1

An Overview of Dynamic Spectrum Sharing:Ongoing Initiatives, Challenges, and a Roadmap for

Future ResearchSudeep Bhattarai∗, Jung-Min “Jerry” Park∗, Bo Gao†, Kaigui Bian‡, William Lehr§∗Bradley Department of Electrical and Computer Engineering, Virginia Tech

† Institute of Computing Technology, Chinese Academy of Sciences‡ Institute of Network Computing and Information Systems, Peking University

§ Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology

Abstract—We are in the midst of a major paradigm shift in howwe manage radio spectrum. This paradigm shift is necessitatedby the growth of wireless services of all types and the demandpressure imposed on limited spectrum resources under legacymanagement regimes. The shift is feasible because of advances inradio and networking technologies that make it possible to sharespectrum dynamically in all possible dimensions—i.e., acrossfrequencies, time, location, users, uses, and networks. Realizingthe full potential of this shift to Dynamic Spectrum Sharing willrequire the co-evolution of wireless technologies, markets, andregulatory policies; a process which is occurring on a global scale.This paper provides a current overview of major technologicaland regulatory reforms that are leading the way toward a globalparadigm shift to more flexible, dynamic, market-based waysto manage and share radio spectrum resources. We focus oncurrent efforts to implement database-driven approaches formanaging the shared co-existence of users with heterogeneousaccess and interference protection rights, and discuss openresearch challenges.

I. INTRODUCTION

Radio spectrum is not only a key enabler of technologicalinnovations in wireless communications, but it also plays animportant role as an economic growth engine, as highlighted inthe 2012 U.S. President’s Council of Advisors on Science andTechnology (PCAST) report,“Realizing the full potential ofgovernment-held spectrum to spur economic growth” [1]. Theimpact of spectrum on the national economy is expected to in-crease as the proliferation of wireless devices and applicationsof all types and for all uses accelerates. This includes legacyand new users; communication and sensing applications; wide-area and local-area networks; commercial and governmentusers; etc. As the demand for spectrum continues to skyrocket,it will become increasingly difficult, if not impossible, tomeet that demand through the legacy spectrum policy basedon the assignment of siloed, exclusive-use spectrum bandsto particular applications. For instance, according to Cisco,there will be a ten-fold increase in U.S. mobile data trafficbetween 2014 and 2019 [2]. Moreover, the legacy management

This work was partially sponsored by NSF through grants 1314598,1265886, 1431244, and 1547241, and by the industry affiliates of theBroadband Wireless Access & Applications Center and the Wireless@VirginiaTech group.

regime is inadequately flexible, making it difficult to transi-tion spectrum resources to new uses, users, and technologiesas market conditions shift further aggravating the artificialspectrum scarcity associated with outmoded, legacy regulatoryframeworks.

What is needed is a paradigm shift toward a world inwhich spectrum is shared more intensively and flexibly—orequivalently, dynamically—among all classes of users anduses. This includes both Incumbent Users (IUs), or thosewith legacy access rights to spectrum, and Secondary Users(SUs), or those who are seeking access to additional spectrum.Realizing this paradigm shift requires the co-evolution ofradio networks, wireless markets and business models, and theregulatory rules and mechanisms—or, regimes—that governhow spectrum is shared among all classes of IUs and SUs.

Realizing that enabling Dynamic Shared Spectrum is one ofthe key strategies for mitigating the artificial spectrum shortageproblem, academia, the wireless industry, regulators, and otherstakeholders in the U.S. and a number of other countrieshave undertaken initiatives to break-down the legacy silos ofexclusive-spectrum usage models and address the technicaland policy challenges in expanding spectrum sharing options.

In this paper, we provide a comprehensive overview ofthe current status of significant regulatory initiatives under-way globally to facilitate the transition toward a regime ofDynamic Shared Spectrum, with a special focus on database-driven models, which have been shown to offer promise asa cost-effective and reliable approach for managing sharingamong multiple classes of users with heterogeneous accessrights and radio network technologies [3]–[6]. In Section II,we summarize some of the earlier literature on spectrumsharing and clarify some of our terminology. We follow thisin Section III with a summary of the ongoing spectrumreform efforts within the U.S. and several other countries.In Section IV, we address, in general terms, a key concernof spectrum management—the need to manage interferenceamong heterogeneous users and discuss how that challenge ischanging as we move toward more dynamic sharing models. InSection V, we discuss how the challenge is being addressed inseveral important spectrum bands. In Section VI, we consideranother important challenge confronting dynamic spectrummanagement—the need to protect the confidentiality and se-

2

curity of spectrum users in light of management frameworksthat require sharing significant information about the locationand usage of spectrum resources. In Section VII, we identifysome of the open research questions. Section VIII providessummary conclusions.

II. SPECTRUM ACCESS AND MANAGEMENT REGIMES

Spectrum resources are artificially scarce, in part, becausestatic, legacy regulatory regimes inhibit the adoption of tech-nologies and usage practices that would allow spectrum tobe shared more intensively. Policymakers have recognizedfor many years that legacy management regimes need to bereformed to allow greater scope for market-forces to directhow spectrum resources are used and to create incentives andopportunities for the commercialization of innovative and moreefficient radio technologies [7], [8].

The critical need for increasing commercial access to sharedspectrum was emphasized in the National Broadband Plan,[9] which was unveiled by the U.S. Federal CommunicationsCommission (FCC) in 2010. The subsequent U.S. PresidentialMemoranda—Unleashing the Wireless Broadband Revolution[10] and Expanding America’s Leadership in Wireless Innova-tion [11]—issued executive mandates to implement policies toexpand access to shared spectrum. Under the auspices of theWhite House, the FCC and the National Telecommunicationsand Information Administration (NTIA) are taking aggres-sive steps to realize the vision outlined in the PresidentialMemoranda. In the U.S., the recently completed AdvancedWireless Services (AWS)-3 auction [12], progress on enablingshared access to TV white spaces [13], and ongoing progressin the FCC’s 3.5 GHz and 5 GHz proceedings [14], [15] showpromise in advancing this vision of increased spectrum sharingin multiple bands, including between commercial and federalgovernment users.

Regulatory bodies in other countries have also put inmotion spectrum-related initiatives, and, in some cases, haveestablished regulations with the aim of improving spectrumutilization efficiency through shared spectrum access. Theseefforts include the studies and initiatives undertaken by theUnited Kingdom’s Office of Communications (Ofcom) [16]–[18], Industry Canada [19], [20], Infocomm DevelopmentAuthority of Singapore (IDA) [21], [22], China’s IMT-20201

[23], Radio Spectrum Policy Group in Europe [24]–[26], andthe European Communications Office [27].

In multiple ways, all of these initiatives represent progresstoward enabling Dynamic Spectrum Access (DSA). In theengineering and technical standards literature, DSA is oftenused to refer to the “real-time adjustment of spectrum utiliza-tion in response to changing circumstances and objectives”[28], and is usually assumed to be enabled by or coincidentto the use of Cognitive Radio (CR) and/or Software DefinedRadio (SDR) capabilities or technologies. CR has been defined

1IMT-2020 (5G) Promotion Group was jointly established in February 2013by three ministries of China based on the original IMT-Advanced PromotionGroup. The members include the main operators, vendors, universities, andresearch institutes in China. The Promotion Group is the major platformto promote 5G technology research in China and to facilitate internationalcommunication and cooperation.

as “an approach to wireless engineering wherein the radio,radio network, or wireless system is endowed with awareness,reason, and agency to intelligently adapt operational aspectsof the radio, radio network, or wireless system” [29]. SDRsimplement radio functionality in software rather than hard-ware, thereby facilitating the adaptive functionality that char-acterizes CRs. In the full-blown scenario, CR-“smart” radiosystems would collectively sense and analyze their local radioenvironment, negotiate optimal sharing arrangements, and thenadapt their radio operating parameters (i.e., frequency, power,modulation, transmission timing, direction of transmission,and other waveform characteristics) to maximize shared useof local spectrum resources. The concept of shared/dynamicspectrum access represented by this vision is presented inFigure 1(a).

Although significant progress has been made in developingCR, SDR, and other smart radio technologies, we are stillfar from being able to actually realize the full-blown scenariodescribed above. Realizing this vision depends on the com-mercial deployment of new radio technologies, the adoptionof new spectrum access regime, and regulatory reforms. Inthe past five years, a number of surveys have been publishedthat describe the current status of enabling technologies suchas CRs and SDRs [30], [31], -or provide a taxonomy of thevarious ways in which spectrum may be managed so as toenable more extensive spectrum sharing [32], [33]. One suchexample is illustrated in a figure reproduced from the METISproject [34], a recently concluded European project that wasfocused on developing technologies for 5G (see Figure 1(b)).

Figure 1(b) lays out various regulatory rights regimes formanaging spectrum access, ranging from exclusive licensedspectrum (used by cellular operators and television broad-casters) to unlicensed spectrum access (used by Wi-Fi andBluetooth). In the exclusive licensed regime, a single operatormanages how spectrum is shared among the spectrum users;whereas in the unlicensed regime, sharing is uncoordinated.These two models represent points on a continuum of potentialsharing regimes, wherein different tiers of users may havedifferent rights to access the spectrum and to protection frompotential interference caused by other users. One sharingmodel that has the features of both of the aforementionedregimes is Licensed Shared Access (LSA). In LSA, an IUwith previously exclusive-usage rights tolerates shared accessfrom a new SU, who is allowed to share pursuant to aframework that ensures mutual protection from interference.Current implementations of this framework rely on a database-driven mechanism to enforce the sharing arrangement [33].

In these and other regulatory frameworks, DSA is morethan just a technical vision. It also encompasses a frameworkfor managing the spectrum access and usage rights, includingprotection from interference and other management rights(e.g., the right to exclude or pre-empt other users, the rightto sub-lease or transfer management of the spectrum; aswell as obligations to obey operating rules). A full-featuredDSA management regime will have technical, regulatory, andbusiness/market mechanisms in place to enable spectrum re-sources to be dynamically reallocated and shared across users(e.g., government and commercial) and uses (e.g., sensing and

3

(a) Dynamic spectrum sharing via geoloca-tion databases and sensing.

(b) Spectrum access schemes and authorization regimes.

Fig. 1: Dynamic spectrum sharing and spectrum access schemes

communications) on a more fine-grained and granular basisalong any potential technical dimension (i.e., frequency, time,space, direction of transmission, etc.) and under a varietyof differing rights models (e.g., real-time spectrum markets,administered sharing among multiple tiers of PU and SUswith changing usage rights, etc.). In this paper, we use DSAmore loosely to refer to the full range of business, regulatory,and/or technically enabled ways in which enhanced sharingmodels are being enabled. In so doing, we diverge from thetechnical literature that restricts DSA to refer to spectrumusage paradigms that require or make use of CR or SDRtechnologies or capabilities. As argued elsewhere, we needto evolve toward these expanded sharing models, and in sodoing, will enhance the likelihood that CR, SDR, and other ad-vanced radio technologies will be commercialized successfully[35]. Our broader interpretation of what constitutes DSA isintended to highlight how the matrix of regulatory reforms andevolving sharing concepts discussed in subsequent sectionsare contributing to the expansion of options and capabilitiesfor sharing spectrum more flexibly and dynamically. In thenext section, we review some of the regulatory initiativesunderway.

III. RECENT SPECTRUM INITIATIVES

A. Spectrum Initiatives in the U.S.

1) TV band: In September 2010, the FCC issued final rulesto allow low power unlicensed devices to operate on unusedchannels in the TV broadcast bands (often called the TVwhite spaces (TVWS)) in the U.S. [13]. Concerned about thetechnical capabilities of sensing and the risk of interference,the FCC mandated a database-driven approach, attesting tothe challenges of Cognitive Radio (CR) systems, even whenemployed to detect the presence of high-power, high-site (TVtransmitters with large HAAT (height above average terrain)),and fixed TV stations. The TVWS devices must register witha database that dictates how and when they can access thespectrum. To obtain spectrum availability information from thedatabase, a TVWS device needs to submit a spectrum query to

it that contains its operational parameters (e.g., type of device,location, etc.). Two IEEE standards, namely IEEE 802.22 andIEEE 802.11af, were developed to enable communications inthe TVWS.

2) AWS-3 band: In January 2015, the FCC completedan auction of AWS-3 licenses in the 1695 − 1710 MHz,1755− 1780 MHz, and 2155− 2180 MHz bands (collectivelycalled “AWS-3” bands) [12]. The incumbents of this band arefederal systems, including the federal meteorological-satellite(MetSat) systems. Cellular service providers will share thisband with the incumbents based on manual coordination ofprotection zones to protect the federal systems [36].

3) 3.5 GHz band: Per its recent Report and Order andSecond Further Notice of Proposed Rule Making (FNPRM)[14], the FCC has opened up the 3.5 GHz (3550 − −3700MHz) band to SU access. This band will now be home to thenew Citizens Broadband Radio Service (CBRS). The entrantusers will share the spectrum among themselves and incum-bents through a three-tiered access model composed of theIncumbent Access (IA), Priority Access (PA) and General Au-thorized Access (GAA) tiers (see Figure 2). The harmoniouscoexistence among the three tiers of users is ensured throughthe employment of an automated frequency assignment andcontrol database mechanism known as the Spectrum AccessSystem (SAS). The FNPRM [14] also prescribes the use ofa network of spectrum sensors, called Environmental SensingCapability (ESC), to detect the presence of IUs and aid theSAS in assessing the spectrum environment.

