基于改进Raft共识算法和PBFT共识算法的双层共识算法

针对目前应用于联盟链中的实用拜占庭（PBFT）共识算法可扩展性不足、通信开销增长过大、难以适用于大规模网络节点环境等问题，提出了一种基于改进Raft共识算法和PBFT共识算法的双层共识算法（DL_RBFT）。首先将区块链中的节点分成若干小组，组成下层共识网络，然后小组的组长再构成上层共识网络，形成一个双层共识网络结构；在下层共识网络的小组内部使用引入监督机制和声誉机制来改进Raft共识算法，在初始组长的选举流程引入了蚁群算法，使选举效率始终维持在较高水平；在上层共识网络中，使用PBFT共识算法进行共识。改进后的Raft共识算法具备了抗拜占庭节点攻击的能力，提升了算法的安全性。实验结果分析表明，相较于传统的PBFT共识算法，在100个节点的情况下，DL_RBFT将共识时延降低了两个数量级，吞吐量也提升了一个数量级，与其余改进算法相比也有着明显优势。因此DL_RBFT共识算法拥有良好的可扩展性，可以广泛应用于联盟链的各种场景中。

Double layer consensus algorithm based on improved Raft consensus algorithm and PBFT

This paper proposed a dual-layer consensus algorithm(DL_RBFT) based on the enhanced Raft and practical Byzantine fault tolerance(PBFT) consensus algorithms to address scalability issues, excessive communication overhead, and challenges in large-scale network node environments in use within consortium blockchains. Initially, blockchain nodes were divided into several groups to form a lower-layer consensus network. Then, leaders from these groups constituted an upper-layer consensus network, creating a dual-layer consensus structure. Within the lower-layer consensus network, it cmployed an improved Raft consensus algorithm with introduced supervision and reputation mechanisms, while the leader election processed incorporates ant colony optimization to maintain high efficiency. In the upper-layer consensus network, the PBFT consensus algorithm was utilized for consensus. The enhanced Raft consensus algorithm exhibited resistance to Byzantine node attacks, enhancing the algorithm''s security. Experimental results indicate that compared to traditional PBFT consensus algorithms, DL_RBFT reduces consensus latency by two orders of magnitude and improves throughput by one order of magnitude in a 100-node scenario. In comparison to other enhanced algorithms, DL_RBFT demonstrates significant advantages. Therefore, DL_RBFT consensus algorithm exhibits strong scalability and can be widely applied in various consortium blockchain scenarios.