Implementation method and system for privacy calculation of unbalanced architecture
By distributing tasks and cryptographic computing in the privacy computing center, the problem of resource waste and expansion in the existing technology is solved, unified control of computing resources and data security is realized, and the participation threshold and cost of privacy computing is reduced.
Patent Information
- Application Number
- CN202411069967.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-06
- Publication Date
- 2024-11-08
AI Technical Summary
The existing privacy computing system requires both or multiple parties to deploy the same server resources, resulting in waste of resources and difficulty in expanding capacity, and the inability to achieve unified centralized control of computing resources.
Adopting an unbalanced architecture, the privacy computing center receives encrypted data from multiple parties, performs task distribution and cryptographic calculations, realizes dynamic expansion and scaling of resources, and ensures data security through homomorphic/semi-homomorphic encryption algorithms.
It reduces the resource requirements and costs of privacy computing participants, realizes unified management and control of computing resources and data security, and avoids the synchronization problem of resource expansion in traditional systems.
Smart Images

Figure CN118916893A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to a privacy computing implementation method and system for an unbalanced architecture. Background Art
[0002] In the information age, the rapid growth and widespread application of data have brought about huge computing needs. However, this is accompanied by increasing attention to privacy and security. Traditional computing models often involve centralized processing, which may lead to the risk of leakage of users' sensitive information during the computing process. To solve this problem, privacy computing came into being, and its evolution process has demonstrated the unremitting pursuit of privacy protection at the intersection of past and present.
[0003] Privacy computing technology is a type of computing method and technology that aims to protect individual privacy during computing. Its goal is to minimize the exposure of user sensitive information while meeting computing needs. Traditional computing models usually require centralized processing, which can easily lead to the risk of personal privacy leakage. Privacy computing technology uses different computing methods to ensure that the user's sensitive information is protected to the greatest extent during the computing process.
[0004] Existing privacy computing systems basically adopt a symmetric computing model, that is, the computing and processing pressure of data is evenly distributed between two or more parties. Therefore, to complete the privacy computing task, both or more parties need to deploy the same privacy computing system (or heterogeneous privacy computing systems that can be interconnected) and purchase the same server resource configuration. However, the privacy computing system solutions commonly used in the industry currently consume a lot of computing resources, and the server configuration facilities are relatively high. In some scenarios, insufficient customer resources may cause business problems.
[0005] Therefore, in general, the existing technology has the following problems: 1) When implementing business scenarios through the privacy computing system, both or multiple parties need to rely on the same server configuration resources, which results in a certain waste of resources or the party with insufficient resources cannot continue to promote business implementation. 2) When the amount of computing tasks increases and resource expansion is required, both or multiple parties must simultaneously perform computing expansion resources to complete the task. 3) Computing resources are peer-to-peer and are independently controlled by each party, and cannot be centrally controlled.
[0006] In response to the above technical problems, the present invention proposes a privacy computing implementation method and system with an unbalanced architecture. Summary of the invention
[0007] The purpose of the present invention is to address the defects of the prior art and provide a privacy computing implementation method and system with an unbalanced architecture.
[0008] In order to achieve the above purpose, the present invention adopts the following technical solutions:
[0009] A privacy computing implementation method of a non-equilibrium architecture, comprising:
[0010] S1. The privacy computing center receives the privacy computing data encrypted by multiple privacy computing nodes;
[0011] S2. The privacy computing center creates computing tasks based on the received data, distributes the tasks evenly according to the computing tasks, and performs confidential computing on the tasks to obtain confidential results;
[0012] S3. The privacy computing center summarizes the confidential results;
[0013] S4. The privacy computing center synchronizes the aggregated confidential results to the corresponding multi-party privacy computing nodes, and the multi-party privacy computing nodes decrypt the obtained confidential results to obtain the required plaintext data.
[0014] Furthermore, the step S2 also includes calculating the required resources according to the created computing task, and judging whether the calculated resources are sufficient. If so, the tasks are evenly distributed and the task dense calculation is performed; if not, resource expansion processing is performed.