IA users include authorized federal and grandfathered fixedsatellite service users currently operating in the 3.5 GHz band.These users will be protected from harmful interference fromPA and GAA users. The PA tier consists of Priority AccessLicensees (PALs) that will be assigned using a competitivebidding process within the 3550 − 3650 MHz portion of theband. Each PAL is defined as a non-renewable authorization touse a 10 MHz channel in a single census tract for up to three-years. At maximum, a total of seven PALs may be assigned inany given census tract with up to four PALs going to any single

4

Fig. 2: Three-tiered access model of the CBRS band.

applicant. Applicants may acquire up to two-consecutive PALterms in any given license area during the first auction. TheGAA tier is licensed-by-rule to enable open, flexible access tothe band for the widest possible group of potential users. GAAusers are permitted to use any portion of the 3550−3700 MHzband not assigned to a higher tier user and may also operateopportunistically on unused PA channels.

The envisioned SAS for the U.S. 3.5 GHz-based CBRS bandsupports coexistence among heterogeneous wireless accesstechnologies, specifically PAL and GAA users, in a centralizedfashion. It maintains a registry of all coexisting devices in theband along with their geo-operational status. SAS acts as ahighly automated spectrum access coordinator that facilitatescoexistence among heterogeneous devices across the band andprevents interference events by enforcing the band’s usage pol-icy. It protects higher tier users from interference from lowertier users, while simultaneously seeking to ensure maximalspectrum availability for lower tier PAL and GAA users andtheir harmonious coexistence [14]. The SAS architecture willbe designed based on the specific use-case scenario of the3.5 GHz band to support a harmonious coexistence amongheterogeneous wireless access technologies of similar/differentspectrum access hierarchy.

4) 5 GHz band: In 2013, the FCC, in its Notice ofProposed Rule Making (NPRM), announced that it intends tomodify rules that govern the operation of Unlicensed NationalInformation Infrastructure (U-NII) devices and make availablean additional 195 MHz of spectrum in the 5 GHz band [37].The NPRM prescribed sharing the Intelligent TransportationSystem (ITS) band (5.85−5.925 GHz) with unlicensed devicesand allow the latter (specifically 802.11ac/802.11ax systems)to have access to more wideband channels (80 MHz and 160MHz). Later, in 2014, the FCC released First Report and Or-der [38] with the goal of increasing the utility of 5 GHz bandby modifying certain U-NII rules and testing procedures thatensure that U-NII devices do not cause harmful interferenceto the IUs (Dedicated Short Range Communications (DSRC)systems) of these bands. Currently, the FCC is in the processof creating a new set of rules for this band which would beultimately referred to as the U-NII-4 band. For more detailsof activities surrounding U-NII-4 band, specifically sharing

between 802.11 and DSRC, readers are referred to [39].In the ongoing proceedings for the 5 GHz band, there is

a growing contention between unlicensed LTE (including Li-cense Assisted Access (LAA) and LTE-Unlicensed (LTE-U))and Wi-Fi stakeholders for access to the band. Several industryentities, mainly Wi-Fi stakeholders, have raised concerns overthe introduction of LTE-U/LAA in the unlicensed bands. Be-cause of the ongoing, sometimes contentious, debate betweenthe LTE and Wi-Fi communities regarding the coexistenceof the two technologies, the FCC, in May 2015, formallysought comments on a range of topics related to the LTE-U/LAA technologies [40]. More recently, in August 2015,the LTE-U and Wi-Fi stakeholders held a meeting to discussthese coexistence issues [41]. We will provide more details onthe coexistence issues between Wi-Fi and unlicensed LTE inSection V.

5) Millimeter wave bands: In August of 2013, the FCCchanged its rules in the 57− 64 GHz band, commonly knownas the 60 GHz band, to improve the use of unlicensed spectrumfor high-capacity, short-range outdoor backhaul, especially forsmall cells [42]. The 60 GHz band is allocated on a co-primarybasis to the federal mobile, fixed, inter-satellite and radio-location services; and to non-federal fixed, mobile and radio-location services. Under Part 18 of the new rules, industrial,scientific and medical (ISM) equipment may also operate inthe 60 GHz band at 61.25 GHz ±250 MHz.

On October 16, 2003, the FCC adopted a Report andOrder [43] establishing service rules to promote non-federaldevelopment and use of the millimeter wave spectrum in the71 − 76 GHz, 81 − 86 GHz and 92 − 95 GHz bands, whichare allocated to non-federal government and federal govern-ment users on a co-primary basis. Specifically, the Reportand Order permits the issuance of an unlimited number ofnon-exclusive, nationwide licenses to non-federal governmententities for all 12.9 GHz of spectrum. It did not requireprior coordination among non-federal government licenseesasserting that interference is unlikely due to the “pencil-beam”nature of the transmissions in this service. However, realizingthat this scheme will delay and perhaps hinder industry effortsto use the 70/80 GHz band as anticipated, the FCC, on March3, 2005, issued Memorandum Opinion and Order [44] andchanged the original decision. The new rules require non-federal government users to finish all interference analysesprior to equipment installation. Furthermore, the FCC recentlyissued an NPRM for seeking comments on authorizing mobileoperations in the 28 GHz, 37 GHz, 39 GHz, and 64−71 GHzbands [45].

As summarized above, the NTIA and the FCC have putinto action several initiatives with the aim to expand wirelessbroadband use, increase sharing, and ultimately meet the 500MHz goal specified in the 2010 Presidential Memorandum[10]. The NTIA’s fifth interim progress report released in 2015states, “Between October 2010 and September 2014, NTIAand the FCC formally recommended or otherwise identifiedfor study for potential reallocation up to 589 MHz.” [46].This total includes 335 MHz in federal or shared bands andbetween 152− 254 MHz in non-federal bands. An additional960 MHz is slated for potential future study for repurposing.

5

Fig. 3: Timeline of spectrum initiatives.

All together, the NTIA and the FCC have made available orare investigating for potential re-purposing between 1, 447 and1, 549 MHz of bandwidth under 6 GHz.

B. Spectrum Initiatives Outside the U.S.

1) European Commission: In November of 2012, theEuropean Commission solicited opinions on spectrum sharingissues concerning LSA which is the key concept that has beenstudied for realizing spectrum sharing in various bands inEurope [47]. LSA ensures guarantees in terms of spectrumaccess and interference protection to the incumbent(s) aswell to the LSA licensees, and hence provides a predictablequality of service to both parties [48]. Spectrum sharing viaLSA can be realized across frequency, time, and geographicaldimensions.

2) United Kingdom: As part of the 2015 World RadioConference (WRC-15) preparatory process, the European Con-ference of Postal and Telecommunications Administrations(CEPT) analyzed possible sharing of Wi-Fi with incumbentusers in 5350 − 5470, 5725 − 5850 and 5850 − 5925 MHzbands in order to develop a European Common Position—i.e., defining a common set of rules for governing accessto the 5 GHz band. In addition, Ofcom recently issued acall for inputs, which sought views on the bands that areto be discussed under WRC-15 agenda item 1.1 and soughtviews on the suitability of these bands for use by mobile orwireless broadband including the 5 GHz bands for Wi-Fi [49].Furthermore, in February of 2015, Ofcom finalized its decisionto allow SUs to access the unused parts of radio spectrum inthe 470− 790 MHz band through dynamic sharing controlledby a spectrum database [50]. Under this plan, the spectrumthat is not utilized by Digital Terrestrial Television (DTT)(including local TV) and PMSE services is shared with TVWSdevices on a license-exempt basis.

3) France: In France, The Agence Nationale desFrequences (ANFR) is considering sharing the 2.3 GHz, 5.8GHz, 17.7−19.7 GHz bands [51]. Incumbents of the 2.3 GHzbands are telemetry and other defense applications, and theexpected secondary users are mobile/cellular service providers.ANFR is considering opening up this band in the regimeof the LSA framework, which affords guaranteed access tospectrum and protection against harmful interference to boththe incumbents as well as the LSA licensees. In the 5.8 GHzband, a geolocation-based approach is being considered tosupport the coexistence between the road tolling applicationand Intelligent Transportation Systems. The 17.7− 19.7 GHzband is being considered for spectrum sharing between fixedservice microwave services (as IUs) and uncoordinated fixedsatellite services (as SUs).

4) Canada: In 2012, Industry Canada (IC)—the govern-ment entity in charge of spectrum management in Canada—released its policy decision to enable access to TVWS [20].In Feb. 2015, IC published a specification describing thetechnical and operational requirements for TVWS devices,which broadly follows the U.S. requirements in terms ofequipment types and technical characteristics [52]. Currently,IC is in the process of defining rules for certifying the databaseand the TVWS devices.

5) Singapore: Singapore’s IDA, in November 2014, ap-proved the rules enabling access to TVWS based on a license-exempt basis provided that devices comply with the technicalrequirements specified by the IDA, contact a licensed databaseto obtain channel availability, and are registered with theIDA following a comprehensive validation process [22]. Thedevice types and requirements are broadly in line with theU.S. model, although Singapore allows for variable effectiveisotropic radiated power (EIRP) levels.

6

6) China: China is actively studying how much spectrumin which bands will be needed to support 5G in its domesticmarket. In 2013, a promotional group called IMT-2020 (5G)was established to define relevant standards and requirementsand to facilitate the development of 5G systems in China. Itsprimary goal is to start the commercialization of 5G networksin China by 2020. So far, IMT-2020 has identified 450− 470MHz, 698 − 806 MHz and 3400 − 3600 MHz bands ascandidate bands for 5G development [23]. Several other bandsin the range 6−100 GHz are also being considered for furtherstudies on channel measurements, modeling and coexistence.

7) New Zealand: Starting in November 2014, Radio Spec-trum Management—the government entity in charge of spec-trum in New Zealand—initiated a temporary arrangement foraccess to TVWS in New Zealand, which allows interestedparties to obtain licenses for operation of TVWS devices atchannels that will be specified in the license [53]. This TVWSaccess plan does not employ the database-driven spectrumsharing approach, but devices are required to be complaintwith relevant regulatory standards which are similar to theones adopted by the FCC.

8) Field Trials of LSA in Europe: In April of 2013, thefirst field trial of shared use of the 2.3 GHz band with alive LTE network was successfully demonstrated in Finland[54] by taking into consideration the inputs from all stake-holders including regulators, incumbents, mobile operators andequipment-supplying industries. Later, in May of 2015, Nokiaperformed another field trial using LTE network and LSAcontroller, and rebutted the potentials of LSA in realizingeffective spectrum sharing [55]. Altogether, it is expected thatthe 2.3 − 2.4 GHz band will be one of the first bands tobe opened up for LSA-based sharing in Europe. However,implementing LSA in this band will entail some challenges asthe band is currently used for various incumbent applications,including government services as well as program makingand special events (PMSE) services, in different Europeancountries.

A timeline of spectrum initiatives, both inside and outsidethe U.S., is shown in Figure 3.

IV. PROTECTING INCUMBENTS

To ensure the viability of spectrum sharing, certain keyrequirements must be met. One of those requirements ismanaging radio interference. For wireless communications towork, receivers need to be able to disentangle their desiredsignals from background noise which may come from ei-ther unintentional (e.g., radiating power lines) or intentionaltransmitters operating either in the same (in-band) or other(out-of-band) frequencies. Historically, the basis for exclusiveassignments was to ensure that the radio users of the spectrumwould be protected from harmful interference from transmis-sions from unaffiliated intentional transmitters. The operatorwith exclusive access could be expected to manage in-bandinterference from its affiliated users, but needed protectionfrom unaffiliated users transmitting in the spectrum as wellas from out-of-band interference from users in other bands.

If the goal is to expand access to spectrum, then animportant challenge is to ensure adequate protection both for

IUs, the legacy rights holders and users of the spectrum, andthe SUs, or new spectrum users who will be accessing thespectrum. Multiple schemes are feasible for managing multipletiers of users. For example, public safety access may havethe right to preempt other traffic, whether the public safetyusers are IUs or SUs in a particular band, and whether theother users are commercial or government users, etc. Moretypically, it is often assumed that the IUs have primary usageand interference protection rights by virtue of their legacyincumbency. They may have to tolerate interference from otherco-primary IUs. The new users, or SUs, have secondary usageand interference protection rights. The SUs may access anduse the spectrum so long as they do not interfere with theIUs, but may be protected from still lower-tier users and mayneed to tolerate interference from same-tier SUs.

