially set 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 are in good agreement with the Monte Carlo simulation results. The actual TV violation is always kept below the target level.

3.2.2 Multiple Secondary User Combining the model for the cumulative effect of multichannel interference in (2.3) and the primary protection requirement defined in (2.4), we obtain the following constraint for a multiple secondary user transmitting on unoccupied CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 34

TV channels:

i NÁ i,x i,x min i,x i,y q Pr Ptv Ptv “tvItv Psu ú“su(fx y) Gn , 2 Ø Y ≠ ≠ Ø ≠ Z y X≠ n=1 ] ÿœ i ÿ ^ (3.6) i [ \ where NÁ denotes the total number of active WSDs transmitting on all adja- cent channels inside the dominant interference region centered at pixel i. Xi≠ 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 aggregate interference from the dominant interference region. There are other candidate distributions for the interference approximation, but lognormal distribution exhibits a good match of the upper tail of the actual interference distribution i i NÁ while being analytically tractable. Denote Ga = n=1 Gn. The parameters µ ‡ of its lognormal approximation, Gi and Gi can be obtained by cumulant a a q matching [8]: i 2 Ÿ1(Ga)=exp‚ µ i +‚‡ i /2 , (3.7) Ga Ga Ë È Ÿ Gi ‡2 µ ‡2 . 2( a)= exp( i ) 1 exp 2 Gi + i (3.8) Ga ≠ ‚ ‚ a Ga i Ë È 1 2 The cumulant Ÿm(Ga) is given by ‚ ‚ ‚ R‘ i i m Ÿm(Ga)=2fi⁄ µm(Gf )µm(G◊) gr,su(r)rdr. (3.9) ⁄d0

Here ⁄ is the secondary user density and µm is the mth raw cumulant. R‘ is the radius of the dominant interference region and d0 is the minimal physical distance between the secondary user and the TV receiver. Then we can fol- low the methodology described in the previous section for single secondary user to convert (3.6) into the logarithmic domain and obtain the permissible secondary transmit power for each available TV channel. As shown in Fig 3.4, the results from the proposed approach closely match the Monte Carlo simulation and adapt to the variation of WSD density. In comparison, the deterministic reference geometry approach proposed in the regulatory frame- work [21] is overly pessimistic and underestimates the spectrum opportunity for 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- CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 35

50 Ref Geo @ N+1 Proposed Approach @ N+1 40 Simulation @ N+1 Ref Geo @ N+5 30 Proposed Approach @ N+5 Simulation @ N+5




−10 Permissible SU transmit power (dBm) −20

−30 0 200 400 600 800 1000 1200 1400 1600 SU density per km2

Figure 3.4: Permissible transmit power with different WSD densities ⁄. q1 =0.99, q2ú = 0.95 [8].

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

m ¸ max xi xk i i=1 ¸ ÿ ÿœLi k k Si s.t. Pr N + I qú, tv eff,i Æ “ Ø 2 ; 0 < (3.10) 0 x¸ c , Æ i Æ i ¸ ÿœLi ¸ 0 x cÕ , Æ i Æ i f , a . ’ k œFi ’ i œA k Here, Ieff,i is the aggregate interference from both co-channel and adjacent ¸ channels. xi is the number of WSDs allowed to transmit on channel f¸ in pixel ai. i is the set of unoccupied TV channels in pixel ai, and m is the L k total number of pixels of the studied area. Ntv is the noise floor and Si is the received TV signal power. ci defines the total number of WSDs in pixel ai, while ciÕ set the limit on the maximum number of users that can be admitted on each TV channel. Finally, represent the set of broadcasting TV channels in Fi pixel a and the set of pixels of the studied area. i A CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 36

Empirical CDF 1 q2 with contr. on both ACI and CCI 0.9 q2 with contr. on only ACI 0.8 q2 with contr. on only CCI q1 0.7


