A practical byzantine fault tolerance improvement algorithm based on credit grouping-classification

Blockchain is a distributed system that maintains a public ledger using consensus protocols. The practical byzantine fault tolerance (PBFT) protocol is vital in consortium blockchains. However, there are still problems with system centralization, consensus inefficiency, high communication complexity, and other issues in PBFT and its improved algorithms. These restrict its widespread application in realistic environments. In this work, we propose a grouping-classification practical byzantine fault tolerance (GC-PBFT) improvement algorithm based on credit. First, we construct a credit model to dynamically evaluate the credit of nodes and rank them in descending order. Second, a grouping-classification model is built to divide nodes into three subgroups. Then, primary node, supervisory node, and consensus node are selected from these subgroups randomly, ensuring that the node roles are constantly updated to improve the security and fairness of the system. Additionally, the consensus process is simplified to improve the efficiency of the consensus and reduce communication complexity. Finally, experiments show that compared with PBFT and its improved algorithms, the GC-PBFT algorithm reduces communication complexity from \(O(N^2)\) $O(N^2)$ to O(N). In addition, active and honest nodes can obtain higher credit, which lowers the possibility of malicious nodes being selected as consensus nodes, leading to higher consensus efficiency and better decentralization characteristics for the system.