An Adaptive Reputation Update Mechanism for Primary Nodes in PBFT

Since primary nodes play a critical leadership role in PBFT (Practical Byzantine Fault Tolerance) consensus algorithm, it is mandatory to perform an update process for eliminating the malicious primary nodes so as to maintain the security level of PBFT. As it is convenient to calculate node reputations, most of existing updating schemes select the node with the highest reputation as the primary node in every consensus round. However, the round-by-round reputation update mechanism inevitably increases the communication overhead and consensus latency. To reduce the number of update times, we propose an algorithm based on an adaptive reputation mechanism, named Commit Block PBFT or CBPBFT for short. In this mechanism, a node’s reputation is determined by the number of its committed blocks during the most recent period as a primary node. As the reputation update adaptively occurs according to the number of blocks committed by the primary node, the mechanism significantly decreases the frequency of reputation updates, and hence increases the consensus performance. Experimental results show that, CBPBFT reduces the probability of a Byzantine node gaining leadership by 15% and decreases communication latency by 15% to 32% in comparison with the state-of-the art.