Dynamic Fractional Frequency Reuse Diversity Design for ...

Research ArticleDynamic Fractional Frequency Reuse Diversity Design forIntercell Interference Mitigation in Nonorthogonal MultipleAccess Multicellular Networks

Kashif Mehmood 1 Muhammad Tabish Niaz2 and Hyung Seok Kim 1

1Department of Information and Communication Engineering Sejong University Seoul Republic of Korea2Smart Device Engineering Sejong University Seoul Republic of Korea

Correspondence should be addressed to Hyung Seok Kim hyungkimsejongedu

Received 16 March 2018 Revised 7 June 2018 Accepted 25 June 2018 Published 15 July 2018

Academic Editor Kostas Peppas

Copyright copy 2018 Kashif Mehmood 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

Nonorthogonal multiple access (NOMA) is one of the few promising techniques that can ensure the achievement of benefitsforeseen in next-generation 5Gwireless networks and beyond By using superposition codingNOMAallowsmultiple users to sharethe same time and frequency resources thereby enhancing user connectivity spectral efficiency and a considerable increase in userthroughput Interference mitigation is an important consideration in NOMA and is considerably more influencing in multicellularenvironments First a brief description of the impairments that can arise in a NOMA cellular network along with responsiblefactors is provided Second different approaches adopted to minimize these impairments are discussed Finally a possible solutionis proposed that consists of a coordinated approach between the individual cells in the NOMA domain to minimize interferencesand improve user throughput Adaptive fractional frequency reuse (FFR) is used to allocate distinct frequency resources to edgeusers of different cells to minimize intercell interference in NOMA Simulation results prove that the proposed NOMA schemeplays an important role in minimizing impairment effects and enhancing the SINR and the throughput performance of edge userswhile ensuring fairness in its design

1 Introduction

Wireless cellular networks have seen unprecedented growthin the last decade in terms of increasing demand for user datarates as well as massive connectivity for users Multimediaapplications and services have seen a gradual and expectedincrease leading to the design of specific standards witha focus on seamless and smooth user experience Next-generation wireless networks including Long-Term Evolu-tion (LTE amp LTE-A) were designed considering the growinguser capacity needs and efficient use of the available spectrumto accommodate these users LTE only offers a couple offold improvement in user capacity over third-generation (3G)networks and will be insufficient considering the expectedgrowth

Orthogonal multiple access (OMA) has been used widelyin current and previous generations of wireless cellularnetworks for user access Frequency resources are allocated

in a disjoint manner to minimize interuser interferencethereby maximizing user throughput and connectivity up toa certain limit as allowed by the availability of frequencyresources Multiple OMA techniques currently being usedinclude frequency-division multiple access (FDMA) time-division multiple access (TDMA) code-division multipleaccess (CDMA) and orthogonal frequency-divisionmultipleaccess (OFDM)

One of the most important and challenging criterion fornext-generation (5G) cellular networks is for them to be ableto provide user throughput 1000times more than that of current4G network deployments To fulfill these requirementsnonorthogonal multiple access (NOMA) with a successiveinterference cancellation (SIC) receiver was presented as oneof the several promising candidate radio access techniques forfuture cellular networks OMA users are separated based ona resource division mechanism whereas a resource sharingapproach is adopted for NOMA schemes Resource sharing

HindawiWireless Communications and Mobile ComputingVolume 2018 Article ID 1231047 18 pageshttpsdoiorg10115520181231047

2 Wireless Communications and Mobile Computing

is accomplished by using the superposition principle where acomposite signal is constructed from individual user signalsand mapped onto a common frequency resource as opposedto one-to-one user resource mapping in OMA NOMAprovidesmassive connectivity as well as throughput enhance-ment obtained by sharing a single resource by multiple users[1] Spectral efficiency is an embedded advantage of NOMAover OMA schemes provided by the superposition of usersover a common resource

In NOMA user clustering is performed to pair userswith diverse channel responses together to maximize userthroughput gain and user capacity [1] Another addedadvantage of clustering process is the simplification of SICat the receiving end All paired users are then mappedonto orthogonal frequency resources to avoid interferencebetween user clusters which is also known as interclusterinterference Interference effectively reduces the benefitsoffered by NOMA over OMA Interference experienced canbe due to a number of factors Firstly incorrect channel stateinformation (CSI) causes errors during SIC decoding at theNOMA receiver for one ormore users depending on the userwho reported the wrong CSI Then clustering method canlead to the wrong pairing of users causing errors in the SICprocess Also user density in a cluster affects the complexityof SIC and the throughput limit of each user and finally thenumber of clusters defines the amount of bandwidth availablefor each cluster and eventually individual user throughputgains

For a multicell NOMA network another major source ofinterference that occurs between different clusters of adjacentcells is known as intercell interference (ICI) which willbe considered in the proposed work For the multicellularwireless network the available spectrum is allocated amongstdifferent clusters in a cell and spectral efficiency as well asuser capacity enhancement is achieved by employing thefrequency reuse scheme [2] NOMA can offer significantimprovements but only if ICI and cluster interferences aremanaged efficiently In this paper ICI isolation is achievedby frequency reuse diversity along with the efficient designof user clustering as well as efficient resource utilizationamongst NOMA users

11 Related Work In recent years many NOMA schemeshave been devised such as interleave-division multiple access(IDMA) [7] low-density spreading CDMA (LDS-CDMA)[8] pattern-division multiple access (PDMA) [9ndash14] sparse-code multiple access (SCMA) [15ndash17] multiuser sharingaccess (MUSA) [18ndash20] bit-division multiplexing (BDM)[21] low-density spreading OFDM (LDS-OFDM) [22]resource-spread multiple access (RSMA) [23 24] interleave-grid multiple access (IGMA) [25] multiuser bit-interleavedcoded modulation with iterative decoding (MU-BICM-ID)[26] and power domain nonorthogonal multiple access (PD-NOMA) [5 27ndash32]

PD-NOMA is considered in which users are distin-guished on the basis of allocated transmit power on a sharedresource [28] Power allocation in NOMA is carried outconsidering the detection process via SIC at the receiverhence a user with a poor channel gain ie cell edge (CE)

user is allocated a higher power and a lower power is assignedto the user with a better channel gain ie cell center (CC)user [28] CE users will be the ones most affected by theICI as they will have to cancel out any possible spilloversignals intended for CE users of adjacent cells Higher powersare allocated to CE users to ensure throughput as wellas sufficient received power at the base station (BS) fordetection [28] Both CC and CE users are then mappedonto respective frequency resource blocks (RBs)TheNOMAreceiver employs simultaneous multiuser detection (MUD)to identify and estimate each userrsquos data eliminating anyeffects of channel distortion or interuser interference withSIC [27] SIC is a technique used to successively extracta single user signal by the cancellation of unwanted usersignals in the received signal SIC performs optimally wheneach user experiences a channel (ie resource block) that isconsidered distinct from other users [1] However a rapidlychanging channel itself poses a problem of employing acomplex channel estimation algorithm to find the channelresponse for each user This feature makes NOMA a suitablecandidate for outdoor cellular networks as the path lossvaries considerably in outdoor scenarios as compared toindoors

FFR [2] and soft frequency reuse (SFR) [33] are twostudied solutions employing frequency reuse to improvespectrum efficiency and to reduce potential ICI in 4GnetworksThe available spectrum is divided into subchannelsthat are further divided into cell center and cell edge groupsBy allocating different power levels to users in each groupICI can be controlled effectively FFR provides a high networkthroughput as well as a high edge user SINR as compared toSFR which provides a balance between resource efficiencyand interference reduction The major difference betweenFFR and SFR is that in SFR cell center users can alsouse the spectrum allocated to cell edge users This causesconsiderably more interference to both center and edge userswhen compared with the FFR caseThis makes FFR a consid-erably better choice when interference reduction is of primeimportance In downlink OFDMA system performance isusually limited by ICI the edge users are the ones mostaffected by it FFR was explored as a possible solution in[2 33] to address the performance issues of edge users dueto interference Edge zones in cells are assigned a largerreuse factor in this scheme to reduce interference In staticFFR the reuse factor was decided at the time of frequencyplanning of a network this is inefficient anddoes not take intoconsideration the changing conditions that affect the edgezones of cells Therefore by using static FFR a satisfactorysystem performance for edge users cannot be achieved inreal environments due to the ever-changing channel andinterfering conditions This drawback of static FFR has alsobeen highlighted in [34] which studied a dynamic adap-tive frequency-division algorithm to improve cell averagethroughput especially the edge user throughput forOFDMAThe results achieved by this scheme are however restricted toscenarios inwhich a single user ismapped onto an orthogonalfrequency resource and no superposition coding is takingplace Detailed analysis of existing schemes for NOMAintercell interference management has been performed in

Wireless Communications and Mobile Computing 3

Section 2 Hence there is a need to develop a new schemefor ICIminimization for NOMAmulticellular environmentsand the prime focus of our work is to address this need

12 Contribution As already discussed the ICI mitigationscheme introduced for OFDMA systems [34] cannot beapplied to a NOMA scenario due to the fundamentaldifferences in the multiaccess approach for users Powerallocation diversity exists for NOMA users and is the basisof differentiating users in power domain Using SIC calls fora consideration of user clustering as well as the efficiencyof the SIC process User clustering performed to reducethe complexity and latency of the SIC process causes anoverload of users on a single resource block (RB) Thesefactors are a driving force in developing a strategy by usingthe advantages of frequency reuse diversity for ICImitigationin NOMA multicellular networks FFR is used to partitionthe system bandwidth into the center and edge bands andchannel allocation to the respective user will be performedby the band starting from edge users A combined powerand frequency allocation design are proposed that ensuresmaximum user performance for both edge and center usersby allocating more power and frequency channels respec-tively The adaptive nature of power and channel allocationas per fairness criteria ensures service to edge users beforecenter users A novel ICI mitigation approach is proposedthat includes the implementation of FFR by cell divisionfollowed by user classification into clusters and then resourceallocation A detailed discussion of the proposed design alongwith its implications will be performed in Section 4 Thefollowing contributions are made in this paper

(i) An FFR-based user clustering technique for NOMAuser distribution is proposed that starts with cellularsegmentation as per the discussed criterion followedby user classification as CE or CC Fairness is consid-ered while servicing these users to ensure symmetricservice to all users in the cellular service area as wellas reaping the benefits of NOMA

(ii) A dynamic power and frequency allocation schemefor NOMA users with proportional fairness for CCand CE users is proposed CE users are prioritizedwhile allocating resources to meet fairness criteriasince CC users have better service and channel con-ditions as compared to CE users

