Data security encryption processing system based on block chain

Abstract

The invention discloses a data security encryption processing system based on a block chain, and aims to improve data privacy protection and calculation efficiency. The system comprises a data acquisition and preprocessing module used for acquiring and preprocessing graph data and multi-modal data; the graph data homomorphic encryption module is used for encrypting graph data by using a post-quantum homomorphic encryption algorithm and supporting graph neural network calculation; the multi-modal data encryption module is used for encrypting texts and sensor data by adopting a hybrid encryption method; the graph neural network calculation module is used for updating node features of the encrypted data; the encrypted data fusion module is used for fusing the encrypted calculation result and the multi-modal data; and the block chain storage and verification module stores the encrypted data in a block chain to ensure data integrity. The system can process multi-source encrypted data efficiently and safely, and data privacy and integrity are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of blockchain technology, data encryption technology and graph neural network, and particularly relates to a data security encryption processing system based on blockchain. BACKGROUND

[0002] At present, with the rapid development of big data, blockchain and graph neural network technologies, data security encryption processing has become an important research direction in the field of information technology. As a decentralized and tamper-proof distributed ledger technology, blockchain technology is widely used in finance, supply chain, Internet of Things and other fields. However, the data stored and transmitted in the blockchain often involves privacy and sensitive information, therefore, how to ensure the effective calculation and sharing of data while ensuring data security has become a core problem in current technology.

[0003] In terms of data encryption technology, homomorphic encryption is considered an important method to solve the problem of data privacy. It allows direct processing and calculation of data in an encrypted state, avoiding the privacy leakage problem caused by the need to decrypt data for calculation in traditional encryption methods. In particular, post-quantum homomorphic encryption algorithms, as encryption schemes against quantum computing threats, have been proposed and gradually applied to protect sensitive data. However, post-quantum homomorphic encryption algorithms still face challenges in computational efficiency and performance optimization in practical applications, especially in complex data structure and large-scale data processing scenarios, and have not yet achieved an ideal balance.

[0004] Graph neural network technology has gained widespread attention in recent years, as it can effectively process and analyze graph structured data, learning the representation and features of graphs through the association between nodes. However, when the graph data involves sensitive information, how to encrypt during the training and inference process of the graph neural network to ensure the privacy and security of the calculation results remains a technical problem that has not been solved.

[0005] In addition, with the widespread application of multi-modal data, how to protect the privacy of these data in an encrypted state and effectively perform fusion processing has become a problem that needs to be solved. Currently, using hybrid encryption methods to encrypt multi-modal data is a common solution, but how to ensure effective fusion and calculation between data in an encrypted state still presents a great challenge.

[0006] Therefore, how to provide a data security encryption processing system based on blockchain is a problem that needs to be solved by those skilled in the art. SUMMARY

[0007] One purpose of the present application is to propose a blockchain-based data security encryption processing system. In view of the obvious defects of the prior art in the security and efficiency of data encryption processing combined with blockchain, a solution capable of graph neural network calculation and multi-modal data processing in an encrypted state is proposed. The present application uses a post-quantum homomorphic encryption algorithm to encrypt graph data, and combines a hybrid encryption method to perform privacy-enhanced encryption on multi-modal data, ensuring that data can still be effectively calculated and analyzed in an encrypted state. Through the blockchain storage mechanism, the data is ensured to be tamper-proof, and the integrity and effectiveness of the encrypted data are verified through the smart contract, improving the data privacy protection while ensuring the calculation efficiency. The present application has the technical advantages of high efficiency, strong privacy protection and strong adaptability, and has significant application potential in the field of sensitive data encryption calculation and large-scale data processing.

[0008] A blockchain-based data security encryption processing system according to an embodiment of the present application comprises the following modules:

[0009] A data acquisition and preprocessing module acquires data from multiple data sources, including graph data and multi-modal data, and preprocesses the acquired data to obtain a standardized input data set.

[0010] A graph data homomorphic encryption module performs homomorphic encryption processing on graph data based on the standardized input data set, encrypts each node in the graph using a post-quantum homomorphic encryption algorithm, ensures that the data can be calculated by a graph neural network in an encrypted state, and obtains encrypted graph data.

