摘要
针对在无线传感器网络中传感器节点本身能量有限的特性,提出一种基于D-S证据理论的组合数据融合算法.先对传感器网络的当前值依据各组数据的标准差进行聚类,然后对每一类数据组,用D-S证据推理算法进行融合,将其结果看成一个虚拟传感器节点数据,最后通过计算马哈诺比斯距离得出虚拟节点数据向量的异常值,把它作为加权权重进行加权融合.仿真试验表明:该算法识别目标的可信度高于D-S推理法,且在计算复杂度上也有明显优势.
Considering the characteristics of energy constrained of sensor node itself in wireless sensor networks,a combination of data fusion algorithm was proposed base on D-S evidence theory.Firstly,the current value of Sensor network data was classified to some cluster based on the standard deviation,and then each category of data sets was fused by using D-S evidential reasoning algorithm,and the result would be treated as a virtual sensor node data.Finally,the outlier of virtual node data vectors was obtained by calculating Mahalanobis distance and weighted integration was completed via the outlier.Simulation results demonstrate that the credibility of identifying the target with the algorithm is higher than D-S reasoning method,and the algorithm in computational complexity has also a clear advantage.
出处
《微电子学与计算机》
CSCD
北大核心
2011年第1期95-98,共4页
Microelectronics & Computer
基金
湖北省教育厅科研项目(B20083003)
关键词
无线传感器网络
数据融合
聚类分析
加权融合
wireless sensor networks
data fusion
cluster analysis
weighted integration