(iii) An FFR-based interference coordination scheme isproposed which makes use of the NOMA for provid-ing user access to the network in dense multicell net-works and meeting the guaranteed minimum servicerequirement for weak users in the network

(iv) A detailed analysis is performed depicting the advan-tages offered by NOMA over OMA and the signifi-cance of the frequency diversity technique used forcatering ICI User throughput analysis is performedto prove the benefits of the proposed scheme with afocus on selection criterion for cell segmentation andits impact onNOMAaswell as ICI experienced by CEusers

(v) It is proven by simulation results in Section 5 that theproposed scheme isolates edge users of neighboringcells effectively in a multicellular NOMA environ-ment to reduce experienced ICI The resource alloca-tion scheme considers the userrsquos fairness criteria andenhances NOMA capacity as well as the throughputfor CE as well as the CC users

(vi) A comparison is performed with existing ICI man-agement schemes (see Section 2) leading to a betterunderstanding of focused role and advantages of theproposed design

The rest of the paper is organized as follows In Section 2different approaches adopted for the mitigation of ICI arediscussed along with any related works using that approachIn Section 3 the system model for NOMA and FFR designfor analyzing the proposed scheme is described In Section 4the proposed adaptive FFR-based ICI mitigation techniquefor a NOMAmulticellular environment is described in detailSection 5 presents the simulation design for verifying theproposed scheme and performance results are discussed incomparison to proposed goals Concluding remarks alongwith possible future extensions and improvements of ourwork have been discussed in Section 6

2 Interference Mitigation Approaches

Interference plays a significant part in influencing communi-cation system design as well as robustness Major interferingsources include natural factors which influence channelbehavior like fog rain and pollution and channel specificfactors like user density as well as the clutter differencesof the covered areas Channel estimation techniques havebeen developed for a different channel and area types butall they can give are the instantaneous estimates Here weconsider the interference between users of different cellsusing the same channel ie ICI Therefore interferencemanagement (IM) is considered a critical part of a robustcommunication system design IM includes three main cat-egories of handling unwanted interference (i) interferenceavoidance (IAv) (ii) interference coordination (ICo) and (iii)interference cancellation (ICa) Interference avoidance (IAv)is considered to try to isolate the interfering entities fromthe intended users Interference coordination (ICo) includesthe design of a coordinated design to control resource allo-cation with an objective of minimizing the unintended usersignals Interference cancellation (ICa) takes an approach tocancel out the interfering parts of the received signal IAvcannot be applied in NOMA design due to the sharing ofresources amongst multiple users Proposed design consistsof a coordinated design where NOMA users are allocatedresources (channel or RBs and power levels) with a focus oninterference minimization

In a multicellular scenario the received signal becomeseven more complex due to the superposition of CE usersignals of different cells This makes interference removalmethods and channel modeling schemes used in single-cellnetworks inapplicable IM schemes are in use since longin wireless networks and following is the brief discussion


UE 1 UE 2 UE e

FreqPo

wer

P1

Pe

Freq

Pow

er

P2

Pe

Backhaul

NOMA-JT

Desired SignalInterference Signal

(a)

UE 1 UE e

Freq

Pow

er

P1

Freq

Pow

er

P2

Pe

Backhaul

NOMA-DCSUE 2


(b)

UE 1 UE 2 UE e

Freq

Pow

er

P1

Pe

Freq

Pow

er

P2

NOMA-CS

Dynamic Switching

Coordinated Silencing


(c)

UE 1 UE 2 UE 4

Freq

Pow

er

P1

P3

Freq

Pow

er

P2

P4

NOMA-CBUE 3


(d)

Figure 1 Multicell NOMA ICI mitigation solutions (a) NOMA-JT (b) NOMA-DCS (c) NOMA-CS and (d) NOMA-CB

of those schemes when applied to NOMA ICI causes asignificant decrease in CE user performance in multicellarchitecture as compared to single-cell design In this sectionrecent research efforts that combine IM approaches with

multicell NOMA are discussed Figure 1 shows some of theinterference coordination scenarios in a multicell networkICI is the main issue in multicell NOMA networks as itreduces a cell edge userrsquos performance Multicell techniques


are used to harness the effect of ICI These techniques can bebroadly categorized as coordinated processing (CP) and jointprocessing (JP) [6]This classification is based onwhether thedata messages desired by the users should be shared amongstmultiple BSs or not A single BS serves a user at any givenmoment in the case of the CP whereas in the JP multiple BSsservice a single user

21 Joint Processing (JP) In NOMA-JP user data is sharedamongstmultiple BSs before the user gets its data from one ormultiple BSs at the same time Two different approaches areusually considered here namely joint transmission (JT) anddynamic channel selection (DCS) JT is a technique in whichmultiple BSs serve a user simultaneously instead of disruptingeach other andDCS is a technique inwhichmultiple BSs havedata for the user but only one of them serve the user at a timeIn NOMA-JT edge users receive and process signals frommultiple BSs and interference can be effectively cancelledalong with improvement in edge user rates It gives the effectof a MIMO transmission as a transmission as a single useris receiving from multiple different transmitters and ICI canbe effectively cancelled as in the single-cell MIMO approach[5] A drawback of this technique is the CSI sharing overheadthat should be accurately available on all transmitters Acoordinated superposition coding (CSC) scheme is detailedin [3] which eliminates the CSI overhead altogether Thiscoordination between cells provides CEusers with a sufficienttransmission rate without any effect on CC user rates InNOMA-DCS despite user data availability at multiple BSsselected BS provides service to CE users whereas the CCusers are served unaffected by the corresponding BS Thiseliminates ICI as only one BSrsquos signal has user data henceCE users will consider signals from other BSs as only noise Itsimplifies the detection process but coordination is requiredbetween BSs for this scheme to work effectivelyThe selectionof a serving BS will be based on channel conditions amongstother factors Joint processing schemes require backhaulcoordination amongst different candidate and serving BSs ofthe network to decide the transmission mode as well as thesharing of CSI information for aiding the decision Signalingoverhead makes these techniques inefficient to implement innext-generation networks

A general architecture of JP techniques for a two-cellNOMA design is shown in Figures 1(a) and 1(b) where edgeusers are serviced jointly by both BSs (JT) or by a singleselected BS (DCS) as per selected mode of operation In JTmode multiple BSs can use Alamouti coded [4] signals totransmit simultaneously to edge users to enhance perfor-mance as well as throughput Center users will be transmittedtheir required signals as it is without any degradation dueto joint transmission to edge users Figure 1(a) shows thediscussed design where the edge user UE e is being served bytwo BSs jointly whereas UEs 1 and 2 are center users beingserved individually by respective BSs Similarly for DCSmode a single BS will be selected based on the mentionedcriteria to serve the edge UE while signaling backhaul is usedto intimate the network and other neighboring BSs of theselection decisionThis has been depicted in Figure 1(b) withonly one BS serving the edge user

22 Coordinated Processing (CP) In NOMA-CP user data isonly available at one BS and is not shared amongst multipleBSs although network information and CSI are usuallyshared for coordination Two different approaches can beapplied when CP is used coordinated beamforming (CB)and coordinated scheduling (CS) In CB data are available atonly one BS and the beamforming (BF) decision is made onbased on global CSI whichmust be accurate and this poses apossible drawback In [35] a possible solution is proposed inwhich joint optimization of BF vectors for BSs is performedsuch that there are no ICI and intercluster interference Aninterference alignment (IA) based CB algorithm is proposedthat uses only edge user channel information and as thenumber of users increases ICI is minimized without theneed for any CSI An interference channel alignment basedalgorithm is also mentioned but it requires CSI informationto operate In CS different BSs communicate with each otherto serve NOMA users with low ICI thereby ensuring properservice to CE users Only one of the coordinating BS willtransmit a composite NOMA signal to both its CE and CCusers whereas the other BSs will only serve their CC usersby sending their intended signals only instead of a compositesignal To the best of our knowledge no prior work hasbeen done utilizing CS approach with respect to multicellularNOMA networks

Figures 1(c) and 1(d) depict a CP based transmissionand interference mitigation approach for CS and CB designswhere a coordinated approach is adopted for CS and adirected BF is done to edge users of BSs respectively InCB mode BF will be done for CE and CS users withdifferent precoding design and BF vectors in order to satisfyminimum ICI for CE users Edge users UE 3 amp 4 will beisolated in Figure 1(d) from each otherrsquos beams due to designspecifications already selected to minimize the experiencedICI In CSmode depending on the channel conditions as wellas the ICI experienced by CE users edge users will be servedvia NOMA or traditional OMA techniques In Figure 1(c)UE c was receiving neighboring BS signals as well beforecoordinated silencing was activated Interfering neighboringBS will now only serve its CC users and ICI to UE c will beminimized as a result A challenging task here will be theselection of users to be scheduled by eachBS from the set of allregistered users which is an NP-hard optimization problem

These schemes were originally detailed for LTE and LTE-A networks which have been modified to accommodatechanges in NOMA schemes Figure 1 shows some of theinterference coordination scenarios in a multicell NOMAnetwork For better understanding a comparison of usercapacity amongst these schemes is also shown in Table 1where a two-cell architecture is considered inwhich each userand BS has T antennas There are also T user clusters in theproposed NOMA setup From [36] it is known that NOMAcan already support 2T users whereas OMA only supports Tusers

Themajor disadvantage of utilizing joint and coordinatedtransmission schemes for ICI mitigation is their inherentdependency on accurateCSI aswell as user channel allocationinformation This emphasizes the need for accurate acqui-sition along with efficient channel allocation for acquiring


Decoder

Decoder

Decoder

Decoder

Decoder

Decoder

Frequency

Pow

er

Ｂ1 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ1

Ｂ1 (Ｒ1 + Ｒ2) + Ｑ1

Ｂ1 Ｒ1+ Ｑ1

Ｒ3

Ｒ3

Ｒ3

Ｂ1

Ｂ1

Ｒ2

Ｒ2

Ｒ1

Ｂ3 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ3Ｓ3

Ｂ2 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ2Ｓ2

Ｓ2

Ｂ2 (Ｒ1 + Ｒ2) + Ｑ2

-

-

-

Ｂ1

Ｂ2 Ｂ2

Ｂ3

Ｂ1 gt Ｂ2 gt Ｂ3

０1

０2

０3

Figure 2 Single-cell NOMA network

Table 1 Multicell noma ici techniques

NOMA-CS

NOMA-CB NOMA-DCS NOMA-JT

Transmissionpoints 1 1 1 (selectable) ge 2

Sharedinformation

CSIscheduling CSI BF CSI data CSI data

BFBackhaulType Non-ideal Non-ideal Ideal Ideal

Number ofsupportedusers

lt 4T 4(T-1) 3T 3T or 4T

References [3] [4] [5 6]

maximum benefits from discussing approaches CSI cannotalways be estimated accurately for all the users or BSs whichhighlights the need for an alternate solution to the ICIproblem with minimum or no depending on the channelstate Efficient channel estimation techniques are needed tofully utilize the advantage offered by discussing schemesDiscussed schemes either require a large amount of datacooperation between users or a need for an accurate synchro-nization of channel state as well as task coordination betweenusers This can become traffic intensive for cellular networksso a novel technique is required which guarantees perfor-mance enhancement as well as interference minimization forNOMA users to extract maximum benefits over OMA InSection 4 a new FFR-based ICI minimization and avoidance

scheme is proposed which serve as the required alternative tothe discussed schemes

