WO2019222993A1

WO2019222993A1 - Blockchain consensus method based on trust relationship

Info

Publication number: WO2019222993A1
Application number: PCT/CN2018/088355
Authority: WO
Inventors: 李挥; 王贤桂; 王菡; 马军锋; 伊鹏; 李恪聃; 黄建森; 李科浇
Original assignee: 北京大学深圳研究生院
Priority date: 2018-05-25
Filing date: 2018-05-25
Publication date: 2019-11-28
Also published as: CN109964242B; CN109964242A

Abstract

The present invention is applicable to the field of improvements in Internet technology, and provides a blockchain consensus method based on a trust relationship, the method comprising: S1. quantifying a trust relationship between nodes according to trading and block data; S2. constructing a trust relationship graph and generating a trust matrix according to the trust relationship between the nodes; S3. iteratively calculating a trust value of each node using the trust relationship between the nodes of a whole network; and S4. for each round of consensus, randomly selecting a representative node as a primary node having the bookkeeping rights, and the primary node selecting several trading creation blocks from a trading pit. A representative mechanism is used to professionalize a bookkeeping node, thereby reducing the consensus cost, reducing energy consumption, and improving the efficiency of node consensus and an expansion capability of an algorithm at the same time. A representative node is selected based on a trust relationship, without depending on a token of a blockchain, thereby preventing the bookkeeping rights from being concentrated to a few "rich people".

Description

Blockchain consensus method based on trust relationship

Technical field

The invention belongs to the field of Internet technology improvement, and particularly relates to a blockchain consensus method based on a trust relationship.

Background technique

With the technological revolution and industrial changes in the global information technology field, the Internet has gradually evolved from "information Internet" to "value Internet". In the "Information Internet" era, information on the Internet is open and transparent, but it also becomes unreliable because it can be tampered with at will, and requires a third party to provide a guarantee of trust. Once the third-party platform that provides trust fails, the trust it provides becomes a bubble. In order to solve the trust problem generated during the development of the Internet, blockchain technology came into being.

Blockchain technology is an integrated application model of distributed data storage systems, point-to-point transmission, consensus mechanisms, encryption algorithms and other technologies. It can achieve trust and value transfer on the Internet that cannot be achieved by traditional Internet. It is based on the principle of cryptography rather than credit, which enables any agreed parties to trade directly without the participation of third-party intermediaries. On the other hand, there is almost no single point of failure in the blockchain, and the data on the chain is stored on countless machine nodes around the world, making the data "stable", "trustworthy" and "immutable", which re-gives the network The value of the data can be trusted.

The blockchain has the characteristics of decentralization. Each node in the network needs to follow a certain protocol to ensure the consistency of the node data. This protocol is called a consensus protocol or consensus algorithm. Consensus algorithm is a key part of blockchain technology, which directly affects the security and efficiency of blockchain products. A secure and efficient consensus protocol is an important research issue in the field of blockchain. The PoW consensus algorithm in the Bitcoin system creatively implements data consistency between nodes in an untrusted decentralized network. It proves that the system is completely decentralized and has the advantages of simplicity and ease of implementation, but it also has resources. Waste and poor performance. In addition, in small-scale applications of the blockchain network, a small number of nodes can master more than 50% of the computing power, which will degrade the PoW consensus from decentralization to centralization, and even affect the security of the consensus. Adopted in large application scenarios. The PoS consensus algorithm in the peer-to-peer currency system solves the problem of waste of resources in the PoW consensus through the proof-of-stake mechanism. However, similar to the concentration of PoW consensus in a few nodes, the PoS consensus will degenerate into a centralized consensus due to the concentration of equity. The DPoS consensus algorithm in the BitShares system proposes an authorized share certification mechanism based on the PoS consensus. The node authorizes its own rights to other nodes. The top 101 nodes with the highest rights become the representative nodes. The representative nodes have exactly the same rights as each other. Bookkeeping rights. The representative mechanism reduces the operating cost of the network and can increase the rate of block generation.

