CN113723954A - Method for detecting and supervising abnormal transaction nodes in block chain - Google Patents
Method for detecting and supervising abnormal transaction nodes in block chain Download PDFInfo
- Publication number
- CN113723954A CN113723954A CN202110657813.8A CN202110657813A CN113723954A CN 113723954 A CN113723954 A CN 113723954A CN 202110657813 A CN202110657813 A CN 202110657813A CN 113723954 A CN113723954 A CN 113723954A
- Authority
- CN
- China
- Prior art keywords
- transaction
- node
- nodes
- identity
- chain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/382—Payment protocols; Details thereof insuring higher security of transaction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/389—Keeping log of transactions for guaranteeing non-repudiation of a transaction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
- G06Q20/401—Transaction verification
- G06Q20/4014—Identity check for transactions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computer Security & Cryptography (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Technology Law (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention belongs to the technical field of block chains, and particularly relates to a method for detecting and supervising abnormal transaction nodes in a block chain. The method comprises the following steps: designing a novel cross-chain digital identity model, and performing fusion operation on transaction data of a plurality of heterogeneous chains through a unified identity entry, a cross-chain gateway and a distributed storage technology, wherein the operation mode provides a unified channel for generation, inquiry, authentication, supervision and the like of all transactions; constructing a graph of a dynamic structure formed by transaction and address nodes in a block chain by adopting an artificial intelligence technology of a graph neural network, constructing composite characteristics of target address nodes by gathering characteristic information of peripheral address nodes, and detecting and monitoring abnormal behaviors in the whole system from a point-to-point communication and transaction space structure of the block chain; the method can effectively improve the efficiency of detecting the block chain abnormity, and provides technical support for the application of the block chain in wider scenes.
Description
Technical Field
The invention belongs to the technical field of block chains, and particularly relates to a cross-chain transaction data detection and supervision method.
Background
The block chain realizes value transmission in a distributed network, becomes an infrastructure for constructing a novel internet, and realizes numerous applications in the fields of cross-border clearing and settlement, copyright protection, public governance, supply chain finance, logistics tracing and the like. In the field of finance, many countries are advancing to research and develop legal Central Bank Digital Currencies (CBDC), and establish international standards for Digital currencies based on block chains, so as to lay a foundation for Digital transactions of currencies. However, current technology and regulation do not support large-scale application of blockchains in various fields.
The above problems with blockchain decentralized, anonymous, and unsupervised transaction features have made the administration of blockchain transaction data increasingly urgent. However, the anonymity of the blockchain transaction makes it very difficult to supervise the blockchain transaction, most of the existing blockchain transaction supervision is de-anonymization of transaction data and aggregation of the transaction data, and usually only the combination and analysis of single blockchain information on a chain and under the chain are considered, mainly for obtaining the correlation between a target address and an identity; moreover, the types of all platforms of the current block chain are numerous, and the relevance of the transaction among various digital assets is difficult to establish; the block chain-based digital asset transaction forms a graph-like structure between nodes, and the existing analysis method cannot dynamically analyze the structure, the attribute, the change and the like between transaction data. Therefore, the problems of how to detect and analyze abnormal behaviors of users based on cross-chain transaction data of various blockchain platforms, how to establish a cross-chain unified digital identity system and associate transaction data, and how to efficiently model, represent and detect the blockchain transaction data become the key of high-quality application of blockchain.
Disclosure of Invention
The invention aims to provide a method for quickly and efficiently detecting and monitoring abnormal transaction nodes in a block chain.
The invention provides a method for detecting and supervising abnormal transaction nodes in a block chain, which comprises the following steps: firstly, a novel cross-chain digital identity model is designed, a plurality of heterogeneous chains are subjected to the fusion operation of transaction data through a unified identity entry, a cross-chain gateway and a distributed storage technology, and the operation mode provides a unified channel for the generation, inquiry, authentication, supervision and the like of all transactions. Secondly, a dynamic structure formed by transaction and address nodes in the block chain is constructed by adopting an artificial intelligence means of a graph neural network, a target address node is constructed by combining characteristic information of peripheral address nodes, and abnormal behaviors in the whole system are detected and monitored from a point-to-point communication and transaction space structure of the block chain. Meanwhile, the technical scheme provided by the invention is embedded between different block chains in a component form, has lower coupling with the existing functional components of the heterogeneous block chains, and is convenient to deploy and maintain.