3 System Model

31 Single-Cell NOMA Consider an n-user downlinkNOMA system and assume that all users experiencedifferent channel responses The BS transmits 119899 differentsuperimposed signals which are multiplexed nonorthogo-nally in the power domain for each user using a single sharedfrequency resource as shown in Figure 2 Each user receivesthe composite signal consisting of all user signals and extractsits own signal using SIC A user classifies all signals except itsown as interference and cancels them out before retrieving itsown signalThere must be a considerable separation betweenuser signals so that SIC is able to separate and decode signalsfor all users This is ensured by the power allocation schemein NOMA that allocates power levels accordingly Usersnear the BS are allocated low power levels as they will havea better channel condition as compared to far users thatwill experience more fading and path loss Figure 2 showsa three-user downlink NOMA scenario with users havingchannel gains h1 h2 and h3 where h1 is the highest andh3 is the lowest Power allocations will be as shown withthe highest power allocated to UE3 and the lowest powerallocated to UE1 as it already has a strong channel responseThis ensures that far users having a weak channel gain willreceive lower interference levels from users having strongchannel gains due to the lower power allocated by NOMAMoreover strong users will receive more interference fromweak users due to higher power allocation by NOMA but as


Power

BS 1 BS 2

Cell-center user

ICI cancellation Detection Desired

signals

Treat Interference as

noise

NoiseInter-

channel interference

cancelled

Inter-channel

interference from BS 1

Superimposed signal for BS 2

Cell-edge users

Cell-center user

Superimposed signal Freq

Power

SIC DetectionDesired signalsIntra-cluster

interference

Freq

Figure 3 Multicell NOMA network

strong users have a better channel condition they will easilydecode via SIC A composite NOMA signal constructed forthe network in Figure 2 is represented as

119909 = 11990111199091 + 11990121199092 + 11990131199093 (1)

Now the received signal at each UEi can be described belowas

119910119894 = ℎ119894119909 + 119908119894 (2)

where hi is the channel response to the ith user and wi is thereceived noise including external and internal interferencesas well

32 Multicell NOMA In this section a multicellular down-link NOMA network and a SIC receiver design for thereception of the NOMA composite signal at each UE ismodeled as depicted in Figure 3 Two types of users aredefined in a multicellular setup CC users are near the BS andCE users are near the boundary of cell coverage In amulticellnetwork all users especially CE users will experience ICIirrespective of whether OMA or NOMA is used Howeverin the case of NOMA ICI is much worse as edge users willexperience ICI all the time as compared to OMA in whichonly some time slots or frequency bands will be affected InNOMA a key feature is that channel difference is usuallyused to pair users into clusters NOMA normally pairs users

experiencing strong and weak channel responses together toease the process of SIC As per assumption cell center usersdo not suffer from any ICI and only edge users are affected

Consider the downlink of a multicell NOMA scenariowith 119873 different cells and 119870 users in each cell The totalsystem bandwidth is denoted as 119861 and it will be furtherdivided into 119871 total subbands For simplicity the number ofreceiver antennas at user terminal is taken as 1 Each BS hasa total transmission power limit of Pmax Resource allocationfor each user is performed in terms of subchannels and thisgives us the benefit of multiuser diversity in the frequencydomain Now the multiuser scheduler maps a set of users119880119887 = 119906119887(1) 119906119887(2) 119906119887(3) 119906119887(119898119887) to a frequency block119887(1 lt 119887 lt 119871) Here 119906119887(119895) represents the jth (1 lt 119895 lt 119898119887)user index scheduled at frequency block b and mb denotethe total number of scheduled users at scheduled users atfrequency block b In the downlink BS will channel code andmodulate each user 119906119887(119895)th data independently of each otherThe available signal xb at a frequency resource b is the sumof 119906119887(119895)th coded modulation symbol 119904119887(119906119887(119895)) Therefore119904119887(119906119887(119895)) of allmb users is a superposition expressed as

119909119887 = 119898119887sum119895=1

radic119901119887 (119906119887 (119895))119904119887 (119906119887 (119895)) (3)

where 119864[|119904119887(119906119887(119895))|2] = 1 and 119901119887(119906119887(119895)) is the power levelassigned to user 119906119887(119895) for transmission at frequency block


119887 In Figure 3 CE users will have more interference as wellas more signal fading and hence a larger portion of availablepower in the composite signal must be allocated to them formeeting performance goals Consequently center users willbe allocated less power as compared to edge users becausecenter user signals will have less fading impact as comparedto edge users The received signal for each user 119906119887(119895) at afrequency block 119887 119910119887(119906119887(119895)) is shown as

119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)

where ℎ119887(119906119887(119895)) and 119908119887(119906119887(119895)) are the channel responseand the noise plus ICI contributions in the received powerrespectively experienced by user 119906119887(119895) at frequency block 119887Channel coefficientsmodeled as propagation loss shadowingloss and instantaneous fading loss coefficients will be keptconstant within a frequency block ICI is treated by theSIC receiver as white noise and it performs maximum ratiocombining (MRC) on received combined signal 119910119887(119906119887(119895)) as

119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)

where 119892119887(119906119887(119895)) = ℎ119887(119906119887(119895))2 and 119911119887(119906119887(119895)) are the equiv-alent channel gain and noise plus ICI afterMRC respectivelyThe average power level of channel gain is denoted as119899119887(119906119887(119895)) = 119864[|119911119887(119906119887(119895))|2] For NOMA signal receptioneach UE implements SIC to recover its individual signalfrom the superposed received signal The decoding order forSIC depends on the ratio between the channel gain and theinterference seen by each user that includes noise and ICI asalready discussedHenceNOMAuserwill in sequence detectsignals of all those users whose turn comes before decodingits own individual signal from the composite signal

If ICI can only be experienced from adjacent cells byeither a CC or CE user in the ith cell and no interference isencountered via SIC or other clusters in a cell the signal-to-interference-plus-noise ratio (SINR) for NOMA users onfrequency block 119887 is calculated as

119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)

ForNOMAusers in a cluster the achievable user rate for eachuser UEi can be represented as [1]

119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)

where 120583 is the number of channels assigned to the user and 119871is the bandwidth of each channel

4 Proposed Joint ICI Minimization ampResource Allocation Scheme

Frequency reuse schemes have since long been used toenhance user capacity and for efficient use of the allotted

frequency spectrum Frequency reuse has also found itsapplications for the minimization of interference betweenadjacent cells ensuring better performance for edge users asthey are the prime victims of this interference FFR and SFRare two possibilities when using frequency reuse diversity tocancel ICI [37] As previously discussed FFR ismore effectiveinminimizing ICI because of its isolation of channels for edgeand center users so it will be used in the proposed solutioninstead of SFR [2 33] SFR takes precedence over FFR (orstrict FFR) due to its greater resource efficiency due to thesharing of resources amongst CE and CC users [38]

Diversity in frequency reuse is used for the minimizationof ICI and different reuse factors are used for center andedge zones of cells [37ndash39] Frequency isolation is establishedby using a higher reuse factor in edge zones of cells ina multicellular environment Orthogonality is achieved byusing FFR in NOMA and this can also be known as an OMAover NOMA system where a feature inherent to orthogonalaccess for eliminating interference is used Static FFR wasfirst proposed inwhich fixed frequency and power allocationswere made to the edge and center users where a changingchannel will result in ICI Hence an adaptive FFR schemeis needed that will allocate power and frequency resourcesto the respective cell center and edge users depending oninherent channel conditions FFR is implemented along withNOMA by firstly dividing each cell into edge and centerregions and then allocating resources including power andspectrum to those resources User clusters are formed togroup users together in respective zones to reap the benefitsof NOMA as shown in previous works Cell zoning leadsto a degradation in performance of the proposed NOMAdue to the division of resources but this is consideredas a tradeoff in the proposed work In the case whereICI mitigation is not performed considerable performancedegradation has been observed when compared with thecases in which ICI mitigation is performed The proposeddesign is suboptimal in the sense that it tries to improve theinterference cancellation performance of proposed schemebased NOMA better than conventional NOMA in return forslightly reduced performance benefits of NOMA Anotheradded benefit of the proposed scheme is a need for low usercoordination as well as the accuracy of CSI information atrespective users

A multicell FFR scheme has been shown in Figure 4where the edge and center zone division has been donealong with bandwidth allocations to each zone A fractionof the band (BC) has been allocated to users in central zonewith a reuse factor of 1 Edge zones have been allocated afraction of the band from the edge user band with a reusefactor of 3 to avoid ICI with neighboring cells NOMA powerallocation factor for edge users will be higher as compared tocenter users to compensate for the reduction in bandwidthallocation Due to the separation of the band for edgeusers of neighboring cells ICI power will decrease whereassignal power will increase due to higher power allocationleading to an improved SINR for edge users for proposedNOMA-FFRbased ICIminimization scheme Salient featuresof proposed scheme have been discussed in detail in thissection


Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc

Figure 4 Fractional frequency reuse (FFR)

41 Cell Zoning As discussed NOMA clusters are made insuch away that users with the highest and lowest channel gainare paired together as CSI diversity improves SIC andNOMAperformance However in this case we will pair users a littlebroader as edge and center users by partitioning each NOMAcell A user categorization method is devised for categorizingavailable cell users into near and far zones It will help usin understanding the effects of ICI on these different usertypes One of the important parameter to consider in thisregard is the radius of coverage for center users denoted asrc which defines the boundary of the near (central) regionof the cell as per consideration To categorize users as centeror edge an estimate of their distance from the transmitteris required in each cell which can be difficult to acquireaccurately in practical scenarios Instead we will use twodifferent approaches

(i) Received SINR from the serving cell is compared tothreshold SINR value

(ii) Difference between received powers from serving andneighboring cells is compared to a threshold powerlevel