0.5 F(x) 0.4




0 0.9 0.92 0.94 0.96 0.98 1 Location Probability

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

By applying Fenton-Wickson lognormal approximation and the relaxation of the constraint as proposed in [42], the original optimization problem can be simplified into a linear optimization problem and solved by standard numerical solvers. The numerical results indicate that a large number of low-power sec- ondary devices can be active simultaneously without violating TV reception and 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 other hand, can only transmit at limited locations and causes mostly co-channel in- terference to the TV receivers. Identifying these limiting factors that dominate the TVWS availability could significantly simplify the regulation process with little risk for compromising the protection of TV reception and efficiency of TVWS utilization [9]. To evaluate the capacity of a low-power secondary system, we include the self-interference and channel contention among secondary users in the as- sessment. We assume the secondary system are multiple pairs of randomly deployed secondary access points and user equipment. Each user will try to access any unoccupied TV channels assisted by a geolocation database, but has to compete for those channels with other secondary users following the CSMA/CA protocol. As a worst-case scenario, we assume each user has full load and infinite buffer size. Instead of simply using the maximum permissible power, an optimal trans- CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 37

200 TVWS in Ptv=−65dbm Rural area TVWS in Ptv=−71dbm Rural area 175 TVWS in Ptv=−65dbm Urban area TVWS in Ptv=−71dbm Urban area ISM band in Rural area ISM band in Urban area 150



75 Maximum average throughput per user (Mbps)

50 4 8 12 16 20 24 28 32 Unoccupied TV channel number

Figure 3.6: Maximum average throughput vs. unoccupied TV channel number and differ- 2 2 ent TV signal Ptv. ⁄ = 100 WSDs/km in rural, 3000 WSDs/km in urban, with 2 channel aggregation [10].

opt mit power Psu is chosen by an exhaustive search in the simulation such that the average throughput is maximized:

opt Psu = arg max TPavg(Psu), (3.11) Psu

where Psuú denotes the WSD maximum permissible transmit power due to either 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 higher throughput given a larger number of unoccupied TV channels. This could be explained by the fact that the secondary interference and network congestion, rather than the primary interference constraint, is the major limiting factor for secondary capacity except in rural areas with low TV signal quality. This effect is illustrated by Fig. 3.7, which shows that the optimal secondary trans- mit power is considerably lower than the permissible level according to the primary protection constraint [10]. CHAPTER 3. SECONDARY ACCESS IN TV WHITE SPACE 38

30 Max permissible in Rural area Max permissible in Urban area 25 Optimal value in Rural area Optimal value in Urban area




5 Secondary user transmit power (dbm)

0 −71 −70 −69 −68 −67 −66 −65 TV signal strength (dbm)

Figure 3.7: Relation between permissible power and optimal power at different TV signal strength. ⁄ = 30 WSDs/km2 in rural, 3000 WSDs/km2 in urban, 2 channel aggregation, unoccupied TV channel number = 8 [10]. Chapter 4

Cellular Content Distribution Network in a Re-farmed UHF Band

4.1 Feasibility Analysis

In the feasibility analysis, we have assumed two fixed TV channel allocation schemes. One is to broadcast all TV channels over SFN regardless of channel popularity. The other is to broadcast only the four most popular TV channels over SFN and unicast the rest of the less popular TV channels. Therefore, the key 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

m Ê(·i)P¯ i=0 qi SINRbroad = , (4.1) m (1 Ê(· ))P¯ ≠ i + N i=1q qi 0 where ·i is the propagation delay,qP¯ is average power associated with base station i, and qi represents the propagation loss to the base station i which accounts for distance-based path loss and shadowing. The total number of cells in the MBSFN area is given by m. For a given ·, the weight function of the constructive portion of a received SFN signal is given by [63] [64]:




300 BW required by current DVB−T deployment: 320 MHz 250

Urban Broadcast 4*1 legacy Rx antenna Urban Broadcast 4*4 diversity 200 Urban Broadcast 8*8 diversity

150 Required Bandwidth (MHz)



0 500 1000 1500 Inter Site Distance (m)

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

P¯ /q0(ri) SINRuni(ri,X)= (4.3) mÕ ¯ l=1 XlP /ql(ri)+N0 where X is the interference collisionq vector conditioned on the network load x and mÕ is the number of interfering base stations (sites allocated with the same spectrum for unicast). The network load x in the system is obtained by solving the fixed point equation

