Research On Load Balancing Among Shards Of Blockchain Adapting To Differences Of Nodes

With the decentralized advantage and the support of cryptography,blockchain technology has challenged the traditional trust system.It has changed the way of value sharing,and has promoted the transformation of infomation Internet into value Internet.However,there are many unresolved problems.Decentralization,security and scalability cannot be satisfied at the same time,which is called the impossible triangle of blockchain.Public blockchairis like Bitcoin and Ethereum are relatively mature in terms of decentralization and security,but they both have difficulties in achieving substantial breakthroughs in scalability.Therefore,the current infrastructure of the blockchain cannot meet the requirement of large-scale applications.So researchers have proposed various blockchain expansion plans such as sharding,'DAG,status channel and side chain.The sharding technology can solve the problem of scalability without reducing the degree of decentralization and has a good application prospect.The sharding technology is still in the initial and exploratory stage,and there are still many problems to be solved.For example,in blockchain sharding,if'a certain shard is congested and generates a large amount of transaction information,it will cause the load of each shard to be unbalanced,the transaction throughput of the blockchain system will reduce.The current sharding scheme usually assigns nodes to different shards randomly.This random allocation strategy ignores performance differences among nodes,resulting in large performance differences among different shards in the blockchain.If a shard with a heavy load has a poor processing capacity,the transaction throughput of the entire blockchain system will reduce with the congestion.This thesis proposes a multi-round verification scheme based on node rating,according to performance differences among nodes.This scheme is named as NRMRV(Node Rating Multi-Round Verify).According to the performance of the node,the speed and result of the verification transaction,the node is scored,and then the load of each shard in each round is obtained by multi-round verification.In the next round of verification,the node with a high score is assigned to the shard with a heavy load.So the performance of the shard will be improved and the imbalance of transaction load among shards will be solved.Therefore the transaction throughput of the blockchain system can be improved.This thesis compares NRMRV and the scheme of assigning nodes randomly in the simulation environment of the laboratory,in aspects of transaction throughput and security.The result shows that the blockchain sharding system with NRMRV balances loads of different shards,improves the transaction throughput,and has the ability of anti-attacking by recognizing dishonest nodes in the aspect of security.