Smart Spectrum Technologies for Mobile Information Systems · 2019. 8. 7. · Smart Spectrum...

Mobile Information Systems

Smart Spectrum Technologies for Mobile Information Systems

Guest Editors Miguel Loacutepez-Beniacutetez Janne Lehtomaumlki Kenta Umebayashi and Fernando Casadevall

Smart Spectrum Technologies forMobile Information Systems



Guest Editors Miguel Loacutepez-Beniacutetez Janne LehtomaumlkiKenta Umebayashi and Fernando Casadevall

Copyright copy 2016 Hindawi Publishing Corporation All rights reserved

This is a special issue published in ldquoMobile Information Systemsrdquo All articles are open access articles distributed under the Creative Com-mons Attribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

Editor-in-ChiefDavid Taniar Monash University Australia

Editorial Board

Markos Anastassopoulos UKClaudio Agostino Ardagna ItalyJose M Barcelo-Ordinas SpainRaquel Barco SpainAlessandro Bazzi ItalyPaolo Bellavista ItalyCarlos T Calafate SpainMariacutea Calderon SpainMarcello Caleffi ItalyJuan C Cano SpainSalvatore Carta ItalyYuh-Shyan Chen TaiwanMassimo Condoluci UKAntonio de la Oliva Spain

Jesus Fontecha SpainJorge Garcia Duque SpainRomeo Giuliano ItalyFrancesco Gringoli ItalySergio Ilarri SpainPeter Jung GermanyAxel Kuumlpper GermanyDik Lun Lee Hong KongHua Lu DenmarkSergio Mascetti ItalyElio Masciari ItalyFranco Mazzenga ItalyEduardo Mena SpainMassimo Merro Italy

Jose F Monserrat SpainFrancesco Palmieri ItalyJose Juan Pazos-Arias SpainVicent Pla SpainDaniele Riboni ItalyPedro M Ruiz SpainMichele Ruta ItalyCarmen Santoro ItalyStefania Sardellitti ItalyFloriano Scioscia ItalyLuis J G Villalba SpainLaurence T Yang CanadaJinglan Zhang Australia

Contents

Smart Spectrum Technologies for Mobile Information SystemsMiguel Loacutepez-Beniacutetez Janne Lehtomaumlki Kenta Umebayashi and Fernando CasadevallVolume 2016 Article ID 3402450 2 pages

CBRS Spectrum Sharing between LTE-U andWiFi AMultiarmed Bandit ApproachImtiaz Parvez M G S Sriyananda İsmail Guumlvenccedil Mehdi Bennis and Arif SarwatVolume 2016 Article ID 5909801 12 pages

Spectrum Assignment Algorithm for Cognitive Machine-to-Machine NetworksSoheil Rostami Sajad Alabadi Soheir Noori Hayder Ahmed Shihab Kamran Arshad and Predrag RapajicVolume 2016 Article ID 3282505 8 pages

A Survey of the DVB-T Spectrum Opportunities for Cognitive Mobile UsersLaacuteszloacute Csurgai-Horvaacuteth Istvaacuten Rieger and Joacutezsef KerteacuteszVolume 2016 Article ID 3234618 11 pages

ETSI-Standard Reconfigurable Mobile Device for Supporting the Licensed Shared AccessKyunghoon Kim Yong Jin Donghyun Kum Seungwon Choi Markus Mueck and Vladimir IvanovVolume 2016 Article ID 8035876 11 pages

Licensed Shared Access System Possibilities for Public SafetyKalle Laumlhetkangas Harri Saarnisaari and Ari HulkkonenVolume 2016 Article ID 4313527 12 pages

PSUN An OFDM-Pulsed Radar Coexistence Technique with Application to 35 GHz LTESeungmo Kim Junsung Choi and Carl DietrichVolume 2016 Article ID 7480460 13 pages

EditorialSmart Spectrum Technologies for Mobile Information Systems

Miguel Loacutepez-Beniacutetez1 Janne Lehtomaumlki2 Kenta Umebayashi3 and Fernando Casadevall4

1Department of Electrical Engineering and Electronics University of Liverpool Liverpool L69 3GJ UK2Centre for Wireless Communications University of Oulu 90014 Oulu Finland3Department of Electrical and Electronic Engineering Tokyo University of Agriculture and Technology Fuchu 184-8588 Japan4Department of Signal Theory and Communications Technical University of Catalonia 08034 Barcelona Spain

Correspondence should be addressed to Miguel Lopez-Benıtez mlopez-benitezliverpoolacuk

Received 28 July 2016 Accepted 31 July 2016

Copyright copy 2016 Miguel Lopez-Benıtez et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

Despite being one of the most important resources of mobileinformation systems the radio frequency spectrum has usu-ally been sparsely exploited as a result of the static spectrumallocation policies traditionally enforced by spectrum regu-lators This situation has recently led to the development ofnovel smart technologies to improve the efficiency of spec-trum utilization Relying on the principles of dynamic spec-trum access and sharing and addressing all layers of thecommunication protocol stack smart spectrum technologiesenable the coexistence of multiple mobile wireless systemswithin the same spectrumband and therefore offer the poten-tial for a smarter and more efficient exploitation of the radiospectrum in a wide range of scenarios The research commu-nity has been working over the last years to overcome manyof the technical challenges posed by the development of smartspectrum technologiesThis issue compiles some of the latestadvances in the field

In response to the open call for papers we receivedregular papers as well as extended versions of outstandingpapers presented at the 2nd IEEE Intentional Workshop onSmart Spectrum (IWSS 2016) held in conjunction with theIEEEWireless Communications andNetworkingConference(WCNC 2016) in Doha Qatar on April 3 2016 All submis-sions have undergone a rigorous reviewprocess and as a resultsix high-quality papers have been selected for publication inthis special issue

The paper titled ldquoPSUN An OFDM-Pulsed Radar Coex-istence Technique with Application to 35 GHz LTErdquo by SKim et al (an extended version of the paper receiving theIEEE IWSS 2016 Best Paper Award) analyzes the performance

of Precoded SUbcarrier Nulling (PSUN) as a coexistencemechanism between 5G Long-Term Evolution (LTE) sys-tems and federal military radars in the 35 GHz CitizensBroadband Radio Service (CBRS) band The pulsed radarinterference can be suppressed by introducing null tones inthe transmitted OFDM signal (PSUN) in addition to settingto zero (pulse-blanking) the received time-domain samplesaffected by pulsed interference In this context S Kim et alanalyze the impact of imperfect radar pulse prediction onthe performance of a PSUN OFDM system and discuss thefeasibility of 5G applications using 35 GHz LTE with PSUN

The paper titled ldquoCBRS Spectrum Sharing between LTE-U and WiFi A Multi-Armed Bandit Approachrdquo by I Parvezet al considers the spectral coexistence between LTE unli-censed (LTE-U) andWiFi systems in the 35GHzCBRS bandGiven the contention-based channel access mechanism ofWiFi systems an unconstrained operation of LTE systemsin the same band may prevent WiFi systems from accessingthe spectrum To enable a fair coexistence LTE systems canintroduce transmission gaps to allow for WiFi operation IParvez et al propose amultiarmed bandit based adaptive LTEduty cycle selection method for the dynamic optimization ofthese transmission gaps which is combined with a downlinkpower control technique for an improved aggregate capacityand energy efficiency

The paper titled ldquoLicensed SharedAccess SystemPossibil-ities for Public Safetyrdquo by K Lahetkangas et al explores thepossibilities of the Licensed Shared Access (LSA) concept asan approach for spectrum sharing between public safety andcommercial radio systems taking into account the particular

Hindawi Publishing CorporationMobile Information SystemsVolume 2016 Article ID 3402450 2 pageshttpdxdoiorg10115520163402450

2 Mobile Information Systems

features of public safety systems discussing the advantagesand disadvantages of several spectrum sharing alternativesand providing illustrative results on the potential benefits

The paper titled ldquoETSI-Standard Reconfigurable MobileDevice for Supporting the Licensed Shared Accessrdquo by KKim et al presents an implementation of a reconfigurablemobile device for LSA The prototype implements a proce-dure to transfer control signals among the software entitiesof the device in compliance with the reference model of theETSI standard reconfigurable architecture

The paper titled ldquoSpectrum Assignment Algorithm forCognitive Machine-to-Machine Networksrdquo by S Rostamiet al proposes a novel aggregation-based spectrum assign-ment algorithm for cognitive machine-to-machine networksS Rostami et al develop a genetic algorithm taking intoaccount practical constraints such as cochannel interferenceand maximum aggregation span and analyze its benefits interms of spectrum utilization and network capacity

The paper titled ldquoA Survey of the DVB-T SpectrumOpportunities for Cognitive Mobile Usersrdquo by L Csurgai-Horvath et al presents an experimental study of the poten-tial opportunities offered by the terrestrial Digital VideoBroadcasting (DVB-T) TV band for mobile cognitive radioapplications L Csurgai-Horvath et al perform a widebandspectrum survey employing a mobile measurement platformin a urban environment where the received signal powerand its statistics are analyzed in order to identify potentialopportunities for mobile cognitive radio systems

Acknowledgments

We highly appreciate the effort of all the authors in preparingand submitting their papers to this special issue as well as thededication of the anonymous reviewers whose voluntary andinvaluable work has contributed to the overall quality of thisissue

Miguel Lopez-BenıtezJanne Lehtomaki

Kenta UmebayashiFernando Casadevall

Research ArticleCBRS Spectrum Sharing between LTE-U and WiFiA Multiarmed Bandit Approach

Imtiaz Parvez1 M G S Sriyananda1 Esmail Guumlvenccedil2 Mehdi Bennis3 and Arif Sarwat1

1Department of Electrical amp Computer Engineering Florida International University Miami FL 33174 USA2Department of Electrical amp Computer Engineering North Carolina State University Raleigh NC 27513 USA3Department of Communications Engineering University of Oulu 90014 Oulu Finland

Correspondence should be addressed to Arif Sarwat asarwatfiuedu

Received 31 March 2016 Revised 14 June 2016 Accepted 19 July 2016

Academic Editor Miguel Lopez-Benıtez

Copyright copy 2016 Imtiaz Parvez et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

The surge of mobile devices such as smartphone and tablets requires additional capacity To achieve ubiquitous and high data rateInternet connectivity effective spectrum sharing and utilization of the wireless spectrum carry critical importance In this paper weconsider the use of unlicensed LTE (LTE-U) technology in the 35 GHzCitizens BroadbandRadio Service (CBRS) band and developamultiarmed bandit (MAB) based spectrum sharing technique for a smooth coexistence withWiFi In particular we consider LTE-U to operate as a General Authorized Access (GAA) user herebyMAB is used to adaptively optimize the transmission duty cycle ofLTE-U transmissions Additionally we incorporate downlink power control which yields a high energy efficiency and interferencesuppression Simulation results demonstrate a significant improvement in the aggregate capacity (approximately 33) and cell-edgethroughput of coexisting LTE-U and WiFi networks for different base station densities and user densities

1 Introduction

Due to the proliferation of mobile devices and diverse mobileapplications the exponentially increasingmobile data is dou-bled approximately every year [1] The 4G Long-Term Evolu-tion (LTE) has recently emerged as a powerful technology toprovide broadband data rates On the other hand to satisfythe throughput demand of broadband LTE networks in theupcoming years larger bandwidth is needed [2 3] Since thelicensed spectrum is expensive and limited extending theoperation of LTE in the underutilized unlicensed bands isrecently getting significant attention which requires effectivecoexistence with other technologies such as WiFi in thesebands

Recently the Federal Communications Commission(FCC) in the United States has been working on opening a150MHz of spectrum in the 35 GHz band for sharing amongmultiple technologies which is also commonly referred to asthe Citizen Broadband Radio Service (CBRS) However theuse of this spectrum is subject to regularity requirementswhere the incumbent military and meteorological radar

systems have to be protected [4 5] In the CBRS band thereare three kinds of users with hierarchical priority IncumbentAccess (IA) users (tier-1) Prioritized Access License (PAL)users (tier-2) and General Authorized Access (GAA) users(tier-3) as illustrated in Figure 1 In the current scenariothe expansion of unlicensed LTE (LTE-U) as PAL or GAAuser in the CBRS band is an enticing choice because ofhigh penetration at 35 GHz clean channel and wide amountof spectrum [6] The Third-Generation Partnership Project(3GPP) standardization group has been recently working onstandardizing the licensed-assisted access (LAA) technologyin the 5GHz spectrum [7 8] The main goal is to developa global single framework of LAA of LTE in the unlicensedbands where operation of LTE will not critically affect theperformance of WiFi networks in the same carrier In theinitial phase only downlink (DL) operation LTE-A (LTEAdvanced) Carrier Aggregation (CA) in the unlicensed bandwas considered while deferring the simultaneous operationof DL and uplink (UL) to the next phase

Another option for the operation of LTE in the unlicensedspectrum is through a prestandard approach referred to



for secondary use by PAL

Federal RLS and ARNS use Federal RLS3 sites only

Tier-1

Tier-2

Tier-3

Pote

ntia

l ban

ds fo

r LTE

-U

depl

oym

ent

3700 MHz3650MHz3550MHz

150MHz channel for use by GAAA minimum of 80 MHz and maximum of

Floating 70 MHz spectrum

Figure 1 CBRS spectrum with 3 types of users

LTE-U where LTE base stations leave transmission gaps forfacilitating coexistence with WiFi networks Development ofLTE-U technology is led by the industry consortium knownas the LTE-U Forum LTE-Umainly focuses on the operationof unlicensed LTE in the regions (eg USA China) wherelisten before talk (LBT) is not mandatory LTE-U definesthe operation of primary cell in a licensed band with oneor two secondary cells (SCells) each 20MHz in the 5GHzunlicensed band U-NII-1 andor U-NII-3 bands spanning5150ndash5250MHz and 5725ndash5825MHz respectively Howeverboth the LTE-U and LAA need licensed band for controlplane Similar to the 5GHz band CBRS band can be utilizedfor LTE-U operation in the absence of IA users such as radarsignal

In our study we consider the coexistence problemof LTE-U andWiFi networks in the CBRS bands SinceWiFi adopts acontention based medium access control with random back-off [9] for channel access and LTE uses dynamic schedulingfor users the unrestrained LTE operation in the same bandwill generate continuous interference on WiFi service Tooperate LTE-U and WiFi simultaneously in the same unli-censed spectrum fair and reasonable coexistencemechanismis indispensable The adverse impact on DL and UL WiFitransmissions due to LTE deployment in the same bandis analyzed in [10ndash12] emphasizing the need for rigorousstudies In this regard discrete mechanisms such as dynamicchannel selection retaining transmission gaps transmissionduty cycle manipulation and LBT have been proposed inthe literature for harmonious coexistence with improvedperformance To select resources dynamically learn from theenvironment and adaptively modify transmission parame-ters for performance improvement variousmachine learningbased techniques [13ndash16] have been introduced

In this paper we introduce a reinforcement learning(MAB) based adaptive duty cycle section for the coexistencebetween LTE-U and WiFi Multiarmed Bandit (MAB) is amachine learning technique designed to maximize the long-term rewards through learning provided that each agentis rewarded after pulling an arm Basically MAB [17 18]problem resembles a gambler (agent) with a finite number ofslot machines in which the gambler wants to maximum hisrewards over a time horizon Upon pulling an arm a rewardis attained with prior unknown distribution The goal is topull arms sequentially so that the accumulated rewards overthe gambling period are maximized However the problem

involves the exploration versus exploitation trade-off that istaking actions to yield immediate higher reward on the onehand and taking actions that would give rewards in the futureon the other hand

In our technique we use a multiarm bandit (MAB)algorithm for selecting appropriate duty cycle Using a 3GPPcompliant Time Division Duplex- (TDD-) LTE and Beaconenabled IEEE 802 systems in the 35 GHz band we simu-late and evaluate the coexistence performance for differentpercentage of transmission gaps We found a significantthroughput improvement for both systems ensuring harmo-nious coexistence The objectives subsequently the gains ofthis study are not limited to throughput enhancements Thebenefits that are achieved in different dimensions with the aidof MAB scheme and the other supporting techniques like PCcan be summarized as follows

(1) Proper coexistence is achieved due to the dynamicexploring and exploitation byMAB So our techniqueis adaptive

(2) The aggregate capacity is improved Due to theapplication ofMAB algorithm optimal or suboptimalsolutions are achieved

(3) Using DL PC higher capacity values are achievedunder dense UE and STA configurations

(4) Higher energy efficiency is also achieved with PCwhich always attempts to reduce the transmissionpower while increasing the energy efficiency

(5) With the use of learning algorithm a high degree ofefficiency is achieved

To the best of our knowledge our work is the first study thatintroduces MAB for improving the coexistence of LTE andWiFi in the unlicensed bands

The rest of the paper is organized as follows Section 2pro-vides a literature review of coexistence of LTE-U andWiFi InSection 3 we provide our systemmodel and problem formu-lation for LTE andWiFi coexistence Section 4 introduces theproposedMABbased dynamic duty cycle selection approachSimulation results with various parameter configurations arepresented in Section 5 Finally Section 6 provides concludingremarks

2 Related Works

21 Coexistence among Unlicensed LTE and WiFi In theliterature several studies can be found that investigate theperformance of LTE and WiFi coexistence in the unlicensedbands In [19] coexistence performance of LTE and WiFihas been investigated in 900MHz considering single floorand multifloor indoor office scenarios It is shown that theperformance of WiFi is heavily affected when WiFi and LTEoperate simultaneously in the unlicensed spectrum

To facilitate harmonious coexistence between LTE-U andWiFi in the same band mainly three techniques have beenproposed in the literature (1) listen before talk (LBT) (2)dynamic channel selection and (3) coexistence gaps InEurope and Japan LBT is mandatory for data offloading in

Mobile Information Systems 3

unlicensed band The usage of LBT has been justified in[20] with different choice of LBT schemes In [21] LBT ispresented considering interradio access technology (RAT)and intra-RAT In this technique energy detection based LBTis proposed to handle inter-RAT interference whereas crosscorrelation based LBT is used to handle intra-RAT interfer-ence However LBT is not mandatory in USA and Chinawhere alternative coexistence techniques can be explored

In [22]Qualcommpresents an effective channel selectionpolicy based on interference level If the interference of theoccupied channel exceeds a certain level LTE-U changes thechannel provided that the interference is measured beforeand during the operation and both at the user equipment(UE) and the network side On the other hand in [6]adaptive bandwidth channel allocation offered by LTE andLeast Congested Channel Search (LCCS) has been suggestedfor channel selection Dynamic channel selection requiresfree or low-interference channel to utilize Since same bandwill be shared by other cellular service providers as well asdifferent technologies such as WiFi finding of clean channelmay not be practical

In [23] blank subframe allocation by LTE has beenproposed where LTE is restrained from transmitting andWiFi keeps on transmission A similar technique has beenproposed in [24] where certain subframes of LTE-U arereserved for WiFi transmission Qualcomm has proposedCarrier Sensing Adaptive Transmission (CSAT) [22] for LTE-U MAC scheduling in which a fraction of TDD duty cycle isused for LTE-U transmission and the rest is used for othertechnologies The cyclic ONOFF ratio can be adaptivelyadjusted based on the activity ofWiFi during the OFF periodIn this paper we focus on the dynamic optimization of coex-istence gaptransmission time along with DL power control

Uplink (UL) power control has been investigated onthe performance of LTE-WiFi coexistence in [25 26] How-ever DL power control in coexistence problem has notbeen explored yet considering uncoordinated LTE and WiFisystems The DL power control enhances performance byreducing interferences which is demonstrated in [27ndash29] Inour study we optimize both the transmission time and DLpower using machine learning technique

Reinforcement algorithm such as Q-learning multiarmbandit and value iteration is effective variant of machinelearning which has been applied for optimization problemsof cellular systems such as channel selection mobility man-agement resource allocation and rate adoption In [13]Q-learning based duty cycle adjustment is presented tofacilitate the sharing of the channel and to increase theoverall throughput In [30] aMAB based distributed channelselection is proposed to use vacant cellular channels in deviceto device (D2D) communication To enhance handoverprocess and increase throughput MAB techniques basedcontext-aware mobility management scheme is studied in[31] In [32] dynamic rate adaptation and channel selectionfrom free primary users have been proposed in cognitiveradio systems usingMAB which yields extensive throughputimprovements

In our studywe propose aMABbased dynamic duty cycleselection for unlicensed LTE systems In particular LTE base

Tier‐1 IA system

Tier‐2 PALcontroller

Tier‐3 GAAcontroller

Federal SAS Federal database

Interface

Commercial SAS‐2Commercial SAS‐1

Tier‐2 PALRAN user

Tier‐3 GAA Tier‐3 GAA Tier‐3 GAAuser‐1 user‐2 user‐2 CB

RS w

ith li

cens

ed sh

ared

acce

ss (L

SA)

middot middot middot

Figure 2 Users access priority

stations (BSs) measure the utilization of the channel based onchannel status information (CSI) learn the channel utiliza-tion of WiFi (current and previous) select the optimum dutycycle and transmission power and perform transmissionunder this duty cycle which results in effective sharing ofwireless spectrum with WiFi networks Due to this dynamiclearning our technique is adaptive and it improves aggregatecapacity and energy efficiency This is the first time we areapplying MAB for coexisting operation of LTE and WiFi

22 CBRS Spectrum Sharing The CBRS spectrum is com-posed of 150MHz bandwidth divided into two chunks80MHz and 70MHz Based on the architecture of CBRSband the spectrum users are prioritized into three groupswith decreasing interference protection requirements as illus-trated in Figure 2

The IA users in tier-1 such as military radars havethe most protection mainly through geographical exclusionzones [33] that averts other users from transmiting in thevicinity of IA users While the NTIA in April 2015 [5 34]shrunk the earlier exclusion zones in [33] by 77 they stillcover several of the Nationrsquos largest cities [35] The mainchallenge of PAL users in tier-2 have is to protect the IAusers and other PAL users from interference To facilitatethis a spectrum access system (SAS) [36] is utilized whichgrants spectrum access to users based on their locationsThe network providers can purchase PAL licenses in givengeographical areas which consist of census tracts Up to a70MHz of PAL spectrum will be available with chunks of10MHz channels which will be auctioned if there is moredemand from providers than the available spectrum Finallytier-3 users are GAAusers which are allowed to operate in thespectrum that are not used by IA and PAL tiers In areas withno IA and PAL activity GAA users may have access to whole150MHz while in areas with PAL activity but outside of IAexclusion zones at least 80MHz of bandwidth will always beavailable for GAA use

Since spectrum is limited and expensive wireless serviceprovider (LTE WiFi) will be interested to operate in CBRSband as GAA users In the GAA band LTE needs to coexistwith other cellular operators as well as other technologiessuch as WiFi Besides that Licensed Shared Access (LSA)concept [37 38] allows an incumbent spectrum user to share


LTE BS

LTE-U UE

LTE-U UE

LTE BS

WiFi AP

WiFi STA

Desired signalInterference

TE-U UE

WiFi STA

(a) Interference on LTE-U DL and WiFi UL

LTE-U UE

LTE-U UE


LTE BS

LTE BS

WiFi AP

WiFi STA

E-U UE

WiFi STA

(b) Interference on LTE-U UL and WiFi DL

Figure 3 DL and UL interference scenarios for LTE-UWiFi transmissions

spectrum with licensed users with defined rights to accessa portion of spectrum at a given location and time Thisalso requires to develop coexistence mechanism betweenmobile network operators (MNOs) and other technologists(licensedunlicensed) such as WiFi In this study we focuson the coexistence of LTE and WiFi in the 35 GHz CBRSspectrum For this study for simplicity we assume that thecoexistence with IA and PAL users are already maintainedthrough a SAS database and we only consider coexistenceamong LTE-U and WiFi users in the GAA bands

3 System Model and Problem Formulation

To evaluate the coexistence performance of LTE-UwithWiFiin the unlicensed band a collocated LTE-U andWiFi networkscenario is consideredThe sets of LTE-UBSsWiFiAPs LTE-UUEs for BS 119894 andWiFi STAs forAP119908 are given byB

119871B119882

Q119894119871 and Q119908

119882 respectively Q

119871= Q1119871Q2119871 Q119894

119871 Q

|B119871|

119871

and Q119882= Q1119882Q2119882 Q119908

119882 Q

|B119882|

119882 represent the sets of

all UEs and STAs For LTE-U TDD-LTE is considered Forsynchronization of WiFi STAs with the corresponding APs aperiodic beacon transmission is used as in [13]

31 Interference on DL and UL Transmissions Interferencecaused to LTE-UUE and LTE-U BS during DL and UL trans-missions is shown in Figure 3 A TDD frame structure similarto that in [39 Figure 62] is considered for all the BSs andUEswith synchronous operation As shown in Figure 3(a) in thesimultaneous operation of an LTE-U within a WiFi coveragearea the DL LTE-U radio link experiences interference fromother LTE-U DL and WiFi UL transmissions As the sametimeWiFi UL suffers fromnear LTE-U transmission Duringan UL transmission subframe shown in Figure 3(b) LTE-U BS is interfered by the UL transmission of LTE-U UEsas well as the DL transmissions of WiFi Similarly WiFiDL transmission is interfered by other LTE-U ULs wherethe DL received signal of a WiFi STA is interfered by otherLTE-U UL transmissions In the coexistence scenarios with

high density of WiFi users WiFi transmissions get delayeddegrading their capacity performance due to the use of carriersense multiple access with collision avoidance (CSMACA)mechanism [40] This is an additional degradation otherthan the performance reduction experienced due to LTE-Utransmissions operated on the same spectrumand this is validonly for WiFi APs and STAs

32 Duty Cycle of LTE-U In the case of designing a duty cyclefor LTE-Umultiple LTETDD frames are considered For thatpurpose five consecutive LTE frames [39 Figure 62(a)] areused to construct a duty cycle Similar to [13] the LTE-UtransmissionONOFF condition is used to define a duty cyclewhich is shown in Figure 4 (eg 40 duty cycle during thefirst two consecutive LTE-U frames transmission is turnedon and it is turned off during the following three frames) Oneout of these two configurations is used by the UEs and BSin an LTE cell during a duty cycle period According to thisstructure a constant ULDL duty cycle value is maintained

33 Capacity Calculation and Power Control For any BS 119894 isinQ119871 there are N119894 resource blocks (RBs) for the DL For a

given UE 119906 associated with BS 119894 119899119894119906RBs are allocated where

N119894 = sum|Q119894119871|

119906=1119899119894119906 119901119894119904119903 119901119887119904119903 119901119886119904119903 and 119901119902

119904119903are transmit power

values associated with RB 119903 and the transmit power index 119904from the LTE-U BS 119894 LTE-U BS 119887 (119894 = 119887) WiFi AP 119886 andWiFi STA 119902 119894th BS is considered as the desired BS where theBSs indexed by 119887 are the interference generating BSs For anyAP UE or STA total transmit power is equally distributedamong all RBsHowever in every BS the total transmit poweris dynamically changed for every duty cycle according toMAB algorithm ℎ119894

119906119903 ℎ119887119906119903 ℎ119886119906119903 and ℎ119902

119906119903are the channel gain

values from BS 119894 to UE 119906 from BS 119887 to UE 119906 from AP 119886

to UE 119906 and from WiFi STA 119902 to UE 119906 respectively Allchannel gain values are calculated considering path lossesand shadowing In that case interference generated to UE119906 from BSs APs and STAs are given by 119868119906BS 119868

119906

AP and 119868119906

STArespectively Since a synchronized transmission is considered


80 Percentage ofthe duty cyclefor an LTE-U transmission

6040

20

LTE TDD frame LTE TDD frame LTE TDD frame LTE TDD frame LTE TDD frame

Duty cycle

Figure 4 Structure of the duty cycle for LTE-U transmissions

there is no interference from the UL transmission of LTE-U UEs Noise variance is denoted by 1205902 The Signal-to-Interference-plus-Noise Ratio (SINR) expression for UE 119906

served by BS 119894 on RB 119903 at time interval 119896 is given as

SINR119894119906119903[119896]

=119901119894119904ℎ119894119906119903

sum119887isinB119871119894

119901119887119904ℎ119887119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

BS

+ sum119886isinB119882

119901119886119904ℎ119886119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

AP

+ sum119902isinQ119882

119901119902119904ℎ119902119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

STA

+ 1205902 (1)

where 119887 119894 isin B119871

The amount of successfully transmitted data bits119873119861from

119894th LTE-U BS during 119879OFDM time interval 119896 within an activeDL subframes of a duty cycle is given by

119873119894

119861=

K119894

sum119896

sum

119906isinQ119894119871

119877119906

sum119903

119882119894

119906119903log2(1 + SINR119894

119906119903[119896]) 119879OFDM (2)

where119879OFDM is the orthogonal frequency divisionmultiplex-ing (OFDM) symbol duration 119879119894Tx = K119894119879OFDM and K119894 isthe total number of transmit 119879OFDM time intervals for theconsidered duty cycle The total allocated bandwidth for RB119903 for UE 119906 served by BS 119894 is 119882119894

119906119903 The average capacity over

a duty cycle period is used as a performance measure in thisstudy as in [13] The DL capacity 119862

119894of LTE-U BS 119894 is given as

119862119894=

119873119894119861

119879119894Tx + 119879119894

Wait (3)

where 119879119894Wait is the waiting time due to silent subframeallocation

The capacity 119862119894in (3) is used as a performance mea-

sure for each LTE-U BS Since the transmit power of oneBS contributes to the interference power of the other BSneighboring BSs are coupled in terms of interference Thegoal of every BS is to maximize 119862

119894while minimizing the DL

transmit power 119901119894119904 forall119894 isin B

119871 By minimizing the transmit

power values 119901119894119904and 119901119887

119904 the goal is to achieve a comparatively

higher energy efficiency than the case of constantDL transmitpower In the same time a reduction in interference is alsoexpected while guaranteeing a minimum capacity Moreover119875min le 119901

119887

119904le 119875max where 119875min and 119875max are the minimum and

maximum transmit power constraints respectivelyThemin-imum capacity corresponding to a given action is denoted by

119862min119895

The objective is to maximize the average capacity whileminimizing the transmit power which can be written as

maximizesum|B119871|

119894=1119862119894

1003816100381610038161003816B1198711003816100381610038161003816

(4)

minimize 119901119894

119904forall119894 isin B

119871(5)

subject to 119901119894

119904 119901119887

119904 le 119875max

forall119894 119887 isin B119871 119894 = 119887 119904 isin 119878

(6)

119901119894

119904 119901119887

119904 ge 119875min


(7)

119862119894gt 119862

min119895

forall119894 isin B119871 forall119895 isin 119869 (8)

In the case of energy efficiency several parameter config-urations are considered for (8) as

119862119894

119901119894119904

gt119862min119895

119901119894119904

or119862119894

119901119894119904

gt119862min119895

119875min

or119862119894

119901119894119904

gt119862min119895

119875max

(9)

Due to the same denominator 119862119894119901119894119904gt 119862min119895

119901119894119904is simplified

to (8) which can be used as a proportional measure ofenergy efficiencyThe problem is reformulated defining a newobjective to maximize energy efficiency as follows


119894=1(119862119894119901119894119904)

1003816100381610038161003816B1198711003816100381610038161003816

subject to (6) (7) and (9)

(10)

4 MAB Techniques forLTE-U WiFi Coexistence

In a MAB problem an agent selects an action (also knownas arm) and observes the corresponding rewardThe rewardsfor given actionarms are random variables with unknowndistribution The goal of MAB is to design action selection


(1) Initialization(2) Set the minimum capacity values 119862min

119895 forall119895 isin 119869 Exploration steps119872 Beta (1 1) 120572119894

119895and 120573119894

119895where forall119895 119895 isin 119869

Select 119889119894119895 forall119895 isin 119869 update 119904 119899

1198940(119889119894119895) V1198940(119889119894119895) and accumulated hypothesisreward 119877

119894(119889119894119895) based on 119862

119894gt 119862min119895

(3) if 120572119894119895(119898) = 120573119894

119895(119898) forall(119897 119898) isin 119872 then

(4) Exploration(5) for119898 = 1 2 3 119872 do(6) Select 119889119894

119895 119889119894119895isin D119894 119895 isin U(1 |D

119894|) cap 119869 and update 119904 (8)

(7) Execute 119889119894119895 119901119894119904 observe 119862

119894and update 119899

119894119898(119889119894119895)

(8) if 119862119894gt 119862min119895

then(9) Reward 119877

119894(119889119894119895) = 119877119894(119889119894119895) + 1

(10) Update 119904 (119904 larr 119904 minus 1) and V119894119898(119889119894119895) (11)

(11) Update 120572119894119895(119898) = 120572119894

119895(119898) + 1

(12) else(13) Reward 119877

119894(119889119894119895) = 119877119894(119889119894119895) + 0

(14) Update 119904 (119904 larr 119904 + 1) and V119894119898(119889119894119895) (11)

(15) Update 120573119894119895(119898) = 120573119894

119895(119898) + 1

(16) end if(17) if 119877

119894(119889119894

119895) = 119877119894(119889119894

119886) 119889119894119895 119889119894

119886isin D119894 119895 = 119886 forall119895 119886 isin 119869

then(18) Select 119889119894

119896 119889119894119896isin D119894 119896 isin U(1 |D

119894|) cap 119869

(19) else(20) Select 119889119894

119896 (12)

(21) end if(22) Exploitation(23) for 119897 = 1 2 3 119871 do(24) Execute the actionA

119894= 119889119894119896 119901119894119904

(25) end for(26) end for(27) end if

Algorithm 1 Multiarm bandit (Thomson sampling)

strategies to maximize accumulate rewards over a given timehorizon However the strategies need to achieve a trade-offbetween exploration (selection of suboptimal actions to learntheir average rewards) and exploitation (selection of actionswhich have provided maximum rewards so far)

In order to dynamically optimize LTE-U transmissionparameters (ie duty cycle and transmit power) a variantof MAB learning techniques called Thomson sampling [4142] algorithm is applied The scenario is formulated asa multiagent problem G = B

119871 A119894119894isinB119871

119862119894119894isinB119871

considering the BSs as players whereA

119894is the action set for

player 119894 During the entire process each BS needs to strikea balance between exploration and exploitation where thereare119872 exploration and 119871 exploitation steps indexed with 1198981 le 119898 le 119872 and 119897 1 le 119897 le 119871 respectively

(i) Agents LTE-U BSsB119871

(ii) Action The action set of agent 119894 A119894is defined as

A119894= 119889119894119895 119901119894119904119895isin119869119904isin119878

119889119894119895 119901119894119904 is the pair of duty cycle

and transmit power elements Configurations of dutycycles are used as part of the action spaceD whereDis common for all players A given BS 119894 selects 119889119894

119895 119889119894119895isin

D according to Algorithm 1 where 119869 = 1 2 |D|119895 isin 119869 and 119869 isin Z+ Probability spaces of positive

integers are denoted byZ+The set of first elements ofthe action vectorD

119894= 1198891198941 1198891198942 119889119894

|D| of BS 119894 is asso-ciated with the duty cycles as 20 40 80respectively The transmit power values set P isrepresented as 119878 = 1 2 |P| 119904 isin 119878 and119878 isin Z+ 119901119894

119904is the transmit power of player 119894 where

P119894= 1199011198941 1199011198942 119901119894

|P| For each action A119894 there is

a distribution Beta (120572119894119895 120573119894119895) forall119895 isin 119869 where 120572119894

119895and

120573119894119895are the shape parameter However in the case of

power control (PC) if119862119894gt 119862

min119895

119904 is decreased by one(119904 larr 119904minus1) reducing the transmit power119901119894

119904by one level

for the next step 119898 + 1 and vice versa Further when119862119894gt 119862min119895

a reward is achieved And for 119862119894gt 119862min119895

120572119894119895is incremented otherwise 120573119894

119895is incremented

(iii) Decision Function The DL capacity of a BS 119894 119862119894is

used as the utility function In order to select a dutycycle a decision function based on the policy UCB1[43] is used where the accumulated rewards achieveddue to values given by 119862

119894are exploited The decision

value for the duty cycle 119889119894119895related to the exploration


step119898 of BS 119894 V119894119898(119889119894119895) is given in (11) while 119889119894

119896based

on the decision is given in (12)

V119894119898(119889119894

119895) = 119909119894119898(119889119894

119895) + radic

2 ln (119898 +1003816100381610038161003816D119894

1003816100381610038161003816)

119899119894119898(119889119894119895)

(11)

119889119894

119896= argmax119889119894

119895isinD119894

(V119894119898(119889119894

119895)) (12)

where 119909119894119898(119889119894119895) = 119877

119894(119889119894119895)119899119894119898(119889119894119895) The argument of

the maximum value is given by arg max(sdot) 119909119894119898(119889119894119895)

119877119894(119889119894119895) and 119899

119894119898(119889119894119895) are the average reward obtained

from 119889119894119895during the exploration step 119898 total rewards

gained form the same 119889119894119895 and the total number of

times 119889119894119895has been played respectively Selection of 119904

is totally independent of the decision function

The multiagent learning problem is addressed using aMAB approach In the contextual MAB problem handled bythe Thomson sampling algorithm [41] current and previousinformation (ie history) is used for the selection of anarm or action Initially 119889119894

119895 forall119895 isin 119869 are played once with

119901119894119904= 119901119894|P| Based on the accumulated reward 119877

119894(119889119894119895) the

parameters 119904 1198991198940(119889119894119895) and V

1198940(119889119894119895) are updated In the learning

process the accumulated reward is used to play the role of theaccumulated hypothesis defined in [44] Subsequently agentsbalance between 119872 exploration and 119871 exploitations stepsDuring the exploration steps 119889119894

119895is selected randomly where

119889119894119895 119889119894119895isin D119894 119895 isin U(1 |D

119894|)cap119869 where a uniformdistribution

with the minimum and maximum values 1199091and 119909

2is given

by U(1199091 1199092) 119904 is decided based on the last available values

of (8) Subsequently the same set of parameters is updatedAt the end of each exploration step based on (8) and theaccumulated rewards an action is selected Then the sameaction is repeatedly played for all the 119871 exploitation steps ofthat particular exploration step as explained in Algorithm 1

5 Simulation Results

For LTE-U TDD-LTE is considered and it is assumedthat all LTE-U UEs are synchronized in both time andfrequency domain as in [13] with the serving BSs A beacon istransmitted periodically for the purpose of synchronizationof WiFi STAs with the corresponding APs To evaluate theperformance an architecture containing two independentlyoperated layers of cellular deployments is considered asshown in Figure 5 Hexagonal cells with omnidirectionalantennas are assumed LTE-U layer encompasses |B

119871| = 7

BSs and |Q119871| UEs where the WiFi layer includes |B

119882| =

7 APs and |Q119882| WiFi STAs In each cell for each APBS

STAsUEs are dropped at random locations All of them areassumed to be uniformly distributed within the cells of theirserving BSs having a mobility speed of 3 kmh and a randomwalk mobility model We consider a nonfull buffer traffic forbothWiFi and LTE networks where the packet arrivals at thetransmitter queues follow a Poisson distribution The traffic

minus50 0 50

Dist

ance

(m)

100

50

0

minus50

Distance (m)

