邊緣運算Part 1：邊緣運算技術全解析 - 淺談股海

以簡單的方式講解邊緣運算，以及比較雲端運算與邊緣運算的差異，同時了解邊緣運算未來的 ... 邊緣運算是一種計算架構，透過分散式的方式將運算位置從過去中心移至邊緣 ...  ​EdgecomputingPublished:2019 Author:D.Y.Viewdetailedstockprice​OperatesontheedgeEdgeComputingPart1 Fullanalysisofedgecomputingtechnology​Edgecomputingisacomputingarchitecturethatmovescomputinglocationsfromthecenterofthepasttoedgenodesinadistributedmanner.Edgenodesaredevicesthatareclosertousers.Theedgecomputingapplicationsthatmostexpertsnowrecognizearenotadirectreplacement.Cloudcomputingisdoneincooperationbetweenedgecomputingandcloudcomputing.Theedgeisresponsibleforpreliminarydataprocessinganddetermineswhetheritneedstoresponddirectlyorfurtherhandedovertothecloud.Thecloudisresponsiblefordatastorageandhigherperformanceandlowtimeliness.Computing,suchastheimprovementofmachinelearning.IntheeraoftheInternetofThings,wehavemoredevicesthatneedtorespondinatimelymanner,andalargeramountofinformationneedstobeprocessed.Therefore,edgecomputinghasemergedtoachievelowlatency,reducetransmissioncosts,andreducecloudload.CurrentpracticalapplicationssuchasAmazon Continuetodevelopthechipbusiness,hopingtotransfersomeofthefunctionsofitsownsmartspeakerEchototerminalequipmentinthefuture.​​Cloudcomputing​Edgecomputing​Cloudcomputingvs.edgecomputing​Thecloudhasgoodstorageandcomputingfunctions,butthetransmissionrateisrelativelyinadequate.Sincedatamustbetransmittedfromtheterminaldevicetotheclouddatacenteraftermultiplelayersoftransmission,itismoretime-consumingthanedgenodecomputing.Second,theInternetofThingsdeviceswillincreasesignificantlyinthefuture.AccordingtoIBM'sstatistics,therewereapproximately15billionInternetofThingsdevicesworldwidein2018,whichisestimatedtodoubleto55billionby2022,andtheamountofdatacollectedwillalsoincreasedramatically.Ifyoustillrelyonlyoncloudcomputing,itwillcostalotofdatatomeetthescaleoftheInternetofThings,andthecloudwillalsobeoverloaded.Edgecomputingthereforehastheadvantagesofreducinglatencyandsharingthehugeamountofdata.However,therearestillsomeproblemsthatneedtobeovercomeinedgecomputing.JustlikeIoTdevices,informationsecuritywillbeabigproblem.Comparedwiththepastcloud,informationsecuritycanbeuniformlyimprovedandmaintainedbythedatacenter,butwhentherearemorecomputingnodesAtthistime,therewillbemoreopportunitiestotakeadvantageoftheemptiness,whichisalsothepartthatlargermanufacturersneedtoactivelyovercome.​Edgenodeintroduction​Thereisnounifiedplacefortheinstallationofedgecomputing.Therearetwomaintypes:​1.EdgeServer:​UsingBladeSeverorindustrialcomputersasedgecomputingnodedevices,andinstallingthesedevicesclosetotheterminaldevices,thisapproachiscurrentlypraisedbymanycompanies,becausethismethodhasseveraladvantagesincludingIntegratethedataofvariousdevices,cross-analyze,andeasytoexpand.Somepeopleevenproposetoinstallcomputingequipmentonmobiletelecommunicationsbasestationstoensurethepopularityofedgecomputing.Theseedgecomputingdevicesinstalledonmobiletelecommunicationsbasestationscanbeusedforreal-timejudgmentandanalysiswhentheyarenotdrivingontheroad.​​2.Terminalequipment:​SomepeoplehavealsoproposedthatcomplexAIcalculationscanbeperformeddirectlyintheterminalIoTdevice,whichmaybeself-drivingorvariousmachines,butthismethodisnotgoodforexpansionanddataintegration,butitcansaveinstallationcosts.​​Applicationfield​Inthefuture,edgecomputingwillbeusedinvariousIoTspaces,anditwillalsobeamajortechnologythatpromotesthepopularizationofIoT,especiallyinenvironmentsthatrequiretimelyresponseorhavealargeamountofdatatobeprocessed.Thefollowingisabrieflistofthemainapplicationareas:1.Smartfactory:​Itistheearliestapplicationfieldofedgecomputing.Throughedgecomputing,itcanquicklydeterminewhethervarioussemi-finishedproductsmeetthestandardsineachstep,andeffectivelysupervisetheproductionprocess,whilepredictingproductionqualityandspeed. 2.Smartcity:​Therewillbealargeamountofdatainthefuturecities,soedgecomputingwillberequiredforpreliminaryandtimelyprocessing,suchastheuseofedgecomputingtoimprovecitysafety,camerareal-timeimagerecognition,andtheimplementationofsmartbuildings. 3.Self-driving:​Autonomousdrivinghasbecomeextremelyimportantbecauseoftherequirementsoftimelyresponse,anditisalsothefieldthatmostexpertsthinkthattheyhavetousewhentheymentionedgecomputing.Inaddition,inordertojudgethesurroundingroadconditions,self-drivingcarswillinstallalargenumberofsensorsaroundthecartocollectalargenumberofsurroundingenvironmentaldataimages.Theprocessingofhugedatamakesedgecomputingextremelyimportant,whichshowsthatedgecomputingisveryimportantforself-drivingcars.Thepromotionofwillbecomeindispensable.​4.LogisticsCenter:​Thedailyvolumeofgoodshandledbythelogisticscenterisincreasing.Inthepastfewyears,automationhasbeencontinuouslypromotedtoreducelaborcosts.Inthefuture,edgecomputingcanbeusedinconveyorbeltmanagement,goodsdistribution,andsortingmachinestospeedupthepickingspeed..​​Marketsize EdgecomputingisamethodtosolvedatatransmissionandtoovercomethehugedataprocessingneedsoftheInternetofEverythinginthefuture.IBMestimatesthattherewillbe15billionInternetofThingsdevicesin2019,whichwillincreasebymorethanthreeyearsto2022.Tripleto55billion.Themarketsizeofedgecomputingisestimatedtobe1.272billionU.S.dollarsin2018and6.959billionU.S.dollarsin2024.​55billionin2022 NumberofIoTdevices15billionin20196.959billion(USD)in2024 EdgeComputingOpportunity20181.272billion(USD)​​relatedarticles​EdgeComputingPart2:IntegrationofEdgeComputingConcepts​Edgecomputingrelatedstocks