基于贝叶斯网络的量化信任评估方法

摘要

随着云计算的不断发展，物联网逐步涉及各行各业，其中包含大量的感知信息、个人或群体的隐私信息。此外，物联网最直接、最严峻的安全隐患是网络中参与信息采集与数据融合的恶意节点，以合法身份发送虚假信息、窃听发送指令等，所以保障物联网安全刻不容缓，尤其是确保节点之间的信任关系。为决定新节点是否可以加入网络，以及排除网络中已有的恶意节点，利用贝叶斯网络量化评估节点间的信任概率，通过节点信任状态分级，融合先验信任概率，分配信任条件概率，推理预测评估节点的信任概率，确定信任等级。通过仿真实验，结果证明了该评估方法的有效性，并在一定程度上降低了评估的主观性。%With the continuous development of cloud computing,Internet of Things ( IoT) gradually involves in all walks of life,which contains large amounts of sensitive information,privacy information. In addition,the most direct and serious security risks are malicious nodes involving in information acquisition and data fusion,which send false information and eavesdrop instructions sent with legal identi-ty. Therefore,it is greatly urgent to ensure the security of IoT,especially trust relationship among nodes. In order to determine whether to allow the new node to join the network and to remove the existing malicious nodes,a quantitative trust assessment method is proposed based on Bayesian network. Through classification of trust status of nodes,integration of trust priori probability and allocation of condi-tional probability,the trust probability of assessment nodes could be predicted and inference to determine the trust level. The simulation re-sults show the effectiveness of assessment method and that the assessment subjectivity can be reduced to some extent.