An Improved Hashgraph Consensus Algorithm Based on Sharding Technology

Hashgraph, as one of the most popular DAG-based consensus algorithms, has the potential to overcome the throughput bottleneck of blockchain technology with its efficient consensus mechanism. However, the consensus process of Hashgraph suffers from issues such as multiple steps, complex procedures, poor stability, and duplicate transaction packaging. To address these problems, this paper proposes some solutions. Firstly, this model integrates sharding concepts to optimize the gossip communication protocol. It divides the network into sub-networks based on the comprehensive reputation evaluation value(CREV) of nodes, reducing the number of neighboring nodes and enhancing the efficiency of event propagation by mitigating blind communication. Secondly, within each sharding, a fair and random election of a leader node is conducted using the comprehensive reputation evaluation value to govern the consensus process. This reduces the steps of the algorithm and improves consensus efficiency while ensuring security. Lastly, a transaction sending mechanism based on Verifiable Random Function(VRF) is proposed to prevent duplicate transactions from entering events and improve the system’s effective throughput. Experimental results demonstrate that this model outperforms Hashgraph in terms of effective throughput, latency, and event propagation efficiency.