The invention provides a method for detecting and supervising abnormal transaction nodes in a block chain, which comprises the following specific steps:
design and construction of cross-chain digital identity model
The traditional central identity model authenticates the identity based on a centralized certificate authority, and all authorization and management of the identity are determined by the centralized authority; the identity management mode makes the data and the state corresponding to the identity difficult to communicate among all platforms and easily causes privacy disclosure; therefore, it is necessary to improve a centralized identity model, integrate dispersed identity data, avoid repeated authentication, and reduce the leakage of personal information; therefore, transaction information generated by the same identity can be fused, and supervision and examination are facilitated; the Decentralized digital Identity refers to the highest management authority of the principal to own the Identity, and some open international standards organizations are setting up relevant protocol standards for Decentralized Identity, such as Decentralized Identity Foundation (DIF) and World Wide Web Consortium (W3C); decentralized mechanisms based on the block chains can manage the scattered identity data and trace identity authentication and use processes; and based on the key and certificate infrastructure going to the center, the security and the usability of the identity data can be greatly improved; when the application of the front block chain is developed rapidly, the ecological healthy development of the block chain is impacted due to the anonymized transaction property, and communication barriers still exist for decentralized identities among heterogeneous block chains; therefore, the heterogeneous blockchain provides a new requirement for the decentralized identity model, namely how to detect and supervise abnormal transaction behaviors in various complex and variable blockchain application ecology; thus, the present invention designs a cross-chain digital identity (CCDI) model for heterogeneous blockchains; the CCDI model mainly comprises an identity data module, a network exchange module and a decentralized storage module:
(1) the identity data module comprises the functions of initialization, authorization, inquiry, authentication, supervision and the like of CCDI; a user obtains a digital certificate through an identity issuer, and a public and private key pair is generated by adopting an asymmetric encryption mode, wherein the public key is used for authenticating the identity of the user to an identity verifier; the identity supervision function can inspect the transaction of the CCDI participating in the heterogeneous chain, so that a supervisor can conveniently and uniformly gather all transaction data of the identity; CCDI is a string of several fields; wherein, the chainID field refers to a unique identifier of a heterogeneous chain, the Method field refers to a predefined function such as initialization, inquiry, authentication and supervision, and the Parameter field refers to parameters required by the function; each CCDI is defined by its particular control subject, which may be an individual, a group organization, or something, a concept; the CCDI character string is analyzed into CCDIDocument, which comprises a CCDI main body, an encryption and decryption protocol, an authorization main body and the like; the controlling principal of CCDI can grant authority to other principal through CCDI or CCDIDocument, for example, the supervisor can delegate part of authority managed by the controlling principal to the lower level employee, so that the CCDI of the lower level employee can temporarily obtain higher authority;
(2) a network switching module including a function of a cross-link gateway; the cross-chain gateway is used for realizing the routing of information and the processing of transaction between heterogeneous chains; the transaction processing mainly comprises initializing the transaction generated by the user and forming the parameters transmitted by the user into blockchain transaction; when the module monitors the transaction information, the module analyzes the transaction information and distributes the transaction information to a corresponding block chain for processing according to the parameters in the CCDI;
(3) a decentralized storage module, which mainly includes heterogeneous block chain storage, and can also be supplemented by various other distributed storages, such as an interplanetary file system, which is commonly used, and writes and deploys application programs and isolates large-scale data by using various network transmission control protocols and distributed hash tables; an etherhouse based distributed storage and content distribution service platform Swarm may also be used.
