conditionsyoutubemp3mp4

IEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015Abstract—Inrecentyearstherehasbeenanexponentialincreaseinthegrowthinmultimediaapplications,andinparticularinvideoapplications.
Understandingthebehaviorofthevideotrafficandtherequirementsforthenetworkhelpsnetworkadministratorstoimprovethetraffic.
Inthiswork,aquantitativeanalysisisperformedbyexperimentation,inordertoevaluatethebehaviorandimpactofvideotrafficonWANnetworks.
WeproposeaWANtestbedcomposedbyavideotrafficserverandseveralclientstations.
Thisarticleintroducesascenariothatallowstoinjectmulticastvideotraffic,compressedwithseveralcodecs.
Fromcapturingvideotraffic,weidentifiedseveralinterestingperformancemetrics,suchasmulticastthroughput,interframespaceandframesizedistributions,andthenumberofframes.
Weincludedetailedcontributionsontheimpactproducedbyseveralfactors,suchastheconfigurationoftheresolutionofthevideo,thevideoclass,thecodecusedforthecompression,andtheuseofmulticasttrafficwhentherearerestrictionsonthebandwidth,asinacorporaterealWANlinkofsomefewMbps.
Thisstudyfacilitatesthecomparisonoftheresultswiththoseobtainedfromanalyticalstudiesandmodellingfordifferentcontexts.
IndexTerms——multicasttraffic,videocodec,WANtestbed,I.
INTRODUCTIONODAY,itispossibletoreceivedigitalTVwithhighdefinitionservicesandagreaternumberofchannels.
Recordingandsharingvideoswithmobilephoneshasalsobeenwidespread.
LargecompaniesandorganizationsareusingvideoconferencingapplicationssuchasCiscoTelePresenceandWebEx,fortheface-to-facecollaborationindifferentgeographicalregions,andevenwithinthesamecompany.
Consumersareincreasinglydemandingintermsofthequalityandperformanceofthevideo-basedproducts,andtherefore,thereisastrongincentiveforcontinuousimprovementinmultimediatechnologies.
Accordingtoaclassificationproposedin[1],thevideoS.
PéreziswiththeNationalTechnologicalUniversity,Mendoza,5500Argentine(e-mail:santiagocp@frm.
utn.
edu.
ar).
H.
FacchiniiswiththeNationalTechnologicalUniversity,Mendoza,5500Argentine(e-mail:higinifac@frm.
utn.
edu.
ar).
J.
CamposiswiththeZaragozaUniversity,Zaragoza,50018Spain(e-mail:jcampos@unizar.
es).
C.
TaffernaberryiswiththeNationalTechnologicalUniversity,Mendoza,5500Argentine(e-mail:carlos.
taffe@gmail.
com).
F.
HidalgoiswiththeNationalTechnologicalUniversity,Mendoza,5500Argentine(e-mail:fabianhdlg@gmail.
com).
S.
MéndeziswiththeNationalTechnologicalUniversity,Mendoza,5500Argentine(e-mail:se.
sebastian@gmail.
com).
trafficcanbe(amongothers):IPTVBroadcast,Livetransmissionofvideoevent(webcast),IPvideosurveillance,Interactivevideoconference,andVideoondemand.
Thesetypesoftrafficillustratethevariablesthatneedtobequantifiedforanydeploymentofvideoandmultimedia:directionality,throughput,latencyandjittertolerance,aswellasthenumberofchannelsandusers.
Anotherkeymetricisthetoleranceforerror.
Theprincipalcontributionsofthisarticleare:i)tospecifyandexperimentonanewWANtestbedasscenario,ii)toquantifymetricsoverthestandardmulticastmechanism,andiii)thecontributionsandconclusionsmadeusingaWANtestbedwithrealequipment.
Therestofthisdocumentisstructuredasfollows.
SectionIIintroducessomebackgroundandrelatedwork.
SectionIIIprovidesageneralviewofvideocodecs.
SectionIVdescribeshardwareandsoftwareresourcesoftheWANtestbed.
SectionVpresentstheresultsobtainedfortheperformancemetricsandthestatisticaldistributions.
AndsectionVIsummarizesthemostsignificantconclusions.
II.
BACKGROUNDANDRELATEDWORKThedataservicesareontheirwaytomeettheneedsofmanyusersinthenetwork.
Thevoiceservicesarealreadyconsideredasanecessitybymostpeople.