R 2r dr x =min (fl BR +fl BR ) , 1 . HD HD SD SD · R2 [Pr(X )BW ESE(r, X)] C ⁄0 X |x uni D (4.4) q Here, ESE(r, x) is spectral efficiency averaged over shadow fading. fl is the traffic intensity of unicast TV viewers in the cell and BR is the bit rate requirement of the TV channel. Note that the network load x is thus dependent on the total bandwidth available for unicast BWuni and the traffic intensity of unicast TV viewers in a cell watching either HD or SD programs (flHD and flSD). Based on our analysis in a typical Swedish urban and sparsely populated area, we concluded that it is possible to deliver the TV content using cellular infrastructure with UHF television band. In dense urban areas, considerable CHAPTER 4. CELLULAR CONTENT DISTRIBUTION NETWORK IN A RE-FARMED UHF BAND 41


160 Rural Unicast with 4x1 legacy Rx antenna, 0.1% blocking Rural Unicast with 4x1 legacy Rx antenna, 5% blocking 140 Rural Unicast with 4x4 MIMO, 0.1% blocking Rural Unicast with 4x4 MIMO, 5% blocking 120



60 Required Bandwidth (MHz) 40


0 4 6 8 10 12 14 16 Inter Site Distance (Km)

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 TV channels 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 as in the urban area due to the diminishing SFN gain in areas with large inter-site distances. Therefore, only a few most popular TV channels are broadcasted. Unicasting the rest of the channels to the few viewers achieves the best result with around 200-150 MHz spectrum saving (see Fig.4.2). More spectrum saving 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 from broadcast to unicast. Instead, the decision on the resources allocation between unicast and broadcast should be dependent on the local viewing demand and cellular infrastructure density. Furthermore, the resource allocation decision in a certain area will affect its neighbors’ spectrum efficiency due to potential interference and/or SFN gains. Consequently, it will affect the spectrum effi- CHAPTER 4. CELLULAR CONTENT DISTRIBUTION NETWORK IN A RE-FARMED UHF BAND 42

450 Urban Suburban Rural 400



250 Total local spectrum requirement Spectrum occupied by assisting cells 200


100 Niche CH SFN1 Niche CH Unicast Spectrum requirement per cell (MHz) 50 POP CH SFN1 POP CH SFN2 0 0 10 20 30 40 50 60 70 80 Distance from urban center (km)

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

ciency of the whole network. The discrepancy between the resource allocation in different locations will inevitably reduce the overall spectrum efficiency of the CellTV network. Thus, we have developed a framework that optimizes the frequency allocation, the transmission method of each TV channel, and the association of base stations to different SFNs. As a result the overall spectrum requirement of the CellTV system is minimized under the constraint that all TV channels must be delivered either by broadcasting with good coverage or by unicast with low blocking probability [12]. For the quantitative analysis, we have used the population distribution data and 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 high density of base stations. Similar trends have been identified in other Swedish cities as well. By applying this optimization framework to a continuous area, we have found that, while the overall spectrum requirement for CellTV is 185 MHz and there are positive spectrum savings in both urban and rural areas, there is a local spectrum shortage in suburban areas where the transmission modes for most of the TV channels switch from broadcast to unicast. The spectrum shortage 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 the same spectrum band as the one occupied by the SFN to ensure that an accept- CHAPTER 4. CELLULAR CONTENT DISTRIBUTION NETWORK IN A RE-FARMED UHF BAND 43

4 10 BW Req = 200 MHz BW Req = 160 MHz BW Req = 120 MHz BW Req = 80 MHz

2 3 10 BW Req = 40 MHz BW Req = 20 MHz Stockholm Statistics

2 10


Population Density per Km 10

0 10 1000 2000 3000 4000 5000 6000 7000 ISD (m)

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 unicast in rural areas more efficient [12], To gain a deeper into the impact of the heterogeneous environment on the spectrum requirement of CellTV, we extend the case study to a more generalized environment. As illustrated in Fig.4.4, the spectrum requirement is more sensitive to the inter-site distance when the base station density is high. In contrast, in rural areas with large inter-site distance, the spectrum requirement 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 TV channels 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, only the most popular TV channels may still be broadcasted and the rest should be transmitted via unicast links. Thus, the spectrum requirement is mainly dependent on the population density [13]. CHAPTER 4. CELLULAR CONTENT DISTRIBUTION NETWORK IN A RE-FARMED UHF BAND 44