BSAP

WiFi

Area boundariesLTE-U

LTE-U WiFiCells

Figure 5 Cellular coverage layout used in LTE-U and WiFi coex-istence simulations

arrival rates for LTE-U and WiFi are 120582LTE = 120582WiFi = 25

packetsecondThe LTE and WiFi IEEE 80211n medium access control

(MAC) and physical (PHY) layers are modeled in which aPHY layer abstraction is used for Shannon capacity calcula-tions of WiFi and LTE-U The time granularity of each WiFiOFDM symbol duration is 4 120583s which we use to periodicallycapture the number of successfully received bits [13] For bothtechnologies wireless channel is modeled according to [45]when the systems are operated in the 35 GHz band IndoorHotspot (InH) scenario is considered with path loss andshadowing parameters FTP TrafficModel-2 [45] is employedfor either WiFi or LTE-U with a noise spectral power densityof minus95 dBmHz

In each transmission time interval (TTI) DL SINR isreported to the corresponding BS Based on the number ofLTE-U UEs waiting and requesting UL transmission duringone subframe bandwidth is equally shared among them-selves The simulation parameters for LTE-U transmissionsare summarized in Table 1 TDD configuration 1 [39 Figure62(a)] is used for the LTE-U frames having a 50ms totalduty cycle period Minimum required capacity level 119862min

119895is

10Mbps and the set of power levels isP119894= 1199011198941 1199011198942 119901119894

|P| =

8 13 18 23 dBmFor WiFi CSMACA with enhanced distributed channel

access (EDCA) and clear channel assessment (CCA) has been


Table 1 LTE MACPHY parameters

Parameter ValueFrequency 35 GHzTransmission scheme OFDMBandwidth 20MHzDL Tx power 23 dBmUL Tx power PL Based TPCFrame duration 10msScheduling Round RobinUL base power level 119875

0minus106 dBm

TTI 1ms

Table 2 WiFi MACPHY parameters

Parameter ValueFrequency 35 GHzTransmission scheme OFDMBandwidth 20MHzDLUL Tx power 23 dBmAccess category Best effortMAC protocol EDCACCA channel sensing threshold minus82 dBmCCA energy detection threshold minus62 dBmNo of service bits in PPDU 16 bitsNo of tail bits in PPDU 12 bits

Backoff type Fixed contentionwindow

Contention window size U(0 31)

Noise figure 6 [39]Beacon interval 100msBeacon OFDM symbol detection threshold 10 dBBeacon error ratio threshold 15

implemented All WiFi STAs with traffic in their queue willcompete for channel access after receiving a beacon transmis-sion Without reception of a signal beacon transmission orreception will not be initiated The WiFi STA will sense thechannel andwill transmit if it is idle Otherwise transmissionwill be backed off and the next transmission will be initiatedafter a backoff time Random backoff timemechanism is usedfor this study All the parameters for the WiFi transmissionare summarized in Table 2

51 Aggregate Capacity with MAB Aggregate capacity ofstand-alone WiFi coexisting LTE-U (80 duty cycle) andWiFi (with no MAB algorithm) and MAB based coexistenceof LTE-U and WiFi are presented in Figure 7 The aggregatenumbers of WiFi APs and LTE BSs in all scenarios are keptconstant For the WiFi only deployment we replace all theLTE BSs in Figure 5 with WiFi APs It is notable that withthe use of MAB the overall capacity is increased significantlyfrom stand-alone WiFi operation and simultaneous opera-tion of LTE-U and WiFi (without MAB) Also we found thatwith the increase of intersite distance (ISD) in Figure 5 the

MAB

LTE WiFi

Scenario 1

Scenario 2

20 MHz

10 MHz 10 MHz

Figure 6 Scenario with two cases

capacity deceases This is because of higher serving area perAPsSTA within the ISDs

The WiFi throughput performance with and withoutMAB algorithm is shown in Figure 8 where it is noted thatMAB algorithm improves the WiFi throughput over the twoother scenarios Moreover with the increase of ISD capacitydegrades for all cases The effect of LTE packet arrival rate onaggregate capacity is shown in Figure 9 We found that theaggregate throughput of coexisting LTE and WiFi networksis maximized for 120582

119871= 25 but then it decreases for larger

values of 120582119871due to increased interference levels Also for full

buffer LTE traffic (120582119871= 0) the coexisting system with MAB

has degraded performance compared to coexisting systemwithout MAB

Impact of energy detection threshold on aggregate capac-ity is shown in Figure 10 It is observed that minus62 dBmthreshold provides best performance for all scenarios Sens-ing threshold less than minus62 dBm makes WiFi back off fromtransmission in the presence of LTE transmission and resultsin lower aggregate capacity On the other hand sensingthreshold more than minus62 dBm allows WiFi to transmit in thepresence of LTE operation which reduces aggregate capacitydue to higher interference

For Figure 11 we consider a scenario with two cases asdescribed in Figure 6 In scenario 1 we consider simultaneousoperation of LTE-U and WiFi using MAB on 20MHz band-width On the other hand in scenario 2 stand-alone LTE (ie100 duty cycle) andWiFi are operating on separate 10MHzbandwidth We find that the overall capacity using MAB isimproved significantly when compared with the aggregatecapacity of two stand-alone systems This reflects how thespectral efficiency can be improved usingMAB andmotivatessharing of wireless spectrum among LTE andWiFi networksrather than deploying them separately

The impact of LTE-U UEs and WiFi STAs density onaggregate capacity is given in Figure 12 We find that theaggregate capacity improves for the reductions of users inboth services Comparatively high sensitivity could be seenwhen the density of STAs is changed When the densitiesare reduced particularly the STAs a significant increasein capacity is achieved under reduced interference condi-tions However this reduction is further contributed by theCSMACAmechanism as well Also it is notable that capacitydecreases with the increase of ISD

52 Cell-Edge Performance In Figure 13 5th percentile LTEthroughput for different user densities of STAs is representedWe found that with the increase of STAs 5th percentile UEthroughput reduces due tomore interference caused by STAs


Only WiFiWiFi and LTE-U (80 duty cycle)WiFi and LTE-U (with MAB)

2

4

6

8

10

12

Agg

rega

te ca

paci

ty (b

ps)

times107

50 500250

ISD (m)

Figure 7 Aggregate capacity of coexisting WiFi and LTE-U (80duty cycle) MAB based coexisting LTE-U and WiFi and stand-alone WiFi system for different ISDs


1

2

3

4

5

6

7

8

WiF

i cap

acity

(bps

)

times107

50 500250

ISD (m)

Figure 8 WiFi capacity of coexisting WiFi and LTE-U (80 dutycycle) MAB based coexisting LTE-U and WiFi and stand-aloneWiFi system for different ISDs

However with the increment of UEs the effect of STA densityreducesThismeans that for higher density of UEs and STAsfewer LTE users will experience higher capacity

53 Energy Efficiency Performance Aggregate capacity of|Q119894119871| = 10 and |Q119908

119882| = 10 is presented in Figure 14 for different

power control techniques Four parameter settings are usedfor PC In the first instance noPC is considered In the secondcase PC is used by replacing the parameters in Step (7) of the

LTE traffic arrival rate (packetsecond)

04

06

08

1

12

14

16

18

2

22

Agg

rega

te ca

paci

ty (b

ps)

120582L = 0 120582L = 15 120582L = 25 120582L = 35


times108

Figure 9 Aggregate capacity of coexisting WiFi and LTE-U (80duty cycle) MAB based coexisting LTE-U and WiFi and stand-alone WiFi system for different LTE traffic arrival rates

Only WiFiWiFi and LTE-U (80 duty cycle)WiFi and LTE-U (MAB)

times107

5

6

7

8

9

10

11

12

Agg

rega

te ca

paci

ty (b

ps)

minus62 minus72minus52

Energy detection threshold (dBm)

Figure 10 Aggregate capacity of coexisting sytem ofWiFi and LTE-U (80 duty cycle) MAB based coexisting LTE-U and WiFi andstand-alone WiFi system for various energy detection thresholds

Algorithm 1 with 119862119894119901119894119904gt 119862min119895

119875min where 119875min = 8 dBmFor the third and forth cases parameters are replaced with119862119894119901119894119904gt 119862min119895

119875max and 119862119894 gt 119862min119895

where 119875max = 23 dBmThe set of power levels is defined asP

119894= 1199011198941 1199011198942 119901119894

|P| =

8 11 14 17 20 23 dBm where 119875min = 8 dBm and 119875max =

23 dBm So in the second and third cases a given level ofenergy efficiency is aimed at In the last case according to theexplanation given for (9) the level is dynamically adjusted It


50 500250

ISD (m)

0

2

4

6

8

10

12

Capa

city

(bps

)

LTE-U (10MHz)WiFi (10MHz)LTE-U (10MHz) + WiFi (10MHz) (scenario 2)MAB (20MHz) (scenario 1)

times107

Figure 11 Capacity of 10 STAs orand 10 UEs under stand-aloneWiFi stand-alone LTE coexisting stand-alone WiFi and LTE-U(scenario 1) and MAB based coexisting LTE-U and WiFi (scenario2) for different bandwidths and ISDs

times108

50 500250

ISD (m)

05

1

15

2

25

Agg

rega

te ca

paci

ty (b

ps)

5UEs 5 STAs5UEs 10 STAs

10 UEs 5 STAs10 UEs 10 STAs

Figure 12 Capacity ofMAB based coexistence for different UEs andSTAs ratios and ISDs

is noted that the best and worst performances are found for119875max and 119875min For MAB with PC optimum result is found

In Figure 15 different numbers of UEs are considered toevaluate energy efficiency performance For all the densitiesthe least efficiency is achieved with no PC In the mostdense scenario the best efficiency can be observed under thesecond configuration 119862min

119895119875min [see (9)] As it is expected

with the reduction of densities energy efficiency is increasedHowever after a certain average energy efficiency level nosignificant improvements could be observed

5STA10 STA15 STA

5 1510

Number of UEs

09

1

11

12

13

14

15

5th

perc

entil

e thr

ough

put (

bps)

times107

Figure 13 5th percentile throughput ofMAB based coexisting LTE-U and WiFi for different UEs and STAs ratios

PC configuration

Total (WiFi + LTE-U)WiFiLTE-U

No PC Pmin Pmax PC2

4

6

8

10

12

14

Capa

city

(bps

)

times107

Figure 14 Capacity of 10 UEs and 10 STAs under different PCconfigurations

6 Conclusion

In this paper a MAB based dynamic duty cycle selectionmethod was proposed to facilitate spectrum sharing betweenWiFi and LTE-U in the same unlicensed band Performanceof the proposed algorithm was further enhanced by using aDL PC technique Subsequently the proposed concept wasextended to optimize energy efficiency Considerable gainsin overall throughputs could be achieved via the proposedMAB while ensuring a minimum capacity for LTE-U basedservices in the same band Significant gains in terms of energyefficiency could be achieved where it is observed that the


No PCPC

PminPmax

5 1510

Number of UEs

108

109

Ener

gy effi

cien

cy (b

itsjo

ule)

Figure 15 Energy efficiency under different PC configurations forvarious numbers of UEs (with 10 STAs)

gains under different parameter settings with PC are muchhigher than those with no PC Our future work includesextending our framework to scenarios with IA and PAL usersin the same spectrum

Competing Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors would like to thank Nadisanka Rupasinghe fordeveloping an earlier version of the simulator used in thispaperThis researchwas supported in part by theUSNationalScience Foundation (NSF) under Grants nos ACI-1541108and AST-1443999 and Presidential Fellowship under FloridaInternational University

References

[1] NSN Whitepaper ldquoEnhance mobile networks to deliver 1000times more capacity by 2020rdquo Tech Rep 2013

[2] M SimsekM Bennis and I Guvenc ldquoEnhanced intercell inter-ference coordination inHetNets single vsmultiflow approachrdquoin Proceedings of the IEEE Globecom Workshops (GC Wkshpsrsquo13) pp 725ndash729 Atlanta Ga USA December 2013

[3] M Simsek M Bennis and I Guvenc ldquoLearning basedfrequency- and time-domain inter-cell interference coordina-tion in HetNetsrdquo IEEE Transactions on Vehicular Technologyvol 64 no 10 pp 4589ndash4602 2015

[4] FCC ldquoAmendment of the commissions rules with regard tocommercial operations in the 3550ndash3650MHz bandrdquo Noticeof Proposed Rulemaking and Order 2012 httpsappsfccgovedocs publicattachmatchDA-15-955A1 Rcdpdf

[5] FCC ldquoAmendment of the commissions rules with regard tocommercial operations in the 3550ndash3650MHz bandrdquo Report

and Order and Second Further Notice of Proposed Rulemaking12-354 2015 httpsappsfccgovedocs publicattachmatchFCC-15-47A1pdf

[6] R Zhang M Wang L X Cai Z Zheng X S Shen and L-LXie ldquoLTE-unlicensed the future of spectrum aggregation forcellular networksrdquo IEEE Wireless Communications vol 22 no3 pp 150ndash159 2015

[7] ldquoStudy on licensed-assisted access using LTErdquo Tech Rep RP-141397 3GPP Study Item Edinburgh Scotland 2014

[8] 3GPP ldquoStudy on licensed-assisted access to unlicensed spec-trumrdquo Tech Rep TR 36899 3GPP Athens Greece 2015

[9] L Cai X Shen J WMark and Y Xiao ldquoVoice capacity analysisof WLAN with unbalanced trafficrdquo in Proceedings of the 2ndInternational Conference on Quality of Service in HeterogeneousWiredWireless Networks (QSHINE rsquo05) pp 8ndash9 LakeVista FlaUSA August 2005

[10] F M Abinader E P L Almeida F S Chaves et al ldquoEnablingthe coexistence of LTE and Wi-Fi in unlicensed bandsrdquo IEEECommunications Magazine vol 52 no 11 pp 54ndash61 2014

[11] I Parvez N Islam N Rupasinghe A I Sarwat and I GuvencldquoLAA-based LTE and ZigBee coexistence for unlicensed-bandsmart grid communicationsrdquo inProceedings of the SoutheastCon2016 pp 1ndash6 Norfolk Va USA March-April 2016

[12] N Rupasinghe and I Guvenc ldquoLicensed-assisted access forWiFi-LTE coexistence in the unlicensed spectrumrdquo in Proceed-ings of the IEEE GlobecomWorkshops (GCWkshps rsquo14) pp 894ndash899 Austin Tex USA December 2014

[13] N Rupasinghe and I Gulvenc ldquoReinforcement learning forlicensed-assisted access of LTE in the unlicensed spectrumrdquo inProceedings of the IEEEWireless Communications and Network-ing Conference (WCNC rsquo15) pp 1279ndash1284 New Orleans LaUSA March 2015

[14] M G S Sriyananda I Parvez I Guvenc M Bennis and A ISarwat ldquoMulti-Armed Bandit for LTE-U and WiFi coexistencein unlicensed bandsrdquo in Proceedings of the IEEE Wireless Com-munications and Networking Conference (WCNC rsquo16) DohaQatar April 2016

[15] T Ran S Sun B Rong and M Kadoch ldquoGame theorybased multi-tier spectrum sharing for LTE-A heterogeneousnetworksrdquo in Proceedings of the IEEE International ConferenceonCommunications (ICC rsquo15) pp 3033ndash3038 LondonUK June2015

[16] F Shams G Bacci and M Luise ldquoA Q-learning game-theory-based algorithm to improve the energy efficiency of a multiplerelay-aided networkrdquo inProceedings of the 31st General Assemblyand Scientific Symposium of the International Union of RadioScience (URSI GASS rsquo14) pp 1ndash4 XXXIth URSI August 2014

[17] J C Gittins ldquoBandit processes and dynamic allocation indicesrdquoJournal of the Royal Statistical SocietymdashSeries BMethodologicalvol 41 no 2 pp 148ndash177 1979

[18] P Auer N Cesa-Bianchi and P Fischer ldquoFinite-time analysis ofthe multiarmed bandit problemrdquoMachine Learning vol 47 no2 pp 235ndash256 2002

[19] A M Cavalcante E Almeida R D Vieira et al ldquoPerformanceevaluation of LTE and Wi-Fi coexistence in unlicensed bandsrdquoin Proceedings of the IEEE 77th Vehicular Technology Conference(VTC Spring rsquo13) pp 1ndash6 Dresden Gramany June 2013

[20] R Kwan R Pazhyannur J Seymour et al ldquoFair co-existence ofLicensed Assisted Access LTE (LAA-LTE) and Wi-Fi in unli-censed spectrumrdquo in Proceedings of the 7th Computer Scienceand Electronic Engineering (CEEC rsquo15) pp 13ndash18 ColchesterUK September 2015


[21] N Whitepaper ldquoViews on LAA for unlicensed spectrummdashscenarios and initial evaluation resultsrdquo Tech Rep RWS-140026 3GPP RAN1 Standard Contribution Sophia AntipolisFrance 2014

[22] Qualcomm ldquoQualcomm research LTE in unlicensed spectrumharmonious coexistence with WiFirdquo Tech Rep 3GPP RAN1Standard Contribution 2014

[23] E Almeida A M Cavalcante R C D Paiva et al ldquoEnablingLTEWiFi coexistence by LTE blank subframe allocationrdquo inProceedings of the IEEE International Conference on Communi-cations (ICC rsquo13) pp 5083ndash5088 IEEE BudapestHungary June2013

[24] T Nihtila V Tykhomyrov O Alanen et al ldquoSystem perfor-mance of LTE and IEEE 80211 coexisting on a shared frequencybandrdquo in Proceedings of the IEEE Wireless Communications andNetworking Conference (WCNC rsquo13) pp 1038ndash1043 ShanghaiChina April 2013

[25] F S Chaves E P L Almeida R D Vieira et al ldquoLTE ULpower control for the improvement of LTEWi-Fi coexistencerdquoin Proceedings of the IEEE 78th Vehicular Technology Conference(VTC Fall rsquo13) pp 1ndash6 September 2013


[27] X Xu G Kutrolli and R Mathar ldquoDynamic downlink powercontrol strategies for LTE femtocellsrdquo in Proceedings of the 7thNext Generation Mobile Applications Services and TechnologiesConference pp 181ndash186 September 2013

[28] ZWangW Xiong C Dong JWang and S Li ldquoA novel down-link power control scheme in LTE heterogeneous networkrdquo inProceedings of the International Conference on ComputationalProblem-Solving (ICCP rsquo11) pp 241ndash245 Chengdu ChinaOctober 2011

[29] T Zahir K Arshad Y Ko and KMoessner ldquoA downlink powercontrol scheme for interference avoidance in femtocellsrdquo inProceedings of the 7th International Wireless CommunicationsandMobile Computing Conference (IWCMC rsquo11) pp 1222ndash1226July 2011

[30] S Maghsudi and S Stanczak ldquoChannel selection for network-assisted D2D communication via no-regret bandit learningwith calibrated forecastingrdquo IEEE Transactions on WirelessCommunications vol 14 no 3 pp 1309ndash1322 2015

[31] M Simsek M Bennis and I Guvenc ldquoMobility managementin HetNets a learning-based perspectiverdquo EURASIP Journalon Wireless Communications and Networking vol 2015 no 1article 26 pp 1ndash13 2015

[32] R Combes and A Proutiere ldquoDynamic rate and channelselection in cognitive radio systemsrdquo IEEE Journal on SelectedAreas in Communications vol 33 no 5 pp 910ndash921 2015

[33] G Locke and L E Strickling ldquoAn assessment of the near-termviability of accommodating wireless broadband systems in the1675ndash1710MHz 1755ndash1780MHz 3500ndash3650MHz and 4200ndash4220MHz 4380ndash4400MHz bandsrdquo Report 2010 httpswwwntiadocgovfilesntiapublicationsfasttrackevaluation 11152010pdf

[34] P R Atkins ldquoNTIA letter office of engineering and technologyFCCrdquo GN Docket No 12-354 2015 httpwwwntiadocgovfilesntiapublicationsntia letter docket no 12-354pdf

[35] L Stefani ldquoThe FCC Raises the Curtain on the CitizensBroadband Radio Servicerdquo CommLawBlog Article May 2015

httpwwwcommlawblogcom201505articlesunlicensed-operations-and-emerthe-fcc-raises-the-curtain-on-the-citi-zens-broadband-radio-service

[36] FCC ldquo35 GHz Spectrum Access System Workshoprdquo Washing-ton DC USA 2014 httpswwwfccgovnews-eventsevents20140135-ghz-spectrum-access-system-workshop

[37] ldquoRSPG opinion on licensed shared accessrdquo Tech Rep RSPG13-538 European Commission Radio Spectrum Policy Group2013

[38] ECC ldquoLicensed shared accessrdquo Tech Rep ECC 205 2014[39] S Sesia I Toufik and M Baker LTEmdashThe UMTS Long Term

Evolution From Theory to Practice John Wiley amp Sons NewYork NY USA 2009

[40] E Perahia and R Stacey LTE The UMTS Long Term EvolutionFromTheory to Practice CambridgeUniversity PressNewYorkNY USA 2008

[41] S Agrawal and N Goyal ldquoAnalysis of thompson samplingfor the multi-armed bandit problemrdquo httpsarxivorgabs11111797

[42] N Gupta O-C Granmo and A Agrawala ldquoThompson sam-pling for dynamic multi-armed banditsrdquo in Proceedings ofthe 10th International Conference on Machine Learning andApplications (ICMLA rsquo11) vol 1 pp 484ndash489Honolulu HawaiiUSA December 2011

[43] P Auer N Cesa-Bianchi and P Fischer ldquoFinite-time analysis ofthe multiarmed bandit problemrdquoMachine Learning vol 47 no2-3 pp 235ndash256 2002

[44] J Langford and T Zhang ldquoThe epoch-greedy algorithm formultiarmed bandits with side informationrdquo in Advances inNeural Information Processing Systems J C Platt D KollerY Singer and S T Roweis Eds vol 20 pp 817ndash824 CurranAssociates 2008

[45] 3GPP ldquoEvolved Universal Terrestrial Radio Access (E-UTRA)further advancements for E-UTRA physical layer aspects(release 9)rdquo Tech Rep TR36814 V900 3GPP 2010

Research ArticleSpectrum Assignment Algorithm for CognitiveMachine-to-Machine Networks

Soheil Rostami1 Sajad Alabadi1 Soheir Noori2 Hayder Ahmed Shihab3

Kamran Arshad4 and Predrag Rapajic1

1Department of Engineering Science University of Greenwich London UK2Department of Computer Science University of Karbala Karbala Iraq3School of Engineering and Informatics University of Sussex Brighton UK4Department of Electrical Engineering Ajman University of Science amp Technology Ajman UAE

Correspondence should be addressed to Soheil Rostami srostamigreacuk


Academic Editor Fernando Casadevall

Copyright copy 2016 Soheil Rostami et alThis is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

A novel aggregation-based spectrum assignment algorithm for Cognitive Machine-To-Machine (CM2M) networks is proposedThe introduced algorithm takes practical constraints including interference to the Licensed Users (LUs) co-channel interference(CCI) among CM2M devices and Maximum Aggregation Span (MAS) into consideration Simulation results show clearly thatthe proposed algorithm outperforms State-Of-The-Art (SOTA) algorithms in terms of spectrum utilisation and network capacityFurthermore the convergence analysis of the proposed algorithm verifies its high convergence rate

1 Introduction

Today there are around 4 billion M2M devices in the worldwhile in 2022 the number is expected to reach 50 billion[1] According to Cisco systems currently a single M2Mdevice can generate as much traffic as 3 basic-feature phonesin addition emerging applications and services of M2Mnetworks are expected to increase average traffic per devicefrom 70MB per month in 2014 to 366MB per month in 2018[2] Because of the growth rate of the number of devicesand high demand of data traffic future M2M networks willface many challenges especially with the so-called spectrumscarcity problem

Cognitive Radio (CR) is introduced as a promising solu-tion to tackle spectrum scarcity problem in M2M networksCRhas become one of themost intensively studied paradigmsin wireless communications In CR unlicensed users exploitCR technology to opportunistically access licensed spectrumas long as interference to LUs is kept at an acceptable level [3]A number of M2M applications (such as smart grid health-care and car parking) can benefit from the combination

of CR and M2M communications [1] CM2M networkscan improve spectrum utilisation and energy efficiency inM2M networks [4] The CM2M device can interact with theradio environment by either performing spectrum sensingor accessing spectrum databases or both of them to detectspectrum opportunities [4] After sensing CM2M deviceutilises the discovered unused spectrum according to thedevice requirements

Furthermore TV bands (VHFUHF) which have highlyfavourable propagation characteristics are traditionallyreserved to broadcasters But after the transition from theanalogue broadcast television system to the digital one ahuge number of TV channels (also known as TV WhiteSpaces (TVWS)) are freed up and unused In September 2010the Federal Communications Commission (FCC) releasedsignificant rule to enable unlicensed broadband wirelessdevices to use TVWS Unfortunately due to spectrumfragmentation and as a result of an inefficient command andcontrol spectrum management approach a continuous widesegment of TVWS is rare in many countries including theUnited Kingdom



Available subcarrier

Unavailable subcarrier

Frequency

Figure 1 Subcarrier distribution over spectrum [7]

As CM2M network can sense and be aware of its radioenvironment the aggregation of narrow spectrum oppor-tunities becomes possible Spectrum aggregation provideswider bandwidth and higher throughput for the CM2Mdevices CM2M devices can access discontinuous portionsof the TVWS simultaneously by means of DiscontinuousOrthogonal Frequency Division Multiplexing (DOFDM) [56]

DOFDM is a multicarrier modulation technique andis a variant of OFDM used to aggregate discontinuoussegments of spectrum The main difference between OFDMand DOFDM is ONOFF subcarrier information block [7]A multiple segments of spectrum can be occupied by otherCM2M devices or LUs As a result these subcarriers are off-limits to the CM2M devices [6] Thus to avoid interferingwith these other transmissions the subcarrier within theirvicinity is turned off and unusable for CM2M devices asshown in Figure 1 Moreover available (usable) subcarriersare located in the unoccupied segments of spectrum whichare determined by spectrum broker

Spectrum aggregation is one of the most important LTE-advanced technologies from physical layer perspective andstandardised in LTE Release 10 [8] However in spite ofstandardisation of spectrum aggregation little effort has beenmade to optimise spectrum aggregation by exploiting CRtechnology in M2M networks There is limited literatureavailable on spectrum assignment among CM2M deviceshaving spectrum aggregation capabilities

In [9] an Aggregation-Aware Spectrum AssignmentAlgorithm (AASAA) is proposed to aggregate discrete spec-trum fragments in a greedy manner The algorithm in [9]utilises the first available aggregation range from the lowfrequency side and assumes that all users have the samebandwidth requirement

Huang et al [10] proposed a prediction based spectrumaggregation scheme to increase the capacity and decreasethe reallocation overhead The proposed scheme is referredto as Maximum Satisfaction Algorithm (MSA) for spectrumassignment The main idea is to assign spectrum for theuser with larger bandwidth requirement first leaving betterspectrum bands for remaining users while taking intoconsideration different bandwidth requirements of users andchannel state statistics However MSA does not enhancespectrum utilisation by reusing spectrum within unlicensednetwork that is CCI is neglected in MSA

Recently genetic algorithm (GA) is used for spectrumallocation [11] Ye et al [11] introduced a GA based spectrum

assignment in CR networks but spectrum aggregation capa-bility of users is not considered

For CM2M networks existing spectrum assignment andaggregation solutions are not applicable directly as practicalissues such as Maximum Aggregation Span (MAS) mustbe taken into account Furthermore in aggregation-basedspectrum assignment a major challenge is to manage CCIamong CM2M devices which is not taken into account in theexisting literature The major contributions of this study aretwofold

(1) To prevent multiple CM2M devices from collidingin the overlapping portions of the spectrum a cen-tralised approach is applied Furthermore an integeroptimisation problem to maximise cell throughputis formulated considering CCI and MAS in anaggregation-aware CM2M network

(2) As the spectrum assignment problem is inherentlyseen as an NP-hard optimisation problem evolution-ary approaches can be applied to solve this challeng-ing problem In this article GA is used to solve theaggregation-aware spectrum assignment because ofits simplicity robustness and fast convergence of thealgorithm [12]

This article is organised as follows In Section 2 the spec-trum assignment and aggregation models are presented Theproposed algorithm is explained in Section 3 Simulationresults are discussed in Section 4 followed by conclusions inSection 5

2 System Model

21 Spectrum Assignment Model We assume a CM2M net-work consisting of 119873 CM2M devices defined as Φ =

1206011 1206012 120601

119873 competing for119872 nonoverlapping orthogonal

channels Γ = 1205741 1205742 120574

119872 in uplink All spectrum

assignment and access procedures are controlled by a centralentity called spectrum broker We assume that distributedsensing mechanism and measurement conducted by eachdevice is forwarded to the spectrum broker [13] A spectrumoccupancy map that is constructed at the spectrum brokerand CCI among CM2M devices is determined Furthermorethe spectrum broker can lease single or multiple channels for120601119899isin Φ in a limited geographical region for a certain amount

of time Finally a base station can transmit data to 120601119899in the

assigned channels Figure 2 depicts systemmodel used in thisarticle

We define the channel availabilitymatrix L = 119897119899119898| 119897119899119898isin

0 1119873times119872

as an 119873 times 119872 binary matrix representing channelavailability where 119897

119899119898= 1 if and only if 120574

119898is available to 120601

119899

and 119897119899119898

= 0 otherwise Each 120601119899is associated with a set of

available channels at its location defined as Γ119899sub Γ that is

Γ119899= 120574119898| 119897119899119898

= 0 Due to the different interference rangeof each LU (which depends on LUrsquos transmit power and thephysical distance) at the location of each CM2M device Γ

119899of

different CM2M devices may be different [14] According tothe sharing agreement any 120574

119898isin Γ can be reused by a group of

CM2M devices in the vicinity defined byΦ119898such thatΦ

119898sub


Spectrum broker

CM2M deviceTV

TV broadcast stationCM2M base station

Figure 2 Architecture diagram of CM2M network operating inTVWS

Φ if CM2Mdevices are located outside the interference rangeof LUs that is Φ

119898= 120601119899| 119897119899119898

= 0The interference constraint matrix C = 119888

119899119896119898| 119888119899119896119898

isin

0 1119873times119873times119872

is an119873times119873times119872 binary matrix representing theinterference constraint among CM2M devices where 119888

119899119896119898=

1 if 120601119899and 120601

119896would interfere with each other on 120574

119898 and

119888119899119896119898

= 0 otherwise It should be noted that for 119899 = 119896 119888119899119899119898

=

1minus119897119899119898

Value of 119888119899119896119898

depends on the distance between120601119899and

120601119896 Interference constraint also depends on 120574

119898as power and

transmission rules vary greatly in different frequency bandsThe bandwidth requirements of all CM2Mdevices are diversebecause of different quality of service requirements for eachdeviceWedefineR = 119903

1198991times119873

as device requested bandwidthvector where 119903

119899represents bandwidth demand of 120601

119899

In a dynamic environment channels availability andinterference constraint matrix both vary continually in thisstudy we assume that spectrum availability is static or variesslowly in each scheduling time slot that is allmatrices remainconstant during the scheduling period In our proposedsolution a subset of CM2M devices is scheduled during eachtime slot and the available spectrum is allocated among themwithout causing interference to LUs

22 Spectrum Aggregation Model In the traditional spec-trum assignment each channel is composed of a continuousspectrum fragment thus it is not feasible for users to utilisesmall spectrum fragments which are smaller than the usersbandwidth demand For instance assume a CM2M networkwhere every machine requires 4MHz channel bandwidthand the available spectrum consists of two spectrum frag-ments of 4MHz and four spectrum fragments of 2MHz(Figure 3) For continuous spectrum allocation the 2MHzspectrum fragments cannot be utilised by any machineTherefore a continuous spectrum assignment mode canonly support two devices for communication (2 times 4MHz)However spectrum aggregation-enabled device can exploitfragmented segments of the spectrum by using specialisedair interface techniques such as DOFDM In Figure 3 if anumber of small spectrum fragments are aggregated into awider channel then 16MHz of unused spectrum is availableto support four CM2M devices (4 times 4MHz)

Due to the limited aggregation capabilities of the RFfront-end only channels that reside within a range of MAS

can be aggregated With this constraint some spectrumfragments may not be aggregated because their span islarger than MAS Our proposed algorithm takes MAS intoconsideration For the sake of simplicity we make followingassumptions

(1) All CM2M devices have the same aggregation capa-bility (ie MAS for all devices is the same)

(2) Guard band between adjacent channels is neglected(3) Bandwidth requirement of each device and band-

width of each channel are an integer multiple ofsubchannel bandwidth Δ which is the smallest unitof bandwidth (in fact the smaller fragments woulddemand excessive filtering to limit adjacent channelinterference) that is

119903119899= 120596119899sdot Δ 120596

119899isin N 1 le 119899 le 119873

BW119898= 120581119898sdot Δ 120581

119898isin N 1 le 119898 le 119872

(1)

where N is the set of natural numbers 120596119899is the

number of requested subchannels by 120601119899 120581119898

is thenumber of subchannels in 120574

119898 and BW

119898is the

bandwidth of 120574119898

The total available spectrum (ie119872 channels) is subdividedinto multiple number of subchannels If the available spec-trum band consists of C subchannels (ie total availablebandwidth isC sdot Δ) then

120574119898=

120581119898

⋃

119894=1

119894119898

120581119898=BW119898

Δ

where 1 le 119898 le 119872

C =119872

sum

119898=1

120581119898

(2)

where 120574119898

has 120581119898

subchannels and 119894119898

represents the 119894thsubchannel of 120574

119898 Each

119894119898can be represented in an interval

defined as [F119871119894119898F119867119894119898] where F119871

119894119898and F119867

119894119898are the lowest

and highest frequency of 119894119898

F119867

119894119898minusF119871

119894119898= Δ for 1 le 119894 le 120581

119898 1 le 119898 le 119872 (3)

Based on this new subchannel indexingmatrices L andC canbe rewritten as

Llowast = 119897lowast119899c | 119897lowast

119899c = 119897119899119898119873timesC

Clowast = 119888lowast119899119896c | 119888

lowast

119899119896c = 119888119899119896119898119873times119873timesC

(4)

if1 le c le 120581

1for 119898 = 1

119898minus1

sum

119895=1

120581119895lt c le

119898

sum

119895=1

120581119895

for 1 lt 119898 le 119872(5)


Aggregating spectrum

Available spectrum

Unavailable spectrum

1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

2M

Hz

2M

Hz

2M

Hz

2M

Hz

3M

Hz

4M

Hz

4M

Hz

Figure 3 Aggregation of disjoint spectrum fragments

where c represents index of each subchannel within theavailable spectrum

The subchannel assignment matrix A = 119886119899c | 119886119899c isin

0 1119873timesC is an119873timesC binarymatrix representing subchannels

assigned to CM2M devices for aggregation such that 119886119899c = 1

if and only if subchannel c is available to 120601119899and 0 otherwise

We define the reward vector B = 119887119899= Δ sdot sum

Cc 119886119899c119873times1 to

represent total bandwidth that is allocated to each CM2Mdevice during scheduling time period for a given subchannelassignment

3 Problem Formulation

31 Optimisation Problem One of the key objectives of thedeployment of CM2M network is to enhance the spectrumutilisation To consider this crucial goal we define networkutilisation tomaximise the total bandwidth that is assigned toCM2Mdevices and referred to asMaximising Sumof Reward(MSR)

MSR =119873

sum

119899=1

119887119899 (6)

To maximise MSR the spectrum aggregation problem can bedefined as a constrained optimisation problem as follows

max119886

119873

sum

119899=1

119887119899

(7)

subject to 119887119899= Δ sdot

C

sum

c=1

119886119899c

=

0 if 120601119899is rejected

119903119899

if 120601119899is accepted

for 1 le 119899 le 119873

(8)

F119867

119889119905minusF119871

119890119891le MAS (9)

119886119899c = 0

if 119897lowast119899c = 0 for 1 le 119899 le 119873 1 le c le C

(10)

119886119899c sdot 119886119896c = 0

if 119888lowast119899119896c = 1 for 1 le 119899 119896 le 119873 1 le c le C

(11)

Expression (8) assures that rewarded bandwidth 119887119899to each

accepted 120601119899must be equal to 120601

119899rsquos bandwidth demand 119903

119899 if

CM2M network cannot satisfy 120601119899rsquos bandwidth request 120601

119899is

rejected and 119887119899= 0 If F119871

119890119891(1 le 119890 le 120581

119891and 1 le 119891 le 119872) is

the lowest frequency of an initial aggregated subchannel andF119867119889119905

(1 le 119889 le 120581119905and 1 le t le 119872) is the highest frequency

of a terminative subchannel (9) guarantees that the rangeof allocated spectrum is equal to or less than MAS A mustsatisfy the interference constraints (10) and (11) expressions(10) and (11) guarantee that there is no harmful interferenceto LUs and other CM2M devices respectively

32 Spectrum Aggregation Algorithm Based on GeneticAlgorithm Traditionally the spectrum assignment problemhas been classified as an NP-hard problem [12] HereinGA is employed to solve the aggregation-based spectrumassignment problem in order to obtain faster convergenceGA is a stochastic search method that mimics the process ofnatural evolution In addition it is easy to encode solutionsof spectrum assignment problem to chromosomes in GAand compare the fitness value of each solution The specificoperations of the proposed algorithm referred to as MSRAlgorithm (MSRA) can be described through the followingsteps

(1) Encoding In MSRA a chromosome represents a pos-sible conflict-free subchannel assignment In order todecrease search space (by reducing redundancy in thedata) and obtain faster solutions similar approach asdescribed in [12] is adopted in this article We applya mapping process between A and the chromosomesbased on the characteristics of Llowast and Clowast Only thoseelements of A are encoded whose correspondingelements in Llowast take the value of 1 that is 119886

119899c = 0where (119899 c) satisfies 119897lowast

119899c = 0 As a result of thismapping the chromosome length is equal to thenumber of nonzero elements of Llowast and the searchspace is greatly reduced Based on a given Llowast lengthof the chromosome can be calculated assum119873

119894=1sum

C119895=1119897lowast

119894119895

(2) Initialisation During initialisation process the initialpopulation is randomly generated based on a binarycoding mechanism as applied in [12] The size of thepopulation depends on |Φ| and |Γ| for larger |Φ| and|Γ| population size should be increased where | sdot |indicates cardinality of a set


(3) Selection The fitness value of each individual ofthe current population according to MSRA criteriadefined in (6) is computed According to the indi-viduals fitness value excellent individuals are selectedand remain in the next generation The chromosomewith largest fitness value replaces the one with a smallfitness value by the selection process

(4) Genetic Operators To maintain high fitness valuesof all chromosomes in a successive population thecrossover and mutation operators are applied Tworandomly selected chromosomes are chosen in eachiteration as the parents and the crossover of theparent chromosomes is carried out at probability ofcrossover rate In addition to selection and crossoveroperations mutation at certain mutation rate is per-formed to maintain genetic diversity

(5) Termination The stop criteria of GA are checked ineach iteration If they can not be satisfied step (3)and step (4) are repeated The number of maximumiterations and the difference of fitness value are usedas the criteria to determine the termination of GA

The population of chromosomes generated after initiali-sation selection crossover and mutation may not satisfythe given constraints defined in (8)ndash(11) To find feasiblechromosomes that satisfy all constraints a constraint-freeprocess is applied that has the following steps (in order)

(1) Bandwidth Requirements The vector B as given inSection 22 is calculated 119887

119899should be equal to either

119903119899or zero otherwise all genomes related to 120601

119899are

changed to zero(2) MAS To satisfy the hardware limitations of the

transceiver expression (9) should be satisfied other-wise all genomes related to 120601

119899are changed to zero

(3) No Interference to LUs Expression (10) guarantees thatCM2M devices transmissions do not interfere LUstransmissions ensuring that CM2M network doesnot harm LUs performance If expression (10) is notsatisfied all genomes related to120601