[0011] A multi-modal data privacy-enhanced encryption module performs privacy-enhanced encryption processing on multi-modal data based on the standardized input data set, including text and sensor data, and uses a hybrid encryption method to encrypt each type of modal data to obtain encrypted multi-modal data.

[0012] A graph neural network calculation module inputs the encrypted graph data and encrypted multi-modal data into a graph neural network, and the homomorphically encrypted graph data is calculated by the graph neural network model in an encrypted state to update the node features and obtain encrypted calculation results.

[0013] An encrypted data fusion module fuses the encrypted calculation results and the encrypted multi-modal data to obtain comprehensive encrypted data output results.

[0014] A blockchain storage and verification module stores the comprehensive encrypted data output results in a blockchain, ensures the data tamper-proof through the decentralized storage mechanism of the blockchain, and verifies the integrity and effectiveness of the encrypted data through a smart contract.

[0015] Optionally, the data acquisition and preprocessing module specifically includes:

[0016] S1.1: Collect raw data from multiple data sources to obtain graph data and multimodal data;

[0017] S1.2: The collected raw graph data is formatted to convert the graph data into a standard graph structure, which includes node data and edge data, and is normalized according to the properties of the graph;

[0018] S1.3: Preprocess the collected raw multimodal data to obtain a standardized multimodal dataset;

[0019] S1.4: The preprocessed graph data and multimodal data are uniformly formatted to obtain a standardized dataset.

[0020] Optionally, the graph data homomorphic encryption module specifically includes:

[0021] S2.1: Based on a standardized input dataset, extract node features from the graph structure. The node features include node attribute data. The graph structure includes nodes and edges, where edges represent the connection relationships between nodes.

[0022] S2.2: Select a post-quantum homomorphic encryption algorithm, wherein the post-quantum homomorphic encryption algorithm adopts a lattice-based encryption mechanism, and the encryption process uses an encryption function to encrypt the feature data of each node to obtain encrypted node feature data, wherein the encryption function is a mapping of homomorphic encryption operations;

[0023] S2.3: Apply addition and multiplication operations from the homomorphic encryption algorithm to each encrypted node feature data. The addition operation is used for the aggregation of node features, and the multiplication operation is used for the calculation of the interaction relationship between features to obtain the encrypted processing result.

[0024] S2.4: Homomorphic encryption is performed on the edge relationships according to the structure of the graph. The edge relationships represent the connection information between nodes in the graph. The same post-quantum homomorphic encryption algorithm is used to encrypt the connection information of each edge to obtain encrypted edge data. The encrypted edge data and the encrypted node features together constitute the encrypted graph data.

[0025] S2.5: Generate encrypted graph data, which includes encrypted node features and encrypted edge data.

[0026] Optionally, the multimodal data privacy enhancement encryption module specifically includes:

[0027] S3.1: Based on a standardized input dataset, extract multimodal data from multiple data sources, including text data and sensor data, wherein the text data includes natural language text data and the sensor data includes data streams from multiple sensors;

[0028] S3.2: Encrypt the text data using a symmetric encryption algorithm, wherein the symmetric encryption algorithm is an Advanced Encryption Standard (AES). The text data is encrypted using a symmetric key, which is generated using an encryption key derivation function.

[0029] S3.3: Encrypt the sensor data using a hash algorithm. The hash function used is SHA-256. The sensor data is encrypted using the hash function.

[0030] S3.4: Generate encrypted multimodal data, wherein the encrypted text data includes encrypted text content and an encrypted symmetric key, and the encrypted sensor data is a hash encryption result.

[0031] Optionally, the graph neural network computing module specifically includes:

[0032] S4.1: Input the encrypted graph data and encrypted multimodal data into the graph neural network model. The encrypted graph data includes encrypted node features and encrypted edge data. The encrypted multimodal data includes encrypted text data and encrypted sensor data.

[0033] S4.2: In the graph neural network model, the encrypted node feature data is encrypted using a graph convolutional layer. The graph convolutional layer uses a homomorphic encryption algorithm to encrypt and aggregate the features of neighboring nodes to obtain the encrypted calculation result of each node.

