基于贝叶斯的改进WSNs信任评估模型

基于贝叶斯和熵，提出一种改进的WSNs信任评估模型。考虑到非入侵因素带来的网络异常行为，引入异常衰减因子，利用修正后的贝叶斯方程估算直接信任，并利用滑窗和自适应遗忘因子进行更新。根据直接信任的置信水平确定其是否足够可信来作为综合信任，减少网络能耗，并降低恶意反馈的影响。如果直接信任不足够可信，计算间接信任来获得综合信任，利用熵来对不同的推荐赋予权重，克服主观分配权重带来的局限性，加强模型的适应性。仿真实验表明，该模型能够有效检测恶意节点，具有较高的检测率和较低的误检率，同时在很大程度上降低了网络的能量消耗。

An improved trust evaluation model based on Bayesian for WSNs

Based on Bayesian and entropy,an improved trust evaluation model for wireless sensor networks(WSNs) is proposed. Considering the abnormal behavior caused by non-invasive factors,the abnormal discount factor is in?troduced to modify the Bayesian equation which is used to estimate the direct trust. The sliding window and adap?tive forgetting factor is used for the update of the direct trust. Then,according to the confidence level,whether the direct trust is credible enough to be a comprehensive trust can be determined,which effectively reduces energy con?sumption and the impact of malicious feedback. Finally,if the direct trust is not reliable enough,the comprehensive trust should be calculated,which uses entropy to assign weights to different recommendation. Simulation results show that the proposed model can detect malicious nodes effectively and reduce energy consumption of the network to a great extent.