Switching condition between unicast and SFN (MCC = 6dB) 4 10 Unicast all channels Broadcast Popular CHs only Broadcast all CHs Stockholm Statisitcs

2 3 10

2 10


Population Density per Km 10

0 10 1000 2000 3000 4000 5000 6000 7000 ISD (m)

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


New linear TV channels New VoD channels 350 DVB−T2 (3 SFNs 256QAM 5/6 coding)

300 Spectrum Savings

250 Spectrum requirement (MHz) 200

150 60 70 80 90 100 110 120 130 Number of TV channels

Figure 4.6: Spectrum requirement for CellTV in the Greater Stockholm area with different numbers of TV channels [13]. Chapter 5


This chapter presents the comparative analysis of the two alternatives from the perspectives of regulatory/business challenges and the effect of evolving trends in multimedia consumption. We will discuss the implication of those challenges 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 comparison of the potential benefits of each framework on the following aspects: • Capacity The major limitation of TV white space is its unsuitability for wide-range secondary 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 due to aggregated interference constraint and a higher number of occupied TV channels. On the other hand, the spectrum saving by CellTV is the highest in urban areas where the value of the extra capacity is also higher due to a greater demand for mobile broadband. Thus, CellTV has a bet- ter capability to accommodate more data traffic for mobile broadband and also more TV contents with better quality for TV service in UHF television band. • Flexibility The 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 is less 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


improved quality or a larger audience base. Both systems can flexibly adapt 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 broadband services in CellTV is achieved on the network layer, it can better adapt to future changes in demand and regulatory landscapes as compared to the physical layer sharing between secondary system and DTT network in TV white space. • Societal benefits In the TV white space approach, the integrity of the TV broadcasting service is maintained by the implementation of strict primary protection requirements. Although secondary access in TV white space is not suit- able for cellular-type system, it can still improve the wireless broadband coverage in rural and suburban area by exploiting the excellent propaga- tion conditions in the UHF band and providing indoor-to-outdoor radio coverage. In the CellTV system, certain business models and regulation incentives must be developed to ensure that the mobile operators provide -to-air access for the public TV channels. Cellular infrastructure is regarded as less reliable than the DTT network in emergency or disas- trous situations. But this concern is outweighed by the improvement in mobile broadband service that helps to reach the EU Digital Agenda’s target for near-universal broadband coverage. • End-user experience In 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 CellTV system not only enables cross-platform content accessibility but also in- teractivity for non-linear services. Wide-area broadband coverage also ensures 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 the U.S. and the UK. In contrast, the cellular content distribution system is still in the conceptual development stage, although its core technical component, i.e., LTE with eMBMS, has already been deployed by a few mobile operator on a limited scale and can be easily adjusted according to 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 CHAPTER 5. DISCUSSION 47

DTT Penetration (%)

89 88

60 48 38 40 30 21 9 5


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 wireless industry. On the other hand, handsets with LTE eMBMS capability is expected to reach the market by the year 2014. The more challenging issue for CellTV is the development of the MIMO antenna that can be easily deployed in the UHF band. • Business Since secondary access in TV white space is unsuitable for providing cellular-like broadband coverage and the short-range secondary system lacks a decisive advantage over traditional Wi-Fi in the ISM band, the wireless industry has so far exhibited limited enthusiasm in its commer- cialization. For CellTV, mobile operators are facing the same challenge that they experienced in previous attempts to provide mobile TV ser- vice, i.e., to develop a viable for delivering TV services over cellular network. In particular, to provide those ’free-to-air’ public channels, it may require cooperation among multiple operators to utilize the spectrum efficiently and thus complicate the control of their own sub- scribers. Nevertheless, the growing trends of on-demand viewing and the improved capability for end-user device caching and pre-fetching may be the key differentiations from previous business models centered around a linear broadcasting service. Besides the possibility of extra revenue sources, a re-farmed UHF band is essential for providing additional mo- bile broadband capacity and it is a strong incentive for mobile operators to invest in the network infrastructure deployment and user equipment upgrades necessary for the implementation of the CellTV system. • Regulation Despite the strong initiative from regulators in both Europe and the U.S. CHAPTER 5. DISCUSSION 48

to develop secondary systems in TV white space, the decision to reallo- cate more spectrum in the post-digital-switchover UHF band to mobile service has reinforced the reluctancy of the wireless industry to invest in this technology and hindered its commercialization process. The dis- cussion on the possibility of a re-farmed UHF band is still at a very early stage. An official investigation led by the European Commission has just started at the beginning of 2014 and will be finished by the end of 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 the technical issues to a political one. The fact that the dependence on the DTT service varies significantly across European countries would only aggravates the problems in creating a harmonized decision (see Fig.5.1).