(4) CCI Expression (11) guarantees that there is no harm-ful interference to other CM2M devices If expression(11) is not satisfied one of two conflicted devicesis chosen at random and then all genomes of theselected device are changed to zero

To achieve higher spectrum utilisation and faster conver-gence after each generation MSRA assigns all unassignedspectra to remaining CM2M devices randomly wheneverpossible At the same time MSRA guarantees that all theconstraints defined in (8)ndash(11) are satisfied at all time


In this section a set of system-level performance resultsare presented in order to compare and show the efficiencyof MSRA over MSA [10] AASAA [9] and RCAA Thesimulation results demonstrate high potential of the proposed

Table 1 Simulation parameters

Parameter ValueΔ 1MHzMAS 40MHzBW119898

Δ sdot 119880(1 20)

119903119899

Δ sdot 119880(1 20)

Total transmit power 26 dBm (400mW)Scheduling time slot 1msTraffic model BackloggedPopulation size 20Number of generations 10Mutation rate 001Crossover rate 08

method in terms of spectrum utilisation and system capacityTo assess the performance of network independent of eachdevicersquos traffic distribution model backlogged traffic model(known as full-buffer model) is used where packet queuelength of every device is much longer than what can bescheduled during each scheduling time slot

Due to the random nature of the channel bandwidth andthe devices bandwidth demand Monte Carlo simulationsare performed and each simulation scenario is repeated100000 timesThe default parameters used in the simulationsare listed in Table 1 where 119880(1 20) represents the discreteuniform random integer numbers between 1 and 20 Each ofthe channels is modeled as flat Rayleigh channel with pathloss model of PL = 1281 + 376 log

10119877 (119877 is in km) and

penetration loss of 20 dB The mean and standard deviationof log-normal fading are zero and 8 dB respectively Inour simulation model the CM2M devices located randomlywithout restrictions within a rectangular area of 2 kmtimes1 kmAll channels are randomly selected between 54MHz and806MHz television frequencies (channels 2ndash69) Typicallythe number of M2M devices is very high in each cell butin this study because of high computational complexityof SOTA solutions smaller number of M2M devices isconsidered for comparison purposes

To investigate the simulation results effectively the fol-lowing terms are defined and used in our analysis

(1) Spectrum Utilisation It is referred to as U which isdefined as the ratio of the sumof rewarded bandwidthto the sum of all available bandwidths that is

U =sum119873

119899=1119887119899

sum119872

119898=1BW119898

(12)

(2) Network Load It is referred to asLwhich is defined asthe ratio of the sum of all CM2M devices bandwidthrequirements to the sum of all available bandwidthsthat is

L =sum119873

119899=1119903119899

sum119872

119898=1BW119898

(13)

6 Mobile Information SystemsSp

ectr

um u

tilisa

tion

()

Network load

100

80

60

40

20

0

05 1 15 2 25 3 35 4 45

MSRAMSA

AASAARCAA

Figure 4 The impact of varying network load conditions onspectrum utilisation (scenario I without CCI)

(3) Number of Rejected Devices Rejected devices arethose machines that are not assigned any spectrum ina certain scheduling time slot

41 Scenario I Without CCI In this scenario the perfor-mance of MSRA is compared with the SOTA algorithmsincluding MSA [10] AASAA [9] and RCAA when CCIamong CM2M devices is not considered Therefore weassume that CM2M devices transmissions do not overlapwith the transmission of other CM2Mdevices using the samechannel

For 119872 = 30 L increases by increasing the number ofCM2M devices from 5 to 60 Figure 4 shows that when thenumber of CM2M devices increases the spectrum utilisationalso increases in all three methods but MSRA utilises allavailable whitespaces in various network loading conditionsmore efficiently than MSA AASAA and RCAA This canbe explained by the fact that in case of higher L networkcan allocate better segments of spectrum to users becauseof higher multiuser diversity In addition because of usingstochastic search method MSRA achieves near to optimumsolution in comparison to other SOTA solutions which arebased on approximate algorithms For MSRA when L ishigher than 3 CM2M network becomes saturated due tothe lack of available spectrum However for the rest of themethods there are still unassigned spectrum slices

42 Scenario IIWithCCI In this scenario CCI exists amongCM2M devices and we compare our algorithm MSRA withAASAA and RCAA As MSA inherently does not considerCCI for that reason we do not includeMSA for comparison

Spec

trum

util

isatio

n (

)

Network load

100

80

60

40

20

0

MSRAAASAARCAA

05 1 15 2 25 3 35 454 555

Figure 5 The impact of varying network load conditions onspectrum utilisation (scenario II with CCI)

Figure 5 shows the spectrum utilisation according to dif-ferent network loads by increasing the number of CM2Mdevices from 5 to 55 when there are only seven availablechannels (ie 119872 = 7) As shown in Figure 5 MSRAoutperforms AASAA and RCAA for different network loadsSimilar to Scenario I MSRA utilises TVWS even better thanprevious scenario because some CM2M devices in networkmay reuse spectrum that is used by other devices in CM2Mnetwork

Figure 6 represents the number of rejectedCM2Mdeviceswhen the network load increases The number of rejectedCM2M devices increases with the network load MSRA hasfewer numbers of rejected CM2M devices (or more satisfieddevices) than AASAA and RCAA of different network loadsMSRA optimises spectrum utilisation by admitting deviceswith better channel quality to the network and allocates thespectrum resources effectively Furthermore MSRA does notassign any spectrum resources to the devices that has leastcontribution to overall network throughput Figure 6 impliesthat MSRA increases the capacity of network (which is veryvital for M2M networks because of a very large number ofdevices) Our approach may starve some of devices whichare located far from the base station in our future work wewill optimise network performance based on proportionalfairness objective function to guarantee the fairness amongdevices

43 Convergence of MSRA Because of the nature of geneticprogramming it is arguably impossible to make formalguarantees about the number of fitness evaluations neededfor an algorithm to find an optimal solutionHowever hereincomputer experiments are performed to show the impact of


Network load05 1 15 2 25 3 35 454 555

MSRAAASAARCAA

Num

ber o

f rej

ecte

d de

vice

s

45

40

35

30

25

20

15

10

5

0

Figure 6 The impact of varying network load conditions on thenumber of rejected CM2M devices (scenario II with CCI)

Table 2 System parameters

Parameter Value119872 10119873 200Processor Intel Core i7-3667U 200GHzMemory (RAM) 4GBOS Windows 7 (64-bit)Simulator MATLAB R2011a (64-bit)

the number of generations on the performance of MSRAThe system parameters used in the section for simulation arelisted in Table 2 For the purpose of convergence studies weassume119873 = 200 and119872 = 10

Figure 7 shows the best fitness value (MSRA) for apopulation in a different number of generations As shown inFigure 7 the performance of algorithm is enhanced when thenumber of generations increases however this is at the costof increased processing time After roughly 34 generationsthe fitness value saturates at optimal value which shows theeffectiveness of using GA for spectrum assignment usingspectrum aggregation

Moreover Figure 8 illustrates distribution of processingtime for MSRA to find an optimal solution As shown inFigure 8 at 85 of time MSRA finds an optimum solution inless than scheduling time slot (1ms) and 15 takes more thanscheduling time slot Additionally MSRA can be optimisedto use fewer processor resources so that it can execute morerapidly

Furthermore Lobo et al [15] provided a theoreticaland empirical analysis of the time complexity of traditional

The b

est fi

tnes

s val

ue o

f MSR

A (M

Hz)

Number of generations

270

265

260

255

250

245

0 20 40 60 80 100

Figure 7 The impact of the number of generations on MSRAresults

Freq

uenc

y (

)

Convergence time (ms)

tclt1

1lttclt2

2lttclt3

3lttclt4

4lttc

100

80

60

40

20

0

Figure 8 Distribution of processing time for MSRA to find anoptimal solution

simple GAs According to [15] GA has time complexitiesof O(sum119873

119894=1sum

C119895=1119897lowast

119894119895) which is dependent on length of each

chromosome The linear time complexity for GA occursbecause the population sizing grows with the square root ofchromosome length The time complexity presented hereinis for the worst-case scenario when the population size isassumed to be fixed and maximum of rest of generations


5 Conclusion

This article introduces an aggregation-aware spectrumassignment algorithm using genetic algorithmThe proposedalgorithm maximises the spectrum utilisation to CM2Mdevices as a criterion to realise spectrum assignment More-over the introduced algorithm takes into account the real-istic constraints of co-channel interference and MaximumAggregation Span Performance of the proposed algorithmis validated by simulations and results are compared withalgorithms available in the literatureThe proposed algorithmdecreases the number of rejected devices and improvesthe spectrum utilisation of CM2M network Our algorithmincreases the capacity of network which is very vital forM2Mnetworks For future work we will investigate the impact ofthe various parameters used in genetic algorithm to solvethe introduced utilisation function in particular populationsize crossover rate and mutation rate are the parametersthat will be investigated in our study in addition we willfurther work on developing genetic algorithm based methodto assign spectrum to CM2M devices in an energy-efficientmanner

Competing Interests

The authors declare that they have no competing interests

References

[1] R Lu X Li X Liang X Shen and X Lin ldquoGRS thegreen reliability and security of emerging machine to machinecommunicationsrdquo IEEE Communications Magazine vol 49 no4 pp 28ndash35 2011

[2] ldquoCisco visual networking index Global mobile data trafficforecast update 2014ndash2019 white paperrdquo 2015 httpwwwciscocomcenussolutionscollateralservice-providervisual-net-working-index-vnimobile-white-paper-c11-520862html

[3] S Rostami K Arshad and K Moessner ldquoOrder-statistic basedspectrum sensing for cognitive radiordquo IEEE CommunicationsLetters vol 16 no 5 pp 592ndash595 2012

[4] Y Zhang R Yu M Nekovee Y Liu S Xie and S GjessingldquoCognitive machine-to-machine communications visions andpotentials for the smart gridrdquo IEEE Network vol 26 no 3 pp6ndash13 2012

[5] M Wylie-Green ldquoDynamic spectrum sensing by multibandOFDM radio for interference mitigationrdquo in Proceedings of the1st IEEE International Symposium on New Frontiers in DynamicSpectrum Access Networks (DySPAN rsquo05) pp 619ndash625 IEEEBaltimore Md USA November 2005

[6] J D Poston and W D Horne ldquoDiscontiguous OFDM consid-erations for dynamic spectrum access in idle TV channelsrdquo inProceedings of the 1st IEEE International Symposium on NewFrontiers in Dynamic Spectrum Access Networks (DySPAN rsquo05)pp 607ndash610 Baltimore Md USA November 2005

[7] R Rajbanshi A M Wyglinski and G J Minden ldquoAn effi-cient implementation of NC-OFDM transceivers for cognitiveradiosrdquo in Proceedings of the 1st International Conference onCognitive Radio Oriented Wireless Networks and Communica-tions (CROWNCOM rsquo06) pp 1ndash5Mykonos Island Greece June2006

[8] 3GPP ldquoLTE evolved universal terrestrial radio access (e-utra)physical layer proceduresrdquo Tech Rep 3GPP TS 36213 version1010 Release 10 3GPP 2010 httpwww3gpporg

[9] D Chen Q Zhang and W Jia ldquoAggregation aware spectrumassignment in cognitive ad-hoc networksrdquo in Proceedings ofthe 3rd International Conference on Cognitive Radio OrientedWireless Networks and Communications (CrownCom rsquo08) pp 1ndash6 May 2008

[10] F Huang W Wang H Luo G Yu and Z Zhang ldquoPrediction-based Spectrum aggregation with hardware limitation in cog-nitive radio networksrdquo in Proceedings of the IEEE 71st VehicularTechnology Conference (VTC rsquo10) pp 1ndash5 May 2010

[11] F Ye R Yang and Y Li ldquoGenetic algorithm based spectrumassignment model in cognitive radio networksrdquo in Proceedingsof the 2nd International Conference on Information Engineeringand Computer Science (ICIECS rsquo10) pp 1ndash4 Wuhan ChinaDecember 2010

[12] Z Zhao Z Peng S Zheng and J Shang ldquoCognitive radio spec-trum allocation using evolutionary algorithmsrdquo IEEE Transac-tions on Wireless Communications vol 8 no 9 pp 4421ndash44252009

[13] K Arshad M A Imran and K Moessner ldquoCollaborativespectrum sensing optimisation algorithms for cognitive radionetworksrdquo International Journal of Digital Multimedia Broad-casting vol 2010 Article ID 424036 20 pages 2010

[14] Y Li L Zhao C Wang A Daneshmand and Q Hu ldquoAggre-gation-based spectrum allocation algorithm in cognitive radionetworksrdquo in Proceedings of the IEEE Network Operations andManagement Symposium (NOMS rsquo12) pp 506ndash509 IEEEMauiHawaii USA April 2012

[15] F G Lobo D E Goldberg and M Pelikan ldquoTime complexityof genetic algorithms on exponentially scaled problemsrdquo inProceedings of the Genetic and Evolutionary Computation Con-ference (GECCO rsquo00) pp 151ndash158 Morgan-Kaufmann 2000

Research ArticleA Survey of the DVB-T Spectrum Opportunities forCognitive Mobile Users

Laacuteszloacute Csurgai-Horvaacuteth Istvaacuten Rieger and Joacutezsef Kerteacutesz

Department of Broadband Infocommunications and Electromagnetic Theory Budapest University of Technology and EconomicsEgry J Street 18 Budapest 1111 Hungary

Correspondence should be addressed to Laszlo Csurgai-Horvath csurgaihvtbmehu

Received 18 February 2016 Revised 30 May 2016 Accepted 5 July 2016

Academic Editor Janne Lehtomaki

Copyright copy 2016 Laszlo Csurgai-Horvath et alThis is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

Cognitive radio (CR) systems are designed to utilize the available radio spectrum in an efficient and intelligent manner TerrestrialDigital Video Broadcasting (DVB-T) frequency bands are one of the future candidates for cognitive radio applications especiallybecause after digital television transition the TV white spaces (TVWS) became available for radio communication This paperdeals with the survey of the DVB-T spectrum wideband measurements were performed on mobile platform in order to studythe variation of the radio signal power in city area aboard a moving vehicle The measurement environment was a densely built-inregionwhere the properDVB-T receivingwas guaranteed by threeTV transmitters utilizing three central channel frequencies using610 746 and 770MHz In our paper the methods the applied antenna and measurement devices will be presented together withsimulated andmeasured fading statisticsThe final result is an estimation of the cognitive DVB-T spectrum utilization opportunityfurthermore a scenario is also proposed for secondary channel usage

1 Introduction

Cognitive radio is an emerging technology to utilize theradio spectrum with high efficiency The main owners ofthe spectrum the primary users (PUs) are not constrainedduring their operation while the secondary users (SUs)can operate in the same frequency band if the spectrumis free [1] It is very important to avoid the degrading ofPUrsquos quality of service (QoS) during the cognitive channelusage whereas an acceptable level of service should also beprovided for the secondary users Several technologies shouldbe applied to guarantee thesemdashsometimes contradictorymdashrequirements [2] Sensing of the spectrum and detectingthe available channels are some of the main tasks of a CRsystem The frequency range that can be utilized by theCR devices depends on the local frequency regulation andtherefore it may vary in different countries In the crowdedradio spectrum it is not a simple task to find the appropriateradio bands for cognitive terrestrial devices [3 4] This paperconcentrates on the terrestrial television bands and theirsecondary usage

In the literature numerous works are presented aboutspectrum measurements and on different technologies to

support cognitive users in better utilization of the availablebandwidth TV white space is also of a great interest due tothe digital TV transition that recently took place in severalcountries In the following an overview of this research fieldwill be given in order to put our research into context

In [5] despite the actual theory that the capacity of theradio spectrum is already achieved the underutilization ofthe spectrum is highlighted and the importance of cognitiveradio techniques is shown The paper is focusing on majortechnologies for opportunistic spectrum access through ahierarchical model approach that adopts the primary andsecondary user structure Spectrum sensing is the key tech-nology to estimating the availability of the licensed spectrumfor secondary usage In [6] the various spectrum occupancymodels used in different research campaigns worldwide werestudied and compared The authors evaluate the percentageof the whole spectrum occupied by different services Long-and short-term statistics are presented showing most of thecommercial terrestrial frequency bands (GSM TV broad-casting 3G etc) utilizing the available spectrum almostbelow 20ndash40 The experiments have been conducted invarious locations such as US Europe New Zealand South



Africa China Singapore and Vietnam A similar study wasperformed in Chicago New York Washington DC and afew rural locations in 2005 between 30 and 3000MHz [7] Ina large business like Chicago low spectrum occupancy wasobserved indicating that a DSS (Dynamic Spectrum Sharing)radio system could access a huge amount of prime spec-trum as there are large unoccupied contiguous spectrumblocks The paper [8] collects previous research work carriedout worldwide and compares it with spectrum occupancymeasurements at the University of Hull UK The collectedhistorical measurements are covering also the 30ndash3000MHzband and they confirmed the generally low occupancy ofthe investigated spectrum The measurements in the UKwere performed with a similar hardware configuration towhat we also applied during our research work and willbe detailed later (spectrum analyser and computer) thefrequency range was 80ndash2700MHz For DVB-T spectrummeasurements in [9] several results can be found especiallyfor occupancy estimations serving as input for outdoor REM(Radio Environment Maps) The measurement setup wassimilar to the campaign performed in Budapest but the latterresearch is focusing also on fade duration statistics and itsconsequences as it will be later demonstrated The cellularand theUHFVHFTV bandwere studied in [10] forMalaysiaand actual spectrum utilization statistics are provided withstatic measurements The low duty cycle of the spectrumoccupancy was also proved by this study A comparativespectrum occupancy study was carried out in BarcelonaSpain andPoznan Poland [11]Themeasurement setupswereharmonized to obtain comparable results by concentratingon the problem of the efficient noise floor estimation Asa result differences have been obtained in the TETRAbands due to the different spectrum allocation regulations inthese countries This study highlights that efficient spectrumdetection is always required in order to avoid the congestionsdue to different local regulatory rules The change of theUHF TV band spectrum availability due to digital transitionin Greece is studied in [12] They proved that the spectrumavailability was significantly increased after the analogueswitch-off Furthermore the risk of LTE-4G interference toTV services and vice versa is also pointed out accordingto the spectrum measurements they carried out A generaland detailed discussion on different approaches to spectrumoccupancy measurements is provided in the relating ITUreport SM2256 [13] Unlicensed communication in the UHFband has also a great actuality Measurements in Italy Spainand Romania are presented in [14 15] in order to estimatepractical parameters to ensure the feasible and harmlessunlicensed communication in the UHF TV bands Specialdevices like wireless microphones may also utilize this bandunder strict regulatory control [16] that is also increasing theimportance of accurate spectrum sensing methods

In the present paper we demonstrate mobile measure-ments in the DVB-T spectrum by concentrating on theoccupancy statistics that can be inferred from the channelfading dynamicsWe significantly extended our former paper[17] with technical details and additional measurement routefurthermore results and conclusions are amended

SU route

Cognitive spectrum usage PU3

PU1

PU2

Figure 1 Fixed PUs and a moving SU for smart DVB-T spectrumutilization

DVB-T users are the primary owners of the televisionreceivers [18 19] In large cities like Budapest where weconducted our measurements the sufficient service requiresseveral multiplexed channels and usually more than onetransmit station DVB-T receivers are the primary users ofthis spectrum and the service provider takes care of thesufficient quality of service at the whole geographical region[20] Nevertheless in densely built-in areas and especiallyin case of hilly areas the received signal level could belocally insufficient to receive the DVB-T signal properly Inthis case by applying smart spectrum sensing technologies asecondarymobile user has an opportunity to utilize this spec-trum for different kind of short-distance communicationslike accessing locally transmitted traffic information and car-to-car communications or for general type of data transferA hypothetical scenario is depicted in Figure 1

Therefore our main goal during this survey was to inves-tigate the frequency band of the terrestrial digital televisionbroadcasting between 400 and 900MHz to have an overviewof the possibilities formobile CR applications [21] In order toachieve this goal the appropriate measurement devices hadto be selected and also designed if off-the-shelf equipmentwas not available The air interface was a custom designedwide band discone antenna For sensing the radio spectruma handheld spectrum analyser was applied As the mea-surement campaign was planned for mobile measurementsaboard a vehicle an appropriate and safe mechanical setupwas needed The route and the speed of movement wererecorded by a GPS-based navigation system

The main target of this research was twofold primarilyreceived power time series was recorded in a wide DVB-Tband while a vehicle was moving in city area Secondly byprocessing the measured data first- and second-order statis-tics were derived allowing inferring the CR opportunities inthis band

2 Measurement Location and Modelling

In the time of the measurements (122013 and 032014) inBudapest three DVB-T transmitters were operating Eachof them has multiplex channels with the standard 8MHzbandwidth providing the sufficient receiving conditions overthe whole city It is worthy of note that in the majority of the


Table 1 DVB-T transmitters in Budapest

UHF channels [MHz] Max ERP [kWdBm]CH Starting Centre Ending Szechenyi Hill 1 Harmashatar Hill 2 Szava Street 338 606 610 614 10080 95698 6267955 742 746 750 39876 9870 7168558 766 770 774 10080 74687 56675

Location LatLonASL 47∘29101584018∘581015840457m 47∘33101584019∘00443m 47∘28101584019∘071015840120m

1

2

3

Figure 2 DVB-T transmitters in Budapest (map source Google)

European countries the transition from analogue to digitalTV broadcasting technologies was finished (see for example[22]) and there are only a few countries where this is still anongoing process

In Table 1 the main transmitter parameters can be foundfor Budapest

The transmitter locations are depicted in the map shownin Figure 2 denoted with 1 2 and 3 signs It is worthmentioning that the left side of the city is hilly while the rightside is flat however transmitter 3 can be found on elevatedlocationThe arrangement of the transmitters and their powerradiated ensure the location-independent receiving despitethe geographical variability

For a first and rough estimation of the received signalpower at the different geographical positions the Okumura-Hata channel model [23] was selected to illustrate the capa-bilities and limitations of such calculations This model isvalid for 150ndash1500MHz frequency range therefore it is wellapplicable for DVB-T It is an empirical model suitable tocalculate the path loss 119871

119880for different urban areas The ℎ

119879

height of the transmit antenna and the ℎ119877receiver antenna

height are also input parameters of the model

119871119880= 6955 + 2616 log

10

119891[MHz]minus 1382 log

10

ℎ119879minus 119862119867

+ [449 minus 655 log10

ℎ119879] log10

119863[km]

(1)

119862119867is the antenna height coefficient and it is for small and

medium cities

119862119867= 08 + (11 log

10

119891[MHz]minus 07) ℎ

119877

minus 156 log10

119891[MHz]

(2)

and for big cities

119862119867

=

829 log10

(154ℎ119877)2

minus 11 150 le 119891[MHz]le 200

32 log10

(1175ℎ119877)2

minus 497 200 le 119891[MHz]le 1500

(3)

The model has limitations in range (1ndash20 km) and trans-mitter antenna height (30ndash200m) By taking into accountthat the sea level height of the city (river floor) is 90m themodel could be applied for a rough estimation of the receivedsignal level In the following this calculation is presentedwhere we considered big city model coefficients and providereceived signal power map for each transmitter frequency

To calculate with the Okumura-Hata model we posi-tioned three transmitters into a hypothetical square of 20 lowast20 km the origin of this area was N47∘251015840 and E18∘541015840The positions of the transmitters are representing their realgeographical places relatively to this origin The gain of thetransmitter antennas was selected uniformly 15 dB and thereceiver location was 3m respectively The result is depictedin Figure 3 where the transmitters are numbered accordingto Table 1

The modelled signal level in the rectangular area visu-alizes the received power at different locations produced bythe DVB-T transmitters Besides the Okumura-Hata modelthe Walfisch-Ikegami and the Lee models are compared andtested for different geographical areas in [24] In this paperthe goal of the modelling was to get a quantitative overviewof the received signal power field and therefore we selectedfor our calculations one of the best known models

Nevertheless the effect of the local variation of the envi-ronment for example shadowing of buildings reflectionsand local interferences is not visible in Figure 3 In order togenerate a more accurate power map a detailed geolocationmap would be required containing an exact database of theobject positions and dimensions across the city but such adatabase was not available for the authors

The lack of the fine structure and the variation of thesignal level on a specific route require a different approachThe description of this method and its conclusions is thefollowing subject of this paper


0 5 10 15 200

5

10

15

20

(dBm)

2

1

3

y(k

m)

x (km)

minus55 minus50 minus45 minus40 minus35 minus30 minus25

(a)

0

5

10

15

20

1

2

3

y(k

m)

0 5 10 15 20x (km)

(dBm)minus55 minus50 minus45 minus40 minus35 minus30 minus25

(b)

0 5 10 15 200

5

10

15

20

1

2

3

y(k

m)

x (km)


(c)

Figure 3 DVB-T signal power at 610MHz (a) 746MHz (b) and 770MHz (c) calculated with Okumura-Hata model

3 Receiver Antenna Design forSpectrum Sensing

Our goal was to build an all-purpose system that is capableof wide range spectral observations between 04 and 3GHzIn [25] for a similar measurement a commercially available25ndash1300MHz antennawas proposed but for our purposes weselected a customized antenna that has a broader bandwidthTherefore a special wideband antenna was designed [26] at

our department whose omnidirectional characteristic wasone of the most important requests (see Figure 4)

The requirements are well fulfilled by a discone antennathat consists of a flat disc on the top of a conical part Withinthis structure the wideband operation is mainly determinedby the conical structure The drawing and final dimensionsof the antenna can be found in Figure 4 Before antennafabrication computer simulations were done in order toprove the performance and check the main parameters


Main antenna dimensions

Cone max diameter 210mm

Cone angle 60∘

Disc diameter 150mm

Total height (wo connector) 180mm

Feed pinDisc

Copper cone Teflon holder

Cone

Coax cable

N connector

Figure 4 Antenna dimensions and simulated characteristics at 746MHz

05 1 15 2 25 3

0

2

Frequency (GHz)

Gai

n (d

Bi)

minus2

minus4

minus6

Figure 5 Simulated antenna gain and a two-channel measurement setup

The simulated antenna of a characteristic at 746MHzis depicted in Figure 4 while variation of the gain withfrequency is depicted in Figure 5 The latter figure alsoillustrates a two-antenna system assembled on the top of acar ready for mobile measurements The gain of the antennais slightly varying with the frequency and according tothe simulation it is nearly 2 dB in the investigated DVB-Tfrequency band

4 Mobile Sensing of the DVB-T Spectrum

Spectrum sensing is a secondary userrsquos task when his opera-tion is based on CR technology SUs should discover usually

a wide frequency band before they can utilize any spectraThis is an indispensable process because the main ownersof the spectrum the Pus cannot be disturbed or restrictedin their operation The air interface of this kind of sensing isusually a wideband and omnidirectional antenna Widebandsensing requires intelligent programmable received signaldetection that allows scanning the selected frequency rangeand performing fast energy detection at the single frequen-cies During our work we applied professional measurementdevices for similar purposes in order to explore the DVB-T spectrum in a larger geographical area The measurementcould be a base to qualify the DVB-T spectrum for mobilecognitive radio applications


GPS Spectrumanalyser

Figure 6 Mobile spectrum measurement setup

This section provides the detailed measurement setup forour experiments and then time series and different statisticswill be presented

In Section 2 we have seen that the modelled receivedsignal map especially in absence of a geolocation databaseof terrestrial objects cannot provide sufficient informationabout the local variability of the signal level In order toinvestigate the exact time series of the DVB-T signal poweraboard a moving vehicle a measurement with location-tagging was designed and conducted As spectrum sensingdevice a type of Agilent N9340B Handheld RF spectrumanalyser was utilized For our research purposes the flexibil-ity and precision of such ameasurement tool were an obvioussolutionThe investigated frequency band is supported by theapplied device [27] and its built-in memory was able to storethe measurement data through the whole route

Themeasurement setup for the mobile system is depictedin Figure 6 and it has the following main blocks

(i) A car equipped with a single discone antenna (seeSection 3)

(ii) A GPS device to record the route and the movingspeed (Mitac P560 PDA)

(iii) A portable spectrum analyser [27] with data storagecapability (Agilent N9340B)

(iv) A notebook to archive measurement files

To have a first look of the measured data a waterfalldiagram is a good opportunity (see Figure 8) depicting thereceived signal power in the complete frequency band for thetotal measurement period

In order to survey the DVB-T frequency band duringmovement two measurements were conducted in the cityarea of Budapest The routes are depicted in Figure 7 alsodenoting their length and duration

In order to cover the whole frequency band of the TVtransmitters the following spectrum analyser settings wereapplied

(i) Starting frequency 590MHz(ii) Stop frequency 800MHz(iii) Span 210MHz(iv) Span time 2 sec(v) Attenuation 10 dB

(vi) Bandwidth 100 kHz(vii) Reference noise power minus109 dBm

10 dB attenuation was required to keep the measuredsignal level within the analysermeasurement rangeThe 590ndash800MHz frequency band was sensed with 1022MHz stepsthus for example for a 8MHz DVB-T channel 176 sampleswere collected The spectrum analyser stores the measuredreceived power in floating point data type with two decimalplaces The antenna was connected with RG-58 type cable of3m length therefore the cable attenuation was 09 dB

TV transmitters 1 and 3 were closed by the routes(their places are marked on the maps) The speed of the carwas slightly varying but it was kept during the route as stableas possible

After processing the measurements the spectrogram andthe time series of the received power for three TV channelsare providing the first overview of the investigated spectrumIn the spectrogram and even more clearly in the receivedpower time series the strong variations of the signal levelsare well observable (Figures 8-9)

The results are indicating that the conditions of properDVB-T receiving do not always exist As the measurementwas performed in densely built-in city area and we con-sidered the movement of the car different type of channelimpairments may arise The shadowing interference andmultipath propagation could decrease the quality of serviceHowever the Okumura-Hata propagation model is a well-known tool to calculate the received signal level in built-inareas [28 29] this is a general model and cannot substitutethe real measurements like the present one allowing derivinga more accurate characterization of the mobile propagationchannel For proper DVB-T receiving primary users require50 dB120583V signal level or considering a 50Ω termination from(4) this level is minus57 dBm [30]

RPmindBm= RPmin

dB120583Vminus 90 minus 20 log (radic119885Ω)

= minus57 dBm(4)

More detailed discussion about the planning of DVB-Tservice area and the minimum field strength requirementscan be found in [31]

We will apply this threshold as an opportunity indicatorfor secondary channel usage On the other hand it shouldbe also considered that in order to minimise the harmfulinterference caused by the cognitive secondary user devicesthe TV signal sensing margin should be much lower thanthat of TV receivers required for high quality receiving [32]The hidden node problem when a primary user with goodreceiving conditions is interfered by a secondary transmittingdevice [33] is one of the reasons that cognitive devices areusually operating with lower sensing margin Neverthelessthis kind of problem is beyond the scope of this paperthe abovementioned minus57 dBm will be for us the measureof the local DVB-T signal quality As the goal of thispaper is a survey of the TVWS the investigation of somestatistical properties of the received signal time series willlead to the estimation of the secondary channel utilization


3

(a)

1

(b)

Figure 7 (a) Route 1 (229 km 58min 122013) (b) Route 2 (349 km 588min 032014) (map sources Google)

Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

010

0

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

0 10 20 30 40 50 60Time (min)minus10

minus20

minus30

minus40

minus50

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus60

minus70

minus80

minus90

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz

Figure 8 Spectrogram and received power time series at TV channel centre frequencies (Route 1)

opportunities We emphasize that for an operational cog-nitive radio application a lower sensing margin should berequired Furthermore especially to avoid the interferenceadditional techniques would be also desirable for examplepilot detection cyclostationary feature detection or cyclicprefix and autocorrelation detection [32]

To find the probability of the minimal received signallevel the Cumulative Distribution Function (CDF) of theattenuation could help To estimate a realistic receivingcondition an increased antenna gain should be appliedbecause the discone antenna is only an experimental deviceand it does not represent correctly the antenna of a standardDVB-T receiverThe applied discone antenna has sim2 dB gainnevertheless for real DVB-T receiving an antenna with 10ndash12 dB gain is recommended [34] and usually applied by PUs

The CDF of the received power indicates the probabilitythat the signal level is less than or equal to a certain value as itis depicted in Figure 10 for the two different routes If we take

into account that a standard PU has a receiving antenna withan additional 10 dB gain compared to the discone antenna inthe measurement according to (4) the probability values atminus57 minus 10 = minus67 dB are representing the thresholds of theimproper receiving conditions

One can see that the probability of insufficient DVB-T signal level is relatively high in Figure 10 these valuesare indicated for each channel Contrarily in case of thiscondition the spectrum could be utilized by the secondaryusers for their own purposes by applying CR technologies

Another aspect of the estimation of the channel impair-ment is the fade duration statistics [35]While the attenuationstatistics inform us about the probability that the fadingdepth exceeds a specified level the length of the individualfade events and thus the possible outage periods could bedetermined only from the fade duration distribution Theduration of fades can be calculated from the attenuation timeseries therefore the received power time series (see Figures 8


Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

0

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus40

minus50

minus60

minus70

minus80

minus90

0 10 20 30 40 50 60Time (min)

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)


0

01

02

03

04

05

06

07

08

09

1

Received power (dBm)

Prob

abili

ty

Route 1

Improper receiving conditions probabilities

minus20minus30minus40minus50minus60minus70minus80minus90

At 610MHz 008At 746MHz 022At 770MHz 015

610MHz 746MHz770MHz

0

01

02

03

04

05

06

07

08

09

1

Prob

abili

ty

Route 2

Received power (dBm)minus40minus50minus60minus70minus80minus90

Improper receiving conditions probabilities At 610MHz 038At 746MHz 066At 770MHz 044

610MHz 746MHz770MHz

Figure 10 CDF of received power and probabilities of improper receiving conditions

and 9) should be converted For this conversion the highestmeasured received power value in the DVB-T channel wasconsidered as a reference (zero attenuation) level

Besides the fade duration in cognitive radio applicationsthe level crossing rate as another dynamics aspect of thechannel is studied in [36] for Rayleigh and Rician fastfading channels The effect of imperfections in the radioenvironment map (REM) information on the performance

of cognitive radio (CR) systems was investigated in [37] Inopportunistic channel allocation algorithms [38] the durationof fade event may play an important role Therefore inour paper we propose fade duration statistics as a tool foropportunity length estimation

Figure 11 indicates the probability of fade durations at15 dB and 20 dB attenuation levels for 10 and 60 secondsrespectively We proved with our measurements and with the


Time (sec)

Prob

abili

tyRoute 1 Route 2

100

100

10minus1

10minus2

Prob

abili

ty

100

10minus1

10minus2

15dB20dB25dB

30dB35dB

15dB20dB25dB

30dB35dB

101 102

Time (sec)100 101 102

012 (D = 10 sec)002 (D = 60 sec)017 (D = 10 sec)003 (D = 60 sec)017 (D = 10 sec)003 (D = 60 sec)

610MHz

746MHz

770MHz



610MHz

746MHz

770MHz


Frequency FrequencyP (d gt D) | Th = 15dB P (d gt D) | Th = 20dB P (d gt D) | Th = 15dB P (d gt D) | Th = 20dB

Figure 11 Fade duration distribution of the 610MHz channel and probabilities of 10 and 60 sec fade events (all channels)

relating fade duration statistics that aboard a moving devicein city area the DVB-T spectrum can be used for secondarypurposes even for several seconds or for a minute durationCalculating with one-hour travelling the opportunity forsecondary channel usage during this journey is severalminutes in 10 s quanta and even some complete minutesThese are significant values that should be taken into accountif secondary channel utilization of the DVB-T spectra isplanned

For the calculations above we appliedminus57 dBm thresholdthat is according to the literature the signal level requiredfor the error-free DVB-T reception Our proposal is that thesecondary usage of the spectrum is a reality when the servicequality is insufficient for the primary users Contrarily forcognitive radio applications the protection of primary userrsquosservice quality is a key issue The appearance of secondaryusers may cause significant interference in the TVWS there-fore an advanced spectrum sensing technique is essential Astudy about this emerging technology [39] discusses that thesensing threshold is minus1128 dBm for 8MHz wide channelsshowing that high quality sensing technique is inevitable ina real CR application

5 Conclusions

In this paper we presented wideband mobile DVB-T spec-trum measurements to study the variation of the received

signal power in the TV channel frequencies Our suggestionis that for cognitive radio applications the same frequencyband is applicable if the service quality for the PUs is insuf-ficient It may happen in densely built-in city areas that dueto shadowing reflections or interference the DVB-T signalquality is improper for primary usage This fact has beenproved by the measurements In this case of short-distancecommunications for example for car-to-car data transfer oraccess local traffic information databases or even for self-driving vehicles the DVB-T spectrum could be utilized Inthe paper the antenna design for spectrum detection theapplied spectrum sensing hardware measurement methodsand their statistics were shown After the evaluation of theresults it was proven that for mobile CR users it is possible toutilize the DVB-T band with intelligent devices for secondarypurposes even without decreasing the QoS of the primaryusers

Competing Interests


References

[1] I F Akyildiz W-Y Lee M C Vuran and S Mohanty ldquoNeXtgenerationdynamic spectrum accesscognitive radio wirelessnetworks a surveyrdquo Computer Networks vol 50 no 13 pp2127ndash2159 2006


[2] O Simeone J Gambini Y Bar-Ness and U SpagnolinildquoCooperation and cognitive radiordquo in Proceedings of the IEEEInternational Conference on Communications (ICC rsquo07) pp6511ndash6515 Glasgow UK June 2007

[3] E Axell G Leus and E G Larsson ldquoOverview of spectrumsensing for cognitive radiordquo in Proceedings of the 2nd Interna-tional Workshop on Cognitive Information Processing (CIP rsquo10)pp 322ndash327 Elba Italy June 2010

[4] A Garhwal and P P Bhattacharya ldquoA survey on spectrumsensing techniques in cognitive radiordquo International Journal ofComputer Science and Communication Networks vol 1 no 2pp 196ndash206 2011

[5] Q Zhao and B M Sadler ldquoA survey of dynamic spectrumaccessrdquo IEEE Signal Processing Magazine vol 24 no 3 pp 79ndash89 2007

[6] D Das and S Das ldquoA survey on spectrum occupancy measure-ment for cognitive radiordquo Wireless Personal Communicationsvol 85 no 4 pp 2581ndash2598 2015

[7] M A McHenry P A Tenhula D McCloskey D A Robersonand C S Hood ldquoChicago spectrum occupancy measurementsamp analysis and a long-term studies proposalrdquo in Proceedingsof the 1st International Workshop on Technology and Policy forAccessing Spectrum (TAPAS rsquo06) article 1 ACM Boston MassUSA 2006

[8] M Mehdawi N Riley M Ammar and M Zolfaghari ldquoCom-paring historical and current spectrum occupancy measure-ments in the context of cognitive radiordquo in Proceedings of the20th Telecommunications Forum (TELFOR rsquo12) pp 623ndash626Belgrade Serbia November 2012

[9] A Kliks P Kryszkiewicz K Cichon A Umbert J Perez-Romero and F Casadevall ldquoDVB-T channels measurementsfor the deployment of outdoor REM databasesrdquo Journal ofTelecommunications and Information Technology no 3 pp 42ndash52 2014

[10] S Jayavalan H Hafizal N M Aripin et al ldquoMeasurements andanalysis of spectrum occupancy in the cellular and TV bandsrdquoLecture Notes on Software Engineering vol 2 no 2 pp 133ndash1382014