[0015] Furthermore, before step S1, the step further includes:
[0016] The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
[0017] Furthermore, the step S2 of evenly distributing the tasks and performing data secret state calculation specifically includes: evenly distributing the tasks to the computing cluster, and the computing cluster calculating the data secret state corresponding to the tasks.
[0018] Furthermore, the step S3 also includes reporting the obtained confidential result.
[0019] Accordingly, a privacy computing implementation system with a non-equilibrium architecture is also provided, including:
[0020] A receiving module, used for the privacy computing center to receive the privacy computing data encrypted by multiple privacy computing nodes;
[0021] A creation module is used by the privacy computing center to create computing tasks based on the received data, distribute the tasks evenly according to the computing tasks, and perform confidential computing on the tasks to obtain confidential results;
[0022] The aggregation module is used by the privacy computing center to aggregate confidential results;
[0023] The decryption module is used for the privacy computing center to synchronize the aggregated confidential results to the corresponding multi-party privacy computing nodes. The multi-party privacy computing nodes decrypt the obtained confidential results respectively to obtain the required plaintext data.
[0024] Furthermore, the creation module also includes calculating the required resources according to the created computing tasks, and judging whether the calculated resources are sufficient. If so, the tasks are evenly distributed and the task is calculated densely; if not, resource expansion is performed.
[0025] Furthermore, it also includes:
[0026] The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
[0027] Furthermore, the creation module distributes the tasks evenly and performs data secret state calculation specifically as follows: distributes the tasks evenly to the computing cluster, and the computing cluster calculates the data secret state corresponding to the tasks.
[0028] Furthermore, the summary module also includes reporting the obtained confidential results.
[0029] Compared with the prior art, the present invention has the following beneficial effects:
[0030] 1. By transferring the computing power of privacy computing to the computing center, privacy computing participants only need fixed lightweight resources to achieve the computing goals of large-scale privacy computing tasks, thereby reducing the participation threshold and cost of privacy computing.
[0031] 2. The computing center can dynamically expand or shrink computing resources based on the evaluation of computing tasks, thus solving the traditional problem of simultaneous expansion or shrinkage of all participants and achieving unified management and control of computing resources.
[0032] 3. When the computing power of the participants is transferred to a third party, the data security and privacy can still be guaranteed through data encryption and encryption algorithms such as homomorphic / semi-homomorphic encryption, and data confidential computing can be achieved. The central party cannot crack the encrypted data. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 This is a flow chart of a privacy computing implementation method of a non-balanced architecture provided in Example 1;
[0034] Figure 2 It is a flow chart of the overall architecture design provided in Example 1;
[0035] Figure 3 It is a flowchart of the overall architecture design of the specific example provided in Example 2. DETAILED DESCRIPTION
[0036] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.
[0037] The purpose of the present invention is to address the defects of the prior art and provide a privacy computing implementation method and system with an unbalanced architecture.
[0038] Embodiment 1
[0039] This embodiment provides a privacy computing implementation method of an unbalanced architecture, such as Figure 1 As shown, including:
[0040] S1. The privacy computing center receives the privacy computing data encrypted by multiple privacy computing nodes;
[0041] S2. The privacy computing center creates computing tasks based on the received data, distributes the tasks evenly according to the computing tasks, and performs confidential computing on the tasks to obtain confidential results;
[0042] S3. The privacy computing center summarizes the confidential results;
[0043] S4. The privacy computing center synchronizes the aggregated confidential results to the corresponding multi-party privacy computing nodes, and the multi-party privacy computing nodes decrypt the obtained confidential results to obtain the required plaintext data.
[0044] Under the unbalanced architecture of this embodiment, in addition to the two or more parties involved, a new computing center role is added. The two or more parties involved encrypt and calculate the local data and upload it to the central computing party, which completes the resource-consuming confidential computing tasks. The central computing party has a set of computing resource clusters, which centralizes the computing resources that originally required simultaneous management and control by multiple parties to the central party for management and control, and dynamically expands and shrinks the capacity according to the scale of the computing tasks.
