Blockchain technology provides tamper-resistant and collusion-resistant features for existing cloud storage and cloud computing systems, significantly meeting the business requirements of cross-domain collaboration. However, current consensus protocols exhibit deficiencies in high scalability and high concurrency, particularly due to serialized consensus mechanisms that underutilize node capabilities and involve complex processes. While existing sharding solutions enhance performance, they compromise security. To address the scalability and concurrency challenges of blockchain consensus protocols in cloud service scenarios, this paper proposes an innovative parallelized consensus protocol named ParallelBFT. It utilizes a reputation-based dynamic grouping mechanism to dynamically select consensus nodes, ordering nodes, and voting nodes for each consensus phase. Additionally, it decouples stages such as transaction ordering, transaction validation, and signature collection into independent pipelines. By adopting a parallel asynchronous pipelined consensus approach, the protocol significantly enhances consensus performance while mitigating the risk of single consensus group failures, thereby improving overall security. Experimental analysis demonstrates that under sufficient node performance, the protocol fully utilizes resources to achieve high-concurrency transaction processing, outperforming serialized consensus protocols and existing parallel solutions with 30% and 15% respectively.
Tianxin et al. (2026) studied this question.