Andthedataservices,video,andTVarerapidlybecominganessentialpartofthedailylifeofconsumers.
Itisexpectedforthe2018that65%ofthetotalwillbeassociatedwithvideoapplications[2].
VariousapplicationsmakeuseofgrowingvideotrafficintheLANandWANnetworks.
EachoneofthemhasitsownspecialcharacteristicsanddemandstoensureanadequatelevelofQoS.
Thereisalargeamountofexperimentalworkandsimulationcarriedoutthatexhibitsthebehaviorofeachcase,fromananalysisofthecaptureoftraffic.
Forexample,in[3]theauthorsevaluatetheperformanceofthreestateoftheartvideocodecsonsyntheticvideos.
Anextensivenumberofexperimentsareconductedtostudytheeffectofframerateandresolutiononcodec'sperformanceforsyntheticvideos.
Whilein[4]acomparativeassessmentispresentedforthetwovideocodingstandards:H.
265/MPEG-HEVC(High-EfficiencyVideoCoding),H.
264/MPEG-AVC(AdvancedVideoCoding),andalsooftheVP9proprietaryvideocodingschemeusinganexperimentaltestbed.
In[5]theimpactoftheH.
264videocodeconthematchperformanceofautomatedfacerecognitioninsurveillanceandmobilevideoapplicationsisassessed.
OtherworksevaluatedthebehaviorwithanS.
Pérez,H.
Facchini,J.
Campos,C.
Taffernaberry,F.
Hidalgo,S.
MéndezBehaviorofCodecsforMulticastVideoTrafficusingWANTestBedTIEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015experimentalstudyofmultimediatrafficperformanceinmeshnetworks,forperformanceevaluationandanalysisofwirelessnetworks[6].
Intheseandotherresearcharticles[2-18]therearefewpapersavailableonsimulationandexperimentalstudiesonWANnetworks.
Andtotheknowledgeoftheauthors,therearenoproposalsthatcombine,inaWANtestbed,withconstraintsoftheselinks,theproblematicofthevariantsofthecodecsforvideotraffic,withtheuseofmulticastflow.
Thisnewexperimentationscenarioallowedustoobtaintherequireddetailofseveralperformancemetricsforthiscontext.
Ourscenarioassumescertainconditions,suchasforexample,thenon-existenceofvoicetrafficorgeneraltraffic(both,usually,inaverylowproportionwithrespecttothevideotraffic),inordertofacilitatethecomparisonandimpactofdifferentcodecsonthemulticasttraffic.
III.
BRIEFDESCRIPTIONOFCODECSCodecisanacronymforcoder-decoder.
Describesaspecificationdevelopedinsoftware,hardware,oracombinationofboth,abletotransformafilewithadatastreamorasignal.
Codecscanencodethefloworthesignalfortransmission,storageorencryption,andrecoverordecryptthesamewayforthereproductionormanipulationinaformatmoresuitablefortheseoperations.
Fig.
1showsapartialprogressionoftherecommendationsofITU(InternationalCommunicationUnion)andMPEG(MovingPicturesExpertsGroup)standards.
Inourcasestudywehaveworkedwiththefollowing3codecs:A.
H.
264/MPEG-4AVCH.
264/MPEG-4AVCprovidesasignificantadvancementintheefficiencyofcompressiontoachieveareductionofaround2timesinthebitratecomparedtoMPEG-2andMPEG-4simpleprofile.
IntheformaltestingbytheJVT(ITU),H.
264gaveanimprovementoftheefficiencyof1.
5xorhigherin78%ofthecasesand77%inthosewhoshowedimprovement2xorgreaterandupto4xinsomecases.
The2ximprovementallowedH.
264thecreationofnewmarketopportunities,suchas:VideoVHS-quality600Kbps.
ThiscanenablethedeliveryofvideoondemandviaADSLlines.
Providesexcellentclarityfortheprofilescoveredbyextensionsofrangeoffidelitywhichextendsthelevelsto"loseless"orveryclosetothisandsupportschroma4:4:4andbitdepthofupto12.
MP4-AVCismoreefficientthan"VisualCoding"(part2),MP4-AVCprovidesbetterqualityatthesamesamplingrateorequalqualityatlowestrates.
Fig.
1.
ViewoftheprogressionofITUandMPEGstandards.
B.
H.
