CBG-PBFT: Grouped Credit-Driven Byzantine Fault Tolerance Algorithm for Consortium Blockchain

Abstract

To address issues such as high communication overhead in classical PBFT on consortium chains, the lack of behavioral criteria for primary node selection, and inflexible governance of faulty nodes, this paper proposes a packet-based reputation-driven Byzantine fault tolerance algorithm. The algorithm constructs a node reputation model based on three dimensions—effective response performance, transaction completion rate, and online activity—and groups nodes according to their reputation scores, establishing a hierarchical consensus process comprising "in-group consensus, inter-group verification, and global confirmation." Additionally, the algorithm employs a dynamic node management mechanism to downgrade, isolate, and recover faulty nodes. Simulation experiments demonstrate that across a range of 4 to 24 nodes, CBG-PBFT reduces consensus latency compared to classical PBFT while exhibiting greater stability in throughput and robustness.