基于网络自聚类的PBFT算法改进

Improved PBFT algorithm based on network self clustering

联盟链是区块链技术在实际行业应用的主要形式,其共识机制多采用实用拜占庭容错算法(PBFT),在节点数量大时共识成功率与共识效率不高,存在扩展性问题。为此,提出一种基于网络自聚类拜占庭容错共识算法NAC-PBFT。利用行业应用中网络结构、系统节点等确知信息,在联盟链审核节点时指定种子节点,再以种子节点为中心自聚类为若干分组,组内通过优化实用拜占庭容错算法选举出代理人,由各组代理人共同完成全局共识。其中,组内选举时,通过定义可信度指标衡量节点作为筛选候选代理人的标准,确保每次选出的代理人具有良好的状态。通过对系统分析与性能测试,NAC-PBFT算法能有效降低消息量,在共识时间、系统吞吐量指标上有更好的表现,具备较好的扩展性。

Consortium chain is the main application form of blockchain technology in the actual industry,and its consensus mechanism mostly adopts the practical Byzantine fault-tolerant algorithm(PBFT).However,if the number of nodes is too large the success rate and efficiency of consensus will lower,and have scalability problems.Therefore,this paper proposed a Byzantine fault-tolerant consensus algorithm NAC-PBFT based on network self clustering.Using the known information of network structure and system node in industry application,the seed node designated in the audit node of consortium chain,and taken as the center and self clustered into several groups.It elected the agents by optimizing the practical Byzantine fault-tolerant algorithm in the group,and the agents in each group completed the global consensus.During the intra group election,the credibility index was defined to measure the node,and it was used as the standard for selecting candidate agents to ensure that the agents selected each time were in good condition.Through the system analysis and performance test,NAC-PBFT algorithm can effectively reduce the amount of information,has better performance in consensus time and system throughput index,and has better scalability.