263/MPEG-4PART2MPEG4Part2calledMPEG4VISUAL,belongstothefamilyofstandardsMPEG-4ISO/IEC.
Thereareseveralimplementationsofthisstandard,beingDIVX,Xvid,NeroDigitalthemostpopular.
MPEG4VISUALwasputonthemarketwithafamilyofconfigurationscalledparties.
Wearedealingwithpart2whichsupportsthreeprofiles,theseare:SimpleProfile,AdvancedSimpleProfileandAdvancedStudioProfile.
ASPwastheprofileusedinourtests,thisallowstheuseofthefollowingtypesofvisualobjects:Simple(rectangularvideothatusesframesintraandpredicted)andSimpleAdvanced(rectangularvideo,improvedcompressionandbidirectionalframes).
Sixcompressionlevelsareallowed(0to5).
Levels0to3havedataratesfrom128to768kbps,levels4and5addedinterlacedencodingtoachieverateof3to8Mbps.
MPEG4VISUALhasagoodsupportformovingimage.
Concerningtheclarity,itplacesfrommoderatetoverygoodtakingintoaccountthatthesamplingislimitedto4:2:0andthatMP4isalossycompressionformat.
Bothvideointerleavingandprogressivearesupported.
C.
MPEG-2MPEG-2waspublishedasISOstandard13818.
MPEG-2istypicallyusedtoencodeaudioandvideoforbroadcastingsignals,includingdigitalterrestrialTV,satelliteorcable.
MPEG-2withsomemodifications,isalsothecodingformatusedbycommercialdiscsandDVDSVCDmovies.
MPEG-2issimilartoMPEG-1,butalsoprovidessupportforinterlacedvideo(theformatusedbytheTV).
MPEG-2introducesanddefinesTransportStreams,whicharedesignedtocarrydigitalvideoandaudiothroughunpredictableandunstableenvironments,andareusedintelevisionbroadcasts.
Withsomeenhancements,MPEG-2isalsothecurrentstandardforHDTVbroadcasts.
IV.
EXPERIMENTATIONSCENARIOANDRESOURCESInthepresentworkweexperimentallyevaluatedtheperformanceofstreamingofvideostoredonaWANtestbed.
A.
WorkingTopologyFig.
2showstheworkingtopology.
ItusesaPCasserverand6PC'sasclients.
InthistopologythelinksindicatedwithcontinuouslineareoftypeFastEthernetwithatransmissionrateof100Mbps,whilethelinksindicatedwithdashedlinearededicatedwithserialinterfacestoatransmissionrateof2Mbps.
FortheoperationoftheOSPFroutingprotocolv2betweenroutersR1toR6isconfigured.
Forthecaseofmulticast,routingprotocolPIMdensemodeforthesamerouterswasconfigured.
B.
HardwareResourcesofExperimentationForthescenarioshowninthetopologyofworkweuseddesktopcomputerswiththefollowingfeatures:Processor:AMDAthlon(tm)IIX2250at3GHzwith2GBofRAM,andOperatingSystem:Windows7Professional32-bit.
RoutersR1,R2,R3andR4wereCiscoModel2811,andRoutersR5andR6wereCiscoMultilayerModelWS-CS3750switchs.
IEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015Fig.
2.
ViewofthetopologyoftheWANtestbed.
FinallyfortheconnectionoftherouterstothePCsCiscoLayer2CatalystModelWS-2950-24switchswereused.
C.
SoftwareResourcesofExperimentationFortheexperimentaldevelopmentUnrealMediaServer[19]v.
11.
0wasused,andmediaplayerv.
6.
1asstreamingclientsoftware.
Itisamulti-protocol,highperformanceandsmallresourcesfootprintsoftwareplatformforstreamingliveandondemandaudiovideocontentoverIPnetworks.
ItstreamswithvarietyofstreamingprotocolstodelivercontenttoFlashPlayer,Silverlight,WindowsMediaPlayer,UnrealMediaPlayer,mobiledevicesandSet-Topboxes.
TheserversupportsUMS(proprietary,DirectShow-based,codec-independent)protocolforstreamingtoUnrealMediaPlayerinunicastandmulticastmodes,andanymultimediafileformat,encodedwithanycodec.
Supportedcontainerformatsincludebutarenotlimitedto:MP4,ASF,AVI,MKV,MPEG,WMV,FLV,OGG,MP3,3GP,MOV.
UnrealLiveServersupportsanypossiblecapturedeviceattachedtoaPC.
