Multi-party data security calculation method and system based on TEE and encrypted mirror image

Abstract

The invention relates to a multi-party data security computing method and system based on TEE and an encrypted mirror image, and belongs to the technical field of data security computing, and the system sequentially comprises an access layer, a security authentication and control layer, a data and mirror image security layer and a security computing environment layer from top to bottom, the method comprises the steps of encrypting and uploading a data demander program mirror image, encrypting and uploading participant data, performing safety calculation in TEE and encrypting and downloading a data demander result. In a word, the method has the advantages of realizing full-process security protection of multi-party data security calculation, effectively reducing the risk of data leakage, providing intellectual property protection of calculation codes, preventing algorithm leakage and reverse engineering, and ensuring commercial confidential security of users.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data security computing, and particularly relates to a multi-party data security computing method and system based on TEE and encrypted images. BACKGROUND

[0002] The data security field covers multiple key technical directions such as security protection, privacy computing, and secure transmission in the whole life cycle of data. From the technical evolution trend, data security is developing from single data encryption storage to more complex multi-party collaborative computing. The traditional "data island" mode cannot meet the collaborative development needs of the digital economy. How to realize secure data collaboration under the premise of protecting the privacy of each party's data has become a key technical challenge.

[0003] Under this background, multi-party data security computing technology based on trusted execution environment (TEE) emerged as the times require. This technology creates a secure area (Enclave) isolated from the outside in the CPU, ensuring that data and code are not stolen or tampered with by the external environment during the computing process. The multi-party data security computing system realized based on this technology transmits the data of each party to the TEE computing party, decrypts and computes in the TEE, and outputs the computing result in the TEE, forming a secure computing link from input to output. However, the existing scheme still has significant deficiencies in realizing end-to-end security: 1) the key management mechanism is weak, lacking secure key distribution based on hardware authentication, and data may be stolen or leaked during transmission and storage; 2) there is a lack of effective protection mechanism for the intellectual property rights of computing code, and users need to upload the computing logic in plaintext to the platform in advance, which has a risk of leakage. SUMMARY

[0004] The purpose of the present application is to overcome the deficiencies in the prior art and to provide a multi-party data security computing method and system based on TEE and encrypted images. The system adopts a layered security architecture design, which includes an access layer, a security authentication and control layer, a data and image security layer, and a secure computing environment layer from top to bottom, and each layer cooperates to build an end-to-end multi-party data security computing system.

[0005] The technical scheme of the present application is as follows:

[0006] A multi-party data security computing system based on TEE and encrypted images, characterized by comprising:

[0007] from top to bottom, an access layer, a security authentication and control layer, a data and image security layer, and a secure computing environment layer;

[0008] The access layer is the interaction interface between the system and external entities, including data demanders, data providers, and computing resource providers.

[0009] The security authentication and control layer is the intelligent management and control center of the system, and is used for safety management and control, task scheduling and policy execution of the system;

[0010] The data and image security layer is used for safety protection and management of data and algorithm logic;

[0011] The data and image security layer includes an image layered encryption module, an image key management service and an encrypted image warehouse;

[0012] The image layered encryption module runs in a data demand side, is responsible for layered encryption processing of algorithm images of the data demand side, independently encrypts each image layer, and registers image keys to the image key management service;

[0013] The image key management service runs in the data demand side, and is specially used for managing key information required by image encryption;

[0014] The encrypted image warehouse runs in a computing resource provider, stores encrypted algorithm images, provides image submission, pulling, verification and version management services, supports image integrity checking and access control;

[0015] The secure computing environment layer is used for providing a hardware-level security basis and a runtime environment of the system.

[0016] Further, the data and image security layer further includes

[0017] A data encryption transmission module: the data encryption transmission module is divided into a client and a server, runs the client in a data demand side and a data provider, and runs the server in a computing resource provider, the data encryption transmission module is responsible for encrypting original data of the data demand side and the data provider, and safely uploading the original data to the computing resource provider, and downloading TEE calculation results to the data demand side and decrypting the TEE calculation results, and safely transmitting data keys to a data key management service through a remote authentication security channel;

[0018] A data key management service: running in a TEE security environment of the computing resource provider, the data key management service is specially used for managing keys required by data encryption, is responsible for receiving and safely storing data keys from the data encryption transmission module, and provides decryption keys for the TEE environment during calculation;

[0019] A ciphertext data storage module: running in the computing resource provider, the ciphertext data storage module is used for storing encrypted data uploaded by each participant, and the data always keeps in an encrypted state during storage.