These techniques represent a composite user classificationcriterion that will be used to classify users in each cell ofour network as CE or CC Due to the difficulty in distancemeasurement between each individual user and cell centerSINR is instead used SINR at any given distance from theserving cell is directly dependent on the distance betweenthe user and the transmitting source and therefore it canbe used instead of distance to categorize users We define atotal of 119869 interfering adjacent cells for each cell and for eachuser either of the above-mentioned approaches are used tocategorize them into cell zones depending on the amount ofICI experienced For each user we have a serving cell andan arbitrary number of adjacent interfering cells SINR foreach user is represented as (6) and we can define the above-mentioned approaches for the kth user in each i-th cell as

(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)

where STH is defined as the decision threshold for the divisionof cells into edge and center zones and it depends on the CSIas well as the user density in each cell Users near the cellcenter experience almost negligible ICI and the second termin (8a) is close to zero which gives us only the SINR of eachuser to be used for comparison We can use (8a) for CE aswell as CC users but as ICI effects are minimal for CC users(8b) can be used instead for simplicity The average SINRof every user is calculated and categorization is performedaccordingly as near users will have a better SINR as comparedto far users Now each cell can label its users as a center oredge user depending on an SINR threshold already calculatedto depict the channel model and conditions Hence users aredivided into two groupsUE for the edge users andUC for thecenter users

42 Resource Allocation Users are considered to be uni-formly distributed in the service area of a cell as is thecase in most practical scenarios and as shown in [40]After distinguishing the edge and center users power andfrequency channel allocation will take place ensuring thatuser fairness is maintained across the cell for each cell inthe network In NOMA power allocation is carried outjointly and for a single user it not only limits the achievablethroughput of that user but all users in the NOMA clusterWe will consider the tradeoff between allocatingmore poweror bandwidth to users depending on their requirement andwhether they are on the edge or central zone of the cellCE users must be allocated more power to enable them toeffectively communicate with the transmitting sourceThis isbecause theywill be the onesmost affected by ICI due to thembeing in the edge zone of the cells CC users will be allocatedlower power levels as compared toCEusers because they havebetter channel conditions and a better SINR Power allocationto the edge and center users is done keeping in mind thatthe sufficient power difference exists between them in orderto ensure signal recovery via SIC at respective receivers Thiscan be depicted by the following condition which needs to besatisfied at each individual user

119875119894ℎ119894minus1 minus 119894minus1sum119895=1

119875119895ℎ119894minus1 ge 119875119898119886119909 (9)

wherePi is the allocated power level to the ith user in aNOMAcluster and hi is the normalized channel gain experiencedby that user whereas Pmax is the power budget for NOMAcluster As implied by (9) transmit power allocated to anyuser must be greater than the sum of transmit power ofall users with a relatively stronger channel defined as thenecessary condition for SIC decoding in NOMA systems [1]This ensures that sufficient power separation exists withineach cluster so that users can successfully distinguish eachuserrsquos signal in composite received NOMA signal It has beenassumed for simplicityrsquos sake that SIC is done perfectly withno error propagation so that focused analysis of ICI canbe performed which is the prime target of the proposedalgorithm

The available frequency spectrum will be divided intocell center and cell edge zones FFR will be applied by


assigning 13rd of the total spectrum B denoted as BE toCE users and remaining 23rd denoted as BC to CC usersin meeting their respective throughput requirements as wellfor compensating ICI All channels in the CE spectrummustalways be orthogonal to the CE channels in neighboringcells for ICI reduction The following parameters are definedin the resource allocation process used to ensure fairnessthroughput performance and ICI mitigation for the NOMAcellular network

(i) Edge user minimum rate threshold (Rmin)

(ii) Center user maximum rate threshold (Rmax)

(iii) Maximum cell power threshold (Pmax)

These parameters will ensure fair and efficient allocation ofpower and frequency resources to edge and center NOMAusers ICI will be eliminated by using FFR in the proposedscheme whereas throughput for all NOMA users will beensured by keeping a specific channel dependent rate limiton both central and edge users Channels from the availablespectrum will be allocated to edge users by considering theminimum rate requirement Rmin which will depend on thechannel conditions as well the available spectrum and powerallocations Rmin will ensure edge users get sufficient servicerates in proposed network design and it will also dictatethe amount of power which will be required for each edgeuser in NOMA setup Similarly channel allocation to centerusers will be carried out considering the Rmax rate which isneeded to restrict rates for center users remain within a limitwhen they will be operating in a NOMA mode along withedge users As base stations have specified power levels fordifferentmodes of operationPmax is defined as themaximumtransmitting power that can be allocated to a NOMA userEdge users will be allocated higher power levels in NOMA tocompensate for the path loss they will endure due to largerdistances as compared to center users Power allocation toNOMA edge users will be kept under this practical limit ofPmax

For an optimal solution the water-filling approach canbe used for power allocation to center and edge users asper their channel gains However this would require aniterative process starting from an initial assignment of powersto all users and then gradually refining power allocationfor each user Convergence will depend on the numberof users as well as the defined maximum average sumrate This process is computationally complex and dependson knowledge of already allocated powers to users in thebeginning Alternate solutions include firstly the allocationof fixed power to all users depending on a fixed allocationfactor which is adjustable and users will have informationabout their allocated power Secondly the fractional powerallocation approach can be used that compensates the chan-nel variations for userswith adaptive power controlThe latterapproach when used will make fair power allocation to usersin our design possible with low complexity and user feedback

Power is allocated to each user by using a proportionalfairness (PF) based technique [41] which will make surethat the resource allocation satisfies the given constraints

(1) Divide total bandwidth B into BC amp BE with a(2) total of L channels(3) for each ub in Ub(4) if 119906119887 997888rarr 119880119864 do(5) Assign a single channel(6) BE = BEndash1(7) if Pnb gt Pmax do(8) if BE = 120601 do(9) Assign another channel to reduce(10) required power amp meet Rmin(11) BE = BEndash1(12) else do(13) Set total assigned power of CE user(14) group to Pmax(15) end(16) end(17) UE = UE ndash 1(18) else if 119906119887 997888rarr 119880119862 do(19) for channels in BE(20) Map a center user on the same channel as(21) an edge user(22) Allocate power as per Rmax and considering(23) power allocation of edge user as well on(24) the same channel using (10)(25) PC = PE ndash 1(26) BC = BC ndash 1(27) end(28) for channels in BC(29) Assign channel and power to remaining(30) users from BC as per rate requirement Rmax(31) BC = BC ndash 1(32) end(33) end(34) ub = ub ndash 1(35) End

Algorithm 1 Fair resource allocation for edge and center users

Total transmission power allocated for each NOMA user 119896at frequency resource 119887 in an nth cell is given as

119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)

where Pnb and (119892119887(119895)119899119887(119895)) represent the total transmitpower of all users and the channel gain for the jth user forfrequency block 119887 respectively Ub is the set of users mappedto a single frequency resource 120573 is the decaying factor and avalue of 120573 = 0 will result in an equal distribution of powerto all users irrespective of channel gains Allocated powerwill decrease with the improvement in channel conditions ofthe NOMA users This signifies the role of channel gain andnoise along with ICI in the selection of power levels for edgeand center users Respective channel and power selectionschemes are described below and shown in Algorithm 1

421 Edge Users Edge users are the ones most likely to beaffected by ICI so their performance is prioritized to achieve


throughput and capacity gains which NOMA offers In thefirst stage channel and power will be allocated to edge usersto meet the rate requirements as defined by Rmin whose valuewill depend on channel conditions as well as a user fairnesscriterion All users in edge user group (UE) will be allocateda single channel and the power level will be derived fromthe minimum rate requirement for edge users using (7) Theessential condition for the SIC process (9) must be keptin mind during power allocation Another condition to besatisfied in this step is the bound as already defined in theform of Pmax which is the maximum transmit power of thetransmitting source in the cell The cumulative power of allusers in an nth cell for each frequency channel b should beless than the maximum transmit power in that cell as definedbelow

119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)

To ensure the validity of (11) multiple channels are allocatedto edge users so that a lower power level can be allocatedfor each channel The bandwidth allocated to each edgeuser will therefore increase and a lower power level will besufficient for them to meet the minimum rate conditionThisdecrease in power will also benefit in terms of interferencereduction between cells as compared to normal conditionsTherefore a fair resource allocation is achieved for edge usersby considering Pmax and Rmin bounds as well as ICI to bereduced for all the cells in the NOMA network

To evaluate cell edge performance we define an instan-taneous user rate for UEi obtained from (7) at any timeinstant t as Rit Edge users will be considered in outage whenRit lt Rmin so we can define average outage probability forproposed design as

119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)

whereP(119877119894119905 lt 119877119898119894119899) gives us the probability that an edge userwill be unable to meet the minimum rate criterion as definedin the proposed resource allocation scheme

422 Center Users Users close to the transmitting sourcereceive a high SNR as well as a low interfering power fromICI due to considerable path loss After sufficient powerand resource allocation to edge users center users will beallocated resources from the cell center resource pool Asingle center user is mapped onto the same channel as anedge user for maximizing capacity spectral efficiency andmaximum throughput for that channel ensured by NOMAFor each channel a NOMA cluster size of two is used where aCE edge user will be paired with a CC user but this does nothold true for all cases CC users can be paired together as wellon the same channel after ICI coordination is achieved Thisresource allocation to center users is performed consideringthe Rmax constraint as defined before and must be met in allcases for user fairness Power allocation values to center userswill be calculated by (8a) and (8b) and allocated accordinglyto ensure rate requirements Remaining channels will be

allocated to any center users left within the already describedconstraints

NOMA offers user throughput and capacity enhance-ments through sharing of spectrum resources by multipleusers [1] However in proposed algorithm bandwidth foredge and center users is somehow isolated This will lead toa loss in performance advantages offered by NOMA Inter-ferences experienced by NOMA users will become a con-siderable factor for diverse cellular environments currentlyin deployment due to high user density and small cell sizesThis will effectively reduce the user performance benefitsoffered by NOMA over OMAThis is a performance tradeoffintroduced when trying to minimize ICI by the proposedalgorithm Attempts have been made to compensate for theloss in performance by allocating more channels to centerusers or edge users and more power to edge users Thisaffects user performance but will considerably compensatefor the reduction in performance due to ICI experiencedby edge users A PF fairness-based scheduler is introducedwhich will serve the edge users on priority for meeting raterequirements

