CN116389040A - Reputation-based block chain consensus method, device and computer equipment - Google Patents

Abstract

The present invention relates to the technical field of blockchain, and provides a reputation-based blockchain consensus method, device and computer equipment. The method includes: counting historical consensus voting information of nodes, and according to the historical consensus voting information and malicious voting Perform logistic regression calculation on the penalty weight to obtain the current trust degree of the node; use the reputation correction threshold to carry out weight balance on the previous round reputation value and the current trust degree of the node to determine the current round reputation value; The current round of reputation value of the node is used for the current round of consensus. Adopting this method can avoid rights and interests crushing attacks, bribery attacks, Sybil attacks, etc., thereby improving robustness and ensuring decentralization.

Description

Reputation-based block chain consensus method, device and computer equipment

technical field

The invention belongs to the technical field of block chains, and in particular relates to a reputation-based block chain consensus method.

Background technique

Blockchain technology, also known as distributed ledger technology, can solve the problem of agreement between untrusted parties. It has the characteristics of decentralization, traceability, anonymity, and auditability. The consensus protocol is used to ensure the consistency of data copies maintained by all nodes in the blockchain network, avoiding data inconsistency and information asymmetry, and is an algorithm used to drive the blockchain forward. At present, blockchain technology uses three mainstream consensus protocols, including PoW (Proof of Work, workload proof mechanism) protocol, PoS (Proof of Stake, equity certification mechanism) protocol and PBFT (Practical Byzantine Fault Tolerance, Byzantine fault tolerance protocol) protocol. Among them, the PoW protocol is usually called the Nakamoto consensus, and the nodes in the consensus must repeatedly calculate the hash value to reach a consensus. The calculation of the hash value makes it necessary to consume computing power to generate blocks in PoW, and as the block height increases, the consumed computing power also increases, which causes unnecessary waste. In addition, in order to ensure security, transactions in PoW must wait for six blocks to be confirmed, which greatly reduces the throughput of the PoW protocol.

The core principle of the consensus reached by the PoS protocol is based on the equity held by the nodes rather than the computing power in PoW, which has been applied to PPCoin, which can effectively avoid the waste of computing power in the PoW consensus. However, since PPCoin uses coin age as a weight, the coin's "age" will increase even if the node is not connected to the network. As shown in Figure 1, node J accumulates up to 70% of the weight by intentionally going offline from the network. It allows node J to gradually dominate the consensus and even determine the generation of new blocks. In other words, a node dominates the bookkeeping of the blockchain, which does not conform to the decentralized nature of the blockchain. The derivative protocol PoA combines PoS with Nakamoto consensus, so that only online nodes can obtain mining income and transaction fees. However, these protocols rely on the rights and interests of nodes to reach consensus, which cannot guarantee the robustness of the blockchain. As shown in Figure 2, node O with low stake can try different sequences of stakeholders (fork) to maximize its own interests, because fork does not consume any resources, which is called stake crushing attack in the blockchain . Node P with high equity will continue to bribe other nodes, and once their equity reaches more than 50%, they can control the entire blockchain to launch double-spending attacks, which is always beneficial as long as the bribe amount is less than the commodity transaction price However, this is known as a bribery attack in blockchain. There is also a random election competition mechanism Follow-the-satoshi algorithm, which differs from PoS in that it uses secure multi-party computing to keep the leader unpredictable, but it assumes that most nodes participating in the joint coin speculation agreement It will not be destroyed by the opponent for a long time. In general, several protocols derived from the above-mentioned PoS-based protocols are vulnerable to rights smashing attacks, bribery attacks, and security assumptions.

The PBFT protocol is improved from the BFT (Byzantine Fault Tolerance) protocol. Compared with the PoW protocol and the PoS protocol, it can achieve good fault tolerance and excellent performance. But in the PBFT protocol, the nodes in the blockchain are equal and can participate at will. Therefore, it is easy for malicious nodes to create any number of pseudonyms, as shown in Figure 3, and then use these pseudonyms to vote for their own improper interests (that is, double consumption), thus easily destroying the security of the network, and even controlling the entire network system, which is known as a Sybil attack in blockchain. Although it has been proposed to combine the Nakamoto consensus with the PoS or BFT protocol, and to completely replace the Nakamoto consensus with the BFT protocol to avoid Sybil attacks. However, all of these protocols achieve consensus based on a node’s recent resources, stake, or external trust, potentially creating an undesirable hierarchy among nodes and compromising the decentralization of the blockchain system.

Contents of the invention

Based on this, it is necessary to address the above technical problems and provide a reputation-based blockchain consensus method, device, computer equipment and storage medium.

The present invention provides a reputation-based block chain consensus method, including:

Statistical historical consensus voting information of the node, performing logistic regression calculation according to the historical consensus voting information and malicious voting penalty weight, to obtain the current trust degree of the node;

performing weight balance on the last round of reputation value of the node and the current trust degree by using the reputation correction threshold to determine the current round of reputation value;

A current round of consensus is performed according to the current round of reputation values of each of the nodes.

In one of the embodiments, the weight balancing of the previous round of reputation value and the current trust degree of the node by using the reputation correction threshold to determine the current round of reputation value includes:

Calculate the difference between the absolute value of the current trust degree and the reputation value of the previous round, and calculate the cumulative deviation of the current round according to the difference between the absolute value and the cumulative deviation determined by the previous round of consensus;

calculating a reputation correction threshold according to the difference between the absolute values and the cumulative deviation of the current round;

Perform weight balance on the last round of reputation value and the current trust degree by using the reputation correction threshold to determine the current round of reputation value.

In one of the embodiments, the weight balancing of the previous round of reputation value and the current trust degree of the node by using the reputation correction threshold, before determining the current round of reputation value, further includes:

Judging whether the node has offline non-voting behavior in the historical consensus voting according to the historical consensus voting information;

When it is determined that there is an offline non-voting behavior, the current trust degree is retained, and when it is determined that there is no offline non-voting behavior, the reputation consumption of the current trust degree is performed using the block generation time.

In one of the embodiments, the logistic regression calculation formula is as follows:

表示节点i在第x轮共识中是否正常参与投票，正常投票为1，未正常投票为0；τ _x表示节点i在第x轮共识中是否恶意投票，恶意投票为1，未恶意投票为0；υ表示恶意投票惩罚权重；e为自然对数。Among them, repu(i) _cur is the current trust degree; n represents the current nth round of consensus; μ is the cumulative number of consensus rounds participated by node i;

Indicates whether node i participates in voting normally in the x-th round of consensus, normal voting is 1, and abnormal voting is 0; τ _x indicates whether node i voted maliciously in the x-th round of consensus, malicious voting is 1, and non-malicious voting is 0 ; υ represents the malicious voting penalty weight; e is the natural logarithm.

In one of the embodiments, the reputation consumption function is as follows:

=0表示节点i在第x轮共识中离线未投票；e为自然对数；ΔB表示区块生成时间。in,

=0 means that node i was offline and did not vote in the x-round consensus; e is the natural logarithm; ΔB means the block generation time.

In one of the embodiments, the calculation formula of the weight balance is as follows:

Among them, t represents the current t-round consensus; repu(i) _t represents the reputation value of the t-round consensus; repu(i) _t-1 represents the reputation value of the t-1 round; γ represents the reputation correction threshold, and the initial value of γ The value is 1.

In one of the embodiments, the calculation formula of the reputation correction threshold is as follows:

Among them, thershold is the set threshold, thershold =0.25; c is the definition parameter, c=0.6; φ _t repu(i) indicates the difference of the absolute value of node i, ψ _t repu(i) indicates the cumulative deviation of the current round, ψ _{t -1} repu(i) represents the cumulative bias determined by the last round of consensus.

A reputation-based blockchain consensus device, including:

The reputation determination module is used to count the historical consensus voting information of the node, and perform logistic regression calculation according to the historical consensus voting information and the malicious voting penalty weight to obtain the current trust degree of the node;

A reputation correction module, configured to use a reputation correction threshold to perform weight balance on the previous round of reputation value of the node and the current trust degree, and determine the current round of reputation value;

A consensus module, configured to perform a current round of consensus according to the current round of reputation values of each of the nodes.

The present invention also provides a computer device, the computer device includes a processor and a memory, the memory stores a computer program, and when the processor executes the computer program, the steps of the above reputation-based block chain consensus method are realized.

The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above reputation-based block chain consensus method are realized.