Although blockchain consensus has been developing, its performance is still very limited and it is difficult to achieve second-level consensus. Researchers and developers are turning their attention to traditional distributed consensus algorithms. The Paxos algorithm proposed by Leslie Lamport in 1990 is currently the most efficient consensus algorithm based on message passing, and it is also the most commonly used consensus algorithm in engineering practice. Paxso can efficiently achieve data consistency between nodes, but it cannot allow malicious nodes to undermine the consensus process, so it can only be used in trusted intranets. In 1999, Castro and Liskov proposed a PBFT consensus algorithm that tolerates fewer malicious nodes than 1/3 of the total number of nodes in the network. Under this mechanism, the system throughput can reach more than 100,000. The Hyperledger project verified the possibility of PBFT consensus in the alliance chain network with very few malicious nodes, and achieved good results. However, the communication cost of O (n ² ) of the PBFT consensus itself limits the network size of the blockchain. As the number of nodes increases, its performance decreases rapidly, so the scalability of the PBFT consensus is poor. The domestic small ant blockchain development team proposed the dBFT protocol, and the bookkeeper was selected through equity, and the bookkeeper reached a consensus through the Byzantine fault tolerance algorithm, thereby reducing the network operation cost and improving the efficiency of the consensus algorithm. Existing consensus algorithms that introduce a representative mechanism are limited by the token mechanism of the blockchain application, and cannot be used in some blockchain systems where tokens do not exist.

In September 2015, one year after the official establishment of the ant project, Onchain released a white paper on the ant consensus algorithm, which proposed an improved Byzantine fault tolerance algorithm-authorized Byzantine fault tolerance (dBFT) as being applicable to the blockchain System consensus module.

The dBFT consensus is formed on the basis of the PBFT consensus. The original C / S architecture response mode is modified to a P2P mode suitable for peer-to-peer networks and nodes can dynamically enter and exit. Nodes select bookkeepers based on equity, and then the bookkeepers reach consensus through a Byzantine fault tolerance algorithm. The dBFT consensus sets system nodes into two roles: ordinary nodes and accounting nodes. The accounting node is an important node in the consensus process. It is responsible for collecting and serializing transaction messages in the network and recording them into the distributed ledger maintained by the entire network. Ordinary nodes do not participate in the consensus process, and only use the system to perform operations such as transfers and transactions. Accept the data in the ledger, but be able to see the full consensus process. The initial state consistency of the nodes is achieved through block synchronization and view replacement. The dBFT system works in a time-synchronized state. Each round of consensus generates a block within a certain time interval. The generation of each block may go through different views. In one view, there is a node that acts as the speaker (leader), and the other nodes are members (followers). When the speaker receives block authentication support from more than a certain number of members, a consensus on a new block can be reached. The dBFT consensus is applicable to the alliance chain and the private chain. The digital certificate is used to solve the authentication problem of the identity of the consensus node.

The main advantages of the dBFT consensus are: first, the use of professional bookkeepers improves the efficiency of block generation and has good scalability; second, the bookkeeping is completed by multiple people in cooperation, and each block is It is final and will not fork.

The above consensus also has certain flaws. First, the selection of representative nodes depends on equity and tokens. Second, when one-third or more of the bookkeepers stop working, the system will not be able to provide services. When one-third or more of the bookkeepers cooperate with each other, and all other bookkeepers are split into two network islands. At times, malicious bookkeepers can fork the system, but leave behind cryptographic evidence.

The Larimer team presented the DPoS white paper and its first application, Bitshares, in March 2014. Due to the emergence of mining pools and ASIC mining machines, the Bitcoin system using the PoW consensus has become more and more centralized, and to some extent it has deviated from the original decentralization of "one CPU, one vote". The Larimer team believes that the decentralization of the PoS consensus can rely on a certain number of representatives instead of all shareholders, and all shareholders vote together to elect a representative. The DPoS consensus expects to return the accounting rights to those who hold digital currencies, and let everyone who holds bitcoin currency BTS vote on candidate representatives in the entire system resources. The 101 people who received the most votes become the representatives. In a random and unpredictable order, they have obtained the right to package and calculate blocks. All representatives will receive 10% of the transaction fee equivalent to an average block as compensation. If an average block contains 100 shares as transaction fee, one representative will receive 1 share as compensation. It can be understood that the rights are evenly distributed among the 101 CPUs, and each CPU has completely equal rights with each other. The DPoS consensus is similar to the US parliamentary system, except that the election process can occur at any time. Therefore, the accounting and verification nodes in the DPoS consensus are reduced from a worldwide scale to a certain number of nodes, which can achieve a second-level delay. However, the entire consensus mechanism still relies on tokens, and in order to achieve a certain level of security, a large number of block confirmations are required. 51% of the 100 blocks after the transaction is written into the block are produced. Can be considered safe on the main chain.

