作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2023, Vol. 49 ›› Issue (8): 29-36. doi: 10.19678/j.issn.1000-3428.0066292

• 热点与综述 • 上一篇    下一篇

多层次实用拜占庭容错算法改进

王春东, 王翔宇   

  1. 天津理工大学 计算机科学与工程学院, 天津 300384
  • 收稿日期:2022-11-18 出版日期:2023-08-15 发布日期:2023-08-15
  • 作者简介:

    王春东(1969—),男,教授、博士生导师,CCF会员,主研方向为网络信息安全、智能信息处理、区块链技术

    王翔宇,硕士研究生

  • 基金资助:
    国家自然科学基金联合基金项目(U1536122); “科技助力经济2020”重点专项(SQ2020YFF0413781); 天津市科委重大专项(15ZXDSGX00030)

Improved Multi-Storey Practical Byzantine Fault Tolerance Algorithm

Chundong WANG, Xiangyu WANG   

  1. School of Computer Science and Engineering, Tianjin University of Technology, Tianjin 300384, China
  • Received:2022-11-18 Online:2023-08-15 Published:2023-08-15

摘要:

应用于联盟链的实用拜占庭容错(PBFT)共识算法存在共识节点选择和多节点共识效率较差的问题,为此,提出一种多层次实用拜占庭优化共识算法MS-PBFT。根据节点所属机构的不同对节点进行分组,并将各组内的节点划分为主网层和次网层。引入一种积分机制,根据各节点自身的性能以及在系统中的表现为其赋予初始积分和信誉值,以监督节点在共识过程中的行为,根据初始积分与信誉值对节点进行分类,使其加入不同的层次中。改进上层节点和主节点的选取方式并增加一种弹劾机制,及时更换掉线或存在恶意行为的领导节点,从而保证节点的可靠性同时提高共识效率。通过先在次网层达成局部共识进而在主网层实现全局共识的方式,降低节点的通信复杂度。实验结果表明,与DGPBFT、PBFT、RAFT算法相比,MS-PBFT算法可以提高数据吞吐量并降低共识时延,利用所提节点选举机制,节点的共识成功率可以达到98.6%,且共识效率比PBFT算法平均提高33%。

关键词: 联盟链, 实用拜占庭容错算法, 信誉值, 全局共识, 局部共识

Abstract:

The Practical Byzantine Fault Tolerant(PBFT) consensus algorithm that is applied to alliance chains has poor efficiency for consensus node selection and multi-node consensus. Therefore, a multi-storey practical Byzantine optimized consensus algorithm called MS-PBFT is proposed. The nodes are grouped according to their respective organizations, and the nodes in each group are divided into primary and secondary network layers. An integration mechanism is introduced to assign initial points and reputation values to each node based on its performance and performance in the system to monitor the behavior of the node during the consensus process. Nodes are classified based on initial points and reputation values and added to different levels, improving the selection method of upper level nodes and master nodes and adding an impeachment mechanism to timely replace leaders who are offline or have malicious behavior, which ensures the reliability of nodes and improves consensus efficiency. Achieving local consensus at the secondary network layer and achieving global consensus at the primary network layer reduces the communication complexity of nodes. Experimental results show that compared to the DGPBFT, PBFT, and RAFT algorithms, the MS-PBFT algorithm can improve data throughput and reduce consensus latency.Using the proposed node election mechanism, the consensus success rate of nodes can reach 98.6%, and the consensus efficiency is improved by an average of 33% compared to the PBFT algorithm.

Key words: consortium chain, Practical Byzantine Fault Tolerance(PBFT) algorithm, reputation value, global consensus, local consensus