CapturingnetworkstreamsoverRTSP,RTMP,MPEG2-TS,HLSandMMSprotocolsissupportedaswell.
UnrealLiveServerencodes/transcodescapturedaudio-videowithH.
264,VC1,AAC,MP3,WMAcodecsandstreamsitoverUMSprotocoltoUnrealMediaServer.
D.
CODECsforExperimentationWeusedthreedifferentcodecs,thosedescribedinsectionIII,andacommercialvideo[20]of29swiththeconfigurationsgiveninTablesI,IIandIII.
V.
EVALUATINGTHEEXPERIMENTATIONSCENARIOA.
AnalysisofthePerformanceMetricsUsingthetestbeddetailedintheprevioussection,aseriesofexperimentswereperformedtransmittingavideoencodedwithMPEG-4/AVC,MPEG-4orMPEG-2,fromtheservertothenetworkinmulticastformattosixPCsdistributedinthenetworkaccordingtothetopologyofFig.
2.
TABLEIPARAMETERSETTINGOFH.
264/MPEG-4AVCTABLEIIPARAMETERSETTINGOFMPEG-4VISUALMeasurementsweretakenwithWireshark[21]sniffersoftwareontheserverandoneachclientPC,toobtainthefollowingmetrics:totaltimeofthevideo,totalnumberofpackets(oframes),totalnumberofbytes,averagesizeofpackets,averageinterframetimeofpacketsandeffectivetransferrateofeachcodec.
IEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015TABLEIIIPARAMETERSETTINGOFMPEG-2TableIVsummarizesthemeasurementsthatwerecaptured.
ThetimevideohasminimaldifferencesbetweenplayingindifferentPCsasaccordingtothedifferentcodecs.
Thetotalnumberofpackets,transmittedbytheserverandreceivedineachPC,arevirtuallythesameforacodecinparticularanddifferentPCsasshowninFig.
3.
Adifferencehoweverisnoticeableinthenumberofpacketsfordifferentcodecs,whenrepresentingtheaverageofallPCsforthesamecodec.
MPEG-4Visualistheonethatusedthemostnumberofpackets,withlittledifferenceonMPEG-4AVC;whileforMPEG-2thenumberofpacketsismuchlower(33.
54%less).
Concerningthetotalnumberofbytes,transmittedbytheserverandreceivedineachPC,theyarevirtuallythesameforacodecinparticularanddifferentPCsasshowninFigure4.
Asinthecaseofthetotalnumberofpackets,MPEG-4AVCistheonewhichtransmittedmorenumberofbytes(orusedalargerfilesize),withlittledifferencewithMPEG-2;whileforMPEG-4Visual,thenumberofbytesismuchlower(45.
75%less).
WithregardtotheaveragepacketsizetransmittedbytheserverandreceivedineachPCitisnotedthattheyarevirtuallythesameforacodecinparticularanddifferentPCsasshowninFig.
5.
Onthecontrary,fortheaveragesizeofpacketsbetweendifferentcodecMPEG-4Visualhasthesmallestsize(63.
58%lessthanMPG2).
AnotherimportantparameteristheaverageinterframetimeofpacketstransmittedandreceivedbytheserverineachPC.
TheyarequitesimilarforacodecinparticularanddifferentPCs,asshowninFig.
6.
Howeverithighlightsadifferenceinaverageinterframetimesbetweendifferentcodecs.
TheaverageofallPCsforMPEG-2ishighercomparedtoothercodecs.
MPEG-4AVChastheshortestinterframetime(alower30.
24%).
TABLEIVSUMMARYOFTHEMEASUREMENTSFig.
3.
PacketnumberforPC.
Fig.
4.
VideoMbytesforPC.
Fig.
5.
Averagepacketsize(bytes)porPC.
IEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015FinallyweanalyzedtheaveragetransmissionratebetweentheserverandeachPC,andnotedthattheyarevirtuallythesameforacodecinparticular,asshowninFig.
7.
Butwhencomparingaveragespeedsbetweendifferentcodec,MPEG-4Visualistheonewiththeleastbitraterespecttotheothervideos(oneless46.
29%).
B.
AnalysisoftheStatisticalDistributionsWealsoanalyzedthestatisticaldistributionsofframesizesandofinterframespacesforeachcodec,ofthevideomulticasttrafficononeofthestations.
TheFig.
