专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

2. 发明公开

EP1780955A1 Monitoring method and apparatus of processing of a data stream with high rate/flow 审中-公开
标题翻译：观察的用于以高速率处理数据流的方法和设备
公开(公告)号：EP1780955A1
公开(公告)日：2007-05-02
申请号：EP05023580.3
申请日：2005-10-28
申请人： SIEMENS AKTIENGESELLSCHAFT
发明人： Freire, Mario , Garcia, Nuno , Hajduczenia, Marek , Monteiro, Paulo , Silva, Henrique
IPC分类号： H04L12/56 , H04L12/26
CPC分类号： H04L47/10 , H04L41/142 , H04L43/08 , H04L47/11 , H04L47/12 , H04L47/2441
摘要： This aim of this invention is achieved by a monitoring method of processing of a data stream that is sequentially inputted at a high rate/flow in an element where said processing is executed, wherein:
- said data stream has at least one valuable self-similarity property that interacts with performance of said processing;
- storing and estimation means are used to provide a self-similarity degree after measuring values in a sampled data set from the inputted data stream;
- said self-similarity degree is defined as Hurst parameter that is estimated by mean of an Embedded Branching Process (EBP) with related crossing tree structure und crossing levels;

characterized in that the invention provides further new features:
a modified Embedded Branching Process (called mEBP) is used, wherein:
- said sampled data set is comprising a first previous stored aggregated sample of a plurality of incoming data and a second current sample of a single new incoming data;
- for estimating a new Hurst parameter, said crossing tree structure for the previous stored aggregated sample is actualized according to the new incoming data;
- said actualization of calculated crossing tree structure is provided by selecting a limited and adaptable number of crossing levels for measuring values, by using minimal storing means for the previous stored aggregated sample as well as by computing in real-time at least one mean value of the previous stored aggregated sample with a measured value of the current sample in order to minimize the estimation time of the new Hurst parameter.
摘要翻译：本发明的这一目的是通过一种数据流没被以高速率依次输入的处理的监视方法来实现/在元件，其中所述处理被执行时，worin流中： - 所述数据流具有至少一个有价值的自相似属性做了与所述处理的性能间行为; - 存储与估计装置被用来测量在采样数据从输入的数据流中设置的值之后提供的自相似性程度; - 所述自相似性度被定义为Hurst参数所做的是通过与相关交叉树结构和交叉水平嵌入式分支过程（EBP）的平均估计; 其特征在于DASS本发明提供进一步的新的特点：修饰的嵌入式分支过程（称为MEBP）时，worin： - 所述采样数据集是包括输入数据的多个一第一先前存储的聚集和样品的第二电流采样单一的新到来的数据; - 用于估计新Hurst参数，所述交叉树结构为前存储的聚集样品现实化gemäß到新的输入数据; - 所述计算的交叉树结构的实现是通过选择交叉电平的限制，并且可适应数量用于测量值，通过使用最小的存储装置，用于先前所存储的聚集样品以及通过在实时中的至少一个均值计算提供先前所存储的聚集样品与当前采样的，以便最小化新Hurst参数的估计时间测量值。

IPRDB

热门服务

关于我们

友情链接

联系方式