概率信任驱动的分层异步BFT协议OA

To address the challenges of node heterogeneity,dynamic communication,and frequent failures in decentralized physical infrastructure networks,a probabilistic trust-driven hierarchical asynchronous Byzantine Fault Tolerance(PT-HABFT)protocol is proposed.The proposed protocol integrates a Gaussian-Bayesian reputation model that combines node geolocation,network latency,and historical behavior into multidimen-sional feature vectors to achieve dynamic trust evaluation and access control.An asynchronous hierarchical structure with decoupled sub-layer local consensus and upper-layer global aggregation is adopted,and a pipe-line verification mechanism is designed to support multi-stage concurrent processing.Furthermore,a reputa-tion-driven weight adjustment and view-switching strategy is developed to enhance robustness.Theoretical a-nalysis demonstrates that the proposed protocol satisfies both consistency and liveness under Byzantine fault tolerance assumptions.Experiments on the Bamboo framework show that,with 128 nodes,PT-HABFT a-chieves approximately eightfold increase throughput and 58％lower communication overhead compared with the HotStuff protocol.