The main focus of the proposed design is to minimizeICI by isolating edge user bands in neighboring cells in amulticell environment An alternate scheme to deal withICI is proposed which does not have a dependency on CSIas other schemes already discussed in Section 3 A slightlymodified NOMA is implemented with OMA functionalityalso being used in order to cater for ICI This will have appli-cation feasibility in dense network deployments in futuregeneration networks like ultradense networks (UDN) Dueto a massive number of users channel state will be severelyaffected and a hybrid approach will be needed to compensatefor the introduced interferences User performance will becompensated by the allocation of more resources (power orchannels) depending on experienced interference levels aswell as target rate requirements Simulation results indicatethe suitability of the proposed scheme for ICI compensationin multicell environments by employing modified NOMAscheme

5 Performance Evaluation amp Results

In this section the proposed interference minimizationtechnique will be analyzed and the system performance willbe discussed The premise of the superiority of NOMA overOMA is proven along with the effectiveness of frequencyreuse diversity as a basis for ICI mitigation in FFR designThe minimum rate requirement criterion is proven to be aneffective condition in ensuring fairness in the NOMA systemLink-level simulations were performed in MATLAB withparameters given in Table 2

51 Simulation Setup Amulticellular network configurationis simulated with diverse parameters to create a realisticenvironment The network model consisting of 19 hexagonalcells (radius = 500 m) arranged in a wraparound manner ofneighbor relations is used for simulations Users are assumedto be distributed uniformly in each cell in either its edge or


Table 2 Simulation parameters

Cell layout 19 hexagonal cells 3 sectors per cellCell Radius 500 mShadowing factor Log-Normal with 8-dB Std DevShadowing correlation 045 (intercell) 1(intracell)Channel modeling 6-tap typical urban (TU6)Path loss model 1336 + 35 log10 (d[km]) dBThermal noise density -174 dBmHzBS transmit power 46ndash50 dBmSystem bandwidth 20 MHzChannel bandwidth 200 KHzNo of users 30 per cell (10 per sector)Sub-channels 100120572decay(decay factor) 06Frequency reuse factor 1(center) 13(edge)

OMA Subcarriers = 64 CP = 14 FFT =64-point

Digital modulation 8 16 64-PSKMaximum NOMA UserClustering Order 2

central zone depending on the cell zoning boundary Eachcell has exactly 30 users who are distributed randomly ineach of the cell zones with a maximum of 10 users per sectorUsers in each cell will suffer intercell residue power frominterfering cells fromfirst-order neighbors onlyThe availablespectrum of 20MHz is divided into subchannels of 200 KHzeach of which will be allocated to demanding users via ourresource allocation algorithm Wireless channel is assumedto be a dense urban design based on a 6-tap typical urban(TU) channel environment with Rayleigh fading Edge userswill experience ICI fromneighboring cells which is treated asnoiseThe distance-dependent path loss with a decay factor of35 is experienced by all users especially affecting edge usersalong with log-normal shadowing losses with a standarddeviation of 8dB At the receiver Turbo codes with a (13)root are used for error correction for ensuring data integrityChannel estimation is assumed to be idealwhich is performedvia pilot symbols embedded in OFDM design Moreoverperformance comparisons of the proposed NOMA basedICI mitigation scheme are performed with traditional OMAand NOMA based designs with no ICI mitigation techniqueNOMA design with the proposed scheme is also comparedwith available ICImanagement techniques to supplement theperformance of NOMA in medical environments

During the network setup phase each user selects itsserving BS based on the strongest received SNR from allthe communicating BSs Each BS is then divided into a cellzone (center or edge) depending on its proximity to its BSlocation A zone division distance of half the radius of eachcell is used initially and is later refined depending on theSNR threshold during simulations Users are respectivelyallocated to a cell zone depending on this zoning criterionas mentioned in the previous section In case of NOMA

users are prepared with a cluster size of 2 for simplicity FFRis then implemented in each cell after cell zoning has beencompleted Frequency reuse factors (RF) of 1 (for center zoneusers) and 3 (for edge zone users) are respectively usedto effectively represent the ICI scenario within the NOMAscheme In simulations only edge users are considered tobe affected by the ICI this is relatively a safe assumptionconsidering the cell zoning process and the distance betweencenter zones of neighboring cells Edge and center usersare then allocated appropriate resources as per the fairnesscriterion depicted as the conditions mentioned before Theexact values of these parameters depend on specific channelconditions and are determined analytically Both users withina cluster are then allocated appropriate powers as per theirCSI such that they can bemultiplexed together usingNOMAAt each UE SIC is performed to extract its data from thesuperposed signal Network simulations are then performedto confirm the benefits of the proposed scheme

52 Simulation Results To evaluate the proposed algorithmall mentioned premises as well as assumptions will beanalyzed OMA and NOMA are compared to establish thesuperiority of NOMA performance over OMA consideringthe user fairness conditions already mentioned The impactof reuse factor diversity on user SNR is discussed to provethe effectiveness of FFR in the proposed solution to ICI Ananalysis of the proposed algorithm is performed with respectto the relation between user and power ratios for the centerand edge zones with a focus on throughput performance Itis concluded from our discussions and the results that theproposed NOMA design outperforms conventional NOMAin terms of interference and throughput performance

521 Fair NOMA versus OMA For fair NOMA the powerallocated to the center and edge users will not be fixed butit will be carried out in such a way as to ensure symmetry inperformance between the center and edge users For edgeUEtheminimum rate requirement (Rmin) and themaximum raterequirement (Rmax) will be considered to ensure appropriateservice and fairness in NOMA design Fair NOMA willoffer a higher capacity than fixed power NOMA and OMAHowever this behavior will change as SNR increases and forconsiderably larger values the capacity performance of fixedand fairNOMAapproacheswill be almost similar as shown inFigure 5 This behavior has also been highlighted in [42] andas the SNR approaches infinity no matter how much poweris allocated to the stronger user the capacity increase will beconstant

522 Frequency Reuse Diversity FFR makes use of the fre-quency reuse concept to distinguish cell zoneswhere differentreuse factors are used for center and edge zone users to isolateICI for edge users Frequency reuse diversity is the key featurein enabling the proposed algorithm to effectively minimizeany interference from neighboring cells Figure 6 depicts theeffects of choosing different reuse factors with respect touser SNR for the center and edge users It clearly shows thatcenter users having a lower reuse factor (RF = 1) will perform


Fixed-Power NOMAFair-NOMAOMA

2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR

Figure 5 Capacity performance of fair versus fixed power NOMA

Center User Group (RF = 1)Edge User Group (RF = 3)

50400 10 20 30 60 70minus20 minus10minus30Average user SINR (dB)

0

01

02

03

04

05

06

07

08

09

1

CDF

Figure 6 Frequency reuse diversity analysis

considerably better than the edge users (RF = 3) Edge userswill experience higher ICI as compared to center users andthis explains their distribution behavior The dependency ofreceived SINR by users in the center and edge zones overfrequency reuse diversity has been depicted in experimentalobservations Center users due to better channel conditionsalong with low interference factors will have higher SINRswhich is shown to be gt -10dB for more than 95 of the usersEdge users will experience ICI from a larger number of cellsdue to a higher reuse factor This results in a lower SINR ascompared to center users and is shown in Figure 6 to be gt 0dBmore than 95 of the users in that regionThis shows thatedge users with a higher reuse factor (RF = 3) will experienceworst SINR as compared to center users with a lower reuse

CRR = 01CRR = 03

CRR = 05CRR = 07

40300 10 20 50 60minus20 minus10minus30Average user SINR (dB)

0

01

02

03

04

05

06

07

08

09

1

CDF

Figure 7 Center ratio analysis of user SNR

factor (RF= 1) This also provides a solid basis for NOMAclustering due to a significant difference in SINRs of users inboth cell zones

523 Dynamic Fair NOMA FFR FFR performance dependson howwell the cell zone division has been performed as wellas the amount of power has been allocated to each user groupEdge users will be allocated more power as per the NOMArequirement considering the channel degradation due to alarger distance from cell center degradation due to a largerdistance from cell center and associated path losses Twoparameters have been defined for analyzing the proposeddesign (i) center power ratio (CPR) that is the ratio of powerallocated to center users to total transmit power (ii) Centerradius ratio (CRR) that is the ratio of the radius of the centralzone to cell radius (iii) edge radius ratio (ERR) which isthe ratio of edge radius and cell radius Figure 7 plots theSINR distribution of users with different center radius ratiosWhen CRR = 01 the central region of the cell is very smallas compared to the edge region This will cause a majority ofusers (about 97) to have an SINR value greater than or equalto -10 dB As the center radius value is increased more userswill be distributed in a relatively larger center zone causingbetter SINR values for users This can be seen for CRR = 07cases in which about the same number of users have SINRvalues greater than or equal to a much lower value of -20 dB

Figures 8 and 9 depict the throughput values for users fordifferent power ratios as the central radius ratio is alteredFor each CPR value simulations are performed for differentvalues of CRR to obtain the throughput behavior of thecenter and edge user groups along with their cumulative sumFor center group users the throughput will increase withan increase in CRR as the number of users in the centralregion will increase and more channels will be allocatedto central users Interestingly a decrease in throughput isobserved as we cross the half-radius threshold of the cell


CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio

Figure 8 CPR impact on center user throughput

CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)

Figure 9 CPR impact on edge user throughput

and increase CRR beyond this point This is due to theincrease in ICI observed by the center users as they are nowgradually getting closer to center zones of other cells Centralusers are allotted the same frequencies in all cells and ICIcannot be ignored if center zones are greater than the half-cell radius This is also partially due to the userrsquos fairnesscriterion due to the imposition of maximum rate limit Rmaxon center users which is an integral part of our algorithmFor edge group users as the values of CRR increase there isa gradual decrease in overall throughput of edge users andthis is simple to perceive as the central zone is becomingbigger and a larger number of users will be registered ascentral users The overall average throughput of edge userswill always be declining due to the decreasing number of usersin edge zones as compared to the central zones As per ouralgorithm edge users are already on orthogonal frequency

ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)

Figure 10 ERR impact on center user throughput

bands with a frequency reuse factor of 3 so ICI is alreadyreduced by using the FFR technique We will make sure thatedge users receive proper service which is determined bythe minimum rate threshold Rmin by allocating appropriateresources consistently

Figures 10 and 11 show the effects of the changing edgezone radius on user throughputs Edge zone radius variedbetween 5 and 20 of the whole cell radius while observingits impact on user throughputs and affecting factors For thecenter group an increase in user throughputs is observeduntil it crosses the half-radius limit or the overlap with edgezone starts for different ERR values Afterward a decreaseis observed due to ICI experienced by central users as wellas due to any false classifications of edge and central usersincorrectly into the wrong zone due to interzonal overlapbetween central and edge zones For the edge group differentERR values are adopted and CRR is altered to observe userthroughput behavior As the central zone radius of the cellincreases more users are included in the central zone ascompared to the edge zone which will cause a decrease inthroughput for edge users A steeper decline is observed afterthe specific value of CRR for each ERR value due to theoverlap of both edge and center bands which will lead userfalse classifications in both bands For both center and edgeusers the fairness criterion is also enforced respectively andwill also limit the achievable throughput by both user groupsand once it has been achieved a decline is observed after thatpoint

