Research on Blockchain Consensus Protocols: New Advances in Asynchronous BFT

Blockchain technology, as a decentralized distributed ledger technology, is fundamentally based on consensus mechanisms. The performance of the consensus mechanism directly affects the scalability and security of the blockchain. Recently, Asynchronous Byzantine Fault Tolerance (BFT) consensus mechanisms have become a research hotspot due to their advantages in handling network delays and node failures.

This study explores the latest developments in blockchain consensus protocols, focusing on the progress of the Asynchronous Byzantine Fault Tolerant State Machine Replication (BFT SMR) protocol. Currently, sMVBA is the fastest asynchronous MVBA protocol, with an expected delay of 10δ. The study also proposes two new protocol designs: 2PAC (2-Phase Asynchronous Consensus) and Ultra-Fast Pipelined Blocks, which demonstrate significant improvements in throughput and delay.

In the asynchronous BFT model, the system consists of n = 3f + 1 processes, where f processes may be maliciously compromised. The processes communicate through asynchronous channels, and the message delivery delay is uncontrollable. Each process has a public-private key pair to ensure the authenticity and integrity of the messages.

The goal of the Blockchain Consensus Protocol is to ensure that all honest nodes reach an agreement on the state of the blockchain. It needs to meet basic requirements such as liveness, consistency, and P-quality. However, the currently fastest asynchronous consensus protocol, 2-chain VABA, has multiple potential attacks that affect its consistency and liveness.

Based on the analysis of existing protocols, the research proposes the 2PAC protocol, which includes two variants: 2PAClean and 2PACBIG. 2PAClean achieves over 90% throughput and an expected latency of 9.5δ, with a message complexity of O(n²). 2PACBIG is currently the fastest blockchain consensus protocol with a message complexity of O(n³), with a single MVBA runtime of only 4δ under fault-free conditions.

In addition, the research proposed a super-fast pipelined block design, which significantly reduces the latency of pipelined blocks. Under a fair scheduler, the decision time for pipelined blocks may even be less than that of non-pipelined blocks.

Through theoretical analysis and practical testing, the expected latency of 2PAClean in the worst case is 9.5δ, and in the good case, it is 6δ. In comparison, the expected latency of sMVBA is 10δ, and in the good case, it is 6δ. The throughput of 2PAClean is improved by 80% to 100% compared to the chain-based sMVBA. The single MVBA run time of 2PACBIG is 4δ, which is faster than existing protocols. The ultra-fast pipelined block design allows s2PAClean and s2PACBIG to achieve pipelined block decision times of 4δ and 3δ, respectively.

The evaluation shows that 2PAClean and 2PACBIG perform excellently under various network conditions, especially in high-latency and high-failure-rate environments. 2PAClean strikes a good balance between message delivery latency and computational complexity, while 2PACBIG achieves lower latency through parallelization and optimization of the voting process.

Future research directions include: further optimizing the protocol structure to reduce message transmission and computational overhead; deeply analyzing the security of the new protocol under various attack scenarios; applying the new protocol to actual Blockchain systems to validate its performance in real network environments.

These new protocols are designed to provide new ideas for efficient and secure Blockchain consensus mechanisms, and are expected to promote the more important role of Blockchain technology in the future digital economy. With the deepening of research and ongoing optimization, the new generation of consensus protocols will lay a solid foundation for the development of Blockchain technology.