In terms of resistance to attacks, because the top 100 delegates received the same power, it was not possible to focus power on a single representative by obtaining more than 1% of the votes. Because there are only 100 representatives, it can be imagined that an attacker conducts a denial-of-service attack on each representative who has obtained the right of accounting in turn. Fortunately, the threat of this particular attack is easily mitigated due to the fact that the identity of each representative is its public key rather than its IP address. This will make it harder to target DDoS attacks. The potential direct connection between delegates will make it more difficult for them to create blocks.

Therefore, the main advantages of this consensus algorithm are: first, the number of participating verification and accounting nodes is greatly reduced, and the waste of computing power is further reduced based on the POS consensus, and the cost of network operation is reduced; second, the consensus traffic Small and has good scalability.

The above consensus also has certain flaws. First, the entire consensus mechanism still relies on tokens, and many commercial applications do not require tokens to exist. Secondly, because the above consensus is an improvement on the PoW consensus, it inherits the low block speed characteristics of the PoW consensus, and the system throughput is small.

Summary of the Invention

The purpose of the present invention is to provide a blockchain consensus method based on a trust relationship, which aims to solve the above technical problems.

The present invention is implemented as such, a blockchain consensus method based on a trust relationship, the blockchain consensus method includes the following steps:

S1. Quantify the trust relationship between nodes based on transaction and block data;

S2. Construct a trust relationship graph and generate a trust matrix according to the trust relationship between nodes;

S3. Iteratively calculate the trust value of each node by using the trust relationship of the nodes of the entire network, and its function formula is: T _i = C ^T T _i-1 ;

S4. Each round of consensus randomly selects a representative node as the master node with accounting rights. The master node selects several transactions from the transaction pool to create blocks;

Wherein, C represents confidence matrix, T _i represents the trusted node value vector after iteration i, T _i-1 represents the confidence value vector after the i-1 th iteration.

A further technical solution of the present invention is that most of the attacks in the blockchain are related to transactions and blocks. The attacks include creating false transactions, broadcasting invalid transactions, creating false blocks, and broadcasting invalid blocks.

A further technical solution of the present invention is that, in step S1, the receipt of valid transactions and valid blocks sent by node i by node i will increase node i's trust in node j, otherwise it will reduce the trust. The trust calculation formula is

Among them, g _ij represents the number of valid transactions and blocks that node i received from node j, u _ij represents the number of invalid transactions and blocks that node i received from node j, and β represents a penalty coefficient for invalid data.

A further technical solution of the present invention is: in step S2, in order to prevent a malicious node from giving a higher trust value to other malicious nodes and a lower trust value to a normal node, thereby affecting the selection of a trust representative node, use

Regularize the trust value between nodes to get the final trust matrix C _{n × n} ; where c _ij represents the normalized trust value of node i to node j, t _ij represents the trust degree of node i to node j, and n represents The total number of nodes in the network.

A further technical solution of the present invention is that the trust relationship matrix constructed in the step S2 has random, irreducible and non-periodic properties, so the iterative process T _i = C ^T T _{i-1 of the} node trust value matrix T can converge.

A further technical solution of the present invention is: the trust value between the nodes in step S3 is affected by interactive transactions and block data, and changes dynamically with time from 0 to 1; the k nodes with high trust value Being selected as a representative node, each representative node has the opportunity to obtain accounting rights.

A further technical solution of the present invention is: in step S3, a representative node is selected according to a trust relationship, the representative node participates in a fault-tolerant Byzantine agreement, and a professional accounting node improves the scalability of the consensus protocol.

A further technical solution of the present invention is: considering the network transmission delay in the step S4, the interval from the creation of the block to the final joining of the blockchain is

Block generation cycle

Therefore, at the beginning of the new round of consensus, other nodes have received the previous block data.

A further technical solution of the present invention is that in step S4, the blockchain network is considered to be relatively stable over a period of time, so the period t _{0 of the} update representative is much larger than the period t of the block creation, which improves the efficiency of consensus.

A further technical solution of the present invention is: in the step S4, the representative nodes create a block through a fault-tolerant Byzantine agreement, and then spread it to other nodes in the blockchain network. The nodes are based on the rule that the longest chain is the main chain. Ensure eventual consistency of node data.

The beneficial effect of the present invention is that a representative mechanism is adopted to select some nodes as representative nodes participating in consensus according to the trust relationship between the nodes. Representative selection is a cyclical behavior, which means that as long as any node works strictly and honestly, it will have the opportunity to become a representative node, obtain accounting rights and create blocks, so it has better decentralization characteristics than equity-based representative mechanisms. . Adopting a representative mechanism to professionalize bookkeeping nodes reduces the cost of consensus and saves energy consumption, while improving the efficiency of node consensus and the scalability of the algorithm. By monitoring the interactions between the nodes and the validity of the block data, a node with a high trust value is selected as the representative node, which does not rely on the token system and avoids the shortcomings of concentrating accounting rights on a few "rich people". The trust value is formed by the iterative convergence of the trust matrix, which is essentially a probability, which is the basis for the nodes on the entire network to judge a certain node, and has relative stability. The trust relationship between nodes changes slowly, so the update cycle of a representative node is much longer than the block generation cycle. On the other hand, the number of representative nodes is relatively fixed, meaning that the number of nodes in the network does not affect the efficiency of running fault-tolerant Byzantine protocols between the representative nodes.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagram of a consensus algorithm network model according to an embodiment of the present invention.

FIG. 2 is a flowchart of a blockchain-based consensus method based on a trust relationship according to an embodiment of the present invention.

FIG. 3 is a schematic diagram of a consensus algorithm according to an embodiment of the present invention.

Detailed ways

Blockchain is a distributed architecture. Multiple computers connected to each other form a peer-to-peer network, which operates in accordance with a consensus protocol to jointly process requests submitted by users. As shown in Figure 1, in a peer-to-peer network, each node is in a peer-to-peer position, serving as both a server and a client, with the same functions. For a peer-to-peer network, the number of nodes is usually not fixed, and nodes may join or leave the network at any time. In addition, there may be some untrustworthy malicious nodes in the network. Malicious Byzantine nodes can refuse to respond to the interactive information of other nodes and even falsely respond to other node's requests. In order to reduce the cost of consensus and improve the efficiency of consensus, the consensus algorithm selects some nodes with high trust values to participate in consensus through the trust relationship between nodes, and these nodes become representative nodes. Therefore, the network model of the algorithm contains four types of nodes: ordinary nodes, Byzantine nodes, ordinary representative nodes, and Byzantine representative nodes. Nodes use encryption technology to ensure communication security. Each message contains a key signature, digest, and verification information. Byzantine nodes cannot brute force encrypted messages, cannot forge signatures, and cannot find messages with the same digest.

The process of the consensus protocol is shown in Figure 2. It consists of (1) quantifying the trust relationship between nodes, (2) generating a trust relationship graph and trust matrix, (3) calculating the node trust value and selecting a representative node, and (4) a block generation protocol. Nodes monitor the transactions and block information of neighboring nodes to establish a trust relationship and quantify the trust relationship as a real number between 0 and 1. A larger value indicates a higher degree of trust. When a node sends valid transaction and block information to other nodes, it will increase the trust between nodes, otherwise it will reduce the trust. Then, based on the trust relationship between nodes in the entire network, a trust relationship graph is constructed, and a trust relationship matrix is generated. Then, based on the idea of search engine web page ranking, the nodes are compared to web pages, and the trust relationship between nodes is compared to the hyperlink relationship between web pages, and the trust value is iteratively calculated for each node. The k nodes with a high trust value are elected as the representative nodes, and they have the opportunity to obtain accounting rights and create blocks. Byzantine protocols with 1/3 fault tolerance are run between the nodes to ensure the security and availability of the consensus algorithm.

