EB-bft:Research On Scalable BFT Consensus Protocol In Block Chain

At present,blockchain technology is more frequently applied to various fields of business,which solves the problems of information isolation and decentralized construction in processing business and enables cross-regional,cross-organizational and cross-departmental business to realize information sharing and transformation However,with the increase of business in blockchain system,performance of the blockchain becomes the bottleneck that limits the business throughput.In the scenario where the trading volume varies greatly and the performance of the nodes in the blockchain network is uneven,how to improve the throughput of blockchain and reduce the transaction delay has become an important research topicPreviously,there have been many studies on increasing throughput through batch processing in distributed systems.Although the blockchain is also distributed,it is still fundamentally different from traditional distributed systems.Researches on improving throughput in distributed systems cannot be copied to that in the blockchain,but there are still some good ideas to learn from.There are also related researches in the field of blockchain on increasing the throughput by adjusting the number of batch consensus,but these studies are limited to proving that throughput can be increased through batch consensus.No one has proposed a specific solution to dynamically adjust the number of batch consensus based on the current physical resource status of the blockchainThis paper proposes a scalable BFT consensus protocol for blockchain system based on improving the PBFT consensus protocol.PBFT protocol can dynamically change the number of participating consensus nodes and the number of batch consensus transactions according to changes of business transaction volume and physical resources.First,the data structure of transactions in the blockchain is changed to meet the needs of current nodes to transfer physical resources to each other.Then,the ARIMA model is used to predict the length of transaction request queue of the master node,and the LSTM model is used to predict the throughput of the blockchain system,as the basis for subsequent dynamic adjustment.According to the performance of the nodes,the short board nodes in the block chain network are identified and their participation in consensus is suspended to improve the efficiency of the blockchain.According to the transaction reception rate of the current master node,the number of consensus nodes is dynamically scaled to improve the block chain throughput.The number of consensus transactions is dynamically scaled according to the physical resources and network resource conditions of nodes in the current blockchain network.Finally,a series of experiments are conducted to verify that the scalable BFT consensus protocol proposed in this paper can adjust the number of nodes participating in consensus and the number of batch consensus transactions in a timely,dynamic and fine-grained manner according to the transaction volume of the business,so as to increase the block chainthroughput and reduce the transaction delay.When the upper limit of the transaction time delay is set to 15 seconds,the throughput of blockchain can reach 20000bps.