Research Article Interference Mitigation Framework for ... · Research Article Interference Mitigation Framework for ... of interference are generated, ... includes techniques like

Hindawi Publishing CorporationInternational Journal of Antennas and PropagationVolume 2013 Article ID 645929 15 pageshttpdxdoiorg1011552013645929

Research ArticleInterference Mitigation Framework forCellular Mobile Radio Networks

Wolfgang Mennerich1 Michael Grieger2 Wolfgang Zirwas3 and Gerhard Fettweis2

1 Focubeam GmbH Germany2Vodafone Chair Mobile Communications Systems TU Dresden Germany3Nokia Siemens Networks Germany

Correspondence should be addressed to Wolfgang Mennerich wolfgangmennerichalumnitu-berlinde

Received 7 May 2012 Accepted 10 August 2012

Academic Editor Sumei Sun

Copyright copy 2013 Wolfgang Mennerich 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

For todayrsquos cellular mobile communication networks the needed capacity is hard to realize without much more of (expensive)bandwidth Thus new standards like LTE were developed LTE advanced is in discussion as the successor of LTE and cooperativemultipoint transmission (CoMP) is one of the hot topics to increase the systemrsquos capacity System simulations often show only weakgains of the signal-to-interference ratio due to high interference from noncooperating cells in the downlink This paper presentsan interference mitigation framework to overcome the hardest issue that is the low penetration rate of mobile stations that can beserved from a cluster composed of their strongest cells in the networkThe results obtained from simulation tools are discussed withvalues resulting from testbed on the TUDresdenThey show that the theoretical ideas can be transferred into gains on real systems

1 Introduction

From the beginning of mobile radio communication there isa constant effort to reduce intercell interference to an accept-able level Since the success of the worldwide deployedGlobalSystem Mobile (GSM) [1] the enormous increased numberof users together with over proportionally increasing trafficdemands have put this topic into the main focus of research

Wireless transmitters generally spread power over largeareas and at very low frequencies even worldwide communi-cation is possible [2] In current cellular deployments carrierfrequencies ranging from a few hundredMHz to several GHzare being used At these frequency bands the radiation of asingle cell will typically cover smaller areas but cell diametersof 100 km and more are still possible in rural areas In urbanareas intersite distance (ISD) of few 100m to few km are quitecommon leading to a so-called interference limited scenariowhere the achievable performance is mainly bounded by theintercell interference (ICI) and not the receiver noise [3]

Interference power decreases with distance which in thebeginning of cellular communications has been the mainmeans for overcoming intercell interference leading to the

well-known frequency reuse concept Important enhance-ments were the introduction of frequency reuse 1 systemsby code division multiple access (CDMA) in the so-called3rd generation of mobile systems On the downside theperformance of this scheme suffers from the near far effectand a variable soft capacity of the system [4]

In the standardization of the current 3GPP standardnamed LTE CDMAwas replaced in the downlink by orthog-onal frequency division multiple access (OFDMA) beingagain a frequency reuse 1 system but providing additionallymultiuser scheduling gains by fast adaptation to frequencyselective channel conditions [5] This concept allows also fastadaptation to interference conditions and thereby approachesthe performance of the so-called inter-cell interference coor-dination (ICIC) schemes ICIC harmonizes resources usagebetween cell edge and cell centre users over neighbouringcells [6] depending on their mutual interference Note ICIChas generated also a lot of interest under the name of frac-tional frequency partitioning (FFP) with a different powerallocation for cell center and cell edge users on differentfrequency subbands

2 International Journal of Antennas and Propagation

In long-term evolution (LTE) Release 10 and now 11there is a lot of discussions about the so-called enhancedICIC (eICIC) schemes in the context of heterogeneous net-works comprising pico- andmacro- or femto- andmacro-sta-tions By transmitting the so-called almost blank subframes(ABS)mdashalmost blank because the common reference signals(CRS) have to be transmitted continuously for backwardcompatibilitymdashin time domain subframes with different levelof interference are generated providing some options forenhanced scheduling

In 2000 Baier et al [7] proposed the idea of cooperativetransmission systemsmdashalso known as networkmultiple inputmultiple output (MIMO) or for example in 3rd genera-tion partnership project (3GPP) as cooperative multi pointtransmission (CoMP) This is a much more fundamentalsolution partly being inspired from the research in the areaof MIMO systems Many groups predicted large theoreticalperformance gains with spectral efficiencies being severalfactors higher than that of conventionalmobile radio systemsFor example for 2 and 4 antennas per cell (BS) and mobilestation (MS) 10th of bitssHzcell have been promised atsignal to noise and interference ratios (SINRs) of less than 10to 20 dB [8ndash10]

Common to all these approaches is that a networkwide precoder over all cells including all MSs is assumedwhich eliminates any interference after proper precoding Inpractical mobile radio systems network wide precoding isunrealistic as networks might span a whole city or even awhole country Neither an according backhaul network canbe implemented nor channel estimation as well as the report-ing of all channel components is feasible For that reason itis necessary to partition the network into so-called coop-eration areas (CAs) connecting a limited number of cellsUnfortunately this step turns the interference free systemback into an interference limited one because of the remain-ing intercooperation area interference For quite sometimea lot of research concentrated on the area of clustering [11]applying different types of optimization algorithms to thisnonconvex optimization problem

Despite the research effort there is still a significantgap between the theoretical predicted gains and the resultsobserved from system level simulations so far [12] Onemainactivity in this area is the so-called 3GPP LTE CoMP studyitem which started in LTE Release 10 has been re-invokedfor LTE Release 11 and finally has been converted into a 3GPPLTE working item in September 2011 [13] While CoMPincludesmany different kinds of cooperation like coordinatedscheduling (CS) coordinated beamforming (CB) dynamic orfast cell selection (DCS) and so forth in our contribution thefocus is mainly on joint precoding (JP) or more specificallyon joint transmission (JT) CoMP JT CoMP is generallyaccepted to be the scheme with the highest potential gainsas signals being interferers in noncooperative scenarios areturned into useful signal power for JT CoMP This is themain benefit over pure interference avoidance strategies suchas CSCB Reference [14] contains the best so far achievedresults for the main 3GPP CoMP schemes In case of JTCoMP the best performance for a 4 times 2MIMO system doingjoint transmission over 9 cells was 4028 bitsHzcell being

far away from the projected 10 bitsHzcell from theoreticalanalysis

So one main issue to be tackled in this contributionis to reveal and analyze a fundamental limit of practicaland realizable CoMP schemes mainly due to the requiredclustering andor due to different assumptions for the radiochannels Note for theoretical analysis Rayleigh channelsare often assumed which are known to have an optimaleigenvalue distribution while macrocellular radio channelslike the spatial channel model extended (SCME) as beingused in 3GPP [15] may suffer from correlation of collocatedantenna elements and certain line of sight (LOS)nonline ofsight (NLOS) probabilities

The important question is whether the optimizationproblem is limited by fundamental reasons for exampleesteeming from clustering or just due to the dimension of theoptimization problemThefirst casewould be in combinationwith its significant complexity almost a killing point againstJT CoMP while in the second case it would be just a matterof doing things right

We have no final answer to this question yet but inthis contribution we propose an overall framework for inter-ference mitigation of which JT CoMP is just one of the build-ing blocks Based on an in depth analysis of interference con-ditions scheduling effects and so forth different conceptsare combined to get much closer to the theoretical results ofnetwork wide cooperation with clustered cooperation areasin macrocellular radio environments

Reference [16] mentions the importance of interferenceand provides an optimal power control provided for twoopportunistic cooperative base stations However the authorsmention that the algorithm does not bring performance ben-efits if there are only cell edge users Besides these ideas [17]suggests a concept to improve the bandwidth efficiency withtwo-way relay nodes where these nodes have multiple anten-nas Upper bounds regarding symbol error probability arederived as well as the order of diversity of the two proposedschemes

The proposed framework is more than CoMP andincludes techniques like interference rejection combining(IRC) multiuser MIMO (MU-MIMO) interference floorshaping wideband beamforming antenna tilting poweradaptation clustering user grouping effective precodingincluding119879119909 diversity per cell and per cooperation area sche-duling optimum channel state information (CSI) estimationand according to low rate but accurate feedback links

Implementation of specific issues like the just mentionedCSI estimation and feedback as well as intersite synchroniza-tion backhaul overhead and so forth are investigated inmoredetail in the EU founded project Artist4G [18] while thiscontribution has its main focus on maximizing system levelgains under ideal conditions In the end both parts have to becombined appropriately to form an overall system concept

Some items like partial CoMP cover shifts or interferencefloor shaping by cooperationwide vertical beam steering havebeen partly described in [19 20] while the CoMP schedulerresults and verification of main parts measurements in thereal world CoMP testbed from TUD [21] are presented herefor the first time

