东南大学研究生英语雅思写作下学期大作业Word文件下载.docx
《东南大学研究生英语雅思写作下学期大作业Word文件下载.docx》由会员分享,可在线阅读,更多相关《东南大学研究生英语雅思写作下学期大作业Word文件下载.docx(6页珍藏版)》请在冰点文库上搜索。
Weconsiderahierarchicalnetworkthatconsistsofmobileusers,atwo-tieredcellularnetwork(namelysmallcellsandmacrocells)andcentralrouters,eachofwhichfollowsaPoissonpointprocess(PPP).Inthisscenario,smallcellswithlimited-capacitybackhaulareabletocachecontentunderagivensetofrandomizedcachingpoliciesandstorageconstraints.Moreover,weconsiderthreedifferentcontentpopularitymodels,namelyfixedcontentpopularity,distance-dependentandload-dependent,inordertomodelthespatio-temporalbehaviorofusers’contentrequestpatterns.Wederiveexpressionsfortheaveragedelayofusersassumingperfectknowledgeofcontentpopularitydistributionsandrandomizedcachingpolicies.Althoughthetrendoftheaveragedelayforallthreecontentpopularitymodelsisessentiallyidentical,ourresultsshowthattheoverallperformanceofcached-enabledheterogeneousnetworkscanbesubstantiallyimproved,especiallyundertheloaddependentcontentpopularitymodel.
Besides,Becauseofthelimitationofresearchconditions,thetotalnetworkdelay,networkcostandoptimizationofnetworkparametersarenotanalyzed.
Keywords:
edgecaching,Poissonpointprocess,stochasticgeometry,mobilewirelessnetworks,5G
Introduction
Itisknownthatcontentcachingin5Gheterogeneouswirelessnetworksimprovesthesystemperformance,andisofhighimportanceinlimited-backhaulscenarios.Mostexistingliteraturefocusesonthecharacterizationofkeyperformancemetricsneglectingthebackhaullimitationsandthespatio-temporalcontentpopularityprofiles.Inthiswork,weanalyzethegainsofcachinginheterogeneousnetworkdeployment,andconsidertheaveragedelayasaperformancemetric.
Firstlyweusepassionpointprocess(PPP)methodtobuildthemodel.Thisheterogeneousnetworkconsistsofmobileterminals(users),cache-enabledsmallbasestations(SBSs),macrobasestations(MBSs)andcentralrouters.Inthisnetworksetting,ausermayexperiencedelaysduetodownlinktransmissions,backhaulandcaches.
Moreover,inordertocapturethespatio-temporalcontentaccesspatternsofusers,wesupposefixedcontentpopularity,distance-dependentandload-dependentcontentpopularities.Assumingthatthecontentpopularitydistributionisperfectlyknownatthesmallbasestations,weexplorethreedifferentcachingpoliciesbasedoncontent-popularityandrandomization.Andfinally,wedrawourconclusion.
Methodology
First,ourResearchobjectiveistogettheaveragedelayinacache-enabledtwo-tieredcellularnetworkmodeledbystochasticgeometry,thefigureoneisanillustrationoftheconsideredsystemmodel,themethodologypartincludingthefollowingparts.
Thefirstisthetopology,thecentralrouters,MBSs,SBSs,USERsaremodeledbytheindependentPPPmodels,thesecondisthesignalmodel,forthetractabilityfortheproblem,weassumethatthetypicaluserexperiencetherayleighfadingandthestandardpowerlawpathloss,alsoweassumethatthenetworkisinterference-limited,thatistosaytheinterferencepowerdominatesoverthenoisepower.
Thenitisthecachingmodel,whenauserhasacontentrequest,weassumethattherequestisdrawnfromacontinuouszipfdistribution,inthefunction,thereisacontentpopularityparameter,basedonthisparameter,wehavethreecachingmodels,Fixedmeansthatthecontentpopularityisidenticalforallusers,allSBSsobservethesamedistribution,Distance-dependentmeansthateachuserhasadistance-dependentsteepnessfactor,thefactorisrelatedtotheaveragedistancebetweentheSBSanditsusers.
Basedonthecachingmodels,hereweconsiderthreecachingpolicies:
StdPopmeanstheSBSscachethemostpopularcontentfromthecatalogue,theUnirandmeansthecontentsarecacheduniformlyatrandom,theMixPoppolicymeansthatpartstoragecachethemostpopularcontents,theotherpartcachethecontentatrandom.
Thelastpartisthedelay,ourmetricinthispaper,thedelayincludingthefollowingthreeparts,WhentheBSsdeliverthecontentstotheirintendedmobileusers,itincursthedownlinkdelay,WhenthecontentrequestedarenotcachedintheBSs,theBSshavetogetthecontentformthecentralroutersthroughthebackhaul,thenitcausesdelay,whenthecontentrequestedinthecaches,thecontentneedtobetakenoutfromthecaches,itcausesdelay.
ReportingandDiscussingResults
Inthissection,wenumericallyvalidateourapproximationsderivedintheprevioussection.Theimpactofcriticalsystemparametersarediscussedasfollows.
Fig.(a)ImpactofMBSdensity.(b)Impactofsmallcelldensity.
(c)ImpactoftargetSIR.(d)Impactofstoragesize.
