CN110138597B

CN110138597B - Block chain DPOS (distributed DPOS) consensus mechanism improvement method based on credit integration and node clustering

Info

Publication number: CN110138597B
Application number: CN201910306559.XA
Authority: CN
Inventors: 冯国瑞; 刘万利
Original assignee: University of Shanghai for Science and Technology
Current assignee: University of Shanghai for Science and Technology
Priority date: 2019-04-17
Filing date: 2019-04-17
Publication date: 2021-11-05
Anticipated expiration: 2039-04-17
Also published as: CN110138597A

Abstract

The invention discloses a block chain DPOS (distributed data processing system) consensus mechanism improvement method based on credit integration and node clustering. The invention defines basic concepts such as credit points and the like aiming at the problems that voting of a DPOS consensus mechanism is not positive, malicious agent nodes cannot be eliminated in time and the like. For each node which is added into the block chain network for the first time, credit integration is initialized to be 100, under credit reward and punishment, the probability that an abnormal node becomes a proxy node can be reduced by voting and counting up the credit integration again for the nodes in the whole network, the nodes in the whole network are clustered by referring to a plurality of characteristic values of the nodes through a K-Means clustering algorithm in data mining, a group with relatively high trust degree has higher probability to win competition rights in the next round of competition, and the generation of invalid blocks or malicious blocks in the block chain network is further reduced.

Description

Block chain DPOS (distributed DPOS) consensus mechanism improvement method based on credit integration and node clustering

Technical Field

The invention relates to a block chain DPOS (distributed data processing system) consensus mechanism improvement method based on credit integration and node clustering.

Background

The blockchain is essentially a decentralized database, and is used as the underlying technology of bitcoin, namely a series of data blocks which are generated by correlation through a cryptographic method, wherein each data block contains information of a batch of bitcoin network transactions for verifying the validity of the information and generating the next blockchain. The consensus mechanism is the core of the block chain, completes the verification and confirmation of the transaction in a short time through the voting of the special nodes, and solves the problem of how to achieve consensus in a network which is lack of trust and is freely opened.

Consensus refers to the result of multiple parties not associated with each other agreeing on a problem. The consensus in the blockchain is mainly studied by accounting right assignment problems and check problems after blockgeneration. At present, the consensus mechanism is researched, and the consensus algorithm existing in the block chain system mainly includes POW, POS, POL, DPOS, and the like. POW (Proof-Of-Work), a workload certification consensus mechanism, has been successfully applied in bitcoin, which can solve the data consistency problem in a completely open, free network, but the mechanism consumes a lot Of computing power and other resources, and the data agreement time is long, which is difficult to meet the general business requirements; the main idea Of POS (Proof-Of-stamp) is that the probability Of obtaining block accounting right by a node is proportional to tokens held by the node, and POS reduces consumption caused by mathematical operation to a certain extent, but is still not suitable for some business environments with higher requirement on consensus time; POL (Proof-Of-Luck) is a common knowledge mechanism built on a trusted execution environment, which can greatly improve the efficiency Of block generation, but this also puts higher requirements on the processor, and needs corresponding hardware support; DPOS (deleted-Proof-Of-Stake) is a consensus mechanism capable Of realizing block chain second-level verification, wherein the DPOS selects the proxy nodes in a voting mode, and finally the proxy nodes complete block generation and verification.

Disclosure of Invention

The invention aims to provide an improved block chain DPOS consensus mechanism method based on credit points and node clustering, aiming at the defects of the existing block chain consensus mechanism in practical application.

As the DPOS consensus mechanism has the problems that voting is not positive, malicious agent nodes cannot be eliminated in time and the like, the invention defines basic concepts such as credit points and the like. Under credit reward and punishment, the probability that the abnormal node becomes the proxy node can be reduced by voting and counting credit integrals through the nodes of the whole network, the nodes of the whole network are clustered by using a K-Means clustering algorithm in data mining and referring to a plurality of attributes of the nodes, a group with relatively high trust degree has higher probability of winning competition right in the next round of competition, and the generation of invalid or malicious blocks is further reduced.

The K-Means clustering algorithm is selected because the K-Means clustering algorithm is an unsupervised clustering algorithm, is concise and clear to realize, and has good clustering effect. The core idea of the K-Means clustering algorithm is as follows: firstly, randomly selecting K seed points, then calculating the distance from all the points to the seed points, bringing the points into a seed point group with the closest distance, moving the seed points to the center of the seed group after all the points are brought into the group, and finally repeating the process until the seed points do not move.

