降低跨分片交易比例的区块链分片方法

Blockchain sharding method for reducing cross-shard transaction proportion

针对分片技术在优化区块链性能时引起的跨分片交易比例过高、跨分片交易验证复杂的问题,提出一种降低跨分片交易比例的区块链分片方法。首先,从数据分片的角度出发,构造区块链交易分片模型,并给出分片性能评价指标;其次,针对区块链中长期积累的交易数据,从账号关联性角度出发,构造发送方和接收方的交易频次集合;最后,设计考虑交易频次的区块链交易分片算法(FBTS),从而解决交易分片中跨分片比例过高的问题。在分片粒度为2、3、5、7、15、20、30和50的情况下,所提算法在跨分片交易比例、账号平均跨分片次数、账号的加权平均跨分片次数等性能指标方面均优于随机分片算法(RSA)和取模分片算法(MSA)。另外所提算法的大多数账号和交易量都集中分布在低跨分片次数处,说明交易的完成不需要多次跨分片。实验结果表明,所提算法可以有效地降低跨分片交易比例,缩短跨分片的交易的时延。

To address the problems of high cross-shard transaction proportion and complex cross-shard transaction validation in optimizing blockchain performance,a blockchain sharding method for reducing cross-shard transaction proportion was proposed.Firstly,from the perspective of data sharding,a blockchain transaction sharding model was constructed,and evaluation indicators for sharding performance were provided.Then,for the long-term historical transaction data in blockchain,the sets of transaction frequencies for the sender and the receiver were constructed from the perspective of accounts’correlation.Finally,a Frequency-considered Blockchain Transcation Sharding algorithm(FBTS)was designed to solve the problem of high cross-shard proportion in transaction sharding.The proposed algorithm was compared with Random Sharding Algorithm(RSA)and Modular Sharding Algorithm(MSA)under the sharding size of 2,3,5,7,15,20,30 and 50.The proposed algorithm outperformed RSA and MSA in terms of performance indicators such as cross-shard transaction proportion,average cross-shard number of accounts,and weighted average cross-shard number of accounts.In addition,the most accounts and transactions were concentrated at low cross-shard number,indicating that the completion of transaction does not involve multiple shards.The experimental results show that proposed algorithm can effectively reduce the cross-shard transaction proportion and shorten the delay of cross-shard transaction.