WSNS中基于Fusion-Bayes的离群点检测

随着无线传感器网络技术的发展,数据采集量越来越大,维数也不断提高。然而现有的离群点检测算法多是面向单维或低维度数据,对此文中提出了基于Fusion-Bayes的离群点检测算法。该检测方法首先利用数据转换技术将不同数据属性转换成统一格式,使得各属性可以进行融合运算;然后再利用贝叶斯方法对融合后的属性进行离群点检测。通过实验得出,多维数据属性融合后的检测结果相比于单维属性或低维属性的检测更加准确、效果更好。

WSNS Outlier Detection Based on Fusion-Bayes

With the rapid development of wireless sensor network,the amount of data acquisition as well as the dimensions of data increases constantly.However,the existing outlier detection algorithms are mostly used for single or low dimensional data.This paper presents an outlier detection technique based on Fusion-Bayes.This method first uses the data fusion to converse data from different sources into the same form,and then uses a Bayesian method to detect outliers of different attributes.Experiments show multidimensional data attribute fusion detection is more accurate and the more effective compared to the single dimension attribute or low dimension attribute detection.