Double-layer Byzantine fault-tolerant grouping consensus algorithm based on raft

Abstract

Addressing the scalability issues, excessive communication overhead, and challenges in adapting to large-scale network node environments faced by the Practical Byzantine Fault Tolerance (PBFT) consensus algorithm currently employed in consortium blockchains, this paper proposes a Double Layer Consensus Algorithm Based on RAFT and PBFT Consensus Algorithms (DLCA_R_P). The nodes in the blockchain are initially divided into several groups to form the lower-layer consensus network. Subsequently, the leaders of these groups constitute the upper-layer consensus network, creating a dual-layer consensus network structure. Within the lower-layer consensus network, the PBFT consensus algorithm is employed for consensus among the groups, while the primary accountants form the upper-layer RAFT consensus network. The algorithm incorporates a supervision mechanism and a reputation mechanism to enhance the security of the consensus network. Additionally, a grouping mechanism is introduced to transform the consensus network into a dynamic structure. Experimental results analysis demonstrates that compared to traditional PBFT consensus algorithms, DLCA_R_P reduces consensus latency by two orders of magnitude and improves throughput by one order of magnitude in a scenario with 100 nodes. Furthermore, it exhibits significant advantages over other improved algorithms. Thus, the DLCA_R_P consensus algorithm exhibits excellent scalability and can be widely applied in various scenarios within consortium blockchains.