Distributed data security collaborative computing platform system oriented to privacy computing
By constructing a distributed data security collaborative computing platform system, adaptive secure collaborative computing throughout the entire lifecycle is realized, solving the problem that security and efficiency are difficult to adaptively adjust in large-scale distributed collaborative computing, and forming an intelligent collaborative system that is self-aware, self-evaluating, and self-adjusting.
Patent Information
- Application Number
- CN202511326064.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-17
- Publication Date
- 2025-11-18
AI Technical Summary
Existing privacy computing technologies lack organic integration in large-scale distributed collaborative computing scenarios, making it difficult to cope with complex and ever-changing security environments. The security status of computing nodes and network threats are not considered in a unified manner, making it difficult to adaptively adjust security and efficiency, and increasing the complexity of operation and management.
We will build a distributed data security collaborative computing platform system for privacy-preserving computing. Through deep collaboration of modules such as node security data collection, network known dangerous data collection, feature vector privacy processing, threat detection, privacy computing, and multi-dimensional security management, we will achieve adaptive security collaborative computing throughout the entire lifecycle.
It achieves adaptive secure collaborative computing throughout the entire lifecycle, accurately identifies risks and triggers differentiated responses, ensures data privacy and computing efficiency, and forms an intelligent collaborative system that is self-aware, self-evaluating, and self-adjusting, thus solving the problems of privacy leakage, low computing efficiency, and lack of global security control.
Smart Images
Figure HDA0005599746500000011 
Figure HDA0005599746500000021
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more specifically, to a distributed data security collaborative computing platform system for privacy-oriented computing. Background Technology
[0002] With the deepening of digital transformation, distributed computing has become the mainstream paradigm for unlocking the value of data. Against this backdrop, privacy-preserving computing technologies such as federated learning, secure multi-party computation, and homomorphic encryption have emerged, providing crucial technical support for cross-institutional data collaboration and aiming to realize the vision of "data remaining within its domain while value flows freely."
[0003] Federated learning effectively reduces the risk of privacy breaches by exchanging model parameters instead of raw data; secure multi-party computation enables multiple participants to perform collaborative computations without disclosing their respective inputs; and homomorphic encryption allows direct execution of operations in ciphertext, further enhancing data protection capabilities. The individual application of these technologies has achieved significant results in specific scenarios. However, in actual large-scale distributed collaborative computing scenarios, these technologies are often deployed independently, lacking organic integration, and are difficult to cope with complex and ever-changing security environments. The security status of the computing nodes themselves, known network threats, and dynamic risks during the collaborative process have not been included in a unified consideration, resulting in limitations of existing solutions in dealing with global security threats. At the same time, the system lacks transparent control and dynamic optimization capabilities for the entire computing lifecycle, making it difficult to adaptively adjust the balance between efficiency and security. This, to some extent, restricts the maximization of collaborative computing efficiency and increases the complexity of operation and management.
[0004] In view of this, the present invention proposes a distributed data security collaborative computing platform system for privacy computing to solve the above problems. Summary of the Invention
[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution, including:
[0006] The node security acquisition module is used to acquire node security datasets; the node security datasets include CPU utilization and memory utilization.
[0007] The network known danger data acquisition module is used to collect a dataset of known network dangers; the dataset includes malicious IP addresses, malicious file hash values, and abnormal network traffic patterns.
[0008] The feature vector privacy processing module is used to preprocess the node security dataset and the dangerous dataset to obtain a joint feature vector;
[0009] Furthermore, the specific steps for preprocessing based on the node security dataset and the dangerous dataset to obtain the joint feature vector include:
[0010] C1: The basic dataset is cleaned by removing outliers, and all sub-data items in the basic dataset are normalized to the range [0, 1] according to the normalization formula;
[0011] C2: Standardize the CPU utilization and memory utilization to obtain a numerical feature vector;
[0012] C3: Convert malicious IP addresses into binary signature vectors;
[0013] C4: Convert the hash value of a malicious file into a hash feature vector;
[0014] C5: Convert abnormal network traffic patterns into traffic feature vectors;
[0015] C6: Pack the numerical feature vector, binary feature vector, hash feature vector and traffic feature vector to obtain the joint feature vector, and send the joint feature vector to the threat detection module and privacy computing module;
[0016] The threat detection module is used to analyze joint feature vectors and malicious node information based on machine learning models, detect high-risk data and behaviors, and execute proactive defense measures.
