基于消息相似度的拜占庭容错共识算法

A Byzantine Fault Tolerant Consensus Algorithm Based on Similarity-of-message

为保证入侵容忍的数据可用性和可靠性,区块链系统需要使用拜占庭容错算法对数据进行复制备份,以避免受被入侵节点攻击而导致的数据不可用,甚至数据丢失。针对Quorum模式拜占庭容错算法,提出了在高并发场景下实现平滑降级的Q/U-GD算法,该算法使用消息相似度机制来对Q/U算法进行优化。设计的消息相似度机制通过提升客户端和副本节点接受的消息的相似度,减少触发不必要的修复操作,实现性能平滑降级。实验结果表明,Q/U-GD算法的吞吐量在高并发场景下降级平缓,且明显高于Q/U算法。

To ensure the intrusion-tolerant availability and reliability of data,the blockchain system is required to use Byzantine fault-tolerant algorithm to replicate data,to avoid the unavailability of data or even loss of data caused by attacks from intrusion nodes.For mode quorum Byzantine fault-tolerant algorithm,a Q/U-GD algorithm is proposed to achieve smooth degradation in high concurrency scenarios.Q/U algorithm is optimized with the Similarity-of-Message(SOM for short)mechanism by Q/U-GD algorithm.The designed SOM reduces the trigger of unnecessary repair operations by enhancing the similarity of messages received by clients and replica nodes.The smooth degradation of performance is realized.The experiment results show that the throughput of Q/U-GD algorithm degrades smoothly in high concurrent scenarios,and is significantly higher than that of Q/U.