International Journal of Antennas and Propagation 3

The main outcome of the overall proposed concept is afactor of 24 higher spectral efficiency than achieved so far byMUMIMO and about 18 times higher than for the up to datebest 3GPP JT CoMP scheme as being given in [13] This wethink is really a great progress

2 Contribution of This Paper

In the introduction an overview about the issues relevantfor interference mitigation based on JP CoMP was givenThis chapter presents an interference mitigation frameworkdesigned to overcome the low penetration rate ofMS that canbe served by their individual set of network wide strongestBSs To reach that goal several concepts are combined totackle each issue that hinders a higher penetration rate ofMSswith high SIRs These concepts can be roughly categorizedinto the following points

(i) Oversized Clusters Oversized clusters provide ahigher probability that MSs find all of their strongestBSs in a single cluster

(ii) Partial User Centric Assignment It is sufficient if MShas only a subset of its serving BSs as its network widestrongest ones This increases the probability to findmultiple MSs to be served jointly from the same BScluster

(iii) Overlapping Clustering The clusters are overlappingsuch that around the border of two clusters a thirdone is added to serve the cluster-edge-MSs as centerMSs

(iv) Interference Floor Shaping Residual intercooperationarea interference is reduced by a suitable combinationof wideband (WB) beamforming and per WB beamindividual vertical antenna tiltingThis is done utiliz-ing the multiple antenna elements (AEs) mounted atBSs

3 Considered Network and Signal Model

We consider a downlink scenario within a cellular networkwith transmitting BSs and receiving MSs The cells areorganized in sites with three cells per site each covering asector of 120∘ as illustrated in Figure 2 consisting of threeadjacent sites Each BS covers a single cell The mathematicalnotation is listed in Table 2

First a simple single input single outpout (SISO) radiochannel between MS 119898 and BS 119887 each with only one AEis introduced as complex scalar value ℎ119898119887 If MSs and BSshavemultiple AEs denoted by119873119877119883 and119873119879119883 respectively thechannel ℎ119898119887 becomes a matrix H119898119887 isin C119873119877119883times119873119879119883 All of thesevalues represent the channel for a certain set of 12 subcarriersin the base band accordingly physical resource block (PRB)in case of LTE Radio channels from multiple BSs denotedby a set for example B = 1198871 119887119861 are concatenatedhorizontally Similar the radio channels to multiple MSs in

a set (eg M = 1198981 119898119872) are concatenated verticallyThis and the concatenation to a matrixHMB is shown in

B = 1198871 119887119861M = 1198981 119898119872997904rArr[[[

H1198981B

H119898119872B

]]] = [HM1198871sdot sdot sdotHM119887119861

] (1)

31 Joint PrecodingTransmission In the followingM119887 is theset of MSs that are jointly served by BS 119887 We assume furtherthat MSs have only one receive antenna that is 119873119877119883 = 1Thus H119898119887 = h119898119887 isin C1119873119879119883 = [h119898119887[1] sdot sdot sdot h119898119887[119873 119879119883]] Toseparate all jointly served MSs in the setM119887 BS 119887 uses a pre-coding vector v119898 isin C119873119879119883times1 = [v119898[1] sdot sdot sdot v119898[119873 119879119883]]119879 for eachMS 119898 isin M119887 The precoding vectors are columns from thejoint precoding matrixV119887 used at BS 119887 It is calculated basedon the channel matrix HM119887119887

and could be for example themoore penrose pseudo-inverse ofHM119887119887

denoted byHdaggerM119887119887

Intracell interference received at MS 119898 is caused by all

other data streams transmitted by BS 119887 for MSs 119896 with 119896 isinM119887 119898 Finally an additive white and Gaussian distributednoise signal 119899 with power 1205902119899 is assumed

If MS119898 is served just from BS 119887 that serves all MSs in setM119887 the signal 119910119898 received at MS119898 can be expressed by119910119898 = h119898119887v119898119889119898 + sum

119896isinM119887119898

h119898119887v119896119889119896 + or119868119898 + 119899 (2)

with 119889119898 being the transmitted data symbol for MS 119898 and

interferenceor119868119898 from other cells If the transmit power is

denoted by 119901119898 = |119889119898|2 the wanted signal power 119878(119898 119887)intra cell interference 119868(119898 119887) and the interference 119868(119898 119902)from another BSs 119902 at MS119898 are given with119878 (119898 119887) = 1199011198981003817100381710038171003817h119898119887v119898100381710038171003817100381722119868 (119898 119887) = sum

119896isinM119887119898

1199011198961003817100381710038171003817h119898119887v119896100381710038171003817100381722119868 (119898 119902) = sum119896isinM119902

11990111989610038171003817100381710038171003817h119898119902v1198961003817100381710038171003817100381722 119902 = 119887 (3)

Interference between these MSs mostly depends on theprecoding algorithm like for exampleMMSEor zero forcingThese algorithms strive to suppress interference power belowthe noise or in the case of zero forcing to remove it completelyIn reality an interference floor remains due to impairmentssuch as channel estimation errors and inter-carrier interfer-ence The error floor depends also on the precoding algo-rithmrsquos ldquorobustnessrdquo For instance the zero forcing algorithmis quite sensitive against precoding errors and more robustsolutions are available [22 23] especially for multiple userseparation [24 25]We focus onmethodical gains achieved bycooperation betweenmultiple distributed BSs and not on thejoint precoding itself Thus we assume that joint precodingalgorithms channel prediction and estimation and so forthare already applied in an optimal manner [26ndash28]


32 Precoding Normalization Loss Data streams of all MSsM119887 served by BS 119887 are transmitted in parallel The transmitpower used at that BS is limited by 119875BS

max yielding to asum power constraint for all signals transmitted in parallelFurthermore each single AE at the BS has also a maximumtransmit power given by 119875AE

max Hence we have a secondsum power constraint These power limitations can either beconsideredwithin the joint precoding algorithmas presentedfor example in [29 30] or by an additional normalizationbetween all jointly transmitted data streams Hence theprecoding vectors like v1015840119898 in this article are normalizedThe application of a separated precoding and normalizationis typically suboptimal and results in reduced SINR Thisprecoding normalization loss (PNL) is denoted by 120572(119898 119887)

v119898 v1015840119898 forall119898119875AEmax ⩾ sum

119898isinM119887

11990111989610038161003816100381610038161003816v1015840119898 [119896]100381610038161003816100381610038162 forall119896119875BSmax ⩾ sum

119898isinM119887

11990111989610038171003817100381710038171003817v10158401198981003817100381710038171003817100381722 forall119887997904rArr 120572 (119898 119887) = 10038171003817100381710038171003817h119898119887v101584011989810038171003817100381710038171003817221003817100381710038171003817h119898119887v119898100381710038171003817100381722

(4)

33 Signal-to-Interference Ratio This article is about inter-ference mitigation Thus we assume an interference limitedsystem with noise power 1205902119899 being negligible compared tothe interference power Hence the signal to interferenceratio (SIR) is used in the following The SIR seen at MS 119898which is served by BS 119887 can be clearly described using theaforementioned notation

SIR (119898 119887) = 120572 (119898 119887) 119878 (119898 119887)119868 (119898 119887) + sum119902 =119887 119868 (119898 119902) (5)

34 Cooperation between Base Stations A BS cluster isdescribed by a setB = 1198871 119887Ω ofΩ cooperating BSs withcardinality |B| = Ω The setMB = 1198981 119898119872 of119872MSsis served jointly by all BSs inB Equivalently asM119887 is the setof all MSs served by BS 119887 MB is defined as the union of allserved MSs from the cooperating BSs inB

MB = ⋃119887isinB

M119887 (6)

The SIR defined in (5) is now extended

SIR (119898B) asymp 120572 (119898B) 119878 (119898B)119868 (119898B) + sum119902notinB 119868 (119898 119902) (7)

Assuming that a certainMS119898 is assigned to BS 1198871 the BSindices 1198871 1198872 fulfil a special sorting of the BSs

(i) The BS to which MS 119898 would be connected withoutany cooperation is denoted by 1198871 Typically this is theBS with the strongest radio channel for MS119898

(ii) The set of all BSs at the same site Bsite with BS 1198871being one of them is denoted by Bsite

1 consisting ofBSs 1198871 1198872 1198873 Note that BS 1198871 is connected to MS 119898because it has the strongest radio channel to it butBSs 1198872 1198873 have in general an arbitrary connectivity toMS 119898 since they are only connected to MS 119898 nowbecause they are located at the same site as BS 119887

(iii) A CA is a special cluster consisting of three adjacentsites that is Bsite

1 Bsite2 Bsite

3 like illustrated inFigure 2 Thus the comprehensive set C contains 9BSs 1198871 1198879 All BSs 119887119899 with 119899 gt 9 and sites Bsite