5.3 Summary

Overall, the technology and regulation framework for secondary access in TV white 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 wireless industry and has an unpromising prospect for any large-scale commercial de- ployment. In contrast, a converged content distribution platform could bring numerous benefits for both the mobile industry and the end-user as discussed above. Although it is still a considerable challenge to develop a sound busi- ness strategy for providing CellTV service, the real barriers to the harmonized re-farming process may come from the political domain, i.e., the strong oppo- sition from the incumbents and the country differentiation within the European Union [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 spectrum band 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 TV consumption and has enjoyed exclusive access to this spectrum in the past decades. However, the consumption of linear mainstream TV content broad- casted by the DTT service is gradually giving way to on-demand service with diversified 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 TV in the living room. At the same time, the growing video consumption on mo- bile platforms via wireless broadband is creating a significant challenge for mobile operators to provide sufficient capacity with the existing infrastructure and spectrum. In light of these developing trends, a series of recent regulatory decision has opened up the possibilities for wireless broadband service to access the UHF broadcast band. This dissertation has studied two possible approaches for reusing the spectrum by the wireless broadband service with distinctive timeframes and scopes. The short-term option, with the assumption of limited changes in TV consumption patterns in the near future, is spectrum-sharing between legacy TV broadcast networks and secondary systems for wireless broadband services. The long term option instead envisages a converged all-IP platform based on cellular networks operating in a re-farmed UHF broadcast band 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-


tion but also underestimation of the TV white space availability. A statistical model and analytical algorithm is proposed to augment the existing regula- tion framework in estimation accuracy and computation efficiency. Applying these models, we have demonstrated that there is a considerable potential for low-power secondary system. Due to the cumulative effect of multi-channel interference, 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/CA protocol is able to achieve a throughput comparable to that of a traditional Wi-Fi of similar bandwidth in the ISM band. In fact, our analysis shows that the capacity of the secondary system is mostly limited by self-interference and network congestion within the secondary system rather than by primary protection constraint. Being a novel concept, the converged content distribution platform based on cellular network operating in a re-farmed UHF band is only beginning to emerge as the regulation environment and video consumption trends have evolved 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 TV viewers. At the same time, it is able to use the provide extra capacity for other mobile services in the UHF broadcast band. We have addressed the deployment issue of CellTV in an inhomogeneous environment by developing a resource management framework. In a case study of the Swedish environment, we have identified a significant amount of spectrum saving in dense urban areas by employing broadcast over single frequency network consisting of closely deployed base stations. Considerable spectrum saving is also observed in rural and sparsely populated area as the limited number of viewers per cell ensures the efficiency of unicast for the niche TV channels in an on-demand fashion. Finally, we have demonstrated the flexibility advantage of the CellTV system which enables it to adapt to the future changes in TV consumption trends more efficiently. Overall, we conclude that substantial additional capacity could be gained by allowing other wireless broadband systems to access the UHF broadcast band. Nevertheless, both of the investigated approaches for spectrum access are facing great opportunities and challenges. The technical and regulation framework is maturing gradually for secondary access in TV white space using geo-location databases. However, the strong initiative taken by the regulator has thus far failed to stimulate enough interest from the industry, due to the uncertainty in future spectrum allocation in the UHF band and the fact that TV white space is less suited for cellular-type secondary systems. Thus we can foresee a limited commercial success of secondary access in TV white space in the near future, mostly in the form of short-range system offloading and as a 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 changes and can potentially provide access for mobile broadband to a considerable amount of spectrum in the UHF band. These are strong incentives for mobile operators to promote the converged platform as a long-term strategy. But the spectrum re-farming would be a long and difficult process, which is further complicated by the strong opposition from the incumbent and the heteroge- neous situation among the EU member states. More quantitative evidence on the benefits of CellTV is needed to influence future regulatory decisions and persuade the mobile operator to invest in the necessary infrastructure and user device upgrades.