[0034] S4.3: Apply the activation function in the graph neural network to the encrypted node feature data for calculation. The activation function is used to perform a nonlinear transformation on the encrypted node features to generate updated encrypted node features.

[0035] S4.4: Output the updated encrypted node features and the encrypted edge data together. The encrypted calculation result includes the updated encrypted node features and the encrypted edge data.

[0036] Optionally, the encrypted data fusion module specifically includes:

[0037] S5.1: Match the encrypted calculation results with the encrypted multimodal data. The encrypted calculation results include encrypted node features and encrypted edge data obtained from the graph neural network model. The encrypted multimodal data includes encrypted text data and encrypted sensor data.

[0038] S5.2: Perform data alignment processing on the encrypted calculation results and the encrypted multimodal data. The data alignment processing includes synchronizing based on node features and the timestamps of the multimodal data to obtain matched encrypted data.

[0039] S5.3: Perform an encryption fusion operation on the matched encrypted data. The encryption fusion operation includes encrypting and merging the encrypted node features and encrypted multimodal data through a homomorphic encryption mechanism. The merged data remains encrypted, resulting in encrypted fused data.

[0040] S5.4: The encrypted fused data is aggregated with encrypted information. The aggregation process includes adding the encrypted data using homomorphic encryption, weighting and summing the different features of the encrypted data, and obtaining the encrypted fusion result of each node.

[0041] S5.5: Based on the encryption fusion result, a comprehensive encrypted data output result is obtained.

[0042] Optionally, the blockchain storage and verification module specifically includes:

[0043] S6.1: The comprehensive encrypted data output result is formatted using an encryption protocol. The encrypted data output result includes encrypted graph data and encrypted multimodal data, resulting in formatted encrypted data.

[0044] S6.2: Store the formatted encrypted data output results in the blockchain, which is a decentralized distributed ledger, through which data is recorded to multiple distributed nodes;

[0045] S6.3: Generate a new block in the blockchain, the block including a hash value storing encrypted data and related timestamp information, the hash value being generated by an encryption algorithm and used to uniquely identify the encrypted data;

[0046] S6.4: Verify the stored encrypted data through smart contracts in the blockchain network. The smart contracts check the integrity and validity of the encrypted data through predefined verification rules.

[0047] The beneficial effects of this invention are:

[0048] This invention combines graph neural network homomorphic encryption and privacy-enhancing multimodal encryption technologies to achieve secure computation and analysis of graph and multimodal data in an encrypted state, significantly improving data privacy protection and computational efficiency. It employs a post-quantum homomorphic encryption algorithm to encrypt graph data and uses a hybrid encryption method to protect the privacy of multimodal data, ensuring that sensitive information is not leaked during storage, transmission, and computation. By utilizing the decentralized storage mechanism of blockchain and combining it with smart contracts to verify the integrity and validity of encrypted data, it effectively solves the security and efficiency problems of blockchain data encryption processing in existing technologies. Compared with traditional technologies, this invention improves computational efficiency and reliability while ensuring data privacy, demonstrating superior adaptability and effectiveness in large-scale sensitive data processing and distributed storage environments. Attached Figure Description

[0049] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0050] Figure 1 This is a flowchart of a blockchain-based data security encryption processing system proposed in this invention;

[0051] Figure 2 This is a flowchart of the data privacy protection and encryption process proposed in this invention, which integrates post-quantum homomorphic encryption algorithm and hybrid encryption method. Detailed Implementation

[0052] Combination Figures 1-2 The present invention will be described in further detail below. These accompanying drawings are simplified schematic diagrams, illustrating only the basic structure of the invention and showing the main components relevant to the invention. Figure 1 and Figure 2 The present invention provides a blockchain-based data security encryption processing system, which includes the following modules:

[0053] Data acquisition and preprocessing module: Acquires data from multiple data sources, including graph data and multimodal data, and preprocesses the acquired data to obtain a standardized input dataset;

[0054] Graph data homomorphic encryption module: Based on a standardized input dataset, it performs homomorphic encryption on graph data. It uses a post-quantum homomorphic encryption algorithm to encrypt each node in the graph, ensuring that the data can be used for graph neural network computation in the encrypted state to obtain encrypted graph data.