[0020] Further, the security authentication and control layer includes:

[0021] TEE Remote Authentication Module: Based on TEE hardware authentication capabilities, it performs remote authentication and integrity verification on TEE servers;

[0022] Access control policy module: The attribute-based access control model defines security policies, implements dynamic authorization management for data access, key invocation and task execution processes, and supports fine-grained permission control, accurate to data fields and operation types;

[0023] Secure Computing Scheduling Module: Responsible for orchestrating and scheduling secure computing tasks, and scheduling TEE nodes to execute secure computing processes based on remote authentication results and access control policies.

[0024] Furthermore, the secure computing environment layer includes:

[0025] TEE Server: A physical server equipped with TEE technology, providing CPU-level hardware security isolation capabilities and creating an Enclave security zone to protect the computing process.

[0026] Confidential container runtime: Runs within the TEE secure zone, responsible for decrypting and loading encrypted images, ensuring that the code and data within the container are protected by hardware during runtime, and starting the confidential container to execute custom algorithms submitted by the data requester;

[0027] Secure transmission module: Uses a TEE remote authentication secure channel to perform encrypted transmission of sensitive data between nodes;

[0028] Full-disk encrypted hard drive: The storage mounted on the TEE container is encrypted to protect the storage security of data written to disk during algorithm execution.

[0029] This invention also proposes a secure computation method for a multi-party data secure computation system based on TEE and encrypted mirrors, comprising the following steps:

[0030] S1: Encrypted upload of data requester's program image: The data requester builds an encrypted image using a custom algorithm and uploads it;

[0031] S2: Encrypted data upload by participating parties: The data requester and data provider respectively encrypt their data to be calculated and upload it to the TEE environment;

[0032] S3: Secure computation within a TEE: Securely decrypts images and data, executes computation logic, and encrypts computation results within a physical server with TEE functionality provided by the computing resource provider;

[0033] S4: Encrypted Download of Results for Data Requesters: Data requesters securely download and decrypt the final calculation results.

[0034] Furthermore, the data requester program image encryption upload process in step S1 includes the following steps:

[0035] S11: The data requester packages the custom algorithm or model code into a container image;

[0036] S12: The image layer encryption module generates a symmetric encryption key independently for each image layer of the container image, performs layer encryption processing on the container image, and generates an encrypted image;

[0037] S13: The mirror key management service generates a corresponding public and private key for each pair of encryption keys and stores the private key to the data requester.

[0038] S14: The image encryption module uses a public key to encrypt a symmetric encryption key and generates a sealing key;

[0039] S15: Push the encrypted image to the encrypted image repository of the computing resource provider.

[0040] Furthermore, the data encryption and uploading process for participants in step S2 includes the following steps:

[0041] S21: The data provider or data requester generates a dedicated data key for the data to be uploaded;

[0042] S22: Encrypt the original data using the generated data key;

[0043] S23: Perform TEE remote authentication on the data key management service of the computing resource provider and obtain the public key of the TEE environment;

[0044] S24: Encrypt data keys using the public key of the TEE environment;

[0045] S25: Data key management service that securely transmits data keys to computing resource providers via a TEE remote authentication secure channel;

[0046] S26: The encrypted original data is uploaded to the encrypted data storage module of the computing resource provider through the secure transmission module, and remains encrypted at all times.

[0047] Furthermore, the secure computation process within the TEE in step S3 includes the following steps:

[0048] S31: The computing resource provider implements access control policies for algorithm images and datasets to ensure that permissions have been granted;

[0049] S32: The computing resource provider starts the TEE environment, generates a remote authentication report and a TEE key pair, and sends the TEE public key in the remote authentication report and TEE key pair to the data requester. The data requester verifies the remote authentication report and thus confirms that the computing resource provider provides a valid TEE environment.