(II) abnormal transaction data detection based on graph neural network
Due to the decentralized and anonymous characteristics of the block chain, digital assets and other applications based on the block chain are very easy to be attacked illegally, and meanwhile, behaviors of encrypting digital currency for paying illegal commodities, money laundering and the like are increased, so that the method has great significance for carrying out abnormal detection on the block chain application; the anomaly detection in the blockchain aims at transaction nodes in the blockchain network, and a transaction mode generated by the anomaly detection in the blockchain network has obvious anomaly characteristics; the abnormal detection of the block chain transaction data mainly comprises the following steps:
(1) extracting blockchain transaction data: this step first requires the extraction of transaction data in a series of blockchains, the transactions in bitcoin usually include transfer records between two nodes (addresses); if the intelligent contract exists in the blockchain system, the transaction data generally comprises a common transfer record, the creation of the intelligent contract and the calling of the intelligent contract; after data extraction is finished, performing feature processing on blockchain transaction data, and defining feature types of blockchain transactions in table 1;
TABLE 1 blockchain transaction characteristics categories
| Feature name | Meaning of characteristic |
| Receiving a transaction value | Sum of all transactions received by the node |
| Issuing a transaction value | Sum of all transactions issued by a node |
| Receiving transaction amount | Summation of all transaction quantities received by node |
| Issuing a transaction amount | Sum of all transaction amounts issued by node |
| Receiving transaction value by a node | Receiving the amount of the node transaction |
| Node issuing transaction value | Amount of money sent to node transaction |
| Total value of transactions between nodes | Sum of money of transactions occurring between nodes |
| The node receives the transaction amount | Number of received node transactions |
| Node sends out transaction amount | Amount of transaction sent to node |
| Total number of transactions between nodes | Sum of the number of transactions that occur between pairs of nodes |
| Total amount of mined ore | Number of node-generated blocks |
| Total volume of mine excavation transactions | Node generates number of transactions in block |
(2) Establishing a block chain link point and transaction data graph: representing the transaction relation between the block chain nodes by graph data; we represent the relationship between the block link points as a directed acyclic graph, which consists of a subgraph between any two nodes, which can be represented as G ═ (V, E, F), where V represents the set of all nodes in the graph, and the number of nodes is represented as N ═ V |; e is an edge between all nodes, meaning a transaction that exists between any two nodes; f is the characteristic vector of all nodes, the complete characteristic data is shown in table 1 and is input in a vector form;
(3) anomaly detection using a graph neural network: the method adopts a graph attention network model (GAT) in a graph neural network to carry out anomaly detection; the attention mechanism in the deep neural network is inspired by the mechanism of human beings on information in cognitive science; due to limited information processing capability, a human being can selectively focus on a part of complete information while ignoring other information; for example, when a picture is watched, the visual attention focus is usually put on a foreground object with richer semantic information, so that the attention to background information is reduced; for another example, in natural language, we focus more on some important words; therefore, the core of the attention mechanism is to perform weight distribution on given information, and the high weight means that the system needs to perform key processing; nowadays, the attention mechanism has been regarded as a more expressive information fusion means, which has been widely applied in computer vision and natural language processing, and meanwhile, has been greatly developed and applied in scenes such as visual question answering, visual reasoning, language model, machine translation, machine question answering and the like.
In the block chain transaction network constructed by the invention, a certain node is taken as a core, and the relationship between the node and surrounding address nodes and characteristic data of the node per se are considered (see table 1). It can be appreciated that the core node is affected differently by different neighboring nodes due to the effects of the transaction amount, etc. On the other hand, the transaction relationship between nodes can also be represented by a graph. The detection of abnormal nodes can be realized through the characteristics of the nodes and the node relation. To this end, the invention proposes to use a graph attention network GAT to aggregate information characteristics and node relationships of address nodes in a blockchain network. In learning the characteristics of the address node, the GAT assigns different weight values to the node's own characteristics and the neighboring nodes' characteristics. Assume a central node of viIts initial feature vector is h0All the characteristic values of the central node are listed in the first table; its feature vector at the l-th layer is hiAfter a polymerization operation centered on the attention mechanismOutputting the new feature vector h 'of the center node'iThis aggregation operation is called Graph Attention Layer (GAL). Let neighbor node viTo vjThe weight coefficients in between are:
eij＝a(Whi，Whj)， (1)
wherein W is the parameter of the node characteristic change of the layer, and a (-) is a function of the correlation degree of the calculation node. In principle, we can compute any one node in the graph to node viBut to simplify the computation we limit it to first order neighbors, while we also factor in the node itself. In the selection of a (-) above, the correlation degree is calculated by using the inner product of vectors, but actually, the output is only a scalar quantity to represent the correlation degree of the two. Here we have chosen a single fully connected layer:
e(ij)＝Leaky ReLU(aT[Whi||Whj])， (2)
where a is a weight parameter, and the activation function selects LeakyReLU. For better weight assignment, the weights are assigned by soft max (·) function
And (3) carrying out unified normalization processing on all the calculated correlation degrees:
where α is a normalized weight coefficient, and the sum of all neighbor weights is 1. The complete weight coefficient calculation formula is given by the following formula:
after the calculation of the weight coefficients is completed, the node v is weighted and summed according to the attention mechanismiThe new feature vector is:
at this moment, features around the nodes and self features are fused through one-time feature aggregation, and related relations are learned; in addition, the feature vector is also changed from the initial feature into a feature vector fusing the node feature and the node relation.
