实用拜占庭容错算法的改进研究

针对实用拜占庭容错算法（PBFT）存在的通信复杂度高、主节点选取简单、对拜占庭节点缺乏惩罚机制的不足，提出了一种基于节点可靠性评估的改进拜占庭容错算法（reliability-based Byzantine fault tolerant algorithm，RB-PBFT），引入节点基础配置评分机制及信誉评分机制，得到各节点的可靠性评分，评估节点的可靠性并将各节点标记为诚实、故障、恶意三种不同信任状态。根据节点的可靠性评分选取主节点并组建共识群组参与共识，以减少参与共识过程的节点数目，降低通信复杂度，提高系统效率。根据节点的不同信任状态设置节点管控机制，对节点进行分类处理，解决缺乏恶意节点惩罚机制的问题。实验表明，RB-PBFT算法较于PBFT算法，在算法通信复杂度、安全性、公平性及容错性等方面均有一定提升。

Improved Study of Practical Byzantine Fault-Tolerant Algorithm

Aiming at overcoming the shortcomings of the practical Byzantine fault tolerant algorithm, such as high communication complexity, simple primary node selection policy and lack of a penalty mechanism for Byzantine nodes, an improved reliability-based Byzantine fault tolerant algorithm（RB-PBFT） is proposed. The improved algorithm introduces a computerized system. This computerized system is used to calculate the reliability scores of the nodes. Based on this score, the reliability of nodes is evaluated and each node is marked with three different trust states：honest, faulty, and malicious. Then, the primary node is selected and a consensus group is formed based on the reliability score of the nodes. Only the nodes in the consensus group can participate in the consensus. By reducing the number of nodes participating in the consensus process, the communication complexity is reduced and the system efficiency is improved. Finally, the node control mechanism is set according to the different trust states of the nodes. This mechanism classifies nodes and solves the problem of lack of penalty mechanism for malicious nodes. Experiments show that the RB-PBFT algorithm has improved in terms of algorithm communication complexity, security, fairness and fault tolerance compared with the PBFT algorithm.