国际会议演讲稿修订版.docx
《国际会议演讲稿修订版.docx》由会员分享,可在线阅读,更多相关《国际会议演讲稿修订版.docx(8页珍藏版)》请在冰点文库上搜索。
国际会议演讲稿修订版
HUAsystemofficeroom【HUA16H-TTMS2A-HUAS8Q8-HUAH1688】
国际会议演讲稿
自我介绍
Thankyou,Mr./Ms.Chair./professor
Mynameissangqian.Iamveryhonoredtobeheretodooralpresentation.
IamaMasterstudentfromHohaiUniversityandIamcurrentlydoingsomeresearchonphysicallayersecurity.
Today,Iwouldliketosharewithyousomeofmyresearchonrelayselectionincooperativecommunication.(external/ekstrnl;kstrnl/)
内容安排:
Mypresentationincludesthesefiveparts.
First,somebackgroundinformationaboutthisresearch;
Second,systemmodelwehavedone;
Third,NN-basedrelayselectionschemewehaveproposed
Forth,Simulationandresultsanalysis
Andlast,someconclusionswehavegot
P4
Partone,introduction
Firstly,Iwouldliketogiveyouabitofbackground.
Differingfromthetraditionalcryptographictechniquesbasedonsecretkeys,wecanmakeuseofwirelesschannelcharacteristicstoenhancephysicallayersecurity.
Cooperativecommunicationhasbeenwidelyrecognizedasaneffectivewaytocombatwirelessfadingandprovidediversitygainwhichisoneoftheresearchhotspots.
Machinelearningasanemergingtechnologyhasbeenwidelyappliedinimageprocessing,cancerprediction,stockanalysisandotherfields.Sowhynottryitinwirelesscommunication?
P5:
Next,Iwanttotalkalittlebitaboutpresentstudy
Recentstudiesondeeplearningforwirelesscommunicationsystemshaveproposedalternativeapproachestoenhancecertainpartsoftheconventionalcommunicationsystemsuchasmodulationrecognition、channelencodinganddecoding、channelestimationanddetectionandanautoencoderwhichcanreplacethetotalsystemwithanovelarchitecture
【modulationrecognition:
AnNNarchitectureformodulationrecognitionthatconsistsofa4-layerNNandtwotwo-layerNNs。
channelencodinganddecoding:
AplainDNNarchitectureforchanneldecodingtodecodekbits
messagesfromNbitsnoisycodewords。
channelestimationanddetection:
Adense-Netforsymbol-to-symboldetectioncanadoptlongshort-termmemory(LSTM)todetectanestimatedsymbol.
Autoencoder:
theautoencodercanrepresenttheentirecommunicationsystemandjointlyoptimizethetransmitterandreceiveroveranAWGNchannel.
】
P6
SowhydidweconductthisresearchWell,wewanttoexploitthepotentialbenefitsofdeeplearninginenhancingphysicallayersecurityincooperative(/k'prtv/wirelesscommunicationandreducethefeedbackoverheadinlimitedspectrumresoucebyourourproposedscheme.
P8
Nowletmemoveontoparttwo-systemmodel
Here,youcanseeafigurewhichisasystemmodel.
Thisisthesource;thesearetherelaynodesandthisisthedestination,thisistheeavesdropper
Thewholeprocessofcooperativewirelesscommunicationcanbedividedintotwophases.
Inthefirstphase,thesourcebroadcaststhesignaltotheoptimalrelaywhichguaranteesperfectsecurity.AsshowninFig1,
representsafadingcoefficientofthechannelfromthesourcetotherelaynode(
.)
Inthesecondphase,theoptimalrelayforwardsascaledversionofitsreceivedsignaltothedestinationinthepresenceoftheeavesdropper,wheretheoptimalrelayisconsideredtoadoptamplify-and-forward(AF)relayscheme.
Inthisfigure,
representsafadingcoefficientofthechannelfromtherelay
tothedestination
representsafadingcoefficientofthechannelfromtherelay
totheeavesdropper.
P9:
Hereyoucanseesomefollowingexpressions.Iamnotgoingtowasteourprecioustimeonthelengthyderivation.Iwouldliketoinviteyoutodirectlytakealookattheequationinitsfinalform.
Thisistheoptimalindexoftheselectedrelaywiththeconventionalrelayselectionscheme.Amaongthisexpression
representstheachievablesecrecyrateofsystemmodelwhenthe
relayisselected.
P11
Nowletmemovetopartthree-----NN-basedRelaySelection
Hereyoucanseeafigurewhichshowsconventional3-layerneuralnetwork.Itconsistsofinputlayer,hiddenlayer1,hidden(/'h?
dn/)layer2andoutputlayer.Neuralnetworkcanlearnfeaturesfromrawdataautomaticallyandadjustparameters(/prmt(r)z/)flexibly(/'fleksbli/)suchasweightsandbiases.
Incomplex(/'kmpleks/)conditions(scenarios(/s'nɑ?
r/),)Neuralnetworkhaspromisingapplicationsinrelayselectionforseveralreasons.
First,thedeepnetworkhassuperior(/supr/)learningabilitydespite(/d'spat/)thecomplexchannelconditions.
Second,Neuralnetworkcanhandlelargedatasetsbecauseofdistributed(/d'strbjtd/)andparallel(/'prlel/)computing(/km'pjut/s,whichensurecomputation(/kmpj'te()n/)speedandprocessingcapacity(/k'pst/).