Repeating the above process K times to obtain K-order feature vector hi finalThe vector fuses K-order neighbor characteristics and node relations of the central node. And finally, classifying the node types by using a full connection layer:
wherein, WfinalParameters of the fully connected layer, bfinalFor biasing, sigmoid () is an activation function used to output probabilities. If the transaction of a certain ratio needs to be detected whether to be abnormal, the edges of the transaction nodes are analyzed:
y＝sigmoid(Wfinal func(hi final，hj final)， (7)
wherein. func (·) correlation functions, combining the characteristics of two nodes, can use concatenation, hadamard product, etc.
And judging whether the generated transaction and the node have abnormity or not according to the final result.
When a new node is added, inductive learning can be performed according to the existing characteristics, and the new node is directly judged.
In the invention, the abnormal detection is carried out by fusing the cross-chain transaction data in the heterogeneous block chain. Cross-chain transactions in a converged heterogeneous blockchain require a uniform cross-chain digital identity entry, thereby forming a consistent transaction execution and detection path.
In the invention, an artificial intelligence algorithm is introduced, a block chain transaction network is combined with a graph neural network, cross-chain transaction data is taken as a basis, a graph attention network is taken as a technical means, and the detection and supervision of transaction data and address nodes are provided.
In the invention, transaction data of a plurality of heterogeneous chains are aggregated, and an entrance of cross-chain identity transaction data is established, so that a supervisor can obtain a query path of a user main body executing transaction on the heterogeneous chains.
According to the invention, the characteristics and spatial correlation between related transactions and nodes can be established by combining with the dynamic change of the blockchain transaction network, and the abnormal information in the blockchain network is identified by using an artificial intelligence method.
Compared with the prior art, the method and the system have the advantages that the identity data among the heterogeneous block chains is efficiently integrated, a data source is provided for the abnormal detection method based on the graph attention network, so that a block chain application builder can detect and supervise abnormal transaction nodes in the application ecology, and the healthy and long-term development of the block chain application is facilitated.
Drawings
Fig. 1 is a block chain anomaly detection framework for a network based on graph attention.
FIG. 2 is a cross-chain digital identity processing architecture.
FIG. 3 shows a block chain abnormal transaction node detection procedure.
Fig. 4 is a CCDI diagram.
Fig. 5 is a block link point weight diagram.
Detailed Description
The invention is further illustrated by the following figures and examples in conjunction with the description. This example is intended to illustrate the invention and is not intended to limit the scope of the invention. Further, it should be understood that various changes or modifications can be made by those skilled in the art after reading the contents of the present invention, and such equivalents should fall within the scope of the appended claims of the present application.