The above reputation-based blockchain consensus method, device, computer equipment and storage medium achieve consensus by introducing reputation as the voting weight of consensus nodes. Since the total reputation of nodes will not change, it can effectively resist Sybil attacks. Moreover, logistic regression is used to dynamically update the reputation of nodes based on the behavior of nodes, and punish malicious nodes to reduce the probability of them launching stake smashing attacks and bribery attacks. At the same time, using the reputation correction threshold to correct the reputation makes the reputation grow very slowly, which ensures the decentralized nature of the blockchain.

Description of drawings

Figure 1 is a schematic diagram of the centralization of blockchain node rights and interests caused by the accumulation of currency age in an embodiment.

Fig. 2 is a schematic diagram of a low-stake node trying to fork a high-stake node to perform bribery in an embodiment.

Fig. 3 is a schematic diagram of a malicious node implementing a Sybil attack in an embodiment.

Fig. 4 is a schematic flow diagram of a blockchain consensus method based on reputation in an embodiment.

Fig. 5 is a schematic diagram of the comparison curve of the delay performance between the consensus protocol of the present invention and the PBFT protocol under the change of the block size in an embodiment.

Fig. 6 is a schematic diagram of the throughput performance curve of the consensus protocol of the present invention and the PBFT protocol under the change of the block size in an embodiment.

Fig. 7 is a schematic diagram of the delay performance curve of the consensus protocol of the present invention and the PBFT protocol under the change of the number of nodes in an embodiment.

Fig. 8 is a schematic diagram of the throughput performance curve of the consensus protocol of the present invention compared with the PBFT protocol under the change of the number of nodes in an embodiment.

Fig. 9 is a structural block diagram of a reputation-based block chain consensus device in an embodiment.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

In one embodiment, as shown in Figure 4, a reputation-based block chain consensus method is provided, comprising the following steps:

In step S101, the historical consensus voting information of the nodes is counted, and logistic regression calculation is performed according to the historical consensus voting information and malicious voting penalty weights to obtain the current trust degree of the node.

Step S103, using the reputation correction threshold to weight balance the previous round of reputation value and current trust degree of the node, and determine the current round of reputation value.

Among them, the historical consensus voting information refers to the voting situation that the nodes have agreed on before the current round of consensus, including whether they normally participate in voting in each round of consensus, whether they vote maliciously in each round of consensus, etc. The malicious voting penalty weight is the preset penalty weight for malicious voting. It is set based on actual needs. The greater the malicious voting penalty weight, the greater the penalty for malicious voting of nodes. The main commonly used model for studying dichotomous problems is the logistic regression model, which describes the relationship between a set of independent variables and a binary dependent variable. It uses a sigmoid function based on linear regression, so that the prediction result is in the [0,1] interval. In this embodiment, the dynamic authorization of reputation can be realized by establishing a logistic regression model to evaluate the change of node reputation and its consensus behavior. As part of the consensus protocol, this model can be enforced in each node participating in the consensus. The reputation is calculated independently, and the calculation of the reputation is determined by the behavior of the nodes when implementing the consensus, and can be broadcast simultaneously with the consensus information. According to the degree of influence of node behavior on the block consensus result, this embodiment divides the behavior of nodes into the following three types: #good: normally receive and broadcast the transaction information list in the network, and send the same information list to other nodes. #bad: Receive and broadcast transaction information lists in the network normally, and send different information lists or forged information lists to other nodes. And #offline: Do not receive and broadcast the list of transaction information in the network. Since the behavior of nodes in each round of consensus is not fixed, the reputation of this embodiment is a process of dynamic change. In the model, the set reputation value R is a real number from 0 to 1, and the larger the reputation value, the higher the credibility. For the newly added consensus nodes in the system, their initial reputation value is set to 0.5.

The reputation correction threshold is a threshold set for correcting the growth rate of reputation. Through the reputation correction threshold, the reputation value of the node will not grow too fast, avoiding the reputation centralization in the initial stage of the network, because it is not known at the beginning Whether the node has a tendency to do evil, so the initial reputation correction threshold γ is set to 1. The last round of reputation value is the reputation of the node in the last round of consensus, and the reputation of the node at the beginning of the t-th round of consensus is equal to the reputation at the end of the t-1 round of consensus. The current round of reputation value is the finalized reputation of the current round of consensus, and the current trust degree is the reputation determined before modification.

