Design And Performance Optimization Of A Hyperledger-based Permissioned Blockchain System

With the rapid development of blockchain technology,the research focus of it has evolved from the virtual currency-oriented permissionless blockchain to the industry application-oriented permissioned blockchain.Compared with the permissionless blockchain,the permissioned blockchain owns more rigorous authentication,smaller network scale,and higher system throughput.Based on the open source blockchain framework Hyperledger Fabric,developers can implement a specific blockchain system conveniently.Therefore,the research focus of this thesis is how to design and implement a Hyperledger-based permissioned blockchain system for specific applications,and perform system modeling,performance analysis and performance optimization on the system.Firstly,this thesis designs and implements a permissionless blockchain system based on Hyperledger Fabric after analyzing the system requirements of emission trading and pointing out several drawbacks of the traditional system architecture.The system is divided into the application layer,HTTP server layer and blockchain system layer according to the specific functions.Next,this thesis introduces the design and implementation of each layer in detail,including the interface design,ledger structure design,smart contract design and the blockchain system design.Finally,the system functions are verified by calling the corresponding HTTP interfaces.Secondly,this thesis carries out system modeling,performance analysis and performance optimization based on the proposed blockchain system.The thesis analyzes the system request processing flow in detail and decomposes it into multiple processing phases.Then,a mathematical model which analyzes the performance of the blockchain system based on Generalised Stochastic Petri Nets(GSPN)is proposed.To verify the proposed system model,we set up an actual system on a cloud server to conduct related experiments and we compare the analysis results and the experimental results under the different system configurations.Next,the performance bottleneck is identified based on this GSPN model and then the thesis proposes two effective approaches to optimize the system performance.(1)Based on the research on the network configuration parameters of the blockchain system,a queuing theory-based determination approach is proposed to make the system reach the maximum throughput.(2)For the load balance-based optimization approach,this thesis proposes a resource allocation-based system model to connect the computing resource and the throughput.Moreover,the optimal number of loading nodes and the maximum throughput are obtained through the theoretical analysis.Finally,the effectiveness of the strategies above is proved by extensive experiments.