According to the detection method of the block chain abnormal transaction node, a block chain abnormal detection frame based on a graph attention network is designed according to the method, and the frame refers to a graph 1, uses cross-chain digital identity CCDI among heterogeneous block chains, fuses and gathers cross-chain transaction data, and then extracts the cross-chain transaction data into transaction and contract data respectively. And then preprocessing and analyzing the extracted data, and inputting the data into a power network for attention and carrying out anomaly detection. Meanwhile, a cross-chain digital identity processing architecture in an anomaly detection framework is designed, and the cross-chain digital identity processing architecture is shown in fig. 2, wherein the cross-chain digital identity processing architecture comprises an identity data module, a network exchange module and a decentralized storage module. The identity data module comprises the functions of initialization, authorization, inquiry, authentication, supervision and the like of the CCDI. The network switching module contains the functionality of a cross-link gateway. The decentralized memory module mainly comprises a heterogeneous block chain memory function. According to this anomaly detection framework, its execution steps are as follows, see fig. 3, fully described below:
(1) a transaction supervisor initiates a supervision instruction through CCDI through a supervision interface provided by an identity verification mechanism, wherein a field contained in the CCDI comprises a ChainID field which refers to a unique identifier of a heterogeneous chain, a Method field which refers to a predefined function such as initialization, inquiry, authentication and supervision, and a Parameter field which refers to a Parameter required by the function;
(2) after receiving the CCDI of the monitoring party, the identity verifying party analyzes the CCDI into CCDIDocument, obtains an identity object which the identity verifying party intends to monitor, extracts a parameter field in the CCDI through an identity data module in the figure 2, and inquires the access authority of the identity on a heterogeneous chain through an identity information obtaining interface;
(3) the identity data module transmits the supervision instruction to a decentralized storage block chain, inquires the transaction data corresponding to the identity and returns the transaction data to the supervisor according to the inquiry;
(4) after the monitoring party obtains the transaction data of the cross-chain digital identity, the transaction data is input into the attention network;
(5) the graph attention network constructs a graph of the transaction data, constructs a feature vector of the graph, and calculates a weight coefficient according to features such as transaction amount and transaction times among nodes as shown in fig. 5; after the calculation of the weight coefficient is completed, obtaining a new feature vector of the node according to the idea of attention mechanism weighted summation, wherein the feature vector of the node is fused with the relationship and the features of the connection node, and then classifying and inducing the node and the transaction;
(6) after repeated iteration for many times, the feature vector of the node already contains the features and node relations of the multi-order neighbor nodes, then the node and the transaction are classified and summarized, and finally the graph attention network returns the abnormal transaction node to the supervisor.
Through the implementation mode and the steps, after initiating transaction detection on a certain cross-link digital identity, a transaction supervisor integrates transaction data and inputs the transaction data into a graph attention network through an identity and transaction processing module, and after deep learning, abnormal transaction nodes are identified and detected.
Claims (2)
1. A method for detecting and supervising abnormal transaction nodes in a blockchain, comprising:
firstly, designing a novel cross-chain digital identity model, and performing fusion operation of transaction data on a plurality of heterogeneous chains through a unified identity entry, a cross-chain gateway and a distributed storage technology, wherein the operation mode provides a unified channel for generation, inquiry, authentication, supervision and the like of all transactions;
secondly, constructing a graph of a dynamic structure formed by transaction and address nodes in a block chain by adopting an artificial intelligence technology of a graph neural network, constructing composite characteristics of target address nodes by gathering characteristic information of peripheral address nodes, and detecting and monitoring abnormal behaviors in the whole system from a point-to-point communication and transaction space structure of the block chain;
meanwhile, the content is embedded between different blockchains in a component form, and the coupling with the existing functional components of the heterogeneous blockchains is low, so that the deployment and the maintenance are convenient.
2. The method for detecting and supervising abnormal transaction nodes in a blockchain according to claim 1, comprising the following specific steps:
design and construction of cross-chain digital identity model
Designing a cross-chain digital identity model aiming at the heterogeneous block chain, marking as a CCDI model, and comprising an identity authentication module, a network switching module and a decentralized storage module; wherein:
