An Improved Method of Blockchain Consortium Chain Consensus Mechanism Based on Random Forest Model

Abstract

As blockchain technology has attracted more and more attention from all walks of life, the industry application prospects of alliance chains are very broad. Consensus algorithms are very important in blockchain applications. Now the alliance chain mainly uses the PBFT consensus algorithm, but the main node selection step of the algorithm needs to be maintained by all nodes in the alliance chain, which has high consumption and high latency performance and low security. The problem. This paper uses the integrated learning random forest model of machine learning by adding a credit scoring mechanism, taking the characteristic data of some influencing factors of the alliance chain that affect the selection of the master node as input, the selection of the master node as the observation sample, and the training sample to obtain the prediction model. The predicted master node is used to replace the master consensus node selection step in the PBFT algorithm to complete the consensus, and RFBFT (Random Forest Byzantine Fault-Tolerant Algorithm) is proposed. The high accuracy of the experimental random forest prediction model ensures the accuracy and safety of the master node selection. Comparing the algorithms before and after the improvement, the consumption of RFBFT is reduced by 20% and the delay is reduced by 19%, which improves the operation of the alliance chain. Performance and safety.