[0050] S33: After the data requester verifies the TEE environment, it uses the public key of the TEE environment to encrypt the private key stored in step S13 and returns it to the TEE environment through a secure channel.

[0051] S34: The TEE process uses its own private key to decrypt the private key of the data requester in step S33, and then uses the private key of the data requester to decrypt the encapsulation key generated in step S14 to obtain the symmetric encryption key. After that, the symmetric encryption key is used to decrypt the encrypted image in the TEE environment and start the container program.

[0052] S35: Programs within the TEE environment request data keys from the data key management service, obtain them, decrypt the stored encrypted data, and prepare input data for computation tasks.

[0053] S36: Execute user-defined calculation logic under the isolation and protection of the TEE server, process the decrypted plaintext data, and generate calculation results;

[0054] S37: A random result key is generated within the TEE environment, and this key is used to perform symmetric encryption on the plaintext calculation result.

[0055] S38: Output the encrypted calculation result to an external storage system;

[0056] S39: The TEE environment requests remote authentication from the data key management service and obtains the TEE public key from the data key management service. It then uses the TEE public key to perform asymmetric encryption protection on the resulting key.

[0057] S310: Upload the encrypted result key to the data key service for secure storage;

[0058] Furthermore, the data requester result encryption and download process in step S4 includes the following steps:

[0059] S41: The data requester initiates TEE remote authentication with the data key management service to verify the authenticity and integrity of the TEE service and obtain the TEE public key;

[0060] S42: The data requester generates a temporary symmetric key locally, which serves as the encapsulation key for the subsequent encryption result key;

[0061] S43: Use the TEE public key to perform asymmetric encryption on the generated symmetric key;

[0062] S44: Send the encrypted symmetric key to the data key service in the TEE, and use the TEE private key to decrypt and obtain the plaintext symmetric key.

[0063] S45: The TEE environment uses the decrypted symmetric key plaintext to symmetrically encrypt the result key it holds, and protects the secure transmission of the result key through a key encapsulation mechanism.

[0064] S46: The data requester uses the locally stored symmetric key to decrypt the encrypted result key received from the TEE and obtain the plaintext result key;

[0065] S47: The data requester downloads the computation result data encrypted with the result key from the TEE environment;

[0066] S48: The data requester uses the obtained result key to decrypt the encrypted calculation result;

[0067] S49: Decryption is complete, and the data requester obtains the final plaintext calculation result.

[0068] Compared with the prior art, the beneficial effects of the present invention are:

[0069] Technical benefits: It achieves full-process security protection for multi-party data secure computation, effectively reduces the risk of data leakage, provides intellectual property protection for computation code, prevents algorithm leakage and reverse engineering, ensures the security of users' business secrets, supports domestic TEE technology, meets national security compliance requirements, and encrypts the storage and transmission of results to avoid the risk of plaintext exposure and ensures the reliability and security of output.

[0070] Economic Benefits: Reducing economic losses from data breaches can decrease losses from security incidents, promote the secure flow and value release of data elements, improve data utilization efficiency, reduce security compliance costs, lower manual auditing costs through automated security protection, drive the development of new data-driven business models, and create new economic value. Attached Figure Description

[0071] Figure 1 This is a system architecture diagram of the present invention;

[0072] Figure 2 This is a flowchart of the calculation method of the present invention;

[0073] Figure 3 This is a flowchart illustrating the data requester program image encryption upload process for the calculation method of this invention.

[0074] Figure 4 This is a flowchart illustrating the process of encrypting and uploading data to the participants in the calculation method of this invention.

[0075] Figure 5This is a flowchart of the secure computation within a TEE using the computation method of this invention;

[0076] Figure 6 This is a flowchart illustrating the encrypted download process of the data requester's result for the calculation method of this invention. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] Example 1

[0079] like Figure 1 As shown, a multi-party data security computing system based on TEE and encrypted mirrors adopts a layered security architecture design, which includes, from top to bottom, an access layer, a security authentication and control layer, a data and mirror security layer, and a secure computing environment layer. The functions of each layer work together to build an end-to-end multi-party data security computing system.

