Lemonshark: Asynchronous DAG-BFT With Early Finality

2026-04-05

AI Summary

This work addresses the high latency inherent in existing DAG-based BFT protocols, where non-leader blocks cannot be independently committed. The paper proposes an asynchronous DAG-BFT protocol that reinterprets the DAG structure at the transaction granularity to identify conditions under which transactions can safely achieve finality before formal block commitment. By integrating a DAG-based asynchronous Byzantine fault-tolerant mechanism, transaction-level finality determination, and a novel leader election and block referencing model, the protocol achieves—without compromising correctness—the first instance of early transaction finality within non-leader blocks. Experimental results demonstrate that this approach reduces end-to-end latency by up to 65% compared to the state-of-the-art asynchronous BFT protocols.

Abstract

DAG-Rider popularized a new paradigm of DAG-BFT protocols, separating dissemination from consensus: all nodes disseminate transactions as blocks that reference previously known blocks, while consensus is reached by electing certain blocks as leaders. This design yields high throughput but confers optimal latency only to leader blocks; non-leader blocks cannot be committed independently. We present Lemonshark, an asynchronous DAG-BFT protocol that reinterprets the DAG at a transactional level and identifies conditions where commitment is sufficient -- but not necessary -- for safe results, enabling nodes to finalize transactions before official commitment, without compromising correctness. Compared to the state-of-the-art asynchronous BFT protocol, Lemonshark reduces latency by up to 65\%.

Problem

• DAG-BFT
• asynchronous consensus
• transaction finality
• latency
• Byzantine fault tolerance

Innovation

• asynchronous BFT
• DAG-based consensus
• early finality
• transaction-level DAG
• low latency