119896

with 119896 gt 3 are not in that CA and cause interferenceat MS119898 Note that in generalBsite

2 andBsite3 that is1198874 1198875 1198876 and 1198877 1198878 1198879 have also arbitrary channel

conditions as already explained above for BS 1198872 1198873Using that numbering scheme the SIR for intra- and

intersite cooperation is given by

SIR (119898 Bsite1 Bsite

2 Bsite3 ) = SIR (119898C)asymp 120572 (119898C) 119878 (119898C)119868 (119898C) + sum119902notinC 119868 (119898 119902)

(8)

4 The Interference Mitigation Framework

In general cooperation between different BSs can be done indifferent ways In this article it is assumed that JP betweenBSs in a CA separates all jointly served MSs in the spatialdomain using theAEs of all cooperatingBSs as one single anddistributed antenna arrayThe precoding itself can be done ina so-called central unit (CU) like the one sketched in Figure 9with119873119879119883 = 2 AEs for each of the three BSs in the cluster

Nowwe introduce the interferencemitigation frameworkstep by step The specific parameter settings given beloware an example for the explanation of the main conceptsThe values like the size of the CAs and so forth have beenchosen based on practical considerations but might be fur-ther optimized or adapted in case of other scenarios

41 User Centric Assignment For high SIRs the clusteringshould be organized such that a cluster contains the networkwide 120596 strongest BSs for each of its served MSsThis individ-ual list of network wide strongest BSs is referred to as anMSrsquosldquowish listrdquo in the following The assignment is called ldquousercentricrdquo because it depends on the users that is the MSs andtheir wish lists In contrast the assignment is network centricif cooperation areas are predefined on network level withoutadaptation to any wish lists For user centric assignment theinterference from adjacent clusters is weak as the strongestBSs belong to the serving cluster and all other BSs have lowerpower than the serving BSs In realistic scenarios user centricassignment is hard to realize because the BSs according to anMSrsquos wish list often do not belong to a single predefined clus-ter and vary between different MSs Since in scenarios withintrasite cooperation only BSs on the same site can cooperateMSs will often have strong BSs at adjacent sites that causestrong inter site interference Above all this motivates the


introduction of the aforementioned CAs because then moreof the strongest BSs might be in a single cluster that is a CA

Regarding the SIR expressions as expressed in (8) the goalof user centric assignments is to minimize the right side ofthe denominator by maximizing the connectivity h119898B ofan MSrsquos serving clusterB1198871 119887Ω arg max

B|B|=Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (9)

Within the optimization in (9) the number 120596 of requestedstrongest BS in the serving cluster is equal to the total amountΩ of BSs in a cluster If the argument of the max-functionismdashby chancemdashidentical to the 119899rsquoth siteBsite

119899 intrasite coop-eration fulfils these requirements for user centric assignmentwith120596 = 3 but often the strongest120596 BSs are not collected at asingle site In that case for example two sites with 6 BSsmustcooperate to have at least the strongest 120596 BSs in the servingCAThis case is referred to as partial user centric assignmentand formulated in1198871 119887Ω ni arg max

B|B|=120596⩽Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (10)

However the strongest BSs are often distributed over a largerarea than just between two adjacent sites within a singleCA This illustrates the challenge of (partial) user centricassignment Recalling the indexing introduced in Section 34there is the following ldquouser centric conditionrdquo concerning theconnectivity 10038171003817100381710038171003817h11989811988711989910038171003817100381710038171003817 ⩾ 10038171003817100381710038171003817h11989811988711989610038171003817100381710038171003817 119899 ⩽ 120596 lt 119896 (11)

Equation (10) fulfils the user centric condition (11) With120596 = Ω also (9) does so In the following the latter case with120596 = Ω is referred to as ldquofull user centric assignmentrdquo and120596 lt Ω is called ldquopartial user centric assignmentrdquo Due to theallocation of the strongest 120596 BS the SIR depends not only onthe cardinality of B For example recall the indexing fromSection 34 1198871 1198873 belongs to the site cluster Bsite

1 while1198874 1198876 and 1198877 1198879 belong toBsite2 andBsite

3 respectivelyFurthermore it is assumed that the connectivity from BSsthat are not members of 1198871 1198879 is not relevant that is theinterference floor fromBSs that are located further away fromMS119898 Now we compare the SIRs resulting from two differentserving clusters for MS119898 under some further assumptions10038171003817100381710038171003817h11989811988711989910038171003817100381710038171003817 ⩾ 10038171003817100381710038171003817h11989811988711989610038171003817100381710038171003817 119899 ⩽ 3 lt 11989610038171003817100381710038171003817h11989811988711989910038171003817100381710038171003817 ≫ 10038171003817100381710038171003817h11989811988711989610038171003817100381710038171003817 119899 ⩽ 9 lt 11989610038171003817100381710038171003817h11989811988711989910038171003817100381710038171003817 asymp 10038171003817100381710038171003817h11989811988711989610038171003817100381710038171003817 119899 119896 isin 4 9997904rArr SIR (119898Bsite

1 ) lt SIR (119898 Bsite1 Bsite

2 )SIR (119898 Bsite

1 Bsite2 ) asymp SIR (119898 Bsite

1 Bsite3 )


119899 ) ≪ SIR (119898 Bsite1 Bsite

2 Bsite3 )119899 isin 2 3

(12)

The example above makes clear that large cooperation gainscan only be realized if all of the relevant interferers are in theserving cluster and contribute to the wanted signal for anMSthat is the signal gain alonewill not result in strong SIR gainsThus the most important issue is to combine the relevantinterferers that is the network wide strongest BSs into theserving cluster for each MS The SIR can be increased moreby interference mitigation of all relevant interferers than dueto a signal gain In the following three concepts are presentedthat exploit these issues

(i) Oversized Clusters Partial user centric assignment isexplicitly permitted The conditions for user centricassignment are reduced to 120596 lt Ω The MSs shouldfind now their 120596 strongest BSs within a single clusterof sizeΩThis is the notion behind the introduction ofCAs because in their central areaMSs are surroundedwith cooperating BSs

(ii) Partial Channel Reporting To realize such large clus-ters only channels of the 120596 strongest BSs are reportedfrom theMSs and used as transmitters for the desiredsignal The channel coefficients to all unreported BSsare assumed to be zero This concept is referred to asldquopartial channel reporting (PCR)rdquo in the following

(iii) Overlapping Clusters For MSs positioned on theborder between two CAs a third (overlapping) CAwith its center area at that border is added Sinceit operates on orthogonal resources for exampleanother frequency sub-band interference towards thefirst two CAs is completely avoided This overlappingscenario is illustrated in Figure 7

42 OversizedClustering Note that PCRwill introduce intra-CA interference at signal power level as it would also appearfrom adjacent and strong BSs that do not belong to theserving CA but for an MS it is unimportant whether inter-ference comes from an unreported BS inside of its servingcluster or any other BSs in the network The advantage liesin the relaxed condition that only 120596 out of Ω BSs must bethe strongest ones for the MS This increases the so-calledldquopenetration raterdquo that is the percentage of MSs that can beserved that way Note that the penetration rate also dependson the probability to find enough MSs with identical wishlists an issue which is addressed later

One reason for the very low probability that MSs can beserved in a full user centric assignment is the huge numberof possibilities for wish lists of each single MS Consideringa network of 119873BS BSs the number of possibilities to select agroup consisting ofΩBSs is given by the binomial coefficient(119873BSΩ ) = 119873BSΩ (119873BS minus Ω) (13)

Neglecting pathloss and assuming an equal distribution of theconnectivity between each BS and MS the probability thata certain group of Ω BSs is the optimum serving set of anarbitrary MS can be estimated just by the inverted binomialcoefficient (119873BS

Ω)minus1


Table 1 Simulation parameter

Parameter ValueNumber of cells (BSs) 57Number of sites 19Cells (BSs) per cellsite 3Sector width 120∘

Physical AEs per BS 4WB beams per BS 2Hrz WB beam direction plusmn30∘ from broadsideVrt WB beam direction 10∘15∘ centeredgePower adapt WB beams 0minus6 dB centeredgeNumber of subcarriers 32Bandwidth per subcarrier 180 kHz119879119909AEs 4119877119909AEs 1Joint precoding ZeroforcingChannelmodell SCMEISD 500mAntenna tilting 10∘ and 15∘

Penetration loss NoCSI Ideal

Table 2 Mathematical notation

Meaning NotationScalars functions 119886Vectors matrix elements aElement 119896 of vector a119899 a119899[119896]Matrices ASets A

