简介:AnovelalgorithmcalledColonyLocationAlgorithm(CLA)isproposed.Itmimicsthephenomenainbioticconmunitythatcoloniesofspeciescouldbelocatedintheplacesmostsuitabletotheirgrowth.Thefactorsworkingonthespecieslocationsuchasthenutrientofsoil,resourcecompetitionbetweenspecies,growthanddeclineprocess,andeffectonenvironmentwereconsideredinCLAviathenutrientfunction,growthanddeclinerates,environmentevaluationandfertilizationstrategy.CLAwasappliedtosolvetheclassicalassignmentproblems.ThecomputationresultsshowthatCLAcanachievetheoptimalsolutionwithhigherpossibilityandshorterrunningtime.
简介:录像加速器在分享象YouTube那样的网站的录像为更好的用户经验被开发。PPLive录像加速器(PPVA)在中国有用户的最大的数字,基于对等(P2P)系统。在PPVA的录像和同伴的数字由比哪个多次大的数量级在需求(VoD)上的传统的P2P录像系统。作为结果,尽管资源是足够的,由于不公平赋值,服务的质量几乎不能满足所有用户。在这份报纸,我们在PPVA专注于基本资源的任务:存储和带宽。存储赋值的问题作为一个随机的变量关于请求的数字作为一个非线性的程序(NLP)被提出。当平均数比变化大得多时,请求的变化的影响不是可以忽略、比例的途径的结果表演仅仅是适当的。关于怎么定位录像到适当同伴的标准也被介绍,考虑抑制例如全部的带宽,带宽竞争的概率和在录像之间的公平的利用。而且,分配的启发式的算法以集中、分布式的时尚上载带宽被建议并且在录像之中关于平衡与广泛地使用的策略(相等的分配)作比较。模拟结果证明两个算法能导致重要表演改进。
简介:
简介:Basedonthewell-knownLeverrieralgorithm,asimpleexplicitsolutiontorightfactorizationofalinearsystemisestablished.Thissolutionisexpressedbythecontrollabilitymatrixofthegivensystemandasymmetricoperatormatrix.ApplicationsofthissolutiontoatypeofgeneralizedSylvestermatrixequationsandtheproblemofparametriceigenstructureassignmentbystatefeedbackareinvestigated,andgeneralcompleteparametricsolutionstothesetwoproblemsarededuced.Thesenewsolutionsaresimple,andpossessdesirablestructuralpropertieswhichrenderthesolutionsreadilyimplementable.Anexampledemonstratestheeffectoftheproposedresults.
简介:Thepoleassignmentinaspecifieddiskbystatefeedbackforuncertaindelta-operatorsystemsisstudied.BymakinguseofalgebraRiccatiequations,asufficientandnecessaryconditionofpoleassignmentforakindofparameteruncertaindelta-operatorsysteminaspecifieddiskbystatefeedbackispresented.Andthedesignmethodofstatefeedbackcontrollerisalsodeveloped.Theproposedmethodcanunifysomepreviousrelatedresultsofcontinuousanddiscretetimesystemsintothedeltaframework.Theefficiencyofthedesignmethodisillustratedbyanumericalexample.
简介:ThewavelengthassignmentwithmultiplemulticastrequestsinfixedroutingWDMnetworkisstudied.Anewmulticastdynamicwavelengthassignmentalgorithmispresentedbasedonmatchingdegree.First,thewavelengthmatchingdegreebetweenavailablewavelengthsandmulticastroutingtreesisintroducedintothealgorithm.Then,thewavelengthassignmentistranslatedintothemaximumweightmatchinginbipartitegraph,andthismatchingproblemissolvedbyusinganextendedKuhn-Munkresalgorithm.Thesimulationresultsprovethattheoveralloptimalwavelengthassignmentschemeisobtainedinpolynomialtime.Atthesametime,theproposedalgorithmcanreducetheconnectingblockingprobabilityandimprovethesystemresourceutilization.
