中文摘要：

针对传统信任模型易遭受恶意推荐攻击及提供少数正常服务即可赚取高信誉度的问题,充分考虑无人值守WSN高度自治特点,提出了一种具有激励机制的信任管理模型.利用节点信誉服从Beta分布的理论,计算节点的信任值及其置信度.采用贝叶斯估计方法,将对节点的直接信任与来自邻居节点的推荐信息融合,计算节点综合信任.引入对评价行为的奖惩,激励节点持续地提供真实可信的服务,抵制对信任管理系统的恶意推荐攻击.仿真实验表明,与RFSN相比,引入置信度可避免不良节点通过少数正常行为获得高信任值,在通信量与存储空间显著减少,运算量相似的情况下,判别节点可信性更准确,适合节点资源受限无人值守的传感器网络.

英文摘要：

In order to address the problems that traditional trust model is vulnerable to attacks of malicious evaluation and to provide a few effective services to earn high reputation,and fully consider the high degree of autonomy in unattended WSN,a trust management model with incentive mechanism is proposed in this paper（TMM-IM）.In the model nodes＇ trust values and their confidence are computed with Beta distribution theory followed by node＇s reputation.The comprehensive trust level for a node is computed from the diffusion of directed trust of the node and recommendation trust from its neighbors using Bayesian estimation method.Nodes can be motivated to continuously provide trustworthy services by introducing the reward and punishment mechanism for evaluating nodes＇ behavior so that malicious recommendation attack to trust management system can be held back.Our simulation results show that some misbehavior nodes are prevented from obtaining high trust value by a little of normal behaviors through introducing confidence level.And our method is more precise in discrimination of nodes＇ trustworthiness in the case where the amount of communication and the storage space are significantly reduced,and the amount of computation is nearly the same compared with RFSN.And our model can be suitable for WSN with resource-constrained nodes.