[0017] Furthermore, the steps for detecting high-risk data and behaviors and implementing proactive defense measures based on machine learning models to analyze joint feature vectors and malicious node information include:
[0018] B1: Based on the joint feature vector, the instantaneous threat score of a single node is calculated using a machine learning model;
[0019] B2: Based on the instantaneous threat score, compare it with the preset threshold range [P1, P2] to determine the current threat level; when the instantaneous threat score is less than P1, the node is determined to be low-risk, and the node is monitored normally with a low-risk flag output; when the instantaneous threat score is greater than or equal to P1 and less than P2, the node is determined to be medium-risk, and a medium-risk flag output; when the instantaneous threat score is greater than or equal to P2, the node is determined to be high-risk, the node is isolated from the computer network, and all its unprocessed data is transferred to a sandbox for in-depth analysis, while a high-risk flag is output.
[0020] B3: Package low-risk, medium-risk, and high-risk labels to obtain the threat level, and send the threat to the distributed collaborative optimization module and the visual management module;
[0021] The privacy computing module is used to achieve secure collaborative computing among distributed nodes while protecting the original data from leakage, based on the integration of federated learning, secure multi-party computation and lightweight homomorphic encryption technology.
[0022] Furthermore, by integrating federated learning, secure multi-party computation, and lightweight homomorphic encryption technologies, the steps to achieve secure collaborative computation among distributed nodes while protecting the original data from leakage include:
[0023] D1: Each edge node trains a local model using local data until the local model reaches the required confidence level;
[0024] D2: The node sends the encrypted model parameters to the aggregation server. The server uses the SMPC protocol to calculate the global model, ensuring that the original data is not leaked.
[0025] D3: Lightweight homomorphic encryption is allowed for encrypted computation on sensitive data to ensure data privacy;
[0026] D4: Package the trained model parameters and joint statistics to obtain privacy computation results, and send the privacy computation results to the distributed collaborative optimization module and the multi-dimensional security management module;
[0027] The multi-dimensional security management module is used to quantitatively assess the security risks of distributed nodes and monitor data flow based on privacy computing results.
[0028] Furthermore, based on the results of privacy-preserving computation, the steps for quantitatively assessing the security risks of distributed nodes and monitoring data flow include:
[0029] M1: The attack success rate is obtained by calculating parameters through simulated attack tests and inverting the attack success rate.
[0030] M2: Based on step M1 and the data encryption strength, the probability of node privacy leakage is calculated;
[0031] M3: Calculates a node credit score based on the probability of node privacy leakage and the threat level;
[0032] M4: Based on the node credit score in step M3, set a scoring threshold range [L1, L2]; when the node credit score is greater than or equal to L2, the node is determined to be a trusted node; when the node credit score is greater than or equal to L1 and less than L2, the node is determined to be a suspicious node, and the monitoring frequency for the node is increased; when the node credit score is less than L1, the node is determined to be a malicious node, and the complete violation log is stored on the blockchain as evidence.
[0033] M5: Sends malicious node information to the threat detection module, and simultaneously packages the privacy leakage probability and node credit score, and sends them to the distributed collaborative optimization module and the visualization management module;
[0034] The distributed collaborative optimization module is used to dynamically optimize the collaborative scheme based on threat level and node credit score;
[0035] Furthermore, based on threat level and node credit score, the steps for dynamically optimizing the collaboration scheme include:
[0036] E1: When a node has a low threat level and is a trusted node, the system is considered to be in a safe state. The node participates in collaborative computing and uses all of the node's original data for local model training.
[0037] E2: When a node is identified as low-risk and is a suspicious node, or a node is identified as medium-risk and is a trusted node, the system is deemed to have a minor anomaly. Data from medium-risk or suspicious nodes is sampled and verified, and an enhanced encryption channel is enabled.
