Research And Implementation Of Sharding Technology Of BlockChain In Consortium BlockChain Scenarios

The blockchain becomes a research hotspot,since the characteristics of blockchain,such as decentralization,traceability and tamperproofing,can break the traditional trust mode and reduce the trust cost among social entities.In particular,the consortium chain has attracted extensive attention due to its characteristics of safety,supervision and operability.However,the single chain structure in consortium chain is the performance bottleneck,which severely blocks the development of consortium chain in the real world.The shard technology can break through the limitation of single chain structure,obtaining high concurrency from multiple independent consensus groups,called shard,and linearly improving the performance.However,there are still two issues when applying shard technology in consortium chain:low data synchronization efficiency between different shards,and byzantine nodes gathering when shards are generated.To overcome the two issues,this thesis redesigns a state snapshot ledger based on the directed acyclic graph(DAG)and improves the consensus protocol.Moreover,this thesis proposes a history score-based node classifies algorithm to constrain the byzantine node gathering,which can improve the efficiency of data synchronization between shards and improved the security of consensus.Besides,based on the open source project md_blockchain,this thesis builds a system where our algorithm is implemented.The main contents of this thesis are as follows:1.To overcome the low efficiency of data synchronization between shards,a DAG based state snapshot ledger is redesigned,and the shard consensus protocol is improved.This thesis implements concurrently cross-shard transactions confirmation by the state snapshot ledger,which can break through the bottleneck of singlechain structure in traditional consortium chain.Moreover,in consensus protocol,this thesis submits the cross-shard transaction and then updates the state,which can simplify the verification of cross-shard transactions.The simulation results show that,our proposed protocol has the advantage of linear performance improvement.Compared with the existing shard consensus protocol,our protocol has higher throughput.In addition,under the same throughput,our proposed shard protocol has a lower cross-shard transaction confirmation delay time compared with existing protocols.2.This thesis proposes a history score-based fast node classifies algorithm,solving the problem that the random sharding of Byzantine nodes may lead to the paralysis of partial shard,and then lead to security problems.The proposed algorithm classifies nodes according to history scores.Then,to avoid the aggregation of Byzantines in partial shards,the algorithm randomizes different nodes to shards,which can prevent breakdown caused by attacking.Simulation results show that,in the case of nodes continuously bad behavior of node,our algorithm can quickly classify nodes to byzantine nodes.In addition,under different Byzantine fault tolerance,the algorithm has a lower shard crash rate compared with existing algorithms.3.Based on the open source consortium chain platform md_blockchain,this thesis builds a system to implement our proposed shard solutions(the DAG-based global state ledger structure,the fast node classifies algorithm based on history points and the twolayer shard consensus protocol).Moreover,the test results of this system show that this prototype system can efficiently reduce the confirm latency time of cross-shard transaction,which can effectively improve the cross-shard transaction.Moreover,when the nodes accumulate enough history points,this system can prevent the aggregation of Byzantine nodes to shard,ensuring the system security.