基于PBFT算法改进的去主节点共识机制优化

Optimization of Consensus Mechanism without Master Node Based on Improved PBFT Algorithm

实用拜占庭容错算法PBFT在达成一致性共识过程中,若系统中存在拜占庭或者宕机节点,就会触发视图转换协议,导致算法的通信开销增加,降低算法的共识效率。此外,PBFT随机选取系统节点为主节点,存在一定概率选中非正常节点,导致经过五阶段通信后,结果无法达成共识,需要转换视图,重新随机选取节点进行一致性共识,故提出N-P PBFT算法,省去主节点角色。实验表明,通过引入外部服务担任主节点,优化了算法的一致性共识流程,降低了算法通信开销,从而优化了系统的共识效率。

In the traditional Byzantine Fault Tolerance(PBFT)algorithm,in the process of reaching a consensus,if there are Byzantine or downtime nodes in the system,the view conversion protocol will be triggered,which will increase the communication overhead of the algorithm and reduce the consensus efficiency of the algorithm.Besides,PBFT randomly selects the system node as the master node,and there is a certain probability to select abnormal nodes,resulting in the failure to reach a consensus after five stages of communication,and it is necessary to convert views and randomly select nodes for consensus.Therefore,the N-P PBFT algorithm is proposed to omit the role of the master node,and by introducing external services as the master node,the consensus process of the algorithm is optimized,the communication overhead of the algorithm is reduced,and the consensus efficiency of the system is optimized.