Performance enhancement for NOMA using the pro-posed algorithm can be clearly identified by comparingcumulative distribution functions (CDF) of both center andedge user groups for different power ratio values It canbe clearly seen in Figure 12 that by applying the proposedscheme a significant improvement is observed that hasdifferent implications for both center and edge user groupsFor center user group a lower power ratio is required forNOMA implementation to the proposed design as compared


ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio

Figure 11 ERR impact on edge user throughput

Proposed NOMAConventional NOMA

Center Edge

0

02

04

06

08

1

CDF

01 02 03 04 05 06 07 08 090 1Power Ratio (Center amp Edge User Groups)

Figure 12 Power allocation impact on center and edge user groups

to the conventional NOMA system Power allocation for edgeusers in the proposed scheme is greater than conventionalNOMA to ensure compensation of the ICI experienced byedge users CC users will have a higher bandwidth availableto them as compared to CE users after the implementation ofFFR in the proposed NOMA design thereby providing CCusers more freedom in the frequency domain

In Figure 13 spectral performance trends of CC and CEusers is depicted under different transmission modes and thecomparison is performed with proposed and conventionalNOMA schemes for benchmarking No impact is observedfor CC users in either of the mentioned schemes with achange in location within the center zone of the cells dueto considerably lower levels of ICI A point to observe hereis that our scheme provides just enough (but still higherthan OMA) spectral efficiency to center users due to a

CenterNOMA-CBCenterNOMA-JTCenterNOMACenterNOMA-FFRCenterOMA

EdgeNOMA-CBEdgeNOMA-JTEdgeNOMAEdgeNOMA-FFREdgeOMA

0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)

100 150 200 25050Edge User Distance (m)

Figure 13 Spectral performance comparison of ICI mitigationschemes

hybrid design and clustering limitations in user pairings(lack of CSI diversity amongst users) Generally a decreasein performance for OMA and NOMA is observed with thechanging location of CE users as no ICI mitigation is appliedNOMA-JT matches the performance of NOMA-CB with anincrease in gain as the CE users get closer to the cell boundarybecause CE user can now take advantage of the link fromthe neighboring BS to increase its SINR via data sharingOMA outperforms all the schemes when CE users are closeto BS mainly due to the remaining interuser interferencein all NOMA schemes Proposed scheme outperforms OMAas well as conventional NOMA schemes in the edge zoneof the cell owing to better ICI handling capability as wellas low signaling overhead and data sharing requirement asin schemes like NOMA-CB and JT This provides a rathersimplistic ICI mitigation design as compared to CB and JTbased designs due to limited coordination required amongstusers which leads to savings in computational capability andinformation acquisition design

In Figure 14 outage performance of the proposed schemeis compared with OMA and available NOMA approachesfor handling ICI Outage performance of proposed schemecan be analyzed by obtaining probability of edge users beingin outage from (12) for the proposed scheme For con-ventional NOMA and OMA schemes outage probabilitieswere approximated as shown in detail by Oviedo [42] Finalformulas have been included for reference in the AppendixExpectedly OMA and conventional NOMA design have ahigher chance of being in outage due to inability to caterfor any experienced ICI by users in edge zone NOMA-CB requires a complex beamforming and a beam steeringmechanism in order to effectively cancel out any effects of ICIon edge users leading to an improved outage performance


NOMA-JTNOMA-CBOMA

Conventional NOMAProposed NOMA-FFR

10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)

Figure 14 Outage performance comparison of CE users

Similarly NOMA-JT improves outage performance due tothe inherent data sharing in CE users amongst neighbor-ing cells leading to an improved SINR as transmit poweris increased Lastly the proposed scheme outperforms allprevious schemes by employing cell zoning as well as ICIaware power allocation and user clustering in respective cellzones for CC and CE users NOMA-FFR suffers in terms ofbetter spectral utilization as compared to other ICImitigationtechniques but makes up for it by improving the interferencehandling capability of users

6 Conclusion

In this paper the importance of interference mitigation inthe multicellular downlink NOMA design was demonstratedand different possible techniques were discussed that can beused to minimize and isolate ICI to improve edge user per-formance A proportional fairness-based channel allocationand power control algorithm were then proposed to achieveICI minimization by exploiting a rather known techniqueof FFR Numerical results indicate that NOMA design withthe proposed scheme improves the user performance forboth edge and center users Power allocations have a directimpact on achieving user rates as expected in NOMA aswell as compensation of the experienced interference inenvironment Effects of an efficient selection of cell zoningwith respect to user density were also discussed and it hasbeen emphasized that the selection of cell zoning thresholdplays a key role in ensuring service toNOMAusers especiallyusers in the edge zone Factors that have an impact on ICI inNOMA include power and resource allocation cell zoningand a suitable selection of fairness thresholds for edge andcenter users ICI can further be minimized by using differentmodified forms of FFR (eg SFR DFFR) and will be theprime focus of any future enhancements in this work CoMPtechniques can also be used for cell edge so that interference

effects can be minimized by mutual information sharingSIC error minimization for NOMA is another potential areaof research that can be exploited for enhancing intraclusterperformance in NOMA

Appendix

Outage Probability of NOMA and OMA

Outage performance analysis has already been performed forNOMA and OMA systems in previous works like Oviedo[42] which has been utilized in this paper for comparisonpurposes According to channel gain information (|ℎ2| gt|ℎ1|) 1198801198641 and 1198801198642 will be present in the edge and centerregion of the cell respectively Power allocation to 1198801198641 willbe more to compensate for low channel gain and vice versa

Here outage probability expressions have been presentedfrom [42] for OMA users that are given with respect tochannel gain experiencedwithin a cluster by users in differentcell zones

1198751198741198721198601 = 1 minus exp[minus2 (4119877119898119894119899 minus 1)120573119875119898119886119909 ] (A1)

1198751198741198721198602 = 1 + exp[minus2 (4119877119898119894119899 minus 1)120573119875119898119886119909 ]minus 2 exp[minus2 (4119877119898119894119899 minus 1)120573119875119898119886119909 ]

(A2)

ForNOMAusers outage probability can be found in a similarway by following expressions

1198751198731198741198721198601 = 1 + exp [minus1205722120573 ]minus 2120573 int

infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where

1205721 = 2119877119898119894119899 minus 1119909119875119898119886119909 + 2119877119898119894119899 (1 minus radic1 + 119909119875119898119886119909)1205722 = 4119877119898119894119899 minus 22119875119898119886119909 + radic 4119877119898119894119899 minus 121198751198981198861199092 +

(4119877119898119894119899 minus 2)2411987511989811988611990921198751198731198741198721198602 = 1 + exp[minus2 (4119877119898119894119899 minus 1)120573119875119898119886119909 ] minus 2sdot exp[minus2 (2119877119898119894119899 minus 1)120573119875119898119886119909 ] + (2119877119898119894119899 minus 1) exp[[

(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909

[erf 119888⟨(2119877119898119894119899 + 1)2radic120573119875119898119886119909 ⟩minus erf 119888⟨(6119877119898119894119899 + 1)2radic120573119875119898119886119909 ⟩]

(A4)


Proof See [42] Appendix C

Average outage probabilities have been calculated byconsidering 120573 = 1 for all the users in a particular region ofthe cell

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

The authors declare that they have no conflicts of interest

Acknowledgments

The National Research Foundation of Korea (NRF)grant funded by the Korea government (MSIP)(2016R1A2B4008457) and the Strengthening R amp DCapability Program of Sejong University supported thiswork

References

[1] M S Ali H Tabassum and E Hossain ldquoDynamic User Clus-tering and Power Allocation for Uplink and Downlink Non-Orthogonal Multiple Access (NOMA) Systemsrdquo IEEE Accessvol 4 pp 6325ndash6343 2016

[2] 3GPPTSG-RANR1-050738 ldquoFFR Interferencemitigation con-siderations and results on frequency reuserdquo September 2005

[3] J Choi ldquoNon-orthogonal multiple access in downlink coordi-nated two-point systemsrdquo IEEECommunications Letters vol 18no 2 pp 313ndash316 2014

[4] S M Alamouti ldquoA simple transmit diversity technique forwireless communicationsrdquo IEEE Journal on Selected Areas inCommunications vol 16 no 8 pp 1451ndash1458 1998

[5] S Han Chih-Lin I Z Xu and Q Sun ldquoEnergy Efficiencyand Spectrum Efficiency Co-Design From NOMA to NetworkNOMArdquo IEEE MMTC E-Letter vol 9 no 5 pp 21ndash24 2014

[6] D Lee H Seo B Clerckx et al ldquoCoordinated multipoint trans-mission and reception in LTE-advanced deployment scenariosand operational challengesrdquo IEEE Communications Magazinevol 50 no 2 pp 148ndash155 2012

[7] L Ping L Liu K Wu and W K Leung ldquoInterleave-divisionmultiple-accessrdquo IEEE Transactions on Wireless Communica-tions vol 5 no 4 pp 938ndash947 2006

[8] R Hoshyar F P Wathan and R Tafazolli ldquoNovel low-densitysignature for synchronous CDMA systems over AWGN chan-nelrdquo IEEE Transactions on Signal Processing vol 56 no 4 pp1616ndash1626 2008

[9] X Dai S Chen S Sun et al ldquoSuccessive interference can-celation amenable multiple access (SAMA) for future wirelesscommunicationsrdquo in Proceedings of the 2014 IEEE InternationalConference on Communication Systems IEEE ICCS 2014 pp222ndash226 China November 2014

[10] S Chen B Ren Q Gao S Kang S Sun and K Niu ldquoPatterndivision multiple access-a novel nonorthogonal multiple accessfor fifth-generation radio networksrdquo IEEE Transactions onVehicular Technology vol 66 no 4 pp 3185ndash3196 2017

[11] B Ren X Yue W Tang et al ldquoAdvanced IDD receiver forPDMA uplink systemrdquo in Proceedings of the 2016 IEEECICInternational Conference on Communications in China ICCC2016 China July 2016

[12] J Zeng B Li X Su L Rong and R Xing ldquoPattern divisionmultiple access (PDMA) for cellular future radio accessrdquo inProceedings of the 2015 International Conference on WirelessCommunications amp Signal Processing (WCSP) pp 1ndash5 NanjingChina October 2015