Cardinality ofA |A||A| = 119873 A119873

Vector a transposed a119879

Vector a hermitian a119867

Pseudo-inverse of matrix A Adagger

Thus the probability that an arbitrary set of BSs isselected and belongs to any of 119873C clusters that is thepenetration rate is given by 119873C(119873BSΩ )minus1 The penetrationrate is increased using the relaxed assigning condition 120596 ltΩ according to the PCR concept Thus there are ( Ω120596 )possibilities that anMS has its strongest120596BSs within a clusterconsisting of Ω BSs Furthermore the fact that in generalmultiple MSs with an identical wish list are needed to form agroup of jointly servedMSs would drastically reduce the pen-etration rate but the aforementioned relaxed condition 120596 ltΩ allows grouping MSs with different wish lists since onlytheir 120596 strongest BSs must be in the serving cluster of sizeΩ

Figure 3 contains results from a system level simulation(details in Table 1) and shows the penetration rate of MSswith 3 6 and 9 of their strongest BSs in their serving CA Asignificant increased penetration rate can already be expectedif MSs should have their strongest 6 BSs in a single CA If 120596

is decreased further to 3 in combination with CAs of cardi-nalityΩ = 9 we have penetration rates of more than 90

43 Overlapping Clustering If each possible cluster wouldbe realizable we have 119873C rarr (119873BS

Ω) and the penetration

rate approaches one but with a nonoverlapping clusteringthe number of clusters that can be defined in parallel from119873BS BSs is limited to119873BSΩThus overlapping clusters allowa higher probability that the wish list of a certain MS isidentical to one of these clusters On a first glance the idea ofoverlapping clustering seems similar to fractional frequencypartitioning but the overlapping clustering is chosen suchthat each BS contributes to multiple CAs that overlap eachother The frequency subbands are assigned to these overlap-ping clusters such that each BS offers the complete systembandwidth distributed between all clusters it is contributingtoThus theMSsmust be divided into subgroups served onlyby a fraction of the systemrsquos bandwidth If these subbands arechosen big enough that eachMS can use as much subcarriersas needed for its purposes the MSs will not suffer from anybandwidth limitation Of course themultiplexing gainmightbe reduced by the reduced bandwidth here

The overlapping clustering consists of different clusteringlayers so called ldquocover shiftsrdquo built from a basic clusteringthat consists of nonoverlapping CAs as illustrated in Figure 2Each of these clustering layers is ldquoshiftedrdquo into differentdirections to be located directly on the borders between twoadjacent CAs from the basic clustering As aforementionedoverlapping clusters use different frequency subbands toavoid interference between different cover shifts The shapeof a CA according to Figure 2 allows to built 6 of such covershifts as illustrated in Figure 4 from left to rightThe triangleshave their edges on the three sites forming a CA The centerarea of each CA is highlighted within the triangles From leftto the right Figure 4 depicts the basic clustering with moreand more cover shifts being added to fill all gaps betweenthe CAs of the basic clustering With all six cover shifts theoverlapping is such that each border between two CAs in forexample cover shift ldquoArdquo is the centre area of another covershift ldquoBrdquo as illustrated in Figure 7

44 Partial Channel Reporting As aforementioned the highcooperation overhead might become an issue in CAs due tointersite cooperation For full channel knowledge within acluster each channel coefficient on each subcarrier for eachserved MS must be known at each of the cooperating BSsFurthermore signals to be transmitted jointly according toFigure 9 must first be distributed between cooperating sitesThe PCR concept reduces this overhead significantly butsince the full channel matrix between all transmit and receiveantennas is needed for JP the CU assumes all unreportedradio channel coefficients simply to be zero

A channel coefficient of zero means that no power canbe transmitted Thus the joint precoding is calculated withwrong assumptions and MSs will see unknown interferencefrom all unreported BSs

As an example we consider a channel matrix HMBB

between a cluster B = 1 5 and the jointly servedMSs in MB = 1 5 In addition we assume one AE


at each BS and MS and that the first MS has only reportedthe channels corresponding to BSs 1 2 and 5 Thus theresulting precoding matrix VB contains errors such thatthe underlined entries (corresponding to the AEs fromunreported BSs) contain interference for the first MS

HMBB = (ℎ11 ℎ12 ℎ13 ℎ14 ℎ15ℎ21 ℎ22 ℎ23 ℎ24 ℎ25ℎ31 ℎ32 ℎ33 ℎ34 ℎ35ℎ41 ℎ42 ℎ43 ℎ44 ℎ45ℎ51 ℎ52 ℎ53 ℎ54 ℎ55)997904rArr VB = (11990711 11990712 11990713 11990714 1199071511990721 11990722 11990723 11990724 11990725

v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)

Since this kind of interference appears also in the data streamfor the first MS that is the first column in V the PCRconcept bounds the achievable SIR but in noncooperativescenarios the unreported BSs would be interferers as wellThus the SIR bound is not really limiting the capacity in realworld scenarios

It is important that each BS in the CA is reported by atleast one MS to make sure that the channel matrix does notcontain any rows which are completely zeroThis can be gua-ranteed if the scheduling applies a simple rule

(i) The setMB is composed such that each BS inB hasat least one MS of which it is the strongest BS inB

(ii) Each MS reports at least its strongest BS in the CAThis is easy to fulfil because MSs are generally connected totheir strongest BS in the network If each BS contributing toB adds one of ldquoitsrdquo MSs toMB the wish lists from this set ofMSs will span all BSs inB To increase the systemrsquos capacityalso more than oneMS per BS can be servedThis is of courselimited by the number of AEs locally mounted at BSs

There are a lot of different approaches how an MS canselect the BS whose radio channels are reported In thisarticle two of them are investigated a ldquotop-119873rdquo and a thre-shold-based selection

(i) A top119873 selectionmeans to select the strongest119873 = 120596BSs in the serving cluster to be reported

(ii) For threshold-based selection the MS reports allBSs of the serving CA that have a radio channelconnectivity stronger than a certain threshold belowthe strongest BS

While Top 119873 selection yields to a variable SIR gain witha fixed cooperation overhead the threshold-based selectionhas a variable cooperation overhead but tries to keep acertain cooperation gain which is only limited by inter-CAinterference

With the same simulations as used for Figure 3 (seeTable 1 for details) PCR in combination with an overlappingclustering according to Figure 4 has been simulated Espe-cially forMSs with a low SINR the gains due to partial CoMPare significant and higher than those of intrasite cooperation

45 Scheduling between Cover Shifts The division into covershifts with frequency subbands motivates a load balancingbetween them and the presented clustering in combinationwith PCR allows a certain degree of freedom there becausethe scheduling of MSs between cover shifts can also bereduced to the reported BS

With 120596 = 3 and the top-119873 selection approach thereported BSs can be distributed between one two or threedifferent sites If all of the reportedBSs belong to the same sitethe corresponding MS can be scheduled to each cover shiftbecause the overlapping clustering as illustrated in Figure 4was designed such that each site contributes to a CA in eachcover shift If the three strongest BSs are distributed betweentwo sites there are still two cover shifts left where these MSscould be scheduled to The cover shift is only fixed if allthree reported BSs belong to different sites Figure 8 sketchesthe possible cover shifts for different locations of an MSrsquosthree network wide strongest BSs denoted by the three darkhexagons Of course the number of sites that are includedin a MSrsquos set of reported BSs depends also on 120596 Until now120596 = 3 with a top-119873 selection approach was assumed but ifthe selection is done threshold based 120596 only limited by Ω

Figure 6 compares the top-119873 and threshold-based selec-tion approaches It comes out that a threshold of about 10 dBresults in intrasite cooperation for 80 of all MS Vice versathis means that the SIR is limited to 10 dB for 80 of allMSs due to intersite interference This coincides with thelow gains that are often seen in such simulations as alreadymentioned in the introduction of this articleThe first idea forload balancing is to schedule MSs that can be scheduled onlyto one single cover shift first SecondMSs with two ldquoallowedrdquocover shifts are distributed between the different cover shiftswith the goal in mind to equalize the number of MSs in eachof these cover shifts In a third step all MSs that are free intheir choice are added As long as there are enough of thelatter type ofMSs to schedule an equal load between all covershifts can be reached This technique is described more indetail in [31] More powerful scheduling ideas are sketchedbriefly later in this article

46 Interference Floor Shaping Both overlapping clustersand PCR exploit the fact that the main cooperation gainarises from a significant interference reduction and less froma signal gain Despite oversized and overlapping clustersthere is still a considerable amount of inter-CA interferenceespecially for CA edge users In the following we focus on themitigation of that interference part

Multiple antennas at BSs can be used for a static widebandbeamforming before joint preprocessing is applied Thisallows steering the transmit power more towards the centerof a CA for mitigating inter-CA interference The concept ispresented in Figure 1The depicted beams are divided into theso-called ldquocentre beamsrdquo and ldquoedge-beamsrdquo Each BS has forexample 4 AEs and uses them to provide 2 wideband beamsAdditionally to the horizontally beamforming edge beamshave a stronger down tilt than centre beams and are reducedin their transmit power to further mitigate the interferencepower to adjacent CAs