As shown in Figures 2 and 3, the flowchart of the consensus method based on the trust relationship provided by the present invention is detailed as follows:

Step S1: Quantify the trust relationship between nodes according to the transaction and block data; the trust relationship itself is difficult to define. A node with a high degree of trust in a computer system usually means that it can independently, safely and reliably complete the specified function in a specific environment. Node. Since transactions and blocks are the most important data in the blockchain, the consensus method quantifies the trust relationship between nodes based on the transactions and block data that are exchanged between nodes. In a blockchain network, nodes are interacting with transactions and block data at all times, constantly processing transaction requests submitted by users, verifying the validity of transactions, and broadcasting valid transactions to other nodes. Most of the blockchain's attack behaviors are related to transactions and blocks. There are mainly malicious acts such as creating false transactions, broadcasting invalid transactions, creating false blocks, and broadcasting invalid blocks. Therefore, consensus algorithms implement nodes based on transactions and block data. Quantification of trust relationships is effective. During the operation of a node, it will count the data information of the interaction between its neighboring nodes. Assuming that node i receives g _ij valid transactions and blocks from node j for a certain period of time, and u _ij invalid transactions and blocks, then The trust degree of node i to node j is shown in formula (1). β indicates the penalty coefficient of invalid data, which can be adjusted, usually greater than 1, indicating that receiving invalid data has a greater impact on the trust value than receiving valid data. Because the blockchain network is usually large, there are many cases where no nodes interact with each other. At this time, the trust relationship is quantified to 0.5.

According to the description of the trust relationship between nodes according to formula (1), the receipt of valid transactions and valid blocks sent by node i by node i will increase node i's trust in node j, otherwise it will reduce the trust.

Step S2, constructing a trust relationship graph and generating a trust matrix according to the trust relationship between nodes; in the World Wide Web, web pages are linked to each other by hyperlinks, and more valuable web pages are usually linked by many web pages, otherwise they are rarely linked . The search engine sorts related web pages according to the link relationship between web pages and presents them to users in an appropriate order. The invention draws on the webpage link relationship diagram, compares nodes to a webpage, and trust relationships to a hyperlink relationship, and establishes a trust relationship map according to the mutual trust relationship between nodes.

A trust relationship graph is a directed graph in which the weight of an edge represents the degree of trust between the two vertices corresponding to the edge. According to the information relationship graph, a trust relationship matrix D _{n × n} between nodes can be obtained, where n represents the total number of nodes in the network, and each element d _{ij of the} matrix represents the degree of trust between node i and node j. In order to prevent malicious nodes from giving higher trust values to other malicious nodes and lower trust values to normal nodes, which affects the selection of trust representative nodes, formula (2) is used to regularize the trust values between nodes to obtain the final trust matrix. C _{n × n} . Each element represents the direct trust relationship between nodes. The value of the relationship with high trust is close to 1, the value of the relationship with low trust is close to 0, and the value of the relationship with little interaction between nodes is close to 0.5.

Step S3: Iteratively calculate the trust value of each node by using the trust relationship of the nodes of the entire network; the nodes can obtain the direct trust relationship between them by directly monitoring the behavior of other nodes, and then iteratively calculate the trust relationship of each node by using the trust relationship of the entire network. The trust value of each node is shown in formula (3). Where C represents the above-mentioned trust matrix, T _i represents the node trust value matrix after the i-th iteration, and T _i-1 represents the node trust value matrix after the i-1 iteration.

T _i = C ^T T _i-1 Formula (3)

It can be proved that the iterative relationship of formula (3) can finally converge, and the convergence result is the trust value of all nodes calculated according to the trust relationship of the entire network. The k nodes with high trust value are selected as the representative nodes and have the opportunity to obtain accounting rights.