In any case, the sharing framework that provides differinglevels of interference protection to different classes of spec-trum users needs to be enforced so that the rules are respectedby all spectrum users. In general, mechanisms for interferenceprotection protect incumbents (or higher-tier users in multi-tiered access models) from interference generated by lower-tier users. In general, mechanisms for incumbent protectioncan be classified into two categories: i) ex ante (a.k.a. pre-ventive) mechanisms, ii) ex post (a.k.a. punitive) mechanisms[56], [57]. Ex ante mechanisms are designed to reduce theprobability of occurrence of harmful interference events, whileex post mechanisms are designed to identify and/or adjudicatemalicious or selfish behavior after harmful interference eventshave occurred. Ex ante and ex post approaches work in tandem(but not in isolation), and thus the choice of an ex anteapproach affects how ex post enforcement is carried out [57].2

In the following sub-sections, we discuss various technicaland regulatory approaches that have been employed to managethe collective interference environment, and will discuss waysin which enforcement mechanisms may change to enable moredynamic sharing models. To simplify the exposition, we willpresume that IUs have a right to interference protection fromSUs, whose right to access the spectrum is contingent on theircompliance to the IU interference protection requirement.

A. Ex Ante (Preventive) Approaches

Ex ante approaches are mechanisms used to prevent interfer-ence events to the IUs. Examples include the use of exclusionzones (EZs), policy based radios [58], secure radio middle-ware [59], tamper-resistance techniques [60], radio integrityassessment techniques [61] and hardware-based compliancemodules [62]. Among these measures, exclusion zones (a.k.a.protection zones) is the primary ex ante spectrum enforcementscheme used by the regulators to protect IUs from SU-generated interference [63]. The legacy notion of an exclusionzone (EZ) is a static spatial separation region defined aroundan IU, where co-channel (and possibly also adjacent-channel)transmissions by SUs are not allowed. An EZ needs to achieve

2The characterization of ex post/ex ante mechanisms as preventive/punitiveis an over-simplification to facilitate discussion. For example, today’s ex postenforcement is tomorrow’s ex ante enforcement; and “carrot” incentives maybe viewed as negative penalties (e.g., credits for a history of non-interferingoperations).

7

two, seemingly opposing, objectives: (i) protect IUs frominterference caused by SUs; and (ii) enable efficient utilizationof spectrum by the SUs. However, the legacy notion of EZs isinept at achieving those objectives, primarily due to the factthat it is overly conservative and static [57], [63], [64]. Thenotion of a static EZ implies that it has to protect incumbentsfrom the union of likely interference scenarios, resulting in aworst-case and very conservative solution.

1) Legacy Notion of Incumbent Protection Zones: Inorder to protect incumbents (i.e., TV broadcast stations) op-erating in the TV bands, the U.S. FCC employs F-curvesthat define a protected service contour around an incumbent.An F-curve, F (x, y), ensures a probabilistic guarantee that,inside the TV coverage region, the received TV signal is abovea given threshold x% of the time in y% of the locations[13], [65]. Then, an appropriate safety margin is added to theprotected service contour, often in the form of a minimumseparation distance from the edge of the protected servicecontour, to derive appropriate EZs for both co-channel andadjacent-channel SU operations.

In the AWS-3 band, the NTIA recently prescribed Co-ordination Zones (CZs) for sharing these bands with wire-less broadband systems (WBSs) [36]. A CZ is not an EZwhere SUs are prohibited, but it is the area inside which aWBS may operate provided that it meets the requirementsfor coordination with federal incumbents [66]. Specifically,NTIA’s rules require each AWS-3 licensee, prior to its firstoperations in its AWS-3 licensed area, to reach a coordinationarrangement with each Federal agency [67]. After a successfulcoordination arrangement, the AWS-3 licensee will be notifiedwith operating conditions that specify the terms in which thelicensee may begin operations, or denial of request. Even ifan AWS-3 licensee is allowed to operate inside a CZ, it musttolerate possible interference from the IUs. For further detailson coordination procedures in the AWS-3 band, readers arereferred to [36].

The legacy protection zones, as they are defined today,often result in an overly conservative approach for incumbentprotection that unnecessarily limits the SUs’ spectrum accessopportunities. A good example of this can be seen in theU.S. TV bands. To account for possible deep fades, theIEEE 802.22 working group specifications require detectorsto have a sensitivity of −116 dBm which corresponds to asafety margin of roughly 20 dB (equivalent to an increasedradius of an EZ up to 110 km) [68], [69]. However, in mostsituations, the probability of such deep fades is very low,and hence the specifications unnecessarily constrain spectrumaccess opportunities for the SUs. Another example can be seenin the proposed use of EZs to protect incumbents in the 3.5GHz band. In [70], the NTIA proposed the use of EZs thatcordoned off vast areas inland of the east and west coast of theU.S. to protect the navy’s ship-borne radar systems. Some haveestimated that nearly 60% of the U.S. population reside insidethese EZs [57]. The deployment of such overly conservativeEZs can significantly reduce the economic benefits of spectrumsharing, and may seriously hinder its adoption due to the lackof interest from potential SU wireless industry stakeholders.

2) New Models for Incumbent Protection: To fully reapthe benefits of spectrum sharing, there is a need to employincumbent protection approaches that strike an optimal balancebetween two key objectives—viz., reliably protect IUs frominterference and maximize spectrum access opportunities forSUs. Achieving such a balance is not feasible with the legacynotion of EZs, and a more agile approach is needed. Onesuch approach is the use of multi-tiered dynamic incumbentprotection zones (MIPZ) proposed by Bhattarai et al. [63].

MIPZ is a novel framework for systematically designing anEZ that can adjust its boundaries by dynamically adapting tochanges in the interference environment. MIPZ is designed fordatabase-driven spectrum sharing (e.g., SAS), and it protectsIUs by providing a probabilistic guarantee of interferenceprotection. Specifically, MIPZ consists of three zones—(i) NoAccess Zone (NAZ) where SU operation is strictly prohibited,(ii) Limited Access Zone (LAZ) where only a “limited”number of SUs are allowed to operate, and (iii) UnlimitedAccess Zone (UAZ) where SUs have unencumbered access tothe spectrum. Figure 4 shows a simplified (considering circularEZs), as well as a practical (considering irregular EZs), modelof the MIPZ framework. For technical details pertaining tothe computation of NAZ-LAZ (inner) and LAZ-UAZ (outer)boundaries, readers are referred to [63].

On one hand, stringent technical requirements and risks ofinterference to the IUs (especially passive IUs) have madespectrum sharing approaches that rely solely on sensing lesspopular recently [71]. On the other hand, experimental resultsfrom recent studies have shown that sharing approaches thatrely solely on databases alone may offer inaccurate and staleinformation of spectrum availability [72]–[74]. State-of-the-art research has shown that incorporating real-time sensingdata with a database-driven approach is more effective inmaximizing spectrum access opportunities for SUs than usingeither a sensing or a database approach in isolation [63],[71], [72], [75], [76]. Hence, researchers proposed a con-cept called Radio Environment Map (REM) [73], [77]–[79],which incorporates a statistical fusion of multiple sourcesof incumbent characteristics and offers improved spectrumaccess opportunities for SUs. A recent study is the work ofBo et al. [6] who proposed a multi-tiered spectrum sharingframework, similar to MIPZ, that incorporates sensing resultsinto a database-driven sharing approach and refines the EZboundary to take advantage of the fallow spectrum that is notcaptured by database-driven sharing alone. Authors of [6] alsostudied strategies to incentivize spectrum sensing that can beincorporated in their proposed framework.

3) Protecting Passive Incumbents: In the preceding para-graphs, we focused on protecting incumbents that activelyemit RF energy. In this section, however, we focus on avery different scenario that involves spectrum sharing betweenpassive sensing systems (as IUs) and active communicationsystems (as SUs). In this scenario, the IUs are receiver-onlypassive systems, such as systems for earth exploration satelliteservice (EESS), radio astronomy (RA), or remote sensing.Such passive systems strive to observe extremely faint signalsthat are emitted by distant non-coordinating transmitters. Thesignal-to-noise ratio (SNR) of the signal that needs to be de-

8

(a) Simplified model (b) Practical model

Fig. 4: MIPZ framework

tected can be as low as -60 dB [80]. This constraint forces thepassive systems to use very sensitive receiver-antenna systemswhich, in turn, make them highly vulnerable to interference,much more so than active communication systems [81].

There are other issues that make spectrum sharing betweenactive and passive systems different from sharing betweenactive systems. For example, geographic separation does notapply to many EESS whose footprint covers a large areaon the Earth’s surface; frequency separation may not beapplicable to RA systems because radio astronomers areoften interested in a large swath of the radio spectrum asdifferent physical processes produce electromagnetic radiationat different frequencies; and time sharing may not be feasiblefor passive systems that require continuous sensing. Therefore,spectrum sharing between passive and active systems requiresan understanding of the operational and functional nature ofeach passive system and identification of the most suitabledimension (in time, frequency, or space), which should be usedto carry out spectrum sharing with active users.

One existing approach for protecting terrestrial passivesystems is to locate them in an area that is far away fromhigh population density areas and is devoid of industrial de-velopment (e.g., RA telescope at Green Bank, WV, USA). Themajority of interference that is detected in such remote sitesis due to extra-terrestrial sources, interference propagationthrough the ionosphere, or tropospheric scatter from remotetransmitters [82]. To mitigate such interference, unilateraltechniques such as excision, cancellation and anti-coincidenceare often employed [83]. However, the efficacy of the legacyunilateral approaches is inherently limited by the lack ofcoordination with the active sources of interference. Moreimportantly, such unilateral approaches are not conducive toefficient utilization of the spectrum.

To enable efficient sharing between incumbent passive sys-tems and secondary active systems, a bilateral approach thatconsiders coordination between the two parties is needed, andthis coordination can be handled by a spectrum managemententity such as a SAS. Cooperative interference mitigationtechniques coordinate the timing and regional use of the radiospectrum in a more dynamic manner. Active systems cancooperate by briefly interrupting or synchronizing radio trans-missions to accommodate the passive measurements wheneverand wherever needed.

Another scenario in which time-sharing can be employedis the coexistence of EESS systems and active secondarysystems. To protect low-orbit EESS systems, SAS can employa no access zone (i.e., a region where no SUs are allowedto transmit) that moves in synchronization with the satellitefootprint’s movement, similar in concept to a dynamic exclu-sion zone. In essence, spectrum access is being coordinatedthrough the SAS, and the IUs and the SUs are time-sharing thespectrum. This notion of time-sharing is possible because mostpassive sensing applications require access to particular seg-ments of the spectrum for relatively short durations only, andthese durations can be pre-scheduled based on the informationabout the satellite’s spectrum usage, orbital path, velocity, andthe location and size of its instantaneous footprint on the earth.

B. Ex Post (Punitive) ApproachesIn general, ex post enforcement approaches involve one or

more of the following four enforcement steps:1) Monitoring/Logging Interference Events: This step

involves collecting information about users’ activities that canlater be used in adjudication procedures, if deemed neces-sary. In essence, logging the details of interference events isequivalent to collecting forensic evidence, which cannot berepudiated by the suspected interferer, during the adjudicationprocess.

2) Identifying Non-Complaint Transmitters: This step in-volves the reliable identification and, if possible, authenticationof the interference source. In the literature, identificationmechanisms at the upper layers of the protocol stack havebeen used to authenticate or uniquely identify transmitters.However, these approaches are of limited value in ex postenforcement, because the enforcement entity (e.g., FCC) maynot know the upper layers of the protocol stack used by theinterference source, and, as a consequence, is unable to decodethe upper layer signaling needed to identify the interferer. Forthis reason, authentication or identification at the PHY-layeris the most appealing approach for ex post enforcement.

PHY-layer authentication is an active area of research,and several schemes have been proposed, which can bebroadly categorized into two classes: intrinsic and extrinsicapproaches. The former utilizes the transmitter-unique “in-trinsic” characteristics of the waveform as unique signaturesto authenticate/identify transmitters. They include RF finger-printing, and electromagnetic signature identification [84]–[87]. Extrinsic schemes enable a transmitter to “extrinsically”embed an authentication signal (e.g., digital signature) inthe message signal and enable a receiver to extract it. Suchschemes include PHY-layer watermarking [88]–[91] and trans-mitter authentication [92]–[98]. On one hand, the intrinsicapproaches require the blind receiver to have only a littleknowledge about the transmission parameters to authenticatethe transmitter, but they are limited by their low robustnessagainst noise and security attacks [99]. On the other hand, theextrinsic approaches can be made highly robust against noiseand security threats, but they require the blind receiver to havecomplete knowledge about the transmission parameters.

In some ex post enforcement scenarios, the enforcemententity may not have knowledge of the PHY-layer parameters

9

used by the rogue or mal-functioning transmitters. In suchsituations, conventional PHY-layer authentication schemes(e.g., [88], [89], [92]–[95]) cannot be used, because in thoseschemes, the verifier needs to know all of the PHY-layerparameters to correctly authenticate the transmitter. Kumar etal. [100] recently addressed this important issue by proposinga scheme that realizes blind transmitter authentication (BTA).The authors coin the term BTA to refer to the problem ofauthenticating a transmitter by extracting its unique, identi-fiable information from the received signal with little or noknowledge of the PHY-layer transmission parameters.

Note that most of the aforementioned schemes can bereadily circumvented if tamper-resistance or integrity protec-tion mechanisms are not employed to thwart hackers fromremoving or incapacitating the authentication mechanism usedby the transmitter. There are a number of prior studies thathave attempted to address this issue [60], [101].

