The Consensus algorithm is the core of the permissioned blockchain, it directly affects the performance and scalability of the system. Performance is limited by the computing power and network bandwidth of a single leader node. Most existing blockchain systems adopt mesh or star topology. Blockchain performance decreases rapidly as the number of nodes increases. To solve this problem, we first design the n-k cluster tree and corresponding generation algorithm, which supports rapid reconfiguration of nodes. Then we propose the Zebra consensus algorithm, which is a cluster tree-based consensus algorithm. Compared to the PBFT, it has higher throughput and supports more nodes under the same hardware conditions. However, the tree network topology enhances scalability while also increasing latency among nodes. To reduce transaction latency, we designed the Pipeline-Zebra consensus algorithm that further improves the performance of blockchain systems in a tree network topology through parallel message propagation and block validation. The message complexity of the algorithm is O(n). Experimental results show that the performance of the algorithm proposed in this paper can reach 2.25 times that of the PBFT algorithm, and it supports four times the number of nodes under the same hardware.

中文翻译：

---

Zebra：一种集群感知的区块链共识算法


共识算法是许可区块链的核心，它直接影响系统的性能和可扩展性。性能受单个领导节点的计算能力和网络带宽的限制。大多数现有的区块链系统采用网状或星形拓扑。随着节点数量的增加，区块链性能会迅速下降。为了解决这个问题，我们首先设计了 n-k 集群树和相应的生成算法，它支持节点的快速重新配置。然后我们提出了 Zebra 共识算法，这是一种基于集群树的共识算法。与 PBFT 相比，它具有更高的吞吐量，并且在相同的硬件条件下支持更多的节点。但是，树状网络拓扑增强了可扩展性，但也增加了节点之间的延迟。为了减少交易延迟，我们设计了 Pipeline-Zebra 共识算法，该算法通过并行消息传播和区块验证进一步提高了树形网络拓扑中区块链系统的性能。该算法的消息复杂度为 O（n）。实验结果表明，本文提出的算法性能可以达到PBFT算法的2.25倍，并且在相同硬件下支持4倍的节点数。