HML-BFT: Hybrid multi-layer BFT consensus with reputation model for large-scale blockchain

Practical Byzantine Fault Tolerance (PBFT) is the most widely used consensus mechanism in current consortiums and private blockchains. Nevertheless, the poor node scalability of PBFT seriously affects the performance of blockchain. To address this problem, we propose a reputation-based hybrid multi-layer consensus protocol (HML-BFT) by combining two strategies: committee election and hierarchical consensus. In the strategy of committee election, we only make a subset of nodes participate in the consensus, while non-committee members observe the results of the consensus. Besides, we select the consensus committee based on the reputation value of nodes, which can reduce the number of nodes to participate in the consensus, thereby reducing communication complexity. In the strategy of hierarchical consensus, we organize the nodes into different layers, and nodes in different layers perform different consensus processes, thereby improving consensus efficiency. Experimental results show that our protocol greatly reduces the communication complexity of PBFT and enhances its scalability. Through the comparative analysis of experimental data, we find HML-BFT outperforms many algorithms proposed to improve the scalability of PBFT in recent years.