[11] A Kliks P Kryszkiewicz J Perez-Romero A Umbert andF Casadevall ldquoSpectrum occupancy in big cities-comparativestudy Measurement campaigns in Barcelona and Poznanrdquo inProceedings of the 10th International Symposium on WirelessCommunication Systems (ISWCS rsquo13) pp 1ndash5 Ilmenau Ger-many August 2013

[12] P I Lazaridis S Kasampalis Z D Zaharis et al ldquoUHFTVbandspectrum and field-strength measurements before and afteranalogue switch-offrdquo in Proceedings of the 2014 4th InternationalConference on Wireless Communications Vehicular Technol-ogy Information Theory and Aerospace and Electronic Systems(VITAE rsquo14) pp 1ndash5 Aalborg Denmark May 2014

[13] ITU-R ldquoSpectrum occupancy measurements and evaluationrdquoReport ITU-R SM2256 2012

[14] P AngueiraM Fadda JMorgadeMMurroni andV PopesculdquoField measurements for practical unlicensed communicationin the UHF bandrdquo Telecommunication Systems vol 61 no 3 pp443ndash449 2016

[15] M Fadda V PopescuMMurroni P Angueira and JMorgadeldquoOn the feasibility of unlicensed communications in the TVwhite space field measurements in the UHF bandrdquo Interna-tional Journal of Digital Multimedia Broadcasting vol 2015Article ID 319387 8 pages 2015

[16] Federal Communications Commission ldquoSpectrum access forwireless microphone operationsrdquo FCC Record FCC-14-145Federal Communications Commission 2014

[17] L Csurgai-Horvath I Rieger and J Kertesz ldquoMobile accessof the DVB-T channel and the opportunity for cognitivespectrum utilizationrdquo in Proceedings of the 17th InternationalConference on Transparent Optical Networks (ICTON rsquo15) pp1ndash4 Budapest Hungary July 2015

[18] W Van den Broeck and J Pierson Digital Television in EuropeVUBpress Brussels Belgium 2008

[19] U Reimers DVB The Family of International Standards forDigital Video Broadcasting Springer Berlin Germany 2004

[20] D Noguet R Datta P H Lehne M Gautier and G FettweisldquoTVWS regulation and QoSMOS requirementsrdquo in Proceedingsof the 2nd International Conference onWireless CommunicationVehicular Technology Information Theory and Aerospace ampElectronic Systems Technology (Wireless VITAE rsquo11) pp 1ndash5Chennai India February 2011

[21] B Wild and K Ramchandran ldquoDetecting primary receiversfor cognitive radio applicationsrdquo in Proceedings of the 1stIEEE International Symposium on New Frontiers in DynamicSpectrum Access Networks (DySPAN rsquo05) pp 124ndash130 IEEEBaltimore Md USA November 2005

[22] R A Saeed and S J Shellhammer Eds TV White Space Spec-trum Technologies Regulations Standards and ApplicationsCRC Press New York NY USA 2012

[23] MHata ldquoEmpirical formula for propagation loss in landmobileradio servicesrdquo IEEE Transactions on Vehicular Technology vol29 no 3 pp 317ndash325 1980

[24] P M Ghosh Md A Hossain A F M Zainul Abadin and KK Karmakar ldquoComparison among different large scale pathloss models for high sites in urban suburban and rural areasrdquoInternational Journal of Soft Computing and Engineering vol 2no 2 2012

[25] A Martian C Vladeanu I Marcu and I Marghescu ldquoEval-uation of spectrum occupancy in an urban environment in acognitive radio contextrdquo International Journal on Advances inTelecommunications vol 3 no 3-4 2010

[26] K-H Kim J-U Kim and S-O Park ldquoAn ultrawide-banddouble discone antenna with the tapered cylindrical wiresrdquoIEEE Transactions on Antennas and Propagation vol 53 no 10pp 3403ndash3406 2005

[27] Agilent N9340B Handheld RF Spectrum Analyzer (HSA) 3GHz User Manual

[28] ITU ldquoPredictionmethods for the terrestrial landmobile servicein the VHF andUHF bandsrdquo ITU-R Recommendation P 529-2ITU Geneva Switzerland 1995

[29] A Medeisis and A Kajackas ldquoOn the use of the universalOkumura-Hata propagation prediction model in rural areasrdquoin Proceedings of the IEEE 51st Vehicular Technology ConferenceProceedings vol 3 pp 1815ndash1818 Tokyo Japan May 2000

[30] ROVER Laboratories SpA ldquoUnderstanding Digital TVrdquo 2013httpwwwroverinstrumentscom

[31] E P J Tozer Broadcast Engineerrsquos Reference Book Taylor ampFrancis London UK 2012

[32] M Nekovee ldquoA survey of cognitive radio access to TV whitespacesrdquo International Journal of Digital Multimedia Broadcast-ing vol 2010 Article ID 236568 11 pages 2010

[33] Ofcom ldquoStatement on Cognitive Access to Interleaved Spec-trumrdquo July 2009

[34] ITU ldquoDVB-T coverage measurements and verification of plan-ning criteriardquo ITU-R Recommendation SM1875-2 ITU 2014


[35] ITU-R Rec P1623-1 Prediction method of fade dynamics onEarth-space paths 2005

[36] M F Hanif and P J Smith ldquoLevel crossing rates of interferencein cognitive radio networksrdquo IEEE Transactions on WirelessCommunications vol 9 no 4 pp 1283ndash1287 2010

[37] M F Hanif P J Smith andM Shafi ldquoPerformance of cognitiveradio systems with imperfect radio environment map informa-tionrdquo in Proceedings of the Australian Communications TheoryWorkshop (AusCTW rsquo09) pp 61ndash66 IEEE Sydney AustraliaFebruary 2009

[38] H Shatila M Khedr and J H Reed ldquoOpportunistic channelallocation decision making in cognitive radio communica-tionsrdquo International Journal of Communication Systems vol 27no 2 pp 216ndash232 2014

[39] C Kocks A Viessmann P Jung L Chen Q Jing and R Q HuldquoOn spectrum sensing for TV white space in Chinardquo Journal ofComputer Networks and Communications vol 2012 Article ID837495 8 pages 2012

Research ArticleETSI-Standard Reconfigurable Mobile Device forSupporting the Licensed Shared Access

Kyunghoon Kim1 Yong Jin1 Donghyun Kum1 Seungwon Choi1

Markus Mueck2 and Vladimir Ivanov3

1School of Electrical and Computer Engineering Hanyang University Seoul 04763 Republic of Korea2Intel Mobile Communications Group 85579 Munich Germany3Mobile SoC Development Department LG Electronics Inc Seoul 06744 Republic of Korea

Correspondence should be addressed to Seungwon Choi choidsplabhanyangackr



Copyright copy 2016 Kyunghoon Kim et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

In order for a Mobile Device (MD) to support the Licensed Shared Access (LSA) the MD should be reconfigurable meaning thatthe configuration of a MD must be adaptively changed in accordance with the communication standard adopted in a given LSAsystem Based on the standard architecture for reconfigurable MD defined in Working Group (WG) 2 of the Technical Committee(TC) Reconfigurable Radio System (RRS) of the European Telecommunications Standards Institute (ETSI) this paper presentsa procedure to transfer control signals among the software entities of a reconfigurable MD required for implementing the LSAThis paper also presents an implementation of a reconfigurable MD prototype that realizes the proposed procedure The modemand Radio Frequency (RF) part of the prototype MD are implemented with the NVIDIA GeForce GTX Titan Graphic ProcessingUnit (GPU) and the Universal Software Radio Peripheral (USRP) N210 respectively With a preset scenario that consists of fivetime slots from different signal environments we demonstrate superb performance of the reconfigurable MD in comparison to theconventional nonreconfigurable MD in terms of the data receiving rate available in the LSA band at 23ndash24GHz

1 Introduction

Global mobile data traffic is expected to grow up to 243exabytes per month by 2019 which is nearly a tenfoldincrease compared to the traffic in 2014 [1] To cope withthis explosive increase in data traffic various enabling tech-nologies such as full dimensional multiple-input multiple-output device-to-device communication and newwaveformdesigns such as nonorthogonal multiple access have beenactively researched [2 3] In particular the World RadioCommunication conference in 2015 (WRC-15) of the Inter-national Telecommunication Union-Radio (ITU-R) commu-nication sector considers spectrum sharing technology to be akeymethodology that is applicable in the 5thGeneration (5G)mobile communications [4] Among the various spectrumsharing techniques Licensed Shared Access (LSA) which is aframework for sharing the spectrum among a limited numberof users [5] has been the focus of research especially in

Europe The Electronic Communications Committee (ECC)performed a comprehensive study of the regulatory aspectof LSA They also released the results of their research onthe applicability of the LSA concept in the 23ndash24GHz bandusing Time-Division Duplexing (TDD) [6] The CognitiveRadio Trial Environment (CORE) demonstrated an LSA livetest in the LSA band at 23ndash24GHz [7] while Mustonenet al introduced a novel network architecture namely self-organizing networking features [8] to support LSA Duringthis timeWorkingGroup (WG) 1 of theTechnical Committee(TC) on the Reconfigurable Radio System (RRS) of theEuropean Telecommunications Standards Institute (ETSI)has been developing LSA-related standards In addition [9ndash11] introduced an early-stage overview of the LSA systemconcept LSA system requirements and architecture foroperation of mobile broadband systems respectively All theLSA-related developments introduced above however haveonly considered the LSA technology from the viewpoint of



network or infrastructure systems but not from the viewpointof Mobile Device (MD) This is problematic because theprevious work has not specified the functionalities requiredin MDs in order to operate using LSA For example if aMD does not support TDD Long Term Evolution (LTE) atthe frequency band of 23ndash24GHz an additional spectralband for LSA that is 23ndash24GHz [9] would provide verylittle advantage [12] Consequently in order to fully exploitspectrum sharing MD must be able to adaptively change itsconfiguration appropriately for the radio application (RA)defined in a given LSA band Therefore it seems thatreconfigurability is amandatory characteristic ofMD in orderto fully exploit the benefits of LSA-based spectrum sharing

Recently WG2 of TC-RRS of ETSI developed a standardarchitecture and related interfaces for reconfigurableMDs In[13] WG2 released a standard reconfigurable MD architec-ture with its main effort focused on resolving the problemof portability between the RA code and the MD hardwareplatform WG2 has also defined standard interfaces in accor-dance with the standard architecture for reconfigurable MDsin [14 15]

The main contribution of this paper is to show how thereconfiguration of MDs should be achieved for realizing LSAdemonstrated by WG1 of TC-RRS of ETSI in [9] where it isassumed that the target MD is compliant with the standardarchitecture released by WG2 of TC-RRS of ETSI [13] Ifthe target MD is reconfigurable there is no restriction onthe RA in an LSA region For example a MD is configuredwith TDD LTE in the frequency region at 23ndash24GHz inorder for the scenario in [9] to be valid because TDD LTEhas been defined as the designated RA in the LSA regionof the 23ndash24GHz band [12] Since we do not know ingeneral which RA will be adopted in the LSA region theLSA technology is not useful for nonreconfigurable MDsIn order to verify the reconfiguration of MDs for LSA wespecify in this paper which interactions should occur inwhat order among the software entities in the reconfigurableMDs using the ETSI-standard architecture The systematicinteractions among the software entities of the reconfigurableMD are referred to as a ldquoprocedurerdquo in this paper We alsopresent implementation of the reconfigurable MD prototypethat realizes the proposed proceduresThe implemented test-bed using the MD prototype is compliant with the referencemodel of the standard architecture [13] released by WG2 ofTC-RRS of ETSI The modem and Radio Frequency (RF)of the prototype MD are implemented with the NVIDIAGeForce GTX Titan Graphic Processing Unit (GPU) andUniversal Software Radio Peripheral (USRP) N210 respec-tively Assuming the LSA region adopts TDD LTE as shownin [12] we demonstrate superb performance of the reconfig-urable MD compared to a conventional nonreconfigurableMD in terms of the data receiving rate available in theLSA band at 23ndash24GHz In addition to the experimentaltests performed with the implemented test-bed computersimulations have also been presented considering a scenarioof multiple users in an LSA band It was verified through thecomputer simulations that the reconfigurable MDs not onlyincrease the total sum rate itself but also increase the numberof users satisfying a given QoS

The rest of this paper is organized as follows Section 2introduces the standard architecture for a reconfigurableMDdeveloped byWG2of TC-RRS based onwhich the procedureis set up in the following section Section 3 proposes theprocedures that specify the interactions among the softwareentities of the ETSI-standard reconfigurable MD for real-ization of the LSA Section 4 introduces the implementedreconfigurableMDwhile Section 5 presents the experimentalresults obtained from the implementedMDand performanceevaluations obtained from the computer simulations con-sidering the scenario of multiple users Finally Section 6concludes this paper

2 Architectural Model for Reconfigurable MD

WG2 of TC-RRS of ETSI has developed a standard architec-ture for reconfigurable MDs and related interfaces with theintention that any desired Radio Access Technologies (RATs)can be realized in a reconfigurable MD by downloading thetarget RA code from the public domain for example theRadioApp Store [16] regardless of the hardware platformof the MD This section introduces a brief summary of thestandard architecture and related interfaces based on whicha systematic procedure is developed in the following sectionin such a way that the software entities in the reconfigurableMD interact with one another for implementing the LSA

21 Architecture for Reconfigurable MD Figure 1 illustratesthe reconfigurable MD architecture and related interfacesproposed by WG2 of TC-RRS of ETSI As shown in thefigure the architecture consists of a Communication ServicesLayer (CSL) RadioControl Framework (RCF)UnifiedRadioApplications (URAs) and radio platform [13] Although thefour components are shown in the figure the necessarypart of the ETSI standard includes the four entities in CSLthat is the Administrator Mobile Policy Manager (MPM)networking stack and monitor as well as the five entities inRCF that is the Configuration Manager (CM) Radio Con-nection Manager (RCM) Flow Controller (FC) multiradiocontroller (MRC) and Resource Manager (RM) This meansthat the radio platform is vendor-specific and the URA isthe downloaded RA code consisting of functional blocksmetadata and other software needed for the processing ofcontext information [13ndash15]

The functionality of each of the four entities in the CSLcan be summarized as follows Administrator entity requests(un)installation of URA and creates or deletes instances ofURA The MPM entity monitors the radio environmentsand MD capabilities requests (de)activation of URA andprovides information about the URA list The networkingstack entity sends and receives the user data The monitorentity transfers the context information from the URA to theusers or the proper destination entity in a MD

The functionality of each of the five entities in theRCF canbe summarized as followsTheCMentity (un)installs createsor deletes instances of URA and manages access to the radioparameters of the URA The RCM entity (de)activates URAaccording to user requests and manages user data flows TheFC entity sends and receives user data packets and controls


AdministratorMobility

PolicyManager

Networking stack Monitor

Radio Connection

Manager

MultiradioController

Resource Manager

UnifiedRadio

Application

Flow Controller

Communication Services Layer

Radio Control Framework

Multiradio Interface (MURI)

Unified RadioApplication Interface

(URAI)

ReconfigurableRadio FrequencyInterface (RRFI)

RF transceiver

Radio platform

ConfigurationManager

Baseband and others

Figure 1 Reconfigurable MD architecture and related interfaces [13]

the flow of the signaling packets The MRC entity schedulesthe requests for radio resources issued by concurrentlyexecuting URAs as well as detecting and managing theinteroperability problems among the concurrently executedURAs The RM entity manages the computational resourcesin order to share them among the simultaneously activeURAThis guarantees their real-time execution

The RA code that is the software that enforces gen-eration of the transmit RF signals or the decoding of thereceived RF signals becomes a URA once it is downloadedinto a reconfigurable MD Since all RAs exhibit commonbehavior from a reconfigurable MD perspective once theyare downloaded in a reconfigurable MD the downloaded RAcode is called URA which consists of functional blocks thatexhibit the required modem functions of the correspondingRAT

The radio platform shown in Figure 1 is part of the MDhardware that relates to the radio processing capability Itincludes the programmable components hardware acceler-ators RF transceiver and antenna(s)

22 Interfaces for Reconfigurable MD As shown in Figure 1there are three types of interfaces the Multiradio Interface(MURI) Unified Radio Application Interface (URAI) andReconfigurable RF Interface (RRFI) with which entities fromthe CSL RCF and radio platform can interact with oneanother

The MURI interfaces each entity of the CSL and RCFIt provides three types of services administrative servicesaccess control services and data flow services [14]TheURAIinterfaces each entity of the RCF and URA It provides fivetypes of services RA management services user data flowservices multiradio control services resource managementservices and parameter administration services [17] TheRRFI interfaces the URA and the radio platform It providesfive types of services spectrum control services powercontrol services antenna management services transmit(Tx)receive (Rx) chain control services and radio virtualmachine protection services [15]

3 Proposed Procedures for LSA inReconfigurable MD

In this section we present an LSA procedure for reconfig-urable MD in which the architecture is specified as the ETSIstandard briefly summarized in the previous section Theprocedure introduced in this section specifies how the entitiesin the CSL and RCF shown in Figure 1 interact with oneanother

Figure 2 illustrates a conceptual view of realizing LSAin which the basic scenario has been demonstrated by WG1of TC-RRS of ETSI [9] The National Regulation Authority(NRA) shown in Figure 2 manages the LSA Repository insuch a way that it provides the LSA Repository information


LSA Repository

Mobile device

Base station

LSA controller

OAM

CORE network

NRA

Figure 2 Conceptual view of realizing LSA

about LSA license regarding the right of using the LSA bandand receives a report regarding the use of LSA spectrumfrom the LSA Repository The LSA Repository containsa database of spatial and temporal information regardingthe spectrum use of the incumbent user Based on theinformation provided from the LSA Repository the LSAcontroller determines the availability of the spectrum thatcan be shared using LSA In cases when the spectrum isavailable the network management system which is denotedas ldquoOperation Administration and Maintenance (OAM)rdquo inFigure 2 acknowledges the availability of the spectrum to thecorresponding base station

The use case of expanding the bandwidth using LSA hasbeen released by WG1 of TC-RRS of ETSI in [9] This is thebasis of the LSA procedure introduced in this section Theuse case can be summarized as follows Let us first considera case where a Mobile Network Operator (MNO) providinga Frequency Division Duplexing (FDD) LTE service wantsto switch the spectral band from its own FDD LTE bandto the LSA band at a specific time Note that as shown in[12] the LSA region is assumed to be supported with TDDLTE in the band at 23ndash24GHz Assuming the MNO hasheld the individual authorization for using the extra band at23ndash24GHz the LSA controller shown in Figure 2 decideswhich base stations can be granted use of the extra spectralband for the required time period Receiving the informationregarding the availability of the extra spectral band fromthe LSA controller the OAM shown in Figure 2 notifiesthe availability of the spectrum to those base stations whichmay use the extra spectral band at 23ndash24GHz In order toimplement this use case we propose a procedure for updatingthe configuration of MD with a new RA defined in a givenLSA region that is TDD LTE in this use case

Figure 3 illustrates the procedure of updating the config-uration of MD with an arbitrary RA required for LSA Theprocedure shown in Figure 3 can be summarized in the 17steps shown as follows

Step 1 In order to install a new URA the the Administratorsends a DownloadRAPReq signal including the Radio Appli-cation Package (RAP) identification (ID) to the RadioAppStore

Step 2 The Administrator receives a DownloadRAPCnf sig-nal including the RAP ID and RAP from the RadioApp Store

Step 3 Upon the download of RAP from the RadioApp Storethe Administrator sends an InstallRAReq signal including theRAP ID to the CM to request installation of the new RA

Step 4 The CM first performs the URA code certificationprocedure in order to verify its compatibility authenticationand so forth

Step 5 The CM performs installation of URA and transfersan InstallRACnf signal including the URA ID to the Admin-istrator

Step 6 In order to deactivate the current URA the MPMtransfers the RCMHardDeactivateReq signal which includesthe RA ID

Step 7 Upon a request from the RCM the Radio OperatingSystem (ROS) deactivates the designated URA

Step 8 After the ROS completes hard deactivation of theURA the RCM acknowledges completion of the deactivationprocedure by sending a HardDeactivateCnf signal to theMPM

Step 9 In order to create an instance of a newURA theMPMtransfers an InstantiateRAReq signal including the ID of theURA to be instantiated to the CM

Step 10 The CM transfers an RMParameterReq signal andanMRCParameterReq signal including the ID of the URA inorder to get the parameters needed for URA activation to theRM and MRC

Step 11 The CM receives an RMParameterCnf signal includ-ing the ID of the URA and the radio resource parametersfrom the RM

Step 12 The CM receives an MRCParameterCnf signalincluding the ID of the URA and computational resourceparameters from the MRC

Step 13 The CM transfers the URA ID and the receivedparameters for performing theURA instantiation to the ROS

Step 14 After creating an instance the CM transfers anInstantiateRACnf signal including the URA ID to the MPM

Step 15 In order to activate the newURA theMPM transfersan ActivateReq signal including the ID of the URA to theRCM

Step 16 Upon request from the RCM the ROS activates thedesignated URA

Step 17 After the ROS completes activation of the URA theRCM sends an ActivateCnf signal back to the MPM

Note that Steps 3 and 5 utilize the administrative servicesof the MURI [14] Steps 6 8 9 14 15 and 17 make use of the


HardDeactivateReq(R1ID)HardDeactivate(R1ID)

HardDeactivateCnf(R1ID)

InstantiateRAReq(R2ID)RMParameterReq(R2ID)

MRCParameterReq(R2ID)

InstantiateRACnf(R2ID)

ActivateReq(R2ID)Activate(R2ID)

ActivateCnf(R2ID)

Deactivation

Creatinginstance

Activation

DownloadRAPReq(P2ID)

DownloadRAPCnf(P2IDRAP)CreatingRAP(P2ID)

InstallRAReq(P2ID)

Certification

InstallRACnf(R2ID)Installation CreateRA(R2ID)

ResourceManager


Radio ConnectionManager

Mobility PolicyManager

R1 Unified RadioApplication

MultiradioControllerAdministratorRadio Apps

Store

P2 RadioApplication Package

Downloaded


Installed

Instantiated

Active

Active

Deactivated

MRCParameterCnf(R2ID Param2RMParameterCnf(R2ID Param1

InstantiateRA(R2ID Param1 Param2 )

)

)

)

Figure 3 Procedure of MD reconfiguration for implementing LSA

access control services of theMURI [14] Steps 7 and 16 utilizethe radio applicationmanagement services of URAI [17] andSteps 4 and 13 make use of the parameter administrationservices of URAI [17] Steps 10 11 and 12 are related to theinteractions among the entities in the RCF which are vendor-specific

Through the procedure shown in Figure 3 the MDreconfiguration can be achieved by updating the presentURAwith a new one Note that in the use case presented by WG1of TC-RRS of ETSI in [9] the present URA is FDD LTEand the new one is TDD LTE It is also noteworthy that thefeasibility of the standard architecture and related interfacescan be verified from Figure 3 through the observation thatthe desired RA code is first downloaded from the RadioAppStore then installed instantiated and activated in a givenreconfigurable MD

4 Implementation of a ReconfigurableMD for LSA

This section presents implementation of the prototype recon-figuration MD used as a test-bed for obtaining the experi-mental results of LSA introduced in Section 5 The imple-mented prototype system is compliant with the standardarchitecture of ETSI TC-RRS WG2 [13]

Figure 4(a) illustrates a reference model of the recon-figurable MD architecture introduced in [13] According tothe standard architecture of the reconfigurable MD definedby WG2 of TC-RRS of ETSI operations supported by theApplicationProcessor are based onnon-real-time processingThe operations supported by the Radio Computer are basedon real-time processing while the dotted part in betweenthese two parts shown in Figure 4(a) is either non-real-timeor real-time depending upon the vendorrsquos choiceThis optionmeans that the Operating System (OS) of the ApplicationProcessor must be a non-real-time OS such as Android or

iOS while that of the Radio Computer which is referred toas ROS in Figure 4(a) has to be a real-time OS includingRCF as indicated in Figure 4(a) The Application Processorin Figure 4(a) includes the following components (1) a driverthat activates a hardware device such as a camera or speakerin the part of the Application Processor on a given MD and(2) a non-real-time OS for execution of the AdministratorMPM networking stack and Monitor [13] which are partof the CSL as described previously The Radio Computerincludes the following components (1) ROS for executingthe functional blocks of the given RAs (2) a radio platformdriver which is for the ROS to interact with the radioplatform hardware and (3) a radio platform which typicallyconsists of programmable hardware dedicated hardware RFtransceiver and antenna(s)

Figure 4(b) illustrates a block diagram of the reconfig-urableMDprototype architecture that has been implementedas a test-bed based on the architecture shown in Figure 4(a)As shown in Figure 4(b) the Application Processor part ofthe test-bed consists of Ubuntu 1204 [18] and CSL whilethe Radio Computer part consists of a Linux kernel RCFradio platform driver and radio platform For the purposeof experimental tests we have not adopted a real-time OS forthe Radio Computer part because the primary purpose of thetest-bed is to verify the feasibility of the standard architecturefor the functionality of LSA-based spectrum sharing ratherthan the real-time functionality of the RA code executionFurthermore the test-bed system does not include all theentities of the CSL and the RCF defined in the ETSI standardSpecifically in the test-bed system shown in Figure 4(b)CSL consists of an Administrator and MPM only while RCFconsists of CM RCM RM and MRC only Also it can beobserved from Figure 4(b) that the Linux kernel which playsthe role of ROS in the test-bed system supports the executionof the functional blocks of a given RA code The RA codeprepared for our test-bed system consists of FDD LTE and


Driver

Radio platform driver

OS

CommunicationServices Layer

Radio OS

App

1Ap

p 2

App

3

App M

Radio platform

Dedicatedhardware AntennaRF transceiver

RA1

RA2

RA3

RAN


Unified Radio Applications

Programmablehardware

Appl

icat

ion

Proc

esso

rRa

dio

Com

pute

r




(a) Reference model of the ETSI-standard reconfigurable MD architec-ture [13]


Communication Services Layer(Administrator MPM)

Ubuntu1204 (OS)

Linux kernel

CUDA driverRadio PlatformProgrammable

hardware(GPU)

FDD LTE TDD LTE

Radio Control Framework (CM RCM MRC RM)

GbEUHD

RF transceiver(USRP N210)

Implemented with USRP N210

Implemented with CPU and GPU in an

ordinary PC

Appl

icat

ion

Proc

esso

rRa

dio

Com

pute

r(b) Implemented reconfigurable MD test-bed architecture

Figure 4 Block diagram of the reference model and implemented test-bed of a reconfigurable MD

TDD LTE which are compliant with 3GPP Rel 10 [19] TheRA code is executed on a GPU in radio platform of the test-bed GPU in general since it contains a great number ofpowerful threads is appropriate for parallel computing Inorder to utilize the number of threads efficiently the RA codecontaining FDD LTE and TDD LTE has been implementedusing Compute Unified Device Architecture (CUDA) thatis a C-based programming language provided by NVIDIAThe GPU adopted in our test-bed is NVIDIArsquos GeForce GTXTitan that is capable of 4494 GFLOPS using 2688 CUDAcore processor cores [20] In addition the radio platformdriver shown in Figure 4(b) includes the CUDA driver andthe URSP Hardware Driver (UHD) through which the Linuxkernel can access the radio platform consisting of a NVIDIAGeForce GTX Titan GPU and USRP N210 [21] respectively

The key issue in RA code implementation is to maximizethe degree of parallelization among the large number ofthreads in a given GPU In fact the parallelization can beconsidered in multiple layers that is among grids blocksandor threads in a given GPU Note that each grid containsmultiple blocks and each block includes multiple threadsIn order to maximize the degree of parallelization eachfunction block of the RA code should be partitioned intoas many pieces as possible such that we can maximize thenumber of threads to be activated for executing a giventask For example the procedure of channel estimation alongthe frequency axis [19] which is a function block neededin both FDD and TDD LTE has been partitioned in ourRA code implementation in such a way that a single gridcontaining 200 blocks each of which includes 6 threads inthe NVIDIA GeForce GTX Titan GPU has been activated Itmeans that totally 1200 threads are activated in parallel for


GUI

Ordinary PC (CPU and GPU)

GbE

Spectrum analyzer

Figure 5 Photograph of the implemented reconfigurable MD test-bed

the function block of the channel estimation along frequencyaxis Similarly for the function block of channel estimationalong time axis [19] totally 8400 threads that is 14 threads ineach block and 600 blocks in a single grid have been activatedin parallel

Figure 5 illustrates a photograph of the implementedtest-bed of the reconfigurable MD The test-bed realizes thearchitectural model shown in Figure 4(b) As shown in Fig-ure 5 the test-bed system consists of two parts an ordinaryPersonal Computer (PC) and an RF transceiver An ordinaryPC which provides a NVIDIA GeForce GTX Titan GPU andCentral ProcessingUnit (CPU)was used to implement all thecomponents of the reconfigurable MD shown in Figure 4(b)except for the RF transceiver which has been separatelyimplemented with USRP N210 as shown in Figure 5 In our


FDD LTE encoder

Video data stream

PC for eNB

RF transceiver

GbE

TDD LTE encoder

GbE RF transceiver

(a) Functional block diagram of eNB

DecoderVideo data stream

PC for MD

RF transceiver

GbE

(b) Functional block diagram of MD

Figure 6 Functional block diagram of the test-bed system

implementation the RF transceiver is connected with thePC through a Giga-bit Ethernet (GbE) as shown in Figures4(b) and 5 All the functional blocks in a given RA code areexecuted on the NVIDIA GeForce GTX Titan GPU boardin the PC while all the functionalities of the RF transceiverincluding analog-to-digital and digital-to-analog conversionsas well as frequency-up and frequency-down conversionsare performed in the USRP N210 Note that the lower partshown by a dotted line in Figure 4(b) corresponds to the RFtransceiver implemented with USRP N210 while the otherpart shown by a solid line in Figure 4(b) corresponds to allthe other parts of a reconfigurable MD implemented withthe ordinary PC shown in Figure 5 Since an ordinary PConly provides a GPU and CPU the implemented prototypesystem does not include Field Programmable Gate Arrays(FPGA) or Digital Signal Processors (DSP) in the part ofthe radio platform shown in Figure 4(b) while the GPUsupports all the functional blocks required in the FDD LTEand TDD LTE that are needed in the LSA The CPU in thePC was used to realize the functionalities of RCF as well asto control the GPU and USRP through the CUDA driver andUHD in the radio platform driver respectively as mentionedearlier The Graphic User Interface (GUI) shown in Figure 5provides monitoring of the video data stream which is theresult of decoding the received FDD or TDD LTE signalsas well as a set of environmental parameters such as datathroughput and Bit Error Rate (BER)The spectrum analyzershown in Figure 5 was used to observe the center frequencyand bandwidth of the RF signals of FDD and TDD LTE

5 Numerical Results

51 Experimental Tests This subsection presents the exper-imental results of the LTE data throughput obtained froma test-bed consisting of an Evolved Node B (eNB) and MDoperating in the signal environment of the use case consid-ered in Section 3 that is the use case of expanding bandwidthusing LSA In the experimental tests we considered two types

of MD for comparison purposes One is a legacy MD ofwhich the configuration is fixed with FDDLTE and the otheris capable of changing its configuration between FDD LTEand TDD LTE depending on the given signal environmentIn general a MD performs a horizontal handover that isit moves to an adjacent base station when the Quality ofService (QoS) drops down to a preset threshold value If thegiven QoS cannot be satisfied through a horizontal handovera reconfigurable MD performs a vertical handover that is itchanges the present radio application to another one that canbring about satisfactory QoS [12] In this paper the requiredQoS was set up with a preset level of LTE data throughputTherefore when the preset level of the LTE data throughput isnot achieved through a horizontal handover the MD checksthe availability of the TDD LTE of the LSA band in order toperform a vertical handover from FDD LTE to TDD LTE Aswe have implemented a single eNB for simplicity howeverthe reconfigurable MD performs a vertical handover directlywhen the present LTE data throughput becomes lower thanthe threshold level Consequently whenever the QoS is notmaintained assuming the LSAband is available in the presentregion a reconfigurable MD changes its configuration fromFDD LTE to TDD LTE As for the legacy MD the config-uration is always fixed with FDD LTE whether or not theQoS is satisfied In this subsection we have summarized theLTE data throughput obtained from both the reconfigurableMD and legacy MD in a signal environment where the QoSand availability of the LSA band vary as a function of timeFor the experimental tests introduced in this subsectionthe MD prototype shown in Section 4 was used for thereconfigurable MD while the dual mode eNB supportingFDD and TDD LTE shown in our previous work in [22] wasused

Figure 6 illustrates a functional block diagram of the dualmode eNB [22] that supports both FDD and TDD LTE andthat of MD Both eNB and MD were implemented with aPC including a GPU for base band signal processing andUSRP N210 which plays the role of the RF transceiver Asshown in Figure 6(a) eNB encodes the video data streamin accordance with the data format of both FDD and TDDLTE The encoded data are transferred to the RF transceiverof USRP N210 via GbE and radiated through the transmitantennas For FDD LTE the center frequency was set to17 GHz a licensed band with its bandwidth being 10MHzwhile TDD LTE uses 235GHz as its center frequency withits bandwidth being 15MHz For the experimental tests ofLSA eNB transmits the FDD LTE signals continually whilethe TDD LTE signal is transmitted only for a preset periodof time which means eNB in our test-bed system transmitsboth FDD and TDD LTE signals only for a preset period oftime except for the FDD LTE signal which is transmittedfrom eNB Figure 6(b) illustrates a common functional blockdiagram for both reconfigurable MDs and legacy MDsAs shown in Figure 6(b) the RF signal transmitted fromeNB is captured at the receive antenna of MD and thefrequency-down and analog-to-digital are converted at theRF transceiver of USRP N210 Then the FDD andor TDDLTE signal is decoded and retrieved into the video datastream


Table 1 Scenario set up for experimental tests

Time interval QoS LSA band1198791 1199050sim1199051

Satisfied Not available1198792 1199051sim1199052

Not satisfied Not available1198793 1199052sim1199053

Not satisfied Available1198794 1199053sim1199054

Satisfied Available1198795 1199054sim1199055

Satisfied Not available


System parameter FDD LTE TDD LTECommunication standard 3GPP Rel 10Channel coding Turbo coding (coding rate = 12)Center frequency (GHz) 17 235Transmission bandwidth (MHz) 10 15Modulation scheme 16 QAM 64 QAMULDL configuration mdash 6Special subframe configuration mdash 1

Table 1 shows the scenario set up for the experimentaltests in terms of QoS satisfaction and LSA band availabilityEach time interval in Table 1 was set to 60 seconds Theexperimentwas performed for five time intervals starting at 119905

0

and ending at 1199055 For example during the first time interval

1198791 that is from 119905

0to 1199051 the signal environment was set up

in such a way that QoS was satisfied and the LSA band isnot available The condition whether or not QoS is satisfiedis determined as mentioned earlier depending on whetheror not the data throughput at the receiving MD exceeds thepreset threshold value The value for the threshold has beenarbitrarily set up to 10Mbps The signal environment wherethe QoS was satisfied was set up by allocating all the spectralresources of FDD LTE to the target MD The other signalenvironment where QoS was not satisfied was implementedby allocating only a half of the entire spectral resources ofFDD LTE to the target MD For the availability of the LSAband the LSA band becomes available only when the dualmode eNB transmits the video stream data in both FDD andTDDLTEWhen eNB transmits the video streamdata only inFDD LTE the LSA band is not available In our experimentassuming that the LSA band is available for the time intervalsof 1198793and 119879

4 the availability of the LSA band is set up for 119879

3

and 1198794as shown in Table 1 which means the procedure for

the LSA controller to notify the availability of the LSA bandto OAM has been omitted in our experiment Note that sincetheMDnormally operates in FDD LTEmode the availabilityof the LSA band does not have to be checked as long as QoSwith FDD LTE is satisfied Consequently if QoS with FDDLTE is not satisfied the reconfigurable MD starts to set upits configuration with TDD LTE of the LSA band while theconventional nonreconfigurable MD has to stay in FDD LTEmode with unsatisfactory data throughput

Figure 7 shows an image of the experimental test formeasuring the data throughput of the reconfigurable MDand legacy MD The system parameters for FDD andTDD LTE were set up as shown in Table 2 Since the

Antenna for reconfigurable

MD

Antenna for legacy MD

Reconfigurable MD Legacy MDeNodeB

Antenna for eNodeB

Figure 7 Photograph showing the experimental environment forcomparing the received data throughputs of the reconfigurable MDand legacy MD

Table 3 Average throughput with Key Performance Indicator (KPI)value for the reconfigurable MD

MD Time interval (Mbps)11987911198792

1198793

1198794

1198795

ReconfigurableMD 1488 732 1439

(KPI = 1) 1445 1487(KPI = 1)

Legacy MD 1480 733 733 1480 1482

received data throughput for TDD LTE is determined by theuplinkdownlink configuration type and the special subframeconfiguration type the types in Table 2 were set up in such away that the maximum throughput of FDD and TDD LTEbecomes approximately the same

Figure 8 illustrates the throughput values measured at thereceiving MD The data throughput shown in Figure 8 wasobtained from the experimental environment shown in Fig-ure 7 inwhich the eNB andMDuse the systemparameter val-ues shown in Table 2 according to the experimental scenarioshown in Table 1 Table 3 shows an average Rx throughput foreach time interval together with Key Performance Indicator(KPI) which indicateswhether or not the configuration of thereconfigurable MD has been correctly set up in accordancewith a given signal environment More specifically KPItells whether or not the configuration of the reconfigurableMD has been correctly changed from FDDTDD LTE toTDDFDD LTE during the time interval 119879

31198795 Therefore

KPI is set up to 1 or reset to 0 depending on whether the con-figuration of the reconfigurableMD is performed successfullyor not Consequently throughput of the receivingMDwouldhave become greater than 10Mbps145Mbps during the timeinterval of 119879

31198795if the configuration of the reconfigurable

MD was successfully performed that is from FDDTDDLTE to TDDFDD LTE during the time interval of 119879

31198795

The solid line in Figure 8 corresponds to the performanceof the reconfigurable MD while the dotted line correspondsto the legacy MD It can be observed from Figure 8 thatduring the first time slot 119879

1 both the reconfigurable MD and

legacy MD exhibit almost the same maximum throughputs1488M bits per second (bps) and 1480Mbps respectivelywith FDD LTE because the first time slot was set up for


0789

10111213141516

Time (sec)

Thro

ughp

ut (M

bps)

Reconfigurable MDLegacy MD

T1 T2 T3 T4 T5

t1 = 60 t2 = 120 t3 = 180 t4 = 240 t5 = 300

Figure 8 Throughput measured at the receiving MD according tothe experimental scenario shown in Table 1

QoS to be satisfied with FDD LTE Note that with the signalenvironment of QoS being satisfied as mentioned earlierit is implemented by allocating all of the spectral resourcestransmitting eNB to the target MD Note that the maximumthroughput of FDD LTE 1488Mbps can be calculated fromthe system parameters shown in Table 2 as 744336 (numberof 16 QAM symbols per frame) lowast 05 (channel coding rate) lowast4 (number of bits per 16 QAM symbol)10ms (frame length)During the second time slot 119879