CA center beam10 tilt

CA edge beamminus 6 dB 15 tilt

CA edge

CA centerHigh signal power

Site with 3 sectorsand 6 beams

300 3300

3060

90120150180210240270

Mitigatedinterference

Figure 1 The complete interference mitigation framework consists of three adjacent sites Multiple antennas at each of the cooperating BSsperform wideband beamforming to steer the signal more towards the CArsquos center

Figure 2 A CA consists of three adjacent sites

1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()

Figure 3 Penetration rate of user centric served MSs

Regarding the channel matrix the effective channelsresulting from the wideband beams are used instead of thephysical antenna coefficients Thus the number of columnsin a channel matrix between a CA and its servedMSs is equalto the number of wideband beamformers used per BS and thewideband beams are referred to as virtual antenna elements(vAEs) In order to be specific the term ldquophysical antennaelements (pAEs)rdquo instead of AEs is used from now on

Only vAEs are used for joint precoding The correspond-ing matrix is denoted by V Thus each antenna weight in Vmust be applied in the same manner at all pAEs the corres-ponding vAE consists of

Assuming a uniform linear array with a 1205822 antennaspacing (120582 = wavelength) 4 pAEs at each BS and usingsin(30 deg) = 05 the (normalized) beamforming vectors

for each BS directed to the left and right are given byrarr119887119871 =(14)[1 119895 minus 1 minus 119895]119879 and rarr119887119877 = (14)[1 minus 119895 minus 1 119895]119879

respectivelyAccording to Figure 1 the wideband beams should also

have an individual down tilt Different down tilts can berealized if each single pAE of the uniform linear antennaarrays is extended to a vertical antenna array again with forexample 1205822 spacing The influences of the vertical antennaarraymdashbasically describing a so-called ldquoactive antennardquomdashcanbe included easily into the mathematical model if all rows ofthe (now) two-dimensional antenna array are concatenated

to one long row which leads to increased lengths ofrarr119887119877 rarr119887119871

Elements corresponding to different rows of the antennaarray must have an additional phase offset according to thedown tilt of center and edge beams Furthermore 119879119909 poweradaptation for cell edge beams is included by a minus6 dB power-down-scaling of the corresponding beamforming vectors In


Figure 4 The basic clustering is illustrated on the most left picture From left to the right more and more cover shifts are added

1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)

Full CoMPPartial CoMP

Geometry factorIntracell site cooperation

Figure 5 The partial CoMP concepts have a significant gaincompared to intrasite cooperation scenarios especially in the low-SINR regions

total four different extended beams result from differentdown tilts and 119879119909 power reduction denoted by 119888

rarr119887119877 119888rarr119887119871(center beams) and 119890

rarr119887119877 119890rarr119887119871 (edge beams) Figure 12 showsthe positions of all these wideband beamformers To keep thenotation clear the length extension of wideband beamform-ers is suppressed in the following since it has no relevant influ-ence on the mathematical model Furthermore the down tiltcould also be realized by two different pAEs with differentdown tilts but would increase the hardware complexity

In Figure 12 the first 6 wideband beams of a CA aredepicted and numbered The corresponding beamformingvectors are concatenated to a block diagonal matrix

V = [v1 sdot sdot sdot v6]= ([119890rarr119887119877 | 119888rarr119887119871] 0[119888rarr119887119877 | 119888rarr119887119871]0 [119888rarr119887119877 | 119890rarr119887119871]) (15)

1 2 304

05

06

07

08

09

1

Number of involved cell sites

CDF

Threshold 20 dBThreshold 15 dB

Threshold 10 dBStrongest 3

Figure 6 The number of involved sites in the set of reported BSdepends on the number of reported BSs and thus on the thresholdif that selection approach is chosen

Figure 7 Using overlapping CAs so-called ldquocover shiftsrdquo can avoidCA edge MSs

The complete MIMO signal model can be described by

y = HVV = HV (16)

The central joint processing (CJP) calculates the joint precod-ing matrix V based on a channel matrix that is a result of theproduct between two matrices that is H and V Thus H is


F

CD

E A

B

F

CD

E A

B

F

CD

E A

B

Figure 8 Dark hexagons denote the three network wide strongestBSs Depending on their distribution an MS can be scheduled intoone two or six different cover shifts

CJP BS 1

BS 2

BS 3

Figure 9 Scheme of cooperation between 3 BSs with 2 pAEs perBS To do the central joint processing a central unit calculates thejoint precoding weights v119898[1] v119898[6] to transmit the symbol 119889119898for MS119898

changed by VThese changes can also decrease the correlationbetween the individual elements in H compared to H Inother words the rank of a matrix consisting of vAEs might behigher compared to the case when the original pAEs are usedasmatrix elementsThis improvement can occur in particularif V is of reduced size

Figure 10 depicts a possible realization of virtual AEsThe virtual antenna elements with tildes are to be applied ontop of the physical antenna weights on the BSs by physicalbeamformers (Here three BSs are used)

Figure 11 illustrates the results comparing SINR valuesutilizing the WB beams with and without CA-specific downtilt and power adaptation In the latter case a lower SINR canbe observed compared to the CoMP simulations presentedearlier in this article Results presented in [32 33] indicatethat this effect comes froman increased inter-CA interferencebecause the pathloss is not as dominant here compared to theCoMP reference scenario without any WB-beamforming

If the WB-beam down tilt and power adaptation areactivated the results in Figure 11 illustrate the effect ofinterference floor shaping Compared to the CoMP referencea significant SINR gain can be observed confirming theperformance of the proposed interference shaping concept

CJP

BS 1

BS 2

BS 3

Figure 10 Each element of v119898 must be spread to all pAEs locallymounted at the BSs Thus there is much less cooperation overheadon the backhaul

SINR (dB)

CDF

1

08

06

04

02

0

0 20 40

CoMP reference (full cooperation of 3 cell sites)

Beamforming only (no down-tilt no power red)

Beamforming -tilting and -power adaptation

Figure 11 The blue line (diamonds) is the reference scenario as inFigure 5 Red (crosses) denotes the wide band (WB) beamformingswitched on but without any additional downtilt or power adapta-tion Green (circles) contains the results for additional downtilt andpower adaptation

5 Real World Issues

The accuracy of the simulation results that were presentedin the previous section is limited by the accuracy of thechannel models that have been applied However the 3GPP3D antenna model does not accurately describe the verticaldimension of a cellular system since it is constantly consid-ered to have LOS characteristics A more solid modelling


1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877

Figure 12 Numbering scheme for wideband beams at first site in aCA

0 1 2 (Km)

Figure 13 Field trial setup and measurement locations (coloreddots) for the first field trial The color indicates the SNR to thestrongest base station for a downtilt of 6∘

would require the consideration ofmultiple reflection clustersas considered in the horizontal dimension In order tovalidate the assumptions and results of the previous sectionswe thus used extensive large scale field trials with regardto these most critical aspects As depicted in Figure 13 thetestbed consists of a total of 16 base stations that are deployedat seven sites It covers a large part of the down town of theGerman city Dresden an area with a typical urban buildingmorphology that is characterized by large apartments of 20ndash50m height a soccer stadium a train station railway tracts

40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850

Distance from base stations (m)

SNR

(dB)

DT = 6DT = 12

Figure 14 Average SNR (dB) over distance from base stationsAverage over results at 16 BSs The dashed lines show a linear fit

and roadswith up to 4 lanes Four base stations are positionedin about 30m and 12 base stations in about 50m aboveground each equipped with a cross-polarized antenna (58degrees half-power beamwidth and 14 dBi gain)The intersitedistance ranges between 500 and 1000m

In a first field trial we were interested in the impact ofthe downtilt on pathloss and coverage In order to investigatethis aspect the samemeasurement route was traversed twiceFor the first trial a downtilt of 6∘ and for the second trial adowntilt of 12∘ were used During the trial a single UE trans-mitted OFDM pilots with transmit power of 37 dBm overa bandwidth of 54MHz at a carrier frequency of 26GHzThe received signal at all base stations was synchronouslycaptured every 10 s and evaluated for channel and SNRestimation For further details on the evaluation procedurewe refer the reader to [21 34]Themeasurement locations aremarked in Figure 13 as well using coloured dots that indicatethe SNR to the strongest base station for each location Intotal 479measurementswere taken in each trial For technicalreasons the measurement locations for both trials are notexactly the samewhich however does not effect the followingresults since the data is evaluated statistically In Figure 14 weshow how the average SNR depends on the distance to thebase stations for the two different downtilt settings