(1) the identity verification module has the functions of initialization, authorization, inquiry, authentication and supervision of CCDI;
a user obtains a digital certificate through an identity issuer, and a public and private key pair is generated by adopting an asymmetric encryption mode, wherein the public key is used for authenticating the identity of the user to an identity verifier;
the identity supervision function is used for auditing the transaction of the CCDI participating in the heterogeneous chain, so that a supervisor can conveniently and uniformly gather all transaction data of the identity;
CCDI is a string of several fields; the device comprises a ChainID field, a Method field, a Parameter field and a function, wherein the ChainID field represents a unique identifier of a heterogeneous chain, the Method field represents a predefined function such as initialization, inquiry, authentication and supervision, and the Parameter field represents parameters required by the function; each CCDI is defined by its particular controlling entity, which is an individual, group organization, or something, concept; the CCDI character string is analyzed into CCDIDocument, wherein the CCDI character string comprises a main body of CCDI, an encryption and decryption protocol and an authorization main body; the control subject of CCDI grants the right to other subjects through CCDI or CCDIDocument;
(2) the network switching module has the function of a cross-link gateway; the cross-chain gateway is used for realizing the routing of information and the processing of transaction between heterogeneous chains; the transaction processing mainly comprises initializing the transaction generated by the user and forming the parameters transmitted by the user into blockchain transaction; analyzing after monitoring the transaction information, and distributing the transaction information to a corresponding block chain for processing according to parameters in the CCDI;
(3) the decentralized storage module mainly comprises heterogeneous block chain storage, and is supplemented by other various distributed storage, or uses distributed storage based on Etheng and a content distribution service platform Swarm;
(II) abnormal transaction data detection based on graph neural network
The method comprises the following steps:
(1) extracting blockchain transaction data: firstly, extracting transaction data in a series of block chains; transactions in bitcoin typically include a record of transfers between two nodes; if the intelligent contract exists in the blockchain system, the transaction data comprises a common transfer record, the creation of the intelligent contract and the calling of the intelligent contract; after the data is extracted, the characteristics of the blockchain transaction data are processed, and the characteristic types of the blockchain transaction are defined in table 1:
TABLE 1 blockchain transaction characteristics categories
(2) Establishing a block chain link point and transaction data graph: representing the relationship between the block link points as a directed acyclic graph; the method comprises the following steps that a subgraph between any two nodes is formed and is represented as G ═ (V, E, F), wherein y represents the set of all nodes in the graph, and the number of the nodes is represented as N ═ V |; e is an edge between all nodes, meaning a transaction that exists between any two nodes; f is the feature vector of all nodes;
(3) anomaly detection using a graph neural network: performing anomaly detection by using a graph attention network (GAT) in a graph neural network; specifically, the information characteristics of address nodes in the network aggregation block chain network are noted by using a graph; when learning the characteristics of the address node, the GAT distributes different weight values to the node self characteristics and the neighbor nodes through the characteristics of the nodes; assume a central node of viWhich is at the firstThe feature vector corresponding to the layer l is hiOutputting a new feature vector h 'of the center node after the aggregation operation taking attention mechanism as a core'iThis aggregation operation is called the Graph Attention Layer (GAL);
let neighbor node υiTo upsilonjThe weight coefficients in between are:
eij＝a(Whi，Whj)， (1)
wherein, W is the parameter of the node characteristic change of the layer, and a (-) is a function of the correlation degree of the calculation node; selecting a single fully connected layer for a (-):
e(ij)＝Leaky ReLU(aT[Whi||Whj])； (2)
wherein, a is a weight parameter, and the activation function selects LeakyReLU; all calculated correlations are subjected to a uniform normalization process by a softmax (·) function:
alpha is a normalized weight coefficient, and the sum of all neighbor weights is 1; the complete weight coefficient calculation formula is as follows:
after the calculation of the weight coefficient is completed, according to the idea of weighted summation of an attention mechanism, a node upsiloniThe new feature vector is:
at this moment, features around the nodes and self features are fused through one-time feature aggregation, and related relations are learned; in addition, the feature vector is changed from the initial feature into a feature vector fusing the node feature and the node relationship;
repeating the above process K times to obtain K-order feature vector hi finalThe vector integrates K-order neighbor characteristics of the central node and node relation; and finally, classifying the node types by using a full connection layer:
