Research On Blockchain Sharding Method Based On Transaction Feature Analysis

Low throughput and long transaction latency on blockchain are one of the main issues that restrict its application implementation.The demand for high scalability is strong,and solutions include sharding,DAG,state channels,side chains,and other methods.Among them,sharding technology improves scalability without weakening decentralization,and has good application prospects.This paper uses sharding technology to optimize blockchain performance and constructs a blockchain transaction sharding model.This model can achieve parallel processing of transactions between shards and distributed storage of transaction data,so it effectively improves system throughput and reduces redundant data storage.On this basis,from the perspective of transaction sharding,two transaction sharding strategies are proposed to solve problems of the high proportion of cross-shard transactions and the hot data aggregation.The main research work of this paper is as follows:(1)A blockchain transaction sharding model is established to optimize blockchain performance from the perspective of transaction sharding.This model changes the strategy of traditional blockchain for transaction data processing,and uses sharding technology to divide transaction data into multiple shards and allocate them to nodes in different shards for processing.Multiple shards process transactions in parallel,so it linearly improves the business throughput of blockchain.(2)In order to reduce the high cross-shard transaction proportion caused by sharding,a blockchain transcation sharding algorithm based on transaction frequency analysis(FBTS)is proposed.This algorithm analyzes the transaction frequency of accounts from the perspective of account correlation,and allocates the accounts with high transaction frequency to the same shard,in order to reduce the proportion of cross-shard transactions.The experimental results show that under different shard size,compared to the random sharding strategy and the modular sharding strategy,the FBTS algorithm reduces the cross-shard transaction ratio by about 20%-40%.In addition,this algorithm outperforms existing algorithms in terms of performance indicators such as the high cross-shard account proportion,the account proportion with different cross-shard times,the transaction volume proportion with different cross-shard times,and the average cross-shard times of accounts.Therefore,the FBTS algorithm can effectively reduce the proportion of cross-shard transactions and decrease the cross-shard communication costs.(3)To avoid the problem of hot data aggregation in shard caused by static sharding strategy,a blockchain sharding algorithm based on transaction volume adaptive equalization(AETS)is proposed.This algorithm belongs to the dynamic sharding strategy.By analyzing the transaction volume and transaction frequency between shards,the new accounts generated by transactions are prioritized into shards with less transaction volume.After multiple rounds of dynamic sharding,the data volume load balance between shards is achieved.The experimental results show that the AETS algorithm can achieve balance in the transaction data volume between shards,solve the problem of hot data aggregation,and be suitable for dynamic sharding application scenarios without significantly improving the proportion of cross-shard transactions and the average cross-shard times of accounts.The results indicate that the transaction sharding model constructed in this paper can improve blockchain throughput in parallel and reduce redundant data storage.The FBTS algorithm and AETS algorithm proposed based on the blockchain transaction sharding model can effectively reduce the proportion of cross-shard transactions,avoid hot data aggregation,and further optimize the system performance of blockchain.