In order to evaluate the relationship between SNR anddistance to the base station we determined the distance toall 16 base stations for each measurement location Figure 14shows the average SNR that was measured at differentdistances to the base stations However for each location onlythe SNRof those base stationswas consideredwhich antennaswere facing to the measurement location with a tolerance of30∘ Figure 14 also shows that the SNR in dBdecreases linearlywith distance with an average SNR degradation (pathlossincrease) of 21 dB and 37 dB every 100mand for a downtilt of6∘ and 12∘ respectively The height of the base stations has a


rather small effect on the average pathloss in our field trialThe results are useful to validate the results of our systemlevel simulations and they show that an appropriate settingof the downtilts in a cover shift allows for separation ofcooperation areas in a way that BSs that are facing outsidethe cooperation areas should apply a larger downtilt in orderto decrease the interference to other cooperation areas Asdescribed previously coverage holes that would occur due tothis strong isolation of cooperation areas would be closed byusing different cover shifts on several layers that are separatedby using different resources for each layer Layer specificcover shifts could be realized by vertical beamforming Ourgoal in the field trial that is described in the following wasto explore how well different cooperation areas could beseparated using these techniques Since vertical beamformingwas not available in the employed field trial system weconcentrated on the separation of a single cooperation areafrom neighboring base stations The general field trial setupis the same as shown in Figure 14 The cooperation area ofinterest was made up of the three sites Hbf-Sud Hbf andLenneplatz Thus in order to achieve a good coverage andlarge cooperation gains inside of the cooperation area a ratherlow downtilt of 7∘ was chosen at base station 2 6 8 that arefacing inward At all other base stations of the cooperationarea a larger downtilt of 17∘ was chosen in order to minimizethe interference to other cooperation areas The downtilt atall surrounding base stations was 17∘ as well as those wouldbe interferers from surrounding cooperation clusters thattransmit on the same resources Figure 15 shows the result ofa field trial where a single transmitting UE traversed througha wide area of the testbed The colour of the measurementlocations indicates the maximum SNR measured at any basestation of the cooperation cluster A black dot is shown atmeasurement locations where the SNR to all base station inthe cooperation cluster is below 0 dB or where a base stationthat is not part of the cooperation cluster has a higher SNRthan all base stations in the cooperation cluster Comparedto the previous trial in Figure 13 the transmit power wasreduced to 18 dBm thus the SNR is clearly lower as it wouldbe achievable in a cellular downlink withmacrobase stations

An inherent feature of the interferencemitigation conceptbeing proposed is joint transmission of base stations in acooperation clusterThe statistics for the SNR that is instanta-neously achieved at several base stations for all measurementlocations in the cooperation area is shown in Figure 16For example the blue curve shows the CDF of SNR of thestrongest base station at each locationThe other solid curvesshow the CDF of the minimum SNR of instantaneouslyachieved at several base stations For example at 40 of themeasurement locations the SNR was larger than 10 dB attwo base stations of the cooperation cluster instantaneouslyConsidering that the transmit power at a typical base stationwould be about 30 dB higher as that used in this field trialshows that there would be definitely a strong benefit fromcooperation The dashed curves show the SNR CDF of thestrongest interfering base station Thus a partial feedback(and precoding) scheme would typically use cooperation ofno more than four base stations In order to increase thisnumber the presented framework incorporates the option for

0 1 2 (Km)

Figure 15 SNR in cooperation area using the tortoise concepts

1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)

SNR of strongest interferer

0 5 10 15 20 25 30

Figure 16 CDF of SNR at the base stations in the CA The dashedcurve shows the CDF of the SNR of the first interferer from outsideof the CA We see that typically 4 base stations in the CA have ahigher SNR than the first interferer

power control as presented in Section 46 Figure 17 showsthe impact of power setting where base stations that faceout of the cooperation area have a transmit power that isreduced by 6 dB Since this field trial was done in the uplinkthe impact of a reduced transmit power was obtained by areduction of the measured SNR While the reduced transmitpower has only small effect on the SNR in the cooperationcluster the strongest interferer is reduced by 6 dB resulting ina significant increase of the average SINR


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30

Figure 17 The same as Figure 15 but with power control

6 Cooperative SchedulingStrategies-Converting SINR intoSpectral Efficiency

Applying the above described framework increases the SINRper UE significantly for the single UE per cell scenario asobvious from Figure 11 This is a big step into the directionof noticeable system level performance gains but leaves amain topic unsolved that is the optimum multiuser CoMPscheduling

CoMP scheduling ismdashsimilar as forMU-MIMOmdasha chal-lenge requiring ideally an exhaustive search in all potentialscheduling and precoding dimensions The goal is to maxi-mize the spectral efficiencywhile keeping a predefined degreeof fairness or even better improving the degree of fairnessThis is equivalent to maximizing the constrained sum rateover all simultaneously served data streams per CA wherethe constraint depends on the intended degree of fairness

For JP CoMP the search space generally comprise thefollowing search dimensions

(i) Setup of CAs with the goal to serve as many aspossible MSs from their most appropriate cells asbeing explained above

(ii) In case of overlapping clusters according to the covershift concept the scheduler has to consider differentallocations of MSs to cover shift

(iii) The grouping of simultaneously served MSs per CAis a very important step for finding the mutually mostorthogonal set of users for a certain subcarrier or PRBIndependent of the specifically used linear precodingtechnique like optimum Wiener filtering zero forc-ing signal to noise and leakage minimization and

so forth the PNL 120572 as being introduced in (4) willexplode for a badly conditioned channel matrixH

(iv) User grouping is closely related with determining theoptimum number of data streams per CA that is howmany users are served with how many data streamsGenerally there is a tradeoffWith increasing numberof data streams the sum rate increases up to the pointwhere the precoding normalization loss 120572 overcomesthe multistream throughput gains 120572 combines asplitting loss due the partitioning of the 119879119909 power tomore streams and the loss for orthogonalization of thedata streams

(v) For each user group and rank the optimum 119879119909- and119877119909- beamformers for the BS and MSs have to bederived

(vi) As one of the last steps frequency dependent schedul-ing of MS sets to PRBs or subcarriers has tomaximizeoverall capacity while ensuring at least proportionalfairness for allMSs Taking all these 6 dimensions intoaccount one can formulate the overall optimizationwhich must be done comprising

(a) cover shifts(b) user groups(c) precoder(d) postcoder(e) number of streams and(f) PRBs

Obviously the overall search space easily explodes evenfor small number of MSs requiring a powerful heuristicscheduling strategy

A completely new score based scheduler has been imple-mented taking advantage of the concepts presented in lastsections that is being limited to one single CAThe descrip-tion of all applied scheduling strategies is beyond the scopeof this paper but so far achieved results look very promisingIt was possible to serve 3 MSs per cell with reasonable datarates leading to a mean throughput of about 7 bitsHzcellfor an urban macro 4 times 2MIMO scenario This is more thana factor of 2 larger than what has been reported in 3GPPfor the same scenario as outcome of the CoMP study item(31 bitssHzcell)

7 Conclusion

Interference mitigation in cellular mobile radio systems is achallenging problem for the high number of interferers thatare potentially wide spread between far off BSs Networkwidecooperation iswell known to be extremely powerful in theorybut practical feasible systems have to divide the network intosmaller cooperation areas (ie clusters) As a consequencethere will be inter cooperation area interference spoilingmost of the gains observed for network wide interferencefree transmission To get back to similar performance asfor network wide cooperation we propose a frameworkfor interference mitigation including oversized clusters and


overlapping cooperation areas interference floor shaping bymeans of wideband beamformers and a novel cooperativescore based scheduler The performance of all that conceptshas been investigated by a software simulation frameworkwritten in Matlab and by real world field trials made at theTU-Dresden-LTE-Testbed

The results indicate that the ideas can really be appliedin future networks with significant gains regarding the SIRSINR For the future further fine tuning is planned to reacheven higher gains

Abbreviations

3GPP 3rd generation partnership projectAE Antenna elementBS CellCA Cooperation areaCDMA Code division multiple accessCJP Central joint processingCoMP Cooperative multi point transmissionCSI Channel state informationCU Central unitFFP Fractional frequency partitioningICI Inter cell interferenceICIC Inter cell interference coordinationIRC Interference rejection combiningISD Inter site distanceJP Joint precodingJT Joint transmissionLOS Line of sightLTE Long term evolutionMIMO Multiple input multiple outputMU-MIMO Multi user MIMOMS Mobile stationNLOS Non line of sightOFDMA Orthogonal frequency division multiple

accesspAE Physical antenna elementPCR Partial channel reportingPRB Physical resource blockPNL Precoding normalization lossSCME Spatial channel model extendedSINR Signal to noise and interference ratioSIR Signal to interference ratioSISO Single input single outpoutvAE Virtual antenna element

References

[1] S Redl M Weber and M Oliphant GSM and PersonalCommunications Handbook Artech House 1998

