​
PoS Consensus with StableBFT

Initially, Stable Blockchain leverages StableBFT, a customized PoS consensus protocol built on CometBFT, to ensure high throughput, low latency, and strong reliability across the network. Its strengths include deterministic finality (ensuring immediate block finality without forks), robust fault tolerance (securing the network even if up to 1/3 of validators fail or act maliciously).

To further optimize consensus performance, Stable is planning to implement the following improvements to these two layers in the near future:

  • Decoupled Transaction and Consensus Gossiping: By separating the transaction gossiping layer and the consensus gossiping layer, it becomes possible to prevent network congestion on the transaction side from interfering with consensus communications.
  • Direct Transaction Broadcasting to the Block Proposer: In the existing model, transaction propagation occurs through peer-to-peer gossiping among nodes, leading to high transaction gossip traffic across the network. Stable aims to improve the efficiency of transaction propagation by enabling transactions to be broadcast directly to the block proposer.

​
Future Roadmap: DAG-based Consensus

To significantly accelerate consensus, Stable intends to upgrade its protocol to a DAG-based design capable of delivering up to 5x speed improvements.

Traditional view-based BFT protocols like PBFT and HotStuff are optimized for low latency under stable network conditions. However, they degrade significantly during disruptions, often experiencing long recovery delays after temporary faults.

First-generation DAG-based engines like Narwhal and Tusk demonstrate that decoupling data dissemination from consensus ordering can eliminate single-proposer bottlenecks and also improve robustness under network instability. However, their architecture is not directly compatible with systems like CometBFT, as they diverge from conventional height-based block semantics and mempool designs.

Autobahn offers a PBFT-on-DAG architecture that integrates more naturally with Stable’s consensus layer, while delivering low latency under normal conditions and fast recovery in the face of network faults. Stable team maintains a close relationship with the authors of the Autobahn paper and will leverage Autobahn’s architecture to maximize the performance of StableBFT.

StableBFT, built atop Autobahn, will enable:

  • Parallel proposal processing by eliminating the single-leader limitation.
  • Faster finality by separating data propagation from final ordering.
  • Enhanced resilience against network adversities through robust BFT mechanisms.

This advanced consensus design supports much higher throughput based on the internal proof-of-concept, which has demonstrated over 200,000 TPS (Consensus only) in controlled environments.