Practical Byzantine Fault Tolerance (PBFT) lacks built-in incentives. It cannot mobilise node enthusiasm effectively. Master-node election is crude, insecure and improperly updated. This paper proposes a PBFT algorithm based on dynamic reputation evaluation and master-node term limits. The goal is to reward honest nodes for active consensus participation while ensuring every node can serve as master node. The improved algorithm formulates a reputation score from historical behaviour, consensus attendance and penalty count. The node with the highest reputation becomes master node, guaranteeing fairness. The algorithm also introduces a master-node term mechanism. It limits consecutive master-node selections. Every node thus gains a fair chance to be elected, reducing system centralisation. Experiments show that under low load the improved PBFT throughput rises 15–37% above original PBFT. Under high load the advantage widens to 75%. Average latency drops 30–60% across all load levels. The enhanced protocol boosts throughput, sustains node motivation and curbs master-node centralisation. This supports long-term stable operation.
Qin et al. (2026) studied this question.