[0055] Multimodal data privacy enhancement encryption module: Based on a standardized input dataset, the module performs privacy enhancement encryption processing on multimodal data, which includes text and sensor data. A hybrid encryption method is used to encrypt each type of data to obtain encrypted multimodal data.

[0056] Graph Neural Network Computation Module: Input the encrypted graph data and encrypted multimodal data into the graph neural network. The homomorphically encrypted graph data is computed through the graph neural network model in the encrypted state to update node features and obtain the encrypted computation result.

[0057] Encrypted data fusion module: It fuses the encrypted calculation results with the encrypted multimodal data to obtain a comprehensive encrypted data output result;

[0058] Blockchain storage and verification module: Stores the comprehensive encrypted data output results in the blockchain, ensures the immutability of the data through the decentralized storage mechanism of the blockchain, and verifies the integrity and validity of the encrypted data through smart contracts.

[0059] In this embodiment, the data acquisition and preprocessing module specifically includes:

[0060] S1.1: Collect raw data from multiple data sources to obtain graph data and multimodal data;

[0061] S1.2: The collected raw graph data is formatted to convert the graph data into a standard graph structure, which includes node data and edge data, and is normalized according to the properties of the graph;

[0062] S1.3: Preprocess the collected raw multimodal data to obtain a standardized multimodal dataset;

[0063] S1.4: The preprocessed graph data and multimodal data are uniformly formatted to obtain a standardized dataset.

[0064] In this embodiment, the graph data homomorphic encryption module specifically includes:

[0065] S2.1: Based on a standardized input dataset, extract node features from the graph structure. The node features include node attribute data. The graph structure includes nodes and edges, where edges represent the connection relationships between nodes.

[0066] S2.2: Select a post-quantum homomorphic encryption algorithm, wherein the post-quantum homomorphic encryption algorithm adopts a lattice-based encryption mechanism, and the encryption process uses an encryption function to encrypt the feature data of each node to obtain encrypted node feature data, wherein the encryption function is a mapping of homomorphic encryption operations;

[0067] Specifically, the encryption function utilizes mathematical structures from lattice theory to map node feature data into an encrypted form. This encryption process ensures that the data can undergo homomorphic operations in the encrypted state; that is, when addition and multiplication operations are performed on the encrypted data, the result remains encrypted and consistent with the plaintext operation result. Node feature data can be used for graph neural network computations without decryption, while simultaneously guaranteeing data privacy and security.

[0068] S2.3: Apply addition and multiplication operations from the homomorphic encryption algorithm to each encrypted node feature data. The addition operation is used for the aggregation of node features, and the multiplication operation is used for the calculation of the interaction relationship between features to obtain the encrypted processing result.

[0069] Specifically, when using homomorphic encryption algorithms for addition, the addition operation between node features is used to aggregate the encrypted feature data of adjacent nodes, thereby updating the feature value of each node. This operation ensures that node features can be efficiently aggregated and calculated in the graph neural network under encrypted conditions.

[0070] Multiplication operations are used to calculate the interaction relationships between node features, such as calculating the similarity or weighted relationships between nodes. Through homomorphic encryption, multiplication operations enable the calculation of interactions and relationships between features within encrypted data, yielding encrypted results. These encrypted calculations are used to update node features while maintaining data privacy and security, ensuring that graph neural network computations are performed on encrypted data.

[0071] S2.4: Homomorphic encryption is performed on the edge relationships according to the structure of the graph. The edge relationships represent the connection information between nodes in the graph. The same post-quantum homomorphic encryption algorithm is used to encrypt the connection information of each edge to obtain encrypted edge data. The encrypted edge data and the encrypted node features together constitute the encrypted graph data.

[0072] S2.5: Generate encrypted graph data, which includes encrypted node features and encrypted edge data.

[0073] In this embodiment, the multimodal data privacy enhancement encryption module specifically includes:

[0074] S3.1: Based on a standardized input dataset, extract multimodal data from multiple data sources, including text data and sensor data, wherein the text data includes natural language text data and the sensor data includes data streams from multiple sensors;

[0075] S3.2: Encrypt the text data using a symmetric encryption algorithm, wherein the symmetric encryption algorithm is an Advanced Encryption Standard (AES). The text data is encrypted using a symmetric key, which is generated using an encryption key derivation function.

