Let It TEE: Asynchronous Byzantine Atomic Broadcast with $n\ge 2f+1$

Abstract

Asynchronous Byzantine Atomic Broadcast (ABAB) promises simplicity in implementation as well as increased performance and robustness in comparison to partially synchronous approaches. We adapt the recently proposed DAG-Rider approach to achieve ABAB with $n\ge 2f+1$ processes, of which $f$ are faulty, with only a constant increase in message size. We leverage a small Trusted Execution Environment (TEE) that provides a unique sequential identifier generator (USIG) to implement Reliable Broadcast with $n>f$ processes and show that the quorum-critical proofs still hold when adapting the quorum size to $\lfloor \frac{n}{2}\rfloor +1$. This first USIG-based ABAB preserves the simplicity of DAG-Rider and serves as starting point for further research on TEE-based ABAB.