Specifically, firstly, the current trust degree of the node is calculated according to the logistic regression model by using parameters such as historical consensus voting information and malicious voting penalty weight. Then use the reputation correction threshold to balance the current trust degree and the previous round of reputation value to obtain the corrected reputation value as the reputation value of the node's current round of consensus, that is, the current round of reputation value. Through the correction, the logistic regression model can be avoided. The reputation value grows rapidly and destabilizes the consensus agreement.

In one embodiment, step S101, the calculation formula of the node reputation measurement function based on the logistic regression model is as follows:

表示节点i在第x轮共识中是否正常参与投票，正常投票为1，未正常投票为0；τ _x表示节点i在第x轮共识中是否恶意投票，恶意投票为1，未恶意投票为0；υ表示恶意投票惩罚权重；e为自然对数。本实施例共识协议的运行从时间上被划分为一系列轮数，每一轮产生一个区块，执行相同的步骤。信誉则以每轮开始前的历史值进行计算，通过节点在每轮中的诚实签名和恶意签名行为的判别，进行动态的信任度授信。Among them, repu(i) _cur is the current trust degree; n represents the current nth round of consensus; μ is the cumulative number of consensus rounds participated by node i, that is, the round length;

Indicates whether node i participates in voting normally in the x-th round of consensus, normal voting is 1, and abnormal voting is 0; τ _x indicates whether node i voted maliciously in the x-th round of consensus, malicious voting is 1, and non-malicious voting is 0 ; υ represents the malicious voting penalty weight; e is the natural logarithm. The operation of the consensus protocol in this embodiment is divided into a series of rounds in terms of time, and each round generates a block and performs the same steps. The reputation is calculated based on the historical value before the start of each round, and the dynamic trust degree credit is carried out through the discrimination of the honest signature and malicious signature behavior of the node in each round.

In one embodiment, step S103 includes: calculating the difference between the absolute value of the current trust degree and the previous round of reputation value, calculating the current round of cumulative deviation according to the difference between the absolute value and the cumulative deviation determined by the previous round of consensus; The reputation correction threshold is calculated based on the difference and the cumulative deviation of the current round; the reputation correction threshold is used to balance the weight of the previous round of reputation value and the current trust degree to determine the current round of reputation value.

Specifically, due to the rapid growth of the reputation value of nodes in the logarithmic growth period of a model such as logistic regression, it is impossible to punish malicious nodes that deliberately performed well in the early stage of consensus, which is not conducive to the stable implementation of the consensus protocol. Therefore, in order to avoid this situation, this embodiment further modifies the model. According to the current reputation of the node and the reputation of the last round of consensus, the weight of the reputation in the consensus voting is balanced, and the revised reputation measurement function , that is, the calculation formula of weight balance is as follows:

Among them, t represents the current t-round consensus; repu(i) _t represents the reputation value of the t-round consensus; repu(i) _t-1 represents the reputation value of the t-1 round; γ represents the reputation correction threshold, and the initial value of γ The value is 1, and the increase rate of reputation can be corrected through γ, so that it will not grow too fast, avoiding the centralization of reputation in the initial stage of the network. Among them, the calculation formula of reputation correction threshold γ is as follows:

Wherein, thershold is a set threshold, which is set to prevent γ from being too saturated and approaching 1. In this embodiment, it is preferred that thershold is set to 0.25, which can prevent γ from being too saturated and approaching 1, and also allow repu(i ) _t will not rely too much on the reputation obtained by direct evaluation of a current reputation function. c is the definition parameter used to control the weight of the reaction to the most recent behavior of the node. The larger c is, it means that the weight of the latest reputation deviation given by the user is higher than the previous cumulative reputation deviation. In this embodiment, c=0.6 is preferred. φ _t repu(i) represents the difference between the absolute value of the current trust degree of node i and the reputation value of the previous round. ψ _t repu(i) represents the cumulative deviation of the current round, and ψ _t-1 repu(i) represents the cumulative deviation determined by the previous round of consensus, that is, the final cumulative deviation of the previous t rounds of the node.

In one embodiment, after step S101 and before step S103, it also includes: judging whether the node has an offline non-voting behavior in the historical consensus voting according to the historical consensus voting information; when it is determined that there is an offline non-voting behavior, retain the current trust degree, when When it is determined that there is no offline non-voting behavior, the current trust degree is consumed by using the block generation time.