[0080] The first layer: The access layer is the interface between the system and external entities, including three core participants:

[0081] 1) Data requester: The party that owns the computing and analysis algorithm or model initiates a secure computing task by providing the algorithm code in the form of a container image, and provides its own raw data as needed.

[0082] 2) Data Provider: The party that possesses sensitive data, is willing to participate in multi-party computation but requires that data privacy be guaranteed and that the original data not be leaked.

[0083] 3) Computing resource provider: Provide physical servers with TEE functionality, provide computing power but should not be able to spy on the content of computing tasks.

[0084] The second layer, the security authentication and control layer, is the intelligent management and control center of the system, responsible for system security management, task scheduling, and policy execution.

[0085] 1) TEE Remote Authentication Module: Based on TEE hardware authentication capabilities, it performs remote identity verification and integrity checks on TEE servers to ensure that the computing environment is trustworthy and verifiable.

[0086] 2) Access Control Policy Module: Based on the ABAC (Attribute-Based Access Control) model, this module defines security policies and implements dynamic authorization management for data access, key retrieval, and task execution. It supports fine-grained permission control, accurate to the data field and operation type.

[0087] 3) Secure Computing Scheduling Module: This module is responsible for orchestrating and scheduling secure computing tasks. It combines remote authentication results and access control policies to schedule TEE nodes to execute secure computing processes, thereby achieving efficient resource utilization and task isolation.

[0088] The third layer: The data and image security layer is responsible for the security protection and management of core assets (data and algorithm logic), ensuring the security of data and algorithm logic during storage and distribution.

[0089] 1) Data Encryption Transmission Module: The data encryption module consists of a client and a server. The client runs on the data requester and data provider, while the server runs on the computing resource provider. This module is responsible for encrypting the raw data from the data requester and data provider and securely uploading it to the computing resource provider, as well as downloading the TEE calculation results to the data requester and decrypting them. It also securely transmits the data key to the data key management service through a remote authentication secure channel.

[0090] 2) Data Key Management Service: Running within the TEE security environment of the computing resource provider, this service is dedicated to managing the keys required for data encryption. It is responsible for receiving and securely storing data keys from the data encryption transmission module and providing decryption keys to the TEE environment during computation.

[0091] 3) Encrypted data storage module: running on the computing resource provider, used to store encrypted data uploaded by each participant. The data remains encrypted throughout the storage process.

[0092] 4) Image Layered Encryption Module: Running on the data requester, this module is responsible for performing layered encryption on its own algorithm image. Each image layer is encrypted independently, and the image keys are registered with the image key management service.

[0093] 5) Mirror Key Management Service: Running on the data requester, this service specifically manages the key information required for mirror encryption. During TEE computation, the computing resource provider requests the mirror key from the data requester through a remotely authenticated secure channel, ensuring the security and traceability of the key request process.

[0094] 6) Encrypted Image Repository: Running on the computing resource provider, it stores encrypted algorithm images and provides image submission, retrieval, verification, and version management services. It supports image integrity verification and access control to ensure the security of the image distribution process.

[0095] The fourth layer: the secure computing environment layer provides the system's hardware-level security foundation and runtime environment, and is the foundation of the entire security system.

[0096] 1) TEE Server: A physical server equipped with TEE technology, providing CPU-level hardware security isolation capabilities and creating an Enclave security zone to protect the computing process.

[0097] 2) Confidential Container Runtime: Runs within the TEE secure zone, responsible for decrypting and loading encrypted images, ensuring that code and data within the container are protected by hardware during runtime. It starts the confidential container to execute custom algorithms submitted by the data requester.

[0098] 3) Secure transmission module: Uses TEE remote authentication secure channel to perform encrypted transmission of sensitive data between nodes.

[0099] Full-disk encrypted hard drive: The storage mounted on the TEE container is fully encrypted to protect the storage security of data written to disk during algorithm execution.

[0100] TEE Hardware Expansion Solution: This solution is expandable to support multiple TEE architectures, including Intel SGX, Intel TDX, AMD SEV, Hygon CSV, ARM TrustZone, etc., achieving cross-platform security protection through a unified authentication interface. Replacing different TEE implementations does not affect the overall solution architecture.

