Reputation is not enough: Ensuring strong order-fairness in Byzantine consensus

Abstract

With the development of blockchain technology, Byzantine Fault Tolerance (BFT) is becoming an important research topic. The rPBFT consensus algorithm was introduced to address a range of shortcomings in the PBFT consensus algorithm, including high communication complexity, limited scalability, and a significant decline in performance when the system reaches approximately 100 nodes. Although rPBFT has been widely applied in the FISCO consortium chain, existing methods fail to ensure fairness in the power distribution among consensus member nodes and fine-grained node classification in rPBFT consensus. This work proposes a rPBFT consensus mechanism based on reputation value evaluation and supervision of consensus members. By implementing hierarchical management of nodes based on their reputation values, malicious nodes are eliminated, and supervision of consensus members is realized. The simulation experiment simulates the decision process of a variety of different nodes and consensus member jury with different proportions of 60%, 80% and 100% judges. The results show that the proposed scheme can dynamically update the node reputation value and classify various nodes. On the premise that the jury judges cast no less than 50% of the verdict votes, the malicious nodes in the consensus members can also be eliminated from the group of consensus member nodes. The scheme proposed in this paper effectively improves the fault tolerance of the rPBFT consensus mechanism, maintains the stability of the consortium chain network and ensures the security of the system.