RBFT:基于Raft集群的拜占庭容错共识机制

针对现有联盟链共识机制因可拓展性不足,无法在支持大规模网络的同时满足低时延、高吞吐量和安全性的问题,采用网络分片的思想,提出一种适用于联盟链的带有监督节点的两级共识机制——RBFT。首先对网络节点进行分组,组内采用改进的Raft机制进行共识,然后由每个组内选出的领导者组成网络委员会,网络委员会内部采用PBFT机制进行共识。研究结果表明,在大规模网络环境下,相比PBFT和Raft,RBFT在具备拜占庭容错能力的同时可以保证高共识效率,因而具有更高的扩展性。

RBFT:a new Byzantine fault-tolerant consensus mechanism based on Raft cluster

The existing consensus mechanisms of consortium blockchain are not scalable enough to provide low latency,high throughput and security while supporting large-scale network.A new consensus mechanism called RBFT was proposed to improve scalability,which was a two-level consensus mechanism with supervised nodes based on the idea of network fragmentation.In RBFT,the nodes were firstly divided into several groups.Each group adopted the improved Raft mechanism to reach consensus and select leader.Then,the leaders of each group formed the network committee,and the network committee adopted PBFT mechanism for consensus.Comparative experiments verify that RBFT can tolerant Byzantine fault while ensuring high consensus efficiency in large-scale network compared with PBFT and Raft.