Research And Application Of Key Technologies Of Blockchain Sharding Based On Deep Reinforcement Learning

As an application paradigm combining distributed database,P2 P network,consensus and encryption algorithm,blockchain is characterized by good decentralization and openness,anti tampering,anonymity and traceability.At present,blockchain system is generally limited in performance and capacity,which makes it unable to be applied to more scenarios,resulting in reduced scalability.To solve the scalability problem,researchers propose different solutions,including increasing block size,lightning network,DAG,sharding and so on.Among them,the sharding technology is to divide the blockchain into different "sub networks",that is,all nodes are sharded,and then the transactions are divided into different shards for processing,so as to form the sharded block chain,so as to improve the overall performance of the blockchain.However,there is a lack of research on the performance parameters of the sharded block chain and how to obtain the parameters to optimize the performance of the sharded block chain.In addition to the relevant characteristics of blockchain itself,the sharded blockchain also has higher throughput than the single chain blockchain,so it has a good application prospect in value chain management.With the development of economy,enterprises have formed a value chain in the process of producing products or completing services,such as the field of vehicle parts industry.In order to improve the market share,enterprises usually need to complete the business as soon as possible and reduce the cost as much as possible,thus forming a value chain and improving the efficiency of enterprises.At present,the realization of the value chain of vehicle parts is mostly based on the cloud platform.Although the cloud platform can realize the integration of resources and information in the industry to a certain extent,there are still some related problems such as information opacity and data insecurity,which need to be solved urgently.To sum up,this paper studies the scalability of blockchain and its application in the field of vehicle parts industry(1)First of all,aiming at the scalability of blockchain,this paper chooses sharding technology as the solution.In the overall architecture,the block chain architecture based on pbft is adopted,and the consensus mechanism of the architecture is analyzed.The performance parameters of pbft block chain are designed from four aspects of scalability,delay,decentralization and security.(2)Secondly,a reinforcement learning model of pbft block chain based on ddqn is proposed,and the main factors affecting the performance of block chain are analyzed,including block size,block packing time and number of blocks.According to the assumptions of the model,a method to obtain the ratio of malicious nodes estimation is proposed.According to the factors that affect the performance of block chain,the state space and action space of the model are designed.Ddqn algorithm is selected to solve the parameters(including block size,block packing)when the pbft block chain system achieves the optimal performance(scalability)under the requirements of security,delay and decentralization.(3)Finally,the paper analyzes the particularity and necessity of using ddqn shardblock model in the scenario of vehicle parts value chain,and analyzes and designs the logical structure of parts allocation and parts traceability business in the parts value chain.At the same time,the system framework of accessory value chain based on sub area block chain is designed,and the accessory value chain system based on sub area block chain is implemented according to the design.