[0101] Encryption algorithm adaptation scheme: Supports the parallel operation of international standard encryption algorithms (such as RSA, AES) and Chinese national cryptographic algorithms (SM2, SM4) to meet the security requirements of different scenarios. Symmetric encryption can use AES or SM4, etc., without affecting the overall architecture.

[0102] Deployment method change: Supports container running solutions of Kubernetes + containerd, and can also be applied to other container or virtualization platforms, such as Docker Swarm or OpenStack, to adapt to different infrastructure environments.

[0103] This embodiment innovatively integrates encrypted container image technology, multi-party encrypted data computation technology, and TEE hardware capabilities to construct a trusted execution environment covering the entire chain from "data input to computation environment to algorithm logic to result output." This architecture ensures that data and computation logic are never exposed in plaintext outside the trusted environment at any stage during secure computing tasks, achieving end-to-end security.

[0104] Example 2

[0105] like Figures 2 to 6As shown, a secure computing method for a multi-party data security computing system based on TEE and encrypted images is presented. The core purpose of this solution is to build a secure TEE computing environment to ensure that the data of each participant is not leaked, while supporting data requesters to upload custom algorithm images and protecting the security of the program image throughout the entire process.

[0106] The entire secure computing process involves three main stakeholders:

[0107] Participants Main responsibilities Key services / components Data demander Upload custom algorithm, provide partial data, and finally get the calculation result. Image building and encryption, image key service, data transmission service. Data provider Mainly responsible for providing data. Data transmission service. Computing resource provider As a trusted base, provide TEE hardware environment, which is the core of protecting algorithm image and data from leakage. Data transmission service (running in TEE), container image repository, data key service (running in TEE), custom algorithm task (running in TEE)

[0108] The data requester obtains the computation results through the custom algorithm provided by the requester. The "Image Building and Encryption" service is responsible for building the user-defined algorithm into an algorithm image and encrypting it; the "Image Key Service" is responsible for managing the encryption key of the program image and ensuring the secure distribution and control of the key; and the "Data Transmission Service" is responsible for encrypting and uploading the requester's data to the computing resource provider and obtaining the computing results from the computing resource provider.

[0109] The data provider is primarily responsible for providing the data, while the "data transmission service" is responsible for encrypting and uploading its own data to the computing resource provider, participating in secure computing.

[0110] The computing resource provider, acting as a trusted foundation, provides the TEE hardware environment, which is crucial for protecting program images and data from leakage. The "Data Transmission Service (running on the TEE)" is responsible for receiving encrypted data from participating parties and transmitting the final computation results to the data requester; the "Container Image Repository" stores encrypted program images, awaiting computation calls; the "Data Key Service (running on the TEE)" receives data keys from participating parties and stores them encrypted; and the "Program Computation Service (running on the TEE)" decrypts and executes the program image within the TEE to complete the computation task, fundamentally eliminating the risk of program logic and data leakage.

[0111] The entire process can be divided into the following four main stages (S1 to S4):

[0112] S1. Data Requester Program Image Encryption Upload Process: The data requester builds an encrypted image using a custom algorithm and uploads it.

[0113] S2. Participant Data Encryption Upload Process: The data requester and data provider respectively encrypt their data to be calculated and upload it to the TEE environment.

[0114] S3. Secure computation process within TEE: Securely decrypt the image and data, execute computation logic, and encrypt the computation results in a physical server with TEE functionality provided by the computing resource provider.

[0115] S4. Encrypted download process for data requester results: The data requester securely downloads and decrypts the final calculation results.

[0116] The process for each stage will be explained in detail below.

[0117] S1: Data requester program image encrypted upload process.

[0118] Process description:

[0119] 1) Data requesters package their custom algorithms or model code into container images using the Docker image format to ensure portability and standardization.

[0120] 2) The image layer encryption module generates a symmetric encryption key independently for each image layer of the container image, performs layer encryption processing on the container image, and generates an encrypted image.

[0121] 3) The mirror key management service generates a corresponding public and private key for each pair of encryption keys and stores the private key to the data requester.