简介:InordertoovercometheshortcomingoftheclassicalHungarianalgorithmthatitcanonlysolvetheproblemswherethetotalcostisthesumofthatofeachjob,animprovedHungarianalgorithmisproposedandusedtosolvetheassignmentproblemofserial-parallelsystems.Firstofall,byreplacingparalleljobswithvirtualjobs,theproposedalgorithmconvertstheserial-parallelsystemintoapureserialsystem,wheretheclassicalHungarianalgorithmcanbeusedtogenerateatemporalassignmentplanviaoptimization.Afterwards,theassignmentplanisvalidatedbycheckingwhetherthevirtualjobscanberealizedbyrealjobsthroughlocalsearching.Iftheassignmentplanisnotvalid,theconvertedsystemwillbeadaptedbyadjustingtheparametersofvirtualjobs,andthenbeoptimizedagain.Throughiterativesearching,thevalidoptimalassignmentplancaneventuallybeobtained.Toevaluatetheproposedalgorithm,thevalidoptimalassignmentplanisappliedtolaborallocationofamanufacturingsystemwhichisatypicalserial-parallelsystem.
简介:Channelassignmentisachallengefordistributedcognitivenetworksduetospectrummobilityandlackofcentralizedentity.Wepresentadynamicandefficientalgorithmviaconflictshifting,referredasShifting-basedChannelAssignment(SCA).Inthisalgorithm,thesystemwasmodeledwithaconflictgraph,anduserscannotassignthechannelsthatprimaryusers(legacyusers)andneighborsalreadyoccupied.Inordertoeliminatetheconflictsbetweenneighborsefficiently,secondaryusers(unlicensedusers)trytotransferthemthroughastraightpathtotheboundary,whereconflictsareeasiertosolveastherearelessneighborsforboundaryusers.Actionsinoneshiftareexecutedinslots,andusersactinasynchronousandseparatedmanner.Asaresult,someoftheconflictingchannelsareavoidfromdirectlyabandoned,andforthis,utilityoftheentirenetworkcanbeimproved.Simulationresultsshowthattheproposedalgorithmcanprovidesimilarutilityperformancewhileobviouslyreducingthecommunicationcostthanbargaining-basealgorithms.Insmallscalenetworkswithlowusermobility(under20%),itreduces50%ofthecommunicationoverheadthanthelater.
简介:Intheautomationofidentificationoflandscapefeaturesthevaguenessarisesfromthefactthattheattributesandparametersthatmakeupalandscapevaryoverspaceandscale.Inmostofexistingstudies,thesetwokindsofvaguenessarestudiedseparately.Thispaperinvestigatestheircombinationinidentificationofcoastlandscapeunits.Fuzzysettheoryisusedtodescribethevaguenessofgeomorphicfeaturesduetothecontinuityinspace.Thevaguenessresultedfromthescaleofmeasurementisevaluatedbystatisticindicators.Thedifferencesoffuzzyobjectsderivedfromdataatdifferingresolutions(insizefrom3×3cellsto25×25cells)arestudiedinordertoexaminethesehigher-orderuncertainties.
简介:Thispaperconsiderseigenstructureassignmentinsecond-orderlinearsystemsviaproportionalplusderivativefeedback.Itisshownthattheproblemiscloselyrelatedtoatypeofso-calledsecond-orderSylvestermatrixequations.Throughestablishingtwogeneralparametricsolutionstothistypeofmatrixequations,twocompleteparametricmethodsfortheproposedeigenstructureassignmentproblemarepresented.Bothmethodsgivesimplecompleteparametricexpressionsforthefeedbackgainsandtheclosed-loopeigenvectormatrices.Thefirstonemainlydependsonaseriesofsingularvaluedecompositions,andisthusnurnericallysimpleandreliable;thesecondoneutilizestherightfactorizationofthesystem,andallowsthedosed-loopeigenvaluestobesetundeterminedandsoughtviacertainoptimizationprocedures.Anexampleshowstheeffectivenessoftheproposedapproaches.
简介:ProbabilisticAnalysisandMulticriteriaDecisionforMachineAssignmentProblemwithGeneralServiceTimesWangJing(Inst.ofSys.Eng.,Dalia...
简介:采用借的地址算法能减少在使用的ZigBee网络的孤儿节点散布了地址任务机制(DAAM)。存在借了地址算法能增加地址赋值的成功率,但是他们在成立打破拓扑学引起的网络有象开销和时间的更大的费用那样的缺点。解决如此的问题,我们基于让拓扑学维护工作的拓扑学维护(A2BTM)建议一个更有效的分布式的借的地址任务算法。它在一样的分支从后代节点第一借地址因为孤儿节点和答复立即散布了借的地址消息的请求,到维持开销和时间在借的地址的机制上花了的网络拓扑学和减少。理论并且模拟分析表明那个A2BTM算法以在成立网络花的开销和时间超过DAAM和它的改进算法,在保留地址赋值的更高的成功率的前提上。而且,A2BTM能高效地从弯路现象减少影响。