Among them, since reputation is the manifestation of the probability of a node being selected as a committee and the consensus voting weight, the greater the reputation, the higher the probability of participating in generating blocks, which brings considerable benefits. In order to maintain the number of online nodes in the network, this embodiment further introduces a reputation consumption function. That is, the method for dynamically awarding reputation in this embodiment consists of two functions: a reputation measurement function and a reputation consumption function. Reputation consumption is set to ensure the participation of nodes, not to allow nodes to participate in the generation of blocks by consuming reputation. Because when the number of online members in the committee is too small, the security and stability of the network cannot be guaranteed.

Specifically, the reputation consumption of the node is performed before the reputation is revised, and is mainly determined based on the historical voting situation of the node. When it is judged according to the historical consensus voting information that the node has offline non-voting behavior in the historical consensus voting, the current trust degree is retained without credit consumption, and when it is determined that there is no offline non-voting behavior, the current trust degree is determined by the block generation time. Carry out credit consumption.

In one embodiment, the reputation consumption function is as follows:

=0表示节点i在第x轮共识中离线未投票；e为自然对数；ΔB表示区块生成时间。提高区块生成时间，意味着区块上链的难度增加，这时候理性节点可能会选择暂时离线（#offline），以寻求上链难度更低、机会更大的时候参与区块生成。越来越多的节点开始离线，节点参与共识的比例也越来越低，恶意节点成功的概率就会变得很大。因此在区块生成时间增高的同时，也增大对于不参与区块生成的节点的信誉消耗，将有助于提高节点的参与度，维护区块链网络的运行安全。in,

=0 means that node i was offline and did not vote in the x-round consensus; e is the natural logarithm; ΔB means the block generation time. Improving the block generation time means that the difficulty of block chaining increases. At this time, rational nodes may choose to temporarily go offline (#offline) in order to participate in block generation when the chaining difficulty is lower and the opportunity is greater. More and more nodes are going offline, and the proportion of nodes participating in the consensus is getting lower and lower, and the probability of success of malicious nodes will become very high. Therefore, while the block generation time increases, it also increases the reputation consumption of nodes that do not participate in block generation, which will help increase the participation of nodes and maintain the operational security of the blockchain network.

In step S105, the current round of consensus is performed according to the current round of reputation values of each node.

Specifically, after obtaining the current round reputation value of a node, the current round of consensus can be carried out based on the reputation value during the consensus process. It should be understood that the current round reputation value of each node can be pre-calculated before the consensus, or can be calculated in real time by each node when needed during the consensus process. The consensus in this embodiment is divided into two processes: master node update algorithm and improved PBFT consensus algorithm.

。其中，N是共识节点的集合，N={N₁ , N₂ ,... N_Q }。在每一轮的共识中，有一个主要的主节点Np，其他节点为备份节点。所有的共识节点首先在本地缓存收到的交易信息。而Np将客户端发送的有效交易打包成一个区块。每个节点都会维护一个更新的节点信息表NIL。NIL包含了当前区块链系统中的相邻节点的集合与该节点的信息、信誉（对应于投票权重）以及节点ID，节点的初始信誉为0.5。P是被选为主节点的概率。D是指数分布(C2C网络和大多数社会网络中的信誉分布网络中的声誉分布是指数型)。主节点更新主要在副本节点怀疑主节点或主节点在共识开始后的设置时间内没有发起共识提议时进行，副本节点选择当前节点中信誉值最大的节点作为主节点，并向其他节点广播主节点变更消息。而节点收到主节点变更消息后，计算收到的消息数量是否超过2f(f为拜占庭节点数量)，超过后即对主节点进行更新。The master node update algorithm is to replace the master node according to the reputation of the node. Assume that the higher the reputation value of the node, the higher the probability of being selected as the master node during the master node update process, for example:

. Among them, N is the set of consensus nodes, N={N ₁ , N ₂ ,... N _Q }. In each round of consensus, there is a main master node Np, and other nodes are backup nodes. All consensus nodes first cache the received transaction information locally. And Np packs the valid transactions sent by the client into a block. Each node maintains an updated node information table NIL. NIL contains the set of adjacent nodes in the current blockchain system and the node's information, reputation (corresponding to voting weight) and node ID, and the initial reputation of the node is 0.5. P is the probability of being selected as a master node. D is an exponential distribution (reputation distribution in C2C networks and most social networks is exponential). The update of the master node is mainly carried out when the replica node suspects that the master node or the master node has not initiated a consensus proposal within the set time after the start of the consensus. The replica node selects the node with the highest reputation value among the current nodes as the master node, and broadcasts the master node to other nodes Change message. After the node receives the master node change message, it calculates whether the number of received messages exceeds 2f (f is the number of Byzantine nodes), and updates the master node after exceeding.