[0076] Specifically, to ensure the security of the encryption process, a key derivation function is used to generate the symmetric key. This function converts the original key material into an encryption key through a pre-defined key generation process. The generated symmetric key is used for encryption and decryption operations in the AES encryption algorithm, thereby ensuring the confidentiality of the text data.

[0077] Encrypted text data is stored and transmitted in an encrypted state. Only authorized users or systems can perform decryption operations with the cooperation of the key management system, thereby effectively protecting the privacy and security of the text data.

[0078] S3.3: Encrypt the sensor data using a hash algorithm. The hash function used is SHA-256. The sensor data is encrypted using the hash function.

[0079] S3.4: Generate encrypted multimodal data, wherein the encrypted text data includes encrypted text content and an encrypted symmetric key, and the encrypted sensor data is a hash encryption result.

[0080] In this embodiment, the graph neural network computing module specifically includes:

[0081] S4.1: Input the encrypted graph data and encrypted multimodal data into the graph neural network model. The encrypted graph data includes encrypted node features and encrypted edge data. The encrypted multimodal data includes encrypted text data and encrypted sensor data.

[0082] S4.2: In the graph neural network model, the encrypted node feature data is encrypted using a graph convolutional layer. The graph convolutional layer uses a homomorphic encryption algorithm to encrypt and aggregate the features of neighboring nodes to obtain the encrypted calculation result of each node.

[0083] Specifically, the graph convolutional layer is responsible for handling the aggregation and transmission of information between nodes and their neighbors. In encrypted mode, the graph convolutional operation uses a homomorphic encryption algorithm to encrypt and aggregate the features of neighboring nodes.

[0084] The encrypted aggregation process is as follows: For each node, the graph convolutional layer collects the encrypted features of its neighboring nodes and aggregates them using the addition operation in a homomorphic encryption algorithm. Because a homomorphic encryption algorithm is used, when node features are added in an encrypted state, the calculation result remains encrypted, without the need for decryption. This ensures the protection of data privacy.

[0085] During the aggregation process, the encrypted features of each node are weighted and summed together with the encrypted features of its neighboring nodes to obtain an encrypted feature aggregation result. Through homomorphic encryption algorithms, all computations are performed in an encrypted state, ensuring the privacy and security of the encrypted data.

[0086] The encrypted aggregation results will be used as updated node features for subsequent graph neural network calculations. The entire process of feature aggregation and updating can be completed without decryption, fully ensuring efficient computation and privacy protection of data under encrypted conditions.

[0087] S4.3: Apply the activation function in the graph neural network to the encrypted node feature data for calculation. The activation function is used to perform a nonlinear transformation on the encrypted node features to generate updated encrypted node features.

[0088] S4.4: Output the updated encrypted node features and the encrypted edge data together. The encrypted calculation result includes the updated encrypted node features and the encrypted edge data.

[0089] In this embodiment, the encrypted data fusion module specifically includes:

[0090] S5.1: Match the encrypted calculation results with the encrypted multimodal data. The encrypted calculation results include encrypted node features and encrypted edge data obtained from the graph neural network model. The encrypted multimodal data includes encrypted text data and encrypted sensor data.

[0091] S5.2: Perform data alignment processing on the encrypted calculation results and the encrypted multimodal data. The data alignment processing includes synchronizing based on node features and the timestamps of the multimodal data to obtain matched encrypted data.

[0092] S5.3: Perform an encryption fusion operation on the matched encrypted data. The encryption fusion operation includes encrypting and merging the encrypted node features and encrypted multimodal data through a homomorphic encryption mechanism. The merged data remains encrypted, resulting in encrypted fused data.

[0093] S5.4: The encrypted fused data is aggregated with encrypted information. The aggregation process includes adding the encrypted data using homomorphic encryption, weighting and summing the different features of the encrypted data, and obtaining the encrypted fusion result of each node.

[0094] S5.5: Based on the encryption fusion result, a comprehensive encrypted data output result is obtained.