ImpactofMBSdensityλmc:
ThechangeofaveragedelaywithrespecttotheMBSdensityisgiveninFig.a.Therein,asthenumberofMBSsincreases,weobserveanincrementinaveragedelayatmillisecondlevel.ThisismainlyduetothebackhaulasthedelayinbackhaulisproportionaltothedistanceandaveragenumberofconnectedMBSs.Ontheotherhand,theaveragedelayinSBSsremainsstaticinthissetup.However,wenotethattheaveragedelayexperiencedbyatypicalsmallcelluserisreducedbyaddingcachingcapabilitiesatthebasestations.
Impactofsmallcelldensityλsc:
ThechangeoftheaveragedelaywithrespecttothesmallcelldensityisdepictedinFig.b.SimilarlytothepreviousfigureforMBSdensity,weseethattheaveragedelayincreasesforallkindofsmallcellusers.However,inthisnumericalsetup,therateofincrementindelaywithno-cachingcapabilitiesattheSBSsishigherthanthedelayexperiencedbythetypicaluserswithcache-enabledSBSs.Comparedtothefixedandload-dependentcontentpopularities,thetypicaluserunderload-dependentcontentpopularityexperienceslessdelaywhenthenumberofSBSsincreases.
ImpactoftargetSIRγ:
Inoursetup,yetanotherimportantdesignparameteristhetargetSIR.Inthisregard,theaveragedelayvariationwithrespecttothetargetSIRisillustratedinFig.2c.Asobservedinthefigure,theaveragedelayincreasesbyimposinghighertargetSIRvalues.ThischangeisonlyvisibleinlowvaluesoftargetSIR,whereasthevariationofdelayinhighervaluesoftargetSIRisnegligible.Thismightstemfromthefactthatthedownlinkdelayisnotadominatingfactorinourscenariocomparedtothebackhauldelay.Atypicaluserconnectedtothesmallcellwithnocachingcapabilitiesexperiencesthehighestdelay,whereastheminimumdelayisachievedbyusingMixPoppolicyunderload-dependentcontentpopularity.ThedelayofatypicalMUremainsbetweenaSUwithno-cachingandcachingcapabilitiesatthebasestations.
ImpactofstoragesizeS:
Yetanothercrucialdesignparameterinoursetupisthestoragesize.TheimpactofstoragesizeontheaveragedelayisshowninFig.2d.Indeed,asobservedfromthefigure,dramaticaldecreaseindelayisobservedbyincreasingthestoragesizeofsmallbasestations.Similarlytopreviousobservations,themostsensitivecontentpopularityfortheaveragedelayistheload-dependentcontentpopularity.
Conclusion
Inthiswork,wehavecharacterizedtheaveragedelayofmacrocellusersandsmallcellusersunderbackhaulconstraintsandcachingcapabilitiesatthesmallbasestations.Weconsideramulti-tierheterogeneousnetworkinthetwo-dimensionalEuclideanplaneandatypicalmobileuser,andeachbasestationperfectlyobservesthecontentpopularitiesaccordingtothreedifferentmodels:
fixedmodels,distance-dependentmodelsandload-dependentmodels.Afterthat,somecachingpoliciessuchasstdPoppolicies,unirandpoliciesandmixpoppolicieshavebeenconsidered.Finallytheimpactofcriticalsystemparameterswhichincludemacrobasestationsdensity,smallcelldensity,targetofsignaltointerferenceratioandstoragesizeontheaveragedelayarediscussed.Althoughthetrendoftheaveragedelayforallthreecontentpopularitymodelsisessentiallyidentical,ourresultsshowthattheoverallperformanceofcached-enabledheterogeneousnetworkscanbesubstantiallyimproved,especiallyundertheloaddependentcontentpopularitymodel.Themainconclusionfromthisworkisthatcachingatthesmallbasestationsallowsforbalancingtheaverageaccessdelaytothecontents,especiallyifheterogeneousnetworkdensificationunderlimitedbackhaulisconsidere
Reference
[1]E.Ba¸
stu˘g,M.Bennis,andM.Debbah,“LivingontheEdge:
Theroleofproactivecachingin5Gwirelessnetworks,”IEEECommunicationsMagazine,vol.52,no.8,pp.82–89,August2014.
[2]Z.Chen,J.Lee,T.Q.Quek,andM.Kountouris,“Cooperativecachingandtransmissiondesignincluster-centricsmallcellnetworks,”arXivpreprintarXiv:
1601.00321,2016.
[3]M.Afshang,H.S.Dhillon,andP.H.J.Chong,“Modelingandperformanceanalysisofclustereddevice-to-devicenetworks,”arXivpreprintarXiv:
1508.02668,2015.
[4]B.SerbetciandJ.Goseling,“Onoptimalgeographicalcachinginheterogeneouscellularnetworks,”arXivpreprintarXiv:
1601.07322,2016.
[5]S.Yan,M.Peng,andW.Wang,“Useraccessmodeselectioninfogcomputingbasedradioaccessnetworks,”arXivpreprintarXiv:
1602.00766,2016.
[6]B.BlaszczyszynandA.Giovanidis,“Optimalgeographiccachingincellularnetworks,”inIEEEInternationalConferenceonCommunications(ICC),June2015,pp.3358–3363.