In the improved PBFT consensus algorithm, this embodiment evaluates the reputation value of each node based on the information list NIL provided by each consensus node, which can not only detect malicious nodes among them, but also delete these nodes from NIL. And because of the high punishment for malicious nodes, the reputation of honest nodes will gradually accumulate, so that the right to speak in the consensus process will gradually increase, and the influence of malicious nodes will gradually decrease. The improved PBFT algorithm of this embodiment is as follows:

Input: Transaction Tx.

Output: New block B _n containing transaction Tx.

Step 1: Client c: Broadcast transaction cluster TxCluster to master node 0.

Step 2: Master node 0: Verify whether the transaction Tx in the transaction cluster is valid, and delete it directly if it is invalid. Then pack the valid transaction into the block, and generate the block header B _head according to the message in the block body.

，其中h是当前新区块的高度，d是区块头B_head的摘要，t是当前的时间戳，D₀是当前主节点0的ID，NIL₀是主节点0的节点信息列表。Step 3: master node 0: broadcast a pre-prepare message to each copy node, the content is

, where h is the height of the current new block, d is the summary of the block header B _head , t is the current timestamp, D ₀ is the ID of the current master node 0, and NIL ₀ is the node information list of master node 0.

，D_i是副本节点i的ID ，NIL_i是副本节点i的节点信息列表。Step 4: Replica nodes 1 and 2: After receiving the pre-preparation message sent by the master node 0, they first verify the validity of the new block, and if the verification passes, they send prepare messages to other nodes, the content of which is

, D _i is the ID of replica node i, and NIL _i is the node information list of replica node i.

。Step 5: Replica nodes 1 and 2: Calculate the reputation value R of the node currently sending the message to it, if R ≥ R _thershold , update the consensus status of the transaction information locally and send the pre_commit (pre-submit) information, the content is

.

；Step 6: Master node 0: Compare the pre-commit information received from the copy node, calculate the current reputation of each node according to the improved logistic regression, and update the information list of the local consensus node, and send the consensus result back to the client and All replica nodes, and send a commit message

;

Step 7: Replica nodes 1 and 2: After completing the submission status, update the local consensus node information list, and feed back the consensus results to the client to prepare for the next round of consensus process.

In addition, the condition for a node in this embodiment to update its consensus state is that the reputation value R of the consensus node sending a message to it is sufficiently large. R should not be lower than the set threshold R _thershold , the calculation formula of the threshold R _thershold is as follows:

It can be seen that the consensus process of the improved PBFT consensus algorithm in this embodiment is divided into four main stages: pre-preparation, preparation, pre-submission and submission. By adding a pre-commit phase to the consensus protocol, this not only ensures that the nodes reach consensus correctly, but also improves performance.

Compared with existing technologies, the above reputation-based blockchain consensus method has the following advantages:

Resisting Sybil attacks: Consensus is reached by introducing reputation as the voting weight of consensus nodes. Since the total reputation of these nodes will not change, malicious nodes cannot create a large number of fake nodes to increase their influence and undermine network security, thus effectively avoiding Sybil attacks.

Avoid stake crushing attacks and bribery attacks: Malicious nodes cannot bear high reputation penalties due to the existence of penalty weights. For example, in the first round of consensus, the reputation of a node after casting a malicious vote is only 36% of that of an honest vote. Moreover, the reputation penalty will rise rapidly with the increase of the number of consensus rounds. The existence of such punishment is unacceptable to any rational node, so as to reduce the probability of them launching rights smashing attacks and bribery attacks to avoid attacks.

Guarantee the nature of decentralization: Correction through the reputation correction threshold makes the growth of reputation very slow, and there is an upper limit for reputation between 0 and 1, which guarantees the decentralized nature of the blockchain.