[0038] E3: When a node is identified as having a medium threat level and is considered a suspicious node, the system is deemed to have a risk of privacy leakage. Only trusted nodes are allowed to participate in core computing, and all computing data from participating nodes is encrypted using lightweight homomorphic encryption. In addition, all model parameter exchanges must be conducted through the SMPC protocol to ensure the privacy of the exchanged parameters.
[0039] E4: When a node is a malicious node, the system is determined to be in a high-risk privacy leakage environment. The global collaborative computing task is suspended and unnecessary connections between the server and the external network are cut off. At the same time, the system waits for the intervention of the administrator.
[0040] The visualization and control module is used to provide managers with global situational awareness and real-time control capabilities through a visualization panel that integrates real-time status monitoring and multi-level risk alarms.
[0041] Furthermore, the steps to provide managers with overall situational awareness and immediate control capabilities through a visual dashboard that integrates real-time status monitoring and multi-level risk alerts include:
[0042] F1: When a risk event is detected, a graded alarm is triggered according to the threat level; when it is a medium-risk marker or a suspicious node, an orange warning icon is displayed and an alarm sound is emitted; when it is a high-risk marker or a malicious node, a red warning icon is displayed and a modal alarm pops up.
[0043] F2: Sends visualized data and tiered alarm information to the data management module;
[0044] The data management module is used to store system basic data and send hierarchical alarm information to security management personnel;
[0045] Furthermore, the steps of storing system-based basic data and sending tiered alarm information to security management personnel include:
[0046] Tiered alarm information is sent to the receiving device of safety management personnel via email, and staff are reminded to check their emails via SMS.
[0047] The technical effects and advantages of the distributed data security collaborative computing platform system for privacy-preserving computing of this invention are as follows:
[0048] This invention, by constructing an organic whole with deep collaboration and intelligent linkage among its modules, completely changes the traditional fragmented situation of computing, security, and control in distributed systems. It achieves full lifecycle, adaptive secure collaborative computing. Its core advantage does not stem from technological breakthroughs in a single module, but rather from the system-level gains generated by the ingenious process design and seamless interaction between modules. Specifically, the system first constructs a three-dimensional data perception system covering both intrinsic state and external threats through node security data acquisition modules and network known danger data acquisition modules. This raw data is not used directly, but is standardized and feature extracted by the feature vector privacy processing module, transforming it into feature vectors that both represent the security situation and protect original privacy. This lays a unified and secure data foundation for subsequent intelligent analysis. The threat detection module, as one of the nerve centers of the system, uses these features for real-time analysis and innovatively adopts a hierarchical threat identification mechanism. The value of this mechanism lies not only in accurately identifying risks, but also in transforming abstract threats into different levels of operational languages that can be understood and executed by other modules, thereby triggering differentiated and precise response processes. More importantly, this threat intelligence and response strategy do not exist in isolation, but are deeply integrated with the core computing process. The privacy computing module, as the core of the computing process, integrates federated learning, secure multi-party computation, and homomorphic encryption to ensure that data is "usable but not visible" during the computing process. Its intermediate output is immediately captured by the distributed collaborative optimization module, which acts as the system's intelligent scheduler. It dynamically senses the privacy leakage risks quantified by the multi-dimensional security management module and the threat level from the threat detection module, and adjusts the collaborative strategy in real time based on this.
[0049] In summary, this invention successfully integrates proactive defense, privacy-preserving computing, and dynamic resource optimization through efficient and intelligent interaction processes between modules, forming an intelligent collaborative system capable of self-awareness, self-evaluation, self-adjustment, and self-evolution. Ultimately, it effectively solves core challenges in distributed collaborative computing such as privacy leaks, low computing efficiency, data silos, and lack of global security control, providing a solid foundation for building a trustworthy, efficient, and resilient data collaborative ecosystem. Attached Figure Description
[0050] Figure 1 This is a schematic diagram of the distributed data security collaborative computing platform system for privacy-oriented computing according to the present invention;
[0051] Figure 2 This is a schematic diagram of the distributed data security collaborative computing method for privacy-oriented computing according to the present invention. Detailed Implementation
[0052] 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.
