中文摘要：

针对无线传感器网络中感知节点处理能力较差,获取网络安全态势要素困难的问题,提出一种层次化架构态势要素获取机制。在该架构中,采用支持向量机超球体多分类算法作为基分类器,非负矩阵分解算法作为属性约简的方法,采用模糊分类算法初始化该算法,解决随机初始化导致局部最优的缺陷。在汇聚节点完成对分类规则和属性约简规则的学习,分别在簇头和汇聚节点做分类分析,降低对感知节点性能要求。数据传输前进行属性约简,减小数据传输时的通信开销,提高分类器分类性能。仿真结果表明,该方案有较好的态势要素获取准确度和较低时间复杂度,降低了信息传输过程中的通信开销。

英文摘要：

In wireless sensor network,the processing ability of the sensor nodes is poor.And the security situational element acquisition is also a serious problem.Thus,a hierarchical framework of security situational elements acquisition mechanism was proposed.In this framework,support vector machine hyper sphere multi classification algorithm was introduced as basic classifier.The non-negative matrix factorization algorithm was used as the method of attribute reduction.The fuzzy classification algorithm was used to initialize non-negative matrix factorization,to avoid the local optimal caused by random initialization of nonnegative matrix factorization.In the sink node,classification rules and attribute reduction rules were formed by learning.The classification analyses respectively focused on the cluster head and sink node,which reduced the requirement of the sensor node properties.Attribute reduction was carried out before the data transmission,reducing communication consumption of data transmission and improving the performance of classifiers.Simulation results show,the scheme has preferable accuracy on the situational elements acquisition,and smaller communication overhead in the process of information transmission.