Step S4, each round of consensus randomly selects a representative node as the master node with accounting rights. The master node selects several transactions from the transaction pool to create a block; only the representative node participates in the consensus process, ordinary nodes cannot create blocks, but can see To the full consensus process. A round of consensus includes two parts: selecting a master node and creating a block. First, a node is randomly selected as the master node from the representative nodes. Only the representative node elected as the master node has the right to keep accounts and can create blocks. Then the block passes the verification of other representative nodes and is finally added to the blockchain. Assume that the number of nodes participating in the consensus protocol is k, and f is the number of malicious Byzantine nodes that can be tolerated

When a node starts, it will first synchronize block data through other nodes in the network. Nodes listen to incoming transactions, independently verify the validity of the transactions, store valid transactions in the transaction pool and broadcast to other nodes, and discard invalid transactions. After the node obtains the right to book, it selects several transactions from the transaction pool to form a block, and the hash value of the parent block is recorded in the block header to ensure the orderliness of the block. . When the transaction pool is empty and there are no valid transactions, empty blocks will be generated according to the normal cycle. Considering the delay of network transmission, the interval from the creation of the block to the final addition to the blockchain is

Block generation cycle

As a result, at the beginning of a round of consensus, other nodes have received the previous block data.

It takes three stages to reach consensus on the block data between consensus nodes, namely Pre-Prepare, Prepare and Commit.

(1) Pre-Prepare phase

The master node that obtained the accounting right sends a Pre-Prepare request to all consensus nodes. The request message includes the height h of the current block, the view v, the master node number p, the block block, and the block signature block _p . After receiving the Pre-Prepare request from the master node, the node verifies the accuracy of the message. If the Pre-Prepare request is invalid, it is proposed to change the view and reselect the master node. When a node receives a valid Pre-Prepare request, it sends a Prepare request to other nodes and enters the Prepare state itself. The prepare request contains the height h of the current block, the view v, the node number i, and the signature block _{i of the} block.

(2) Prepare stage

When a node receives 2f Prepare messages, the node enters the Commit state and sends a Commit message to other nodes. The Commit message contains the height h of the current block, the view v, the node number i, and the signature block _{i of the} block.

(3) Commit stage

When the node receives 2f Commit messages, it considers that this round of consensus is complete and writes the block into the blockchain.

The Pre-Prepare message contains the complete block. After receiving the Pre-Prepare message from the master node, the node will save the block content of the consensus locally, and use the hash value to replace the block in the following Prepare and Commit phases. Blocks, reducing communication costs. Through these three stages of negotiation and voting, the consistency of block data between consensus nodes is finally achieved. Only blocks that have reached consensus among the representative nodes can be broadcast to other nodes in the blockchain network.

A representative mechanism is adopted, and some nodes are selected as representative nodes participating in the consensus according to the trust relationship between the nodes. Representative selection is a cyclical behavior, which means that as long as any node works strictly and honestly, it will have the opportunity to become a representative node, obtain accounting rights and create blocks, so it has better decentralization characteristics than equity-based representative mechanisms. .

Adopting a representative mechanism to professionalize bookkeeping nodes reduces the cost of consensus and saves energy consumption, while improving the efficiency of node consensus and the scalability of the algorithm. By monitoring the interaction between nodes and the validity of the block data, the node with a high trust value is selected as the representative node. It does not rely on the token system and avoids the shortcomings of concentrating accounting rights on a few "rich people." Application prospects.

The trust value is formed by the iterative convergence of the trust matrix, which is essentially a probability, which is the basis for the nodes on the entire network to judge a certain node, and has relative stability. The trust relationship between nodes changes slowly, so the update cycle of a representative node is much longer than the block generation cycle. On the other hand, the number of representative nodes is relatively fixed, meaning that the number of nodes in the network does not affect the efficiency of running fault-tolerant Byzantine protocols between the representative nodes.

In order to strictly guarantee the security of the consensus, the fault-tolerant Byzantine algorithm is required to have 1/3 fault tolerance. However, the representative nodes are not completely randomly selected, but are strictly selected based on the trust relationship between the nodes. The trust value is greater than other nodes in the network, so it has a fault tolerance of more than 1/3.

The trust mechanism is introduced into the blockchain in the blockchain. Since transactions and blocks are the most important data, the consensus method proposes a type of node suitable for the blockchain network based on the transactions and block data that are exchanged between nodes. A Quantitative Approach to Trust Relationships. Each node independently judges the trust value of other nodes, and different nodes may have different trust values for the same node. The trust value between nodes is affected by interactive transactions and block data, and changes dynamically over time in the range of 0 to 1. However, the trust value of any node is not absolutely reliable. It is essentially a probability and the basis for other nodes to judge it.

