首页> 外国专利> Data communications performance monitoring using principal component analysis

Data communications performance monitoring using principal component analysis

机译:使用主成分分析进行数据通信性能监控

摘要

A method and apparatus are provided for monitoring data communications performance over at least a portion of a network having a plurality of network users each having, linked to the network, a network termination device for accessing services over the network. In a training phase, remote connections are established to the network termination device of each in a predetermined set of network users served by a predetermined portion of the network to retrieve parameter data generated by the device, the parameter data comprising values for a predetermined set of parameters indicative of data communications performance over that portion of the network. Principal component analysis is applied to the retrieved data to determine the principal component directions of variation. One or more models of data communications performance are then generated for that portion of the network in terms of one or more value ranges in a projection of the retrieved parameter data into a coordinate system defined by two or more of the determined principal components. In a monitoring phase, the same parameters are retrieved from a network termination device of a selected user served by the same portion of the network and projected into the coordinate system of the generated models to expose any significantly divergent behaviour, potentially indicative of a fault condition.
机译:提供了一种用于监视至少一部分网络上的数据通信性能的方法和装置,该网络具有多个网络用户,每个用户具有链接到网络的网络终端设备,用于通过网络访问服务。在训练阶段,建立与网络的预定部分服务的预定网络用户集合中每个网络终端设备的远程连接,以检索由该设备生成的参数数据,该参数数据包括针对一组预定用户的值。指示网络那部分数据通信性能的参数。将主成分分析应用于检索到的数据,以确定主成分变化方向。然后,根据所获取的参数数据到由两个或多个确定的主成分定义的坐标系中的投影中的一个或多个值范围,针对网络的该部分生成一个或多个数据通信性能模型。在监视阶段,从网络的相同部分服务的选定用户的网络终端设备中检索相同的参数,并将其投影到生成的模型的坐标系中,以暴露任何明显不同的行为,从而可能指示故障情况。

著录项

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号