2 the signal environment was

set up for QoS not being satisfied and the LSA band notbeing available as shown in Table 1 Setting the thresholdvalue for determining whether or not QoS is satisfied to be10Mbps at the receiving MD we have allocated only half ofall the spectral resources of eNB to the target MD in order toimplement the signal environment as QoS not being satisfiedIt can be observed that with half of all the spectral resourcestransmitting eNB themaximum throughput is nearly 14882= 744Mbps which is far less than the threshold value of10Mbps During 119879

2 eNB transmits data with only half of the

entire spectral resources with which the throughput cannotexceed the threshold therefore QoS is not satisfied Sincethe signal environment during 119879

2does not provide the LSA

band either both the reconfigurable and legacy MDs cannothelp staying in FDD LTE with nearly the same throughputs732Mbps and 733Mbps respectively During 119879

3 since eNB

transmits the signal in both FDDandTDDLTEmeaning thatthe LSA band is now available the reconfigurable MD canexploit the throughput of TDDLTE 1439Mbps by switchingits configuration from FDD LTE to TDD LTE of the LSAbandThe legacyMD however stays in FDD LTE with only ahalf throughput Note that themaximum throughput of TDDLTE that is 145Mbps available with the system parametersshown in Table 2 can be calculated as 47986 (number of64 QAM symbols per frame) lowast 05 (channel coding rate)lowast 6 (number of bits per 64 QAM symbol)10ms (framelength) During 119879

4 as eNB transmits the signals of FDD LTE

that satisfy the QoS requirement the legacy MD can securethe maximum throughput comparable to the one obtainedduring 119879

1 Since the throughput is maintained above the

threshold the reconfigurable MD stays in TDD LTE Sincethe throughput of TDD LTE has been arbitrarily set up a littlebit lower than that of FDD LTE in our test-bed system thethroughput of the reconfigurable MD happens to be slightlylower than that of legacyMDduring119879

4 During119879

5 as the LSA

band is no longer available the reconfigurable MD changesits configuration back to FDD LTE from TDD LTE with itsthroughput returning to the one obtained during 119879

1 Note

that the lengths of the time intervals could be related to thepossible interferences tofrom primarysecondary users ofthe spectrum In addition since the transition in betweenthe configuration changes takes about 5ndash10ms in our test-bed the lengths of 119879

3and 119879

4where the LSA band is available

should not be too short for the MDs using the LSA bandto exploit the benefit of LSA But it should not be too longbecause otherwise the MDs occupying the LSA band couldinterfere with the primary users

From our experimental tests performed in accordancewith the preset scenario shown in Table 1 it is clear thatin order to fully utilize the benefits of the LSA band theconfiguration of MD should be adjustable to the radioapplication used in the LSA band which is set to TDD LTEin our experiments

52 Computer Simulations In the test-bed implemented forthe experimental tests the number of the reconfigurableMDsand that of legacy MDs were only 1 as shown in Figure 7In this subsection we introduce computer simulations per-formed for a scenario of multiple users in a given LSA bandThe system parameters shown in Table 2 which were usedfor the experimental tests have been adopted again in thesimulations The total number of users which consists of thereconfigurable MDs as well as legacy MDs is set to be 100 inthe simulations For simplicity but without loss of generalitywe assume that the number ofMDs that can be allowed usingthe LSA band is limited to 30 by the NRA shown in Figure 2[5] in our simulations Furthermore the Rx throughput ofeach user has arbitrarily been set up with a random numberbetween 30Kbps and 300Kbps where the threshold valuethat determines whether or not QoS is satisfied has been setup to 100Kbps Therefore those MDs whose throughput isbelow the threshold that is 100Kbps are to apply for theLSA band by changing their configurations from FDD LTEto TDD LTE Among those MDs not more than 30 MDs arerandomly selected for using the LSA band in our simulationsConsequently the Rx throughput of each reconfigurable MDthat has been allowed using the LSA band would be changedfrom a random number between 30Kbps and 100Kbps toanother random number between 100Kbps and 300Kbps ifthe reconfigurable MDs have been accepted to use the LSAband

Figure 9 illustrates accumulated sum rates when theportion of the reconfigurable MDs is 0 10 50 70and 100 of the entire 100 users As shown in Figure 9since the LSA band is not available until the end of 119879

2 the

accumulated sum rates for all the cases are quite comparableAs the LSA band becomes available during the time intervalof 1198793and 119879

4 the sum rates increase more rapidly as the

portion of the reconfigurable MDs is higher Note that the


0 60 120 180 240 3000

1

2

3

4

5

6

7

Time (sec)

Accu

mul

ated

sum

rate

(Gbp

s)

Reconfigurable MD 100Reconfigurable MD 70Reconfigurable MD 50

Reconfigurable MD 10Reconfigurable MD 0

T1 T2 T3 T4 T5

Figure 9 Accumulated sum rates

0 01 02 03 04 05 06 07 08 09 10

010203040506070809

1

Normalized user throughput

CDF



Figure 10 CDF according to the normalized user throughput

number of the reconfigurable MDs whose throughputs areimproved due to the LSA technology increases as the portionof the reconfigurable MDs is higher From Figure 9 it can beobserved that more number of reconfigurable MDs improvesthe accumulated sum rate more conspicuously

Figure 10 illustrates Cumulative Distribution Function(CDF) according to the normalized user throughputs for thecases of the different reconfigurableMD portions that is 010 50 70 and 100 of the entire 100 usersThe normal-ized user throughput has been obtained by normalizing thethroughput of each user with the maximum user throughputAs shown in Figure 10 when the entire user group consistsof purely legacy MDs for instance the Rx throughput ofnearly 70 of the entire users is less than 60 of that of themaximum user throughput In contrast when the entire usergroup consists of the reconfigurable MDs only 30 of theentire user suffers from the low throughput that is 60 ofthat of the maximum user throughput In other words theother 70 of the entire users can enjoy the Rx throughput ofhigher than 60 of that of the maximum user throughputFrom Figure 10 it can be concluded that more number of

the reconfigurable MDs brings about more number of userssatisfying the QoS

6 Conclusion

In order to fully exploit the merits of LSA the configurationof MD should be adjustable to the RA adopted in the LSAbandThis paper shows the performance evaluation of recon-figurable MD in terms of system throughput in comparisonto legacy MD in a preset test signal environment For experi-mental tests we implemented a prototype of reconfigurableMD with a system architecture that is compliant with theETSI-standard reference architecture suggested by WG2 ofETSI TC-RRS [13]The prototypeMD has been implementedusing NVIDIA GeForce GTX Titan GPU and USRP N210 asits modem and RF transceiver respectively In order to setup the configuration of MD in accordance with the radioapplication adopted in the LSA band we also developed asystematic procedure for transferring control signals amongthe software entities defined in the reference architectureThe procedure shown in this paper is based on the usecase of expanding bandwidth using LSA released by WG1of TC-RRS of ETSI in [9] Through the experimental testsperformedwith the prototypeMD and computer simulationsin a simple test environment it has been verified that thereconfigurability of MD is a necessary condition for LSAtechnology to fully obtain its benefits

Competing Interests


Acknowledgments

This research was supported by the MSIP (Ministry ofScience ICT amp Future Planning) Korea under the ITRC(Information Technology Research Center) support program(IITP-2015- H8501-15-1006) supervised by the IITP (Institutefor Information amp Communications Technology Promo-tion)

References

[1] Cisco Visual Networking Index Global Mobile Data TrafficForecast Update 2012ndash2017 vol 6 2013 White Paper

[2] E Hossain and M Hasan ldquo5G cellular key enabling tech-nologies and research challengesrdquo IEEE Instrumentation andMeasurement Magazine vol 18 no 3 pp 11ndash21 2015

[3] W Roh ldquo5G mobile communications for a connected worldand recent RampD resultsrdquo in Proceedings of the Smart RadioSymposium Seoul Republic of Korea June 2015

[4] M Matinmikko H Okkonen M Palola S Yrjola P Ahokan-gas and M Mustonen ldquoSpectrum sharing using licensedshared access the concept and its workflow for LTE-Advancednetworksrdquo IEEEWireless Communications vol 21 no 2 pp 72ndash79 2014

[5] K Jamshid et al ldquoLicensed shared access as complementaryapproach to meet spectrum demands Benefits for next gener-ation cellular systemsrdquo in Proceedings of the ETSI Workshop on


Reconfigurable Radio Systems Cannes France December 2012[6] ldquoElectronic Communications Committee (ECC) Report 205rdquo

Licensed Shared Access (LSA) 2014[7] M Matinmikko M Palola H Saarnisaari et al ldquoCognitive

radio trial environment first live authorized shared access-based spectrum-sharing demonstrationrdquo IEEE Vehicular Tech-nology Magazine vol 8 no 3 pp 30ndash37 2013

[8] M Mustonen T Chen H Saarnisaari M Matinmikko SYrjola and M Palola ldquoCellular architecture enhancement forsupporting the european licensed shared access conceptrdquo IEEEWireless Communications vol 21 no 3 pp 37ndash43 2014

[9] ETSI TR 103113 Mobile Broadband Services in the 2300ndash2400MHz Frequency Band under Licensed Shared AccessRegime vol 111 2013

[10] ETSI TS 103 235 ldquoSystem requirements for operation ofMobileBroadband Systems in the 2 300MHzndash2 400MHz band underLicensed Shared Access (LSA)rdquo V111 2014

[11] ETSI ldquoSystem architecture and high level procedures foroperation of Licensed Shared Access (LSA) in the 2300MHzndash2400MHz bandrdquo ETSI TS 103235 2015 v0012

[12] ETSI TS 136 101 LTE Evolved Universal Terrestrial RadioAccess (E-UTRA) User Equipment (UE) Radio Transmission andReception vol v1270 2015

[13] ETSI EN 303 095 Reconfigurable Radio Systems (RRS) RadioReconfiguration related Architecture for Mobile Devices volv121 2014

[14] ETSI TS 103 146-1 Reconfigurable Radio Systems (RRS) MobileDevice Information Models and Protocols Part 1 MultiradioInterface (MURI) vol v111 2013

[15] ETSI TS 103 146-2 Reconfigurable Radio Systems (RRS) MobileDevice Information Models and Protocols Part 2 ReconfigurableRadio Frequency Interface (RRFI) vol v111 2015

[16] M Mueck V Ivanov S Choi et al ldquoFuture of wireless commu-nication RadioApps and related security and radio computerframeworkrdquo IEEE Wireless Communications vol 19 no 4 pp9ndash16 2012

[17] ETSI ldquoReconfigurable Radio Systems (RRS) multiradio inter-face for Software Defined Radio (SDR) mobile device architec-ture and servicesrdquo ETSI TR 102839 2011 v111

[18] httpwwwubuntucom[19] ETSI TS 136 101 ldquoLTE Evolved Universal Terrestrial Radio

Access (E-UTRA) User Equipment (UE) radio transmission andreception (3GPP TS 36101)rdquo v1060 2012

[20] httpwwwgeforcecomhardwaredesktop-gpusgeforce-gtx-titan

[21] httpwwwettuscomproductdetailsUN210-KIT[22] C Ahn S Bang H Kim et al ldquoImplementation of an SDR

system using anMPI-based GPU cluster forWiMAX and LTErdquoAnalog Integrated Circuits and Signal Processing vol 73 no 2pp 569ndash582 2012

Research ArticleLicensed Shared Access System Possibilities for Public Safety

Kalle Laumlhetkangas1 Harri Saarnisaari1 and Ari Hulkkonen2

1Centre for Wireless Communications University of Oulu 90014 Oulu Finland2BittiumWireless Ltd Tutkijantie 7 90570 Oulu Finland

Correspondence should be addressed to Kalle Lahetkangas kallelaeeoulufi

Received 11 March 2016 Accepted 30 May 2016


Copyright copy 2016 Kalle Lahetkangas et al This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

We investigate the licensed shared access (LSA) concept based spectrum sharing ideas between public safety (PS) and commercialradio systemsWhile the concept of LSA has beenwell developed it has not been thoroughly investigated from the public safety (PS)usersrsquo point of view who have special requirements and also should benefit from the concept Herein we discuss the alternativesfor spectrum sharing between PS and commercial systems In particular we proceed to develop robust solutions for LSA use caseswhere connections to the LSA system may fail We simulate the proposed system with different failure models The results showthat the method offers reliable LSA spectrum sharing in various conditions assuming that the system parameters are set properlyThe paper gives guidelines to set these parameters

1 Introduction

The wireless operators should prepare for 1000 times growthin mobile data over the next 10 years [1 2] This growthis giving pressure for governmental spectrum users whichrarely utilize their spectrum to free up their frequenciesfor commercial use In the United States 500MHz of thespectrum from the federal and nonfederal applications isgoing to be freed completely or by spectrum sharing forcommercial mobile radio systems by the year 2020 [3] Thismay be the direction also in Europe The main interest in theUnited States for spectrum sharing is the spectrum accesssystem (SAS) [3] For spectrum sharing in Europe licensedshared access (LSA) [4ndash7] has gained interest since the LSAsystems can be made operator-specific More specifically theoperators of every country can agree on their own spectrumutilization between the possible secondary users LSA hasbeen proposed as an option for sharing the spectrum with PSin [8]

This work extends our work in [9] and first gives anoverview of how special applications such as public safetyshortly PS hereafter and other governmental users fit intothe possibilities of spectrum sharing with LSA and how toprepare for it The PS has a wide range of different users

and applications needing the spectrum The users are forexample first responders police firefighters border controlandmilitary which are vital for the society One of the criticalissues in deploying commercial technology to these kinds ofspecial applications is the ownership of the spectrum Forexample by the PS being an LSA licensee it can obtain thelegal right to utilize additional LSA spectrum resources whenthey are available Note that the PS can also be an incumbentof other predetermined frequencies for guaranteed resourcesWhile there are multiple choices for PS to utilize spectrumsharing it is also a political decision how the spectrum willbe shared Spectrum sharing principles for public safety havebeen categorized in five different sharing models in [10] andthe spectrum sharing has been extensively studied further in[11] There is also ongoing work on use cases for synergiesbetween commercial military and public safety domains in[12] We examine sharing approaches in the means of ownedspectral resources and their advantages and disadvantages Toour knowledge this issue has not been considered previouslyalthough it may be one of those steps that are needed for therelease of spectrum with LSA and for system developmenttherein

After the review of this novel topic our second contri-bution is planning a more specific system where the PS is



an LSA licensee for LSA spectrum resources Importantly ifthe PS utilizes LSA spectrum resources the PS requires thesharing process to be robust against connection problemsThe fall-back measures for the LSA system are generallypresented only on a high level [7] and they are still in theplanning phaseWhile the LSA systemhas been implementedand demonstrated in the project [4] the trials have not yetincluded any connection breaks inside the LSA system Ourobjective is to plan a system that can be tested in a liveenvironment More specifically we design a highly robustLSA system to be implemented with current commercialtechnology and equipment By robust it is meant that theproposed system is resilient to connection breaks in the LSAsystem that may be reality in real life due to electric breaksand so forth that is in the cases where the PS services areoften needed

We validate our proposed spectrum reservation methodvia simulations We study the duration of time intervalsbetween connection checks for noticing connection breaksand the effect of doing the resource reservations a predeter-mined time before the incumbent transmissions These arethe main system design parameters and the aim is to giveguidelines for selecting them properly

The paper is organized as follows In Section 2 we gothrough the different spectrum sharing possibilities withcommercial domain and PS In Section 3 we present a systemmodel of an LSA system to be built in a live network forthe PS and the key functionalities of the system componentsto overcome connection breaks In Section 4 we presentvalidating simulation results of the LSA systemWe concludethe paper in Section 5

2 Spectrum Sharing Possibilities

In this section we provide an overview of alternatives for thespectrum sharing in the case of PS and a commercial system(CS) The truth is that the PS might not always use their fullspectrum and it might remain available most of the timeat least locally Examples are police patrolling where just asmall voice service part of the spectrum needs to be reservedand military users that often in peace time need large partof the spectrum only in exercises and in special exerciseareas Naturally in the case of increased threat they need itin patrolling in the cities and so forth The temporally andspatially available spectrum could be used for other purposesat those times unused by the PS assuming it will be releasedimmediately back to the PS when needed For example thenonused spectrum can be used to speed up CS transmissionsfor example to ease rush hour data traffic naturally this is ofinterest in areas that have a high mobile traffic and that arenot in isolated areas

In addition the PS may also need complementary oradditional resources for its events and thus it would bebeneficial for them to get spectrum from CSs For examplewhen there is a large fire in a city the demands of the PS userscan grow dramatically especially if they would like to use newservices like live video streaming connections to data bases tocollect information about the area and social media to alarm

people In that case the PS requires their full spectrum andpossibly even more With spectrum sharing the additionalspectrum can preferably be obtained from silent commercialdevicesThe target spectrum bands considered are any bandsthat can be exploited by the PS for example the bandsof mobile operators and wireless camera and microphonesystems

In Figure 1 we plot different options for spectrum sharingin the means of owned spectral resources The differentoptions for allowing the other entity to use the spectrum aredepicted with arrows All the approaches can be grouped asfollows First the sharing framework is designed so that theCS users are the LSA licenseesThis way incumbent is alwaysallowed to use the spectrum and the CS obtains additionalspectrum Second the CS is incumbent and complementaryspectrum is given to the LSA licensee such as the PS Thirdoption is that all the users are using the CS Note that theseideas can also be used in parallel in different situations andareas We briefly list the above spectrum sharing system pos-sibilities and their advantages and disadvantages as follows

The PS Owns a Relatively Wide Spectrum (See Figure 1(a))

(1) The incumbent PS allows CS to use all its spectrumIn some areas where the incumbent does not usuallyhave activity allowing is more or less naturally per-manent In cities the incumbent activity can be morefrequent and allowing happens on a faster time scale

(2) The incumbent PS allows CS to use its free spectrumThe incumbent system might not need the entirespectrum but only parts of it Thus the remainingavailable spectrum can be utilized by the CS

(+) The incumbent has all the control for spectrumutilization

(+) The incumbent has a predictable quality for its appli-cations

(+) CS obtains additional spectrum(minus) No guaranteed additional resources for CS(minus) CS need devices that work using the spectrum of the

incumbent

CS or Other Applications Own the Majority of the Spectrum(See Figures 1(b) and 1(c))

(1) CS gives its available spectrum to the PS (Figure 1(c))(2) CS has the obligation to give enough spectrum to

the other system using the spectrum during criticaloperations (Figures 1(b) and 1(c))

(3) CS has the responsibility to give all its resourcesincluding physical equipment to PS during criticaloperations

(4) Some spectrum can be given for CS by the othersystem but as a tradeoff they can be demanded togive their spectrum to the other system in highlycritical situations


PS CS(1)

(2)

(3)

PS owns a relatively wide spectrum

(a)

LSA (CS)

(2)

(3)

Inc PS owns a narrow spectrum

Inc

(PS)

(b)

Inc (CS)(1)

(3)

LSA licensee PS owns a narrow spectrum

LSA(PS)

(c)

CS PS

PS is a customer for CS

PS sub CS

(d)

Figure 1 We have different options for spectrum sharing We use Inc as an abbreviation for the incumbent of the system (a) The PS ownssufficient number of spectra to support all of its requirements (b)The incumbent PS has only the critical number of spectra and CS has a widespectrum (c) The PS is LSA licensee of CS After the overview we concentrate more specifically on this setting where CS allows spectrumuse to PS (d) The incumbent is a roaming user at the CS network (1) CS allows spectrum use (2) PS allows spectrum use (3) CS is allowedto use the spectrum given that CS is obligated to give spectrum when needed

(+) The LSA licensee obtains additional resources for itsapplications

(minus) If CS is obligated to give spectrum to the other userthe CS cannot have guaranteed resources

CS Has a Complete System (See Figure 1(d) Users Such as PSUtilize the CS Network)

(1) All of the spectrum users PS and CS can be roamingusers of the CS network

(2) The PS can rentobtain the CS network for their ownuse

(+) The PS obtains instant coverage(+) The CS is constantly developing its network(minus) The PS does not have complete control over the CS

network(minus) The systemneeds a priority protocol if the incumbent

users are PS users(minus) There is no coverage or support for all the applications

at every location The PS still needs their own servicein the areas where the CS network cannot support it

(minus) The PS has to trust CS and their security when beingan CS user

The current state of the affair is that the PS and CS havetheir own spectrum and they do not cooperate Here toobtain similar functionalities as the CS the PS requires equalamount of spectrum as CS The first step to this setting iscooperation as illustrated in Figure 1(a) Naturally sharingrules have to be agreed on that is CS PS or both allow

their spectrum to be used by the other one In the followingsubsections we go through the options for spectrum sharingin more detail for LSA systems

21 PS Is the Incumbent In this subsection we consideroptions for when the PS is the incumbent in an LSA systemas for example in Figures 1(a) and 1(b) Here a part of thePS spectrum has been released for CS under the requirementthat they must allow the incumbent PS to use that spectrumwhen and where needed Obviously this situation requiresa political decision but it is listed here as an opportunityIt is discussed in the US that in this scenario the CS andother users can share the spectrum as secondary users [3]Moreover in the US a wide bandwidth of spectrum will bereleased from governmental users to CSs in the upcomingyears Note that the majority of spectra can still be used bythe PS during critical operations

By being the incumbent the PS has all the controlto support its critical and noncritical applications witha predictable quality Here the PS can build its networkinfrastructure and the management system for organizing itsnetwork and services However the PS might not build anationwide network for itself Moreover the PS might notuse its spectrum all the time This leads to free spectrumwhich can be utilized by other applications A possibility isto cooperate with a CS The additional spectrum could beused as a complementary resource by theCS to unload its datatraffic There are multiple possibilities for cooperation

First the PS can allow the CS to use the spectrum atpredetermined times and areas This is applicable when thepossible PS spectrum usage is known in advance This is


the case for example when the PS has scheduled theiroperations In these cases the PS can have the spectrum forthe reserved time and area even if they are not using itWith this method the spectrum is free at given times andthe individual PS users do not need to worry about the CStransmitting at the same timeThis is applicable for examplein some of the military training scenarios and in borderprotection as the military is mostly using their spectrum inknown areas during peace time

As a second option the PS can allow the CS to use thespectrum at all the times when the spectrum is free Thisoption needs a rapid method for the spectrum reservationHere the PS should preferably notify the LSA repository afew moments before the transmission so that the spectrumcan be guaranteed to be free for the PS Another possibilityis for the PS to notify the LSA repository when the trans-mission begins In this setting the PS should accept possibleinterference from the LSA licensee in the beginning of itstransmission Moreover in the scenarios above the fall-backmeasures to handle connection breaks for guaranteeing thepossible incumbent transmission should be expeditious

Third the PS can allow the CS to use the spectrum at thelocations where the spectrum is not currently needed by thePS usersThis option can be accomplished by tracking the PSusers and by reserving the necessary spectrum for them attheir locations This is applicable for example with the firstresponder units whose locating is important also from theoperational perspective

Fourth depending on the applications the PS might notalways need all of its frequencies The PS can allow the CSto use the remaining free frequencies Here the spectrumband can be divided into multiple smaller bands that can beaccessed with the CS according to the need of the PS users

Moreover any combination of the above is also possibleIn these systems however the spectrum is a complementaryresource for the CS when the PS users are silent To startbuilding the system the agreements between the incumbentPS and commercial LSA licensees can be first allowed insmaller areas Then if the CS is able to develop theirapplications in such a way that they do not cause intolerableinterference to the PS operations the agreements are easy toexpand to wider areas

The amount of gain obtained by the CS depends on theactivity of the PS For example if the PS is silent most ofthe time the CS obtains the spectrum most of the time Thegreatest benefit for the PS by owning the spectrum is thecontrol It is possible for the PS to freely use the spectrumfor its own applications In addition it is always possibleto decline the spectrum use of the CS or other spectrumusers However the resources owned by the PS might stillnot be enough to support all the PS operations Moreoverthe PS might not want to reserve a wide spectrum for itsapplications Thus it may be beneficial for the PS to alsoobtain additional resources and services from the CS whenneeded

22 CS Is the Incumbent In this subsection we consideroptions for when the CS is the incumbent in an LSA system

as shown in Figure 1(c) The CS has a wide spectrum andis giving spectrum resources to the PS which only has asmall portion of spectrum reserved for example to voicecommunication Later in this work we will concentrate onlyon this scenario in developing an LSA system for the PSThere are multiple possibilities for cooperation which can allbe implemented in parallel depending on the needs by the PS

First the resources can be shared with an LSA systemWhen the incumbent user comes to the area PS will retreator change its frequency This suits the case when the PS ismostly using the spectrum in the area where the CSs orother incumbent users remain silent This is applicable if thePS uses spectrum mainly for noncritical applications suchas training and has the authority to reserve the spectrumcompletely for itself during critical operations for obtainingspectrum This is the use case for example in military andborder control applications where the PS would requirespectrum for their communication during peace time ThesePS operators can agree onmultiple LSA agreementswithmul-tiple incumbents to obtain multiple spectrum bands Thenthey are able to legally utilize the band that is available WithPS being the LSA licensee the PS users do not necessarilyneed to inform their location to the LSA repository andthe PS users are not tracked for spectrum information Thistype of LSA sharing method brings security in some PSapplications where the location of PS operators should bekept as a secret Another example of resource sharing likethis is a high speed mobile network for the PS at sparselypopulated training areas This kind of high speed networkscan also offer a backup mobile infrastructure for examplein disaster areas and in rescue operations during electricalshortages when a commercial network of the CS is down

Second the CS can be obligated to give spectrum to thePS in areas that are not covered by the CS network Thusthe PS can obtain spectrum for its own use here that is fortraining and for emergency use This option is applicable inthe long termonly if theCS is not building its network in theseareas for example if these areas give no financial benefitOtherwise there is no long-term guarantee of interference-free spectrum for the PS

Third the CS has the obligation to give required spectrumto the PS during critical operations Here the PS can havethe rights of the incumbent during critical operation This isa viable option when the PS is mainly a minor user of thespectrum and critical operations happen rarely The CS canbuild its network using a wide spectrumThen the spectrumis released when the PS users come to the area and need itThis option would require a backdoor for PS to be installedto CS equipment For example by using the backdoor the PScould reserve spectrum or switch off related CS base stationswith alarm signals or via central controller In some PS casesthe spectrum can also be reserved in advance by the basisof the emergency calls which usually happen via CS basestations and near the locations of the required PS needs

23 PS Utilizes CS Network One additional option on theabove scenarios is the following As shown in Figure 1 thePS users can be the roaming users of the CS network [13 14]


LSA server

LSA controller

LSA repository

LSA licenseeAP (PS)

Incumbent manager via IP network

IP network

Closed network

Incumbent

Figure 2 A wireless camera uses the spectrum with LSA licensee that has LSA controllers at every AP

Here the entire spectrum is owned by CS and it is responsiblefor building the network However in order for the PS to beindependent of CS networks a backup system for the mostcritical applications and communication is still needed Notealso that this option is not spectrum sharing in the means ofLSA but is listed here as an opportunity

When the PS users are roaming users at the CS networkthey need priority over the CS users Here the PS shouldobtain the highest priority for its critical applications Inaddition when the PS users are roaming users at the CSnetwork the CS operator needs to be able to support PSapplicationsThe benefit of being a roaming user is the instantcoverage of the CS network in densely built areas Anotherbenefit is that the CS develops its spectrum usage to meet thecurrent requirements better because it is competing for usersHowever the PS does not have full control over the networkwhich reduces the security Moreover there needs to be solidencryption for the PS and the CS network should be builtrobustly

3 System Model

Next we concentrate more specifically on developing the LSAsystem for the PS which acts as an LSA licencee for accessibleLSA spectrum resources as discussed in Section 22 The PSuse case considered here is only for noncritical applicationsThe proposed resource allocation method builds on previousLSA work in [15 16]

We consider an LSA system with an LSA repository LSAcontrollers an LSA licensee and an incumbent user Thesesystem elements and their connections are shown in Figure 2The incumbent is the primary user of the LSA spectrumresources We consider the incumbent to be for exampleemployees of programmemaking and special events serviceswhich are defined in [17 18] The LSA repository collects

maintains and manages up-to-date data on spectrum useThe LSA licensee is a secondary user with a license toutilize the spectrum when incumbent user is silent TheLSA licensee has multiple access points (APs) that utilize theresources The LSA licensee has a network that connects theAPs together In contrast to [15] with one LSA controllerevery AP of PS has its own distributed LSA controllerThus no single device is solely responsible for the spectrumallocations

We also introduce an LSA server to the system The LSAserver is a mediator between the LSA repository and the LSAcontrollers By using a mediator the PS network can be keptclosed from the IP network which provides security Herethe LSA server is the only device of the PS network that canbe connected from the outside The LSA server reports onlythe necessary network information from the LSA licenseenetwork to the LSA repository

The spectrum sharing between the users operates asfollows Incumbent user reserves the spectrum at least apredetermined time before using the spectrum contrary tothe on-demand operation mode for LSA spectrum resourcereservation [6] Thus during a connection break the mostrecent information is still valid for the predetermined timeThe incumbent reserves the resources by connecting the LSArepository with an incumbent manager Then the repositorysends notification of the spectrum reservation to the LSAserver After the LSA server obtains spectrum reservationinformation it forwards the information to the LSA con-trollers of affected APs Finally the LSA controllers computethe protection criteria of incumbent and control the spectrumusage of the APs

In Figure 3 we present more precisely how to implementthis system in a real Long-TermEvolution (LTE) networkWedepict the components and their connections Here LTE APs(eNodeBs) of PS utilize the spectrum as an LSA licensee ThePS has its own closed LTE network where the backhaul is


IP network

Tactical router

LTE access point

(eNodeB)S1

LSA repository

LSA server

Tactical network

Incumbent

transmitterreceiver

Tactical router

LTE access point

(eNodeB)

S1

Incumbent manager

IP network

Lite-EPCDistributed LSA

controller dOMS


controller dOMS

IP network

Figure 3 Two LTE access points in LSA licensee network

built with tactical routers In addition to wired links theserouters also support radio link connections [19] They canalso automatically reroute any given data from the source tothe destination via alternative routes given that the primaryroute fails Every AP is connected to the closed networkvia a lite-EPC and a tactical router The lite-EPCs provideLTE hot spots to the network and emulate the evolvedpacked core functionalities of an LTE network The accesspoints are connected with S1 interface to the lite-EPC Thecomputer with the lite-EPC works also as a distributed LSAcontroller The LSA system components communicate witheach other using http(s) with representational state transferarchitechture The data is formatted using JavaScript objectsWe go through the main functions of the main componentsin the following subsections

31 Incumbent via Incumbent Manager Incumbents of oursystem use a http(s)-based incumbent manager to inform therepository of their spectrum access The reservation messageincludes ldquostartingrdquo and ldquoendingrdquo time of the incumbentstransmission the reserved frequencies (center frequenciesand bandwidths) the location and the type of the usage Thereservation information is used to calculate the protectionzone for incumbent

The incumbent manager allows reserving the spectrumonly for a predetermined time beforehand More specificallyincumbent has to send a reservation message via incumbentmanager to the LSA repository at least a predetermined time119879

119894before its transmission This time can vary for different

types of users Additionally the requirement for reservationof a predetermined time before the incumbent transmissioncan also be voluntary in some of the systems Then ifthe incumbent does not reserve the spectrum on time it

is obligated to possibly tolerate interference from the LSAlicensee for the predetermined time given that there areconnection breaks

32 LSA Repository The LSA repository keeps a database ofup-to-date information about incumbent spectrum reserva-tions and about the conditions for utilizing the spectrumTheLSA repository forwards information about incumbent andits planned use of LSA spectrum resources to the LSA serverwhen the information becomes available The informationsent from the repository also includes the time when it issent The LSA repository can also reply to a request for theincumbent information This reply includes the informationthat is new to the requesting device

Connection checks to the LSA repository happen viaheartbeat signals The devices which check the connectionrequest heartbeat signals periodically from the LSA reposi-tory The LSA repository replies to a heartbeat request witha heartbeat signal If there is no response the connection isbroken Heartbeat response signals include the timewhen theheartbeat response signal is sent

33 LSA Server The LSA server acts as an LSA controller tothe LSA repository It has a strong firewall for separating thePS network from the IP network After obtaining incumbentinformation from the LSA repository the LSA server broad-casts this information to the distributed LSA controllersThe LSA server also saves incumbent information until theinformation expires To obtain robustness for connectionbreaks to this setting any tactical router could act as an LSAserver given that it has an Internet access and given that it hasa programmable interface

The LSA server sends heartbeat requests to the LSArepository between time intervals of 119879check The heartbeatresponses are then forwarded to the LSA controllers TheLSA server notices a connection break to the LSA repositoryif there is no heartbeat signal within time 119879timeout fromthe heartbeat request When this kind of connection breakoccurs the LSA server sends heartbeat failure signals to thelite-EPCs periodically between time intervals of 119879check Thesesignals provide the LSA controllers information whether theconnection break is external or internal

The LSA server tries to reconnect to the LSA repositoryduring a connection break The LSA server requests up-to-date incumbent information from the LSA repository whenbecoming connected to it The LSA server can also answerto a request for incumbent information and replies with theinformation that is new to the requesting device

34 LSA Controller in Lite-EPC Computer The LSA con-trollers control the spectrum utilization of the PS Theyreceive the incumbent information from the LSA serverwhenit becomes available Additionally an LSA controller requestsfor up-to-date incumbent information from the LSA serverwhen becoming connected to the PS network All of the LSAcontrollers save the received incumbent information until itexpires The main task for an LSA controller is to calculatethe protection zone for the incumbent using incumbent


information The calculation is done similarly at every LSAcontroller using the same algorithms as in the centralizedcontroller developed by the project [4] However a lite-EPCcontrols only the AP that is connected to it

35 Distributed Operations Management System We havedepicted distributed operations management system as(dOMS) in Figure 3 The dOMS are distributed per AP andalso work in the same computers as the lite-EPCs Theyare responsible for sharing the spectrum between the otherAPs and include command tool for controlling the AP andthe necessary commission plans with a site manager forvalidating the plans Each of the individual dOMS sendscommand messages to their own APs for the frequencyallocations and power levels In other words every unit ofdOMS controls only their own AP but decides the spectrumsharing together with other units of dOMS

The spectrum sharing between APs is done in dOMSthat keep a list of APs in the vicinity To share the LSAspectrum resources the dOMS utilize signaling methodssimilar to coprimary spectrum sharing [20]The difference to[20] is that the spectrum sharing is done between a single PSoperator without the need to compete with other operatorsThe signalingmessages are sent inside the closed PS network

The dOMS has the task to clear the spectrum beforeincumbent utilizes the spectrum and when the spectrumreservation information becomes invalid due to a connectionbreak Recall that the sending times are included in all ofthe data originating from the LSA repository The spectrumreservation information is valid for time 119879

119894after a successful

heartbeat signal or any other data is sent from the LSArepository

Let 119879empty be the time that it takes to empty the spectrumby the AP after a command from the dOMS If no heartbeatsignal or other data arrives from the LSA repository theLSA spectrum resources are freed after time 119879

119894minus 119879empty from

the sending time of the last successful data from the LSArepository The spectrum can be emptied immediately orgradually by using graceful shutdownwhich gradually lowersthe power level of the APs The dOMS can also order its APto utilize some available backup frequency Alternatively anyother fall-back measure [7] can be used

4 Simulation Setup and Numerical Results

In this section we present our simulation setup and resultsfor our LSA system We use simulations to validate thespectrum reservationmethod setup in the case of connectionbreaks inside the IP network We assume that the closedPS network is built reliably This means that there are noconnection breaks inside the PS network The incumbentis also assumed to utilize the LSA spectrum resources onlyafter a successful reservation This is a conventional methodfor incumbents such as programme making and specialevents services which are required to inform their spectrumutilization to a national telecommunications regulator Theconnection breaks in the LSA systemoccurs in the IP networkbetween the LSA repository and LSA controllers We assume

that the APs of PS with the same frequency are at a longdistance from each otherWe also assume that the APs whichare near each other utilize different frequencies as usualThus no dynamic spectrum sharing is simulated

We use spectrum utilization and valid spectrum knowl-edge of the LSA licensee to measure the performance of theLSA system The latter measure tells us the ratio of time thatthe spectrum reservation information is valid with respectto the total simulation time For example when the valueof it is 05 the spectrum reservation information is valid for50 of the time Recall that the LSA licensee utilizes the freespectrum only when the spectrum knowledge is valid Thusthe incumbent and the LSA licensee share the LSA resourcesperfectly only during this timeTherefore the amount of validspectrum knowledge reflects the LSA system performanceIt also relates directly to the reliability of the LSA systemas the spectrum can be utilized by the LSA licensee duringconnection breaks if the spectrum knowledge is valid

We show how our LSA system design parameters 119879checkand 119879

119894 affect the performance in different network scenarios

with different incumbent activity levels We simulate everyscenario over 1000 iterationswith different connection breaksand incumbents for average results In every scenario wedraw the durations of the incumbent transmissions andconnection breaks from Poisson distributions We draw thenumber of incumbent transmissions and connection breaksfrom normal distributions where the negative values are setto zero The starting times of incumbent user transmissionsand connection breaks are uniformly distributed The ratio-nale for using these simplifying distributions is to obtain first-level insights into our protocol behavior when using differentdesign parameters in different scenariosThe total simulationtime is 12 hours The time to empty spectrum with an orderfrom the dOMS 119879empty is 30 seconds The delay to transmitdata from the LSA repository to the LSA controllers is threeseconds when the connection is working

We model the IP network connection breaks for differentscenarios as follows We model three types of networkconnections They are reliable mediocre and poor and theparameters to simulate them are shown in Table 1 The lastcolumnConnection OK shows the quality of the connectionthat is the ratio of time that the connection is workingbetween the LSA repository and LSA controllers with respectto the total simulation time These ratios are also a pointof reference for valid spectrum knowledge in the currentlyavailable LSA systems More specifically in the current LSAsystems the spectrum is shared perfectly only when theconnection is working The rationale for simulating lowconnection reliabilities comes from the fact that the PS shouldremain functional when the commercial IP networks haveserious connection problems

Similarly wemodel the incumbent activity for three typesof incumbentsThe incumbent types are rare occasional andactive and the parameters to simulate them are shown inTable 2The last column spectrum utilization shows the ratioof time that the incumbent utilizes the spectrumwith respectto the total simulation time


Table 1 The parameters for simulating the connection quality

Mean of connection breaks Variance Mean duration of a connection break Connection OKReliable 0 2 5min 099Mediocre 7 2 20min 073Poor 15 2 60min 029

Table 2 The parameters for simulating the incumbent activity

Mean of transmissions Variance Mean transmission time Spectrum utilizationRare 0 2 40min 006Occasional 5 2 40min 026Active 12 2 40min 050

In the next simulations we study the LSA system perfor-mance with respect to 119879check Recall that the value of 119879check isthe time between heartbeat signal requests

In Figure 4 the incumbent notifies about itself 15minutesbefore its transmission that is 119879

119894= 15min From Fig-

ure 4 we observe that the spectrum knowledge for reliablemediocre and poor internet qualities is higher than 9973 and 29 which are the corresponding percentages oftimes for internet connection working Thus the spectrumcan be utilized by the LSA licensee even during some of theconnection breaks with our reservation method Moreoverwe see that the quality of the internet connection is importantwhen the incumbent informs about its spectrum utilizationon a short notice

From Figure 4 we also see that the spectrum knowledgeby the LSA licensee is higher when 119879check is low that is whenthe connection to the LSA repository is checked more oftenThis is because then it is more likely to get an answer from therepository for validating the connection Therefore with anunreliable internet connection the value of 119879check should beas low as possible to have themost valid spectrumknowledgeHowever from the figure we also see that it is more importantto have a good internet connection than to make the value of119879check as low as possible