[0045] The overall architecture process design of this embodiment is as follows Figure 2 As shown, two privacy computing nodes and one privacy computing center are taken as an example. The two privacy computing nodes include privacy computing node A and privacy computing node B, and the privacy computing center is privacy computing center C.
[0046] Privacy computing node A: One of the participants in the privacy computing business, responsible for reading local data and encrypting and uploading it to the center for confidential computing to ensure data security and privacy.
[0047] Privacy computing node B: One of the participants in the privacy computing business, responsible for reading local data and encrypting and uploading it to the center for confidential computing to ensure data security and privacy.
[0048] Privacy Computing Center C: The central computing party is responsible for providing computing resources, coordinating multiple nodes and scheduling the execution of confidential computing tasks, and dynamically expanding and shrinking computing resources according to computing tasks.
[0049] Before step S1, the method further includes:
[0050] S0. The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
[0051] Privacy computing node A and privacy computing node B each prepare the data needed to participate in privacy computing, read the data from the local data warehouse, and load it into the privacy computing system. Then, the read local data is encrypted using homomorphic encryption, and finally the encrypted data is uploaded to the privacy computing center C for secret computing to ensure data security and privacy.
[0052] The unbalanced architecture of this embodiment is that the computing resources of privacy computing nodes A and privacy computing nodes B are unbalanced with the central node C, and the computing resources are concentrated on the central node C. Privacy computing nodes A and privacy computing nodes B only need to retain a small amount of resources that support encryption and decryption, which shows the imbalance of data server resources.
[0053] In step S1, the privacy computing center receives the privacy computing data encrypted by multiple privacy computing nodes.
[0054] Privacy computing center C receives the encrypted data sent uniformly by privacy computing node A and privacy computing node B after encrypting the local data.
[0055] In step S2, the privacy computing center creates computing tasks based on the received data, distributes the tasks evenly according to the computing tasks, and performs confidential calculations on the tasks to obtain confidential results.
[0056] After receiving the multi-party computing data, the privacy computing center C first creates the current batch of computing tasks and analyzes the resources required for the current task calculation. If the resources are insufficient, the resources are expanded first. If the resources are sufficient, the tasks are evenly distributed to the computing cluster for data secret computing using the homomorphic encryption algorithm.
[0057] The creation process of the current batch of computing tasks is as follows: there is a job table in the central node, and creating a computing task is to create a new record in the job table; the computing task can be understood as, for example, calculating what 1+1 equals, but this calculation is calculated in a secret state using a homomorphic encryption algorithm, and this process is managed through a task maintenance, which is convenient for recording and tracking the status and progress of the task.
[0058] Analyze the current task specifically: For example, evaluate the required resources based on the amount of data currently involved in the computing task and the type of computing algorithm. Internal testing can be done in advance to roughly determine the computing resources required for different data volume levels. For example, if 8C16G is required for 1 million data volumes, then 16C32G is required for a task with 2 million requested data volumes.
[0059] Calculate the current resources specifically: For example, evaluate the required resources based on the amount of data currently involved in the computing task and the type of computing algorithm. Internal testing can be done in advance to roughly determine the computing resources required for different data volume levels. For example, if 8C16G is required for 1 million data volumes, then 16C32G is required for a task with 2 million data volumes.
[0060] The way to expand resources is that the Privacy Computing Center C service is deployed through the K8S cluster, which naturally supports the scaling of computing resources.
[0061] The task balanced distribution involved in this embodiment is that when the computing tasks of privacy computing node A and privacy computing node B are dispatched to privacy computing center C, there is a set of computing resources on privacy computing center C. In order to achieve parallel scheduling, the computing tasks are load balanced on the computing service of the central node, which belongs to load balancing.
[0062] In step S3, the privacy computing center summarizes the confidential results.
[0063] After the distributed confidential computing task is completed, the confidential result of the task will be reported to the machine where the general scheduling service is located. First, the center receives the task when a service receives the request, and then breaks the request into N computing tasks and distributes them to N computing servers. After each server completes the calculation, it reports the calculation result by requesting the reporting interface of a certain service. Then a certain service summarizes all the calculation results as the execution result of this job, among which the data of the privacy computing node A and the privacy computing node B are highlighted.
