Research On Blockchain Consensus Algorithm Based On Practical Byzantine Fault Tolerance

In recent years,blockchain technology has been widely researched at home and abroad.With the development of blockchain technology,its application has also expanded from finance to other fields such as transportation,energy,and medical treatment.As the core of blockchain,the consensus algorithm defines the rules for consensus among nodes on the data state,and is the key to determine the performance of blockchain system.However,current consensus mechanism has increasingly prominent problems in resource consumption,time delay,throughput and other aspects,which affects the development of blockchain technology.Therefore,this paper studies the Practical Byzantine Fault Tolerant(PBFT)consensus algorithm in blockchain.The main work is as follows:Aiming at the lack of incentive mechanism and punishment mechanism in PBFT algorithm,and the system performance is easily affected by faulty nodes and Byzantine nodes,a dynamic trust model is proposed.Based on Bayesian theory,the proposed model evaluates the trust of nodes according to the historical behavior evaluation and current behavior evaluation among nodes,and predicts the subsequent behavior of nodes.Moreover,it designs two mechanisms including a trust consumption mechanism to ensure the participation of nodes and a trust recovery mechanism to give low trust nodes the opportunity to participate in consensus again and accumulate trust,so as to improve the enthusiasm of nodes.In order to reduce the influence of failed nodes and Byzantine nodes on the consensus process,a trust hierarchy mechanism is introduced to assign different levels and permissions to nodes according to their trust degrees.Experimental results show that the model can effectively eliminate Byzantine nodes and improve the reliability of the system.Aiming at the problem of PBFT algorithm that the traffic in the network increases sharply and the consensus efficiency decreases rapidly with the increase of the number of nodes,a group Byzantine Fault Tolerant algorithm based on dynamic trust model(DGBFT)is proposed.The algorithm uses the dynamic trust model to evaluate the trust degree of nodes,improve the rules of selecting primary nodes in polling,and improve the stability of the system.The nodes in the network are randomly grouped by hash modulus to form a multi-center hierarchical network structure.Consensus is carried out within and between groups respectively and consensus results are transmitted layer by layer from bottom to top.By dividing the consensus tasks into each group,communication traffic in the network is reduced and the efficiency of consensus is improved.Experimental results show that the proposed algorithm effectively reduces communication overhead,reduces consensus delay,improves throughput and has good stability.