简介:Thispaperstudiesthetrackingperformanceofthesingle-inputsingle-output(SISO),finitedimensional,linearandtime-invariant(LTI)systemoveranadditivewhiteGaussiannoise(AWGN)channelwithfinitecontrolenergyandchannelinputenergyconstraint.Anewperformanceindexisproposedwhichisminimizedoverallstabilizingtwo-degree-of-freedomcontrollers.Theexplicitexpressionsofthelowerboundofthetrackingperformanceandtheminimumofsignal-to-noiserequiredareobtained.Theresultsshowthatthelowerboundiscorrelatedtotheunstablepole,nonminimumphasezeroandthechannelscalingfactor.Finally,oneexampleisgiventovalidatetheconclusionsbyadoptingthespecialinner-outerfactorization.
简介:我们现在的Fatman,企业规模档案的存储基于从under-utilized网服务者的志愿者贡献资源,通常与业余存储能力在几千上部署了节点。Fatman明确地为提高存在存储资源的利用并且砍倒硬件购买费用被设计。没有违背服务水平目的(SLO)并且没有减少档案的系统的可获得性,最小化费用,系统设计的二个主要担心的问题正在最大化志愿者节点的资源利用。Fatman广泛地越过几datacenters在几万个服务者节点上被部署了,提供超过100PB存储能力并且服务几十内部集体数据的申请。系统由强壮的隔离和预算限制认识到有效存储限额巩固,到没有任何降级,最大地在主人级的SLO上支持资源贡献。它使用新方法由把磁盘失败预言用于minish失败恢复花费的、命名差错知道的数据管理改进数据可靠性,戏剧性地在76.3%把吝啬的时间归结为修理(MTTR)并且减少真实产品工作量上的在35%的文件碰撞比率。
简介:InCAGDandCG,energymodelisoftenusedtocontrolthecurvesandsurfacesshape.Incurve/surfacemodeling,wecangetfaircurve/surfacebyminimizingtheenergyofcurve/surface.However,ourresearchindicatesthatinsomecaseswecan'tgetfaircurves/surfaceusingthecurrentenergymodel.Soanimprovedenergymodelispresentedinthispaper.Examplesarealsoincludedtoshowthatfaircurvescanbeobtainedusingtheimprovedenergymodel.
简介:Parameterizationoftrianglemeshesisafundamentalproblemfortexturemapping,surfacefitting,surfacereconstruction,andmeshediting.Thedeformationoftriangularmeshescausedbytheparameterizedprocessisthemeasurementofparameterization.Traditionalstandardmethodhasitslimitationwhenevaluatingmixturedistortionenergyparameterizations.Thusanevaluationmethodbasesondistortionenergyparameterizationoftriangularmeshesisintroducedforthelimitation.Thenovelmethodemploysanadaptiveexpressionformtothemixtureenergy,andusesaweightfactortorepresentdistortionenergydistribution.Byusingthismethod,wecanevaluateallkindsofparameterizationinauniformmeasurementandacquireamoreintuitiveandclearevaluation.
简介:Asatypicalerasurecodingchoice,Reed-Solomon(RS)codeshavesuchhighrepaircostthatthereisapenaltyforhighreliabilityandstorageefficiency,therebytheyarenotsuitableingeo-distributedstoragesystems.Wepresentanovelfamilyofconcurrentregenerationcodeswithlocalreconstruction(CRL)inthispaper.TheCRLcodesenjoythreebenefits.Firstly,theyareabletominimizethenetworkbandwidthfornoderepair.Secondly,theycanreducethenumberofaccessednodesbycalculatingparitiesfromasubsetofdatachunksandusinganimpliedparitychunk.Thirdly,theyarefasterthanexistingerasurecodesforreconstructioningeo-distributedstoragesystems.Inaddition,wedemonstratehowtheCRLcodesovercomethelimitationsoftheReed-Solomoncodes.Wealsoillustrateanalyticallythattheyareexcellentinthetrade-offbetweenchunklocalityandminimumdistance.Furthermore,wepresenttheoreticalanalysisincludinglatencyanalysisandreliabilityanalysisfortheCRLcodes.Byusingquantitycomparisons,weprovethatCRL(6,2,2)isonly0.657xofAzureLRC(6,2,2),wheretherearesixdatachunks,twoglobalparities,andtwolocalparities,andCRL(10,4,2)isonly0.656xofHDFS-Xorbas(10,4,2),wherethereare10datachunks,fourlocalparities,andtwoglobalparitiesrespectively,intermsofdatareconstructiontimes.OurexperimentalresultsshowtheperformanceofCRLbyconductingperformanceevaluationsinbothtwokindsofenvironments:1)itisatleast57.25%and66.85%morethanitscompetitorsintermsofencodinganddecodingthroughputsinmemory,and2)ithasatleast1.46xand1.21xhigherencodinganddecodingthroughputsthanitscompetitorsinJBOD(JustaBunchOfDisks).WealsoillustratethatCRLis28.79%and30.19%morethanLRConencodinganddecodingthroughputsinageo-distributedenvironment.
