Let It TEE: Asynchronous Byzantine Atomic Broadcast with n ≥ 2f+1
share
Asynchronous Byzantine Atomic Broadcast (ABAB) promises, in comparison to partially synchronous approaches, simplicity in implementation, increased performance, and increased robustness. For partially synchronous approaches, it is well-known that small Trusted Execution Environments (TEE), e.g., MinBFT's unique sequential identifier generator (USIG), are capable of reducing the communication effort while increasing the fault tolerance. For ABAB, the research community assumes that the use of TEEs increases performance and robustness. However, despite the existence of a fault-model compiler, a concrete TEE-based approach is not directly available yet. In this brief announcement, we show that the recently proposed DAG-Rider approach can be transformed to provide ABAB with n≥ 2f+1 processes, of which f are faulty. We leverage MinBFT's USIG to implement Reliable Broadcast with n>f processes and show that the quorum-critical proofs of DAG-Rider still hold when adapting the quorum size to ⌊n/2⌋ + 1.
READ FULL TEXT