[2] K Rawer Wave Propagation in the Ionosphere Kluwer Aca-demic 1993

[3] H Dai A F Molisch and H V Poor ldquoDownlink capacity ofinterference-limitedMIMO systems with joint detectionrdquo IEEETransactions onWireless Communications vol 3 no 2 pp 442ndash453 2004

[4] A Sendonaris and V Veeravalli ldquoThe capacity-coverage trade-off in CDMA systems with soft handoffrdquo in Proceedings of the

31st Asilomar Conference on Signals Systems amp Computers vol1 pp 625ndash629 IEEE November 1997

[5] R Kwan C Leung and J Zhang ldquoProportional fair multiuserscheduling in LTErdquo IEEE Signal Processing Letters vol 16 no 6pp 461ndash464 2009

[6] D Kimura Y Harada and H Seki ldquoDe-centralized dynamicICIC using X2 interfaces for downlink LTE systemsrdquo in Pro-ceedings of the 73rd IEEEVehicular Technology Conference (VTCrsquo11-Spring) pp 1ndash5 IEEE May 2011

[7] P W Baier M Meurer T Weber and H Troeger ldquoJointtransmission (JT) an alternative rationale for the downlink oftime division CDMA using multi-element transmit antennasrdquoin Proceedings of the 6th IEEE International Symposium onSpread Spectrum Techniques and Applications vol 1 pp 1ndash5IEEE September 2000

[8] N JindalW Rhee S Vishwanath S A Jafar and A GoldsmithldquoSum power iterative water-filling for multi-antenna Gaussianbroadcast channelsrdquo IEEE Transactions on Information Theoryvol 51 no 4 pp 1570ndash1580 2005

[9] G J Foschini K Karakayali and R A Valenzuela ldquoCoordi-nating multiple antenna cellular networks to achieve enormousspectral efficiencyrdquo IEE Proceedings vol 153 no 4 pp 548ndash5552006

[10] M K Karakayali G J Foschini and R A Valenzuela ldquoNetworkcoordination for spectrally efficient communications in cellularsystemsrdquo IEEE Wireless Communications vol 13 no 4 pp 56ndash61 2006

[11] C Botella T Svensson X Xu and H Zhang ldquoOn the perfor-mance of joint processing schemes over the cluster areardquo inProceedings of the 71st IEEE Vehicular Technology Conference(VTC rsquo10-Spring) pp 1ndash5 IEEE May 2010

[12] P Marsch and G P Fettweis Coordinated Multi-Point inMobile Communications From Theory to Practice CambridgeUniversity Press 2011

[13] 3GPP Coordinated Multi-Point Operation for LTE TR 368193rd Generation Partnership Project (3GPP) 06 2011

[14] A Osseiran J F Monserrat and W MohrMobile and WirelessCommunications for IMT-Advanced and Beyond Wiley 2011

[15] D S Baum J Hansen and J Salo ldquoAn interim channelmodel for beyond-3G systems extending the 3GPP spatialchannel model (SCM)rdquo in Proceedings of the 61st IEEEVehicularTechnology Conference (VTC rsquo05-Spring) vol 5 pp 3132ndash3136IEEE June 2005

[16] N U Hassan C Yuen and Z Zhang ldquoOptimal power con-trol between two opportunistic cooperative base stationsrdquo inProceedings of the 13th IEEE International Workshop on SignalProcessing Advances in Wireless Communications (SPAWC rsquo12)pp 194ndash198 June 2012

[17] M Eslamifar W H Chin C Yuen and G Y Liang ldquoPer-formance analysis of two-way multiple-antenna relaying withnetwork codingrdquo in Proceedings of the 70th IEEE VehicularTechnology Conference Fall (VTC rsquo09-Fall) pp 1ndash5 IEEESeptember 2009

[18] V DrsquoAmico A Dekorsy A Gouraud et al ldquoARTIST4G a wayforward to the interference problem in futuremobile networksrdquoin Proceedings of Future Network and Mobile Summit pp 1ndash8IEEE June 2010

[19] W Mennerich and W Zirwas ldquoUser centric coordinated multipoint transmissionrdquo in Proceedings of the 72nd IEEE VehicularTechnologyConference (VTC rsquo10-Fall) OttawaCanada Septem-ber 2010


[20] W Mennerich and W Zirwas ldquoImplementation issues of thepartial CoMP conceptrdquo in Proceedings of the 21st InternationalSymposium on Personal Indoor and Mobile Radio Communica-tions (PIMRC rsquo10) pp 1939ndash1944 Istanbul Turkey September2010

[21] M Grieger G Fettweis and P Marsch ldquoLarge scale field trialresults on uplink CoMP with multi antenna base stationsrdquo inProceedings of the IEEE Vehicular Technology Conference (VTCrsquo11-Fall) pp 1ndash5 September 2011

[22] R Hunger F A Dietrich M Joham and W Utschick ldquoRobusttransmit zero-forcing filtersrdquo inProceedings of ITGWorkshop onSmart Antennas pp 130ndash137 IEEE March 2004

[23] T Yoo and A Goldsmith ldquoOn the optimality of multiantennabroadcast scheduling using zero-forcing beamformingrdquo IEEEJournal on Selected Areas in Communications vol 24 no 3 pp528ndash541 2006

[24] C B Peel B M Hochwald and A L Swindlehurst ldquoAvector-perturbation technique for near-capacity multiantennamultiuser communication-part I channel inversion and regu-larizationrdquo IEEE Transactions on Communications vol 53 no1 pp 195ndash202 2005

[25] B M Hochwald C B Peel and A L Swindlehurst ldquoA vector-perturbation technique for near-capacity multiantenna mul-tiuser communication-part II perturbationrdquo IEEE Transactionson Communications vol 53 no 3 pp 537ndash544 2005

[26] M H M Costa ldquoWriting on dirty paperrdquo IEEE Transactions onInformation Theory vol 29 no 3 pp 439ndash441 1983

[27] A Goldsmith S A Jafar N Jindal and S Vishwanath ldquoCapac-ity limits of MIMO channelsrdquo IEEE Journal on Selected Areas inCommunications vol 21 no 5 pp 684ndash702 2003

[28] D Gesbert M Shafi D S Shiu P J Smith and A NaguibldquoFrom theory to practice an overview of MIMO space-timecoded wireless systemsrdquo IEEE Journal on Selected Areas inCommunications vol 21 no 3 pp 281ndash302 2003

[29] W Yu and T Lan ldquoTransmitter optimization for the multi-antenna downlink with per-antenna power constraintsrdquo IEEETransactions on Signal Processing vol 55 no 6 pp 2646ndash26602007

[30] S Kaviani and W Krzymien ldquoOptimal multiuser zero forcingwith per-antenna power constraints for network MIMO coor-dinationrdquo EURASIP Journal on Wireless Communications andNetworking vol 2011 Article ID 190461 2011

[31] W Mennerich and W Zirwas ldquoUser centric scheduling incooperative networksrdquo in Proceedings of the 1st Middle EastConference on Antennas and Propagation (MECAP rsquo10) pp 1ndash5 Cairo Egypt October 2010

[32] W Mennerich and W Zirwas ldquoInterference floor shapingfor cooperative cellular radio systemsrdquo in Proceedings of the74th IEEE Vehicular Technology Conference (VTC rsquo11-Fall) SanFrancisco Calif USA September 2011

[33] WMennerich andW Zirwas ldquoReporting effort for cooperativesystems applying interference floor shapingrdquo in Proceedings of22nd Annual IEEE International Symposium on Personal Indoorand Mobile Radio Communications (PIMRC rsquo11) pp 541ndash545Toronto Canada September 2011

[34] M Grieger and G P Fettweis ldquoImpact of antenna downtilton cooperative uplink detection in a large scale field trialrdquo inProceedings of GLOBECOM Workshops (GLOBECOM rsquo11) pp431ndash435 Houston Tex USA December 2011

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



In long-term evolution (LTE) Release 10 and now 11there is a lot of discussions about the so-called enhancedICIC (eICIC) schemes in the context of heterogeneous net-works comprising pico- andmacro- or femto- andmacro-sta-tions By transmitting the so-called almost blank subframes(ABS)mdashalmost blank because the common reference signals(CRS) have to be transmitted continuously for backwardcompatibilitymdashin time domain subframes with different levelof interference are generated providing some options forenhanced scheduling

In 2000 Baier et al [7] proposed the idea of cooperativetransmission systemsmdashalso known as networkmultiple inputmultiple output (MIMO) or for example in 3rd genera-tion partnership project (3GPP) as cooperative multi pointtransmission (CoMP) This is a much more fundamentalsolution partly being inspired from the research in the areaof MIMO systems Many groups predicted large theoreticalperformance gains with spectral efficiencies being severalfactors higher than that of conventionalmobile radio systemsFor example for 2 and 4 antennas per cell (BS) and mobilestation (MS) 10th of bitssHzcell have been promised atsignal to noise and interference ratios (SINRs) of less than 10to 20 dB [8ndash10]

