基于CNN与BiGRU融合的无线传感器网络数据聚合方法
Data aggregation method of wireless sensor network based on the fusion of CNN and BiGRU
任金金 1任敬敏 1王淑芳 1赵慧芳1
作者信息
- 1. 濮阳医学高等专科学校,公共教学部,河南濮阳 457400
- 折叠
摘要
由于缺乏对无线传感器网络数据特征的综合分析,导致数据聚合延迟偏高,为此,提出基于CNN与BiGRU融合的无线传感器网络数据聚合方法研究.引入了CNN实现对无线传感器网络数据全局以及结构特征的提取,其中,CNN的两个特征提取分支分别对原始数据的全局状态和信息结构进行特征提取.在对长短期记忆网络(LSTM)进行改进的基础上,将门控循环单元网络—BiGRU引入到无线传感器网络数据的聚合研究中,利用前向GRU网络,和后向GRU网络对接无线传感器网络数据特征,实现对数据的有效聚合.在测试结果中,设计方法对于不同规模无线传感器网络数据的聚合延迟稳定在6.5s以内,处于较低水平.
Abstract
Due to the lack of comprehensive analysis of the data characteristics of wireless sensor network,the data aggregation delay is high.Therefore,the data aggregation method of wire-less sensor network based on the fusion of CNN and BiGRU is proposed.CNN is introduced to extract the global and structural features of wireless sensor network data,in which the two fea-ture extraction branches of CNN perform feature extraction on the global state and information structure of the original data respectively.On the basis of the improvement of long and short-term memory network(LSTM),the gating cycle unit network-BiGRU is introduced into the aggregation research of wireless sensor network data,and the forward GRU network and the backward GRU network are used to connect the characteristics of wireless sensor network data to realize the effective aggregation of data.In the test results,the design method for data of wire-less sensor networks stabilized within 6.5s.
关键词
CNN与BiGRU融合/无线传感器网络数据/数据聚合/特征提取/前向GRU网络/后向GRU网络Key words
CNN and BiGRU fusion/wireless sensor network data/data aggregation/feature extraction/forward GRU network/backward GRU network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据