[0095] In this embodiment, the blockchain storage and verification module specifically includes:

[0096] S6.1: The comprehensive encrypted data output result is formatted using an encryption protocol. The encrypted data output result includes encrypted graph data and encrypted multimodal data, resulting in formatted encrypted data.

[0097] S6.2: Store the formatted encrypted data output results in the blockchain, which is a decentralized distributed ledger, through which data is recorded to multiple distributed nodes;

[0098] S6.3: Generate a new block in the blockchain, the block including a hash value storing encrypted data and related timestamp information, the hash value being generated by an encryption algorithm and used to uniquely identify the encrypted data;

[0099] S6.4: Verify the stored encrypted data through smart contracts in the blockchain network. The smart contracts check the integrity and validity of the encrypted data through predefined verification rules.

[0100] Example 1: To verify the feasibility of this invention in practice, a detailed description is provided using a medical data storage and privacy protection scenario in an intelligent medical system as an example. Intelligent medical systems involve a large amount of patient data, medical records, image data, and sensor data. These data must be stored, shared, and analyzed with a high degree of privacy protection and data security. However, existing technologies, while ensuring data privacy, often face problems such as low computational efficiency, insufficient data security, and difficulties in data fusion and analysis. Especially when using blockchain technology for data storage, how to efficiently and securely process and analyze encrypted data on the blockchain remains a pressing technical challenge.

[0101] In this embodiment, to address these issues, a series of experiments will be conducted to verify the blockchain data encryption method proposed in this invention, which combines graph neural network homomorphic encryption with privacy-enhancing multimodal encryption. The innovation of this invention lies in encrypting graph data using a post-quantum homomorphic encryption algorithm and combining it with a hybrid encryption method to enhance the privacy of multimodal data, ensuring data privacy protection while enabling efficient computation and analysis. Through this method, this invention successfully resolves the contradiction between privacy protection and computational efficiency in traditional blockchain technology, improving the security and efficiency of blockchain data storage and encrypted computation.

[0102] In this embodiment, it is assumed that the data being processed is patient diagnosis and treatment data from an intelligent medical system, which includes graph data and multimodal data. The specific implementation steps are as follows:

[0103] Patient data was collected from multiple sources within the hospital, including medical records, medical images, and sensor data collected from wearable devices. Through data cleaning and standardization, the raw data was transformed into a standardized input dataset that met the requirements of subsequent processing. The standardization process included denoising the text data, normalizing the sensor data, and formatting the nodes and edges in the graph data, ensuring the dataset met the input requirements of graph neural network models.

[0104] Homomorphic encryption is applied to the standardized graph data. A post-quantum homomorphic encryption algorithm is used to encrypt each node in the graph data, ensuring that node features can be computed even in the encrypted state. Through addition and multiplication operations of the homomorphic encryption algorithm, the encrypted data can be aggregated and interactively computed to obtain encrypted node features and graph connectivity information. This process ensures that data is not leaked during processing, while allowing the encrypted data to still participate in the computation of the graph neural network.

[0105] The collected multimodal data undergoes privacy-enhancing encryption. A hybrid encryption method is employed: text data uses Advanced Encryption Standard (AES) for symmetric encryption with a 256-bit key; sensor data is encrypted using the SHA-256 hash algorithm. This encryption approach effectively ensures data privacy while preserving the integrity and computational power of the encrypted data.

[0106] Encrypted graph data and multimodal data are input into a graph neural network model. The encrypted graph data is processed by the graph convolutional layers and activation functions of the graph neural network to calculate updated encrypted node features. Throughout the process, a homomorphic encryption mechanism ensures that the calculation results are performed in an encrypted state without exposing any sensitive information. The features of each node are dynamically updated to reflect the patient's latest health status.

[0107] The encrypted computation results are fused with the encrypted multimodal data to obtain a comprehensive encrypted data output. During the fusion process, a homomorphic encryption mechanism is used for data merging, and a weighted summation operation is used to encrypt and aggregate data from different modalities. This process maintains data privacy while effectively combining different modalities to obtain a richer encrypted data output.

