FP‐BFT: A fast pipeline Byzantine consensus algorithm

• Published in: IET blockchain 2023-09, Vol.3 (3), p.123-135
• Main Authors: Liu, Xinlei, Liu, Yang, Li, Xiangyang, Cao, Haohao, Wang, Yaoqi
• Format: Article
• Language: English
• Subjects: Algorithms; Blockchain; Blockchains; Byzantine fault‐tolerance; Committees; Communication; Communications systems; Consensus algorithm; Efficiency; Energy consumption; Fault tolerance; Hash ring network; Nodes; Pipeline framework; Transaction processing
• ISSN: 2634-1573; 2634-1573
• DOI: 10.1049/blc2.12030

The Byzantine Fault Tolerant (BFT) consensus algorithms have been widely applied in the blockchain systems because of their fault tolerance capability to determine system consistency in the presence of malicious nodes. However, the BFT consensus algorithms are confronted with low efficiency and scalability problems caused by multiple rounds of handshake communication. In this paper, a pipeline‐based Fast Pipeline Byzantine Fault Tolerance consensus algorithm (FP‐BFT) is proposed, which adopts a non‐leader pipeline framework to process different rounds of transactions in parallel. By means of randomly selecting 2f+1 nodes to form a committee for one round of transactions, consensus agreement can be reached within the committee via nodes broadcasting and voting. Committee nodes participating in the consensus are chosen by chance to avoid the monopoly of which becomes the block producer. Consensus efficiency and the system throughput can be significantly improved with the pipeline framework. Comparison experiments are conducted to verify the superiority of the FP‐BFT algorithm, and the theoretical proof is given to guarantee the Byzantine fault‐tolerant security. Experimental results show that FP‐BFT has improved the consensus efficiency by decreasing communication overhead to make it better applied both in public blockchain and consortium blockchain systems. In this paper, we propose a pipeline‐based Byzantine fault‐tolerant consensus algorithm FP‐BFT to demonstrate its improvement in throughput, consensus efficiency, and scalability by comparison with the other BFT consensus algorithms including PBFT, BFT‐SMaRt, and HotStuff. Compared with the classical BFT algorithms such as PBFT and BFT‐SMaRt, the proposed FP‐BFT algorithm has significantly improved the throughput due to its pipeline‐based parallel architecture, which has surpassed the performance of traditional BFT algorithms.