Abstract: Based on the object-oriented trust attack graph, the object-oriented rough trust attack threat perception model is put forward; the model describes all attack relations among attributes of trust entity objects. Though the introduction of the rough graph theory, attack methods with the same attack consequences and trust entity objects with the same importance in attack relations are classified into an equivalent class in an object-oriented rough trust attack graph. Based on these equivalent classes, only several characteristic attack paths are obtained,and then the whole attack strategies space can be searched out, thus the contradict between overall grasping attack actions and greatly limiting the scale of attack paths is solved. By defining similar degree of path, all characteristic attack paths which can arrive at an attack goal can be searched out by use of the ant algorithm, and the maximal threat path, which can access to the goal node, can be found out from all these characteristic attack paths. An application study shows that the model can quickly locate the attacked trust entities objects and the related attack methods, and accurately find their positions from all these characteristic attack paths.

Key words: trust, trust security, trust attack, rough graph, trust attack graph

摘要： 基于面向对象的信任攻击图，提出了一种复杂信任环境系统信任攻击威胁感知模型；该模型描述了信任主体对象属性间的所有攻击关系。通过引入粗糙图理论，将面向对象粗糙信任攻击图中具有相同攻击效果的攻击方法，以及在攻击关系中具有相同重要性的信任主体对象划入同一等价类，基于这些等价类，只要获取有限的几条特征攻击路径就能够搜索整个攻击策略空间，从而解决了全面把握攻击动向和限制路径规模之间的矛盾。通过定义路径相似度，采用蚁群算法在论域信任攻击图中搜索到达攻击目标的特征路径，在这些特征路径中找出给目标节点带来的最大威胁的攻击路径。试验证明该方法能够快速定位被攻击的信任主体对象及攻击方式，在各种特征攻击路径中准确找到其所在位置。

关键词: 信任, 信任安全, 信任攻击, 粗糙图, 信任攻击图