Research On Block Structure Supporting High Throughput Blockchain

Blockchain is an only incrementally updated ledger maintained by distributed nodes rather than a centralized organization.Although the prospects of blockchain technology,such as Bitcoin,Ethereum,and Hyperledger Fabric,are very promising and popular in all industries,the current blockchain solutions still face scalability problems,including low transaction throughput and high storage capacity.The main index of blockchain throughput performance is Transaction per Second(TPS).The throughput of traditional blockchain networks will not scale with the increase of the network size.Because the data stored on the blockchain can only be updated incrementally and cannot be deleted,each node needs to maintain all the data of the entire blockchain system,which wastes storage space greatly.So far,many methods,such as Off-chain and Directed Acyclic Graph(DAG)solutions,have been proposed by researchers to address the issue.However,they not only have disadvantages such as the formation of parasitic chains,but also cannot solve the problem of storage overhead.The sharding technology is one of the most promising scalable blockchain solutions,which will effectively solve blockchain performance issues.This thesis proposes a blockchain structure that supports high throughput to solve the problem of scalability.The proposed method is designed based on sharding protocol.This thesis designs and implements blockchain network sharding,transaction sharding and state sharding schemes,and proposes a network sharding size prediction algorithm to predict the optimal shard size,so that traditional blockchains have the advantages of high throughput and saving of storage costs,and can handle large numbers of transactions.The whole blockchain network node of the sharding protocol are randomly divided into each shard,and the identity of each node is given using the idea of proof of stake(PoS),that is the network shard,and then the unconfirmed transactions are divided into transaction sharding according to the account address attribute.Each network sharding node process transactions in parallel and consensus the value of generating subtree merkleroot,then the leader combines the subtree merkleroot and publishes the complete block.State sharding uses a versioned approach to manage account status,and uses the idea of a distributed hash table(DHT)to store account data.Experimental results show that a blockchain network with more than 100,000 nodes can be divided into 1024 shards.The blockchain network with this structure can process 500,000 transactions in about 5 seconds.If the consensus time of the blockchain is about 10 seconds,and the blockchain system's block generation time is 15 seconds,the transaction throughput can reach 33,000 tx/sec.TPS of the proposed blockchain structure is 1650 times than Ethereum which currently has no sharding mechanism,and is 4.5 times than RapidChain based on sharding mechanism.The results of the experiments indicate that the throughput of the proposed protocol increases with the network size,which confirms the scalability of the blockchain structure based on sharding technology.