面向泛在电力物联网的边缘节点感知自适应数据处理方法

Novel Edge-ware Adaptive Data Processing Method for the Ubiquitous Electric Power Internet of Things

泛在电力物联网(UEP-IoT)背景下,采用数据中心集中处理的传统云计算运行方式难以适应急剧扩大的数据规模,且存在对电力数据中心计算能力要求较高的缺点。针对上述问题,在构建基于边缘计算的泛在电力物联网数据处理架构的基础上,提出一种新的感知自适应数据处理方法。首先,通过设计基于计数bloom滤波器的边缘节点数据类型自适应感知机制,从而自动识别泛在电力物联网数据类别与对应字段。其次,设计了一种包含了通用可移植的文件尾(EOF)识别插件以及数据副本管理系统的边缘节点感知自适应处理方法,最终实现了泛在电力物联网数据的有效收集。仿真结果表明,对于相同的数据集,所提方法对辨识数据类别的正确率提升了1.3%以上;与传统数据副本管理方法相比,所提就地管理与集群管理相结合的副本管理方法对数据集的处理平均时间下降了33.33%,表明所提方法能够有效提升对海量UEP-Io T数据的处理性能。

In the ubiquitous electric power internet of things(UEP-IoT),the traditional cloud computing operation mode that uses centralized processing mode in the power data center is difficult to deal with the rapidly expanding data scale and requires high computational capacity for the power data center.Aiming at solving the above problems,in this paper,based on designing an edge computing framework for processing the UEP-Io T data,a novel adaptive data processing method is proposed.Firstly,via design a counting bloom filter-based adaptively sensing mechanism for the edge nodes,the UEP-Io T data categories and the corresponding data segments can be identified automatically.Secondly,an edge-node-aware adaptive data processing scheme,containing a universal portable end-of-file(EOF)identification plug-in and a data-copy management system,is designed to realize the UEP-IoT data collection.The numerical examples show that,for the same data set,the accuracy of the data type identification of our proposed method can be improved by over1.3%.Besides,compared with the traditional data replica management scheme,the average processing time of the proposed combining local management with cluster management scheme for the data set decreases by 33.33%,respectively,which shows that the proposed scheme can effectively improve the processing performance of massive UEP-IoT data.