As the core technology of the blockchain, the consensus mechanism determines how the participating nodes agree on certain data, which is related to the security, scalability and decentralization of the blockchain and many other important characteristics. The quality of consensus design is the key to whether blockchain autonomy can enter a virtuous circle. Consensus mechanisms can be divided into two categories: blockchain consensus mechanisms and classical distributed consensus mechanisms. Although the existing blockchain consensus mechanism achieves non-authorization and the highest degree of decentralization, in order to meet the security of the blockchain system, there is a general problem of low performance. Taking the currently widely used public chain as an example, the TPS (Transactions Per Second) of Bitcoin is not higher than 7, and the TPS of Ethereum is not higher than 25, and its transaction processing capability is much lower than that of credit cards such as Visa. The processing power of the payment system (about 7000 transactions per second). Inefficient throughput limits the application of blockchain systems in real-world scenarios. In order to enhance the usability of the blockchain system, people began to study how to improve the performance of the blockchain by improving the consensus mechanism.
As early as the end of the last century, academia has carried out relatively systematic research on classical distributed consensus mechanisms, for example, Paxos, which is designed for the situation that a collapsed node may occur in the network, PBFT that can tolerate a certain number of Byzantine fault nodes, etc. , they were mainly used in some distributed database systems before the emergence of blockchain. When the number of nodes is small, they can provide better throughput than non-authorized consensus mechanisms such as PoW and PoS. At present, academia and industry have closely combined the classical distributed consensus mechanism with the blockchain. However, due to the high communication complexity of the classical distributed consensus mechanism, as the number of nodes increases, the throughput of the system will also increase with Dip. Therefore, the scalable technology of the classical distributed consensus mechanism has become a current research hotspot.
[0005] According to the different assumptions of the network model, the classical distributed consensus mechanism can be divided into three categories: consensus mechanism based on synchronous network model, consensus mechanism based on partial synchronous network model and consensus mechanism based on asynchronous network model. The synchronous network model (Synchronous model) assumes that the attacker's ability is very limited, and all messages can arrive within a known time T. Under the assumption of the synchronous network model, the protocol operates in "round" (round) units. The time of a round is T, and all messages can be transmitted within T time, so it can be guaranteed that each node has received messages from other nodes before the end of each round; Partially synchronous network model (Partially synchronous model) ) means that under the assumption of this network model, all nodes can achieve synchronization at an unknown point in time, and from this point in time, messages can complete end-to-end transmission within a known time; in the asynchronous network model ( Under the assumption of Asynchronous model), there is no guarantee that all message transmissions between nodes will be successful.
[0006] The consensus mechanism adopted by the existing consortium blockchain system is realized based on the consensus mechanism of the partial synchronous network model, including three important components: node selection, consensus reaching, and node reconfiguration. Among them, node selection and node reconfiguration are an important part of ensuring the activity of the consensus mechanism, which is used to ensure that the leader node is an honest node, and the proportion of honest nodes in the network exceeds a certain proportion. Consensus is an important part of ensuring the consistency of the consensus mechanism, which is used to ensure that all honest nodes agree on the same content.
"System and Method of Scalable Consensus Mechanism Based on Multi-role "