Furthermore, PHY level identification may pose a threat toprivacy or security if access and use of the information isnot appropriately secured and in compliance with higher-layer(network or application layer) policies and protocols. Practicalimplementation of such techniques will need to consider suchcross-layer design issues. (These issues are addressed furtherin Section VI.)

3) Localizing Non-Complaint Transmitters: Once the ma-licious or mal-functioning transmitter has been identified, thenext step is to localize it. In database-driven spectrum sharing,SUs need to register with the database. Hence, it would bestraightforward for the database to know a rogue transmitter’sapproximate location, assuming that the transmitter identifica-tion step has been successfully completed. The challenge isto find out the precise geo-location of the rogue transmitterbecause it is unlikely that the rogue transmitter would provideany cooperation for its location estimation. Thus, the local-ization in cognitive radio networks has to be achieved via anon-interactive technique, e.g., by using localizing techniquessuch as direction of arrival estimation, or by implementingsensors/receivers to measure the received signal strength (RSS)[102]–[105]. The RSS is an indicator of the link distancebetween a transmitter and a receiver. Hence, the informationabout the distances measured between the rogue transmitterand a set of receivers through RSS measurements can bemerged at the regulator to localize the rogue transmitter.

4) Adjudication and Resolution: In the final step of expost enforcement, the non-compliant transmitter is adjudicatedand penalized. Malicious users can be penalized by eitherrestricting their access to the spectrum for a certain durationof time or by imposing economic penalties. The severity of thepunishment should be proportional to the severity of the non-compliant act and the estimated cost of the harm [106], [107].However, the implications of imperfect enforcement must alsobe taken into account when designing an ex post enforcementmethodology. The primary goal of adjudication/penalizationshould be to encourage and, if possible, incentivize SUs toself-regulate and obey access rules, and not to mete out heavy-handed punishments to misbehaving users.

In the U.S., the responsibility of interference resolutionand spectrum enforcement is centralized in the Enforcement

Bureau of the FCC. In carrying out its enforcement activities,the Enforcement Bureau has several punitive measures atits disposal, including imposing monetary forfeitures, issuingcease and desist orders, and revoking operating authority (e.g.,a station license). Traditionally, the Enforcement Bureau hasrelied on a number of enforcement tools, one of which is callsigns and related identifiers. Call signs or call letters havebeen used since the early days of wireless communicationsto uniquely identify transmitting stations. By listening toand deciphering the call letters or related identifiers of aninterfering fixed (but not mobile) transmitter, a regulator canidentify and, if the geographic coordinates associated with thetransmitter are known, locate the transmitter. Another legacyenforcement tool is station licenses. To operate a station, anoperator has to first file an application for a station license tothe FCC. The station license authorizes the licensee to operatethe station for a pre-determined period of time after which thelicense needs to be renewed. The threat of license revocationoften acts as a strong incentive for operators to obey the FCC’srules. Another traditional tool worth mentioning is operatorlicenses. Traditionally, the FCC has required those who operateand maintain transmitter stations to be licensed.

Unfortunately, most, if not all, of the traditional enforcementtools mentioned above have little value in spectrum sharingscenarios [108]. For instance, the utility of traditional call signsfor enforcement activities has dwindled away to almost noth-ing due to the widespread movement from analog transmittersto digital transmitters and from aural to data communications.Station licenses cannot be used as an enforcement tool becausemost SUs (who opportunistically access fallow spectrum)are unlicensed users in the envisioned SAS ecosystem. Fur-ther, operator licensing has been largely phased out in mostservices. This is due to the fact that, in modern systems,procedures for accessing channels are controlled by computerlogic, and minimal or no expertise on the part of the operator isrequired to ensure efficient operation and control interference.

The limitations of legacy approaches for enforcing accessrules in spectrum sharing scenarios necessitate the develop-ment of a new set of enforcement tools. Interference resolutionand enforcement is one of the fundamental requirements thatmust be met to enable government-commercial spectrum shar-ing. The willingness of government agencies to share spectrumon a more expanded and dynamic basis depends on theirconfidence that the applicable regulations and rules regardinginterference will be effective and enforced in an appropriatetime frame [108]. Also, the value of shared governmentspectrum to commercial entities depends on their confidencethat the regulators have applicable rules and tools in place toeffectively control the number of interference incidents andhave the ability to resolve incidents promptly.

To enhance the efficiency of enforcement, it is importantto consider the design of ex ante and ex post frameworksconcurrently to ensure they are mutually re-enforcing so asto maximize the likelihood of compliant behavior and min-imize the total costs imposed of ensuring such compliance.Creative approaches should be considered. For example, itmay be feasible to have negative penalties as part of an expost regime. Under these, radios which establish a record of

10

good performance might be granted additional permissions orexpanded access to spectrum resources as way to incentivizecompliance with access and radio operation rules. Alterna-tively, crowd-sourcing and other techniques for distributing thecosts and responsibilities for enforcement may augment andease the enforcement burden falling on traditional regulatoryinstitutions such as the FCC. How to integrate such noveltechniques into existing rules and institutional managementframeworks poses an open research challenge.

V. COEXISTENCE BETWEEN HETEROGENEOUS WIRELESSTECHNOLOGIES

In the previous section, we focused on the general prob-lem of interference protection as represented by the variousmechanisms used to enforce coexistence between IUs andSUs. In this section, we discuss an equally important problem,viz., techniques for assuring harmonious sharing betweenheterogeneous wireless access technologies. Compared to IU-SU coexistence, SU-SU coexistence has attracted much lessattention from the regulators as well as the research com-munity. This may be partially due to the experience in theISM bands, where diverse technologies, such as Wi-Fi andBluetooth, coexist harmoniously, in most situations, withouta common coexistence mechanism. However, the coexistencesituation in the TV bands and other shared-access bands (e.g.,3.5 and 5 GHz bands) is more complex and challenging dueto a number of reasons, including the use of devices withhigher transmission power, disparity of PHY/MAC strategiesemployed by the secondary systems, and the competitionamong the commercial service providers that aim to augmenttheir existing network capacity with additional capacity tappedfrom those bands.

A. TV Band: Coexistence among SUs

In order to address the coexistence issues among TVWSdevices, industry stakeholders have undertaken active stepstowards standardizing several TVWS technologies, includ-ing ECMA-392 [109], IEEE 802.19.1 [110], 802.22 [111],802.11af [112], and 802.15.4m [113]. In addition, the DySPANStandards Committee formed a new 1900.7 task group (TG) tocreate another standard for TVWS [114]. To address the prob-lem of secondary network coexistence in TVWS, the 802.19.1TG was formed [110], [115]. The purpose of the 802.19.1standard is to enable the family of 802 wireless standards tomost effectively use TVWS by providing standard coexistencemethods among dissimilar or independently operated TVWSnetworks and dissimilar TVWS devices [116]. This standardaddresses coexistence issues among IEEE 802 networks anddevices and will also be useful for non-IEEE 802 networksand TVWS devices. It acts as an interface between coexistingnetworks, providing functionality related to identification of802.19.1-compliant systems (e.g., via registration), obtain-ing/updating coexistence information (e.g., from databases),and using intelligent decision making algorithms to facilitateharmonious coexistence (e.g., decide which actions should betaken by networks to solve coexistence problems).

B. 5 GHz Band: Coexistence between Wi-Fi and LTE-U

Recently, the 5 GHz bands have attracted significant in-terest for launching new wireless applications and servicesbecause of their favorable propagation characteristics and therelative abundance of spectrum therein (580 MHz in the U.S.,455− 605 MHz in Europe, 325 MHz in China [117]). Thereare proposals from multiple industry stakeholders such asQualcomm and others to extend the deployment of LTE-Advanced (LTE-A) to the 5 GHz U-NII bands by exploiting anumber of available technologies, such as carrier aggregation(CA) and supplemental downlink (SDL) [118], [119]. This so-called pre-standard LTE-U approach is applicable to regionsthat do not have a Listen Before Talk (LBT) requirementon accessing the U-NII bands (e.g., US, China, and SouthKorea). For instance, CA/SDL can be used in LTE FDDmode to augment the data-carrying capacity of the downlinkfrom the LTE eNodeB to the LTE user equipment (UE),thus creating a fat downlink pipe. Control and managementfunctions, as well as time-critical communications, are likelyto continue to take place over the licensed (“anchor”) channel.In other regions of the world (e.g., Europe, Japan), regulatorsrequire devices that wish to access the U-NII bands to executethe LBT procedure at the milliseconds scale. Ongoing 3GPPstandardization efforts that aim to make LTE compatible withthe LBT procedure are referred to as LAA. The introduction ofLTE-U (or LAA) in the 5 GHz bands will require current andfuture Wi-Fi technologies operating in these bands to sharespectrum with LTE-U (or LAA) devices.

Coexistence in the 5 GHz bands has been studied in severalprevious works, which focused on different aspects, includingchannel interference, spectrum sharing, and scheduling. Partof the challenge is that technologies like Wi-Fi rely ondistributed access control mechanisms whereas technologieslike LTE take advantage of operator-centralized control, whichproponents of distributed access models argue may give LTE-like technologies an unfair advantage in accessing the sharedspectrum. An investigation of the performance of co-existingLTE and WLAN on the license-exempt band [120] shows thatWLAN technologies hold their transmissions when channelinterference is detected, while LTE reduces its transmissionspeed in order to increase the transmission robustness. Authorsof [121] propose to apply the Request-to-Send/Clear-to-Send(RTS/CTS) protocol to LTE eNodeBs, so that the WLANnetworks do not always yield to LTE transmissions. A largenumber of related works study spectrum sharing. Xing et al.[122] propose an adaptive spectrum sharing technique in whichLTE should adjust its subframe configuration periodicallyto adapt to the traffic load of WLAN system. Authors of[123] propose spectrum sharing using a spectrum consump-tion model. This mechanism considers spectrum managementpolicies, including existing users convey restrictions to newusers, spectrum trading [124], service-level agreement [125],etc. Some works study the spectrum sharing in unlicensedband using game theory [126], [127]. Yun et al. [128] proposea new algorithm to decode two interfering cross technologyOrthogonal Frequency Division Modulation (OFDM) signalswithout alignment in time or frequency. Although their algo-

11

rithm allows LTE and Wi-Fi to transmit simultaneously, theproposed estimation and decoding technology must be appliedto each receiver device.

Enabling a fair coexistence between LTE-U and Wi-Fi ischallenging due to the disparity between their MAC protocolsand the “greedy” nature of LTE-U. LTE-U adopts a schedule-based access approach, whereas Wi-Fi uses contention-basedrandom access. Although LTE-U employs a number of coex-istence techniques (e.g., carrier-sensing adaptive transmission(CSAT)), it still exhibits an inherently more aggressive ap-proach in accessing the spectrum compared to Wi-Fi. Fairnessis critically important in this case as the two coexistingtechnologies have the same spectrum access priority.

C. 5 GHz Band: Coexistence between Wi-Fi and DSRC

Wi-Fi stakeholders in the U.S. have been lobbying thegovernment for access to more spectrum in the 5 GHz bandsin order to deploy their next-generation technologies, such as802.11ac and 802.11ax. In response to the rapidly acceleratingadoption of Wi-Fi, particularly the burgeoning 802.11ac stan-dard, the FCC issued an NPRM in 2013 [15] that recommendsadding 195 MHz of additional spectrum by opening up theITS band (5.850 − 5.925 GHz), where DSRC users are theIUs, to unlicensed devices. Inclusion of the ITS band permitsone additional 80 MHz and one additional 160 MHz channelfor 802.11ac or 802.11ax. However, the realization of thisscenario requires a careful design of the coexistence frame-work. The coexistence of ITS applications with Wi-Fi mayseverely degrade the performance of the former, especiallysafety applications that are very sensitive to communicationlatency [129]. Moreover, when the ITS band was first allocatedin 1999, the FCC’s original intention was for this band tosupport DSRC for ITS exclusively, and hence the ITS protocolstack or the relevant applications designed thereafter were notdesigned to coexist with unlicensed devices.

The coexistence of DSRC and Wi-Fi raises a totally dif-ferent set of challenges than the ones discussed in SectionV-B. In this case, DSRC is the incumbent user, and Wi-Fi isthe secondary user with lower priority. Incumbent protectionis a major challenge because the incumbent users are highlymobile vehicular nodes that utilize spectrum in a dynamicmanner in spatial and temporal dimensions. The difficulty ofthe problem is exacerbated by the fact that 802.11ac/802.11axhas a distinct advantage over DSRC in accessing the spectrumdue to its shorter interframe space (IFS) values, which couldhave a very negative impact on the communication latency ofDSRC safety applications.

D. Coexistence in Spectrum Bands above 6 GHz

Due to the scarcity of spectrum at frequencies below 6 GHz,the use of higher frequencies such as those in the mmWavebands has been proposed for 5G cellular systems. In general,mmWave communication systems cause less interference toneighboring cells operating in the same frequency bandscompared to communication at lower frequencies [130]. Thenoise-limited behavior inherent in mmWave communicationsystems may allow them to share the spectrum without

any prior coordination. However, the requirement of highlydirectional and adaptive transmissions, directional isolationbetween links, and significant possibilities of outage havestrong implications on multiple access, channel structure,synchronization, and receiver design [131]. In this regard,several industry stakeholders and regulatory bodies, includingIMT-2020 in China, OfCom in the UK and FCC in the US,have been actively seeking comments and proposals regardingcoexistence mechanisms for bands above 6 GHz [23], [45].

