Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BBCA-CHAIN: Low Latency, High Throughput BFT Consensus on a DAG

Dahlia Malkhi, Chrysoula Stathakopoulou, Maofan Yin

TL;DR

BBCA-CHAIN tackles the latency-throughput trade-off in DAG-based BFT by introducing BBCA, an abortable Byzantine broadcast with a Complete-Adopt probe, and by leveraging a causal DAG backbone to commit blocks directly. It eliminates vote blocks and enables parallel leader proposals, while decoupling data availability from consensus to improve scalability. The approach achieves high throughput with reduced latency relative to existing DAG protocols and provides formal correctness arguments for safety and liveness under partial synchrony. The work offers a concrete protocol design, pseudocode, and a solid foundation for high-performance replication in blockchain and distributed systems.

Abstract

This paper presents a partially synchronous BFT consensus protocol powered by BBCA, a lightly modified Byzantine Consistent Broadcast (BCB) primitive. BBCA provides a Complete-Adopt semantic through an added probing interface to allow either aborting the broadcast by correct nodes or exclusively, adopting the message consistently in case of a potential delivery. It does not introduce any extra types of messages or additional communication costs to BCB. BBCA is harnessed into BBCA-CHAIN to make direct commits on a chained backbone of a causally ordered graph of blocks, without any additional voting blocks or artificial layering. With the help of Complete-Adopt, the additional knowledge gained from the underlying BCB completely removes the voting latency in popular DAG-based protocols. At the same time, causal ordering allows nodes to propose blocks in parallel and achieve high throughput. BBCA-CHAIN thus closes up the gap between protocols built by consistent broadcasts (e.g., Bullshark) to those without such an abstraction (e.g., PBFT/HotStuff), emphasizing their shared fundamental principles. Using a Bracha-style BCB as an example, we fully specify BBCA-CHAIN with simplicity, serving as a solid basis for high-performance replication systems (and blockchains).

BBCA-CHAIN: Low Latency, High Throughput BFT Consensus on a DAG

TL;DR

BBCA-CHAIN tackles the latency-throughput trade-off in DAG-based BFT by introducing BBCA, an abortable Byzantine broadcast with a Complete-Adopt probe, and by leveraging a causal DAG backbone to commit blocks directly. It eliminates vote blocks and enables parallel leader proposals, while decoupling data availability from consensus to improve scalability. The approach achieves high throughput with reduced latency relative to existing DAG protocols and provides formal correctness arguments for safety and liveness under partial synchrony. The work offers a concrete protocol design, pseudocode, and a solid foundation for high-performance replication in blockchain and distributed systems.

Abstract

This paper presents a partially synchronous BFT consensus protocol powered by BBCA, a lightly modified Byzantine Consistent Broadcast (BCB) primitive. BBCA provides a Complete-Adopt semantic through an added probing interface to allow either aborting the broadcast by correct nodes or exclusively, adopting the message consistently in case of a potential delivery. It does not introduce any extra types of messages or additional communication costs to BCB. BBCA is harnessed into BBCA-CHAIN to make direct commits on a chained backbone of a causally ordered graph of blocks, without any additional voting blocks or artificial layering. With the help of Complete-Adopt, the additional knowledge gained from the underlying BCB completely removes the voting latency in popular DAG-based protocols. At the same time, causal ordering allows nodes to propose blocks in parallel and achieve high throughput. BBCA-CHAIN thus closes up the gap between protocols built by consistent broadcasts (e.g., Bullshark) to those without such an abstraction (e.g., PBFT/HotStuff), emphasizing their shared fundamental principles. Using a Bracha-style BCB as an example, we fully specify BBCA-CHAIN with simplicity, serving as a solid basis for high-performance replication systems (and blockchains).
Paper Structure (27 sections, 14 theorems, 4 figures, 2 tables)

This paper contains 27 sections, 14 theorems, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. The Benefits of Causal Ordering.
  3. Technical Overview
  4. Consensus.
  5. Parallel Leader Proposals.
  6. Causal Broadcast.
  7. The Backbone.
  8. Separating Data Availability.
  9. BBCA
  10. The Complete-Adopt Invariant.
  11. BBCA-Chain
  12. The BBCA Primitive
  13. Relationship to Other Broadcasts
  14. The BBCA-Chain Protocol
  15. Model
  16. ...and 12 more sections

Key Result

lemma thmcounterlemma

If a correct sender BBCA-broadcasts a message $m$ for instance $\mathit{bid}$, and no correct node probes $\mathit{bid}$, eventually every correct node BBCA-completes $m$.

Figures (4)

  • Figure 1: A sequenced backbone on a DAG (thick arrows and solid blocks) and referenced blocks that commit with it (thin arrows and hollow blocks).
  • Figure 2: Reducing Bullshark commit latency (left) via three mechanisms.
  • Figure 3: Decision making latency in BBCA-Chain (left) compared with Bullshark (right).
  • Figure 4: When a view timer expires, each node needs to include a BBCA-probe result inside its new-view block. The left DAG depicts an adopted block, and some correct node may have just BBCA-completed it. The right DAG shows $2f+1$ noadopt responses, proving that no block can be completed in that view.

Theorems & Definitions (28)

  • lemma thmcounterlemma: Validty
  • proof
  • lemma thmcounterlemma: Consistency
  • proof
  • lemma thmcounterlemma: Complete-Adopt
  • proof
  • lemma thmcounterlemma: Integrity
  • proof
  • lemma thmcounterlemma
  • proof
  • ...and 18 more