The mutual trust relationship between nodes is used to iteratively converge to obtain the trust value of nodes in the entire network. A node with a high trust value is selected as the representative node and has the opportunity to obtain the right to account. The representative node usually has a higher trust value, which improves the decentralization characteristics and also enhances the security and reliability of the system. This method of selecting representative nodes does not depend on the token system and avoids the shortcomings of bookkeeping rights concentrated on a few "rich people".

The communication cost of the fault-tolerant Byzantine protocol cannot be used in large-scale blockchain networks and has poor scalability. The trust-based consensus mechanism algorithm proposed in this article, from another perspective, specializes in bookkeeping nodes through the trust mechanism, improves the trustworthiness of bookkeeping nodes, avoids fault-tolerant Byzantine consensus throughout the network, and improves the scalability and efficiency of consensus .

The above description is only the preferred embodiments of the present invention and is not intended to limit the present invention. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention shall be included in the protection of the present invention. Within range.

Claims

A blockchain consensus method based on a trust relationship is characterized in that the blockchain consensus method includes the following steps:

S1. Quantify the trust relationship between nodes based on transaction and block data;

S2. Construct a trust relationship graph and generate a trust matrix according to the trust relationship between nodes;

S3. Iteratively calculate the trust value of each node by using the trust relationship of the nodes of the entire network, and its function formula is: T i = C T T i-1 ;

S4. Each round of consensus randomly selects a representative node as the master node with accounting rights. The master node selects several transactions from the transaction pool to create blocks;

Among them, C represents the trust matrix, T i represents the node trust value vector after the i-th iteration, and T i-1 represents the node trust value vector after the i-1 iteration.
The blockchain consensus method according to claim 1, wherein the majority of attacks in the blockchain are related to transactions and blocks, and the attacks include creating false transactions, broadcasting invalid transactions, creating false blocks, and broadcasting Invalid block.
The blockchain consensus method according to claim 2, wherein in step S1, the receipt of a valid transaction and a valid block sent by node i by node i will increase node i's trust in node j, and vice versa Reduce trust; trust calculation formula is
Among them, g ij represents the number of valid transactions and blocks that node i received from node j, u ij represents the number of invalid transactions and blocks that node i received from node j, and β represents a penalty coefficient for invalid data.
The blockchain consensus method according to claim 3, characterized in that, in step S2, in order to prevent malicious nodes from giving higher trust values to other malicious nodes, lower normal trust values are given to normal nodes, thereby affecting trust representative nodes. Selection, so use
Regularize the trust value between nodes to get the final trust matrix C n × n ; where c ij represents the normalized trust value of node i to node j, t ij represents the trust degree of node i to node j, and n represents The total number of nodes in the network.
The blockchain consensus method according to claim 4, characterized in that the trust relationship matrix constructed in the step S2 has random, irreducible and non-periodic properties, so the iterative process of the node trust value matrix T i = C T T i-1 is able to converge.
The blockchain consensus method according to claim 5, characterized in that the trust value between the nodes in step S3 is affected by the interaction of transactions and block data, and changes dynamically with time in the range of 0 to 1. The k nodes with a high trust value are selected as the representative nodes, and each representative node has the opportunity to obtain the right to book.
The blockchain consensus method according to claim 6, characterized in that in step S3, a representative node is selected according to a trust relationship, the representative node participates in a fault-tolerant Byzantine agreement, specializes a bookkeeping node, and improves the scalability of the consensus protocol.
The blockchain consensus method according to claim 7, characterized in that the network transmission delay is considered in the step S4, and the interval from the creation of a block to the final joining of the blockchain is
Block generation cycle
Therefore, at the beginning of the new round of consensus, other nodes have received the previous block data.
The consensus method for a blockchain according to claim 8, characterized in that the step S4 considers that the blockchain network has relative stability over a period of time, and therefore the period t 0 of the update representative is much larger than the block generation period t, improve the efficiency of consensus.
The blockchain consensus method according to claim 9, characterized in that, in step S4, the representative nodes create a block through a fault-tolerant Byzantine agreement, and then spread it to other nodes in the blockchain network. The rule of the longest chain as the main chain guarantees the final consistency of the node data.