[0053] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in the embodiments of this invention are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.
[0054] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”
[0055] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.
[0056] In practice, the server-side equipment deployed in a privacy-oriented distributed data security collaborative computing platform system may consist of one or more devices. This privacy-oriented distributed data security collaborative computing platform system can be implemented as: a business instance, a virtual machine, or hardware devices. For example, this privacy-oriented distributed data security collaborative computing platform system can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, this privacy-oriented distributed data security collaborative computing platform system can be understood as software deployed on a cloud node, used to provide privacy-oriented distributed data security collaborative computing platform system to various user terminals. Alternatively, this privacy-oriented distributed data security collaborative computing platform system can also be implemented as a virtual machine deployed on one or more devices in a cloud node. This virtual machine contains application software for managing various user terminals. Alternatively, this privacy-oriented distributed data security collaborative computing platform system can also be implemented as a server composed of numerous identical or different types of hardware devices, with one or more hardware devices configured to provide privacy-oriented distributed data security collaborative computing platform system to various user terminals.
[0057] In terms of implementation, the distributed data security collaborative computing platform system for privacy-oriented computing and the user client are mutually compatible. That is, if the distributed data security collaborative computing platform system for privacy-oriented computing is implemented as an application installed on a cloud service platform, then the user client is implemented as a client that establishes a communication connection with the application; or if the distributed data security collaborative computing platform system for privacy-oriented computing is implemented as a website, then the user client is implemented as a webpage; or if the distributed data security collaborative computing platform system for privacy-oriented computing is implemented as a cloud service platform, then the user client is implemented as a mini-program in an instant messaging application.
[0058] like Figure 1 The figure shown is a system architecture diagram of a distributed data security collaborative computing platform system for privacy computing provided in an embodiment of the present invention.
[0059] The distributed data security collaborative computing platform system for privacy-oriented computing described in this invention can be located on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed in the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the distributed data security collaborative computing platform system for privacy-oriented computing may include a node security data acquisition module, a network known danger data acquisition module, a feature vector privacy processing module, a threat detection module, a privacy computing module, a distributed collaborative optimization module, a multi-dimensional security management module, a visualization control module, and a data management module. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.
[0060] In this embodiment of the invention, in the distributed data security collaborative computing platform system for privacy computing, each of the above modules can be implemented independently and can call other modules. Here, "calling" can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. For example, the sharing and evaluation module can call the same information collection module to obtain the information collected by that module. Based on the above characteristics, in the distributed data security collaborative computing platform system for privacy computing provided in this embodiment of the invention, without modifying the program code, the applicability of the distributed data security collaborative computing platform system architecture can be adjusted by adding modules and directly calling them, achieving cluster-based horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the distributed data security collaborative computing platform system for privacy computing. In practical applications, the above modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in a cloud server.
[0061] Example 1
[0062] Please see Figure 1 As shown in this embodiment, the distributed data security collaborative computing platform system for privacy computing includes:
[0063] The node security acquisition module is used to acquire node security datasets; the node security datasets include CPU utilization and memory utilization.
[0064] Further steps in collecting node security datasets include:
[0065] A1: By reading the system monitoring file of the specified node, obtain the idle time and total time of the CPU of the specified node, and calculate the CPU utilization based on the idle time and total time of the CPU;
[0066] A2: By reading the system monitoring file of the specified node, obtain the memory capacity and used memory capacity of the specified node, and calculate the memory utilization rate based on the total memory capacity and used memory capacity;
[0067] The network known danger data acquisition module is used to collect a dataset of known network dangers; the dataset includes malicious IP addresses, malicious file hash values, and abnormal network traffic patterns.