6.2 Future Work

This dissertation has contributed to the increasingly thorough study on the technical aspects of secondary access in TV white space, although there is still a considerable room for improvement in secondary access control and the coexistence of multiple secondary users. As demonstrated by recent research trends, the study on how to realize secondary access is shifting its focus from technical 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-farmed UHF band has only taken its initial step. So far we have only assumed that it would rely entirely on cellular infrastructure. The possibility to combine DTT infrastructure and cellular ones has not been fully exploited yet. The enhance- ment in the computation and storage capability of user devices also deserves further attention, as device caching and pre-fetching add another dimension to the content distribution system. Bibliography

[1] Electronic Communications Committee (ECC) within the European Conference of Postal and Telecommunications Administrations (CEPT), “The European Table of Frequency Allocations and Applications in the 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 Frequency Allocations,” April 2014. [Online]. Available: http://transition.fcc.gov/ oet/spectrum/table/fcctable.pdf [3] L. Backlund, “The importance of UHF Spectrum for DTT 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 Data Traffic 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 Verification of Indoor TV White Space Opportunity Prediction Model,” in 5th Inter- national Conference Cognitive Radio Oriented Wireless Net- works and Communications, June 2010. [6] ——, “A Model for Aggregate Adjacent Channel Interference in TV White Space,” in IEEE 73rd Vehicular Technology Conference, May 2011. [7] L. Shi, K. W. Sung, and J. Zander, “Permissible Transmit Power for Secondary User in TV White Space,,” in 7th International Conference Cognitive Radio Oriented Wireless Networks and Communications, June 2012. [8] ——, “Controlling Aggregate Interference under Adjacent Channel In- terference Constraint in TV White Space,” in 7th International Confer-


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 , October 2012. [10] Y. P. Yang, L. Shi, and J. Zander, “On the Capacity of Wi-Fi System in TV White Space with Aggregate Interference Constraint,” in IEEE International Symposium on Dynamic Spectrum Access Networks , July 2013. [11] L. Shi, E. Obregon, K. W. Sung, J. Zander, and J. Boström, “CellTV - on the Benefit of TV Distribution over Cellular Networks A Case Study,” IEEE Transaction on Broadcasting, vol. 60, no. 1, pp. 73–84, March 2014. [12] L. Shi, K. W. Sung, and J. Zander, “Spectrum Requirement for Cellular TV distribution in UHF Band from Urban to Rural Environment,” in IEEE 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 and Temporal Demand Shaping for Sustainable Wireless Networks,” IEEE Communications Magazine, vol. 52, no. 2, pp. 196–203, February 2014. [18] Cisco, “Cisco Visual Networking Index: Global Mobile Data Traffic 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 on Communicatgoions Workshops, June 2009. BIBLIOGRAPHY 54

[20] Federal Communications Commission, “In the Matter of Unlicensed Operation in the TV Broadcast Bands: Second Memorandum Opinion and Order,” September 2010. [Online]. Available: http://www.fcc.gov [21] Electronic Communications Committee, “Technical and Operational Requirements for the Possible Operation of Cognitive Radio Systems in the White Spaces of the Frequency Band 470-790MHz,” ECC Report 159, 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,” IEEE Transaction on Wireless Communications, vol. 18, no. 4, pp. 22 –31, August 2011. [23] European Telecommunications Standards Institute, “ White Space Devices (WSD); Wireless Access Systems Operating in the 470 MHz to 790 MHz Frequency Band; Harmonized EN Covering the Essential Requirements 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 : the Emergence of a European Dgital Dividend in the United Kingdom and Spain.” in 22nd European Regional Conference of the International Telecommunications Society, 2011. [26] European Broadcasting Union, “Benefits and Limitations of Single Frequency Networks (SFN) for DTT,” June 2013. [Online]. Available: tech.ebu.ch/docs/techreports/tr016.pdf [27] Dagens Nyheter, “ Digital-TV Fär Nu Lämna Frekvenser till Mobilerna,” 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 Algorithms for Cognitive Radio Applications,” IEEE Communications Surveys and Tutorials, vol. 11, no. 1, pp. 116 –130, 2009. [30] R. Murty, R. Chandra, T. Moscibroda, and P. Bahl, “SenseLess: A Database-Driven White Spaces Network,” IEEE Transactions on , vol. 11, no. 2, pp. 189 –203, February 2012. BIBLIOGRAPHY 55

