Performance Optimization Of PBFT Consensus Algorithm Based On C4.5 Decision Tree In Blockchain

In the traditional centralized data management mode,the data is completely managed by "trusted" third parties,which are prone to security problems such as data fraud and tampering.The use of blockchain multi-centered,traceable,difficult to tamper and other original characteristics to achieve reliable management of data.However,in the actual blockchain system application,it is found that the consortium chain based on the PBFT consensus algorithm is not well satisfied with the performance requirements of the application scenario.Therefore,it is of great significance and value to research the performance of PBFT consensus algorithm in the context of the Alliance Chain Hyperledger Fabric open source platform.We will explore the implementation of the PBFT consensus algorithm in the blockchain,and based on this,adopt an algorithm optimization strategy suitable for industrial application requirements.The main work of this paper has the following four points:(1)Firstly,take the Hyperledger Fabric as an example to learn the blockchain concept and data structure.Secondly,the blockchain architecture is discussed in depth,and the concept of Docker container in smart contract operation sandbox environment is introduced.Finally,the principle and shortcomings of the PBFT consensus algorithm are studied in detail,and compared with other blockchain consensus algorithms.(2)For the Hyperledger Fabric blockchain based on the PBFT consensus algorithm,the non-honesty node in the consortium chain scenario is the master node problem.We propose the C4.5 decision tree with high model classification accuracy rate in the blockchain network.The consensus nodes perform a trust evaluation to reduce the non-honest node as the primary node.(3)In order to solve the Hyperledger Fabric blockchain based on the PBFT consensus algorithm in the consortium chain scenario,the consensus nodes need to consume a large amount of network bandwidth resources when performing message broadcast.Based on the evaluation of the trust degree of consensus nodes in the blockchain network,we propose the concept of introducing voting weights.A consensus can be reached only by considering a small number of trusted nodes voting weights,thereby reducing the number of messages broadcast on the network.(4)The PBFT consensus algorithm before and after optimization is applied to the simulation test platform of Supply Chain Financial consortium chain of Hyperledger Fabric.The experimental results show that the throughput,delay,fault tolerance and other performance of the optimized algorithm are significantly improved.In summary,we can provide a new ideas and technical support for enterprises to solve the problem of blockchain application system performance.