由于移动Ad hoc(MANET)网络具有无固定基础设施、通信信道不可靠、节点随机移动等特性,对其安全问题的研究一直富有挑战性。信任机制结合加密技术可以有效缓解各类网络攻击(例如丢包攻击)。考虑到MANET网络的特性,需采取智能化方法来实现信任构建,进而发现攻击者的恶意行为。文章侧重缓解丢包攻击与数据包篡改攻击威胁,提出了一种基于贝叶斯网络的可信度评估构建机制,该机制可有效进行因果推理。文章提出的信任模型可有效区别导致网络丢包的不同原因:是由于网络信道不可靠,还是由于遭受恶意攻击。这样,即可计算出更精确的可信度。最后,仿真结果表明文章所提算法在恶劣网络环境中可行且有效。 Due to its characteristics such as no fixed infrastructure, unreliable communication channels and random movement of nodes, the research on the security of mobile ad hoc (MANET) networks has always been challenging. Trust mechanism combined with encryption technology can effectively alleviate all kinds of network attacks (such as packet loss attacks). Taking into account the characteristics of the MANET network, an intelligent approach needs to be taken to implement trust building to discover the attackers’ malicious behavior. This paper focuses on mitigating the threat of packet loss attacks and packet tampering attacks, and proposes a mechanism for evaluating the credibility of Bayesian networks. This mechanism can effectively perform causal reasoning. The trust model proposed in this paper can effectively distinguish the different causes of network packet loss: whether the network channel is unreliable or due to malicious attacks. In this way, more accurate confidence can be calculated. Finally, the simulation results show that the proposed algorithm is feasible and effective in harsh network environments.