简介:ThispaperinvestigatestheasymptoticalstabilizationofHamiltoniancontrolsystemswithtimedelay.First,Hamiltoniancontrolsystemswithtimedelayareproposed.Second,atwo-to-one(TTO)principleisintroducedthattwodifferentHamiltonianfunctionsaresimultaneouslyenergy-shapingbyonedesiredenergyfunction.Third,anovelmatchingequationisbuiltviatheTTOprinciplefortheHamiltoniancontrolsystemswithtimedelay,whichgeneratesaneffectivecontrollawfortheHamiltoniancontrolsystemswithtimedelay.Finally,anumericalexampleshowstheeffectivenessofproposedmethod.
简介:Duetothelimitationofenergyresources,energyefficiencyisakeyissueinwirelesssensornetworks(WSNs).Clusteringisprovedtobeanimportantwaytorealizehierarchicaltopologycontrol,whichcanimprovethescalabilityandprolongthelifetimeofwirelesssensornetworks.Inthispaper,anenergy-drivenunequalclusteringprotocol(EDUC)forheterogeneouswirelesssensornetworksisproposed.EDUCincludesanunequalclusteringalgorithmandanenergy-drivenadaptiveclusterheadrotationmetho...
简介:Duetotheenergyandresourceconstraintsofawirelesssensornodeinawirelesssensornetwork(WSN),designofenergy-efficientmultipathroutingprotocolsisacrucialconcernforWSNapplications.Toprovidehigh-qualitymonitoringinformation,manyWSNapplicationsrequirehigh-ratedatatransmission.Multipathroutingprotocolsareoftenusedtoincreasethenetworktransmissionrateandthroughput.Althoughlarge-scaleWSNcanbesupportedbyhighbandwidthbackbonenetwork,theWSNremainsthebot...
简介:服务器的差的精力比例为现代数据中心的低精力效率被看作主要来源。我们发现一个应用程序的不同资源配置导致类似的性能,但是有不同精力消费。当表演等价物资源配置(PERC),和它的性能范围被称为相等的区域,我们叫这现象(嗯)。基于PERC,为改进精力效率的一个基本想法是为每应用从PERC选择最有效的配置。然而,当几千个应用程序在围住资源的服务器上同时被运行时,获得最佳的答案不能支持每个应用程序。这里,我们建议一个启发式的计划,CPicker,基于基因编程到改进服务者的精力效率。加快集中,CPicker由首先从有高精力的区域选择配置初始化一张高质量的人口变化。试验CPicker与神谕盒子相比与贪婪途径,和不到4%效率损失相比在17%精力效率改进上面获得的表演。
简介:在黑暗的硅时代,许多核心处理器的许多独立部件在一个薄片上由于电源消费的限制正在变得自愿地不活跃。然而,联网连接保留,在薄片上互联必须仍然被使激活并且浪费可观的精力避免这些不活跃的部件的隔离,伤害整个处理器芯片的精力比例。在这份报纸,我们建议一个新奇图案没有损坏网络连接,提供更多的精力比例的在薄片上连接。达到这个目标,我们重新设计路由器建筑学。新建筑学,DimRouter,支持三个模式:正常、黑暗、暗淡。在暗淡模式,仅仅路由器的部分是活跃的并且当黑暗模式把所有路由器元素放处于睡着了的状态时,提供灵活连接。而且,最大化黑暗路由器的数字,我们也基于抑制度的Steiner建议一个重构算法树。在合成交通下面的评估结果证明新设计能与普通设计相比减少精力消费直到85%。为真实应用程序交通,新设计能也与4%性能改进保存一般水准46%精力消费。
简介:为了避免处理零动力学限制,通过紧接的变换器由双向力量流动引进了一个二倍地喂的风发电机,一为格子方面变换器的新基于精力的建模和控制途径被介绍。在当模特儿的进程期间,格子方面变换器与反馈互联结构被划分成二个分系统,并且模型的交互矩阵考虑具体端口结构。然后,一个基于精力的控制器被建议基于新模型认识到格子方面控制目的。模拟研究在MATLAB/Simulink被执行。在比例积分的控制器和基于精力的控制器之间的比较结果证明一个人能更快获得的后者负担水流变化的集中率和全球稳定性。而且,当与随机的风在2MW风精力变换系统模仿了时,基于精力的控制器为格子方面控制也是能干的。