Key Laboratory of Aerospace Information Security and Trusted Computing, Ministry of Education, School of Cyber Science and Engineering, Wuhan University, Wuhan, China.
PLoS One. 2020 Mar 10;15(3):e0228844. doi: 10.1371/journal.pone.0228844. eCollection 2020.
The data layer devices in the Software Defined Network (SDN) play an important role in packet forwarding. However, whether the forwarding task can be efficiently completed by the node has not attracted enough attention. A method called TrustBlock is proposed in this paper, which introduces trust as a security attribute in SDN routing planning. Besides, in order to enhance the integrity and controllability of trust evaluation, the double-layer blockchain architecture is established. In the first layer, the behavior data of the node is recorded, and then the trust calculation is performed in the second layer. In the evaluation model, nodes' trust is calculated from three aspects: direct trust, indirect trust and historical trust. Firstly, from the perspective of security, blockchain is used to achieve identity authentication of nodes, after that, from the perspective of reliability, the forwarding status is used to calculate the trust value. Secondly, consensus algorithm is used to filter malicious recommendation trust value and prevent colluding attacks. Finally, the adaptive historical trust weight is designed to prevent the periodic attack. In this paper, the entropy method is used to determine the weight of each evaluation attribute, which can avoid the problem that the subjective judgment method is not adaptable to the weight setting. Simulation results show that the detection rate of the TrustBlock is up to 98.89%, which means this model can effectively identify the abnormal nodes in SDN. Moreover, it is attractive in terms of integrity and controllability.
软件定义网络（SDN）中的数据层设备在数据包转发中起着重要作用。然而，节点是否能够高效地完成转发任务并没有引起足够的重视。本文提出了一种名为 TrustBlock 的方法，它将信任作为 SDN 路由规划中的一个安全属性引入其中。此外，为了增强信任评估的完整性和可控性，建立了双层区块链架构。在第一层中，记录节点的行为数据，然后在第二层进行信任计算。在评估模型中，从直接信任、间接信任和历史信任三个方面计算节点的信任。首先，从安全角度出发，利用区块链实现节点的身份认证，然后从可靠性角度出发，利用转发状态计算信任值。其次，利用共识算法过滤恶意推荐信任值，防止共谋攻击。最后，设计自适应历史信任权重，防止周期性攻击。本文采用熵值法确定各评价属性的权重，避免了主观判断方法不适应权重设置的问题。仿真结果表明，TrustBlock 的检测率高达 98.89%，表明该模型能够有效识别 SDN 中的异常节点。此外，它在完整性和可控性方面具有吸引力。