wherein, WfinalParameters of the fully connected layer, bfinalFor biasing, sigmoid () is an activation function for outputting probability; when whether a certain transaction is abnormal or not needs to be detected, the edges of the transaction nodes are analyzed:
y＝sigmoid(Wfinal func(hi final，hj final)+bfinal)， (7)
wherein, the func (-) correlation function combines the characteristics of the two nodes;
and judging whether the generated transaction and the node have abnormity or not according to the final result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110657813.8A CN113723954A (en) | 2021-06-15 | 2021-06-15 | Method for detecting and supervising abnormal transaction nodes in block chain |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110657813.8A CN113723954A (en) | 2021-06-15 | 2021-06-15 | Method for detecting and supervising abnormal transaction nodes in block chain |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113723954A true CN113723954A (en) | 2021-11-30 |
Family
ID=78672886
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110657813.8A Pending CN113723954A (en) | 2021-06-15 | 2021-06-15 | Method for detecting and supervising abnormal transaction nodes in block chain |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113723954A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114048362A (en) * | 2022-01-11 | 2022-02-15 | 国网电子商务有限公司 | Block chain-based power data anomaly detection method, device and system |
| CN114612235A (en) * | 2022-03-09 | 2022-06-10 | 烟台大学 | Block chain abnormal behavior detection method based on graph embedding |
| CN115065458A (en) * | 2022-08-08 | 2022-09-16 | 北京邮电大学 | Electronic commerce transaction system with data encryption transmission |
| CN116955306A (en) * | 2023-06-21 | 2023-10-27 | 东莞市铁石文档科技有限公司 | File management system based on distributed storage |
-
2021
- 2021-06-15 CN CN202110657813.8A patent/CN113723954A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114048362A (en) * | 2022-01-11 | 2022-02-15 | 国网电子商务有限公司 | Block chain-based power data anomaly detection method, device and system |
| CN114612235A (en) * | 2022-03-09 | 2022-06-10 | 烟台大学 | Block chain abnormal behavior detection method based on graph embedding |
| US11803855B2 (en) | 2022-03-09 | 2023-10-31 | Yantai University | Method for detecting block chain abnormal behavior based on graph embedding |
| CN115065458A (en) * | 2022-08-08 | 2022-09-16 | 北京邮电大学 | Electronic commerce transaction system with data encryption transmission |
| CN115065458B (en) * | 2022-08-08 | 2022-11-15 | 山东鼎信数字科技有限公司 | Electronic commerce transaction system with data encryption transmission |
| CN116955306A (en) * | 2023-06-21 | 2023-10-27 | 东莞市铁石文档科技有限公司 | File management system based on distributed storage |
| CN116955306B (en) * | 2023-06-21 | 2024-04-12 | 东莞市铁石文档科技有限公司 | File management system based on distributed storage |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113723954A (en) | Method for detecting and supervising abnormal transaction nodes in block chain | |
| Deebak et al. | Privacy-preserving in smart contracts using blockchain and artificial intelligence for cyber risk measurements | |
| Chen et al. | A survey on blockchain systems: Attacks, defenses, and privacy preservation | |
| Abusorrah et al. | Minimax-regret robust defensive strategy against false data injection attacks | |
| Li et al. | Retracted: Design of multimedia blockchain privacy protection system based on distributed trusted communication | |
| CN115065458B (en) | Electronic commerce transaction system with data encryption transmission | |
| Lin et al. | Incentive edge-based federated learning for false data injection attack detection on power grid state estimation: A novel mechanism design approach | |
| Xueshuo et al. | AWAP: Adaptive weighted attribute propagation enhanced community detection model for bitcoin de-anonymization | |
| Tang et al. | Secure and trusted collaborative learning based on blockchain for artificial intelligence of things | |
| Srivastava et al. | An effective computational technique for taxonomic position of security vulnerability in software development | |
| Anwar et al. | A comprehensive insight into game theory in relevance to cyber security | |
| Lata et al. | A comprehensive survey of fraud detection techniques | |
| Cong et al. | Blockchain applications, challenges and evaluation: a survey | |
| Hindistan et al. | A hybrid approach with gan and dp for privacy preservation of iiot data | |
| Sanda et al. | Long-Range attack detection on permissionless blockchains using Deep Learning | |
| Bokhari et al. | Cybersecurity strategy under uncertainties for an IoE environment | |
| Acevedo-Viloria et al. | Relational graph neural networks for fraud detection in a super-app environment | |
| Lv et al. | Research on modeling of E-banking fraud account identification based on federated learning | |
| Jin et al. | Detecting arbitrage on ethereum through feature fusion and positive-unlabeled learning | |
| Khalil et al. | Fuzzy Logic based security trust evaluation for IoT environments | |
| Mu et al. | A study of the internet financial interest rate risk evaluation index system in cloud computing | |
| Wu et al. | Federated learning for tabular data: Exploring potential risk to privacy | |
| Chivers et al. | Risk profiles and distributed risk assessment | |
| Al-Shabi et al. | Improving blockchain security for the internet of things: challenges and solutions | |
| Hou et al. | Research on cyberspace multi-objective security algorithm and decision mechanism of energy internet |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |