星型区块链架构的TKM分片算法

TKM sharding algorithm for star blockchain architecture

区块链系统的通量严重不足,而解决此问题最有效的一类方案是并行化处理,并行化方案主要为星型架构,当前星型架构对系统中节点的分片方式多为账户随机分片,这种分片方式的系统通量仍然不足。针对此问题,提出了一种基于星型结构的TKM分片算法,该算法将原始K-means聚类算法进行改进,并运用在节点分片上。TKM分片算法将聚类算法与区块链的网络分片技术相结合,使节点根据地理位置进行分片,极大提高邻近节点发生的交易为片内交易的概率,从而提高系统通量,同时在原始算法的基础上引入了时间戳,减少了恶意节点的攻击。仿真实验表明该算法与传统的随机分片算法相比,最大系统通量提高了20%。根据上述通量模型,通过实验得出基于TKM算法的星型区块链系统的最优分片数量。

The throughput of blockchain systems is severely insufficient,and the most effective solution to this problem is parallelization processing.The parallelization scheme is mainly a star architecture.Currently,the star architecture mostly uses account random sharding for node sharding in the system,and the system throughput of this sharding method is still insufficient.In response to this issue,this paper proposed a TKM sharding algorithm based on star structure,which improved the original K-means clustering algorithm and applied it to node sharding.The TKM sharding algorithm combined clustering algorithm with blockchain network sharding technology,allowing nodes to be sharded based on geographical location,greatly increasing the probability of transactions between neighboring nodes being intra chip transactions,thereby improving system throughput.At the same time,it introduced time stamps on the basis of the original algorithm to reduce attacks from malicious nodes.Simulation experiments show that this algorithm improves the maximum system throughput by 20%compared to traditional random sharding algorithms.Based on the above flux model,the optimal number of shards for the star blockchain system based on the TKM algorithm is obtained through experiments.