[0068] Furthermore, the steps for collecting known dangerous datasets from the internet include:
[0069] B1: By reading the threat intelligence source database, a list of malicious IPs is obtained, and the malicious IP addresses are obtained;
[0070] B2: By reading the threat database, malicious file data is obtained, and the hash value of the malicious file is obtained;
[0071] B3: Obtain abnormal traffic characteristics and abnormal network traffic patterns through network monitoring tools;
[0072] The feature vector privacy processing module is used to preprocess the node security dataset and the dangerous dataset to obtain a joint feature vector;
[0073] Furthermore, the specific steps for preprocessing based on the node security dataset and the dangerous dataset to obtain the joint feature vector include:
[0074] C1: The basic dataset is cleaned by removing outliers, and all sub-data items in the basic dataset are normalized to the range [0, 1] according to the normalization formula;
[0075] It should be explained that removing outliers refers to, for example, negative CPU utilization; the base dataset includes a safe node dataset and a dangerous dataset; the specific expression of the normalization formula is: Where X new For normalized values, X is any sub-data item of the basic data. max X represents the historical maximum value of this arbitrary sub-data item. min The historical minimum of this arbitrary sub-data item;
[0076] C2: Standardize the CPU utilization and memory utilization to obtain a numerical feature vector;
[0077] C3: Convert malicious IP addresses into binary signature vectors;
[0078] C4: Convert the hash value of a malicious file into a hash feature vector;
[0079] C5: Convert abnormal network traffic patterns into traffic feature vectors;
[0080] C6: Pack the numerical feature vector, binary feature vector, hash feature vector and traffic feature vector to obtain the joint feature vector, and send the joint feature vector to the threat detection module and privacy computing module;
[0081] The threat detection module is used to analyze joint feature vectors and malicious node information based on machine learning models, detect high-risk data and behaviors, and execute proactive defense measures.
[0082] Furthermore, the steps for analyzing joint feature vectors and malicious node information based on machine learning models to detect high-risk data and behaviors, and to implement proactive defense measures include:
[0083] B1: Based on the joint feature vector, the instantaneous threat score of a single node is calculated using a machine learning model;
[0084] B2: Based on the instantaneous threat score, compare it with the preset threshold range [P1, P2] to determine the current threat level; when the instantaneous threat score is less than P1, the node is determined to be low-risk, and the node is monitored normally with a low-risk flag output; when the instantaneous threat score is greater than or equal to P1 and less than P2, the node is determined to be medium-risk, and a medium-risk flag output; when the instantaneous threat score is greater than or equal to P2, the node is determined to be high-risk, the node is isolated from the computer network, and all its unprocessed data is transferred to a sandbox for in-depth analysis, while a high-risk flag is output.
[0085] It should be noted that the threshold range is manually set and entered;
[0086] B3: Package low-risk, medium-risk, and high-risk labels to obtain the threat level, and send the threat to the distributed collaborative optimization module and the visual management module;
[0087] The privacy computing module is used to achieve secure collaborative computing among distributed nodes while protecting the original data from leakage, based on the integration of federated learning, secure multi-party computation and lightweight homomorphic encryption technology.
[0088] Furthermore, by integrating federated learning, secure multi-party computation, and lightweight homomorphic encryption technologies, the steps to achieve secure collaborative computation among distributed nodes while protecting the original data from leakage include:
[0089] D1: Each edge node trains a local model using local data until the local model reaches the required confidence level;
[0090] D2: The node sends the encrypted model parameters to the aggregation server. The server uses the SMPC protocol to calculate the global model, ensuring that the original data is not leaked.
[0091] D3: Lightweight homomorphic encryption is allowed for encrypted computation on sensitive data to ensure data privacy;
[0092] D4: Package the trained model parameters and joint statistics to obtain privacy computation results, and send the privacy computation results to the distributed collaborative optimization module and the multi-dimensional security management module;
[0093] The multi-dimensional security management module is used to quantitatively assess the security risks of distributed nodes and monitor data flow based on privacy computing results.
[0094] Furthermore, based on the results of privacy-preserving computations, the steps for quantitatively assessing the security risks of distributed nodes and monitoring data flow include:
[0095] M1: The attack success rate is obtained by calculating parameters through simulated attack tests and inverting the attack success rate.
