无参照节点信任评估机制WSN多状态数据融合

Data Fusion of WSN in Many Situations Based on No References and Several factors Nodes Trust Evaluation Mechanism

研究快速准确识别无线传感器网络中的恶意节点进而实现数据安全融合的问题。针对传统的节点异常行为检测与信誉评测模型存在功能单一、考虑因素不全面、信任值算法比较复杂等缺陷,提出一种基于节点自主行为的无参照信任评估模型。每个节点都具有监控权限和被监控义务,临近区域内任意相邻节点彼此进行行为监控,采用相对简便的计算方法获得自身对对方的直接信任值和间接信任值,再根据D-S证据理论的合成法则得到自身对对方的综合信任值,在数据融合时,临近区域内各节点根据彼此的综合信任值自组织成簇,并自动筛选出簇头节点作为数据融合的节点,赋予来自各节点的每个数据一个不同权值,综合信任值设置了权值向量,进行融合处理,进而将融合安全结果传送到基站。仿真结果表明,信任评估模型能快速有效识别恶意节点,避免恶意节点成为簇头节点,通信效率高、节省节点能量的同时,大幅提高了数据融合的安全性和准确性,能适应于WSN不同应用领域和不同部署环境状态下的应用。

The malicious nodes in the wireless sensor network （WSN） should be recognized accurately for the re alization of data fusion safely. The legality and security are the keys in data fusion. Traditionally, the detection and evaluation models of abnormal behaviors are single in function, and the algorithm is complex. According to the weak ness as above, a new no reference trust evaluation model was proposed based on the self behavior of nodes. And each node in the WSN had the authority of monitor and should be monitoT＇ed. It worked mutually in the near area. The algorithm was relatively simple. And the comprehensive trust values are obtained based on I） S evidence theory from self to the other side. The cluster head nodes were selected automatically as the data fusion nodes, and the dif ferent weight was added to each data of the different nodes. Weight vectors were obtained from comprehensive trust data, and the data fusion was completed. The result can be transmitted to the database in security. Simulation result shows that the model can recognize the malicious nodes precisely, which avoids that the malicious nodes are selected as the cluster head nodes. And the communication efficiency is also improved, the computing cost is reduced in the same time, and the security and precision are improved. It can be applied in different areas and different environ ments for WSN.