[0122] 4) The image encryption module uses a public key to encrypt the symmetric encryption key, generating a packaging key. Only the user's private key can decrypt the symmetric key. In this step, the packaging key is included in the encrypted image. Decryption is impossible without the private key.

[0123] 5) Push the encrypted image to the computing resource provider's encrypted image repository. The encrypted image cannot be decrypted and used without the key.

[0124] S2: Participant Data Encryption Upload Process

[0125] Process description:

[0126] 1) The data provider or data requester generates a dedicated encryption key for the data to be uploaded.

[0127] 2) Encrypt the original data using the generated key.

[0128] 3) Perform TEE remote authentication on the data key management service of the computing resource provider and obtain the public key of the TEE environment.

[0129] 4) Data keys are encrypted using the TEE public key, and can only be decrypted using the private key of the TEE environment, thus ensuring the security of data key transmission and storage.

[0130] 5) Securely transmit data keys to the data key management service of the computing resource provider through the TEE remote authentication security channel, which is established based on TEE hardware authentication.

[0131] 6) The encrypted data is uploaded to the encrypted data storage module of the computing resource provider through a secure transmission module, and remains encrypted at all times.

[0132] S3: Secure Computation Process within TEE

[0133] Process description:

[0134] 1) The computing resource provider implements access control policies for algorithm images and datasets to ensure that permissions have been granted.

[0135] 2) The computing resource provider starts the TEE environment, generates a remote authentication report and a TEE key pair, and sends the remote authentication report and TEE public key to the data requester. The data requester verifies the remote authentication report and thus confirms that the computing resource provider provides a valid TEE environment.

[0136] 3) After the data requester verifies the TEE identity (TEE environment), the private key stored in the data requester is encrypted using the TEE public key and returned to the TEE environment through a secure channel.

[0137] 4) In the TEE stage, the private key of the data requester is decrypted using its own private key, and then the encapsulation key is decrypted using the private key stored in the data requester to obtain the symmetric encryption key. After that, the symmetric encryption key is used to decrypt the encrypted image in the TEE environment and start the container program.

[0138] 5) The program within the TEE requests the data key from the data key management service, obtains it, decrypts the stored encrypted data, and prepares input data for the computing task.

[0139] 6) Execute user-defined calculation logic under the isolation and protection of the TEE server, process the decrypted plaintext data, and generate calculation results.

[0140] 7) The TEE generates a random result key, which is then used to perform symmetric encryption on the plaintext calculation result.

[0141] 8) The encrypted calculation results are output to an external storage system, and the ciphertext state ensures the security of the storage.

[0142] 9) The TEE environment requests remote authentication from the data key management service and obtains the TEE public key from the data key management service. The TEE public key is then used to perform asymmetric encryption protection on the resulting key.

[0143] 10) Upload the encrypted result key to the data key service for secure storage to complete the entire calculation process.

[0144] S4: Encrypted Download Process for Data Requesters

[0145] Process description:

[0146] 1) The data requester initiates TEE remote authentication for the data key service to verify the authenticity and integrity of the TEE service and obtain the TEE public key. The TEE public key here is the data key management service's own TEE public and private key, which can be considered to be generated when the service is first started.

[0147] 2) The data requester generates a temporary symmetric key locally, which serves as the encapsulation key for the subsequent encryption result key.

[0148] 3) Use the TEE public key to perform asymmetric encryption on the generated symmetric key to ensure the confidentiality of the key during transmission.

[0149] 4) Send the encrypted symmetric key to the data key service in the TEE, use the TEE private key to decrypt and obtain the plaintext of the symmetric key, and the data requester initiates TEE remote authentication with the data key management service to verify the authenticity and integrity of the TEE service and obtain the TEE private key.

[0150] 5) The TEE uses the decrypted symmetric key to perform symmetric encryption on the result key it holds, and protects the secure transmission of the result key through a key encapsulation mechanism.

[0151] 6) The data requester uses the locally stored symmetric key to decrypt the encrypted result key received from the TEE and obtain the plaintext result key.

[0152] 7) The data requester downloads the computation result data encrypted with the result key from the TEE environment.