[13] B Ren Y Wang X Dai K Niu and W Tang ldquoPattern matrixdesign of PDMA for 5G UL applicationsrdquo China Communica-tions vol 13 pp 159ndash173 2016

[14] P Li Y Jiang S Kang et al ldquoJoint Transmitter and ReceiverDesign for Spatial Pattern DivisionMultiple Access with Large-scale Antennardquo in Proceedings of the 65 China Communicationsampamp SupplementNo 2 2016 submitted for publication PIMRC-2016 oint Transmitter and Receiver Design for Spatial PatternDivision Multiple Access with Large-scale Antenna Ed April2016

[15] H Nikopour and H Baligh ldquoSparse code multiple accessrdquo inProceedings of the IEEE 24th Annual International SymposiumonPersonal Indoor andMobile Radio Communications (PIMRCrsquo13) pp 332ndash336 IEEE London UK September 2013

[16] M Taherzadeh H Nikopour A Bayesteh and H BalighldquoSCMA codebook designrdquo in Proceedings of the 80th IEEEVehicular Technology Conference VTC 2014-Fall CanadaSeptember 2014

[17] 3GPP ldquoHuawei HiSilicon Sparse Code Multiple Access(SCMA) for 5G Radio Transmissionrdquo R1- 162155 April 2016

[18] M Al-Imari P Xiao M A Imran et al ldquoUplink Non-Orthogonal Multiple Access for 5G Wireless Networksrdquo inProceedings of the ISWCS 2014 2014

[19] 3GPP ldquoZTE Discussion on multiple access for new radiointerfacerdquo R1-162226 April 2016

[20] Z Yuan G Yu W Li Y Yuan X Wang and J Xu ldquoMulti-UserShared Access for Internet ofThingsrdquo in Proceedings of the 2016IEEE 83rd Vehicular Technology Conference (VTC Spring) pp1ndash5 Nanjing China May 2016

[21] H Jin K Peng and J Song ldquoBit division multiplexing forbroadcastingrdquo IEEE Transactions on Broadcasting vol 59 no3 pp 539ndash547 2013

[22] White paper ldquov20D-Alternative Multiple access v1rdquo November2015 httpwwwfuture-forumorgzhuanti151105cnindexasp

[23] 3GPP Qualcomm Inc Candidate NR multiple access schemesR1-162202 Apr 2016

[24] 3GPP Qualcomm Inc RSMA R1- 164688 May 2016[25] 3GPP ldquoSamsung Non-Orthogonal Multiple Access Candidate

for NRrdquo R1-163992 May 2016[26] A Li Y Lan X Chen and H Jiang ldquoNon-orthogonal multiple

access (NOMA) for future downlink radio access of 5Grdquo ChinaCommunications vol 12 pp 28ndash37 2015

[27] K Higuchi and Y Kishiyama ldquoNon-orthogonal access withrandom beamforming and intra-beam SIC for cellular MIMOdownlinkrdquo in Proceedings of the 2013 IEEE 78th VehicularTechnology Conference VTC Fall 2013 USA September 2013

[28] K Higuchi and A Benjebbour ldquoNon-Orthogonal MultipleAccess (NOMA) with successive interference cancellation forfuture radio accessrdquo IEICE Transactions on Communicationsvol E98B no 3 pp 403ndash414 2015


[29] 3GPP ldquoStudy on Downlink Multiuser Superposition Transmis-sion (MUST) for LTE (Release 13)rdquo TR36859 December2015

[30] 3GPP ldquoMediaTek Inc CMCC etc New work item proposalDownlink Multiuser Superposition Transmission for LTErdquo RP-160680 March 2016

[31] 3GPP ldquoNTT-DOCOMO Initial views and evaluation result onnon-orthogonal multiple access for NR uplinkrdquo R1-163111 April2016

[32] 3GPP ldquoNTT-DOCOMO Initial views and evaluation results onnon-orthogonal multiple access for NRrdquo R1-165175 May 2016

[33] 3GPP TSG-RANR1-050507 ldquoSFR Soft frequency reuse schemefor UTRAN LTErdquo 3GPP May 2005

[34] Y Umeda and K Higuchi ldquoEfficient adaptive frequency parti-tioning in OFDMA downlink with fractional frequency reuserdquoin Proceedings of the 2011 International Symposium on IntelligentSignal Processing and Communications Systems (ISPACS 2011)pp 1ndash5 Chiang Mai Thailand December 2011

[35] W Shin M Vaezi B Lee D J Love J Lee and H V PoorldquoCoordinated beamforming for multi-cell MIMO-NOMArdquoIEEE Communications Letters vol 21 no 1 pp 84ndash87 2017

[36] Z Ding F Adachi and H V Poor ldquoThe Application of MIMOto Non-Orthogonal Multiple Accessrdquo IEEE Transactions onWireless Communications vol 15 no 1 pp 537ndash552 2016

[37] N Saquib E Hossain and D I Kim ldquoFractional frequencyreuse for interference management in LTE-advanced hetnetsrdquoIEEEWireless CommunicationsMagazine vol 20 no 2 pp 113ndash122 2013

[38] T D Novlan R K Ganti A Ghosh and J G Andrews ldquoAnalyt-ical evaluation of fractional frequency reuse for OFDMA cellu-lar networksrdquo IEEE Transactions on Wireless Communicationsvol 10 no 12 pp 4294ndash4305 2011

[39] T Novlan J G Andrews I Sohn R K Ganti and A GhoshldquoComparison of fractional frequency reuse approaches in theOFDMA cellular downlinkrdquo in Proceedings of the 53rd IEEEGlobal Communications Conference (GLOBECOM rsquo10) pp 1ndash5Miami Fla USA December 2010

[40] Zubin Bharucha and Harald Haas ldquoThe Distribution of PathLosses for Uniformly Distributed Nodes in a Circlerdquo ResearchLetters in Communications vol 2008 pp 1ndash4 2008

[41] N Otao Y Kishiyama and K Higuchi ldquoPerformance of non-orthogonal access with SIC in cellular downlink using pro-portional fair-based resource allocationrdquo in Proceedings of the2012 9th International Symposium on Wireless CommunicationSystems ISWCS 2012 pp 476ndash480 August 2012

[42] J A Oviedo and H R Sadjadpour ldquoA Fair Power AllocationApproach to NOMA in Multiuser SISO Systemsrdquo IEEE Trans-actions on Vehicular Technology vol 66 no 9 pp 7974ndash79852017

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


is accomplished by using the superposition principle where acomposite signal is constructed from individual user signalsand mapped onto a common frequency resource as opposedto one-to-one user resource mapping in OMA NOMAprovidesmassive connectivity as well as throughput enhance-ment obtained by sharing a single resource by multiple users[1] Spectral efficiency is an embedded advantage of NOMAover OMA schemes provided by the superposition of usersover a common resource

In NOMA user clustering is performed to pair userswith diverse channel responses together to maximize userthroughput gain and user capacity [1] Another addedadvantage of clustering process is the simplification of SICat the receiving end All paired users are then mappedonto orthogonal frequency resources to avoid interferencebetween user clusters which is also known as interclusterinterference Interference effectively reduces the benefitsoffered by NOMA over OMA Interference experienced canbe due to a number of factors Firstly incorrect channel stateinformation (CSI) causes errors during SIC decoding at theNOMA receiver for one ormore users depending on the userwho reported the wrong CSI Then clustering method canlead to the wrong pairing of users causing errors in the SICprocess Also user density in a cluster affects the complexityof SIC and the throughput limit of each user and finally thenumber of clusters defines the amount of bandwidth availablefor each cluster and eventually individual user throughputgains

For a multicell NOMA network another major source ofinterference that occurs between different clusters of adjacentcells is known as intercell interference (ICI) which willbe considered in the proposed work For the multicellularwireless network the available spectrum is allocated amongstdifferent clusters in a cell and spectral efficiency as well asuser capacity enhancement is achieved by employing thefrequency reuse scheme [2] NOMA can offer significantimprovements but only if ICI and cluster interferences aremanaged efficiently In this paper ICI isolation is achievedby frequency reuse diversity along with the efficient designof user clustering as well as efficient resource utilizationamongst NOMA users

11 Related Work In recent years many NOMA schemeshave been devised such as interleave-division multiple access(IDMA) [7] low-density spreading CDMA (LDS-CDMA)[8] pattern-division multiple access (PDMA) [9ndash14] sparse-code multiple access (SCMA) [15ndash17] multiuser sharingaccess (MUSA) [18ndash20] bit-division multiplexing (BDM)[21] low-density spreading OFDM (LDS-OFDM) [22]resource-spread multiple access (RSMA) [23 24] interleave-grid multiple access (IGMA) [25] multiuser bit-interleavedcoded modulation with iterative decoding (MU-BICM-ID)[26] and power domain nonorthogonal multiple access (PD-NOMA) [5 27ndash32]

PD-NOMA is considered in which users are distin-guished on the basis of allocated transmit power on a sharedresource [28] Power allocation in NOMA is carried outconsidering the detection process via SIC at the receiverhence a user with a poor channel gain ie cell edge (CE)

user is allocated a higher power and a lower power is assignedto the user with a better channel gain ie cell center (CC)user [28] CE users will be the ones most affected by theICI as they will have to cancel out any possible spilloversignals intended for CE users of adjacent cells Higher powersare allocated to CE users to ensure throughput as wellas sufficient received power at the base station (BS) fordetection [28] Both CC and CE users are then mappedonto respective frequency resource blocks (RBs)TheNOMAreceiver employs simultaneous multiuser detection (MUD)to identify and estimate each userrsquos data eliminating anyeffects of channel distortion or interuser interference withSIC [27] SIC is a technique used to successively extracta single user signal by the cancellation of unwanted usersignals in the received signal SIC performs optimally wheneach user experiences a channel (ie resource block) that isconsidered distinct from other users [1] However a rapidlychanging channel itself poses a problem of employing acomplex channel estimation algorithm to find the channelresponse for each user This feature makes NOMA a suitablecandidate for outdoor cellular networks as the path lossvaries considerably in outdoor scenarios as compared toindoors

