Lei Wei, Xiao Peng, Tt, Guannan Zhang, Chenhao Jiang, Hongyu Li, Lanbo Lin, Yuanwu Xu, Jiayao Liu, Kesu Wang, Bin Wang. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

As the requirements of different application scenarios vary, numerous blockchains have been developed. Among them, chains using Byzantine Fault Tolerance consensus (BFT chains for short) occupy a dominant position. Performance modeling is the most common method for guiding chain design and configuration. However, many lack specific models due to the large number of BFT chains. Providing a general framework for them is promising and should demonstrate two important features that BFT chains may differ in: different consensus stages and different packaging rules. This paper is the first to propose a general framework applicable to various BFT chains. We use the bulk-service model in queuing theory to reflect the above two features of various BFT chains and provide calculation methods for them. We give an expression for the confirmation delay of transactions (TXs for short) in the chain and verify it through experiments on various BFT chains. We also use the model for a brief analysis to assist designers in configuring and designing BFT chains.

With the emergence of various application scenarios, various chains have been developed to meet their requirements. Among them, chains with fixed block intervals (fixed chains for short) occupy an increasingly significant position. Performance has always been a key bottleneck of blockchains and modeling for them is the most common method for performance analysis. But til now, few models for fixed chains exist and they are not precise and applicable enough. This paper proposes a model for fixed chains via the bulk-service queuing theory, which can reflect the real scenario more precisely and apply to the high load. We consider the continuous time and the transaction (TX for short) pool with limited capacity and reflect the chains’ features of fixed intervals and empty blocks to improve accuracy and applicability. We give an expression for three significant measurements: the average confirmation delay of TXs, the blockchain throughput, and the TX rejection rate. We use Ethereum to validate our model. And moreover, we use the model for analysis to assist designers in operating chains.