[0064] In step S4, the privacy computing center synchronizes the aggregated secret results to the corresponding multi-party privacy computing nodes, and the multi-party privacy computing nodes decrypt the obtained secret results to obtain the required plaintext data.
[0065] The aggregated and processed secret results are synchronized to the relevant privacy computing nodes A and B respectively, and then each privacy computing node decrypts the secret results through the homomorphic encryption and decryption method to obtain the required plaintext data. The local result will be simply analyzed and processed before it can be output for application.
[0066] In this embodiment, the confidential results synchronized to the relevant privacy computing nodes A and B are different. The returned data can only be decrypted by privacy computing node A or privacy computing node B, and the other party's data cannot be seen.
[0067] This embodiment has the following beneficial effects:
[0068] 1. By transferring the computing power of privacy computing to the computing center, privacy computing participants only need fixed lightweight resources to achieve the computing goals of large-scale privacy computing tasks, thereby reducing the participation threshold and cost of privacy computing.
[0069] 2. The computing center can dynamically expand or shrink computing resources based on the evaluation of computing tasks, thus solving the traditional problem of simultaneous expansion or shrinkage of all participants and achieving unified management and control of computing resources.
[0070] 3. When the computing power of the participants is transferred to a third party, the data security and privacy can still be guaranteed through data encryption and encryption algorithms such as homomorphic / semi-homomorphic encryption, and data confidential computing can be achieved. The central party cannot crack the encrypted data.
[0071] Embodiment 2
[0072] The privacy computing implementation method of the non-balanced architecture provided in this embodiment is different from that in the first embodiment in that:
[0073] like Figure 3 As shown, this embodiment is described with specific data.
[0074] Party A reads the processed data ga locally, generates a random number ra, calculates ga+ra, and encrypts ga+ra to obtain [ga+ra], and sends the encrypted [ga+ra] and random number ra to the privacy computing center C.
[0075] Party B reads the processed data gb locally, generates a random number rb, calculates gb+rb, and encrypts gb+rb to obtain [gb+rb], and sends the encrypted [gb+rb] and the random number rb to the privacy computing center C.
[0076] The privacy computing center C creates a computing task based on the data sent by participants A and B. The computing task includes ciphertext calculation on participant A’s data: [ga+ra]·[ra]=[ga] and ciphertext calculation on participant B’s data: [gb+rb]·[rb]=[gb].
[0077] The privacy computing center C will encrypt and decrypt the calculated [ga] and [gb] to obtain the encrypted results [Ga] and [Gb].
[0078] The privacy computing center C generates a random number rc and performs encryption calculation on [Ga] and [Gb] to obtain [Ga+rc] and [Gb+rc].
[0079] The privacy computing center C sends the obtained [Ga+rc] and rc to participant A, and sends the obtained [Gb+rc] and rc to participant B.
[0080] Party A decrypts [Ga+rc] based on the obtained [Ga+rc] and rc to obtain Ga+rc, and calculates Ga=Ga+rc-rc, and finally outputs Ga.
[0081] Party B decrypts [Gb+rc] based on the obtained [Gb+rc] and rc to obtain Gb+rc, and calculates Gb=Gb+rc-rc, and finally outputs Gb.
[0082] Embodiment 3
[0083] This embodiment provides a privacy computing implementation system with a non-balanced architecture, including:
[0084] A receiving module, used for the privacy computing center to receive the privacy computing data encrypted by multiple privacy computing nodes;
[0085] A creation module is used by the privacy computing center to create computing tasks based on the received data, distribute the tasks evenly according to the computing tasks, and perform confidential computing on the tasks to obtain confidential results;
[0086] The aggregation module is used by the privacy computing center to aggregate confidential results;
[0087] The decryption module is used for the privacy computing center to synchronize the aggregated confidential results to the corresponding multi-party privacy computing nodes. The multi-party privacy computing nodes decrypt the obtained confidential results respectively to obtain the required plaintext data.