Low communication overhead: the dynamic authorization of node reputation is added in the process of node consensus, and the pre-submission stage is added in the process of node consensus. Although the time complexity of the present invention is still O(N ² ), the total communication volume is (n ² +2n-3), compared with the total number of PBFT communications (2n ² -n-1), the number of communications between nodes in the present invention is reduced.

In addition, the above-mentioned reputation-based blockchain consensus method also has the advantage of high performance. As shown in FIG. 5 , FIG. 6 , FIG. 7 and FIG. 8 , a schematic diagram of the performance experiment results of the consensus protocol of the present invention (hereinafter referred to as RIPPB) and the traditional PBFT protocol is provided.

Referring to Fig. 5 and Fig. 6, it is a schematic diagram of curves comparing performance such as delay and throughput by changing the block size when the number of nodes is fixed at 20. It can be found that the latency (Latency) of RIPPB and PBFT increases significantly as the block size increases, but the growth rate of RIPPB is 30% lower than that of PBFT. In addition, an increase in the block size means an increase in the number of transactions in the block, which increases the throughput (TPS) of the PBFT and RIPPB protocols. Obviously, the throughput of RIPPB is higher than that of PBFT. However, due to the limitation of network capacity, when the block size is close to or exceeds 4M, the growth rate of throughput will be significantly slower.

Referring to Figure 7 and Figure 8, it is a schematic diagram of curves comparing performance such as delay and throughput by changing the number of nodes when the block size is fixed at 1/8MB. It can be found that compared with the PBFT protocol, the RIPPB protocol greatly reduces the delay in generating blocks by consensus nodes. Moreover, the reduction in delay of block generation by consensus nodes becomes more pronounced as the number of consensus nodes increases. Compared with the PBFT protocol, the throughput of RIPPB has been improved in an all-round way, up to 28%, and the decline rate of throughput slows down as the number of consensus nodes increases.

In one embodiment, as shown in Figure 9, a credit-based block chain consensus device is provided, including:

The reputation determination module 901 is used to count the historical consensus voting information of the node, and perform logistic regression calculation according to the historical consensus voting information and malicious voting penalty weight to obtain the current trust degree of the node.

The reputation correction module 903 is configured to use the reputation correction threshold to carry out weight balance between the previous round of reputation value and the current trust degree of the node, and determine the current round of reputation value.

Consensus module 905, configured to perform current round consensus according to the current round reputation value of each node.

In one embodiment, the reputation correction module 903 is also used to calculate the difference between the absolute value of the current trust degree and the previous round of reputation value, and calculate the cumulative deviation of the current round according to the difference between the absolute value and the cumulative deviation determined by the previous round of consensus; The difference between the absolute value and the cumulative deviation of the current round is used to calculate the reputation correction threshold; the reputation correction threshold is used to balance the weight of the previous round of reputation value and the current trust degree to determine the current round of reputation value.

In one embodiment, it also includes a reputation consumption module, which is used to judge whether the node has offline non-voting behavior in the historical consensus voting according to the historical consensus voting information; when it is determined that there is an offline non-voting behavior, the current trust degree is retained; When there is offline non-voting behavior, the current trust degree is consumed by using the block generation time.

For the specific limitations of the reputation-based blockchain consensus device, please refer to the limitations of the reputation-based blockchain consensus method above, and will not be repeated here. Each module in the above reputation-based block chain consensus device can be fully or partially realized by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of the above modules. Based on this understanding, the present invention realizes all or part of the processes in the methods of the above embodiments, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer When the program is executed by the processor, it can realize the steps of the above embodiments of the reputation-based blockchain consensus method. Wherein, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form.

In one embodiment, a computer device is provided, which may be a server, and includes a processor, a memory, and a network interface. Wherein, the processor of the computer device is used to provide calculation and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal via a network connection. The computer program, when executed by the processor, implements a reputation-based blockchain consensus method. Exemplarily, a computer program can be divided into one or more modules, and one or more modules are stored in a memory and executed by a processor to implement the present invention. One or more modules may be a series of computer program instruction segments capable of accomplishing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device. The so-called processor can be a central processing unit (Central Processing Unit, CPU), and can also be other general-purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., and the processor is the control center of the computer device, using various interfaces and lines to connect various parts of the entire computer device.