[0108] Encrypted, comprehensive data is stored in the blockchain to ensure its immutability. The blockchain's decentralized storage mechanism ensures redundant data storage across multiple nodes, and smart contracts verify the integrity of the stored encrypted data. The smart contracts automatically execute verification rules each time data is written to the blockchain, ensuring the validity of the encrypted data on the blockchain.

[0109] To verify the effectiveness of the method of this invention, an experiment was conducted on data encryption processing in an intelligent medical system. The experimental data included the diagnosis and treatment data of 500 patients, involving graph data, text data, and sensor data. The experimental results are as follows:

[0110] Sample Number Pre-encryption computation time (seconds) Post-encryption computation time (seconds) Pre-encryption data storage size (MB) Post-encryption data storage size (MB) 1 15 20 5 10 2 12 17 4.8 9.5 3 14 19 5.2 10.2 4 13 18 5.1 10.1 5 16 21 5.3 10.3

[0111] The experimental data shows that, although encryption increases computation time, compared to traditional data processing methods, the system maintains a reasonable computation time and data storage size while ensuring data privacy. Furthermore, the system can efficiently process large-scale patient data, and the data is stored on the blockchain, guaranteeing its immutability and integrity.

[0112] Based on the above experimental results, the blockchain-based data security encryption processing system of this invention successfully solves the problems of privacy protection and computational efficiency in existing technologies, and has significant practical application value. This method can achieve efficient computation and analysis while ensuring data security, providing an innovative solution for sensitive data processing in intelligent medical systems.

Claims

1. A blockchain-based data security encryption processing system, characterized by, The method comprises the following steps: A data acquisition and preprocessing module acquires data from multiple data sources, including graph data and multi-modal data, and preprocesses the acquired data to obtain a standardized input dataset; A graph data homomorphic encryption module performs homomorphic encryption processing on the graph data based on the standardized input dataset, encrypts each node in the graph using a post-quantum homomorphic encryption algorithm, ensures that the data can be calculated by a graph neural network in an encrypted state, and obtains encrypted graph data; A multi-modal data privacy-enhanced encryption module performs privacy-enhanced encryption processing on the multi-modal data based on the standardized input dataset, the multi-modal data including text and sensor data, and uses a hybrid encryption method to encrypt each type of modal data to obtain encrypted multi-modal data; A graph neural network calculation module inputs the encrypted graph data and encrypted multi-modal data into a graph neural network, and the homomorphically encrypted graph data is calculated by the graph neural network model in an encrypted state to update the node features and obtain encrypted calculation results; An encrypted data fusion module fuses the encrypted calculation results and the encrypted multi-modal data to obtain comprehensive encrypted data output results; A blockchain storage and verification module stores the comprehensive encrypted data output results in a blockchain, ensures the data's non-tamperability through the blockchain's decentralized storage mechanism, and verifies the integrity and validity of the encrypted data through a smart contract.

2. The blockchain-based data security encryption processing system of claim 1, wherein, The data acquisition and preprocessing module specifically comprises: S1.1: Acquire raw data from multiple data sources to obtain graph data and multi-modal data; S1.2: Format the acquired raw graph data, convert the graph data into a standard graph structure, and normalize the graph structure according to the properties of the graph; S1.3: Preprocess the acquired raw multi-modal data to obtain a standardized multi-modal dataset; S1.4: Uniformly format the preprocessed graph data and multi-modal data to obtain a standardized dataset. 3.The blockchain-based data security encryption processing system of claim 1, wherein, The graph data homomorphic encryption module specifically comprises: S2.1: Based on the standardized input dataset, extract node features in the graph structure, including node attribute data, the graph structure including nodes and edges, and the edges representing the connection relationship between nodes; S2.2: Select a post-quantum homomorphic encryption algorithm, wherein the post-quantum homomorphic encryption algorithm uses a lattice-based encryption mechanism, the encryption process uses an encryption function to encrypt the feature data of each node to obtain encrypted node feature data, and the encryption function is a mapping of homomorphic encryption operations; S2.3: Apply addition and multiplication operations in the homomorphic encryption algorithm to each encrypted node feature data, wherein the addition operation is used for node feature aggregation, and the multiplication operation is used for interaction relationship calculation between features to obtain encrypted processing results; S2.4: Homomorphic encryption processing is performed on edge relationship according to the structure of the graph, the edge relationship representing connection information between nodes in the graph, connection information of each edge being encrypted using the same post-quantum homomorphic encryption algorithm to obtain encrypted edge data, the encrypted edge data being used together with the encrypted node features to form encrypted graph data; S2.5: The encrypted graph data is generated, the encrypted graph data including the encrypted node features and the encrypted edge data.