VI. SECURITY AND PRIVACY ISSUES

In the preceding two sections we discussed the importantchallenge of managing interference. Many of the mechanismsdiscussed relied on the collection, analysis, and sharing ofinformation about the radio environment and spectrum usagein order to ensure efficient collective use of the spectrum.Much of this information has the potential to pose a threat tothe confidentiality and privacy of spectrum users. For example,unauthorized access to the location, technical capabilities, orusage behavior of SUs or IUs could pose a significant threatto users strategic interests and privacy. In this section, we turnto another important challenge: the need to protect the privacyand confidentiality of information collected as a byproduct ofspectrum management.

For example, the use of spectrum databases offers a prag-matic approach for enabling spectrum sharing between gov-ernment (IU) and commercial users (SUs), but at the sametime, it incurs a number of security and privacy concernsby unintentionally facilitating the collection and aggregationof sensitive information by adversial SUs [132]. Examplesinclude threats to the operational security (OPSEC) of theincumbents (e.g., Naval radar systems), privacy of the SUs(e.g., by potentially enabling location-tracking of individuals),and attacks targeting the SAS infrastructure.

A. OPSEC Threats to the Incumbents and Countermeasures

Although using geolocation databases is a practical andcost-effective approach for enabling spectrum sharing, it posesa potentially serious OPSEC problem. SUs (queriers), throughseemingly innocuous queries to the database, can infer op-erational attributes (e.g., location, type, times-of-operation,antenna attributes, receiver sensitivity, etc.) of the incumbents,and thus compromise their OPSEC [132]. Figure 5 illustratesthis threat, which is referred to as an inference attack. Amalicious user can infer the sensitive data of incumbentseven if the database does not directly reveal such information.When incumbents are federal government systems, includingmission-critical military systems and public-safety entities,such as in the U.S. 3.5 GHz band, this breach of incumbents’OPSEC is a critical issue. Although potentially less serious,OPSEC concerns may arise among commercial users who maybe concerned about access to strategically sensitive informa-tion by competitors or threats to the privacy of their customersfrom unauthorized access to database information.

The problem of incumbents’ OPSEC cannot be adequatelyaddressed by tightly controlling access to the database be-cause: i) all SUs need to access the database to realize

12

Fig. 5: Indirect information access via inference channel.

spectrum sharing, and ii) identifying malicious SUs is chal-lenging, especially when they are inside-attackers (i.e., le-gitimate/authorized users performing illegitimate/unauthorizedactions). A more viable approach may be to “obfuscate” theinformation revealed by the database in an intelligent mannersuch that a certain level of OPSEC is assured while support-ing efficient use of the spectrum [56]. Several obfuscationschemes, such as perturbation, transfiguration, k-anonymityand k-clustering, are proposed in [132]. Differential privacy[133] is another emerging privacy-preserving paradigm thatprovides a semantic privacy model with strong protectionguarantees: it captures the amount of disclosure that occurs dueto the publication of sensitive data in addition to mandatinghow the published data should look. A generalization ofdifferential privacy, called “geo-indistinguishability”, has beenstudied in [134], [135] for protecting the location privacy ofusers in the context of location based services.

B. Privacy Threats to the SUs and Countermeasures

Another issue in database-driven spectrum sharing is theprivacy of SUs. Since SUs need to send their operationalattributes (e.g., identity, location, antenna parameters, etc.) tothe database for obtaining the spectrum availability informa-tion in their region, their privacy may be threatened by anuntrustworthy/semi-trustworthy database. An example of sucha database is a compromised SAS that is managed by a com-mercial entity, which provides accurate spectrum availabilityinformation but may misuse the querier’s information for itsown benefit. An untrustworthy or a compromised databasecan directly infer a SU’s information including its spectrumusage habits, device type, times of operation, mobility, etc.[136]. One way to counter this threat is to use a two-wayauthentication protocol between the SAS and the SU—itenables both the SAS and the SU to authenticate each other.Homomorphic encryption-based private information retrieval[137] is another approach for accessing data privately froman untrustworthy/semi-trusted database. Other techniques thathave been studied in the context of LBS to preserve the users’location privacy include sending a space- or time-obfuscatedversion of the users’ actual locations [138], [139], hiding some

of the users’ locations by using mix zones [140], sendingfake queries, indistinguishable from real queries, issued fromfake locations to the database [141], and applying k-anonymity[142] to location privacy.

C. Security Threats to the SAS and Remedies

In addition to aforementioned threats to the security andprivacy of the IUs and the SUs, there are other securityconcerns associated with database-driven spectrum sharing.One such concern is the threats against the infrastructurethat govern spectrum access and sharing (i.e., SAS or SAS-like infrastructure). Similar to the Internet’s DNS (DomainName System), it is expected that several physical SASs willbe networked together to form a region-wide infrastructure.Therefore, some of the security threats against DNS serversmay be applicable to the SAS as well. A malicious user maymodify its own device to masquerade as another certifieddevice. An attacker can spoof a spectrum database and canmodify/jam the query sent by another device or modify/jam thedatabase-response that was intended for some other legitimatedevice. Adversaries can maliciously alter the contents of theSAS (a.k.a. SAS poisoning). Malicious groups may redirect alegitimate SAS’s traffic to another bogus server (a.k.a. SASpharming). Similarly, denail-of-service to the legitimate SUscan be caused by bombarding a large number of bogus queriesto the SAS [56].

Most of the aforementioned security threats may be ad-dressed by implementing a two-way encrypted authenticationbetween the SAS and the SU (querier). Authentication thwartsmasquerade and database spoofing attacks. Cryptography-based integrity protection mechanisms (e.g., message authen-tication codes) prevent adversaries from modifying spectrumqueries/responses without being detected, and the same appliesto the unauthorized modification of the SAS contents. In termsof countering threats to the availability of SAS, maintainingredundancy in the SAS’s contents is one straightforward tech-nique. The efficacy of this technique has been demonstrated inseveral attack incidents of the past, including the distributed-denial-of-service (DDoS) attack that was launched againstDNS root servers in October of 2007 [143].

VII. OPEN PROBLEMS AND RESEARCH CHALLENGES

While radio and networking technical capabilities haveimproved substantially and while promising reform initiativesare underway enabling progress toward realization of the DSAvision articulated earlier, much remains to be done. Newradio capabilities breed new demands for access to spectrum,and the continuing process of technical, market, and policyinnovations is continuously creating new SUs seeking accessto spectrum. There are a number of open challenges in both thetechnological and policy domains. Some of these are discussedin the following subsections and summarized in Table I.

A. Spectrum Efficiency and Access

One of the key challenges in dynamic spectrum sharing isthe effective management of spectrum resources. It requires

13

TABLE I: Open problems and research challenges in dynamic spectrum sharing

Category Open Problems and Research Challenges

1. Spectrum allocation/assignment andspectrum management

• New access paradigms and protocols for efficient spectrum use.• Models that incentivize incumbents to share their licensed spectrum or to relocate to other bands.• Sharing between commercial and non-commercial users (e.g., federal-commercial sharing).• Models and techniques, including dynamic spectrum markets/auctions, for assignment of spectrum.• Frameworks for defining dynamic and flexible incumbent protection zones.• Approaches for incorporating real-time sensing results with database-driven spectrum sharing.

2. Metrics to quantify spectrum usage

• Quantifying spectrum efficiency, value of spectrum and fair access to the spectrum.• Defining techniques and standards for spectrum measurement.• Quantitative definition of harmful interference, and its applications.• Quantifying receiver performance in real-world spectrum sharing environment.• Tools for evaluating the economic and technical trade-offs in spectrum sharing.

3. Interference management andcoexistence

• Techniques for enabling harmonious coexistence between heterogeneous wireless technologies.• Adaptive modulation schemes to enable coexistence and interference mitigation.• Realistic propagation models for frequencies that are being considered for new applications.• Mathematical models of interference and techniques for mitigating interference.• Models for radio system performance with trade-off assessment capabilities.• Interference-tolerant waveforms and protocols.• Techniques and policies for protecting passive IUs.

4. Security and enforcement

• Vulnerability studies of flexible spectrum access systems and development of countermeasures.• Investigating the trade-off between OPSEC and implementation complexity in spectrum sharing.• Hardware and software technologies for enforcing spectrum sharing rules.• Techniques and policies for identifying, adjudicating and punishing non-compliant radio devices.• Defining property rights, and mechanisms to enforce those rights.• Automated enforcement mechanisms and compliance certification methods.

5. Radio hardware and software

• Reconfigurable radio hardware, including antennas, amplifiers, filters, tuners, etc.• Improvements in smart radio architectures that support high dynamic range for wideband operation.• Radio hardware that supports operation in the millimeter wave band.• Designing low-power or energy-harvesting devices for sustainability.• Hardware that provide improved geolocation, direction-finding and interference nulling capabilities.• Clearly defined metrics for quantifying advances in radio hardware and software.

6. Protocols and standards

• Frequency-, space-, and time-cognizant protocols that dynamically leverage multi-functional radiohardware and software.

• Standards that support pre-emptive spectrum access for emergency services.• Protocols and standards for carrier aggregation.• Database-access-protocols for database-driven sharing.• Standards for radio propagation measurement for different bands.

7. Experimentation, testing andstandardization

• Development of advanced and adaptable test beds using advances in hardware, software and policy.• Virtual test beds, including the use of computer simulations, to model and assess coexistence

techniques in large-scale spectrum sharing scenario.• Proofs-of-concept demonstrations covering a variety of bands, applications and geographical areas.• Standardization of current/future test beds.

8. Policy, regulatory and economic issues

• Service level agreements for negotiated sharing.• Understanding the possible impact of new radio technologies on health and environment.• Risk assessment techniques for evaluating when and how to share the spectrum between users.• Strategies for designing dynamic spectrum auctions and markets.• Assessment of economic trade-offs in incentivizing spectrum sharing under multiple scenarios.• Strategies to inventivize spectrum sensing.• Devising economic models and processes that can operate on huge datasets of wireless feedback,

rapidly assess spectrum usage, and adjust spectrum sharing parameters in real-time.

advancements in allocation and assignment mechanisms thatnot only facilitate spectrum sharing, but also support mea-surement and dynamic assessment of the costs and benefits ofsharing. It is a multidisciplinary challenge that requires a jointengagement of technical, economic and policy perspectives[144]. Further research is needed to develop and advance ourability to quantify spectrum efficiency, harmful interference,spectrum value and fair access to the spectrum.

To increase the resolution of spectrum availability andutilization, there is a need to augment the spectrum databasecontent with real-time sensing results [145]. Advanced spec-trum sensing techniques are needed to quickly and accuratelyidentify transmission opportunities over a very wide spectrum

pool that may host a large number of different wireless ser-vices. However, designing a framework that enables the mar-riage of database-driven and sensing-driven spectrum sharingapproaches remains an open problem. Furthermore, sharingbetween users with different access-priorities confront novelchallenges that demand study of new access paradigms andprotocols, dynamic and flexible incumbent protection zones,and adaptive models for spectrum allocation and assignment.We summarize open research challenges related to spectrumefficiency and access in Category 1 and 2 of Table I.

14

B. Coexistence and Interference Management

Facilitating harmonious coexistence among heterogeneouswireless technologies is another challenge in dynamic spec-trum sharing (see Category 3 of Table I). First and foremost,specific metrics need to be established for assessing how welldevices are coexisting together. There is also a need to developmodulation schemes that adapt in concert with other systemcomponents to enable interference mitigation/avoidance. Real-istic propagation models, including inferential models (modelsthat allow the prediction of signal loss at one frequency frommeasurements at other), for frequencies that are being consid-ered for new applications enable regulators and policy makersto foresee the merits of coexistence in both technological andnon-technological aspects.

Future spectrum sharing scenarios require coexistenceamong secondary users of the spectrum (e.g., LTE-U and Wi-Fi in the 5 GHz band). Facilitating SU-SU coexistence requiresthe development of techniques that have not been studiedbefore. For example, even with TV white space systems usingdatabase approaches, the issue of how multiple white spacenetworks might co-exist and coordinate with each other is yetto be explored [110]. The databases inform the radios whichTV channels are available—they say nothing about other datanetworks that might be trying to use the available TV channels.

Moreover, spectrum coordination or cooperation betweendisparate networks, such as a government network as an IUand a commercial mobile network as a SU, as required in U.S.3.5 GHz band, is just in the beginning phase. Sharing betweengovernment and commercial entities will require innovationsin technology as well as in business, administrative, and mar-ket institutions and practices. Special-purpose wireless systemsmay be difficult to accommodate within bold new spectrum-use models because of fundamental limitations on frequencyagility due to basic operational requirements, extreme sensi-tivity to interference, or potentially drastic consequences dueto failure of a radio frequency link. Innovative solutions foraccommodating such systems are needed. These systems mayinclude medical devices, surveillance, remote sensing, andpassive systems such as radio telescopes. Furthermore, coex-istence with legacy systems is an additional challenge becauseof backward and forward inter-operability and compatibility.