In Figure 5 the incumbent notifies about itself 60minutesbefore its transmission that is119879

119894= 60minWhen comparing

this figure to Figure 4 we see that the spectrum knowledge isoverall better for every type of internet quality for a greatervalue of 119879

119894 We also can see that setting 119879

119894large is more

important in terms of spectrum knowledge than to set 119879checklow Moreover we observe that the spectrum is known forover 50 of the time when the internet quality is poor thatis when the internet connection is working 29 of the timeTherefore the 119879

119894should be large if the internet quality is low

From Figure 5 we see that the mediocre internet quality isallowable in this setting that is the spectrum can be utilized100 of the time when the 119879check is below 3 minutes Thusgiven that the internet connection to the PS network can bemediocre the PS should utilize frequencies of incumbentswhich are able to report their frequencies reliably in advanceMoreover if the internet connection is poor the PS requireseither additionalmethods for utilizing all of the free spectrum

0 2 4 6 8 10 12 140

01

02

03

04

05

06

07

08

09

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee

Reliable internetMediocre internetPoor internet

Tcheck (min)

Figure 4 The spectrum knowledge of the channel as a functionof 119879check while 119879

119894= 15min with different qualities of internet

connection The incumbent is rare that is it utilizes the channelapproximately 6 of the time

or an incumbent that reports its spectrum utilization evenearlier

In the next simulations we study the LSA system perfor-mance with respect to 119879

119894 with different types of incumbents

and internet qualities Recall that the value of 119879119894indicates the

predetermined time before which the incumbent is requiredto send its spectrum reservation to the LSA repository

In Figure 6 the incumbent is rare and the 119879check isset to be 15 minutes From Figure 6 we see a rise of thespectrum knowledge as a function of 119879

119894 This implies that

when the internet quality is poor the incumbent shouldreserve the spectrum as early as possible This is applicablefor incumbents that know their spectrum needs beforehandor rarely change their frequency allocations and have a static


0 2 4 6 8 10 12 140

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)

Figure 5 The spectrum knowledge of the channel as a function of119879check while 119879119894 = 60min The incumbent is rare

operation An example of this kind of incumbent is anorganizer of programme making special events

In Figure 7 we study how different activity levels of theincumbent affect the LSA system performance We observefrom the results that the spectrum knowledge is higher whenthe incumbent ismore activeThis is because then the incum-bent reserves the spectrum more often and the reservationsinclude the spectrum knowledge However if the incumbentis very active it might be hard for all incumbent applicationsto report the plans at a predetermined time before utilizingthe spectrum Thus the PS with a poor internet connectionshould utilize different methods such as sensing to obtainthe LSA resources with an active incumbent

In Figure 8 we plot the spectrum utilization of the LSAlicensee In this figure we compare the spectrum utilizationby the LSA licensee by using two measures First we plotthe utilized spectrum resources divided by all the resourcesSecond we plot the utilized spectrum resources divided bythe available resources that is the LSA resources that areavailable at the times when the incumbent does not transmitFrom the figure we see that the LSA licensee can utilizethe spectrum less often when the incumbent is more activewhile the available spectrum for the LSA licensee is utilizedrelatively better Therefore as natural it is always preferablefor the LSA licensee that the incumbent does not transmitMoreover the overall spectrum is utilized more effectivelywhen there are more incumbents

In Figure 9 we study the spectrum utilization of thecomplete LSA system This is the utilization of the spectrumby either the LSA licensee or the incumbent We plot theutilized spectrum resources divided by the total spectrumresources We see that the spectrum utilization is inlinewith the spectrum knowledge by the LSA licensee shown inFigure 7 The spectrum is utilized approximately 100 of the

20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)

Figure 6 The spectrum knowledge of the channel as a function of119879

119894while 119879check = 15min The incumbent is rare

20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee

Rare incumbentOccasional incumbentActive incumbent

Ti (min)


119894while119879check = 15minwith different incumbent activity levelsThe

internet connection ismediocre

timewhen the119879119894is over 80We can see that the proposed LSA

systemwithmediocre internet connection to the LSA licenseeis ideal for sharing the spectrum with incumbents such asmobile operators if they can reliably estimate their spectrumneeds 80 minutes beforehand

In Figure 10 we plot the utilized spectrum resourcesdivided by the total spectrum resources for different valuesof119879check with an occasional incumbent andmediocre internetNote that the value of 119879check affects only spectrum utilization


20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

lice

nsee

All resources rare incumbentAvailable resources rare incumbentAll resources occasional incumbentAvailable resources occasional incumbentAll resources active incumbentAvailable resources active incumbent

Ti (min)

Figure 8 LSA resource utilization by the LSA licensee as a functionof 119879119894while 119879check = 15min in amediocre channel

20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m


Ti (min)

Figure 9 LSA resource utilization by the LSA system as a functionof 119879119894while 119879check = 15min in amediocre channel

of the LSA licensee Thus from Figure 10 we notice that theLSA licensee receives more resources with smaller values of119879check This is because the LSA licensee knows more validspectrum information when it checks the connection moreoften However the amount of valid spectrum informationdoes not grow significantly when the 119879check becomes smallerthan 15 seconds From the figure we also see that the valid

20 40 60 80 100 12008

085

09

095

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m

Ti (min)

Tcheck = 15minTcheck = 11minTcheck = 7minTcheck = 3min

Tcheck = 1minTcheck = 15 sTcheck = 5 s

Figure 10 LSA spectrum resource utilization as a function of119879119894with

occasional incumbent in amediocre channel

information does not vary significantly for different values of119879check if the119879119894 is over 80minutesThus the value of119879check canbe set adaptively according to the value of119879

119894 that is according

to the predetermined time before which the incumbent sendsits spectrum reservation to the LSA repository

5 Conclusion

We gave an overview of spectrum sharing possibilitiesbetween PS and CS since there may be a possibility to findmore spectrum for their users in the future While thereare multiple choices for PS to utilize spectrum sharing it isalso a political decision how the spectrum will be sharedTherefore PS should be ready for every scenario If PSowns the spectrum it can rent the free spectrum to CSvia an LSASAS system Another option for providing highquality PS performance is the following We reserve only asmall portion of the spectrum for voice service to PS Welet CS networks utilize the remaining spectrum with thecondition that CS is obligated to release spectrum to PS whenneeded for critical applications We gave multiple options toautomatically reserveCS resources for PS use In addition thePS can be a roaming user at CS network Furthermore PS canbe an LSA licensee of the incumbent CS

Moreover if LSA sharing arrangement is used thereneeds to be a reliable method for spectrum allocation toPS during connection breaks We developed a specific LSAsystem for robustness to overcome short-term connectionbreaks In this system the PS is the LSA licensee and theCS is the incumbent which can be for example when thePS requires additional resources with LSA In our systemthe incumbent reserves the spectrum for a predetermined


time beforehand and is not transmitting during this predeter-mined timeWe validated the reservation system and studiedhow to select suitable durations for the predetermined timesand for time intervals between connection checks Thetime intervals between connection checks can be selectedadaptively based on the network quality and on the timebefore which the incumbent sends its spectrum reservationsThe simulations show that the proposed system is able toreduce the impact of possible connection breaks inside theLSA system

However this method is not alone sufficient for utilizingall the LSA spectrum resources during all connection breaksThere might be a long connection break and no possibilityfor an internet connection In addition the incumbent mightnot always have an internet connection but can still utilize thespectrumTherefore if the PS is an LSA licensee and requiresavailable LSA spectrum resources it needs to develop othermethods to guarantee its own error-free transmission andincumbent protection

To protect the incumbent without internet connectionthere can be additional signals that tell about a connec-tion break and that the incumbent is using the spectrumsuch as errors accumulating to the LSA licensees humanintervention at the base stations local reservation signalswith separate control channels and sensing methods In theupcoming work we will develop the LSA system to coexistwith the already available sensing methods and enable spec-trum sharing and utilization also during major connectionbreaks

Competing Interests


Acknowledgments

The authors would like to acknowledge CORE++ projectconsortium VTT University of Oulu Centria Universityof Applied Sciences Turku University of Applied SciencesNokia PehuTec Bittium Anite Finnish Defence ForcesFICORA and Tekes

References

[1] Cisco ldquoCisco visual networking index global mobile datatraffic forecast update 2015ndash2020rdquo Cisco White Paper 2014httpwwwciscocomcenussolutionscollateralservice-pro-vidervisual-networking-index-vnimobile-white-paper-c11-520862pdf

[2] ldquoThe 1000x mobile data challengerdquo Qualcomm Presentation2013 httpwwwqualcommcommediadocumentsfiles1000x-mobile-data-challengepdf

[3] The White House ldquoRealizing the full potential of government-held spectrum to spur economic growthrdquo Presidents Councilof Advisors on Science and Technology 2012 httpswwwwhitehousegovsitesdefaultfilesmicrositesostppcast spec-trum report final july 20 2012pdf

[4] Core++ project web page June 2016 httpcorewillabfi

[5] The Electronic Communications Committee ldquoLicensed sharedaccess (LSA)rdquo ECC Report 205 The Electronic Communica-tions Committee Copenhagen Denmark 2014 httpwwwerodocdbdkDocsdoc98officialpdfECCREP205PDF

[6] ETSI ldquoReconfigurable radio systems (RRS) System require-ments for operation of mobile broadband systems in the 2300MHzmdash2 400MHz band under licensed shared access (LSA)rdquoETSI TS 103 154V111 October 2014 httpwwwetsiorgdeliveretsi ts103200 103299103235010101 60ts 103235v010101ppdf

[7] ETSI ldquoReconfigurable radio systems (RRS) system architectureand high level procedures for operation of licensed sharedaccess (LSA) in the 2 300MHzndash2 400MHz bandrdquo ETSI TS103 235 V111 October 2015 httpwwwetsiorgdeliveretsits103200 103299103235010101 60ts 103235v010101ppdf

[8] ETSI ldquoReconfigurable radio systems (RRS) use cases forspectrum and network usage among public safety commer-cial and military domainsrdquo Article ID 102900 ETSI TR102 970 V111 2013 httpwwwetsiorgdeliveretsi tr102900102999102970010101 60tr 102970v010101ppdf

[9] K Lahetkangas H Saarnisaari and A Hulkkonen ldquoLicensedshared access system development for public safetyrdquo in Proceed-ings of the European Wireless Conference Oulu Finland May2016

[10] R Ferrus O Sallent G Baldini and L Goratti ldquoPublicsafety communications enhancement through cognitive radioand spectrum sharing principlesrdquo IEEE Vehicular TechnologyMagazine vol 7 no 2 pp 54ndash61 2012

[11] R Ferrus andO SallentMobile Broadband Communications forPublic Safety The Road Ahead Through LTE Technology JohnWiley amp Sons New York NY USA 2015

[12] ETSI ldquoReconfigurable radio systems (RRS) Feasibility studyon inter-domains synergies synergies between civil securitymilitary and commercial domainsrdquo ETSI TR 103 217 June 2016httpsportaletsiorgwebappworkProgramReport WorkItemaspwki id=43285

[13] ldquoUkkoverkot commercial servicerdquo June 2016 httpwwwukkoverkotfi

[14] R Hallahan and J M Peha ldquoEnabling public safety priority useof commercial wireless networksrdquo Homeland Security Affairsvol 9 article 13 2013 httpwwwhsajorgarticles250

[15] M Palola T Rautio M Matinmikko et al ldquoLicensed SharedAccess (LSA) trial demonstration using real LTE networkrdquo inProceedings of the 9th International Conference on CognitiveRadio OrientedWireless Networks (CROWNCOM rsquo14) pp 498ndash502 June 2014

[16] M Palola M Matinmikko J Prokkola et al ldquoLive field trialof Licensed Shared Access (LSA) concept using LTE networkin 23 GHz bandrdquo in Proceedings of the IEEE InternationalSymposium on Dynamic Spectrum Access Networks (DYSPANrsquo14) pp 38ndash47 McLean Va USA April 2014

[17] Electronic Communications Committee ldquoBroadband wirelesssystems usage in 2300ndash2400MHzrdquo ECCReport 172 2012 httpwwwerodocdbdkdocsdoc98officialpdfECCRep172pdf

[18] European Radiocommunications Committee ldquoHandbook onradio equipment and systems videolinks for ENGOB userdquo ERCReport 38 1995 httpwwwerodocdbdkdocsdoc98officialpdfREP038pdf

[19] Elektrobit ldquoEnhancing the link network performance with EBtactical wireless IP network (TACWIN)rdquo EB Defense Newslet-ter December 2014 httpwwwbittiumcomfilephpfid=785


[20] M Jokinen M Makelainen and T Hanninen ldquoDemo co-primary spectrum sharing with inter-operator D2D trialrdquo inProceedings of the 20th Annual International Conference onMobile Computing and Networking pp 291ndash294 September2014

Research ArticlePSUN An OFDM-Pulsed Radar Coexistence Technique withApplication to 35 GHz LTE

Seungmo Kim Junsung Choi and Carl Dietrich

Bradley Department of Electrical and Computer Engineering Virginia Tech Blacksburg VA 24060 USA

Correspondence should be addressed to Seungmo Kim seungmovtedu



Copyright copy 2016 Seungmo Kim et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

This paper proposes Precoded SUbcarrier Nulling (PSUN) an orthogonal frequency-division multiplexing (OFDM) transmissionstrategy for a wireless communications system that needs to coexist with federal military radars generating pulsed signals in the35 GHz band This paper considers existence of Environmental Sensing Capability (ESC) a sensing functionality of the 35 GHzband coexistence architecture which is one of the latest suggestions among stakeholders discussing the 35 GHz band Hence thispaper considers impacts of imperfect sensing for a precise analysis Imperfect sensing occurs due to either a sensing error by anESC or a parameter change by a radar This paper provides a framework that analyzes performance of an OFDM system applyingPSUN with imperfect sensing Our results show that PSUN is still effective in suppressing ICI caused by radar interference evenwith imperfect pulse prediction As an example application PSUN enables LTE downlink to support various use cases of 5G in the35 GHz band

1 Introduction

In 2010 the US National Telecommunications and Informa-tion Administration (NTIA) Fast Track Report [1] identifiedthe 3550ndash3650MHz band to be potentially suitable forcommercial broadband use The NTIA identified it as one ofthe candidate bands in response to the presidentrsquos initiative[2] to identify 500 megahertz of spectrum for commercialwireless broadband In 2012 the Federal CommunicationsCommission (FCC) released a Notice of Proposed Rulemak-ing (NPRM) [3] where they proposed creation of the CitizensBroadband Radio Service (CBRS)The FCC voted to approvethe suggestions developed through two NPRMs [3 4] andadopted rules for managing 150 megahertz in the 3550ndash3700MHz band (the 35 GHz band) in a report and order [5]

The FCC proposes structuring the CBRS according toa three-tiered shared access model comprised of IncumbentAccess (IA) Priority Access (PA) and General AuthorizedAccess (GAA) IA includes federal military radars and fixedsatellite service which are protected from PA and GAAPA operations are protected from GAA operations PriorityAccess License (PAL) three-year authorization to use a 10-megahertz channel in a single census tract will be assigned

in up to 70 megahertz of the 3550ndash3650MHz portion of thebandGAAusewill be allowed throughout the 150-megahertzband GAA users will receive no protection from interferenceof other CBRS users There exist spectrum access systems(SASs) incorporating a dynamic database and interferencemitigation techniques A SAS collects pulse parameters ofthe incumbent radars and provides them with the coexistingCBRS devices In many cases a SAS may not be able toprovide such information directly to the CBRS users due tosecurity concerns related to military radar systems Then aSAS provides such information in an indirect manner forexample query responses to the CBRS users

The NTIA recommends addition of Environmental Sens-ing Capability (ESC) a component for sensing capability[6] The NTIArsquos review of the public record indicates thatmany stakeholders proposed employing sensing techniquesto augment capability of a SAS The inputs from the ESC canbe used by the SAS to direct the PA and GAA tier users toanother channel or if necessary to cease transmissions toavoid potential harmful interference to federal radar systems

In addition the FCC recommends in [3 4] the CBRSsystem to be a small-cell system where each transmitter cankeep its transmitting power low The most popular examples



of small-cell systems so far in practice are Wireless Fidelity(Wi-Fi) and the 3rd Generation Partnership Project (3GPP)Long-Term Evolution (LTE) To the best of our knowledgeit is more challenging to design a small-cell system based onLTE (than Wi-Fi) because as a ldquocellularrdquo system it tends tohave higher requirements for example higher mobility withlower latency Therefore we set LTE as our model system forthe CBRS in the 35 GHz band Contributions of this paperare summarized as follows

(1) This paper proposes Precoded SUbcarrier Nulling(PSUN) an OFDM transmission strategy that effec-tively suppresses pulsed interference from a radarBy applying PSUN at a transmitter (Tx) and pulseblanking (PB) at a receiver (Rx) an LTE systemcan mitigate intercarrier interference (ICI) caused bypulsed interference from coexisting radars It is note-worthy that this paper suggests a coexistence methodwithout modifying the incumbent radarsrsquo operations

(2) This paper provides an analysis framework forOFDM-pulsed radar coexistence To the best of ourknowledge this paper is the first work that considersexistence of ESC in the coexistence problem whichreflects uniqueness of the problem that it is managedby both means of database and spectrum sensingFurthermore the framework takes into account theimpacts of imperfect prediction of radar interference

(3) This paper suggests use cases of the fifth-generation(5G)mobile networks that LTE downlink can supportby using the 35 GHz band based on the analyses andresults that this paper provides

2 Related Work

In [7] a novel radar waveform that minimizes a radarrsquos in-band interference on a coexisting communications systemis proposed This approach assumes that a radar has fullknowledge of the interference channel and modifies its ownsignal vectors in such a way that they fall into the null spaceof the channel matrix between the radar and the coexistingcommunications system In [8] the coexistence scenarioof [7] is extended to more than one interference channelOur work is distinguished from [7 8] because it proposesa strategy that requires no change of the incumbent radarsystem It is ameaningful contribution considering the widelyacknowledged concern about national security and cost ofchanging the incumbent system

In [9 10] opportunistic spectrum sharing between anincumbent radar and a secondary cellular system is studiedThe work specifies applications that are feasible in such acoexistence scenario It is found that noninteractive video ondemand peer-to-peer file sharing file transfers automaticmeter reading and web browsing are feasible while real-time transfers of small files and VoIP are not In [11] it issuggested that the secondary communication system utilizesinformation of the incumbent radar that is provided by adatabase In [12] impacts of interference from shipborneradars to LTE systems are studied An eNodeBrsquos signal-to-interference-plus-noise ratio (SINR) plummets when hit by

radar pulses but an LTE system is able to recover duringthe time between radar pulses Average throughput of userequipment (UE) drops under radar interferenceThe authorsconcluded that theUE throughput loss in the uplink directionis tolerable even with a radar deployed only 50 kilometersaway from the LTE system In [13] the study in [12] isextended The authors studied impacts of shipborne radarsthat operate in the same channel and are located in thevicinity of a 35 GHz macrocell and outdoor small-cell LTEsystems With such additional consideration of out-of-bandeffects of shipborne radars the authors still conclude thatboth macrocell and outdoor small-cell LTE systems canoperate inside current exclusion zones In [14] on the otherhand it is concluded that LTE systems are unable to cope wellwith narrowband bursty interference on the downlink Ourwork is distinguished from [9ndash14] because this paper studieshow to actually cancel radar interference while only feasibilityof coexistence was discussed in the prior studies

In addition this paper provides a generalized analyticalframeworkThis paper takes into consideration a comprehen-sive interplay amongmultiple variables regarding themilitaryradarsrsquo operations such as the number of radars pulseparameters antenna sidelobes and out-of-band emissionswhich will be discussed in Section 3 Moreover impacts ofimperfect prediction of radar interference are measured byappropriate probabilities whichwill be explained in Section 5

Note that this paper is an extension of our previousstudy that was published in [15] The extension is twofold(i) we change the performance metric from bit error rateto maximum data rate to more fairly reflect the impact ofPSUN on an OFDM system performance (ii) we use 35 GHzLTE as a near-term example that serves to illustrate how thetechnique could be applied to operation of future 5G systemsin bands shared with pulsed radars

3 Coexistence Model

This paper discusses the performance of an LTE small-cellsystem that coexists with multiple military radars that rotateand generate pulsed signals Note that this paper focuses onthe downlink of an LTE system where an eNodeB acts as a Txand a UE becomes an Rx

Also this paper assumes that there is no impact of fadingfrom mobility nor multipath since the ICI that is causedby radar interference has far more significant impacts thanDoppler shift and delay spread Therefore we assume thatthe only two channel impairments are radar interference andadditive white Gaussian nose (AWGN) In other words anOFDM symbol goes through an AWGN channel when theLTE system is not interfered by the radar There is a periodof time when the radar beam does not point at the LTEsystem since a radar rotates during this time an LTE systemis assumed to experience an AWGN channel It should benoted that hence the simulation results that are presented inSection 6 do not take fading into consideration

31 Characterization of a Military Radar It is very importantto note that a 35 GHz band coexistence problem is morechallenging than what is often acknowledged This paper


Table 1 Parameters for antenna horizontal sidelobe analysis

Parameter Remark

120579beam

Angle of a radar antennarsquos horizontal beam withmain lobe and sidelobes that cause interference onan LTE system

120579passAngle that a radar antennarsquos horizontal beam passesthrough an LTE cell

120579intfThe total angle that a radar antennarsquos horizontalbeam interferes with an LTE cell

119889 Distance between a radar and an LTE cell119903119888 Diameter of an LTE cell119879rot Radar rotation time

d

rc

Beam rotation

120579intf120579beam

120579pass120579beam 120579beam

Figure 1 Impact of antenna horizontal sidelobes

considers two aspects that increase the impact of a pulsedradarrsquos interference on an LTE cell a radarrsquos antenna sidelobesand out-of-band emissions These analogous spatial andfrequency domain effects are serious due to the extremedifference in transmitting power between radar and LTE

311 Antenna Sidelobes Following the FCCrsquos guideline indesigning a CBRS system coexisting with military radars [3ndash5] a sufficiently large spatial separation must be guaranteedbetween a federal military radar and an LTE system toguarantee a low level of interference from an LTE eNodeB(Tx) to the radar In spite of this large distance from a radaran LTE UE (Rx) cannot avoid radar interference with a veryhigh level due to the much higher transmitting power of aradar The power of a radarrsquos signal received at an LTE Rx isso high that even sidelobes cause significant interference tothe communications system This is interpreted as a greatervalue of horizontal angle of a radarrsquos beam that actually causesinterference on a coexisting LTE system Figure 1 illustratessuch an impact of a radar antennarsquos horizontal sidelobes Itdescribes that the angle of a radar beam 120579beam contains notonly its main lobe but also the sidelobes The value of 120579beamdiffers according to type of radar For instance the antennapattern of a radar analyzed in [1] has cosine pattern withsidelobes that are 144 dB lower than the main lobe

Now we formulate such a coexistence model in whichan LTE system is interfered by a radar that rotates andtransmits pulses Table 1 describes parameters used in theanalysis including those shown in Figure 1 Suppose that a

radar rotates counterclockwise and an LTE system is withininterference range of the radarrsquos signal The angle of rotationduring which the radarrsquos beam passes through a cell of an LTEsystem is given by

120579pass =360∘

sdot 119903119888

2120587119889 (1)

As illustrated in Figure 1 the total angle through which theradar beam interferes with a cell of an LTE system can bewritten as

120579intf = 120579beam + 120579pass (2)

Note that 120579beam differs according to type of radar while 120579passis determined by 119889 and 119903

119888 Then the total interference time

is defined as the time period when a cell of an LTE systemis interfered by a radar within a beam rotation which isobtained by

119879intf =120579intf360

sdot 119879rot (3)

Such an impact of a radarrsquos antenna horizontal sidelobesis evidenced in Figure 5 of [16] The report describes anobserved case in which a wireless communication systemreceives energy from an SPN-43 shipborne radar at a levelthat is approximately 30 dB higher than the noise floor evenwhen the main lobe of the radar antenna is towards thedirection opposite to a cell of the wireless communicationssystem This implies that sidelobes of a radar beam can havea significant impact on operation of a coexisting wirelesscommunications system

312 Out-of-Band Emission Due to extremely high peaktransmitting power of a radar out-of-band emission from aradar operating in a neighboring channel also has a signifi-cant impact on a coexisting LTE system Radars themselvesare separated among different channels to avoid interferingwith each other This spectral separation is enough to protectradars from interference due to other radars but is insufficientto protect a wireless communications system that operateswith a much lower transmitting power

Figure 2 illustrates a simulation result of a radarrsquos out-of-band interference on an LTE system We simulated an LTEsystem operating at 35 GHz and a radar generating pulsesat 35 355 and 36GHz The transmitting powers of a radarand an LTE eNodeB are assumed to be 83 dBm and 23 dBmrespectively The distance between an LTE eNodeB and a UEis 100 meters while the radar is assumed to be separated bydistance of 100 kilometers Also the radarrsquos pulse repetitiontime (PRT) and duty cycle are 1msec and 10 respectivelyA radar has an extremely large bandwidth due to its pulsednature Since transmitting power of a radar is too muchhigher than that of wireless communications Tx it is stillhigher than an LTE eNodeBrsquos signal at a UE even with a50MHzor 100MHzoffsetThis implies thatwemust take intoaccount interference caused by radarsrsquo out-of-band emissionswhen we analyze coexistence between a pulsed radar anda wireless communications system As mentioned earlier a


348 3485 349 3495 35 3505 351 3515 352

0

10

20

30

40A

mpl

itude

(dB)

Radar (in-band)LTE

f (Hz)

minus10

minus20

minus30

times109

Radar (10MHz offset)Radar (5MHz offset)

Figure 2 Impact of out-of-band emissions

radarrsquos out-of-band transmission does not cause significantinterference to another radar in an adjacent band becausetransmitting powers of the radars are similar However to anLTE system an out-of-band radar emission causes significantinterference due to a significant difference in transmittingpower between an LTE eNodeB and a radar

Regarding the simulation setting discussed above it isnoteworthy to elaborate the rationale behind selection of thevalue of path loss exponent that equals 2 In the geography ofthe coexistence model the lengths are significantly differentbetween the two main parts (i) between a radar and an LTEsystem and (ii) between an eNodeB and a UE in an LTEsystem The idea is that the former part is much longer indistance and thusmore affected by the path loss In the formerpart of a coexistence geography the path loss becomes thedominant channel impairment due to the long distance (egtens of kilometers) On the other hand in the latter partradar interference becomes the main channel impairmentsince the path loss does not influence the performance due toshort-distance propagation As mentioned earlier in a LTE-radar coexistence scenario the former part is much longerin length than the latter part Therefore when selecting avalue of the path loss exponent it is the former part that weshould consider more significantly than the latter part Sincethe former part is very likely composed of a long line-of-sightpath it is approximated as 2 to give a conservative estimateeg one that is less favorable to the LTE link

Such interference from out-of-band radars can be inter-preted as a greater number of radars that cause interferencesince radars operating in neighboring channels also causeinterference to an OFDM system Hence there are additionalbursts of interference from the out-of-band radars within anin-band radarrsquos rotation period It is likely that the radars

Table 2 Computation of the total interference time 1198791015840intf

120579beam (deg) 120579intf (deg) 119879intf (msec) 1198791015840

intf (msec)5 107 596 178810 157 874 262230 357 1985 5955

have different values of 119879rot duty cycle and PRT whichmakes the task of an LTE system to track interfering pulsesmore difficult In this paper we reflect the impact of out-of-band interference due to radars on lower and upper adjacentfrequencies in such away that there occurs a threefold increasein the number of OFDM symbols that are hit by a radarpulseTherefore the total length of time that a radar interfereswith an LTE cell within a radar rotation 119879

1015840

intf can be given by1198791015840

intf le 3119879intf Note that 1198791015840

intf = 3119879intf is true when there is nooverlap in time among pulses generated by the three radars

Table 2 demonstrates1198791015840intf according to different values of120579beam assuming that 1198791015840intf = 3119879intf We set 120579beam to 5 10 and30 degrees Let us apply 119879

1015840

intf = 5955msec to the currentLTE standard as an example Within a radar rotation time119879rot = 2 sec 2000 LTE subframes can be transmitted Since 14OFDM symbols are transmitted in a subframe 28000 OFDMsymbols can be transmitted As a result (59552000) times

28000 asymp 8337 out of 28000 OFDM symbols are hit withina rotation of a radar

32 Generalized Expression of Radar Interference In the35 GHz Band radars report their operating parameters (iepulse parameters and position) to a SAS and an ESC alsosenses and sends the parameters to a SAS Based on such acoexistence model the frequency of pulse interference withina certain time can be quantified for use in analysis There arefour factors affecting the frequency (i) the number of radars(ii) PRT of a radar (iii) level of interference from antennasidelobes of a radar and (iv) level of interference caused byout-of-band radars However it is extremely difficult for anESC to keep track of all the four factors since military radarskeep changing their parameters and the radars parametersare even classified in many cases as explained in an armysregulation document [22] To this end this paper generalizesthe frequency of pulse occurrence by defining a quantitycalled the probability of pulsed interference 120588 It is defined tobe the probability that anOFDM system experiences a pulsedinterference within a certain period of time In this way thequantity 120588 generalizes the impacts of all of the four factorsdescribed above

Note that this paper adopts the LTE standardrsquos parametersfor simulating a CBRS system as will be demonstrated inSection 6 and the scope of defining 120588 is 1msec the lengthof a subframe defined in the LTE standard If 120588 = 0 during asimulation of 1000 subframes none of the subframes are hitby a radar pulse If 120588 = 1 on the other hand every subframeexperiences radar interference during the simulation Notethat this analytical framework can be extended to any othertype of OFDM communication without loss of generality Inother words the definition of 120588 can be set within any specified


Table 3 Existing ICI self-cancellation (ISC) schemes and the proposed subcarrier nulling (119871 = 2)

ICI self-cancellation (ISC) scheme Subcarrier allocationData conversion [17] 119883

1015840

(119896) = 119883(119896)1198831015840(119896 + 1) = minus119883(119896) where 119896 is the subcarrier indexSymmetric data conversion 119883

1015840

(119896) = 119883(119896)1198831015840(119873 minus 119896 minus 1) = minus119883(119896) where119873 is the FFT sizeWeighted data conversion [18] 119883

1015840

(119896) = 119883(119896)1198831015840(119896 + 1) = minus120583119883(119896) where 120583 is a real number in [0 1]

Plural weighted data conversion [19] 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883(119896)

Data conjugate 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = minus119883lowast

(119896)

Data rotated and conjugate [20] 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883lowast

(119896)

PSUN 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = 0

time period that can be measured by the number of OFDMsymbols

4 Precoded SUbcarrier Nulling (PSUN)

41 Proposition of PSUN Pulse blanking (PB) is knownto be one of the most effective techniques for suppressingpulsed interference [23ndash25] Unfortunately PB still leavesa significant level of ICI In PB time domain samples ofthe received signal affected by pulsed interference are set tozero The technique deteriorates performance of an OFDMsystem by affecting not only the interfered samples but alsothe desired samples This problem occurs due to the factthat (inverse) Fourier transform provides a time-frequencymapping in such a way that every frequencytime samplecontributes to generating a timefrequency symbol In anOFDMsystem PB takes place in the timedomainwhereas thedata symbols are mapped to the subcarriers in the frequencydomain An OFDM Rx blanks only several samples that areradar-interfered in the time domain However such a partialchange leads to corruption of all the samples in the frequencydomain due to characteristic of the Fourier transform whichstill causes ICIThis paper focuses on suppression of such ICIthat remains after applying PB at an OFDM Rx

This paper suggests that the negative impact of PB can beconsidered a form of time-selective fading Channel codingis usually applied in combination with interleaving anddiversity to mitigate performance degradation due to fading[26] In OFDM systems the main means of combating time-selective fading are block interleaving and antenna diversityHowever our results indicate that neither method can effec-tively mitigate ICI caused by PB Interleaving is ineffectivebecause PB does not result in bursty errors due to the one-to-all mapping characteristic of the Fourier transform Antennadiversity is also not effective against the ICI caused by PBbecause an entire LTE cell is likely to be hit at once by a radarrsquosbeam A multiple-antenna technology can bring no benefitwhen the signals received by all the antennas are interferedwith simultaneously

ICI self-cancellation (ISC) is an aggressive means ofcombating ICI It cancels ICI by allocating precoded 119871 minus

1 redundant subcarriers between data subcarriers whichresults in a 1119871 data rate Based on the work of Zhao andHaggman [17] several ISC schemes have been proposed [18ndash20] Some of the existing ISC schemes are summarized inTable 3 assuming 119871 = 2 Note that 119883(sdot) and 119883

1015840

(sdot) indicate

the original transmitted data symbol and the symbol after ISCprecoding respectively

We discovered that the most effective way of reducingICI induced by PB is to insert null subcarriers instead ofallocating any other types of redundant subcarriers Therationale is illustrated in Figure 3 It is an example that issimplified to clearly demonstrate the impact of location of PBon the level of ICI Figure 3(a) represents an example signalat Tx while Figures 3(b) and 3(c) show two different locationsof PB at Rx The example signal contains three among 64subcarriers around the center (28th 30th and 32nd) thatare set to 1 while all the others are set to 0 Note that thetransmitted signal in Figure 3(a) shows the real part of theoriginal complex signal It is observed from Figure 3 that thelocation of PB has a very significant impact on the level ofICI caused by PB Comparing Figures 3(b) and 3(c) the ICIbecomes more severe as higher-amplitude samples are blankedIn other words the ICI level can be reduced as the timedomain fluctuation gets flatter It is straightforward that thesimplest way of keeping time domain amplitudes low is toreduce the number of subcarriers AnOFDMRx can suppressICI remaining after PB better when a Tx has allocated nullsubcarriers instead of other types of redundancy since use ofnull subcarriers reduces the number of high-energy bins inthe time domain

For this reason an OFDM Tx employing PSUN precodesan OFDM symbol by inserting null tones between data tones sothat the ICI after PB at its Rx can be suppressed This makesPSUN a type of ISC as listed in Table 3 Various mannersof inserting null tones for different purposes have beenstudied in the literature [27ndash29] In this work PSUN allocatesthe null tones in such a way that the radar interference isminimized Figure 4 shows that PSUN outperforms the otherISC schemes Note that for the weighted data conversionscheme the value of 120583 becomes 12 The reason for PSUNrsquoshigher performance is that PSUN yields smaller variation ofan OFDM symbol in the time domain because it transmits asmaller number of subcarriers

42 The Transmission Protocol of PSUN Let 119903 denote thecoding rate of PSUN With the coding rate of 119903 = 1119871 PSUNinserts 119871minus1 null tones between data tones Figure 5 illustrateshow PSUN inserts null tones in an exemplar OFDM symbolwith QPSK and the FFT size of 32 Figure 5(a) demonstratesan OFDM symbol without PSUN Figures 5(b) and 5(c) show


0 10 20 30 40 50 60

0

005

Time

TransmittedA

mpl

itude

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(a) Transmitted

0 10 20 30 40 50 60

0

005

Time

ReceivedPulse blanking

minus005

Am

plitu

de

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(b) Received (PB on low-amplitude samples)

100 20 30 40 50 60

0

005

Time

Received

Am

plitu

de

Pulse blanking

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(c) Received (PB on high-amplitude samples)

Figure 3 Dependency of ICI on the location of PB

examples of precoding the OFDM symbol using PSUN with119903 equal to 12 and 14 respectively PSUN extracts the firsthalffourth of the data tones from the original OFDM symbolgiven in Figure 5(a) Note that this method of taking 1119871 ofits original data is only an example PSUN can do it in variousother ways another example is to extract a data tone in every119871 subcarrier Then PSUN inserts null tones (marked with redsquares) between the data tones which leads to the mappingillustrated in Figures 5(b) and 5(c)

This is where PSUN sacrifices data rate by 1119903 within anOFDM symbol Tominimize such loss of data rate anOFDMTxperforms two important operationswhen adopting PSUNFirst it localizes OFDM symbols to be hit a priori and allocatesnull tones in the symbols only The a priori knowledge aboutradar pulse parameters is provided by a SAS but sensed by

an ESC beforehand Figure 6 shows a subframe in which anOFDM symbol is expected to be hit by a radar pulse Onlythat symbol is precoded with the null subcarriers at Tx beforetransmission Second within the OFDM symbol to be radar-interfered an OFDMTx disables channel coding and shifts thesaved redundancy to PSUN This assumes that for an OFDMsymbol to be radar-interfered the pulsed interference ismoresevere than AWGN This protects the symbol from radarinterference while keeping the total number of transmittedbits the same Multiple OFDM symbols can be hit simulta-neously because an interference pulse can be either shorteror longer than an OFDM symbol In this case the OFDMsymbols are all precoded All the other symbols that are notprecoded are transmitted with channel coding and full datatones


0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3

10minus4

PSUNData conversionSymmetric data conversionWeighted data conversionPlural weighted data conversionData conjugateData rotated and conjugate

(a) Pulse duty cycle of 1

0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3


(b) Pulse duty cycle of 10

Figure 4 Comparison of PSUN to other ISC schemes (QPSK 1024-FFT)

0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(a) Without PSUN

0 5 10 15 20 25 30minus1

minus05

0

05

1

Subcarrier

Am

plitu

de

(b) With PSUN (119903 = 12)

0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(c) With PSUN (119903 = 14)

Figure 5 An OFDM symbol applying PSUN (QPSK 32-FFT)

Figure 6 illustrates PSUN from such a macroscopicstandpoint An OFDM Tx employing PSUN reduces lossof data rate by selecting certain OFDM symbols to insertnull subcarriers According to the FCCrsquos suggestion a prioriknowledge of interference from incumbent radars is available

at an LTE eNodeB Radars report their operating parameters(ie pulse parameters and position) to a SAS and an ESC alsosenses the parameters and sends them to a SAS

Taking LTE as an example of a CBRS system there are14 OFDM symbols in a subframe Figure 5 showed only


OFDM symbol not to be radar-interferedOFDM symbol to be radar-interfered

TimePulsed interference

Subcarriers Subcarriers

Am

plitu

de

Am

plitu

de

Null carriers

middot middot middot middot middot middot


Figure 6 Transmission protocol of PSUN (119903 = 12)

one OFDM symbol that is expected to be hit by a radarpulse In Figure 6 an OFDM symbol to be radar-interferedis highlighted by orange color However there are 13 otherOFDM symbols that are not radar-interfered An OFDM Txapplying PSUN does not precode these OFDM symbols fortwo reasons (i) they undergo AWGN channels against whichchannel coding achieves better protection than PSUN (ii)thus as explained earlier unnecessary loss of data rate canbe avoided by not applying redundancy in subcarriers

It is possible that two or more consecutive OFDMsymbols can be interfered by the same pulse because aninterference pulse can be either shorter or longer than anOFDM symbol depending on the pulsersquos duty cycle In such acase all of the OFDM symbols that are expected to be radar-interfered are precoded

5 Imperfect Pulse Prediction

We discovered that three types of imperfect pulse predictionare possible in a 35 GHz band coexistence framework (i)false prediction (ii) missed prediction and (iii) mislocationFalse alarm and missed detection are defined as an ESCrsquosinaccurate claim of presenceabsence of an interfering radarpulse given that a pulse is in fact absentpresentMislocationis a unique type of imperfect pulse prediction that we suggestin this paper It occurs when an ESC accurately predictsthe location of a pulse interference in terms of subframebut being inaccurate in terms of symbol within a subframeMore specifically it is called a mislocation when an ESCpredicts that an OFDM symbol within a subframe will behit by a radar pulse and in fact the interference actuallyoccurs at the predicted subframe but at a different OFDMsymbol