[0096] M2: Based on step M1 and the data encryption strength, the probability of node privacy leakage is calculated;
[0097] M3: Calculates a node credit score based on the probability of node privacy leakage and the threat level;
[0098] It should be noted that all the above calculated data are normalized data;
[0099] M4: Based on the node credit score in step M3, set a scoring threshold range [L1, L2]; when the node credit score is greater than or equal to L2, the node is determined to be a trusted node; when the node credit score is greater than or equal to L1 and less than L2, the node is determined to be a suspicious node, and the monitoring frequency for the node is increased; when the node credit score is less than L1, the node is determined to be a malicious node, and the complete violation log is stored on the blockchain as evidence.
[0100] M5: Sends malicious node information to the threat detection module, and simultaneously packages the privacy leakage probability and node credit score, and sends them to the distributed collaborative optimization module and the visualization management module;
[0101] The distributed collaborative optimization module is used to dynamically optimize the collaborative scheme based on threat level and node credit score;
[0102] Furthermore, based on threat level and node credit score, the steps for dynamically optimizing the collaboration scheme include:
[0103] E1: When a node has a low threat level and is a trusted node, the system is considered to be in a safe state. The node participates in collaborative computing and uses all of the node's original data for local model training.
[0104] E2: When a node is identified as low-risk and is a suspicious node, or a node is identified as medium-risk and is a trusted node, the system is deemed to have a minor anomaly. Data from medium-risk or suspicious nodes is sampled and verified, and an enhanced encryption channel is enabled.
[0105] E3: When a node is identified as having a medium threat level and is considered a suspicious node, the system is deemed to have a risk of privacy leakage. Only trusted nodes are allowed to participate in core computing, and all computing data from participating nodes is encrypted using lightweight homomorphic encryption. In addition, all model parameter exchanges must be conducted through the SMPC protocol to ensure the privacy of the exchanged parameters.
[0106] E4: When a node is a malicious node, the system is determined to be in a high-risk privacy leakage environment. The global collaborative computing task is suspended and unnecessary connections between the server and the external network are cut off. At the same time, the system waits for the intervention of the administrator.
[0107] The visualization and control module is used to provide managers with global situational awareness and real-time control capabilities through a visualization panel that integrates real-time status monitoring and multi-level risk alarms.
[0108] Furthermore, the steps to provide managers with comprehensive situational awareness and immediate control capabilities through a visual dashboard that integrates real-time status monitoring and multi-level risk alerts include:
[0109] F1: When a risk event is detected, a graded alarm is triggered according to the threat level; when it is a medium-risk marker or a suspicious node, an orange warning icon is displayed and an alarm sound is emitted; when it is a high-risk marker or a malicious node, a red warning icon is displayed and a modal alarm pops up.
[0110] F2: Sends visualized data and tiered alarm information to the data management module;
[0111] The data management module is used to store system basic data and send hierarchical alarm information to security management personnel;
[0112] Furthermore, the steps of storing system-based basic data and sending tiered alarm information to security management personnel include:
[0113] The tiered alarm information is sent to the receiving device of the safety management personnel via email, and staff are reminded to check their emails via SMS.
[0114] It should be explained that the system's basic data includes node security datasets, dangerous datasets, model parameters, joint statistics, node credit scores, and threat levels;
[0115] The beneficial effect of this embodiment lies in the fact that by constructing an organic whole with deep collaboration and intelligent linkage among modules, it completely changes the situation of the separation between computing, security and control in traditional distributed systems, and realizes full life cycle adaptive secure collaborative computing. Its core advantage does not stem from the technological breakthrough of a single module, but rather from the system-level gains generated by the ingenious process design and seamless interaction between modules. Specifically, the system first constructs a three-dimensional data perception system covering the internal state and external threats through the node security data acquisition module and the network known danger data acquisition module. This raw data is not used directly, but is standardized and feature extracted by the feature vector privacy processing module, and transformed into feature vectors that can both represent the security situation and protect the original privacy, laying a unified and secure data foundation for subsequent intelligent analysis. The threat detection module, as one of the nerve centers of the system, uses these features for real-time analysis and innovatively adopts a hierarchical threat identification mechanism. The value of this mechanism lies not only in accurately identifying risks, but also in transforming abstract threats into different levels of operational languages that can be understood and executed by other modules, thereby triggering differentiated and precise response processes. More importantly, this threat intelligence and response strategy does not exist in isolation, but is deeply integrated with the core computing process. The privacy computing module, as the core of the computing process, integrates federated learning, secure multi-party computation, and homomorphic encryption to ensure that data is "usable but not visible" during the computing process. Its intermediate output is immediately captured by the distributed collaborative optimization module, which acts as the system's intelligent scheduler. It dynamically senses the privacy leakage risks quantified by the multi-dimensional security management module and the threat level from the threat detection module, and adjusts the collaborative strategy in real time based on this.