[0153] 8) The data requester uses the obtained result key to decrypt the encrypted calculation result.

[0154] 9) After the decryption operation is completed, the data requester obtains the final plaintext calculation result, completing the entire secure download process.

[0155] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-party data security computing system based on TEE and encrypted mirroring, characterized in that: From top to bottom, it includes the access layer, security authentication and control layer, data and image security layer, and secure computing environment layer; The access layer is the interface between the system and external entities, including data requesters, data providers, and computing resource providers. The security authentication and control layer is the intelligent management and control center of the system, used for system security management, task scheduling and policy execution. The data and mirror security layer is used for the security protection and management of data and algorithm logic; The data and image security layer includes an image layered encryption module, an image key management service, and an encrypted image repository; The image layered encryption module runs on the data requester and is responsible for performing layered encryption on the algorithm image of the data requester. It encrypts each image layer independently and registers the image key to the image key management service. The image key management service runs on the data requester and is specifically designed to manage the key information required for image encryption. The encrypted image repository runs on the computing resource provider, stores encrypted algorithm images, provides image submission, retrieval, verification and version management services, and supports image integrity verification and access control; The secure computing environment layer is used to provide the system's hardware-level security foundation and runtime environment.

2. The multi-party data security computing system based on TEE and encrypted mirroring according to claim 1, characterized in that: The data and mirror security layer also includes: Data encryption transmission module: The data encryption transmission module is divided into a client and a server. The client runs on the data requester and the data provider, and the server runs on the computing resource provider. The data encryption transmission module is responsible for encrypting the original data of the data requester and the data provider and securely uploading it to the computing resource provider, as well as downloading the TEE calculation results to the data requester and decrypting them. It also securely transmits the data key to the data key management service through a remote authentication secure channel. Data Key Management Service: Running in the TEE security environment of the computing resource provider, it is dedicated to managing the keys required for data encryption, receiving and securely storing data keys from the data encryption transmission module, and providing decryption keys for the TEE environment during computation; Encrypted data storage module: Runs on the computing resource provider and is used to store encrypted data uploaded by each participant. The data remains encrypted throughout the storage process.

3. The multi-party data security computing system based on TEE and encrypted mirroring according to claim 2, characterized in that: The security authentication and control layer includes: TEE Remote Authentication Module: Based on TEE hardware authentication capabilities, it performs remote authentication and integrity verification on TEE servers; Access control policy module: The attribute-based access control model defines security policies, implements dynamic authorization management for data access, key invocation and task execution processes, and supports fine-grained permission control, accurate to data fields and operation types; Secure Computing Scheduling Module: Responsible for orchestrating and scheduling secure computing tasks, and scheduling TEE nodes to execute secure computing processes based on remote authentication results and access control policies.

4. A multi-party data security computing system based on TEE and encrypted mirroring according to claim 3, characterized in that: The secure computing environment layer includes: TEE Server: A physical server equipped with TEE technology, providing CPU-level hardware security isolation capabilities and creating an Enclave security zone to protect the computing process. Confidential container runtime: Runs within the TEE secure zone, responsible for decrypting and loading encrypted images, ensuring that the code and data within the container are protected by hardware during runtime, and starting the confidential container to execute custom algorithms submitted by the data requester; Secure transmission module: Uses a TEE remote authentication secure channel to perform encrypted transmission of sensitive data between nodes; Full-disk encrypted hard drive: The storage mounted on the TEE container is encrypted to protect the storage security of data written to disk during algorithm execution.

5. A secure computing method for a multi-party data security computing system based on TEE and encrypted mirroring as described in claim 4, characterized in that: Includes the following steps: S1: Encrypted upload of data requester's program image: The data requester builds an encrypted image using a custom algorithm and uploads it; S2: Encrypted data upload by participating parties: The data requester and data provider respectively encrypt their data to be calculated and upload it to the TEE environment; S3: Secure computation within a TEE: Securely decrypts images and data, executes computation logic, and encrypts computation results within a physical server with TEE functionality provided by the computing resource provider; S4: Encrypted Download of Results for Data Requesters: Data requesters securely download and decrypt the final calculation results.

