Paper 2025/877

Towards Improving Throughput and Scalability of DAG-based BFT SMR

Nibesh Shrestha, Supra Research
Aniket Kate, Purdue University/Supra Research
Abstract

Directed Acyclic Graph (DAG)-based BFT consensus protocols often suffer from limited throughput and scalability due to bandwidth-intensive data replication to all participants. However, it is sufficient to replicate data to a smaller sub-committee of parties that holds an honest majority with high probability. In this work, we introduce tribe-assisted reliable broadcast, a novel primitive that ensures reliable broadcast (RBC) properties within a smaller honest-majority sub-committee—referred to as a clan—drawn from the entire network, called the tribe. Leveraging this primitive, we develop two efficient DAG-based BFT consensus protocols. First, we present a single-clan protocol, in which a single clan is elected from the tribe, and data is disseminated exclusively to this designated clan using tribe-assisted RBC. We then extend this design to a multi-clan setting, where multiple clans are elected and data is distributed within each respective clan via the same mechanism. Experimental results demonstrate that both protocols offer substantial improvements in throughput and latency over existing DAG-based BFT protocols, even at moderately large scales.

Metadata
Available format(s)
PDF
Category
Foundations
Publication info
Preprint.
Keywords
DAG-based BFTsub-committeesthroughputscalability
Contact author(s)
n shrestha @ supraoracles com
aniket @ purdue edu
History
2025-05-19: approved
2025-05-17: received
See all versions
Short URL
https://ia.cr/2025/877
License
Creative Commons Attribution-NonCommercial
CC BY-NC

BibTeX

@misc{cryptoeprint:2025/877,
      author = {Nibesh Shrestha and Aniket Kate},
      title = {Towards Improving Throughput and Scalability of {DAG}-based {BFT} {SMR}},
      howpublished = {Cryptology {ePrint} Archive, Paper 2025/877},
      year = {2025},
      url = {https://eprint.iacr.org/2025/877}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.