移动Ad Hoc网由于没有固定基础设施,使得网络节点间的信任问题变得日益突出。针对已有信任评估模型在证据收集和信任融合方面的不足,提出了一个多维证据融合信任评估模型。通过拓展Watchdog机制,检测节点对包的各种历史处理行为,来扩宽直接信任证据收集的维度;同时使用Beta分布下的贝叶斯统计方法,计算每种行为的可信度,之后通过证据合成规则来融合多种行为,以提高信任评估的准确性。通过仿真实验,验证了该模型的有效性;分析表明,模型中使用信息贴近度量化推荐权重,降低了恶意推荐带来的影响,保证了信任评估的健壮性。
In the mobile Ad Hoc network, there is no infrastructure, so that the trust evaluation is more and more significant. In the existing model of trust evaluation, there are some defects on evidence collection and integration. A trust evaluation model of multidimensional evidence fusion is proposed. It detects the various historical behavior of nodes through the expanding Watch- dog to widen the dimension of direct evidence collecting. Each behavior is calculated by using Bayesian method which meets the Beta distribution. Through the credibility method, all behaviors are merged together to improve the accuracy of assessment of the trust. The simulation results demonstrate the effectiveness of this model. Analysis shows that recommended weight is quantitatively evaluated by information similarity measure, reducing the impact of malicious recommendation, ensuring the trust assessment robustness.