【Author】
Yang, Jian; Jia, Zhenhong; Su, Ruiguo; Wu, Xiaoxiong; Qin, Jiwei
【Abstract】Nowadays Practical Byzantine Fault Tolerance (PBFT) algorithm has become the most extensive consensus algorithm in the alliance chain. However, the PBFT algorithm is usually only applicable to small networks due to high communication complexity and poor scalability. Although there have been many improved algorithms for PBFT in recent years, they ignore fault tolerance and democracy. Therefore, to meet the requirements of a high degree of decentralization and fault tolerance of blockchain-based scenarios. This paper proposes a high fault tolerance consensus algorithm NBFT, which follows the principle of decentralization and democratization of blockchain and ensures the improvement of performance in fault tolerance upper limit and scalability. First, we use the consistent hash algorithm to group the consensus nodes to avoid much communication between nodes, reduce the communication complexity of the network, and improve the scalability of the network. Second, to ensure the fault-tolerant ability of the grouping consensus, the nodal decision broadcast model and threshold vote-counting model are proposed first. Combined with the proposed two models, the joint fault analysis of nodes is carried out, and the fault tolerance upper limit is more than 1/3. Then, the Faulty Number Determined (FND) model is introduced to simulate the experiment, and the results are verified.
【Keywords】Fault tolerant systems; Fault tolerance; Scalability; Complexity theory; Blockchains; Consensus algorithm; Analytical models; High fault tolerance; group consensus; scalability
【摘要】如今，实用拜占庭容错（PBFT）算法已成为联盟链中最广泛的共识算法。然而，由于通信复杂度高、可扩展性差，PBFT算法通常只适用于小型网络。尽管近年来 PBFT 有许多改进的算法，但它们忽略了容错性和民主性。因此，要满足基于区块链的场景对高度去中心化和容错的要求。本文提出了一种高容错共识算法NBFT，该算法遵循区块链去中心化和民主化的原则，保证了容错上限和可扩展性方面的性能提升。首先，我们使用一致性哈希算法对共识节点进行分组，避免节点之间过多的通信，降低网络的通信复杂度，提高网络的可扩展性。其次，为了保证分组共识的容错能力，首先提出了节点决策广播模型和阈值计票模型。结合提出的两种模型，对节点进行联合故障分析，容错上限为1/3以上。然后，引入故障数确定（FND）模型对实验进行仿真，并对结果进行了验证。
【关键词】容错系统；容错；可扩展性；复杂性理论；区块链；共识算法；分析模型；容错性高；团体共识；可扩展性
评论