基于DPoS共识机制的改进  

Improvements based on DPoS consensus mechanism

作　　者：夏棋 谭敏生 朱涛 丁琳 XIA Qi;TAN Min-sheng;ZHU Tao;DING Lin(School of Computer/School of Software,University of South China,Hengyang 421001,China)

机构地区：[1]南华大学计算机学院/软件学院,湖南衡阳421001

出　　处：《计算机工程与设计》2023年第12期3635-3641,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(62006110);湖南省自然科学基金项目(2022JJ50153)。

摘　　要：为降低委托权益证明共识机制中心化,提升系统吞吐量,提出一种全民评分选举、异步出块的共识机制。采用低门槛的参与方式,保证全民享有选举资格,在选举中引入责任连坐的奖惩制度,规范节点行为;在此基础上,利用有向无环图的特点,改进图的无序性,形成伪队列结构的存储模式。实验结果表明,该共识机制去除了节点权益的统治地位,降低了节点参与门槛,提高了系统活跃度;单位时间出块平均提升40%左右;系统数据冗余时,共识时延缩短45%左右。To reduce the centralization of the delegated proof-of-stake consensus mechanism and improve the system throughput,a consensus mechanism with universal scoring election and asynchronous block generation was proposed.A low threshold participation method to ensure that the whole people were eligible for election was adopted in this consensus mechanism,and a reward and punishment system for joint responsibility in the election to standardize the behavior of nodes was introduced.Due to the characteristics of the directed acyclic graph,the disorder of the graph was improved and the storage mode of pseudo-queue was formed.Experimental results show that the dominance of node rights was removed,the threshold for node participation was lowered and the system activity is improved using this model.The block output efficiency per unit time is increased by about 40%on average.When the system data is redundant,the consensus delay is shortened by about 45%.

关 键 词：共识机制 评分 责任连坐 异步 奖惩 有向无环图 伪队列 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]