简介:Theletterproposesathree-layermanageablemediadistributionnetworksystemarchi-tecturecalledMSPnet,whichisbasedonSessionInitiationProtocol[1]andPeertoPeer(SIPP2P)technology.MSPnetperformsapplication-levelstructuredDHTroutingandresourcelocationamongdomainsandunstructuredonesindomain.Exceptformediadistribution,itcanbeusedtosupportavarietyofP2Papplications,includingvideobroadcasting,videoondemand,VoIP,etc.MSPnetiscomposedofthreelayers,namely,thesignalcontrollayer,themanagementlayer,andthemediatransportationlayer.TheMSPnetprototypeconsistsoftheSIPserver,themanagementserver,themediaserver,andthenodeUserAgent(UA).Resultsfromaprototypeexperimentinalarge-scaleInternetenvironmentshowthatMSPnetisfeasible,scalableandmanageable.
简介:地理信息系统(GIS)逐渐地正在管理数据的很大的集合,一个因此集中的数据索引不能总是提供最可伸缩的解决方案。最近,p2p(P2P)网络为以一种完全分散的方式分享信息变得很流行。在这份报纸,新混合P2P空间索引网络(HPSIN)被建议,它把分布式的空铅树与基于的分布式的哈希值桌子(DHT)相结合维持询问效率和系统负担的弦网络平衡。另外,一个简单理论模型基于为HPSIN的打开的排队网络被建立。在模型,作为M/M/1排队处理器假定每个同伴系统的基本特征被捕获,并且平均质问延期的表示在靠近的形式被获得。理论分析和数字计算结果证明那在那里存在在效率和负担平衡之间的折衷的一个最佳点。由将开始索引水平ls的合适的值放为不同网络规模和询问率,HPSIN将完成因此,延期能使适应不同P2P应用程序环境的最小的全面询问。
简介:AnewcompactopticalFanofiltersuitableforbiologicalsensingisproposed,whichpatternsphotoncrystalinsinglecrystallinesiliconnanomembranes(SiNMs)andtransferringontotransparentglasssubstrates.Theeffectsofairholesizeandsiliconthicknessonthetransmissioncharacteristicsofnewfilterarenumericallyinvestigatedbyusingthree-dimensionalfinite-differencetime-domain(FDTD)technique,thespectralresponseisalsostudiedbyback-fillingbioliquid.Theresultsshowthatthedipwavelengthwillshifttowardlongerwavelengthbyeitherreducingairholeradiusorfillingbio-liquid.Thenumberofdipswillincreasewiththeincreaseofsiliconthickness.Thesizeofproposedfiltercanbelessthan1mm2.
简介:Videoobjectextractionisakeytechnologyincontent-basedvideocoding.AnovelvideoobjectextractingalgorithmbytwoDimensional(2-D)mesh-basedmotionanalysisisproposedinthispaper.Firstly,a2-Dmeshfittingtheoriginalframeimageisobtainedviafeaturedetectionalgorithm.Then,higherorderstatisticsmotionanalysisisappliedonthe2-Dmeshrepresentationtogetaninitialmotiondetectionmask.Afterpost-processing,thefinalsegmentingmaskisquicklyobtained.Andhencethevideoobjectiseffectivelyextracted.Experimentalresultsshowthattheproposedalgorithmcombinesthemeritsofmesh-basedsegmentingalgorithmsandpixel-basedsegmentingalgorithms,andherebyachievessatisfactorysubjectiveandobjectiveperformancewhiledramaticallyincreasingthesegmentingspeed.