In order to achieve the purpose, the invention adopts the following technical scheme:

a block chain DPOS consensus mechanism improvement method based on credit score and node clustering comprises the following specific operation steps:

1) initializing credit points: the credit score is a credit parameter endowed by a system when the node joins the blockchain network, is an important expression form of the trust degree of the node, and is initialized to be 100 for each node which is added to the blockchain network for the first time;

2) new transaction occurs in the blockchain network, nodes in the whole network compete for accounting after the transaction is finished, and all information of the transaction is recorded in a newly generated block of the blockchain network by the nodes which compete successfully;

3) the system judges whether a block newly generated in the network is an effective block or not, and judges whether the block is a malicious block or not if the block is not the effective block;

4) the whole network node votes the newly generated blocks in the network and divides the votes into a positive vote and a negative vote, if the votes are positive, the node considers the newly generated blocks as valid blocks, and if the votes are negative, the node considers the newly generated blocks as invalid blocks or even malicious blocks;

5) the system counts the voting result, carries out credit reward and punishment on the nodes of the whole network according to the voting result, and recalculates the credit integral of the nodes;

6) setting a credit point threshold, and when the credit point of the node is lower than 95, the system inhibits the node from becoming a proxy node;

7) taking credit points, whether voting is positive, time for the nodes to join the network and the number of tokens held by the nodes as clustering characteristic values, carrying out normalization processing on the characteristic values, and then utilizing a K-Means clustering algorithm to cluster all the nodes in the whole network into four classes;

8) after clustering is finished, the node with the highest credit score, the most positive voting, the longest time for joining the network and the highest token number has the highest trust level, and the nodes in the group have higher probability to obtain the accounting right in the next round of competition; similarly, the class with the lowest credit score, the most negative vote, the shortest time to join the network and the lowest token number has the lowest confidence level, and the probability that the node in the class will get the right to be billed in the next round of competition is greatly reduced.

Compared with the prior art, the invention has the advantages that:

firstly, credit integral is defined, and the probability that an abnormal node becomes an agent node can be reduced by the nodes of the whole network in a voting mode under the credit reward and punishment; secondly, clustering nodes of the whole network by using a K-Means clustering algorithm in data mining and referring to a plurality of attributes of the nodes, wherein the group with relatively high trust degree has a higher probability of winning the accounting right in the next round of competition, so that the generation of invalid blocks or malicious blocks is effectively reduced; finally, the probability of successful competition of the same type of clustered nodes in the next round of competition accounting is the same, so that the possibility that individual nodes with higher trust degree monopolize accounting right is effectively avoided.

Drawings

Fig. 1 is a flowchart of an improved block chain DPOS consensus method based on credit score and node clustering according to the present invention.

FIG. 2 is a diagram illustrating the effect of clustering the block link points by using K-Means according to the present invention.

Detailed Description

In order to facilitate understanding for those skilled in the art, the present invention will be further described with reference to the accompanying drawings and examples.

As shown in fig. 1, an improved method for block chain DPOS consensus based on credit score and node clustering specifically includes the following steps:

1) credit points are initialized. For each node that first joins the blockchain network, the credit score will be initialized to 100;

4) the nodes of the whole network begin to vote for the newly generated blocks in the network, the nodes can vote for the blocks or vote against the blocks, if the votes are voted for, the nodes consider the newly generated blocks as valid blocks, and if the votes are voted for, the nodes consider the newly generated blocks as invalid blocks or even malicious blocks;

5) the system counts the voting result, carries out credit reward and punishment on the nodes of the whole network according to the voting result, and recalculates the credit integral of the nodes; assuming that the current credit integral of the node is Score, the reward and punishment mode is shown in table 1 below:

table 1 credit reward and punishment mode

7) taking credit points, whether voting is positive, time for the nodes to join the network and the number of tokens held by the nodes as clustering characteristic values, normalizing the characteristic values, then utilizing a K-Means clustering algorithm to Cluster all the nodes in the whole network into four classes, wherein the clustering effect is shown in figure 2, the confidence level of Cluster 1 is the highest, the credit points owned by the central nodes after being subjected to normalization are up to 130, the confidence level of Cluster 2 is the lowest, the trust points owned by the central nodes after being subjected to normalization are only 74, and the average credit points of each class of nodes in ten rounds of competition are shown in the following table 2;

TABLE 2 mean Credit score for each class of nodes

	One wheel	Two wheels	Three-wheel	Four-wheel vehicle	Five wheels	Six-wheel	Seven wheels	Eight wheels	Nine wheels	Ten wheels
											Cluster1	101.06	102.06	103.12	104.18	105.24	106.3	107.3	108.36	109.42	110.48
Cluster2	100	99	99	98	97	97	97	96	96	95
											Cluster3	100.98	101.04	102.02	102.02	102.02	103.02	103.08	104.08	105.06	106.04
Cluster4	100.98	101.98	102.96	103.96	103.96	104.96	104.96	105.96	106.94	107.94

8) After clustering is finished, the nodes in the Cluster 1 have higher probability of obtaining the accounting right in the next round of competition. Similarly, the nodes in Cluster 2 have the lowest trust, and the probability that the nodes in the class obtain the accounting right in the next round of competition is greatly reduced.

Claims

1. A block chain DPOS consensus mechanism improvement method based on credit score and node clustering is characterized by comprising the following specific operation steps:

4) voting a newly generated block in the network by the whole network node, wherein the voting is divided into a vote casting and a vote casting, if the vote casting is carried out, the node considers the newly generated block as a valid block, and if the vote casting is carried out, the node considers the newly generated block as an invalid block or even a malicious block;

8) after clustering is finished, the node with the highest credit score, the most positive voting, the longest time for joining the network and the highest token number has the highest trust level, and the node in the next round of competition has higher probability to obtain the accounting right; similarly, the class with the lowest credit score, the most negative vote, the shortest time to join the network and the lowest token number has the lowest confidence level, and the probability that the node in the class will get the right to be billed in the next round of competition is greatly reduced.