Let us interpret actual impacts of the three types of imper-fect pulse prediction Recall that channel coding and PSUNare countermeasures against AWGN and pulsed interferencerespectively A false alarm is interpreted as a situation wherean OFDM symbol that is not to be radar-interfered is pre-dicted to be radar-interfered and thus precoded with PSUNTherefore in the OFDM symbol redundant bits for channelcoding are removed and null subcarriers are allocated insteadwhich is a weaker protection than channel coding against

AWGN but in fact the symbol is not hit by a radar pulse butgoes through an AWGN channel On the other hand whena missed detection occurs an OFDM symbol to be radar-interfered is not predicted to be radar-interfered and thus notprecoded with PSUN Thus the OFDM symbol is protectedwith channel coding instead which is a weaker protectionthan PSUN against pulsed interference Overall although inthe opposite way either a false alarm or missed detectiondeteriorates performance of an OFDM system that appliesPSUN Most interestingly a mislocation has the impact of afalse alarm and missed detection within a single subframeRecall that a false alarm unnecessarily precodes an OFDMsymbol that will undergo AWGN with PSUN while misseddetection does not precode a symbol that will be hit by aradar pulse Let us assume that an ESC has predicted anOFDM symbol named ldquoArdquo to be hit by a radar pulse andhence has precoded it A mislocation occurs when in factanother OFDM symbol called ldquoBrdquo has actually been hit Theproblem is that OFDM symbol ldquoBrdquo has not been precodedwith null subcarriers since the ESC has predicted it not to behit by a radar pulse but to go through an AWGN channelTherefore a mislocation results in two OFDM symbols thatare incorrectly precoded within a single subframe OFDMsymbol ldquoArdquo has been protected against a radar pulse but hasactually undergone anAWGNwhile ldquoBrdquo has been believed toexperience an AWGN and thus has not been precoded but infact has gone through a radar interference To interpret thissituation a false alarm has occurred at OFDM symbol ldquoArdquowhereas missed detection has happened at ldquoBrdquo This is how amislocation causes a false alarm and missed detection at thesame time within one subframe

Major causes of the above imperfect pulse prediction aretwofold Firstly an ESC can cause sensing errors Secondly anESC can lose track of radarsrsquo pulse parameters The formeraffects false alarm and missed detection while the latterimpacts all of the three types of imperfect pulse prediction

51 Sensing Error by an ESC Typically for a protocol requir-ing spectrum sensing either a matched filter or an energydetector can be used [30 31] This paper assumes that anESC a device with sensing capability uses an energy detectorAssuming that an interference signal from a radar and noiseare both modeled as white Gaussian processes the problemof sensing a radarrsquos pulsed interference signal by an ESC canbe given by the following hypotheses test

1198670 119884 sim N (0 120590

2

0)

1198671 119884 sim N (0 120590

2

0+ 1205902

1)

(4)

where

119884 is an observation sample

1205902

0is power of noise

1205902

1is power of an interference signal


0 02 04 06 08 10

02

04

06

08

1

Miss

ed d

etec

tion

prob

abili

tyP

m

False alarm probability Pfa

ReferenceEbNo = 10dBEbNo = 5dB

EbNo = 4dBEbNo = 0dB

Figure 7 ROCs of the energy detector at an ESC

Since an ESC adopts an energy detector based on theNeyman-Pearson detection theory the probability of falsealarm 119875fa and missed detection 119875

119898 are defined by

119875fa ≜ Pr (1198671| 1198670) = 1 minus Γ(

1

2120578se212059020

)

119875119898≜ Pr (119867

0| 1198671) = 1 minus Γ(

1

2

120578se2 (12059020+ 12059021))

(5)

where 120578se denotes the sensing error threshold and the incom-plete gamma function is given by

Γ (119905 119911) =1

Γ (119905)int

119909

0

119905119905minus1

119890minus119909

119889119909 (6)

A receiver operating characteristic (ROC) curve is usedfor an analysis of interplay between 119875fa and 119875

119898 Figure 7

shows ROCs of (5) according to the energy per bit to noisepower spectral density ratio (EbNo) An increase in thesensing threshold for given signal and noise power valuesmoves the operating point toward the upper direction alongone of the curves in the figure At a high EbNo regime both119875

119898

and119875fa canmaintain low values even if the sensing thresholdchanges much This is not the case for low EbNo

52 Loss of Track of Radarsrsquo Operating Information It isdifficult to track a radarrsquos pulsed signals for the followingtwo reasons Firstly the pulse information might not be fullyavailable to the SAS There has been strong opposition frommilitary stakeholders to provide information to the databaseabout radarsrsquo position or other information that could makethemmore prone to be affected by enemy jammers Secondlya radar may change its pulse parameters and position forvarious purposes such as higher security or avoidance of

interference among radars According to a recent extensivesurvey paper [32] most radar systems have fixed positionand operating parameters However airborne and shipborneradars may not have preplanned routes and therefore anerror region has to be defined for such cases In this casethere occurs a time during which an ESC loses track of aradarrsquos pulse parameters An ESC requires some time to sensea radarrsquos parameter changes during which it cannot avoidproviding outdated information to a SAS

We suggest that an ESCrsquos losing track of radarsrsquo operatinginformation must be understood more seriously than anESCrsquos sensing errors The reason is that it is more likely andcan cause any of the three types of imperfect pulse predictionbut is more difficult to study since it is not a characteristic ofan ESC but that of a radar which is an independent variablein this paper Therefore this paper provides a frameworkfor analyzing this loss of track Values of the false alarmmissed detection and mislocation probabilities 119875fa 119875119898 and119875ml over the interval of [01] are considered so that theanalysis can be generalized over any case in which an ESCloses track of radarsrsquo operating parameters

6 Performance Evaluation

61 Simulation Setup The discussion in [9 10] can beinterpreted that the CBRS system coexisting with the pulseradar utilizes spectrummore efficiently in the downlink thanin the uplink in terms of the data rate per megahertz Hencespectrum sharing with radar would be more appropriate forapplications that require greater capacity in the downlinkthan the uplink which is a typical characteristic of manyapplications Therefore this paper assesses the performanceof the downlink of an LTE system by measuring the numberof bits per second that an LTE UE successfully receivesThe number of transmitted bits differs according to themodulation scheme (In this paperrsquos simulations 16-QAMand 64-QAM were evaluated) We analyze the metric asfunctions of six variables that are chosen to represent threedifferent aspects of coexistence between an LTE Rx andmilitary radars as follows (i) EbNo represents impact ofAWGN (ii) pulse duty cycle and 120588 represent characteristicsof interference by a radar (iii) 119875fa 119875119898 and 119875ml representimpacts of imperfect pulse prediction Each variable gaugesdifferent levels of channel impairment that is AWGN orradar interference It differentiates the bit error rates whichagain directly determines the number of received bits

Table 4 summarizes the simulation parameters for LTEand radar We leverage LTE physical-layer simulations whichare 3GPP compliant [33] The FFT size is set to 1024 but theresults based on this parameter can hold for other valuesof FFT size The reason is that PB is a channel impairmentthat occurs in time domain and LTE is always synchronizedin time regardless of FFT size Coding rates of channelcoding and PSUN are kept identical to be 119903 = 12 for easeof demonstrating the impacts of shifting redundancy fromchannel coding to subcarrier nulling The only two channelimpairments that are considered in this paper are AWGNand radar interference as a result no typical fading effects areconsidered Hence the simulations do not accurately follow



Parameter ValueLTE

FFT size 1024Subcarrier spacing 15 kHzSampling frequency 1536MHzOFDM symbol time 667 120583sSubframe length 1msCP length 52 120583s (1st)469120583s (the following 6)OFDM symbolssubframe 14Modulation 16-QAM 64-QAMChannel coding (133171) convolutional code (119903 = 12)PSUN 119903 = 12

RadarPulse repetition time 1msRotation rate 30 rpm

themodulation and coding scheme (MCS) that are associatedwith channel quality indicator (CQI) In order for LTE tooperate in the 35 GHz band a new set of MCS and CQI mustbe matched Radar pulse repetition time is set identical to anLTE subframe duration (1msec) for accuracy of computationEach simulation is conducted through 10

6 subframesTo elaborate the discussion about a new set of MCS

and CQI we claim that it will be necessary because the35 GHz environment is a totally different one from theprevious spectrum bands in which LTE systems have beenoperating In addition to all the mobility and multipathimpacts design of an LTE system at the 35 GHz band needsto consider pulsed interference generated by radarsHoweverthis exceeds the scope of this paper and will be discussed inour future work In other words the results that are discussedin this paper do not have any impact from the new set ofMCSand CQI

62 Results

621 EbNo Figure 8(a) shows the number of received bitsper second versus EbNo with 16-QAM and 64-QAM Recallthat an OFDM Tx employing PSUN disables channel codingbut puts the redundancy saved fromno channel coding to nullsubcarriers between data subcarriers instead In low EbNoregion AWGN is the predominating channel impairmentthat outweighs radar interference which results in lowereffectiveness of PSUN In other words outperformance ofPSUN over the case without PSUN gets increased as EbNogets higher In thatway radar interference becomes prevailingwhich leads to greater performance advantage of PSUNMoreover such advantage of PSUN gets greater with highermodulation order

622 Pulse Parameters of the Radar Figure 8(b) demon-strates the number of received bits per second versus the dutycycle of a radar pulse We generalized the values of pulse duty

cycle for wider generality of this work although many of thepulsed radars deployed in practice use relatively small valuesof duty cycle for example 01ndash10 It is straightforward thathigher pulse duty cycle yields greater outperformance ofPSUNover the casewithout PSUNAlso similar to the resultswith EbNo above performance advantage gets greater as themodulation order becomes higher

Figure 8(c) illustrates the number of received bits persecond versus the probability that an OFDM symbol is hitby a radar pulse 120588 When 120588 = 0 the performance must bethe same between the cases with and without PSUN sincePSUN does not allocate null subcarriers when no OFDMsymbol is radar-interfered As explained in Section 32 agreater value of 120588 yields a smaller number of received bitsper second Similar to the discussion of pulse duty cyclein Figure 8(b) a greater value of 120588 indicates a more severesituation of radar interference Due to this it still holds truethat outperformance of PSUN increases as 120588 becomes greaterThe performance curve drops faster in 64-QAM than 16-QAM which implies that higher-order modulation is moresensitive to radar interference Nevertheless performanceadvantage of PSUN gets greater as the modulation order getshigher

623 Pulse Prediction Errors So far we have seen the perfor-mances assuming perfect pulse prediction The results shownthrough Figures 8(d) and 8(f) depict how the performanceof an OFDM system is deteriorated with imperfect pulseprediction Figure 8(d) shows the number of received bitsper second versus the probability of false alarm 119875fa It isstraightforward that higher 119875fa decreases the number ofreceived bits per second of an OFDM system employingPSUN while the case without PSUN stays unrelated to thelevel of 119875fa The reason is that with a false alarm an OFDMsymbol is protected by PSUN instead of channel coding butin fact it undergoes an AWGN channel where channel codingis more effective protection than PSUN

Figure 8(e) shows the number of received bits per secondversus the probability of missed detection 119875

119898 As explained

earlier in Section 5 at an OFDM Tx applying PSUN misseddetection is translated as a situation where an OFDM sym-bol is not predicted to be radar-interfered and hence notprecoded with PSUN but in fact hit by a radar pulse Inother words the particular symbol is equipped with channelcoding instead of PSUNandhence contributes to degradationof performance The performance degradation of OFDMRx without PSUN is shown by the gap at zero 119875

119898 As

119875119898increases the performance of PSUN gets closer to the

case without PSUN The performance advantage of PSUNincreases as the modulation order gets higher

Figure 8(f) shows the number of received bits per secondversus the probability of pulsemislocation119875ml Amislocationrefers to a wrong location of to-be-interfered OFDM symbolwithin a subframe Recall that with a mislocation a falsealarm and missed detection occur at the same time withina subframeThis is why performance propensity according to119875ml from Figure 8(f) is nearly linear while the ones accordingto 119875fa and 119875

119898are logarithmic and exponential respectively

as observed from Figures 8(d) and 8(e)


0 2 4 6 8 10 124050607080904050607080

EbNo (dB)

Dat

a rat

e (M

bps)

w PSUN (16-QAM)wo PSUN (16-QAM)w PSUN (64-QAM)wo PSUN (64-QAM)

(a) Versus EbNo (120588 = 08 duty cycle = 01)

0 005 01 015 02 025 035055606570755055606570

Dat

a rat

e (M

bps)

Duty cycle of a radar pulse


(b) Versus duty cycle (EbNo=4 dB120588 = 08)

0 02 04 06 08 150

55

60

65

70

Dat

a rat

e (M

bps)

120588


(c) Versus 120588 (EbNo = 4 dB duty cycle = 01)

0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pfa


(d) Versus 119875fa (duty cycle = 01 120588 = 08EbNo = 4 dB)

0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pm


(e) Versus 119875119898

(duty cycle = 01 120588 = 08EbNo = 4 dB)

0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pml


(f) Versus 119875ml (duty cycle = 01 120588 = 08EbNo = 4 dB)

Figure 8 Data rate versus EbNo the duty cycle of a radar pulse 120588 119875fa 119875119898 and 119875ml

7 Feasibility of 5G Applications Using 35 GHzLTE with PSUN

Fifth-generation (5G) mobile networks will operate in ahighly heterogeneous environment characterized by the exis-tence of multiple types of access technologies over multiplechunks of spectrum bands In other words enabling 5Guse cases and business models requires the allocation ofadditional spectrum for mobile broadband and needs tobe supported by flexible spectrum management capabilitiesBased on the analyses and results of this paper we suggestthat the 35 GHz band can be a usable additional spectrumfor enabling LTE to support several functionalities of 5Gtechnologies

We refer to a white paper [21] issued by the NextGeneration Mobile Networks (NGMN) a mobile telecom-munications association of mobile operators vendors man-ufacturers and research institutes for understanding therepresentative example use cases of 5G and the correspondingrequirement of data rate for each use case A consistent userexperience with respect to throughput needs a minimumdata rate guaranteed everywhere The data rate requirementof a use case is set as the minimum user experienced datarate required for the user to have a quality experience of thetargeted use case The use cases are summarized in Table 5

According to our results LTE with PSUN can fulfill thedownlink requirements of several use cases which are listedunder the category of ldquocandidates for LTE with PSUNrdquo in


Table 5 Data rate requirements for use cases of 5G [21]

Use case Data rate requirement(downlinkuplink)

Candidates for LTE with PSUNMassive low-costlong-rangelow-powerM2M

1ndash100 kbps

Resilience and traffic surge 01ndash1Mbps01ndash1MbpsUltrahigh reliability ampultralow latency

50 kbps to 10Mbpsa few kbpsto 10Mbps

Ultrahigh availability ampreliability 10Mbps10Mbps

Airplanes connectivity 15Mbps75MbpsBroadband access in a crowd 25Mbps50Mbps50+Mbps everywhere 50Mbps25MbpsUltralow latency 50Mbps25Mbps

Others

Broadband like services Up to 200Mbpsmodest (eg500 kbps)

Ultralow-cost broadbandaccess 300Mbps50Mbps

Mobile broadband in vehicles 300Mbps50MbpsBroadband access in denseareas 300Mbps50Mbps

Indoor ultrahigh broadbandaccess 1 Gbps500Mbps

Table 5 While most of the requirements of the selected usecases are set to be 50Mbps our results (Figures 8(a) through8(f)) indicate that LTE with PSUN is capable of supportingdata rates that are higher than 50Mbps and 40Mbps with64-QAM and 16-QAM respectively For example observingFigure 8(a) the required EbNo values for achieving the datarate of 50Mbps are 0 and 1 dB for 64-QAM and 16-QAMrespectively

It is discussed in [9 10] that although average data rateis roughly the same for all file sizes because of interruptionsas a radar rotates average received data rate for smallerfiles may vary depending on when the transmission beginsrelative to the radarrsquos rotation cycleThis effect does not occurduring transmission of larger files that span one or morerotation periods of the radar The authors suggested severalappropriate applications that can tolerate interruptions froma pulsed radar video on demand peer-to-peer file sharingand automatic meter reading or applications that transferlarge enough files so the fluctuations are not noticeable suchas song transfers Among these applications a white paperthat analyzed the mobile traffic pattern of 2015 [34] finds adirection that LTEwith PSUN can target in the 35 GHz bandIt says that mobile video traffic accounted for 55 of totalmobile data traffic in 2015 Mobile video traffic now accountsfor more than half of all mobile data traffic It will be verypromising if LTE with PSUN can support video traffic in the35 GHz band while coexisting with military radar

8 Conclusion

This paper proposes PSUN an OFDM transmission schemeenabling an LTE system to coexist with federalmilitary radarsin the 35 GHz bandThe scheme is comprised of PB at an Rxand precoding of null subcarriers at Tx of an OFDM systemTo maximize data rate OFDM Tx employing PSUN (i)localizes OFDM symbols to be radar-interfered a priori and(ii) shifts redundancy from channel coding to subcarriers intheOFDMsymbolsThis paper considers existence of sensingfunctionality in the 35 GHz band coexistence architectureand hence impacts of imperfect sensing which can occur dueto a sensing error by ESC and parameter changes by a radarResults show that PSUN is still effective in suppressing ICIremaining after PB even with imperfect pulse prediction andas a result enables an LTE system to support various usecases of 5G that require the data rate lower than 50Mbpsin the downlink and relatively larger file size such as videostreaming

Disclosure

This work was presented in part in the 2nd IEEE WCNCInternational Workshop on Smart Spectrum Technologies(IWSS 2016) Doha Qatar on 3 April 2016

Competing Interests


References

[1] NTIA An Assessment of the Near-Term Viability of Accom-modating Wireless Broadband Systems in the 1675ndash1710MHz1755ndash1780MHz 3500ndash3650MHz 4200ndash4220MHz and 4380ndash4400MHz Bands NTIA 2010

[2] Memorandum for the Heads of Executive Departments andAgencies Unleashing the Wireless Broadband Revolution 2010

[3] FCC 12-148 ldquoAmendment of the commisionrsquos rules with regardto commercial operations in the 3550ndash3650MHz bandrdquo Noticeof Proposed Rulemaking in GN Docket 12-354 2012

[4] FCC 14-49 ldquoAmendment of the commissionrsquos rules with regardto commercial operations in the 3550ndash3650MHzbandrdquo FurtherNotice of Proposed Rulemaking in GN Docket 12-354 2015

[5] FCC 15-47 ldquoAmendment of the commissions rules with regardto commercial operations in the 3550ndash3650MHz bandrdquo Reportand Order and Second Further Notice of Proposed Rulemakingin GN Docket 12-354 2015

[6] NTIA ldquoResponse to commercial operations in the 3550ndash3650MHz bandrdquo GN Docket 12-354 2015

[7] S Sodagari A Khawar T C Clancy andRMcGwier ldquoAprojec-tion based approach for radar and telecommunication systemscoexistencerdquo in Proceedings of the IEEE Global CommunicationsConference (GLOBECOM rsquo12) pp 5010ndash5014 Anaheim CalifUSA December 2012

[8] A Khawar A Abdel-Hadi and T C Clancy ldquoSpectrumsharing between S-band radar and LTE cellular system a spatialapproachrdquo in Proceedings of the IEEE International Symposiumon Dynamic Spectrum Access Networks (DYSPAN rsquo14) pp 7ndash14McLean Va USA April 2014


[9] R Saruthirathanaworakun J M Peha and L M CorreialdquoOpportunistic sharing between rotating radar and cellularrdquoIEEE Journal on Selected Areas in Communications vol 30 no10 pp 1900ndash1910 2012

[10] R Saruthirathanaworakun J M Peha and L M CorreialdquoGray-space spectrum sharing betweenmultiple rotating radarsand cellular network hotspotsrdquo in Proceedings of the IEEE 77thVehicular Technology Conference (VTC rsquo13) June 2013

[11] F Paisana J P Miranda N Marchetti and L A DaSilvaldquoDatabase-aided sensing for radar bandsrdquo in Proceedings of theIEEE International Symposium on Dynamic Spectrum AccessNetworks (DYSPAN rsquo14) pp 1ndash6 McLean Va USA April 2014

[12] M Ghorbanzadeh E Visotsky P Moorut W Yang and CClancy ldquoRadar in-band interference effects on macrocell LTEuplink deployments in the US 35 GHz bandrdquo in Proceedingsof the International Conference on Computing Networking andCommunications (ICNC rsquo15) pp 248ndash254 Garden Grove CalifUSA February 2015

[13] M Ghorbanzadeh E Visotsky P Moorut W Yang and CClancy ldquoRadar inband and out-of-band interference into LTEmacro and small cell uplinks in the 35 GHz bandrdquo in Pro-ceedings of the IEEE Wireless Communications and NetworkingConference (WCNC rsquo15) pp 1829ndash1834 March 2015

[14] H-A Safavi-Naeini C Ghosh E Visotsky R Ratasuk and SRoy ldquoImpact and mitigation of narrow-band radar interferencein down-link LTErdquo inProceedings of the IEEE International Con-ference on Communications (ICC rsquo15) pp 2644ndash2649 LondonUK June 2015

[15] S Kim J Choi and C Dietrich ldquoCoexistence between OFDMand pulsed radars in the 35 GHz band with imperfect sensingrdquoin Proceedings of the IEEE Wireless Communications and Net-working Conference Doha Qatar April 2016

[16] M Cotton and R Dalke ldquoSpectrum occupancy measurementsof the 3550ndash3650 Megahertz maritime radar band near SanDiego Californiardquo NTIA Report TR-14-500 2014

[17] Y Zhao and S-G Haggman ldquoSensitivity to Doppler shift andcarrier frequency errors in OFDM systems-the consequencesand solutionsrdquo in Proceedings of the IEEE 46th VehicularTechnology Conference vol 3 pp 1564ndash1568 Atlanta Ga USAMay 1996

[18] Y Fu and C Ko ldquoA new ICI self-cancellation scheme forOFDM systems based on a generalized signal mapperrdquo inProceedings of the 5th International Symposium on WirelessPersonal Multimedia Communications vol 3 pp 995ndash999IEEE 2002

[19] Y-H Peng Y-C Kuo G-R Lee and J-H Wen ldquoPerformanceanalysis of a new ICI-self-cancellation-scheme in OFDM sys-temsrdquo IEEE Transactions on Consumer Electronics vol 53 no4 pp 1333ndash1338 2007

[20] Q Shi Y Fang and M Wang ldquoA novel ICI self-cancellationscheme for OFDM systemsrdquo in Proceedings of the 5th Interna-tional Conference onWireless Communications Networking andMobile Computing (WiCOM rsquo09) pp 1ndash4 IEEE Beijing ChinaSeptember 2009

[21] The Next Generation Mobile Networks NGMN 5G WhitePaper The Next Generation Mobile Networks Ltd FrankfurtGermany 2015

[22] Operations and SignalSecurity Army Regulation 530-1 2005[23] S Brandes Suppression of Mutual Interference in OFDM Based

Overlay Systems Universitat Fridericiana Karlsruhe KarlsruheGermany 2009

[24] S Brandes U Epple and M Schnell ldquoCompensation of theimpact of interference mitigation by pulse blanking in OFDMsystemsrdquo in Proceedings of the IEEE Global TelecommunicationsConference (GLOBECOM rsquo09) pp 1ndash6 Honolulu Hawaii USADecember 2009

[25] U Epple D Shutin and M Schnell ldquoMitigation of impulsivefrequency-selective interference inOFDMbased systemsrdquo IEEEWireless Communications Letters vol 1 no 5 pp 484ndash487 2012

[26] A Goldsmith Wireless Communications Cambridge Univer-sity Cambridge UK 2005

[27] S Ahmed and M Kawai ldquoDynamic null-data subcarrierswitching for OFDM PAPR reduction with low computationaloverheadrdquo Electronics Letters vol 48 no 9 pp 498ndash499 2012

[28] M Ghogho A Swami and G B Giannakis ldquoOptimizednull-subcarrier selection for CFO estimation in OFDM overfrequency-selective fading channelsrdquo in Proceedings of the IEEEGlobal Telecommunicatins Conference (GLOBECOM rsquo01) pp202ndash206 San Antonio Tex USA November 2001

[29] B Wang P-H Ho and C-H Lin ldquoOFDM PAPR reductionby shifting null subcarriers among data subcarriersrdquo IEEECommunications Letters vol 16 no 9 pp 1377ndash1379 2012

[30] H V Poor An Introduction to Signal Detection and EstimationSpringer New York NY USA 2nd edition 1994

[31] JW Chong D K Sung and Y Sung ldquoCross-layer performanceanalysis for CSMACA protocols impact of imperfect sensingrdquoIEEE Transactions on Vehicular Technology vol 59 no 3 pp1100ndash1108 2010

[32] F Paisana N Marchetti and L A Dasilva ldquoRadar TV andcellular bands which spectrum access techniques for whichbandsrdquo IEEE Communications Surveys and Tutorials vol 16no 3 pp 1193ndash1220 2014

[33] 3GPP ldquoFurther advancements for EUTRA physical layeraspects release 9rdquo 3GPP TR 36814 V900 (2010-03) 2010

[34] Cisco ldquoCisco visual networking index globalmobile data trafficforecast updaterdquo White Paper 20152020 2016








Editorial Board




Contents


























Acknowledgments












1 Introduction









Tier-1

Tier-2

Tier-3

Pote

ntia

l ban

ds fo

r LTE

-U

depl

oym

ent

















2 Related Works










Tier‐1 IA system




Interface




RS w

ith li

cens

ed sh

ared

acce

ss (L

SA)








LTE BS

LTE-U UE

LTE-U UE

LTE BS

WiFi AP

WiFi STA


TE-U UE

WiFi STA


LTE-U UE

LTE-U UE


LTE BS

LTE BS

WiFi AP

WiFi STA

E-U UE

WiFi STA






119871B119882

Q119894119871 and Q119908


119871= Q1119871Q2119871 Q119894

119871 Q

|B119871|

119871

and Q119882= Q1119882Q2119882 Q119908

119882 Q

|B119882|








N119894 = sum|Q119894119871|

119906=1119899119894119906 119901119894119904119903 119901119887119904119903 119901119886119904119903 and 119901119902



119906119903 ℎ119887119906119903 ℎ119886119906119903 and ℎ119902




119906

AP and 119868119906




6040

20


Duty cycle




SINR119894119906119903[119896]

=119901119894119904ℎ119894119906119903

sum119887isinB119871119894

119901119887119904ℎ119887119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

BS

+ sum119886isinB119882

119901119886119904ℎ119886119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

AP

+ sum119902isinQ119882

119901119902119904ℎ119902119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

STA

+ 1205902 (1)

where 119887 119894 isin B119871



119873119894

119861=

K119894

sum119896

sum

119906isinQ119894119871

119877119906

sum119903

119882119894

119906119903log2(1 + SINR119894

119906119903[119896]) 119879OFDM (2)





119862119894=

119873119894119861

119879119894Tx + 119879119894

Wait (3)







power values 119901119894119904and 119901119887



119887



119862min119895



119894=1119862119894

1003816100381610038161003816B1198711003816100381610038161003816

(4)

minimize 119901119894


119871(5)

subject to 119901119894

119904 119901119887

119904 le 119875max


(6)

119901119894

119904 119901119887

119904 ge 119875min


(7)

119862119894gt 119862

min119895



119862119894

119901119894119904

gt119862min119895

119901119894119904

or119862119894

119901119894119904

gt119862min119895

119875min

or119862119894

119901119894119904

gt119862min119895

119875max

(9)


119901119894119904is simplified



119894=1(119862119894119901119894119904)

1003816100381610038161003816B1198711003816100381610038161003816


(10)






119895and 120573119894




119894(119889119894119895) based on 119862

119894gt 119862min119895

(3) if 120572119894119895(119898) = 120573119894

119895(119898) forall(119897 119898) isin 119872 then


119895 119889119894119895isin D119894 119895 isin U(1 |D

119894|) cap 119869 and update 119904 (8)

(7) Execute 119889119894119895 119901119894119904 observe 119862

119894and update 119899

119894119898(119889119894119895)

(8) if 119862119894gt 119862min119895


119894(119889119894119895) = 119877119894(119889119894119895) + 1


(11) Update 120572119894119895(119898) = 120572119894

119895(119898) + 1

(12) else(13) Reward 119877

119894(119889119894119895) = 119877119894(119889119894119895) + 0

(14) Update 119904 (119904 larr 119904 + 1) and V119894119898(119889119894119895) (11)

(15) Update 120573119894119895(119898) = 120573119894

119895(119898) + 1

(16) end if(17) if 119877

119894(119889119894

119895) = 119877119894(119889119894

119886) 119889119894119895 119889119894


then(18) Select 119889119894

119896 119889119894119896isin D119894 119896 isin U(1 |D

119894|) cap 119869

(19) else(20) Select 119889119894

119896 (12)


119894= 119889119894119896 119901119894119904





119871 A119894119894isinB119871

119862119894119894isinB119871






A119894= 119889119894119895 119901119894119904119895isin119869119904isin119878



119895 119889119894119895isin



119894= 1198891198941 1198891198942 119889119894



P119894= 1199011198941 1199011198942 119901119894



119895and



min119895


119904by one level











119896based


V119894119898(119889119894

119895) = 119909119894119898(119889119894

119895) + radic

2 ln (119898 +1003816100381610038161003816D119894

1003816100381610038161003816)

119899119894119898(119889119894119895)

(11)

119889119894

119896= argmax119889119894

119895isinD119894

(V119894119898(119889119894

119895)) (12)

where 119909119894119898(119889119894119895) = 119877

119894(119889119894119895)119899119894119898(119889119894119895) The argument of


119877119894(119889119894119895) and 119899









119894(119889119894119895) the

parameters 119904 1198991198940(119889119894119895) and V




119889119894119895 119889119894119895isin D119894 119895 isin U(1 |D



2is given





119871| = 7


119882| =



minus50 0 50

Dist

ance

(m)

100

50

0

minus50

Distance (m)

BSAP

WiFi


LTE-U WiFiCells






119895is


|P| =






0minus106 dBm

TTI 1ms








MAB

LTE WiFi

Scenario 1

Scenario 2

20 MHz

10 MHz 10 MHz














2

4

6

8

10

12

Agg

rega

te ca

paci

ty (b

ps)

times107

50 500250

ISD (m)



1

2

3

4

5

6

7

8

WiF

i cap

acity

(bps

)

times107

50 500250

ISD (m)







04

06

08

1

12

14

16

18

2

22

Agg

rega

te ca

paci

ty (b

ps)

120582L = 0 120582L = 15 120582L = 25 120582L = 35


times108



times107

5

6

7

8

9

10

11

12

Agg

rega

te ca

paci

ty (b

ps)






119875max and 119862119894 gt 119862min119895


119894= 1199011198941 1199011198942 119901119894

|P| =




50 500250

ISD (m)

0

2

4

6

8

10

12

Capa

city

(bps

)


times107


times108

50 500250

ISD (m)

05

1

15

2

25

Agg

rega

te ca

paci

ty (b

ps)








5STA10 STA15 STA

5 1510

Number of UEs

09

1

11

12

13

14

15

5th

perc

entil

e thr

ough

put (

bps)

times107


PC configuration


No PC Pmin Pmax PC2

4

6

8

10

12

14

Capa

city

(bps

)

times107


6 Conclusion



No PCPC

PminPmax

5 1510

Number of UEs

108

109

Ener

gy effi

cien

cy (b

itsjo

ule)



Competing Interests


Acknowledgments


References


























































1 Introduction









Frequency













2 System Model


1206011 1206012 120601


channels Γ = 1205741 1205742 120574






0 1119873times119872


119899119898= 1 if and only if 120574


119899

and 119897119899119898



Γ119899= 120574119898| 119897119899119898


119899of




119898sub


Spectrum broker

CM2M deviceTV




119898= 120601119899| 119897119899119898


119899119896119898| 119888119899119896119898

isin

0 1119873times119873times119872


119899119896119898=

1 if 120601119899and 120601


119898 and

119888119899119896119898


=

1minus119897119899119898

Value of 119888119899119896119898



119898as power and


1198991times119873



119899








119903119899= 120596119899sdot Δ 120596

119899isin N 1 le 119899 le 119873

BW119898= 120581119898sdot Δ 120581

119898isin N 1 le 119898 le 119872

(1)




119898 and BW

119898is the



120574119898=

120581119898

⋃

119894=1

119894119898

120581119898=BW119898

Δ

where 1 le 119898 le 119872

C =119872

sum

119898=1

120581119898

(2)

where 120574119898

has 120581119898



119898 Each



119894119898and F119867



F119867

119894119898minusF119871

119894119898= Δ for 1 le 119894 le 120581

119898 1 le 119898 le 119872 (3)



119899c = 119897119899119898119873timesC

Clowast = 119888lowast119899119896c | 119888

lowast

119899119896c = 119888119899119896119898119873times119873timesC

(4)

if1 le c le 120581

1for 119898 = 1

119898minus1

sum

119895=1

120581119895lt c le

119898

sum

119895=1

120581119895

for 1 lt 119898 le 119872(5)



Available spectrum


1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

2M

Hz

2M

Hz

2M

Hz

2M

Hz

3M

Hz

4M

Hz

4M

Hz








Cc 119886119899c119873times1 to




MSR =119873

sum

119899=1

119887119899 (6)


max119886

119873

sum

119899=1

119887119899

(7)


C

sum

c=1

119886119899c

=


119903119899


for 1 le 119899 le 119873

(8)

F119867

119889119905minusF119871

119890119891le MAS (9)

119886119899c = 0


(10)

119886119899c sdot 119886119896c = 0


(11)




119899 if


119899is

rejected and 119887119899= 0 If F119871

119890119891(1 le 119890 le 120581

119891and 1 le 119891 le 119872) is








119894=1sum

C119895=1119897lowast

119894119895










119899are












Δ sdot 119880(1 20)

119903119899

Δ sdot 119880(1 20)




10119877 (119877 is in km) and




U =sum119873

119899=1119887119899

sum119872

119898=1BW119898

(12)


L =sum119873

119899=1119903119899

sum119872

119898=1BW119898

(13)


ectr

um u

tilisa

tion

()

Network load

100

80

60

40

20

0

05 1 15 2 25 3 35 4 45

MSRAMSA

AASAARCAA






Spec

trum

util

isatio

n (

)

Network load

100

80

60

40

20

0

MSRAAASAARCAA

05 1 15 2 25 3 35 454 555






Network load05 1 15 2 25 3 35 454 555

MSRAAASAARCAA

Num

ber o

f rej

ecte

d de

vice

s

45

40

35

30

25

20

15

10

5

0








The b

est fi

tnes

s val

ue o

f MSR

A (M

Hz)


270

265

260

255

250

245

0 20 40 60 80 100


Freq

uenc

y (

)


tclt1

1lttclt2

2lttclt3

3lttclt4

4lttc

100

80

60

40

20

0



119894=1sum

C119895=1119897lowast




5 Conclusion


Competing Interests


References
























1 Introduction









SU route


PU1

PU2











1

2

3







119879



119871119880= 6955 + 2616 log

10


10

ℎ119879minus 119862119867

+ [449 minus 655 log10

ℎ119879] log10

119863[km]

(1)


medium cities

119862119867= 08 + (11 log

10


119877

minus 156 log10

119891[MHz]

(2)

and for big cities

119862119867

=

829 log10

(154ℎ119877)2

minus 11 150 le 119891[MHz]le 200

32 log10

(1175ℎ119877)2

minus 497 200 le 119891[MHz]le 1500

(3)







0 5 10 15 200

5

10

15

20

(dBm)

2

1

3

y(k

m)

x (km)


(a)

0

5

10

15

20

1

2

3

y(k

m)

0 5 10 15 20x (km)


(b)

0 5 10 15 200

5

10

15

20

1

2

3

y(k

m)

x (km)


(c)









Cone angle 60∘

Disc diameter 150mm


Feed pinDisc


Cone

Coax cable

N connector


05 1 15 2 25 3

0

2

Frequency (GHz)

Gai

n (d

Bi)

minus2

minus4

minus6

























RPmindBm= RPmin


= minus57 dBm(4)




3

(a)

1

(b)


Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

010

0

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

0 10 20 30 40 50 60Time (min)minus10

minus20

minus30

minus40

minus50

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus60

minus70

minus80

minus90

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz









Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

0

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus40

minus50

minus60

minus70

minus80

minus90

0 10 20 30 40 50 60Time (min)

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)


0

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty

Route 1




610MHz 746MHz770MHz

0

01

02

03

04

05

06

07

08

09

1

Prob

abili

ty

Route 2



610MHz 746MHz770MHz







Time (sec)

Prob

abili

tyRoute 1 Route 2

100

100

10minus1

10minus2

Prob

abili

ty

100

10minus1

10minus2

15dB20dB25dB

30dB35dB

15dB20dB25dB

30dB35dB

101 102

Time (sec)100 101 102


610MHz

746MHz

770MHz



610MHz

746MHz

770MHz






5 Conclusions



Competing Interests


References



















































1 Introduction
















PolicyManager


Radio Connection

Manager


Resource Manager

UnifiedRadio

Application

Flow Controller





(URAI)


RF transceiver

Radio platform


Baseband and others











LSA Repository

Mobile device

Base station

LSA controller

OAM

CORE network

NRA






























ActivateCnf(R2ID)

Deactivation

Creatinginstance

Activation



InstallRAReq(P2ID)

Certification


ResourceManager






Store


Downloaded


Installed

Instantiated

Active

Active

Deactivated



)

)

)










Driver


OS


Radio OS

App

1Ap

p 2

App

3

App M

Radio platform


RA1

RA2

RA3

RAN




Appl

icat

ion

Proc

esso

rRa

dio

Com

pute

r







Ubuntu1204 (OS)

Linux kernel


hardware(GPU)

FDD LTE TDD LTE


GbEUHD




ordinary PC

Appl

icat

ion

Proc

esso

rRa

dio

Com

pute






GUI


GbE

Spectrum analyzer





FDD LTE encoder

Video data stream

PC for eNB

RF transceiver

GbE

TDD LTE encoder

GbE RF transceiver



PC for MD

RF transceiver

GbE




5 Numerical Results















0


1198791 that is from 119905




3





MD



Antenna for eNodeB




1198793

1198794

1198795


(KPI = 1) 1445 1487(KPI = 1)

Legacy MD 1480 733 733 1480 1482



31198795 Therefore




31198795





0789

10111213141516

Time (sec)

Thro

ughp

ut (M

bps)


T1 T2 T3 T4 T5

t1 = 60 t2 = 120 t3 = 180 t4 = 240 t5 = 300









3 since eNB






4 During119879

5 as the LSA


1 Note


3and 119879






2 the





0 60 120 180 240 3000

1

2

3

4

5

6

7

Time (sec)

Accu

mul

ated

sum

rate

(Gbp

s)



T1 T2 T3 T4 T5


0 01 02 03 04 05 06 07 08 09 10

010203040506070809

1


CDF







6 Conclusion


Competing Interests


Acknowledgments


References

































1 Introduction





















incumbent







PS CS(1)

(2)

(3)


(a)

LSA (CS)

(2)

(3)


Inc

(PS)

(b)

Inc (CS)(1)

(3)


LSA(PS)

(c)

CS PS


PS sub CS

(d)