C. Hardware, Software, Protocols and Standards

Improving spectral efficiency and radio configurability forhardware and software-defined radios is crucial for enablingthe commercialization of appropriate spectrum sharing cus-tomer and network equipment. This requires advancementsin smart radio architectures that support high dynamic rangefor wideband operation. New technologies and applicationsabove 6 GHz are a promising emerging area. Designing powerefficient radios that support high bandwidth and multi-antennaapplications is another challenging problem that needs to beaddressed.

New advances in the areas of radio hardware, software,signal processing, protocols and access theory need to bedeveloped such that they will work in concert, flexibly andover time to support wireless technologies of diverse needs.

Fundamental limits in these areas also need to be explored.There is also a need to design hardware that provide improvedgeolocation capabilities for indoor applications, direction-finding and interference-nulling capabilities. For sustainabil-ity, research should focus on designing low power energy-harvesting devices.

Besides hardware, there is a need to develop simulationtools and software for evaluating the efficiency and scalabilityof newly proposed architectures. Frequency-, space-, and time-cognizant protocols need to be developed for improving thespectral efficiency. Standards need to be defined for tech-nologies such as carrier aggregation, database-access-protocoland radio propagation measurement for different bands. Theseissues are outlined in Category 5 and 6 of Table I.

D. Security and Enforcement

The successful deployment of new spectrum access tech-nologies, such as cognitive radios, and the realization of theirbenefits will partly depend on the placement of essential secu-rity mechanisms in sufficiently robust form to resist misuse ofthe technologies. The emergence of new spectrum access tech-nologies and spectrum utilization paradigms raise new securitychallenges that have not been studied previously. Vulnerabilitystudies of flexible spectrum access systems and developmentof countermeasures is central to realizing effective spectrumsharing.

There is a need to study obfuscation techniques that meet theOPSEC requirements of incumbent users, especially militaryusers, while enabling efficient spectrum utilization. Studiesmust also focus on thwarting threats against the infrastruc-ture that govern spectrum access and sharing. Furthermore,regulators and policy makers need to understand what datafrom spectrum usage can be collected and analyzed to assessspectrum utilization without infringing on the users’ privacy.Note that more reliable and effective enforcement regimesoften result in greater infringement of the user’s privacy rights.Multidisciplinary research efforts are needed to study thisissue, and to explore the technical and sociological solutionapproaches for balancing the two, seemingly opposing, goalsof enforcement and privacy.

Frequency agile radios combined with compliance andenforcement requirements necessitate research in the followingareas: (i) automating the detection and identification of inter-ference sources; (ii) creating mechanisms for rapidly enforcingpolicy changes on radio devices; (iii) evaluating the sociologyof privacy, enforcement mechanisms, and potential penalties;and (iv) evaluating the economic trade-offs in ex-ante and ex-post mechanisms. As wireless systems become more advanced,certifying new wireless devices (especially those which arebased on software defined radio) for compliance becomes non-trivial because of the radio’s reconfigurable and frequency-agile nature [146]–[148]. Beyond certification, monitoring willbe needed to ensure that deployed systems are in compliance,and when they are not, enforcement procedures will be neededto remedy the problems. Some of the open problems and re-search challenges related to spectrum enforcement, complianceand security are enumerated in Category 4 of Table I.

15

E. Experimentation, Testing and StandardizationTo ensure that the new technologies will not cause harm

to legacy systems, are robust and secure, and are efficientusers of the spectrum, testing of new technology throughlarge-scale experimentation is essential. Such testing will alsoprove helpful in raising stakeholder trust that the new sharingapproaches will work as promised. This requires new tests,measurement solutions, standards and regulatory validation.Measurements and metrics to establish existing and futurelevels of spectrum occupancy and efficiency will also berequired. Development of advanced and adaptable test bedsusing advances in hardware, software and policy; proof-of-concept demonstrations; and standardization of current/futuretest beds are imperative to assess the performance of new tech-nologies. Open problems related to experimentation, testingand standardization are listed in Category 7 of Table I.

F. Regulatory and Policy ChallengesBeyond technical issues, there are also policy-domain chal-

lenges in dynamic spectrum sharing. Future sharing systemsmay employ dynamic spectrum markets in which primarylicensees can sell spectrum access to SUs on a temporarybasis. There exists a need for interdisciplinary research inthe areas of market- and non-market-based mechanisms forspectrum access and usage to efficiently organize the sharingof scarce spectrum resources. For spectrum markets to workefficiently, there has to be sufficiently liquid supply anddemand of spectrum resources to make it worthwhile incurringthe transaction costs associated with administering and makinguse of the markets. Thus, it is also important to consider howto enhance the value of spectrum to SUs. For instance, sharingmight only be of a little value if an IU needs arbitrary accessto the spectrum because this prevents SUs from predicting theavailability of spectrum.

Market barriers (e.g., transaction costs, lack of incentives,strategic concerns) and limited availability of information (e.g.,unavailability of federal-IUs’ operational parameters) all posepolicy challenges for realizing the potential of spectrum shar-ing. Other challenges include authorization constraints (e.g.,regulatory and policy requirements); and lack of incentives forincumbents to share (e.g., incumbents may see future spectrumsharing only as a threat to their own use) [51]. Also, thereare issues related to health and environmental ramifications ofemerging technologies. Interdisciplinary collaboration amongresearchers of diverse backgrounds is crucial in addressingthese challenges.

Finally, the design of radio systems is inherently cross-disciplinary since the technologies are only a part of thelarger system which includes the stakeholders, markets, andinstitutional frameworks within which those technical systemsoperate and evolve. The technologies, business models, mar-kets, and regulatory frameworks need to co-evolve. Category 8of Table I outlines some of the policy, regulatory and economicchallenges in dynamic spectrum sharing.

VIII. CONCLUDING REMARKS

This paper has provided a comprehensive review of im-portant trends, regulatory reform initiatives, and research

challenges that are part of the ongoing systematic efforts tobring about fundamental changes to how we manage andutilize radio spectrum. Most of the legacy spectrum regimesemployed throughout the world are overly static and inflexible,making it difficult, if not impossible, to utilize spectrum to itsfull potential in an efficient manner. Dedicated exclusive-useassignments, premised on archaic notions of radio technology,reduced incentives to use spectrum efficiently and contributedto problems of artificial scarcity, thereby impeding growth andinnovation in wireless services and technologies. The futureneeds to embrace more dynamic models of spectrum sharing,or Dynamic Spectrum Access (DSA), in which spectrummay be shared along multiple technical dimensions (e.g.,frequency, time, space, and direction) and across multipleusage contexts (e.g., commercial/government, legacy/new, li-censed/unlicensed, or multiple classes of spectrum rights hold-ers). Realizing this DSA vision requires the co-evolution oftechnical, regulatory, and business models for managing howspectrum usage rights are administered and enforced. Ren-dering traditional mechanisms such as static exclusion zonesmore dynamic [63], incorporating database technologies, andaugmenting those with sensing capabilities represent distinctbut complementary steps on the path to dynamic spectrumsharing.

REFERENCES

[1] PCAST, “Report to the President Realizing the Full Potential ofGovernment-Held Spectrum to Spur Economic Growth,” Jul. 2012.

[2] Cisco, “Cisco Visual Networking Index: Global Mobile Data TrafficForecast Update, 2014-2019,” Tech. Rep., Feb. 2015.

[3] R. Murty, R. Chandra, T. Moscibroda, and P. Bahl, “Senseless: Adatabase-driven White Spaces Network,” IEEE Transactions on MobileComputing, vol. 11, no. 2, pp. 189–203, 2012.

[4] D. Gurney, G. Buchwald, L. Ecklund, S. Kuffner, and J. Grosspietsch,“Geo-Location Database Techniques for Incumbent Protection in theTV White Space,” in IEEE Symposium on Dynamic Spectrum AccessNetworks (DySPAN), 2008., Oct. 2008, pp. 1–9.

[5] B. Gao, J.-M. Park, and Y. Yang, “Supporting Mobile Users inDatabase-driven Opportunistic Spectrum Access,” in Proceedings ofthe 15th ACM International Symposium on Mobile Ad Hoc Networkingand Computing, ser. MobiHoc ’14. ACM, 2014, pp. 215–224.

[6] B. Gao, S. Bhattarai, J.-M. Park, Y. Yang, M. Liu, K. Zheng, andY. Dou, “Incentivizing Spectrum Sensing in Dynamic Spectrum Shar-ing,” in Proceedings of the 2016 IEEE International Conference onComputer Communications (INFOCOM), Apr. 2016.

[7] FCC, “Report of the Spectrum Efficiency Working Group, ET Docket02-135,” Nov. 2002.

[8] OfCom, “Spectrum Framework Review: a Consultation on Ofcom’sviews as to how radio spectrum should be managed,” Nov. 2004.

[9] FCC, “Connecting America: The National Broadband Plan,” Mar. 2010.[10] The White House, “Unleashing the Wireless Broadband Revolution,”

Jun. 2010.[11] The White House, “Expanding America’s Leadership in Wireless

Innovation,” Jun. 2013.[12] FCC, “Auction 97: Advanced Wireless Services (AWS-3),”

Accessed: 2015-11-09. [Online]. Available: http://wireless.fcc.gov/auctions/default.htm?job=releases year&y=2015&m=6

[13] FCC, “Second Memorandum Opinion and Order (FCC 08-260),” Sep.2010.

[14] FCC, “Report and Order and Second Further Notice of ProposedRulemaking, GN Docket No. 12-354,” Apr. 2015.

[15] FCC, “Revision of Part 15 of the Commissions Rules to PermitUnlicensed National Information Infrastructure (U-NII) Devices in the5 GHz Band (NPRM 13-22),” Feb. 2013.

[16] OfCom, “Geolocation for Cognitive Access: A Discussion on usingGeolocation to enable License-exempt Access to the Interleaved Spec-trum,” Jul. 2009.

16

[17] OfCom, “Implementing Geolocation: Summary of Consultation Re-sponses and Next Steps,” Sep. 2011.

[18] OfCom, “Spectrum Management Strategy: Ofcoms Strategic Directionand Priorities for Managing Spectrum over the Next 10 Years,” Apr.2014.

[19] Industry Canada, “Consultation on a Policy and Technical Frameworkfor the Use of Non-broadcasting Applications in the Television Broad-casting Bands below 698 MHz (SMSE-012-11),” Aug. 2011.

[20] Industry Canada, “Framework for the Use of Certain Non-broadcastingApplications in the Television Broadcasting Bands Below 698 MHz(SMSE-012-12),” Oct. 2012.

[21] Infocomm Development Authority of Singapore, “Trial of White SpaceTechnology Accessing VHF and UHF Bands in Singapore,” Jul. 2010.

[22] Infocomm Development Authority of Singapore, “Regulatory Frame-work for TV White Space Operations in he VHF/UHF Bands,” Jun.2014.

[23] T. Wang, G. Li, J. Ding, Q. Miao, J. Li, and Y. Wang, “5G Spectrum:is China Ready?” IEEE Communications Magazine, vol. 53, no. 7, pp.58–65, Jul. 2015.

[24] European Paliament and Council, “Decision No 243/2012/EU of theEuropean Parliament and of the Council of 14 March 2012 Establishinga Multiannual Radio Spectrum Policy Programme,” Mar. 2012.

[25] Radio Spectrum Policy Group, “RSPG Opinion on Licensed SharedAccess (RSPG13-538),” Nov. 2013.

[26] European Commission, “2013/195/EU: Commission Implementing De-cision of 23 April 2013 Defining the Practical Arrangements,UniformFormats and a Methodology in Relation to the Radio Spectrum In-ventory Established by Decision No 243/2012/EU of the EuropeanParliament and of the Council Establishing a Multiannual RadioSpectrum Policy Programme,” Apr. 2013.

[27] ECC, “Technical and Operational Requirements for the Possible Opera-tion of Cognitive Radio Systems in the ’White Spaces’ of the FrequencyBand 470-790 MHz ,” Jan.

[28] P. A. Tenhula, “Regulatory Framework(s) for Facilitating New Spec-trum Sharing Schemes,” Jul. 2012.

[29] International Telecommunications Union (ITU) , “Definitions of Soft-ware Defined Radio (SDR) and Cognitive Radio Systems (CRS),” Sep.2009.

[30] B. Wang and K. Liu, “Advances in cognitive radio networks: A survey,”IEEE Journal on Selected Topics in Signal Processing, vol. 5, pp. 5–23,Feb. 2011.

[31] E. Hossain, D. Niyato, and D. I. Kim, “Evolution and future trends ofresearch in cognitive radio: a contemporary survey,” Wireless Commu-nications and Mobile Computing, vol. 15, pp. 1530–1564, Aug. 2015.

[32] OECD., “New Approaches to Spectrum Management,” OECD DigitalEconomy Papers, May 2014.