8and9showthedistributionofthe906packetsgroupedbysizeandinterframespacesforMPEG-4AVC,respectively.
IntheFig.
8,asignificantnumberofpacketshasalengthlessthan500bytes(35%)andothershave1442bytes(18%).
Therestofpacketsaredistributedinthedifferentlengthsbetween500to1442bytes.
AndintheFig.
9,thehighestnumberofpacketsisbelow5ms(50%)andbetween68and78ms(32%)ofinterframespaces.
TheFig.
10and11showthedistributionofthe913packetsgroupedbysizeandinterframespacesforMPEG-4Visual,respectively.
IntheFig.
10,amoreimportantnumberofpacketsarebetween200and300bytes(71%)andothergroupbetween700and1000bytes(20%).
AndintheFig.
11,thehighestnumberofpackets(39%)has50msofinterframespace.
Othergrouphasalowerinterframespaceof10ms(21%)andbetween25and35ms(30%).
Finally,Fig.
12and13showthedistributionofthe609packetsgroupedbysizeandinterframespacesforMPEG2,respectively.
IntheFig.
12,thehighestnumberofpacketshas1442bytes(60%).
Therestofthepacketsaredistributedinthedifferentlengthsbetween60to1420bytes.
AndintheFig.
13,thehighestnumberofpacketsisbelow5ms(60%)andintheremainingbetween90and125ms(40%)ofinterframespaces.
Fig.
6.
Averageinterframetime(s)forPC.
Fig.
7.
Bitrate(Mbps)forPC.
Fig.
8.
DistributionofthepacketsgroupedbysizeMPEG-4/AVC.
Fig.
9.
DistributionofthepacketsgroupedbyinterframespaceMPEG-4AVC.
Fig.
10.
DistributionofthepacketsgroupedbysizeMPEG-4/V.
VI.
CONCLUSIONANDFUTUREWORKThisstudyusedanewWANtestbedasscenarioofferingmetricsevaluationwhileusingtypicalvideocodecs.
MetricswereexhaustivelyanalysedformulticastnetworktrafficinlinewiththeexpectationsofarealWAN.
Directmetricsandtheiraverages,andthestatisticaldistributionswerequantifiedoverrealequipment.
ObtainedfiguresshowthatthemulticasttrafficprovidestheQoSandtheperformancethatisexpectedovereachstationusingdifferenttypesofcodecs.
Differencesinthebehaviorofthemulticasttrafficaregivenbydifferencesbetweencodecs,andnotbythemulticasttrafficinitself.
IEEECHILECON2015-ISSNxxxx-xxxx,Pág.
xxxx.
xxx.
SantiagodeChile,28al30deOctubre2015Fig.
11.
DistributionofthepacketsgroupedbyinterframespaceMPEG-4/V.
Statisticalanalysisforthisscenarioshowsthatdifferentcodecsdisplayadifferentbehaviourinthedistributionofpacketlengthsandofinterframespaces.
TheimpactontheoveralltrafficofaWANlinkdependsonthecodecused,itssettingandintheresultingmixturewithotherstraffics.
Resultscomplement,enrichandcanbeusedforcomparisonwithotheranalyticalandofsimulationstudiesovervideotraffic.
Weforeseefuturestudiesofferingaquantitativebehaviourevaluationatdifferent802.
11physicallayers.
ThesestudieswouldpreciselydeterminethebestgeneralnetworkbehaviourforhigherWi-Fivelocities.
Finally,anewlineofstudycouldbedevelopedregardingtheimpactofqueuelengthonmaximumthroughputforeachcodec.
Fig.
12.
DistributionofthepacketsgroupedbysizeMPEG2.
Fig.
13.
DistributionofthepacketsgroupedbyinterframespaceMPEG2.
REFERENCES[1]Available:http://www.
arubanetworks.
com/.
[2]CiscoSystema,"CiscoVisualNetworkingIndex:GlobalMobileDataTrafficForecastUpdate,"2013-18,Available:http://www.
cisco.
com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white_paper_c11-520862.
html.
[3]A.
Abdalla,A.
Mazhar,M.
SalahandSaharKhalaf,"ComparativeStudyofCompressionTechniquesforSyntheticVideos,"TheInternationalJournalofMultimedia&ItsApplications(IJMA),vol.
6,no.
2,April2014.
[4]D.
Grois,D.
Marpea,T.
NguyenaandO.