Common to all these approaches is that a networkwide precoder over all cells including all MSs is assumedwhich eliminates any interference after proper precoding Inpractical mobile radio systems network wide precoding isunrealistic as networks might span a whole city or even awhole country Neither an according backhaul network canbe implemented nor channel estimation as well as the report-ing of all channel components is feasible For that reason itis necessary to partition the network into so-called coop-eration areas (CAs) connecting a limited number of cellsUnfortunately this step turns the interference free systemback into an interference limited one because of the remain-ing intercooperation area interference For quite sometimea lot of research concentrated on the area of clustering [11]applying different types of optimization algorithms to thisnonconvex optimization problem

Despite the research effort there is still a significantgap between the theoretical predicted gains and the resultsobserved from system level simulations so far [12] Onemainactivity in this area is the so-called 3GPP LTE CoMP studyitem which started in LTE Release 10 has been re-invokedfor LTE Release 11 and finally has been converted into a 3GPPLTE working item in September 2011 [13] While CoMPincludesmany different kinds of cooperation like coordinatedscheduling (CS) coordinated beamforming (CB) dynamic orfast cell selection (DCS) and so forth in our contribution thefocus is mainly on joint precoding (JP) or more specificallyon joint transmission (JT) CoMP JT CoMP is generallyaccepted to be the scheme with the highest potential gainsas signals being interferers in noncooperative scenarios areturned into useful signal power for JT CoMP This is themain benefit over pure interference avoidance strategies suchas CSCB Reference [14] contains the best so far achievedresults for the main 3GPP CoMP schemes In case of JTCoMP the best performance for a 4 times 2MIMO system doingjoint transmission over 9 cells was 4028 bitsHzcell being

far away from the projected 10 bitsHzcell from theoreticalanalysis

So one main issue to be tackled in this contributionis to reveal and analyze a fundamental limit of practicaland realizable CoMP schemes mainly due to the requiredclustering andor due to different assumptions for the radiochannels Note for theoretical analysis Rayleigh channelsare often assumed which are known to have an optimaleigenvalue distribution while macrocellular radio channelslike the spatial channel model extended (SCME) as beingused in 3GPP [15] may suffer from correlation of collocatedantenna elements and certain line of sight (LOS)nonline ofsight (NLOS) probabilities

The important question is whether the optimizationproblem is limited by fundamental reasons for exampleesteeming from clustering or just due to the dimension of theoptimization problemThefirst casewould be in combinationwith its significant complexity almost a killing point againstJT CoMP while in the second case it would be just a matterof doing things right

We have no final answer to this question yet but inthis contribution we propose an overall framework for inter-ference mitigation of which JT CoMP is just one of the build-ing blocks Based on an in depth analysis of interference con-ditions scheduling effects and so forth different conceptsare combined to get much closer to the theoretical results ofnetwork wide cooperation with clustered cooperation areasin macrocellular radio environments

Reference [16] mentions the importance of interferenceand provides an optimal power control provided for twoopportunistic cooperative base stations However the authorsmention that the algorithm does not bring performance ben-efits if there are only cell edge users Besides these ideas [17]suggests a concept to improve the bandwidth efficiency withtwo-way relay nodes where these nodes have multiple anten-nas Upper bounds regarding symbol error probability arederived as well as the order of diversity of the two proposedschemes

The proposed framework is more than CoMP andincludes techniques like interference rejection combining(IRC) multiuser MIMO (MU-MIMO) interference floorshaping wideband beamforming antenna tilting poweradaptation clustering user grouping effective precodingincluding119879119909 diversity per cell and per cooperation area sche-duling optimum channel state information (CSI) estimationand according to low rate but accurate feedback links

Implementation of specific issues like the just mentionedCSI estimation and feedback as well as intersite synchroniza-tion backhaul overhead and so forth are investigated inmoredetail in the EU founded project Artist4G [18] while thiscontribution has its main focus on maximizing system levelgains under ideal conditions In the end both parts have to becombined appropriately to form an overall system concept

Some items like partial CoMP cover shifts or interferencefloor shaping by cooperationwide vertical beam steering havebeen partly described in [19 20] while the CoMP schedulerresults and verification of main parts measurements in thereal world CoMP testbed from TUD [21] are presented herefor the first time













B = 1198871 119887119861M = 1198981 119898119872997904rArr[[[

H1198981B

H119898119872B


] (1)







119896isinM119887119898

h119898119887v119896119889119896 + or119868119898 + 119899 (2)




119896isinM119887119898

1199011198961003817100381710038171003817h119898119887v119896100381710038171003817100381722119868 (119898 119902) = sum119896isinM119902

11990111989610038171003817100381710038171003817h119898119902v1198961003817100381710038171003817100381722 119902 = 119887 (3)







119898isinM119887


119898isinM119887


(4)


SIR (119898 119887) = 120572 (119898 119887) 119878 (119898 119887)119868 (119898 119887) + sum119902 =119887 119868 (119898 119902) (5)


MB = ⋃119887isinB

M119887 (6)








1 Bsite2 Bsite


119896







(8)








B|B|=Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (9)



B|B|=120596⩽Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (10)








1 Bsite3 )


119899 ) ≪ SIR (119898 Bsite1 Bsite

2 Bsite3 )119899 isin 2 3

(12)








Ω)minus1


























v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation





















B = 1198871 119887119861M = 1198981 119898119872997904rArr[[[

H1198981B

H119898119872B


] (1)







119896isinM119887119898

h119898119887v119896119889119896 + or119868119898 + 119899 (2)




119896isinM119887119898

1199011198961003817100381710038171003817h119898119887v119896100381710038171003817100381722119868 (119898 119902) = sum119896isinM119902

11990111989610038171003817100381710038171003817h119898119902v1198961003817100381710038171003817100381722 119902 = 119887 (3)







119898isinM119887


119898isinM119887


(4)


SIR (119898 119887) = 120572 (119898 119887) 119878 (119898 119887)119868 (119898 119887) + sum119902 =119887 119868 (119898 119902) (5)


MB = ⋃119887isinB

M119887 (6)








1 Bsite2 Bsite


119896







(8)








B|B|=Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (9)



B|B|=120596⩽Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (10)








1 Bsite3 )


119899 ) ≪ SIR (119898 Bsite1 Bsite

2 Bsite3 )119899 isin 2 3

(12)








Ω)minus1


























v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation














119898isinM119887


119898isinM119887


(4)


SIR (119898 119887) = 120572 (119898 119887) 119878 (119898 119887)119868 (119898 119887) + sum119902 =119887 119868 (119898 119902) (5)


MB = ⋃119887isinB

M119887 (6)








1 Bsite2 Bsite


119896







(8)








B|B|=Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (9)



B|B|=120596⩽Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (10)








1 Bsite3 )


119899 ) ≪ SIR (119898 Bsite1 Bsite

2 Bsite3 )119899 isin 2 3

(12)








Ω)minus1


























v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












B|B|=Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (9)



B|B|=120596⩽Ω

1003817100381710038171003817h119898B100381710038171003817100381722 (10)








1 Bsite3 )


119899 ) ≪ SIR (119898 Bsite1 Bsite

2 Bsite3 )119899 isin 2 3

(12)








Ω)minus1


























v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation


































v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












v31 11990732 11990733 11990734 11990735v41 11990742 11990743 11990744 1199074511990751 11990752 11990753 11990754 11990755)

(14)


















CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












CA edge



300 3300

3060

90120150180210240270




1 2 3 4 5 6 7 8 9

100

80

60

40

20

0

MSs

()












1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











1

08

06

04

02

0

CDF

minus10 0 10 20 30 40SINR (dB)









1 2 304

05

06

07

08

09

1


CDF






y = HVV = HV (16)



F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










F

CD

E A

B

F

CD

E A

B

F

CD

E A

B


CJP BS 1

BS 2

BS 3






CJP

BS 1

BS 2

BS 3


SINR (dB)

CDF

1

08

06

04

02

0

0 20 40





5 Real World Issues



1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










1

23

4

5 6

119888

rarr119887119877

119888

rarr119887119877

119888

rarr119887119871

119890

rarr119887119871

119888

rarr119887119871

119890

rarr119887119877


0 1 2 (Km)



40

35

30

25

20

15

1050 150 250 350 450 550 650 750 850


SNR

(dB)

DT = 6DT = 12








0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












0 1 2 (Km)


1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30




1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










1

09

08

07

06

05

04

03

02

01

0

CDF

SNR (dB)


0 5 10 15 20 25 30
















7 Conclusion





Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation












Abbreviations



References







































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation



























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









Documents

Research Article Interference Mitigation Framework for ... · Research Article Interference Mitigation Framework for ... of interference are generated, ... includes techniques like