[33] M. Matinmikko and M. Mustonen and D. Roberson and J. Paavola andM. Hoyhtya and S. Yrjola and J. Roning, “Overview and comparisonof recent spectrum sharing approaches in regulation and research: Fromopportunistic unlicensed access towards licensed shared access,” in Dy-namic Spectrum Access Networks (DYSPAN), 2014 IEEE InternationalSymposium on, Apr. 2014, pp. 92–102.

[34] METIS, “Deliverable D5.1: Intermediate description of the spectrumneeds and usage principles,” Aug. 2014.

[35] J. M. Chapin and W. H. Lehr, “Cognitive Radios for Dynamic SpectrumAccess - The Path to Market Success for Dynamic Spectrum AccessTechnology,” IEEE Communications Magazine, vol. 45, no. 5, pp. 96–103, May 2007.

[36] FCC and NTIA, “Coordination Procedures in the 1695-1710 MHz and1755-1780 MHz bands (FCC 13-185),” Jul 2014.

[37] FCC, “Notice of Proposed Rulemaking (ET Docket No. 13-49),” Feb.2013.

[38] FCC, “First Report and Order (ET Docket No. 13-49),” Apr. 2014.[39] J. Lansford, J. Kenney, P. Ecclesine, T. Yucek, and P. Spaanderman,

“Final Report of DSRC Coexistence Tiger Team,” Tech. Rep.,Mar. 2015. [Online]. Available: https://mentor.ieee.org/802.11/dcn/15/11-15-0347-00-0reg-final-report-of-dsrc-coexistence-tiger-team-clean.pdf

[40] FCC, “Office of Engineering and Technology and Wireless Telecom-mmunications Bureau Seek Information on Current Trends in LTE-Uand LAA Technology, Docket No. 15-105,” May 2015.

[41] Fierce WirelessTech, “Verizon, Qualcomm lobby for LTE-U, Wi-Fi coexistence scheme,” http://www.fiercewireless.com/tech/story/verizon-qualcomm-lobby-lte-u-wi-fi-coexistence-scheme/2015-07-20,accessed: 11-29-2015.

[42] FCC, “Report and Order, ET Docket No. 07-113,” Aug. 2013.[43] FCC, “Report and Order, WT Docket No. 02-146,” Oct. 2003.

[44] FCC, “Memorandum Opinion and Order, WT Docket No. 02-146,”Mar. 2005.

[45] FCC, “Notice of Proposed Rulemaking, GN Docket No. 14-177,” Oct.2015.

[46] NTIA, “Fifth Interim Progress Report on the Ten-Year Plan andTimetable ,” Apr. 2015.

[47] ECC, “Request for Opinion on Licensed Shared Access (LSA),” Nov.2012.

[48] ECC, “Licensed Shared Access (LSA),” Feb. 2014.[49] OfCom, “Ofcom Consultation on the UK Preparations for the World

Radio Communication Conference 2015 (WRC-15),” Jun. 2014.[50] OfCom, “Implementing TV White Spaces,” Feb. 2015.[51] OfCom, “A Framework for Spectrum Sharing,” Jul. 2015.[52] Industry Canada, “White Space Devices (WSDs) (RSS-222),” Feb.

2015.[53] Radio Spectrum Management, “Television White Space Devices Cer-

tification and Licensing Rules,” Nov. 2014.[54] M. Palola, M. Matinmikko, J. Prokkola, M. Mustonen, M. Heikkila,

T. Kippola, S. Yrjola, V. Hartikainen, L. Tudose, A. Kivinen, J. Paavola,and K. Heiska, “Live field trial of Licensed Shared Access (LSA)concept using LTE network in 2.3 GHz band,” in IEEE Symposiumon Dynamic Spectrum Access Networks (DySPAN), 2014.

[55] S. Yrjola, “Licensed Shared Access (LSA) Field Trial Using LTENetwork and Self Organized Network LSA Controller,” in Communica-tions Technologies and Software Defined Radio (WinnComm-Europe),2015 Wireless Innovation Forum European Confernece on, Oct. 2015.

[56] J.-M. Park, J. Reed, A. Beex, T. Clancy, V. Kumar, and B. Bahrak,“Security and enforcement in spectrum sharing,” Proceedings of theIEEE, vol. 102, no. 3, pp. 270–281, Mar. 2014.

[57] M. Altamimi and M. B. Weiss, “Enforcement and Network Capacityin Spectrum Sharing: Quantifying the Benefits of Different Enforce-ment Scenarios,” in Proceedings of The 41st Research Conference onCommunication, Information and Internet Policy, Mar. 2014.

[58] B. Bahrak, A. Deshpande, and J.-M. . Park, “Spectrum Access PolicyReasoning for Policy-based Cognitive Radios,” Computer Networks,vol. 56, no. 11, pp. 2649–2663, 2012.

[59] C. Li, A. Raghunathan, and N. Jha, “An Architecture for Secure Soft-ware Defined Radio,” in Design, Automation Test in Europe ConferenceExhibition, 2009. DATE ’09., Apr. 2009, pp. 448–453.

[60] S. Xiao, J.-M. Park, and Y. Ye, “Tamper Resistance for SoftwareDefined Radio Software,” in Computer Software and ApplicationsConference, 2009. COMPSAC ’09. 33rd Annual IEEE International,vol. 1, Jul. 2009, pp. 383–391.

[61] C. R. Aguayo Gonzalez and J. H. Reed, “Power Fingerprinting in SDRIntegrity Assessment for Security and Regulatory Compliance,” AnalogIntegr. Circuits Signal Process., vol. 69, no. 2-3, pp. 307–327, Dec.2011.

[62] X. Li, J. Chen, and F. Ng, “Secure Transmission Power of CognitiveRadios for Dynamic Spectrum Access Applications,” in InformationSciences and Systems, 2008. CISS 2008. 42nd Annual Conference on,Mar. 2008, pp. 213–218.

[63] S. Bhattarai, A. Ullah, J.-M. Park, J. Reed, D. Gurney, and B. Gao,“Defining Incumbent Protection Zones on the Fly: Dynamic Boundariesfor Spectrum Sharing,” in Proc. of IEEE Symposium on DynamicSpectrum Access Networks (DySPAN), Sep-Oct 2015.

[64] A. Ullah, S. Bhattarai, J.-M. Park, J. Reed, D. Gurney, and B. Bahrak,“Multi-Tier Exclusion Zones for Dynamic Spectrum Sharing,” in Proc.of IEEE International Conference in Communications (ICC), Jun. 2015.

[65] R. O’Connor, “Understanding Television’s Grade A and Grade BService Contours,” Broadcasting, IEEE Transactions on, vol. 47, no. 3,pp. 309–314, Sep. 2001.

[66] FCC, “Propagation Data Required for the Evaluation of CoordinationDistances in the Frequency Range 0.85-60 GHz (Rec. ITU-R P.620-3),”Jul. 2014.

[67] NTIA, “Portal Opens for AWS-3 Spectrum Shar-ing Coordination,” https://www.ntia.doc.gov/blog/2015/portal-opens-aws-3-spectrum-sharing-coordination, accessed: 11-30-2015.

[68] S. Shellhammer, S. Sai, T. Rahul, and T. James, “Performance of powerdetector sensors of DTV signals in IEEE 802.22 WRANs,” in Proc.of ACM Workshop on Technology & Policy for Accessing Spectrum,2006.

[69] R. Tandra, S. Mishra, and A. Sahai, “What is a Spectrum Hole andWhat Does it Take to Recognize One?” Proceedings of the IEEE,vol. 97, no. 5, May. 2009.

17

[70] National Telecommunications and Information Administration, “AnAssessment of the Near-Term Viability of Accommodating WirelessBroadband Systems in the 1675-1710 MHz, 1755-1780 MHz, 3500-3650 MHz, and 4200-4220 MHz, 4380-4400 MHz Bands,” Oct. 2010.

[71] RYSAVY Research, “Spectrum Sharing: The Promise and the Reality,”Jul. 2012.

[72] J. C. Ribeiro, J. Ribeiro, J. Rodriguez, R. Dionisio, H. Esteves,P. Duarte, and P. Marques, “Testbed for Combination of Local Sensingwith Geolocation Database in Real Environments,” IEEE WirelessCommunications, vol. 19, no. 4, Aug. 2012.

[73] T. Zhang and S. Banerjee, “Inaccurate Spectrum Databases? PublicTransit to its Rescue!” in Proc. ACM HotNets’13, Nov. 2013.

[74] A. Chakraborty and S. R. Das, “Measurement-Augmented SpectrumDatabases for White Space Spectrum,” in Proc. ACM CoNEXT’14,Dec. 2014.

[75] D. Rojerio, J. Ribeiro, P. Marques, and J. Rodriguez, “Combinationof a Geolocation Database Access with Infrastructure Sensing in TVBands,” EURASIP Journal on Wireless Communications and Network-ing, Dec. 2014.

[76] N. Wang, Y. Gao, and B. Evans, “Database-Augmented SpectrumSensing Algorithm for Cognitive Radio,” in Communications (ICC),2015 IEEE International Conference on, Jun. 2015, pp. 7468–7473.

[77] Y. Zhao, B. Le, and J. H. Reed, “Network Support—The RadioEnvironment Map,” in Cognitive Radio Technology. Elsevier, 2006.

[78] H. B. Yilmaz, T. Tugcu, F. Alagoz, and S. Bayhan, “Radio Environ-ment Map as Enabler for Practical Cognitive Radio Networks,” IEEECommunications Magazine, vol. 51, no. 12, Dec. 2013.

[79] Y. Zhao, L. Morales, J. Gaeddert, K. Bae, J.-S. Um, and J. Reed,“Applying Radio Environment Maps to Cognitive Wireless RegionalArea Networks,” in New Frontiers in Dynamic Spectrum Access Net-works, 2007. DySPAN 2007. 2nd IEEE International Symposium on,Apr. 2007, pp. 115–118.

[80] V. Pankonin and R. Price, “Radio Astronomy and Spectrum Manage-ment: The Impact of WARC-79,” Communications, IEEE Transactionson, vol. 29, no. 8, pp. 1228–1237, Aug 1981.

[81] T. Gergely, “Spectrum Access for the Passive Services: The Past andthe Future,” Proceedings of the IEEE, vol. 102, no. 3, pp. 393–398,Mar. 2014.

[82] J. Ford and K. Buch, “RFI mitigation techniques in radio astronomy,”in Geoscience and Remote Sensing Symposium (IGARSS), 2014 IEEEInternational, Jul. 2014, pp. 231–234.

[83] ITU-R, “Techniques for Mitigation of Radio Frequency Interference inRadio Astronomy (Rep. ITU-R RA.2126),” 2007.

[84] V. Brik, S. Banerjee, M. Gruteser, and S. Oh, “Wireless DeviceIdentification with Radiometric Signatures,” in Proceedings of the 14thACM International Conference on Mobile Computing and Networking,ser. MobiCom ’08, 2008, pp. 116–127.

[85] B. Danev and S. Capkun, “Transient-based Identification of WirelessSensor Nodes,” in Information Processing in Sensor Networks, 2009.IPSN 2009. International Conference on, Apr. 2009, pp. 25–36.

[86] W. Hou, X. Wang, and J. Chouinard, “Physical Layer Authenticationin OFDM Systems based on Hypothesis Testing of CFO Estimates,” inCommunications (ICC), 2012 IEEE International Conference on, Jun.2012, pp. 3559–3563.

[87] O. Ureten and N. Serinken, “Wireless Security through RF Finger-printing,” Electrical and Computer Engineering, Canadian Journal of,vol. 32, no. 1, pp. 27–33, 2007.

[88] I. J. Cox, M. Miller, and A. McKellips, “Watermarking as Commu-nications with Side Information,” Proceedings of the IEEE, vol. 87,no. 7, pp. 1127–1141, Jul. 1999.

[89] C. Fei, D. Kundur, and R. Kwong, “Analysis and Sesign of SecureWatermark-based Authentication Systems,” Information Forensics andSecurity, IEEE Transactions on, vol. 1, no. 1, pp. 43–55, Mar. 2006.

[90] N. Goergen, T. Clancy, and T. Newman, “Physical Layer AuthenticationWatermarks through Synthetic Channel Emulation,” in New Frontiersin Dynamic Spectrum, 2010 IEEE Symposium on, Apr. 2010, pp. 1–7.

[91] J. E. Kleider, S. Gifford, S. Chuprun, and B. Fette, “Radio FrequencyWatermarking for OFDM Wireless Networks,” in Acoustics, Speech,and Signal Processing, 2004. Proceedings. (ICASSP ’04). IEEE Inter-national Conference on, vol. 5, May 2004, pp. V–397–400 vol.5.

[92] V. Kumar, J.-M. Park, T. Clancy, and K. Bian, “PHY-layer Authenti-cation by Introducing Controlled Inter Symbol Interference,” in IEEECNS, 2013, pp. 10–18.

[93] Y. Liu, P. Ning, and H. Dai, “Authenticating Primary Users’ Signals inCognitive Radio Networks via Integrated Cryptographic and WirelessLink Signatures,” in IEEE Symp. Security and Privacy, May 2010, pp.286–301.

[94] R. Miller and W. Trappe, “Short paper: ACE: Authenticating theChannel Estimation Process in Wireless Communication Systems,” inProc. ACM WiSec, 2011, pp. 91–96.

