Research On Distributed Consensus For Scenario-related Blockchain Systems

Blockchain is a distributed evidence storage mechanism that achieves consensus on digital evidence among decentralized peers.It build trust and promote efficient collaboration among unknown network participants who do not have a trust relationship at low cost based on its basic features such as decentralization,tamper-resistance,and traceability.Blockchain technology can not only play an essential role in the financial industry as distributed ledgers but also has extensive application scenario prospects in many fields,such as data sharing and information security.Distributed consensus technology,as one of the core technologies of the Blockchain,has gradually become a research hotspot in academia and industry.In the traditional field of distributed computing,consensus technology is often divided into two types: Byzantine fault-tolerance and non-Byzantine fault-tolerance.The non-Byzantine fault-tolerant consensus algorithm represented by Paxos and Raft only considers that nodes fail only by halting;that is,they can only stop working and have no other malicious behaviors present.However,in the diverse application scenarios of the Blockchain,each participant acts in pursuit of maximizing their interests.They may cause adverse behaviors to the regular operation of the system.Therefore,Byzantine faulttolerance is required in the Blockchain system.This research focuses on the Blockchain distributed consensus technology with Byzantine fault-tolerance.It mainly aims at three typical Blockchain application scenarios,namely public Blockchain distributed accounting,consortium Blockchain distributed accounting,and consortium Blockchain data collaboration.Generally speaking,different scenarios often need to customize specific consensus algorithms based on their characteristics and requirements.The design principle is to make trade-offs in terms of security,effectiveness,and decentralization.Focusing on the above three Blockchain application scenarios,the main contributions of this research are summarized as follows:(1)For the scenario of public Blockchain distributed accounting,we propose two consensus algorithms that take accounting efficiency,fair distribution of accounting rights,and savings in computing power into account.In the public Blockchain distributed accounting scenario characterized by free participation and full openness,the high accounting frequency demand caused by large-scale transaction activities among various participants cannot be satisfied by the existing consensus algorithms.Besides,due to the existence of unequal identities,the existing consensus algorithms can easily lead to uneven distribution of accounting rights,which may bring a risk of accounting monopoly.To solve the above problems,we propose the Proof of Participation and Fees(PoPF)consensus algorithm that can shield users from unequal identity and make the distribution of accounting rights fairer.First,users are ranked according to their participation in the public Blockchain,and then different mining difficulties are set for different users according to the ranking.However,PoPF still faces problems such as wasted computing power and low efficiency.Therefore,we propose the Proof of Previous Transactions(PoPT)consensus algorithm.PoPT follows the ranking rules of PoPF for users and performs a consistent hashing-based parallel accounting mechanism.It is verified through experiments that both PoPF and PoPT can save computing resources and make the allocation of accounting rights more fair and reasonable while ensuring safety and efficiency.(2)For the scenario of consortium Blockchain distributed accounting,we propose a consensus algorithm that takes accounting efficiency and nodes' size into account under the premise that the consensus process can be monitored.In the scenario of distributed accounting of consortium Blockchains based on collaboration in a specific scope and authorized the participation of enterprises,the existing consensus algorithms cannot ensure both high efficiency and node scalability without sacrificing security.Furthermore,it is difficult to supervise and deal with malicious behaviors in Blockchain systems.To solve the above problems,we propose a highly efficient and supervised consensus algorithm called Jointgraph based on Directed Acyclic Graph(DAG).Jointgraph introduces a weakly centralized supervisor,and cleverly implements its functional design in the consensus process so that the supervisor will not become a fully centralized role to control the consortium Blockchain.The supervisor can not only monitor the malicious behavior,remove malicious participants in time,but also improve the transaction processing speed of the Blockchain.Experiments show that Jointgraph outperforms Hashgraph(one of the current mainstream DAG-based consensus algorithms)in transaction throughput,confirmation latency,and node scalability.(3)For the scenario of consortium Blockchain data collaboration,we propose a consensus algorithm in which the consensus participants' scope(specific participants to confirm specific transactions)is customizable,the consensus degree is viewable,and consensus subjects is dynamically changeable.In the consortium Blockchain data collaboration scenario that serves “ cloud-edge-end ” flexible collaboration,different data needs to be confirmed in real-time by the participants of different swarms.Therefore,the consensus algorithm for this application scenario should provide the ability for users to customize the consensus participants' scope,query the consensus degree of any transaction in real-time,and dynamically change the consensus subjects.However,the existing consensus algorithms cannot meet these demands.To solve these challenges,we propose a consensus algorithm called Teegraph based on Trusted Execution Environment(TEE)and DAG.In Teegraph,each node first sends the transaction to TEE for pre-verification to avoid malicious nodes from sending different information to different nodes in the same round,thereby avoiding fork attacks.Through the design of the consensus mechanism based on TEE,the complexity of the consensus process is reduced(compared to Hashgraph),and the security of the algorithm is improved,making the Byzantine node fault tolerance rate increased from 33.3 % to 50 %.Also,we provide a storage resource-saving mechanism for Teegraph.Experiments show that Teegraph outperforms Hashgraph and Jointgraph both in transaction throughput and confirmation latency.With the economic globalization,how to efficiently use the Blockchain technology to achieve high-efficiency,low-cost,long-distance trust establishment and value transfer is a significant primary and cutting-edge issue.In this research,combining the prospective research object of the Blockchain with the requirements of different types of Blockchain application scenarios,we make breakthroughs around the distributed consensus technology in the dimensions of distributed accounting and data collaboration in both the public and consortium Blockchains.This research has a positive impetus for a new type of value transfer paradigm based on Blockchain in the future.