[0088] Furthermore, the creation module also includes calculating the required resources according to the created computing tasks, and judging whether the calculated resources are sufficient. If so, the tasks are evenly distributed and the task is calculated densely; if not, resource expansion is performed.
[0089] Furthermore, it also includes:
[0090] The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
[0091] Furthermore, the creation module distributes the tasks evenly and performs data secret state calculation specifically as follows: distributes the tasks evenly to the computing cluster, and the computing cluster calculates the data secret state corresponding to the tasks.
[0092] Furthermore, it also includes reporting the obtained confidential results.
[0093] It should be noted that the privacy computing implementation system with a non-balanced architecture provided in this embodiment is similar to that in Embodiment 1 and will not be elaborated here.
[0094] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims
1. A privacy computing implementation method based on a non-equilibrium architecture. It is characterized in that include: S1. The privacy computing center receives the privacy computing data encrypted by multiple privacy computing nodes; S2. The privacy computing center creates computing tasks based on the received data, distributes the tasks evenly according to the computing tasks, and performs confidential computing on the tasks to obtain confidential results; S3. The privacy computing center summarizes the confidential results; S4. The privacy computing center synchronizes the aggregated confidential results to the corresponding multi-party privacy computing nodes, and the multi-party privacy computing nodes decrypt the obtained confidential results to obtain the required plaintext data.
2. According to the privacy computing implementation method of the non-balanced architecture of claim 1, It is characterized in that The step S2 also includes calculating the required resources according to the created computing task, and judging whether the calculated resources are sufficient. If so, the tasks are evenly distributed and the task dense calculation is performed; if not, resource expansion is performed.
3. According to the privacy computing implementation method of the non-balanced architecture of claim 1, It is characterized in that Before step S1, the following steps are also included: The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
4. According to the privacy computing implementation method of the non-balanced architecture of claim 1, It is characterized in that The step S2 of evenly distributing the tasks and performing data secret state calculation specifically includes: evenly distributing the tasks to the computing cluster, and the computing cluster calculating the data secret state corresponding to the tasks.
5. According to the privacy computing implementation method of a non-balanced architecture as described in claim 1, It is characterized in that The step S3 also includes reporting the obtained confidential result.
6. A privacy computing implementation system with a non-equilibrium architecture, It is characterized in that include: A receiving module, used for the privacy computing center to receive the privacy computing data encrypted by multiple privacy computing nodes; A creation module is used by the privacy computing center to create computing tasks based on the received data, distribute the tasks evenly according to the computing tasks, and perform confidential computing on the tasks to obtain confidential results; The aggregation module is used by the privacy computing center to aggregate confidential results; The decryption module is used for the privacy computing center to synchronize the aggregated confidential results to the corresponding multi-party privacy computing nodes. The multi-party privacy computing nodes decrypt the obtained confidential results respectively to obtain the required plaintext data.
7. According to claim 6, a privacy computing implementation system with a non-balanced architecture, It is characterized in that The creation module also includes calculating the required resources according to the created computing tasks, and judging whether the calculated resources are sufficient. If so, the tasks are evenly distributed and the task dense calculation is performed; if not, resource expansion processing is performed.
8. According to claim 6, a privacy computing implementation system with a non-balanced architecture, It is characterized in that Also includes: The multi-party privacy computing node reads the local data that needs to participate in the privacy computing and encrypts the data that needs to participate in the privacy computing.
9. A privacy computing implementation system with a non-balanced architecture according to claim 6, It is characterized in that The creation module distributes the tasks evenly and performs data density calculation specifically as follows: distributes the tasks evenly to the computing cluster, and the computing cluster calculates the data density corresponding to the tasks.
10. A privacy computing implementation system with a non-balanced architecture according to claim 6, It is characterized in that The summary module also includes reporting the obtained confidential results.
Citation Information
Patent Citations
Task allocation method and system in node cluster, and node
CN107135257A
Load balancing method of server cluster
CN108551474A
Multi-participant resource processing method and privacy computing system
CN113901507A
Multi-party computing control method, device and equipment for providing privacy protection
CN114817982A
Method and device for imbalance privacy set intersection
CN115733602A