LSA server

LSA controller

LSA repository

LSA licenseeAP (PS)


IP network

Closed network

Incumbent




3 System Model








IP network

Tactical router

LTE access point

(eNodeB)S1

LSA repository

LSA server

Tactical network

Incumbent

transmitterreceiver

Tactical router

LTE access point

(eNodeB)

S1

Incumbent manager

IP network


controller dOMS


controller dOMS

IP network















































119894large is more




0 2 4 6 8 10 12 140

01

02

03

04

05

06

07

08

09

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)













0 2 4 6 8 10 12 140

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)






20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)



20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)








20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

lice

nsee


Ti (min)


20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m


Ti (min)



20 40 60 80 100 12008

085

09

095

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m

Ti (min)








5 Conclusion







Competing Interests


Acknowledgments


References






























1 Introduction












2 Related Work






3 Coexistence Model






Parameter Remark

120579beam





d

rc

Beam rotation


120579pass120579beam 120579beam






120579pass =360∘

sdot 119903119888

2120587119889 (1)


120579intf = 120579beam + 120579pass (2)




119879intf =120579intf360

sdot 119879rot (3)





348 3485 349 3495 35 3505 351 3515 352

0

10

20

30

40A

mpl

itude

(dB)

Radar (in-band)LTE

f (Hz)

minus10

minus20

minus30

times109








intf (msec)5 107 596 178810 157 874 262230 357 1985 5955


1015840





1015840








1015840


1015840


1015840



(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119883lowast

(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883lowast

(119896)

PSUN 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = 0







1015840

(sdot) indicate






0 10 20 30 40 50 60

0

005

Time

TransmittedA

mpl

itude

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(a) Transmitted

0 10 20 30 40 50 60

0

005

Time


minus005

Am

plitu

de

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05


100 20 30 40 50 60

0

005

Time

Received

Am

plitu

de

Pulse blanking

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05







0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3

10minus4



0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3




0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(a) Without PSUN

0 5 10 15 20 25 30minus1

minus05

0

05

1

Subcarrier

Am

plitu

de

(b) With PSUN (119903 = 12)

0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(c) With PSUN (119903 = 14)









Am

plitu

de

Am

plitu

de

Null carriers












1198670 119884 sim N (0 120590

2

0)

1198671 119884 sim N (0 120590

2

0+ 1205902

1)

(4)

where


1205902

0is power of noise

1205902



0 02 04 06 08 10

02

04

06

08

1

Miss

ed d

etec

tion

prob

abili

tyP

m







119875fa ≜ Pr (1198671| 1198670) = 1 minus Γ(

1

2120578se212059020

)

119875119898≜ Pr (119867

0| 1198671) = 1 minus Γ(

1

2

120578se2 (12059020+ 12059021))

(5)


Γ (119905 119911) =1

Γ (119905)int

119909

0

119905119905minus1

119890minus119909

119889119909 (6)


119898 Figure 7


119898










Parameter ValueLTE






62 Results







119898 As explained


119898 As







0 2 4 6 8 10 124050607080904050607080

EbNo (dB)

Dat

a rat

e (M

bps)



0 005 01 015 02 025 035055606570755055606570

Dat

a rat

e (M

bps)




0 02 04 06 08 150

55

60

65

70

Dat

a rat

e (M

bps)

120588



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pfa



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pm


(e) Versus 119875119898


0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pml












1ndash100 kbps





Others







8 Conclusion


Disclosure


Competing Interests


References




































Contents


























Acknowledgments












1 Introduction









Tier-1

Tier-2

Tier-3

Pote

ntia

l ban

ds fo

r LTE

-U

depl

oym

ent

















2 Related Works










Tier‐1 IA system




Interface




RS w

ith li

cens

ed sh

ared

acce

ss (L

SA)








LTE BS

LTE-U UE

LTE-U UE

LTE BS

WiFi AP

WiFi STA


TE-U UE

WiFi STA


LTE-U UE

LTE-U UE


LTE BS

LTE BS

WiFi AP

WiFi STA

E-U UE

WiFi STA






119871B119882

Q119894119871 and Q119908


119871= Q1119871Q2119871 Q119894

119871 Q

|B119871|

119871

and Q119882= Q1119882Q2119882 Q119908

119882 Q

|B119882|








N119894 = sum|Q119894119871|

119906=1119899119894119906 119901119894119904119903 119901119887119904119903 119901119886119904119903 and 119901119902



119906119903 ℎ119887119906119903 ℎ119886119906119903 and ℎ119902




119906

AP and 119868119906




6040

20


Duty cycle




SINR119894119906119903[119896]

=119901119894119904ℎ119894119906119903

sum119887isinB119871119894

119901119887119904ℎ119887119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

BS

+ sum119886isinB119882

119901119886119904ℎ119886119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

AP

+ sum119902isinQ119882

119901119902119904ℎ119902119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

STA

+ 1205902 (1)

where 119887 119894 isin B119871



119873119894

119861=

K119894

sum119896

sum

119906isinQ119894119871

119877119906

sum119903

119882119894

119906119903log2(1 + SINR119894

119906119903[119896]) 119879OFDM (2)





119862119894=

119873119894119861

119879119894Tx + 119879119894

Wait (3)







power values 119901119894119904and 119901119887



119887



119862min119895



119894=1119862119894

1003816100381610038161003816B1198711003816100381610038161003816

(4)

minimize 119901119894


119871(5)

subject to 119901119894

119904 119901119887

119904 le 119875max


(6)

119901119894

119904 119901119887

119904 ge 119875min


(7)

119862119894gt 119862

min119895



119862119894

119901119894119904

gt119862min119895

119901119894119904

or119862119894

119901119894119904

gt119862min119895

119875min

or119862119894

119901119894119904

gt119862min119895

119875max

(9)


119901119894119904is simplified



119894=1(119862119894119901119894119904)

1003816100381610038161003816B1198711003816100381610038161003816


(10)






119895and 120573119894




119894(119889119894119895) based on 119862

119894gt 119862min119895

(3) if 120572119894119895(119898) = 120573119894

119895(119898) forall(119897 119898) isin 119872 then


119895 119889119894119895isin D119894 119895 isin U(1 |D

119894|) cap 119869 and update 119904 (8)

(7) Execute 119889119894119895 119901119894119904 observe 119862

119894and update 119899

119894119898(119889119894119895)

(8) if 119862119894gt 119862min119895


119894(119889119894119895) = 119877119894(119889119894119895) + 1


(11) Update 120572119894119895(119898) = 120572119894

119895(119898) + 1

(12) else(13) Reward 119877

119894(119889119894119895) = 119877119894(119889119894119895) + 0

(14) Update 119904 (119904 larr 119904 + 1) and V119894119898(119889119894119895) (11)

(15) Update 120573119894119895(119898) = 120573119894

119895(119898) + 1

(16) end if(17) if 119877

119894(119889119894

119895) = 119877119894(119889119894

119886) 119889119894119895 119889119894


then(18) Select 119889119894

119896 119889119894119896isin D119894 119896 isin U(1 |D

119894|) cap 119869

(19) else(20) Select 119889119894

119896 (12)


119894= 119889119894119896 119901119894119904





119871 A119894119894isinB119871

119862119894119894isinB119871






A119894= 119889119894119895 119901119894119904119895isin119869119904isin119878



119895 119889119894119895isin



119894= 1198891198941 1198891198942 119889119894



P119894= 1199011198941 1199011198942 119901119894



119895and



min119895


119904by one level











119896based


V119894119898(119889119894

119895) = 119909119894119898(119889119894

119895) + radic

2 ln (119898 +1003816100381610038161003816D119894

1003816100381610038161003816)

119899119894119898(119889119894119895)

(11)

119889119894

119896= argmax119889119894

119895isinD119894

(V119894119898(119889119894

119895)) (12)

where 119909119894119898(119889119894119895) = 119877

119894(119889119894119895)119899119894119898(119889119894119895) The argument of


119877119894(119889119894119895) and 119899









119894(119889119894119895) the

parameters 119904 1198991198940(119889119894119895) and V




119889119894119895 119889119894119895isin D119894 119895 isin U(1 |D



2is given





119871| = 7


119882| =



minus50 0 50

Dist

ance

(m)

100

50

0

minus50

Distance (m)

BSAP

WiFi


LTE-U WiFiCells






119895is


|P| =






0minus106 dBm

TTI 1ms








MAB

LTE WiFi

Scenario 1

Scenario 2

20 MHz

10 MHz 10 MHz














2

4

6

8

10

12

Agg

rega

te ca

paci

ty (b

ps)

times107

50 500250

ISD (m)



1

2

3

4

5

6

7

8

WiF

i cap

acity

(bps

)

times107

50 500250

ISD (m)







04

06

08

1

12

14

16

18

2

22

Agg

rega

te ca

paci

ty (b

ps)

120582L = 0 120582L = 15 120582L = 25 120582L = 35


times108



times107

5

6

7

8

9

10

11

12

Agg

rega

te ca

paci

ty (b

ps)






119875max and 119862119894 gt 119862min119895


119894= 1199011198941 1199011198942 119901119894

|P| =




50 500250

ISD (m)

0

2

4

6

8

10

12

Capa

city

(bps

)


times107


times108

50 500250

ISD (m)

05

1

15

2

25

Agg

rega

te ca

paci

ty (b

ps)








5STA10 STA15 STA

5 1510

Number of UEs

09

1

11

12

13

14

15

5th

perc

entil

e thr

ough

put (

bps)

times107


PC configuration


No PC Pmin Pmax PC2

4

6

8

10

12

14

Capa

city

(bps

)

times107


6 Conclusion



No PCPC

PminPmax

5 1510

Number of UEs

108

109

Ener

gy effi

cien

cy (b

itsjo

ule)



Competing Interests


Acknowledgments


References


























































1 Introduction









Frequency













2 System Model


1206011 1206012 120601


channels Γ = 1205741 1205742 120574






0 1119873times119872


119899119898= 1 if and only if 120574


119899

and 119897119899119898



Γ119899= 120574119898| 119897119899119898


119899of




119898sub


Spectrum broker

CM2M deviceTV




119898= 120601119899| 119897119899119898


119899119896119898| 119888119899119896119898

isin

0 1119873times119873times119872


119899119896119898=

1 if 120601119899and 120601


119898 and

119888119899119896119898


=

1minus119897119899119898

Value of 119888119899119896119898



119898as power and


1198991times119873



119899








119903119899= 120596119899sdot Δ 120596

119899isin N 1 le 119899 le 119873

BW119898= 120581119898sdot Δ 120581

119898isin N 1 le 119898 le 119872

(1)




119898 and BW

119898is the



120574119898=

120581119898

⋃

119894=1

119894119898

120581119898=BW119898

Δ

where 1 le 119898 le 119872

C =119872

sum

119898=1

120581119898

(2)

where 120574119898

has 120581119898



119898 Each



119894119898and F119867



F119867

119894119898minusF119871

119894119898= Δ for 1 le 119894 le 120581

119898 1 le 119898 le 119872 (3)



119899c = 119897119899119898119873timesC

Clowast = 119888lowast119899119896c | 119888

lowast

119899119896c = 119888119899119896119898119873times119873timesC

(4)

if1 le c le 120581

1for 119898 = 1

119898minus1

sum

119895=1

120581119895lt c le

119898

sum

119895=1

120581119895

for 1 lt 119898 le 119872(5)



Available spectrum


1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

2M

Hz

2M

Hz

2M

Hz

2M

Hz

3M

Hz

4M

Hz

4M

Hz








Cc 119886119899c119873times1 to




MSR =119873

sum

119899=1

119887119899 (6)


max119886

119873

sum

119899=1

119887119899

(7)


C

sum

c=1

119886119899c

=


119903119899


for 1 le 119899 le 119873

(8)

F119867

119889119905minusF119871

119890119891le MAS (9)

119886119899c = 0


(10)

119886119899c sdot 119886119896c = 0


(11)




119899 if


119899is

rejected and 119887119899= 0 If F119871

119890119891(1 le 119890 le 120581

119891and 1 le 119891 le 119872) is








119894=1sum

C119895=1119897lowast

119894119895










119899are












Δ sdot 119880(1 20)

119903119899

Δ sdot 119880(1 20)




10119877 (119877 is in km) and




U =sum119873

119899=1119887119899

sum119872

119898=1BW119898

(12)


L =sum119873

119899=1119903119899

sum119872

119898=1BW119898

(13)


ectr

um u

tilisa

tion

()

Network load

100

80

60

40

20

0

05 1 15 2 25 3 35 4 45

MSRAMSA

AASAARCAA






Spec

trum

util

isatio

n (

)

Network load

100

80

60

40

20

0

MSRAAASAARCAA

05 1 15 2 25 3 35 454 555






Network load05 1 15 2 25 3 35 454 555

MSRAAASAARCAA

Num

ber o

f rej

ecte

d de

vice

s

45

40

35

30

25

20

15

10

5

0








The b

est fi

tnes

s val

ue o

f MSR

A (M

Hz)


270

265

260

255

250

245

0 20 40 60 80 100


Freq

uenc

y (

)


tclt1

1lttclt2

2lttclt3

3lttclt4

4lttc

100

80

60

40

20

0



119894=1sum

C119895=1119897lowast




5 Conclusion


Competing Interests


References
























1 Introduction









SU route


PU1

PU2











1

2

3







119879



119871119880= 6955 + 2616 log

10


10

ℎ119879minus 119862119867

+ [449 minus 655 log10

ℎ119879] log10

119863[km]

(1)


medium cities

119862119867= 08 + (11 log

10


119877

minus 156 log10

119891[MHz]

(2)

and for big cities

119862119867

=

829 log10

(154ℎ119877)2

minus 11 150 le 119891[MHz]le 200

32 log10

(1175ℎ119877)2

minus 497 200 le 119891[MHz]le 1500

(3)







0 5 10 15 200

5

10

15

20

(dBm)

2

1

3

y(k

m)

x (km)


(a)

0

5

10

15

20

1

2

3

y(k

m)

0 5 10 15 20x (km)


(b)

0 5 10 15 200

5

10

15

20

1

2

3

y(k

m)

x (km)


(c)









Cone angle 60∘

Disc diameter 150mm


Feed pinDisc


Cone

Coax cable

N connector


05 1 15 2 25 3

0

2

Frequency (GHz)

Gai

n (d

Bi)

minus2

minus4

minus6

























RPmindBm= RPmin


= minus57 dBm(4)




3

(a)

1

(b)


Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

010

0

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

0 10 20 30 40 50 60Time (min)minus10

minus20

minus30

minus40

minus50

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus60

minus70

minus80

minus90

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz









Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

0

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus40

minus50

minus60

minus70

minus80

minus90

0 10 20 30 40 50 60Time (min)

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)


0

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty

Route 1




610MHz 746MHz770MHz

0

01

02

03

04

05

06

07

08

09

1

Prob

abili

ty

Route 2



610MHz 746MHz770MHz







Time (sec)

Prob

abili

tyRoute 1 Route 2

100

100

10minus1

10minus2

Prob

abili

ty

100

10minus1

10minus2

15dB20dB25dB

30dB35dB

15dB20dB25dB

30dB35dB

101 102

Time (sec)100 101 102


610MHz

746MHz

770MHz



610MHz

746MHz

770MHz






5 Conclusions



Competing Interests


References



















































1 Introduction
















PolicyManager


Radio Connection

Manager


Resource Manager

UnifiedRadio

Application

Flow Controller





(URAI)


RF transceiver

Radio platform


Baseband and others











LSA Repository

Mobile device

Base station

LSA controller

OAM

CORE network

NRA






























ActivateCnf(R2ID)

Deactivation

Creatinginstance

Activation



InstallRAReq(P2ID)

Certification


ResourceManager






Store


Downloaded


Installed

Instantiated

Active

Active

Deactivated



)

)

)










Driver


OS


Radio OS

App

1Ap

p 2

App

3

App M

Radio platform


RA1

RA2

RA3

RAN




Appl

icat

ion

Proc

esso

rRa

dio

Com

pute

r







Ubuntu1204 (OS)

Linux kernel


hardware(GPU)

FDD LTE TDD LTE


GbEUHD




ordinary PC

Appl

icat

ion

Proc

esso

rRa

dio

Com

pute






GUI


GbE

Spectrum analyzer





FDD LTE encoder

Video data stream

PC for eNB

RF transceiver

GbE

TDD LTE encoder

GbE RF transceiver



PC for MD

RF transceiver

GbE




5 Numerical Results















0


1198791 that is from 119905




3





MD



Antenna for eNodeB




1198793

1198794

1198795


(KPI = 1) 1445 1487(KPI = 1)

Legacy MD 1480 733 733 1480 1482



31198795 Therefore




31198795





0789

10111213141516

Time (sec)

Thro

ughp

ut (M

bps)


T1 T2 T3 T4 T5

t1 = 60 t2 = 120 t3 = 180 t4 = 240 t5 = 300









3 since eNB






4 During119879

5 as the LSA


1 Note


3and 119879






2 the





0 60 120 180 240 3000

1

2

3

4

5

6

7

Time (sec)

Accu

mul

ated

sum

rate

(Gbp

s)



T1 T2 T3 T4 T5


0 01 02 03 04 05 06 07 08 09 10

010203040506070809

1


CDF







6 Conclusion


Competing Interests


Acknowledgments


References

































1 Introduction





















incumbent







PS CS(1)

(2)

(3)


(a)

LSA (CS)

(2)

(3)


Inc

(PS)

(b)

Inc (CS)(1)

(3)


LSA(PS)

(c)

CS PS


PS sub CS

(d)































LSA server

LSA controller

LSA repository

LSA licenseeAP (PS)


IP network

Closed network

Incumbent




3 System Model








IP network

Tactical router

LTE access point

(eNodeB)S1

LSA repository

LSA server

Tactical network

Incumbent

transmitterreceiver

Tactical router

LTE access point

(eNodeB)

S1

Incumbent manager

IP network


controller dOMS


controller dOMS

IP network















































119894large is more




0 2 4 6 8 10 12 140

01

02

03

04

05

06

07

08

09

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)













0 2 4 6 8 10 12 140

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)






20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)



20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)








20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

lice

nsee


Ti (min)


20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m


Ti (min)



20 40 60 80 100 12008

085

09

095

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m

Ti (min)








5 Conclusion







Competing Interests


Acknowledgments


References






























1 Introduction












2 Related Work






3 Coexistence Model






Parameter Remark

120579beam





d

rc

Beam rotation


120579pass120579beam 120579beam






120579pass =360∘

sdot 119903119888

2120587119889 (1)


120579intf = 120579beam + 120579pass (2)




119879intf =120579intf360

sdot 119879rot (3)





348 3485 349 3495 35 3505 351 3515 352

0

10

20

30

40A

mpl

itude

(dB)

Radar (in-band)LTE

f (Hz)

minus10

minus20

minus30

times109








intf (msec)5 107 596 178810 157 874 262230 357 1985 5955


1015840





1015840








1015840


1015840


1015840



(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119883lowast

(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883lowast

(119896)

PSUN 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = 0







1015840

(sdot) indicate






0 10 20 30 40 50 60

0

005

Time

TransmittedA

mpl

itude

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(a) Transmitted

0 10 20 30 40 50 60

0

005

Time


minus005

Am

plitu

de

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05


100 20 30 40 50 60

0

005

Time

Received

Am

plitu

de

Pulse blanking

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05







0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3

10minus4



0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3




0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(a) Without PSUN

0 5 10 15 20 25 30minus1

minus05

0

05

1

Subcarrier

Am

plitu

de

(b) With PSUN (119903 = 12)

0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(c) With PSUN (119903 = 14)









Am

plitu

de

Am

plitu

de

Null carriers












1198670 119884 sim N (0 120590

2

0)

1198671 119884 sim N (0 120590

2

0+ 1205902

1)

(4)

where


1205902

0is power of noise

1205902



0 02 04 06 08 10

02

04

06

08

1

Miss

ed d

etec

tion

prob

abili

tyP

m







119875fa ≜ Pr (1198671| 1198670) = 1 minus Γ(

1

2120578se212059020

)

119875119898≜ Pr (119867

0| 1198671) = 1 minus Γ(

1

2

120578se2 (12059020+ 12059021))

(5)


Γ (119905 119911) =1

Γ (119905)int

119909

0

119905119905minus1

119890minus119909

119889119909 (6)


119898 Figure 7


119898










Parameter ValueLTE






62 Results







119898 As explained


119898 As







0 2 4 6 8 10 124050607080904050607080

EbNo (dB)

Dat

a rat

e (M

bps)



0 005 01 015 02 025 035055606570755055606570

Dat

a rat

e (M

bps)




0 02 04 06 08 150

55

60

65

70

Dat

a rat

e (M

bps)

120588



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pfa



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pm


(e) Versus 119875119898


0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pml












1ndash100 kbps





Others







8 Conclusion


Disclosure


Competing Interests


References






















































Acknowledgments












1 Introduction









Tier-1

Tier-2

Tier-3

Pote

ntia

l ban

ds fo

r LTE

-U

depl

oym

ent

















2 Related Works










Tier‐1 IA system




Interface




RS w

ith li

cens

ed sh

ared

acce

ss (L

SA)








LTE BS

LTE-U UE

LTE-U UE

LTE BS

WiFi AP

WiFi STA


TE-U UE

WiFi STA


LTE-U UE

LTE-U UE


LTE BS

LTE BS

WiFi AP

WiFi STA

E-U UE

WiFi STA






119871B119882

Q119894119871 and Q119908


119871= Q1119871Q2119871 Q119894

119871 Q

|B119871|

119871

and Q119882= Q1119882Q2119882 Q119908

119882 Q

|B119882|








N119894 = sum|Q119894119871|

119906=1119899119894119906 119901119894119904119903 119901119887119904119903 119901119886119904119903 and 119901119902



119906119903 ℎ119887119906119903 ℎ119886119906119903 and ℎ119902




119906

AP and 119868119906




6040

20


Duty cycle




SINR119894119906119903[119896]

=119901119894119904ℎ119894119906119903

sum119887isinB119871119894

119901119887119904ℎ119887119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

BS

+ sum119886isinB119882

119901119886119904ℎ119886119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

AP

+ sum119902isinQ119882

119901119902119904ℎ119902119906119903⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟⏟

119868119906

STA

+ 1205902 (1)

where 119887 119894 isin B119871



119873119894

119861=

K119894

sum119896

sum

119906isinQ119894119871

119877119906

sum119903

119882119894

119906119903log2(1 + SINR119894

119906119903[119896]) 119879OFDM (2)





119862119894=

119873119894119861

119879119894Tx + 119879119894

Wait (3)







power values 119901119894119904and 119901119887



119887



119862min119895



119894=1119862119894

1003816100381610038161003816B1198711003816100381610038161003816

(4)

minimize 119901119894


119871(5)

subject to 119901119894

119904 119901119887

119904 le 119875max


(6)

119901119894

119904 119901119887

119904 ge 119875min


(7)

119862119894gt 119862

min119895



119862119894

119901119894119904

gt119862min119895

119901119894119904

or119862119894

119901119894119904

gt119862min119895

119875min

or119862119894

119901119894119904

gt119862min119895

119875max

(9)


119901119894119904is simplified



119894=1(119862119894119901119894119904)

1003816100381610038161003816B1198711003816100381610038161003816


(10)






119895and 120573119894




119894(119889119894119895) based on 119862

119894gt 119862min119895

(3) if 120572119894119895(119898) = 120573119894

119895(119898) forall(119897 119898) isin 119872 then


119895 119889119894119895isin D119894 119895 isin U(1 |D

119894|) cap 119869 and update 119904 (8)

(7) Execute 119889119894119895 119901119894119904 observe 119862

119894and update 119899

119894119898(119889119894119895)

(8) if 119862119894gt 119862min119895


119894(119889119894119895) = 119877119894(119889119894119895) + 1


(11) Update 120572119894119895(119898) = 120572119894

119895(119898) + 1

(12) else(13) Reward 119877

119894(119889119894119895) = 119877119894(119889119894119895) + 0

(14) Update 119904 (119904 larr 119904 + 1) and V119894119898(119889119894119895) (11)

(15) Update 120573119894119895(119898) = 120573119894

119895(119898) + 1

(16) end if(17) if 119877

119894(119889119894

119895) = 119877119894(119889119894

119886) 119889119894119895 119889119894


then(18) Select 119889119894

119896 119889119894119896isin D119894 119896 isin U(1 |D

119894|) cap 119869

(19) else(20) Select 119889119894

119896 (12)


119894= 119889119894119896 119901119894119904





119871 A119894119894isinB119871

119862119894119894isinB119871






A119894= 119889119894119895 119901119894119904119895isin119869119904isin119878



119895 119889119894119895isin



119894= 1198891198941 1198891198942 119889119894



P119894= 1199011198941 1199011198942 119901119894



119895and



min119895


119904by one level











119896based


V119894119898(119889119894

119895) = 119909119894119898(119889119894

119895) + radic

2 ln (119898 +1003816100381610038161003816D119894

1003816100381610038161003816)

119899119894119898(119889119894119895)

(11)

119889119894

119896= argmax119889119894

119895isinD119894

(V119894119898(119889119894

119895)) (12)

where 119909119894119898(119889119894119895) = 119877

119894(119889119894119895)119899119894119898(119889119894119895) The argument of


119877119894(119889119894119895) and 119899









119894(119889119894119895) the

parameters 119904 1198991198940(119889119894119895) and V




119889119894119895 119889119894119895isin D119894 119895 isin U(1 |D



2is given





119871| = 7


119882| =



minus50 0 50

Dist

ance

(m)

100

50

0

minus50

Distance (m)

BSAP

WiFi


LTE-U WiFiCells






119895is


|P| =






0minus106 dBm

TTI 1ms








MAB

LTE WiFi

Scenario 1

Scenario 2

20 MHz

10 MHz 10 MHz














2

4

6

8

10

12

Agg

rega

te ca

paci

ty (b

ps)

times107

50 500250

ISD (m)



1

2

3

4

5

6

7

8

WiF

i cap

acity

(bps

)

times107

50 500250

ISD (m)







04

06

08

1

12

14

16

18

2

22

Agg

rega

te ca

paci

ty (b

ps)

120582L = 0 120582L = 15 120582L = 25 120582L = 35


times108



times107

5

6

7

8

9

10

11

12

Agg

rega

te ca

paci

ty (b

ps)






119875max and 119862119894 gt 119862min119895


119894= 1199011198941 1199011198942 119901119894

|P| =




50 500250

ISD (m)

0

2

4

6

8

10

12

Capa

city

(bps

)


times107


times108

50 500250

ISD (m)

05

1

15

2

25

Agg

rega

te ca

paci

ty (b

ps)








5STA10 STA15 STA

5 1510

Number of UEs

09

1

11

12

13

14

15

5th

perc

entil

e thr

ough

put (

bps)

times107


PC configuration


No PC Pmin Pmax PC2

4

6

8

10

12

14

Capa

city

(bps

)

times107


6 Conclusion



No PCPC

PminPmax

5 1510

Number of UEs

108

109

Ener

gy effi

cien

cy (b

itsjo

ule)



Competing Interests


Acknowledgments


References


























































1 Introduction









Frequency













2 System Model


1206011 1206012 120601


channels Γ = 1205741 1205742 120574






0 1119873times119872


119899119898= 1 if and only if 120574


119899

and 119897119899119898



Γ119899= 120574119898| 119897119899119898


119899of




119898sub


Spectrum broker

CM2M deviceTV




119898= 120601119899| 119897119899119898


119899119896119898| 119888119899119896119898

isin

0 1119873times119873times119872


119899119896119898=

1 if 120601119899and 120601


119898 and

119888119899119896119898


=

1minus119897119899119898

Value of 119888119899119896119898



119898as power and


1198991times119873



119899








119903119899= 120596119899sdot Δ 120596

119899isin N 1 le 119899 le 119873

BW119898= 120581119898sdot Δ 120581

119898isin N 1 le 119898 le 119872

(1)




119898 and BW

119898is the



120574119898=

120581119898

⋃

119894=1

119894119898

120581119898=BW119898

Δ

where 1 le 119898 le 119872

C =119872

sum

119898=1

120581119898

(2)

where 120574119898

has 120581119898



119898 Each



119894119898and F119867



F119867

119894119898minusF119871

119894119898= Δ for 1 le 119894 le 120581

119898 1 le 119898 le 119872 (3)



119899c = 119897119899119898119873timesC

Clowast = 119888lowast119899119896c | 119888

lowast

119899119896c = 119888119899119896119898119873times119873timesC

(4)

if1 le c le 120581

1for 119898 = 1

119898minus1

sum

119895=1

120581119895lt c le

119898

sum

119895=1

120581119895

for 1 lt 119898 le 119872(5)



Available spectrum


1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

1M

Hz

2M

Hz

2M

Hz

2M

Hz

2M

Hz

3M

Hz

4M

Hz

4M

Hz








Cc 119886119899c119873times1 to




MSR =119873

sum

119899=1

119887119899 (6)


max119886

119873

sum

119899=1

119887119899

(7)


C

sum

c=1

119886119899c

=


119903119899


for 1 le 119899 le 119873

(8)

F119867

119889119905minusF119871

119890119891le MAS (9)

119886119899c = 0


(10)

119886119899c sdot 119886119896c = 0


(11)




119899 if


119899is

rejected and 119887119899= 0 If F119871

119890119891(1 le 119890 le 120581

119891and 1 le 119891 le 119872) is








119894=1sum

C119895=1119897lowast

119894119895










119899are












Δ sdot 119880(1 20)

119903119899

Δ sdot 119880(1 20)




10119877 (119877 is in km) and




U =sum119873

119899=1119887119899

sum119872

119898=1BW119898

(12)


L =sum119873

119899=1119903119899

sum119872

119898=1BW119898

(13)


ectr

um u

tilisa

tion

()

Network load

100

80

60

40

20

0

05 1 15 2 25 3 35 4 45

MSRAMSA

AASAARCAA






Spec

trum

util

isatio

n (

)

Network load

100

80

60

40

20

0

MSRAAASAARCAA

05 1 15 2 25 3 35 454 555






Network load05 1 15 2 25 3 35 454 555

MSRAAASAARCAA

Num

ber o

f rej

ecte

d de

vice

s

45

40

35

30

25

20

15

10

5

0








The b

est fi

tnes

s val

ue o

f MSR

A (M

Hz)


270

265

260

255

250

245

0 20 40 60 80 100


Freq

uenc

y (

)


tclt1

1lttclt2

2lttclt3

3lttclt4

4lttc

100

80

60

40

20

0



119894=1sum

C119895=1119897lowast




5 Conclusion


Competing Interests


References
























1 Introduction









SU route


PU1

PU2











1

2

3







119879



119871119880= 6955 + 2616 log

10


10

ℎ119879minus 119862119867

+ [449 minus 655 log10

ℎ119879] log10

119863[km]

(1)


medium cities

119862119867= 08 + (11 log

10


119877

minus 156 log10

119891[MHz]

(2)

and for big cities

119862119867

=

829 log10

(154ℎ119877)2

minus 11 150 le 119891[MHz]le 200

32 log10

(1175ℎ119877)2

minus 497 200 le 119891[MHz]le 1500

(3)







0 5 10 15 200

5

10

15

20

(dBm)

2

1

3

y(k

m)

x (km)


(a)

0

5

10

15

20

1

2

3

y(k

m)

0 5 10 15 20x (km)


(b)

0 5 10 15 200

5

10

15

20

1

2

3

y(k

m)

x (km)


(c)









Cone angle 60∘

Disc diameter 150mm


Feed pinDisc


Cone

Coax cable

N connector


05 1 15 2 25 3

0

2

Frequency (GHz)

Gai

n (d

Bi)

minus2

minus4

minus6

























RPmindBm= RPmin


= minus57 dBm(4)




3

(a)

1

(b)


Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

010

0

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

0 10 20 30 40 50 60Time (min)minus10

minus20

minus30

minus40

minus50

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus60

minus70

minus80

minus90

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz









Frequency (MHz)

Tim

e (m

in)

590 640 690 740 790

0

10

20

30

40

50

0

minus50

minus100

0

minus50

minus100

0

minus50

minus100

minus40

minus50

minus60

minus70

minus80

minus90

0 10 20 30 40 50 60Time (min)

610MHz

0 10 20 30 40 50 60Time (min)

746MHz

0 10 20 30 40 50 60Time (min)

770MHz

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)

Rece

ived

pow

er(d

Bm)


0

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty

Route 1




610MHz 746MHz770MHz

0

01

02

03

04

05

06

07

08

09

1

Prob

abili

ty

Route 2



610MHz 746MHz770MHz







Time (sec)

Prob

abili

tyRoute 1 Route 2

100

100

10minus1

10minus2

Prob

abili

ty

100

10minus1

10minus2

15dB20dB25dB

30dB35dB

15dB20dB25dB

30dB35dB

101 102

Time (sec)100 101 102


610MHz

746MHz

770MHz



610MHz

746MHz

770MHz






5 Conclusions



Competing Interests


References



















































1 Introduction
















PolicyManager


Radio Connection

Manager


Resource Manager

UnifiedRadio

Application

Flow Controller





(URAI)


RF transceiver

Radio platform


Baseband and others











LSA Repository

Mobile device

Base station

LSA controller

OAM

CORE network

NRA






























ActivateCnf(R2ID)

Deactivation

Creatinginstance

Activation



InstallRAReq(P2ID)

Certification


ResourceManager






Store


Downloaded


Installed

Instantiated

Active

Active

Deactivated



)

)

)










Driver


OS


Radio OS

App

1Ap

p 2

App

3

App M

Radio platform


RA1

RA2

RA3

RAN




Appl

icat

ion

Proc

esso

rRa

dio

Com

pute

r







Ubuntu1204 (OS)

Linux kernel


hardware(GPU)

FDD LTE TDD LTE


GbEUHD




ordinary PC

Appl

icat

ion

Proc

esso

rRa

dio

Com

pute






GUI


GbE

Spectrum analyzer





FDD LTE encoder

Video data stream

PC for eNB

RF transceiver

GbE

TDD LTE encoder

GbE RF transceiver



PC for MD

RF transceiver

GbE




5 Numerical Results















0


1198791 that is from 119905




3





MD



Antenna for eNodeB




1198793

1198794

1198795


(KPI = 1) 1445 1487(KPI = 1)

Legacy MD 1480 733 733 1480 1482



31198795 Therefore




31198795





0789

10111213141516

Time (sec)

Thro

ughp

ut (M

bps)


T1 T2 T3 T4 T5

t1 = 60 t2 = 120 t3 = 180 t4 = 240 t5 = 300









3 since eNB






4 During119879

5 as the LSA


1 Note


3and 119879






2 the





0 60 120 180 240 3000

1

2

3

4

5

6

7

Time (sec)

Accu

mul

ated

sum

rate

(Gbp

s)



T1 T2 T3 T4 T5


0 01 02 03 04 05 06 07 08 09 10

010203040506070809

1


CDF







6 Conclusion


Competing Interests


Acknowledgments


References

































1 Introduction





















incumbent







PS CS(1)

(2)

(3)


(a)

LSA (CS)

(2)

(3)


Inc

(PS)

(b)

Inc (CS)(1)

(3)


LSA(PS)

(c)

CS PS


PS sub CS

(d)































LSA server

LSA controller

LSA repository

LSA licenseeAP (PS)


IP network

Closed network

Incumbent




3 System Model








IP network

Tactical router

LTE access point

(eNodeB)S1

LSA repository

LSA server

Tactical network

Incumbent

transmitterreceiver

Tactical router

LTE access point

(eNodeB)

S1

Incumbent manager

IP network


controller dOMS


controller dOMS

IP network















































119894large is more




0 2 4 6 8 10 12 140

01

02

03

04

05

06

07

08

09

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)













0 2 4 6 8 10 12 140

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Tcheck (min)






20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)



20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

kno

wle

dge b

y th

e LSA

lice

nsee


Ti (min)








20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

lice

nsee


Ti (min)


20 40 60 80 100 1200

02

04

06

08

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m


Ti (min)



20 40 60 80 100 12008

085

09

095

1

Spec

trum

util

izat

ion

by th

e LSA

syste

m

Ti (min)








5 Conclusion







Competing Interests


Acknowledgments


References






























1 Introduction












2 Related Work






3 Coexistence Model






Parameter Remark

120579beam





d

rc

Beam rotation


120579pass120579beam 120579beam






120579pass =360∘

sdot 119903119888

2120587119889 (1)


120579intf = 120579beam + 120579pass (2)




119879intf =120579intf360

sdot 119879rot (3)





348 3485 349 3495 35 3505 351 3515 352

0

10

20

30

40A

mpl

itude

(dB)

Radar (in-band)LTE

f (Hz)

minus10

minus20

minus30

times109








intf (msec)5 107 596 178810 157 874 262230 357 1985 5955


1015840





1015840








1015840


1015840


1015840



(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119883lowast

(119896)


(119896) = 119883(119896)1198831015840(119896 + 1) = minus119890minus1198951205872

119883lowast

(119896)

PSUN 1198831015840

(119896) = 119883(119896)1198831015840(119896 + 1) = 0







1015840

(sdot) indicate






0 10 20 30 40 50 60

0

005

Time

TransmittedA

mpl

itude

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05

(a) Transmitted

0 10 20 30 40 50 60

0

005

Time


minus005

Am

plitu

de

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05


100 20 30 40 50 60

0

005

Time

Received

Am

plitu

de

Pulse blanking

minus005

0 10 20 30 40 50 60

0

05

1

15

Subcarrier

Am

plitu

de

minus05







0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3

10minus4



0 2 4 6 8 10 12EbNo (dB)

Bit e

rror

rate

10minus1

10minus2

10minus3




0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(a) Without PSUN

0 5 10 15 20 25 30minus1

minus05

0

05

1

Subcarrier

Am

plitu

de

(b) With PSUN (119903 = 12)

0 5 10 15 20 25 30

0

05

1

Subcarrier

Am

plitu

de

minus05

minus1

(c) With PSUN (119903 = 14)









Am

plitu

de

Am

plitu

de

Null carriers












1198670 119884 sim N (0 120590

2

0)

1198671 119884 sim N (0 120590

2

0+ 1205902

1)

(4)

where


1205902

0is power of noise

1205902



0 02 04 06 08 10

02

04

06

08

1

Miss

ed d

etec

tion

prob

abili

tyP

m







119875fa ≜ Pr (1198671| 1198670) = 1 minus Γ(

1

2120578se212059020

)

119875119898≜ Pr (119867

0| 1198671) = 1 minus Γ(

1

2

120578se2 (12059020+ 12059021))

(5)


Γ (119905 119911) =1

Γ (119905)int

119909

0

119905119905minus1

119890minus119909

119889119909 (6)


119898 Figure 7


119898










Parameter ValueLTE






62 Results







119898 As explained


119898 As







0 2 4 6 8 10 124050607080904050607080

EbNo (dB)

Dat

a rat

e (M

bps)



0 005 01 015 02 025 035055606570755055606570

Dat

a rat

e (M

bps)




0 02 04 06 08 150

55

60

65

70

Dat

a rat

e (M

bps)

120588



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pfa



0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pm


(e) Versus 119875119898


0 02 04 06 08 1

54565860626466687072

Dat

a rat

e (M

bps)

Pml












1ndash100 kbps





Others







8 Conclusion


Disclosure


Competing Interests


References




































Smart Spectrum Technologies for Mobile Information Systems · 2019. 8. 7. · Smart Spectrum...

Documents

Transcript of Smart Spectrum Technologies for Mobile Information Systems · 2019. 8. 7. · Smart Spectrum...