4. The blockchain-based data security encryption processing system of claim 1, wherein, The multi-modal data privacy-enhanced encryption module specifically includes: S3.1: Based on the standardized input data set, multi-modal data is extracted from multiple data sources, the multi-modal data including text data and sensor data, the text data including natural language text data, and the sensor data including data streams from multiple sensors; S3.2: The text data is encrypted, a symmetric encryption algorithm being used to encrypt the text data, the symmetric encryption algorithm being the advanced encryption standard, the text data being encrypted by a symmetric key, the symmetric key being generated by an encryption key derivation function; S3.3: The sensor data is encrypted, a hash algorithm being used to encrypt the sensor data, the hash function used being SHA-256, the sensor data being encrypted by the hash function; S3.4: The encrypted multi-modal data is generated, the encrypted text data including encrypted text content and an encrypted symmetric key, the encrypted sensor data being a hash encryption result.

5. The blockchain-based data security encryption processing system of claim 1, wherein, The graph neural network calculation module specifically includes: S4.1: The encrypted graph data and the encrypted multi-modal data are input into a graph neural network model, the encrypted graph data including encrypted node features and encrypted edge data, the encrypted multi-modal data including encrypted text data and encrypted sensor data; S4.2: In the graph neural network model, a graph convolution layer is applied to the encrypted node feature data for encrypted calculation, the graph convolution layer performing encrypted aggregation on features of neighbor nodes by a homomorphic encryption algorithm to obtain an encrypted calculation result of each node; S4.3: An activation function in the graph neural network is applied to the encrypted node feature data for calculation, the activation function being used to perform nonlinear transformation on the encrypted node features to generate updated encrypted node features; S4.4: The updated encrypted node features and the encrypted edge data are output together, the encrypted calculation result including the updated encrypted node features and the encrypted edge data.

6. The blockchain-based data security encryption processing system of claim 1, wherein, The encrypted data fusion module specifically includes: S5.1: The encrypted calculation result and the encrypted multi-modal data are matched, the encrypted calculation result including encrypted node features and encrypted edge data obtained from the graph neural network model, the encrypted multi-modal data including encrypted text data and encrypted sensor data; S5.2: Data alignment processing is performed on the encrypted calculation result and the encrypted multi-modal data, the data alignment processing including synchronization according to timestamps of the node features and the multi-modal data to obtain matched encrypted data; S5.3: Perform an encrypted fusion operation on the matched encrypted data, the encrypted fusion operation including encrypted merging of the encrypted node features and the encrypted multi-modal data through a homomorphic encryption mechanism, the merged data remaining in an encrypted state, to obtain encrypted fusion data; S5.4: Perform encrypted information aggregation on the encrypted fusion data, the aggregation process including weighted summation of different features of the encrypted data using homomorphic encryption addition operations, to obtain an encrypted fusion result for each node; S5.5: Obtain a comprehensive encrypted data output result based on the encrypted fusion result.

7. The blockchain-based data security encryption processing system of claim 1, wherein, The blockchain storage and verification module specifically includes: S6.1: Format the comprehensive encrypted data output result through an encryption protocol, the encrypted data output result including encrypted graph data and encrypted multi-modal data, to obtain formatted encrypted data; S6.2: Store the formatted encrypted data output result in a blockchain, the blockchain being a decentralized distributed ledger, with data being recorded to multiple distributed nodes through blockchain nodes; S6.3: Generate a new block in the blockchain, the block including a hash value of the stored encrypted data and related timestamp information, the hash value being generated through an encryption algorithm and used to uniquely identify the encrypted data; S6.4: Verify the stored encrypted data through a smart contract in the blockchain network, the smart contract checking the integrity and validity of the encrypted data through predefined verification rules.

Images