Hadarb,"ComparativeAssessmentofH.
265/MPEG-HEVC,VP9,andH.
264/MPEG-AVCEncodersforLow-DelayVideoApplications,"SPIEProceedings,vol.
9217,ApplicationsofDigitalImageProcessingXXXVII,California,USA,September,2014.
[5]S.
Acharya,B.
Smith,"AnExperimenttoCharacterizeVideosStoredOnTheWeb,"ProceedingsofMultimediaComputingandNetworking,pp.
166-178,California,USA,1998.
[6]Y.
Sun,I.
Sheriff,E.
Belding-RoyerandK.
Almeroth,"AnExperimentalStudyofMultimediaTrafficPerformanceinMeshNetworks,"WorkshoponWirelessTrafficMeasurementsandModeling,pp.
25-30,Seattle,Washington,USA,June,2005.
[7]R.
AdduandV.
Potuvardanam,"EffectofCodecPerformanceonVideoQoEforvideosencodedwithXvid,H.
264andWebM/VP8,"MasterThesis,BlekingeInstituteofTechnology,Karlskrona,Sweden,August2014.
[8]S.
Tanwir,H.
PerrosandB.
Anjum,"AQoSevaluationofvideotrafficmodelsforH.
264AVCvideo,"FifthInternationalConferenceonNextGenerationNetworksandServices(NGNS),pp.
313-320,Casablanca,Morocco,May,2014.
[9]Y.
Shin,M.
Choi,J.
Koo,Y.
Kim,J.
IhmandS.
Choi,"EmpiricalAnalysisofVideoMulticastoverWiFi,"ThirdInternationalConferenceonUbiquitousandFutureNetworks(ICUFN),pp.
381-386,China,2011.
[10]S.
Pérez,H.
Facchini,A.
Dantiacq,G.
CangemiandJ.
Campos,"AnEvaluationofQoSforintensivevideotrafficover802.
11eWLANs-Quantitativeanalysis,"InternationalConferenceonElectronics,CommunicationsandComputers(CONIELECOMP),pp.
8-15,Cholula,Mexico,february,2015.
[11]S.
TanwirandH.
Perros,"AsurveyofVBRvideotrafficmodels,"IEEECommunicationsSurveysandTutorials,vol.
15,no.
4,pp.
1778-1802,2013.
[12]A.
LazarisandP.
Koutsakis,"ModelingmultiplexedtrafficfromH.
264/AVCvideoconferencestreams,"ComputerCommunications,vol.
33,no.
10,pp.
1235-1242,2010.
[13]M.
KrunzandS.
K.
Tripathi,"OnthecharacterizationofVBRMPEGstreams,"inSIGMETRICS,1997,pp.
192-202.
[14]U.
K.
Sarkar,S.
Ramakrishnan,andD.
Sarkar,"Modelingfull-lengthvideousingMarkov-modulatedGamma-basedframework,"IEEE/ACMTransactionsonNetworking,vol.
11,no.
4,pp.
638-649,2003.
[15]M.
Dai,Y.
Zhang,andD.
Loguinov,"AunifiedtrafficmodelforMPEG-4andH.
264videotraces,"IEEETransactionsonMultimedia,vol.
11,no.
5,pp.
1010-1023,August2009.
[16]S.
MaandC.
Ji,"Modelingheterogeneousnetworktrafficinwaveletdomain,"IEEE/ACMTransactionsonNetworking,vol.
9,no.
5,pp.
634-649,October2001.
[17]G.
V.
derAuwera,P.
T.
David,andM.
Reisslein,"Trafficandqualitycharacterizationofsingle-layervideostreamsencodedwithH.
264/MPEG-4advancedvideocodingstandardandscalablevideocodingextension,"IEEETransactionsonBroadcasting,vol.
54,no.
3,pp.
698-718,September2008.
[18]P.
Seeling,M.
Reisslein,andB.
Kulapala,"Networkperformanceevaluationwithframesizeandqualitytracesofsingle-layerandtwo-layervideo:Atutorial,"IEEECommunicationsSurveysandTutorials,vol.
6,no.
3,pp.
58-78,2004.
[19]UnrealStreamingTechnologies,UnrealMediaServer,Losngeles,USA,Available:http://umediaserver.
net/.
[20]Available:https://www.
youtube.
com/watchv=pNxOGqextv4.
[21]Available:https://www.
wireshark.
org/.