基于信用评价模型的Raft共识算法

Raft Consensus Algorithm Based on Credit Evaluation Model

在车联网中,车辆节点间需要交通信息的共享和交互,但目前还存在着节点之间难以高效同步交通数据信息以及恶意节点传播虚假信息的问题。针对上述问题,提出了一种基于信用评价模型的Raft共识算法(CE-Raft)。首先构建信用评价模型,基于孤立森林异常检测算法检测拜占庭车辆节点并将其剔除,生成诚实节点编号表;然后进行领导者选举,通过修改跟随者节点的投票过程,实现诚实节点当选领导者;最后进行日志复制,领导者节点根据诚实节点编号表发送信息同步请求,确保正确的消息在节点间达成共识。实验结果表明,CE-Raft算法能够有效排除拜占庭节点,提高了诚实节点预测准确率,具有较低的时延和较高的吞吐量,使车联网在存在恶意节点的情况下,仍然能安全高效地完成数据共享。

In the Internet of Vehicles,the sharing and interaction of traffic information is required between vehicle nodes.How-ever,at present,there are still problems such as the difficulty in efficiently updating traffic data information between vehicle nodes and the dissemination of false information by malicious vehicle nodes.Aiming at the above problems,this paper proposes a Raft consensus algorithm based on credit evaluation model(CE-Raft).First,the credit evaluation model is constructed,and Byzantine vehicle nodes are detected and eliminated based on the isolated forest anomaly detection algorithm,and then an honest node number table is generated.Then the leader election is performed,and the honest node is elected as the leader by modifying the voting process of the follower node.Finally,log replication is performed,the leader node send an information synchronization request according to the honest node number table to ensure that the correct message reaches a consensus among the nodes.Experimental results show that the CE-Raft algorithm can effectively exclude Byzantine nodes,improve the prediction accuracy of honest nodes,and has lower latency and higher throughput,so that the Internet of Vehicles can still complete data sharing safely and efficiently in the presence of malicious nodes.