联合分布算法对区块链分片的稳定性分析优化研究

在基于Hyperledger Fabric架构的区块链网络中，可以通过分片来提高吞吐量.在分片之前，需要预先评估每个委员会的失败概率，以获得更稳定的分片方案.本文提出了一种联合分布(JD)算法.首先，对于预分片方案中的节点，根据预分片方案中的委员会数量进行随机抽样，然后得到每个委员会中节点的超几何分布；其次，根据节点的超几何分布计算每个委员会中恶意节点的概率，并根据这些概率构建所有委员会的联合分布函数；最后，根据各委员会的联合分布函数，计算整个分片方案的失败概率和失败年数，实现区块链分片的稳定性评估.仿真结果表明，通过计算Wilson置信区间验证了计算失败概率的准确性，与RapidChain的分片稳定性分析算法和Hoeffding边界算法相比，联合分布(JD)算法可以更好地估计区块链分片的稳定性.

Optimization of security of Blockchain sharding by joint distribution algorithm

The throughput can be improved by sharding in the blockchain network based on Hyperledger Fabric architecture, before sharding, it is required to pre evaluate the failure probability of each committee, so as to obtain a more stable sharding scheme. This paper proposes a joint distribution (JD) algorithm. Firstly, for the nodes in the pre sharding scheme, random sampling is performed according to the number of committees in the pre sharding scheme, the hypergeometric distribution of nodes in each committee is then obtained. Secondly, the probability of malicious nodes in each committee is calculated according to the hypergeometric distribution of nodes, and the joint distribution function of all committees is constructed according to these probabilities. Finally, according to the joint distribution function of all committees, the failure probability and failure years of the whole sharding scheme are calculated to realize the stability evaluation of blockchain sharding. The simulation results show that the accuracy of calculating the failure probability is verified by calculating Wilson confidence interval, which is compared with RapidChain''s sharding stability analysis algorithm and Hoeffding boundary algorithm, Joint distribution (JD) algorithm can better estimate the stability of blockchain sharding.