FFR [2] and soft frequency reuse (SFR) [33] are twostudied solutions employing frequency reuse to improvespectrum efficiency and to reduce potential ICI in 4GnetworksThe available spectrum is divided into subchannelsthat are further divided into cell center and cell edge groupsBy allocating different power levels to users in each groupICI can be controlled effectively FFR provides a high networkthroughput as well as a high edge user SINR as compared toSFR which provides a balance between resource efficiencyand interference reduction The major difference betweenFFR and SFR is that in SFR cell center users can alsouse the spectrum allocated to cell edge users This causesconsiderably more interference to both center and edge userswhen compared with the FFR caseThis makes FFR a consid-erably better choice when interference reduction is of primeimportance In downlink OFDMA system performance isusually limited by ICI the edge users are the ones mostaffected by it FFR was explored as a possible solution in[2 33] to address the performance issues of edge users dueto interference Edge zones in cells are assigned a largerreuse factor in this scheme to reduce interference In staticFFR the reuse factor was decided at the time of frequencyplanning of a network this is inefficient anddoes not take intoconsideration the changing conditions that affect the edgezones of cells Therefore by using static FFR a satisfactorysystem performance for edge users cannot be achieved inreal environments due to the ever-changing channel andinterfering conditions This drawback of static FFR has alsobeen highlighted in [34] which studied a dynamic adap-tive frequency-division algorithm to improve cell averagethroughput especially the edge user throughput forOFDMAThe results achieved by this scheme are however restricted toscenarios inwhich a single user ismapped onto an orthogonalfrequency resource and no superposition coding is takingplace Detailed analysis of existing schemes for NOMAintercell interference management has been performed in















UE 1 UE 2 UE e

FreqPo

wer

P1

Pe

Freq

Pow

er

P2

Pe

Backhaul

NOMA-JT


(a)

UE 1 UE e

Freq

Pow

er

P1

Freq

Pow

er

P2

Pe

Backhaul

NOMA-DCSUE 2


(b)

UE 1 UE 2 UE e

Freq

Pow

er

P1

Pe

Freq

Pow

er

P2

NOMA-CS

Dynamic Switching



(c)

UE 1 UE 2 UE 4

Freq

Pow

er

P1

P3

Freq

Pow

er

P2

P4

NOMA-CBUE 3


(d)













Decoder

Decoder

Decoder

Decoder

Decoder

Decoder

Frequency

Pow

er

Ｂ1 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ1

Ｂ1 (Ｒ1 + Ｒ2) + Ｑ1

Ｂ1 Ｒ1+ Ｑ1

Ｒ3

Ｒ3

Ｒ3

Ｂ1

Ｂ1

Ｒ2

Ｒ2

Ｒ1

Ｂ3 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ3Ｓ3

Ｂ2 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ2Ｓ2

Ｓ2

Ｂ2 (Ｒ1 + Ｒ2) + Ｑ2

-

-

-

Ｂ1

Ｂ2 Ｂ2

Ｂ3


０1

０2

０3



NOMA-CS



Sharedinformation




lt 4T 4(T-1) 3T 3T or 4T




3 System Model



Power

BS 1 BS 2

Cell-center user


signals


noise

NoiseInter-


cancelled

Inter-channel



Cell-edge users

Cell-center user


Power


interference

Freq



119909 = 11990111199091 + 11990121199092 + 11990131199093 (1)


119910119894 = ℎ119894119909 + 119908119894 (2)





119909119887 = 119898119887sum119895=1

radic119901119887 (119906119887 (119895))119904119887 (119906119887 (119895)) (3)




119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)


119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)



119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)


119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)








Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia
















UE 1 UE 2 UE e

FreqPo

wer

P1

Pe

Freq

Pow

er

P2

Pe

Backhaul

NOMA-JT


(a)

UE 1 UE e

Freq

Pow

er

P1

Freq

Pow

er

P2

Pe

Backhaul

NOMA-DCSUE 2


(b)

UE 1 UE 2 UE e

Freq

Pow

er

P1

Pe

Freq

Pow

er

P2

NOMA-CS

Dynamic Switching



(c)

UE 1 UE 2 UE 4

Freq

Pow

er

P1

P3

Freq

Pow

er

P2

P4

NOMA-CBUE 3


(d)













Decoder

Decoder

Decoder

Decoder

Decoder

Decoder

Frequency

Pow

er

Ｂ1 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ1

Ｂ1 (Ｒ1 + Ｒ2) + Ｑ1

Ｂ1 Ｒ1+ Ｑ1

Ｒ3

Ｒ3

Ｒ3

Ｂ1

Ｂ1

Ｒ2

Ｒ2

Ｒ1

Ｂ3 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ3Ｓ3

Ｂ2 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ2Ｓ2

Ｓ2

Ｂ2 (Ｒ1 + Ｒ2) + Ｑ2

-

-

-

Ｂ1

Ｂ2 Ｂ2

Ｂ3


０1

０2

０3



NOMA-CS



Sharedinformation




lt 4T 4(T-1) 3T 3T or 4T




3 System Model



Power

BS 1 BS 2

Cell-center user


signals


noise

NoiseInter-


cancelled

Inter-channel



Cell-edge users

Cell-center user


Power


interference

Freq



119909 = 11990111199091 + 11990121199092 + 11990131199093 (1)


119910119894 = ℎ119894119909 + 119908119894 (2)





119909119887 = 119898119887sum119895=1

radic119901119887 (119906119887 (119895))119904119887 (119906119887 (119895)) (3)




119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)


119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)



119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)


119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)








Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia











Decoder

Decoder

Decoder

Decoder

Decoder

Decoder

Frequency

Pow

er

Ｂ1 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ1

Ｂ1 (Ｒ1 + Ｒ2) + Ｑ1

Ｂ1 Ｒ1+ Ｑ1

Ｒ3

Ｒ3

Ｒ3

Ｂ1

Ｂ1

Ｒ2

Ｒ2

Ｒ1

Ｂ3 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ3Ｓ3

Ｂ2 (Ｒ1 + Ｒ2 + Ｒ3) + Ｑ2Ｓ2

Ｓ2

Ｂ2 (Ｒ1 + Ｒ2) + Ｑ2

-

-

-

Ｂ1

Ｂ2 Ｂ2

Ｂ3


０1

０2

０3



NOMA-CS



Sharedinformation




lt 4T 4(T-1) 3T 3T or 4T




3 System Model



Power

BS 1 BS 2

Cell-center user


signals


noise

NoiseInter-


cancelled

Inter-channel



Cell-edge users

Cell-center user


Power


interference

Freq



119909 = 11990111199091 + 11990121199092 + 11990131199093 (1)


119910119894 = ℎ119894119909 + 119908119894 (2)





119909119887 = 119898119887sum119895=1

radic119901119887 (119906119887 (119895))119904119887 (119906119887 (119895)) (3)




119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)


119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)



119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)


119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)








Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Power

BS 1 BS 2

Cell-center user


signals


noise

NoiseInter-


cancelled

Inter-channel



Cell-edge users

Cell-center user


Power


interference

Freq



119909 = 11990111199091 + 11990121199092 + 11990131199093 (1)


119910119894 = ℎ119894119909 + 119908119894 (2)





119909119887 = 119898119887sum119895=1

radic119901119887 (119906119887 (119895))119904119887 (119906119887 (119895)) (3)




119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)


119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)



119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)


119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)








Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




119910119887 (119906119887 (119895)) = ℎ119887 (119906119887 (119895)) 119909119887 + 119908119887 (119906119887 (119895)) (4)


119910119887 (119906119887 (119895)) = ℎ119867119887 (119906119887 (119895)) 119910 (119906119887 (119895))1003817100381710038171003817ℎ1198871003817100381710038171003817= radic119892119887 (119906119887 (119895))119909119887 + 119911119887 (119906119887 (119895))

(5)



119878119868119873119877119895 = (1119898119887)sum119898119887119894=1 1199102119887 (119906119887 (119894))sum119906119887(119894)120598119880119887 1199102119887 (119906119887 (119894)) + 1199082119887 (119906119887 (119895)) (6)


119877119894 = 120583119871 log2(1 + 119875119894ℎ119894sum119894minus1119895=1 119875119894ℎ119894 + 120583) (7)








Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Cell A

Cell A

Cell B

Cell B

Cell C

Cell C

Be

Be

Be

Bc






(119878119873119868119877119894)119896 minus 119869sum119895=1

(119878119873119868119877119895) gt 119878119879119867 (8a)

(119878119873119868119877119894)119896 gt 119878119879119867 (8b)




119875119895ℎ119894minus1 ge 119875119898119886119909 (9)














119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia













119901119887 (119896) = 119875119899119887sum119895isin119880119887(119887) (119892119887 (119895) 119899119887 (119895))minus120573 (119892119887 (119895)119899119887 (119895))

minus120573

(10)





119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




119898119887sum119895=1

119901119887 (119906119887 (119895)) = 119875119899119887 le 119875119898119886119909 (11)



119875119873119865119865119877 = sum119894120598119870sum119895120598119873P (119877119894119905 lt 119877119898119894119899)sum119894120598119870119880119864 (12)






















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia















2

3

4

5

6

7

8

Sum

Cap

acity

15 20 25 30 35 40 6050 6510 45 55SNR




0

01

02

03

04

05

06

07

08

09

1

CDF



CRR = 01CRR = 03

CRR = 05CRR = 07


0

01

02

03

04

05

06

07

08

09

1

CDF






CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



CPR 02CPR 04

CPR 06CPR 08

7

8

9

10

11

12

13

14

15

16

17Th

roug

hput

(Mbp

s)

045 05 055 06 065 07 075 08 085 0904Center Ratio


CPR 02CPR 04

CPR 06CPR 08

045 05 055 06 065 07 075 08 085 0904Center Ratio

0

1

2

3

4

5

6

7

8

Thro

ughp

ut (M

bps)



ERR 5ERR 10

ERR 15ERR 20

045 05 055 06 065 07 075 08 085 0904Center Ratio

135

14

145

15

155

16

165

17

175

Thro

ughp

ut (M

bps)






ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



ERR 5ERR 10

ERR 15ERR 20

0

02

04

06

08

1

12

14

16

18

Thro

ughp

ut (M

bps)

045 05 055 06 065 07 075 08 085 0904Center Ratio



Center Edge

0

02

04

06

08

1

CDF







0

1

2

3

4

5

6

7

8

9

10

Spec

tral

Effi

cien

cy (b

psH

z)






NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



NOMA-JTNOMA-CBOMA


10minus3

10minus2

10minus1

100

Out

age P

roba

bilit

y

3510 15 20 25 30 400 5Transmit SNR (dB)



6 Conclusion



Appendix






(A2)



infin

1205722

exp[minus119909 (1205721 + 1)120573 ] 119889119909 (A3)

where



(2119877119898119894119899 minus 3)24120573119875119898119886119909 ]] radic120587120573119875119898119886119909


(A4)




Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





Data Availability




Acknowledgments


References














































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


Dynamic Fractional Frequency Reuse Diversity Design for ...

Documents

Transcript of Dynamic Fractional Frequency Reuse Diversity Design for ...