Blockchain technology has become a research hotspot in distributed systems, aiming to sustain a decentralized ledger via consensus. Traditional consensus solutions exhibit slow processing speed and response time, resulting in poor performance. To address this issue, several consensus protocols have been proposed. One such popular protocol is HotStuff, a Byzantine fault-tolerant consensus (BFT) that achieves high throughput at the cost of latency. However, its throughput suffers from a proportional decrease with the increase in latency, posing a significant challenge. In this paper, we propose a new protocol called Dolphin that builds upon HotStuff. It operates in a partially synchronous network with $n$ replicas, up to $f$ byzantine faults, where $n \geq 3f+1$ , and achieves higher throughput in high-latency environments by leveraging non-blocking concurrent block generation. Specifically, we formalize our strategy as a generic Asynchronization Procedure Patch and prove that it does not affect the execution process of the original protocol. Theoretical analysis validates that Dolphin preserves the safety, liveness, and responsiveness properties while enhancing the throughput. The evaluation demonstrates that Dolphin typically achieves more than 10x higher throughput in Wide Area Network (WAN) environments with lower latency compared to HotStuff and its variants, and exhibits similar bandwidth utilization to DAG-based protocols such as Narwhal.

Since the emergence of blockchain, how to improve its transaction throughput and reduce transaction latency has always been an important issue. Hostuff has introduced a pipeline mechanism and combined it with a chain structure to improve the performance of blockchain networks. HCA has introduced a revoting mechanism on the basis of Hostuff, further reducing transaction latency, but it has also brought some problems. In HCA, if the leader is malicious, it would be possible to continuously call on the replica nodes to revote, which can lead to network congestion. This paper employs the global perfect coin technology to guarantee that every replica can obtain a globally consistent and the freshest candidate proposal during the Revote phase, thereby improving the robustness of the HCA protocol. The performance improvement of RHCA in attack scenarios has been verified through experiments.