Research On Blockchain Throughput Optimization Based On Improved DAG And PBFT

Blockchain is popular because of its technical advantages such as decentralization,privacy encryption,security,reliability and distribution.However,the influx of large numbers of users has also exposed the drawback of insufficient throughput performance of blockchain.It has been proposed to use DAG(Directed Acyclic Graph)and PBFT(Practical Byzantine Fault Tolerance)algorithms to improve the throughput performance of blockchain,but there are the following problems: the adoption of DAG structure will make the serialization of blockchain transactions complicated and the problem of double spending is difficult to solve;the PBFT algorithm requires consensus nodes to be safe and reliable,so the addition or withdrawal of consensus nodes need to be reinitialized.Based on this,this paper proposes a data structure model based on self-referencing DAG and a PBFT algorithm based on votes to improve throughput performance.The main research contents of this paper are as follows:(1)In terms of data structure,a DAG based on self-referencing is proposed in this paper.Each vertex only stores one transaction in the graph and the storage of single transaction can greatly increase the concurrency of system.Each vertex has several in-degrees and two out-degrees.The in-degrees are references of other transactions to the transaction.The two out-degrees are a reference to the previous transaction of itself and a reference to other transactions in the network.In this DAG model,the self-referencing method can make a single user's transaction block linearly connected and can also allow the entire DAG network to execute transactions in parallel on the premise of effectively avoiding double spending,thereby improving the throughput performance of blockchain.(2)In terms of consensus algorithm,the vote-based PBFT consensus algorithm is proposed in this paper,which can reach a consensus more quickly and efficiently and maintain the stability of consensus nodes and make the algorithm also applicable in public blockchain networks.The vote-based PBFT consensus algorithm includes vote process and consensus process.The vote process selects representative consensus nodes through the vote algorithm.During the verification process,the consensus nodes verify the transactions in the blockchain through the consensus verificationalgorithm.The consensus reached between the consensus nodes represents the final consensus of the blockchain.The proposed algorithm solves the number and stability requirements of the consensus nodes in the PBFT algorithm and ensures the safety and reliability of the consensus nodes.At the same time,the algorithm can effectively improve the throughput performance of blockchain.(3)Based on the Hyperledger Fabric open source framework,this paper proposes a blockchain throughput optimization framework based on improved DAG and PBFT by combining the above two optimization models.The framework can further improve the throughput performance while ensuring the security and reliability of blockchain.In this paper,an experiment is performed on the framework.Performance and reliability analysis are used to evaluate the throughput and security of the framework and the Hyperledger Fabric framework is added as a control group in the experiment.Experiments show that the throughput performance of this framework reaches to between 600 and 1400 which is better than the control group.It demonstrates that the proposed framework can effectively improve the throughput performance of blockchain.And the reliability of the framework is verified through reliability analysis.