基于群体信任的 WSN 异常数据过滤方法

以节点数据的时空相关性为理论依据，通过定量数据与定性知识之间的不确定性转换，在知识层面上比较节点间数据的相似程度，实现对单节点数据的群体信任评估，进而设计了一种实时的WSN异常数据过滤方法，在节点数据采集过程中实时发现可疑数据。仿真实验验证了此方法不但能够实时过滤异常数据，提升WSN的入侵容忍能力，还有较低的通信及计算开销。

Abnormal data filtering approach based on collective trust for WSN

Data security is the major challenge for WSN applications. It's significant in theory and practice to detect and filter false data effectively. Traditional approaches based on symmetric key, public key or polynomial always need large cost in transmission and computation, and could hardly detect the abnormal data caused by hardware of nodes. According to the spatio-temporal correlation of data in WSN, quantitative data can be converted to qualitative knowledge, and col-lective trust of data can be computed based on the comparisons of qualitative knowledge. A real-time outliner filtering approach was proposed to detect and filter abnormal data. Simulation results show that this method cannot only detect and filter the outliner in-time,but also need low cost in transmission and computation.