About CipherBFT
First Open-Source Implementation of Autobahn BFT
Project Overview
CipherBFT is a research project aiming to build the first open-source implementation of the Autobahn BFT algorithm. Currently developed by a team of 4 members, this project bridges the gap between academic research (SOSP 2024) and practical implementation.
The goal is to verify the claims of the Autobahn paper and build a production-quality PoC implementation in Rust.
Why High-Performance L1?
The core issue is L2 sequencer centralization. Most L2s use a single sequencer, creating fundamental vulnerabilities in censorship resistance and liveness.
Multi-proposer L1 architecture enables validators distributed across regions to receive transactions simultaneously. This provides structurally faster inclusion than single sequencer solutions.
We are not rejecting L2. We believe L1 level performance improvement is still a valuable research area.
Technical Approach
Current Consensus Limitations
- CometBFT: Stable but leader-based structure creates throughput bottleneck.
- DAG protocols (Bullshark): High throughput (130K+ TPS) but complex commit rules increase latency.
Autobahn attempts to combine the advantages of both approaches.
Autobahn BFT Key Claims
- 2x latency reduction compared to Bullshark
- HotStuff-level latency achievement
- Fast recovery after failures
CipherBFT Architecture
Data Chain Layer (DCL)
Corresponds to Autobahn's Data Layer. Handles mempool management, Car creation, attestation collection, and Cut formation. Uses BLS12-381 signatures for aggregation.
Consensus Layer (CL)
Handles Cut consensus, view management, and finality guarantee. Integrates Malachite BFT (Tendermint's Rust implementation by Informal Systems). Uses Ed25519 for fast individual verification.
Execution Layer (EL)
Uses revm (Rust EVM), a verified implementation used in major Ethereum infrastructure like Reth and Foundry.
Research Scope
This is currently a research project. The goal is Autobahn paper verification and PoC implementation.
Production deployment is a decision to be made after seeing the validation results. It would be irresponsible to promise production at this stage.
Research Partners
Decipher
Academic ResearchThe leading blockchain research group at Seoul National University and one of Korea's premier blockchain academic communities. Bringing cutting-edge research capabilities and academic rigor to the project.
Visit DecipherB-Harvest
Research & DevelopmentA blockchain research and infrastructure company since 2018, contributing to consensus protocol research and development in the Cosmos ecosystem. Deep expertise in BFT consensus mechanisms and distributed systems.
Visit B-HarvestGet Involved
CipherBFT is an open-source research project. We welcome performance analysis, algorithm improvements, documentation contributions, and bug reports.