Third,variouslibrariesorframeworks,suchasTensorFlow,Theano,andCaffegiveitwideapplications
Inthispaper,theproblemoftherelayselectionismodeledasamulti(/'mlt/,ao)-classificationproblem.Weadoptsimpleneuralnetwork(NN)toselecttheoptimalrelaytoguaranteesperfectsecrecyperformanceofrelaycooperativecommunicationsystem.(enhancephysicallayersecurity)
P12
Beforetrainingtheclassificationmodel,weneedtomakesomepreparationfordeeplearningtoacquireatrainingsetandatestingset.
First,weneedtoproducerealfeaturevectorforeachexampleaccordingtochannelstateinformation;becausethechannelstateinformationmatricesiscomposedofcomplexnumbersbutfeaturevectorsaregenerallycomposedofrealnumbers.Soweneedtochangecomplexnumbersintorealnumberswithabsolute(/'bslut/)valueoperation.Moreover,inordertoimprovetheclassificationperformance(precision),itisnecessarytonormalizethefeaturevectors.
Second,weneedtodesignkeyperformanceindicator(KPI).Inordertoeffectivelypreventtheeavesdropperfrominterceptinginformation,wechooseachievablesecrecyrateastheKPIofsystem.ThisKPIindicates(represents/shows)thedifferenceoftheachievableratefromthesourcetothedestinationandtheachievableratefromthesourcetotheeavesdropper.
Third,wecanmakelabelsforexamplesaccordingtoKPI.theindexoftherelaywhichobtainsthemaximum(/'mksmm/)KPIisregardedastheclasslabeloftheexample.
P13
Classificationmodel
Thispicture(isabout)showsthewholeprocessofbuildingclassificationmodel.
Thewholeprocessofbuildingclassificationcanbedividedintotwophases,namelytrainingphaseandtestingphase.Inthefirstphase,weneedtochoosesuitablehyper(/'hap/)parameterstotrainneuralnetworkmodel.Inthesecondphase,wecanpredict(/pr'dkt/)labelsofoptimalrelayaccordingtoinputdataandassessclassificationperformance.
P15
Nowletmemovetopartfour-----SimulationandResults(/rzlts/)Analysis
Here,youcanseeafigurewhichshowstherelationshipbetweentheaveragetransmit(/trnzmt/)powerofthesourceandtheachievablesecrecyratewithdifferentnumbersofrelays.
Inthisfigure,thebluelinerepresentstheconventionalrelayselectionschemeandtheredlinerepresentstheNN-basedscheme.
Inthisfigure,asthenumbersofrelayandtheaveragetransmitpowerofthesourceincreases,theachievablesecrecyrateincreasesaccordingly,.whichmeansincreasingthenumberofrelayscaneffectivelyimprovethesecretperformance.
Theredlinearealmost(/'lmst/)closetotheblueline,【whchindicatesthatourproposedscheme(i.e.theNN-basedscheme)achievesalmostthesamesecrecyratesasthoseoftheconventionalschemeforallvaluesof
】whichvalidateseffectiveness(/'fektvns/)ofourproposedschemes.
P16
Thistableshowsthethenormalized(/nrmlazd/)meansquare(/skwe/)error(NMSES)valuesofdirrerentrelaynodes.ThevalueofNMSEmeanstheperformancedifferencebetweentheconventionalschemeandourproposedscheme.ThevaluesofNMSEarebelow(/b'l/)negative('negtv/)20(
),whichvalidateseffectivenessofourproposedschemeagain.
P17
Now,letmemovetothelastpart-----Conclusion
Okay,nowwearegoingtotakealookatthelastpart-Conclusion.
P18
Wehavegotthefollowingconclusions.
First,Incomplex(conditions)scenarios,Neuralnetworkhaspromisingapplicationsinrelayselectionforsuperiorlearningability,computationspeedandprocessingcapacity.
Second,Comparedwiththeconventionalrelayselectionscheme,ourproposedschemeachievesalmostthesamesecrecyperformance.
Andlast,Ourproposedschemehasanadvantage(/d'vɑntd/)ofrelativelysmallfeedbackoverhead,indicatingthatproposedschemecanbeappliedtotheconditions(scenarios)wherethefeedbackislimited.
(Iftheconventionalschemeneedsfeedbackof
complexnumbers,NN-basedschemewillonlyneedfeedbackof
realnumbers.Therefore,thefeedbackoverheadofourproposedschemeishalf(/hɑ?
f/)ofthatoftheconventionalscheme,)
Q&A
1、计算复杂度
Computationalcomplexity
Thebiggestdrawbackisthehighlyselectioncomplexitieswithasmallnumberofrelaynodes.
Ifnumberofrelaynodeisbig,itwillhaveaadvantage.Thisneedourfurtherresearch.
Q:
TheexperimentshowsthatsecrecyrateisalmostthesameastraditionalmethodandwhatisthepromotionofusingNNtorelayselection.(whatismeaningofintroducingNNtorelayselection)
A:
Thatourproposedscheme(i.e.theNN-basedscheme)achievesalmostthesameachievablesecrecyrateasthatoftheconventionalschemeindicatesthatourproposedschemeiseffectiveanditcanselectoptimalrelaynodewhichobtainsmaximumachievablesecrecyrate.
Onereason(thefirstreason)isthatAdoptingNNforrelayselectionisanovelidea.
Anotherreasonisthatthespectrumresourceisrelativelimitedandourproposedschemehassmallfeedbackoverhead.
Q:
what’sthemeaningof“perfectsecrecyperformance”What’sthemeaningof“Comparedtotheconventionalrelayselectionscheme”
A:
“perfe