6. The multi-party data security calculation method based on TEE and encrypted mirror as described in claim 5, characterized in that: The data requester program image encryption upload process in step S1 includes the following steps: S11: The data requester packages the custom algorithm or model code into a container image; S12: The image layer encryption module generates a symmetric encryption key independently for each image layer of the container image, performs layer encryption processing on the container image, and generates an encrypted image; S13: The mirror key management service generates a corresponding public and private key for each pair of encryption keys and stores the private key to the data requester. S14: The image encryption module uses a public key to encrypt a symmetric encryption key and generates a sealing key; S15: Push the encrypted image to the encrypted image repository of the computing resource provider.

7. The multi-party data security calculation method based on TEE and encrypted mirror as described in claim 5, characterized in that: The data encryption and uploading process for participants in step S2 includes the following steps: S21: The data provider or data requester generates a dedicated data key for the data to be uploaded; S22: Encrypt the original data using the generated data key; S23: Perform TEE remote authentication on the data key management service of the computing resource provider and obtain the public key of the TEE environment; S24: Encrypt data keys using the public key of the TEE environment; S25: Data key management service that securely transmits data keys to computing resource providers via a TEE remote authentication secure channel; S26: The encrypted original data is uploaded to the encrypted data storage module of the computing resource provider through the secure transmission module, and remains encrypted at all times.

8. The multi-party data security computation method based on TEE and encrypted mirror as described in claim 5, characterized in that: The secure calculation process within the TEE in step S3 includes the following steps: S31: The computing resource provider implements access control policies for algorithm images and datasets to ensure that permissions have been granted; S32: The computing resource provider starts the TEE environment, generates a remote authentication report and a TEE key pair, and sends the TEE public key in the remote authentication report and TEE key pair to the data requester. The data requester verifies the remote authentication report and thus confirms that the computing resource provider provides a valid TEE environment. S33: After the data requester verifies the TEE environment, it uses the public key of the TEE environment to encrypt the private key stored in step S13 and returns it to the TEE environment through a secure channel. S34: The TEE process uses its own private key to decrypt the private key of the data requester in step S33, and then uses the private key of the data requester to decrypt the encapsulation key generated in step S14 to obtain the symmetric encryption key. After that, the symmetric encryption key is used to decrypt the encrypted image in the TEE environment and start the container program. S35: Programs within the TEE environment request data keys from the data key management service, obtain them, decrypt the stored encrypted data, and prepare input data for computation tasks. S36: Execute user-defined calculation logic under the isolation and protection of the TEE server, process the decrypted plaintext data, and generate calculation results; S37: A random result key is generated within the TEE environment, and this key is used to perform symmetric encryption on the plaintext calculation result. S38: Output the encrypted calculation result to an external storage system; S39: The TEE environment requests remote authentication from the data key management service and obtains the TEE public key from the data key management service. It then uses the TEE public key to perform asymmetric encryption protection on the resulting key. S310: Upload the encrypted result key to the data key service for secure storage.

9. A multi-party data security calculation method based on TEE and encrypted mirror as described in claim 6, characterized in that: The data requester result encryption and download process in step S4 includes the following steps: S41: The data requester initiates TEE remote authentication with the data key management service to verify the authenticity and integrity of the TEE service and obtain the TEE public key; S42: The data requester generates a temporary symmetric key locally, which serves as the encapsulation key for the subsequent encryption result key; S43: Use the TEE public key to perform asymmetric encryption on the generated symmetric key; S44: Send the encrypted symmetric key to the data key service in the TEE, and use the TEE private key to decrypt and obtain the plaintext symmetric key. S45: The TEE environment uses the decrypted symmetric key plaintext to symmetrically encrypt the result key it holds, and protects the secure transmission of the result key through a key encapsulation mechanism. S46: The data requester uses the locally stored symmetric key to decrypt the encrypted result key received from the TEE and obtain the plaintext result key; S47: The data requester downloads the computation result data encrypted with the result key from the TEE environment; S48: The data requester uses the obtained result key to decrypt the encrypted calculation result; S49: Decryption is complete, and the data requester obtains the final plaintext calculation result.

Images