🤖 AI Summary
This paper addresses the fundamental trade-off among fault tolerance, communication/computation overhead, round complexity, and liveness in asynchronous Byzantine Fault Tolerant (BFT) consensus. We propose Ocior, the first leaderless, parallelized asynchronous BFT protocol achieving two-round asynchronous finality and optimal resilience (n ≥ 3t + 1). Ocior introduces OciorBLSts—a novel non-interactive threshold signature scheme—and integrates an immediate signature aggregation mechanism. This enables, for the first time in the asynchronous model, linear communication and computation complexity—both O(n)—thereby substantially improving liveness and throughput scalability. The protocol ensures adaptive security under dynamic adversarial conditions while maintaining stable liveness. Experimental evaluation confirms Ocior’s effectiveness in high-concurrency transaction processing environments, demonstrating superior scalability and robustness compared to prior asynchronous BFT designs.
📝 Abstract
In this work, we propose Ocior, a practical asynchronous Byzantine fault-tolerant (BFT) consensus protocol that achieves the optimal performance in resilience, communication, computation, and round complexity. Unlike traditional BFT consensus protocols, Ocior processes incoming transactions individually and concurrently using parallel instances of consensus. While leader-based consensus protocols rely on a designated leader to propose transactions, Ocior is a leaderless consensus protocol that guarantees stable liveness. Ocior achieves: 1) Optimal resilience: Ocior tolerates up to $t$ faulty nodes controlled by an adaptive adversary, for $ngeq 3t+1$. 2) Optimal communication complexity: The total expected communication per transaction is $O(n)$. 3) Optimal (or near-optimal) computation complexity: The total computation per transaction is $O(n)$ in the best case, or $O(n log^2 n)$ in the worst case. 4) Optimal round complexity: A legitimate two-party transaction can be finalized with a good-case latency of two asynchronous rounds, for any $ngeq 3t+1$. The good case in terms of latency refers to the scenario where the transaction is proposed by any (not necessarily designated) honest node. A two-party transaction involves the transfer of digital assets from one user (or group of users) to one or more recipients. To support efficient consensus, we introduce a novel non-interactive threshold signature (TS) scheme called OciorBLSts. It offers fast signature aggregation, and is adaptively secure. OciorBLSts achieves a signature aggregation computation cost of only $O(n)$ for the best case. Moreover, OciorBLSts supports the property of Instantaneous TS Aggregation. This enables real-time aggregation of partial signatures as they arrive, reducing waiting time and improving responsiveness.