[95] X. Tan, K. Borle, W. Du, and B. Chen, “Cryptographic Link Signaturesfor Spectrum Usage Authentication in Cognitive Radio,” in Proc. ACMWiSec, Jun. 2011, pp. 79–90.

[96] X. Wang, Y. Wu, and B. Caron, “Transmitter Identification using Em-bedded Pseudo Random Sequences,” IEEE Trans. Broadcast., vol. 50,pp. 244–252, Sep. 2004.

[97] L. Xiao, L. Greenstein, N. Mandayam, and W. Trappe, “Using thePhysical Layer for Wireless Authentication in Time-Variant Channels,”IEEE Trans. Wireless Commun., vol. 7, no. 7, pp. 2571–2579, Jul. 2008.

[98] P. Yu, J. Baras, and B. Sadler, “Physical-Layer Authentication,” IEEETrans. Inf. Forensics Security, vol. 3, no. 1, pp. 38–51, Mar. 2008.

[99] B. Danev, H. Luecken, S. Capkun, and K. Defrawy, “Attacks onPhysical-Layer Identification,” in ACM WiSec, 2010, pp. 89–98.

[100] V. Kumar, and J-M. Park and K. Bian, “Blind Transmitter Authen-tication for Spectrum Security and Enforcement,” in Computer andCommunications Security (CCS), 21st ACM Conference on, Oct. 2014.

[101] “IEEE Standard for Policy Language Requirements and System Ar-chitectures for Dynamic Spectrum Access Systems,” IEEE Std 1900.5-2011, pp. 1–51, Jan. 2012.

[102] R. Chen, J.-M. Park, and J. Reed, “Defense against Primary UserEmulation Attacks in Cognitive Radio Networks,” IEEE J. Sel. AreasCommun., vol. 26, no. 1, pp. 25–37, Jan. 2008.

[103] S. Liu, Y. Chen, W. Trappe, and L. J. Greenstein, “Non-InteractiveLocalization of Cognitive Radios based on Dynamic Signal StrengthMapping,” in Proc. Sixth Int. Conf. on Wireless On-Demand NetworkSystems and Services, 2009, pp. 77–84.

[104] T. He, C. Huang, B. M. Blum, J. A. Stankovic, and T. Abdelzaher,“Range-Free Localization Schemes for Large Scale Sensor Networks,”in Proc. ACM MobiCom, 2003, pp. 81–95.

[105] A. Dutta and M. Chiang, “See Something, Say Something: Crowd-sourced Enforcement of Spectrum Policies,” Wireless Communications,IEEE Transactions on, vol. PP, no. 99, pp. 1–1, 2015.

[106] K. Woyach, A. Sahai, G. Atia, and V. Saligrama, “Crime and Punish-ment for Cognitive Radios,” in Communication, Control, and Comput-ing, 2008 46th Annual Allerton Conference on, Sep. 2008, pp. 236–243.

[107] M. B. H. Weiss, W. Lehr, M. Altamimi, and L. Cui, “Enforcement inDynamic Spectrum Access Systems,” in 2012 Confernece on TRPC,Mar. 2012.

[108] FCC: Spectrum/Receiver Performance Working Group, “Introductionto Interference Resolution, Enforcement, and Radio Noise,” Tech. Rep.,Jun. 2014.

[109] Standard ECMA-392, “MAC and PHY for Operation in TV WhiteSpace,” Dec 2009.

[110] T. Baykas, M. Kasslin, M. Cummings, H. Kang, J. Kwak, R. Paine,A. Reznik, R. Saeed, and S. Shellhammer, “Developing a Standardfor TV White Space Coexistence: Technical Challenges and SolutionApproaches,” Wireless Communications, IEEE, vol. 19, no. 1, pp. 10–22, Feb. 2012.

[111] “IEEE Standard for Information Technology– Local and MetropolitanArea Networks—Specific Requirements– Part 22: Cognitive WirelessRAN Medium Access Control (MAC) and Physical Layer (PHY)Specifications: Policies and Procedures for Operation in the TV Bands,”IEEE Std 802.22-2011, pp. 1–680, July 2011.

[112] IEEE 802.11 homepage, Accessed: 2015-12-01. [Online]. Available:http://www.ieee802.org/11/

[113] “Project Authorization Request for IEEE Standard for Local andMetropolitan Area Networks Part 15.4: Low Rate Wireless PersonalArea Networks (LR-WPANs) Amendment: TV White Space between54 MHz and 862 MHz Physical Layer,” IEEE Std 802.15.4m, Jul. 2011.

[114] “Project Authorization Request for P1900.7 Radio Interface for WhiteSpace Dynamic Spectrum Access Radio Systems Supporting Fixed andMobile Operation,” Jun. 2011.

[115] T. Baykas, M. Cummings, H. Kang, M. Kasslin, J. Kwak, R. Paine,A. Reznik, R. Saeed, and S. Shellhammer, “Developing a Standardfor TV White Space Coexistence: Technical Challenges and SolutionApproaches,” IEEE 802.19.1 Task Group White Paper, 2011.

[116] “Workshop on TV White Space Coexistence: IEEE 802.19.1Overview,” Jul. 2010, Accessed: 2015-12-01. [Online]. Available:http://ieee802.org/19/pub/Workshop/2 Baykas-NICT.pdf

[117] P. M. and F. S., “LTE Unlicensed and Wi-Fi: Moving beyond Coexis-tence,” Tech. Rep., 2015.

[118] Qualcomm Research, “LTE in Unlicensed Spectrum: HarmoniousCoexistence with Wi-Fi,” Tech. Rep., Jun. 2014.

18

[119] Signals Research Group, “The Prospect of LTE and Wi-Fi SharingUnlicensed Spectrum: Good Fences Make Good Neighbors,” Tech.Rep., Feb. 2015.

[120] T. Nihtila, V. Tykhomyrov, O. Alanen, M. Uusitalo, A. Sorri,M. Moisio, S. Iraji, R. Ratasuk, and N. Mangalvedhe, “SystemPerformance of LTE and IEEE 802.11 Coexisting on a Shared Fre-quency Band,” in Wireless Communications and Networking Confer-ence (WCNC), 2013 IEEE, Apr. 2013, pp. 1038–1043.

[121] R. Ratasuk, M. Uusitalo, N. Mangalvedhe, A. Sorri, S. Iraji, C. Wijting,and A. Ghosh, “License-exempt LTE Deployment in HeterogeneousNetwork,” in Wireless Communication Systems (ISWCS), 2012 Inter-national Symposium on, Aug. 2012, pp. 246–250.

[122] M. Xing, Y. Peng, T. Xia, H. Long, and K. Zheng, “Adaptive SpectrumSharing of LTE Co-Existing with WLAN in Unlicensed FrequencyBands,” in Vehicular Technology Conference (VTC Spring), 2015 IEEE81st, May 2015, pp. 1–5.

[123] J. Stine and C. Caicedo Bastidas, “Enabling Spectrum Sharing viaSpectrum Consumption Models,” Selected Areas in Communications,IEEE Journal on, vol. 33, no. 4, pp. 725–735, April 2015.

[124] C. Caicedo and J. Stine, “Spectrum Markets and Sharing via SpectrumConsumption Models,” in 41st Res. Conf. TPRC, Mar. 2013.

[125] J. Stine and C. Bastidas, “Service Level Agreements with SpectrumConsumption Models,” in Dynamic Spectrum Access Networks (DYS-PAN), 2014 IEEE International Symposium on, Apr. 2014, pp. 206–214.

[126] R. Etkin, A. Parekh, and D. Tse, “Spectrum Sharing for UnlicensedBands,” Selected Areas in Communications, IEEE Journal on, vol. 25,no. 3, pp. 517–528, Apr. 2007.

[127] J. Suris, L. DaSilva, Z. Han, and A. MacKenzie, “Cooperative GameTheory for Distributed Spectrum Sharing,” in Communications, 2007.ICC ’07. IEEE International Conference on, Jun. 2007, pp. 5282–5287.

[128] S. Yun and L. Qiu, “Supporting WiFi and LTE Co-existence,” inComputer Communications (INFOCOM), 2015 IEEE Conference on,Apr. 2015, pp. 810–818.

[129] N. Gupta, A. Prakash, and R. Tripathi, “Medium access controlprotocols for safety applications in vehicular ad-hoc network: Aclassification and comprehensive survey,” Vehicular Communications,vol. 2, no. 4, pp. 223–237, 2015.

[130] Y. Niu, Y. Li, D. Jin, L. Su, and A. V. Vasilakos, “A survey of millimeterwave communications (mmwave) for 5g: opportunities and challenges,”Wireless Networks, vol. 21, no. 8, pp. 2657–2676, 2015.

[131] S. Rangan and T. S. Rappaport and E. Erkip, “Millimeter-Wave CellularWireless Networks: Potentials and Challenges,” Proceedings of theIEEE, vol. 102, no. 3, pp. 366–385, Mar. 2014.

[132] B. Bahrak, S. Bhattarai, A. Ullah, J.-M. Park, J. Reed, and D. Gurney,“Protecting the Primary Users’ Operational Privacy in Spectrum Shar-ing,” in Dynamic Spectrum Access Networks (DYSPAN), 2014 IEEEInternational Symposium on, April 2014, pp. 236–247.

[133] C. Dwork, “Differential privacy,” in International Conference on Au-tomata, Languages and Programming, 2006, pp. 1–12.

[134] M. E. Andres, N. E. Bordenabe, K. Chatzikokolakis, andC. Palamidessi, “Geo-indistinguishability: Differential Privacy forLocation-based Systems,” in Proceedings of the 2013 ACM SIGSACConference on Computer, Communications Security, ser. CCS ’13.ACM, 2013, pp. 901–914.

[135] N. E. Bordenabe, K. Chatzikokolakis, and C. Palamidessi, “OptimalGeo-Indistinguishable Mechanisms for Location Privacy, booktitle =Proceedings of the 2014 ACM SIGSAC Conference on Computer andCommunications Security,” ser. CCS ’14. New York, NY, USA: ACM,2014, pp. 251–262.

[136] Z. Gao, H. Zhu, Y. Liu, M. Li, and Z. Cao, “Location Privacy inDatabase-driven Cognitive Radio Networks: Attacks and Countermea-sures,” in IEEE Proc. of INFOCOM, April 2013, pp. 2751–2759.

[137] Z. Chen, J. Wang, Z. Zhang, and S. Xinxia, “A Fully HomomorphicEncryption Scheme with Better Key Size,” Communications, China,vol. 11, no. 9, pp. 82–92, Sept 2014.

[138] R. Shokri, G. Theodorakopoulos, C. Troncoso, J.-P. Hubaux, and J.-Y.Le Boudec, “Protecting Location Privacy: Optimal Strategy AgainstLocalization Attacks,” in Proc. of the 2012 ACM Conference onComputer and Communications Security.

[139] M. Gruteser and D. Grunwald, “Anonymous Usage of Location-BasedServices Through Spatial and Temporal Cloaking,” in Proceedings ofthe 1st International Conference on Mobile Systems, Applications andServices, ser. MobiSys ’03. ACM, 2003, pp. 31–42.

[140] J. Freudiger, R. Shokri, and J.-P. Hubaux, “On the Optimal Placementof Mix Zones,” Privacy Enhancing Technologies, vol. 5672, pp. 216–234, 2009.

[141] R. Chow and P. Golle, “Faking Contextual Data for Fun, Profit, andPrivacy,” in Proceedings of the 8th ACM Workshop on Privacy in theElectronic Society, ser. WPES ’09. ACM, 2009, pp. 105–108.

[142] B. Gedik and L. Liu, “Protecting Location Privacy with Personalizedk-Anonymity: Architecture and Algorithms,” Mobile Computing, IEEETransactions on, vol. 7, no. 1, pp. 1–18, Jan. 2008.

[143] ICANN, “FactSheet: Root Server Attack on 6 February 2007,” Feb.2007.

[144] NSF, “Enhancing Access to the Radio Spectrum,” Aug. 2010.[Online]. Available: http://www.nsf.gov/mps/ast/ears workshop finalreport ce final corr.pdf

[145] R. Dionisio, J. Ribeiro, P. Marques, and J. Rodriguez, “Combinationof a Geolocation Database Access with Infrastructure Sensing in TVBands,” EURASIP Journal on Wireless Communications and Network-ing, 2014.

[146] G. Baldini, T. Sturman, A. Biswas, R. Leschhorn, G. Godor, andM. Street, “Security Aspects in Software Defined Radio and CognitiveRadio Networks: A Survey and A Way Ahead,” CommunicationsSurveys Tutorials, IEEE, vol. 14, no. 2, pp. 355–379, 2012.

[147] D. Symeonidis and G. Baldini, “European Standardization and SDRCertification,” in Telecommunications (AICT), 2010 Sixth AdvancedInternational Conference on, May 2010, pp. 136–141.

[148] J. Giacomoni and D. Sicker, “Difficulties in Providing Certification andAssurance for Software Defined Radios,” in New Frontiers in DynamicSpectrum Access Networks, 2005. DySPAN 2005. 2005 First IEEEInternational Symposium on, Nov. 2005, pp. 526–538.