基于指数平滑的无线Mesh网络改进信任模型

IMPROVED TRUST MODEL FOR WMN BASED ON EXPONENTIAL SMOOTHING

针对无线Mesh网络的安全问题,提出一种改进的信任模型ESTrust。利用观测节点和待评价节点在若干时间片内的交互次数构造时间序列,通过三次指数平滑法预测下一个时间片两节点的交互次数,取预测值和实际交互次数的相对差值比计算待评价节点的直接信任值。引入熵值法为网络中各观测节点对待评价节点的直接信任值分配加权系数,并对这些直接信任值进行加权运算,得到待评价节点的综合信任值。仿真结果表明:当网络存在恶意节点时,相比AODV路由协议和Beth信任模型,ESTrust模型拥有更好的服务质量和更低的通信代价。

iming at the security problem of WMN, we propose an improved trust model ESTrust. It uses the number of interactions between observation node and evaluating node within several time periods to construct the time series, and predicts the number of interactions between the two nodes within next time period through cubic exponential smoothing method. The relative difference ratio of the predicted and actual number of interactions is used to calculate the direct trust value of the evaluating node. It introduces the entropy method to allocate the weight coefficients on the direct trust values of the evaluating node for every observation node in network, and conducts weighting operations on these direct trust values to get the synthesised trust value of evaluating node. Simulation results show that compared with AODV routing protocol and Beth trust model, the presented ESTrust model has better service quality and lower communication cost when there are malicious nodes in WMN.