Formalizing Accountable Liveness for Provable Consensus Security and Validator Punishment
Introducing Accountable Liveness and the $x$-partially-synchronous model to formally identify and punish consensus-stalling nodes, strengthening BFT security.
Cross-Cluster Consistent Broadcast Enables Efficient Replicated State Machine Interoperability
The new Cross-Cluster Consistent Broadcast (C3B) primitive and PICSOU protocol solve inter-RSM communication, achieving 24x better performance for decentralized systems.
Asymmetric DAG Consensus Achieves Robustness with Heterogeneous Node Trust
By formalizing an asymmetric common core primitive, this new DAG-based consensus protocol enables robust, constant-time finality under heterogeneous trust assumptions.
Epidemic Consensus Protocol Scales Blockchains by Eliminating Fixed Leaders
The Blockchain Epidemic Consensus Protocol (BECP) achieves probabilistic convergence without leaders, fundamentally unlocking massive-scale decentralized network architecture.
Asynchronous Consensus Achieves Low Latency through Optimistic Responsiveness
A new BFT protocol guarantees safety under full asynchrony while achieving synchronous-like speed during periods of network stability, resolving the latency-robustness trade-off.
Asynchronous Partial Vector Agreement Enables Constant-Round Error-Free Byzantine Consensus
Introducing Asynchronous Partial Vector Agreement, a new primitive that enables information-theoretically secure Byzantine consensus with optimal constant-time round complexity.
Adaptive Byzantine Consensus via Decentralized Reinforcement Learning
A reinforcement learning engine enables BFT protocols to dynamically self-optimize, boosting throughput and establishing the first Learned Consensus paradigm.
Hybrid Synchronous Model Unlocks Optimal Low-Latency Byzantine Fault Tolerance
A new BFT protocol leverages a hybrid synchrony model to achieve up to 15x lower latency, preserving high fault tolerance for scalable decentralized systems.
Batch Zero-Knowledge BFT Achieves Scalable Private Federated Learning Consensus
Batch Zero-Knowledge Proofs are integrated into BFT consensus, cutting communication complexity to $O(n)$ and enabling scalable, private decentralized AI.