The memory can be used to store the computer programs and/or modules, and the processor realizes the computer by running or executing the computer programs and/or modules stored in the memory and calling the data stored in the memory Various functions of the device. The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required by at least one function (such as a sound playback function, an image playback function, etc.) and the like; the storage data area may store Data created based on the use of the mobile phone (such as audio data, phonebook, etc.), etc. In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disk, internal memory, plug-in hard disk, smart memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card , a flash memory card (Flash Card), at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

The technical features of the above embodiments can be combined arbitrarily. To make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be It is considered to be within the range described in this specification.

The above-mentioned embodiments only represent several implementation modes of the present application, and the description thereof is relatively specific and detailed, but it should not be construed as limiting the scope of the patent for the invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the scope of protection of the patent application should be based on the appended claims.

Claims (10)

1. A reputation-based blockchain consensus method, comprising:

counting historical consensus voting information of a node, and performing logistic regression calculation according to the historical consensus voting information and malicious voting penalty weight to obtain the current trust degree of the node;

performing weight balance on the last-round reputation value of the node and the current trust degree by using a reputation correction threshold value to determine a current-round reputation value;

and carrying out current round consensus according to the current round credit values of the nodes.

2. The method of claim 1, wherein the weighting the previous round of reputation values for the node and the current trust level using a reputation correction threshold to determine a current round of reputation values comprises:

calculating the difference between the current trust and the absolute value of the credit value of the previous round, and calculating the current round of accumulated deviation according to the difference between the absolute values and the accumulated deviation determined by the consensus of the previous round;

calculating a reputation correction threshold according to the difference between the absolute values and the current wheel accumulated deviation;

and carrying out weight balance on the previous-round reputation value and the current trust degree by utilizing a reputation correction threshold value, and determining the current-round reputation value.

3. The method according to claim 1 or 2, wherein the weighting of the previous round of reputation values of the node and the current trust level using a reputation correction threshold, before determining the current round of reputation values, further comprises:

judging whether the node has offline non-voting behaviors in the history consensus voting according to the history consensus voting information;

and when the offline non-voting behavior is determined to exist, the current trust level is reserved, and when the offline non-voting behavior is determined to not exist, the block generation time is utilized to conduct reputation consumption on the current trust level.

4. The method of claim 1, wherein the logistic regression calculation formula is as follows:

wherein, repu (i) _cur Is the current confidence level; n represents the current nth round consensus; mu is the cumulative number of consensus rounds participated by the node i;

indicating whether the node i normally participates in voting in the x-th round of consensus, wherein the normal voting is 1, and the abnormal voting is 0;τ _x indicating whether the node i maliciously votes in the x-th round of consensus, wherein the maliciously votes are 1, and the maliciously votes are 0;υrepresenting malicious voting penalty weights; e is the natural logarithm.

5. The method of claim 3, wherein the reputation consumption function is as follows:

wherein,,

=0 means node i was not voted offline in the x-th round of consensus; e is natural logarithm; Δb represents the block generation time.

6. The method according to claim 1 or 2, characterized in that the weight equalization is calculated as follows:

wherein t represents the current t-th round consensus; repu (i) _t A reputation value representing a t-th round consensus; repu (i) _t-1 Representing the reputation value of the t-1 th round; gamma denotes a reputation correction threshold, and the initial value of gamma is 1.

7. The method of claim 2, wherein the reputation correction threshold is calculated as:

wherein,,thersholdfor the set threshold value(s),thershold=0.25; c is a defined parameter, c=0.6;φ _t the repu (i) represents the difference in absolute value of node i,ψ _t the repu (i) represents the current wheel accumulated bias,ψ _t-1 the repu (i) represents the accumulated bias of the previous consensus determination.

8. A reputation-based blockchain consensus apparatus, comprising:

the credit determining module is used for counting historical consensus voting information of the node, and carrying out logistic regression calculation according to the historical consensus voting information and malicious voting penalty weight to obtain the current trust degree of the node;

the reputation correction module is used for carrying out weight balance on the previous-round reputation value of the node and the current trust degree by utilizing a reputation correction threshold value to determine the current-round reputation value;

and the consensus module is used for carrying out current round consensus according to the current round credit value of each node.

9. A computer device comprising a processor and a memory, the memory storing a computer program, wherein the processor is configured to implement the blockchain consensus method of the reputation of any of claims 1-7 when executing the computer program.

10. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the reputation blockchain consensus method according to any of claims 1-7.

Images