[31] J. van de Beek, J. Riihijarvi, A. Achtzehn, and P. Mahonen, “TV White Space 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 Unlicensed Cognitive Device Operation on Digital Television Performance in the VHF/UHF Band,” in IEEE Symposium on Radio and Wireless, January 2008, pp. 499 –502. [33] Electronic Communications Committee, “Measurements on the Perfor- mance of DVB-T Receivers in the Presence of Interference from the Mobile Service (Especially from LTE),” ECC Report 148, June 2010. [Online]. Available: http://www.ero.dk [34] B. Randhawa, “Conducted Measurements to Quantify DVBT Interference 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 Dynamic Spectrum, 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 in Finland,” in IEEE Symposium on New Frontiers in Dynamic Spectrum, May 2011. [37] European Broadcasting Union, “The Cumulative Effect of Multiple WSD Interference,” September 2011. [Online]. Available: http: //www.cept.org/meeting-documents [38] H. Karimi, “Geolocation Databases for White Space Devices in the UHF TV Bands: Specification of Maximum Permitted Emission Levels,” in New Frontiers in Dynamic Spectrum Access Networks (DySPAN), 2011 IEEE 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 femtocells with cellular network,” in Personal Indoor and 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 Devices under a Probability Constraint on Aggregated Interference,” in IEEE In- ternational Symposium on Dynamic Spectrum Access Networks, October 2012. 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 secondary system design ,” Radio Engineering, vol. 18, no. 7, pp. 1271–1278, De- cember 2010. [43] K. Koufos, K. Ruttik, and R. Januarytti, “Controlling the Interference from Multiple Secondary Systems at the TV Cell Border,” in IEEE 22nd International Symposium on Personal Indoor and Mobile Radio Com- munications (PIMRC), September 2011. [44] S. Kawade and M. Nekovee, “Broadband Wireless Delivery Using an Inside-Out TV White Space Network Architecture,” in IEEE Global Telecommunications 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 Under Realistic Conditions,” in IEEE International Conference on Communi- cations, 2012. [46] S. A. Lery, W. H. Paik, and R. M. Rast, “Extending HDTV Coverage Using Low Power : Cellular Approach,” IEEE Transactions on Broadcasting, vol. 38, no. 3, pp. 145–150, December 1992. [47] F. Hartung, U. Horn, J. Huschke, M. Kampmann, T. Lohmar, and M. Lundevall, “Delivery of Broadcast Services in 3G Networks,” IEEE Transactions on Broadcasting, vol. 53, no. 1, pp. 188–199, 2007. [48] D. Catrein, J. Huschke, and U. Horn, “Analytic Evaluation of a Hybrid Broadcast-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, Proceedings 2006 31st IEEE Conference on, 2006, pp. 850–851. [50] D. Gomez-Barquero, N. Cardona, A. Bria, and J. Zander, “Affordable Mobile 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 Vehicular Technology, vol. 60, no. 2, pp. 644–655, 2011. [53] U. H. Reimers, “ Quo Vadis - Germany as a DTT Case Study,” February 2014. [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,” in IEEE 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,” in IEEE 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 Network Infrastructure 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 MIMO Radio Channel Characteristics on UHF Band,” in IEEE International Conference on Communications, June 2009. [62] E. Obregon. and J. Zander, “Short Range White Space Utilization in Broadcast Systems for Indoor Environments,” in IEEE Symposium on New 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 ,” IEEE Transactions on Broadcasting, vol. 39, no. 4, pp. 395–401, December 1993. [65] L. Shi, K. W. Sung, and J. Zander, “Opportunities and Challenges for TV and Mobile Broadband in 470-790 MHz,” in 24th European Regional International Telecommunication Society Conference , October 2013.