[0116] In summary, this invention successfully integrates proactive defense, privacy-preserving computing, and dynamic resource optimization through efficient and intelligent interaction processes between modules, forming an intelligent collaborative system capable of self-awareness, self-evaluation, self-adjustment, and self-evolution. Ultimately, it effectively solves core challenges in distributed collaborative computing such as privacy leaks, low computing efficiency, data silos, and lack of global security control, providing a solid foundation for building a trustworthy, efficient, and resilient data collaborative ecosystem.
[0117] Example 2
[0118] Please see Figure 2 As shown, the parts not described in detail in this embodiment are described in Embodiment 1. A distributed data security collaborative computing method for privacy computing is provided. The method includes: S1: collecting secure datasets of nodes;
[0119] S2: Collect known dangerous datasets from the network;
[0120] S3: Preprocess the node security dataset and the dangerous dataset to obtain a joint feature vector;
[0121] S4: Analyze joint feature vectors and malicious node information based on machine learning models to detect high-risk data and behaviors, and implement proactive defense measures;
[0122] S5: Based on the integration of federated learning, secure multi-party computation and lightweight homomorphic encryption technology, it enables secure collaborative computation among distributed nodes while protecting the original data from leakage.
[0123] S6: Based on privacy-preserving computation results, quantitatively assess the security risks of distributed nodes and monitor data flow;
[0124] S7: Dynamically optimizes collaboration schemes based on threat level and node credit score;
[0125] S8: Provides managers with global situational awareness and real-time control capabilities through a visual panel that integrates real-time status monitoring and multi-level risk alarms;
[0126] S9: Stores system basic data and sends tiered alarm information to security management personnel.
[0127] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the present invention.
Claims
1. A distributed data security collaborative computing platform system for privacy-preserving computing, characterized in that, The system includes: a threat detection module, a privacy computing module, a multi-dimensional security management module, a distributed collaborative optimization module, and a visualization and control module, wherein: The threat detection module is used to analyze joint feature vectors and malicious node information based on machine learning models, detect high-risk data and behaviors, and execute proactive defense measures. The privacy computing module is used to achieve secure collaborative computing among distributed nodes while protecting the original data from leakage, based on the integration of federated learning, secure multi-party computation and lightweight homomorphic encryption technology. The multi-dimensional security management module is used to quantitatively assess the security risks of distributed nodes and monitor data flow based on privacy computing results. The distributed collaborative optimization module is used to dynamically optimize the collaborative scheme based on threat level and node credit score; The visualization and control module provides managers with global situational awareness and real-time control capabilities through a visualization panel that integrates real-time status monitoring and multi-level risk alarms.
2. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, The system also includes: a node security data acquisition module, a network known danger data acquisition module, and a data management module, wherein: The node security acquisition module is used to acquire node security datasets; the node security datasets include CPU utilization and memory utilization. The network known danger data acquisition module is used to collect a dataset of known network dangers; the dataset includes malicious IP addresses, malicious file hash values, and abnormal network traffic patterns. The feature vector privacy processing module is used to preprocess the node security dataset and the dangerous dataset to obtain a joint feature vector; The data management module is used to store system basic data and send hierarchical alarm information to security management personnel.
3. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 2, characterized in that, The specific steps for preprocessing the node security dataset and the dangerous dataset to obtain the joint feature vector include: C1: The basic dataset is cleaned by removing outliers, and all sub-data items in the basic dataset are normalized to the range [0, 1] according to the normalization formula; C2: Standardize the CPU utilization and memory utilization to obtain a numerical feature vector; C3: Convert malicious IP addresses into binary signature vectors; C4: Convert the hash value of a malicious file into a hash feature vector; C5: Convert abnormal network traffic patterns into traffic feature vectors; C6: Pack the numerical feature vector, binary feature vector, hash feature vector, and traffic feature vector to obtain a joint feature vector, and send the joint feature vector to the threat detection module and the privacy computing module.
4. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, The steps for detecting high-risk data and behaviors and implementing proactive defense measures based on machine learning models to analyze joint feature vectors and malicious node information include: B1: Based on the joint feature vector, the instantaneous threat score of a single node is calculated using a machine learning model; B2: Based on the instantaneous threat score, compare it with the preset threshold range [P1, P2] to determine the current threat level; when the instantaneous threat score is less than P1, the node is determined to be low-risk, and the node is monitored normally with a low-risk flag output; when the instantaneous threat score is greater than or equal to P1 and less than P2, the node is determined to be medium-risk, and a medium-risk flag output; when the instantaneous threat score is greater than or equal to P2, the node is determined to be high-risk, the node is isolated from the computer network, and all its unprocessed data is transferred to a sandbox for in-depth analysis, while a high-risk flag is output. B3: Package low-risk, medium-risk, and high-risk identifiers to obtain the threat level, and send the threat to the distributed collaborative optimization module and the visual management module.
5. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, By integrating federated learning, secure multi-party computation, and lightweight homomorphic encryption, the steps to achieve secure collaborative computation among distributed nodes while protecting the original data from leakage include: D1: Each edge node trains a local model using local data until the local model reaches the required confidence level; D2: The node sends the encrypted model parameters to the aggregation server. The server uses the SMPC protocol to calculate the global model, ensuring that the original data is not leaked. D3: Lightweight homomorphic encryption is allowed for encrypted computation on sensitive data to ensure data privacy; D4: Package the trained model parameters and joint statistics to obtain privacy computation results, and send the privacy computation results to the distributed collaborative optimization module and the multi-dimensional security management module.
6. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, Based on privacy-preserving computation results, the steps for quantitatively assessing the security risks of distributed nodes and monitoring data flow include: M1: The attack success rate is obtained by calculating parameters through simulated attack tests and inverting the attack success rate. M2: Based on step M1 and the data encryption strength, the probability of node privacy leakage is calculated; M3: Calculates a node credit score based on the probability of node privacy leakage and the threat level; M4: Based on the node credit score in step M3, set a scoring threshold range [L1, L2]; when the node credit score is greater than or equal to L2, the node is determined to be a trusted node; when the node credit score is greater than or equal to L1 and less than L2, the node is determined to be a suspicious node, and the monitoring frequency for the node is increased; when the node credit score is less than L1, the node is determined to be a malicious node, and the complete violation log is stored on the blockchain as evidence. M5: Sends malicious node information to the threat detection module, and simultaneously packages the probability of privacy leakage and the node credit score, and sends them to the distributed collaborative optimization module and the visualization management module.
7. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, The steps for dynamically optimizing the collaboration scheme based on threat level and node credit score include: E1: When a node has a low threat level and is a trusted node, the system is considered to be in a safe state. The node participates in collaborative computing and uses all of the node's original data for local model training. E2: When a node is identified as low-risk and is a suspicious node, or a node is identified as medium-risk and is a trusted node, the system is deemed to have a minor anomaly. Data from medium-risk or suspicious nodes is sampled and verified, and an enhanced encryption channel is enabled. E3: When a node is identified as having a medium threat level and is considered a suspicious node, the system is deemed to have a risk of privacy leakage. Only trusted nodes are allowed to participate in core computing, and all computing data from participating nodes is encrypted using lightweight homomorphic encryption. In addition, all model parameter exchanges must be conducted through the SMPC protocol to ensure the privacy of the exchanged parameters. E4: When a node is malicious, the system is determined to be in a high-risk privacy leakage environment. Global collaborative computing tasks are suspended and unnecessary connections between the server and the external network are cut off, while waiting for administrator intervention.
8. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 1, characterized in that, The steps to provide managers with global situational awareness and immediate control capabilities through a visual dashboard that integrates real-time status monitoring and multi-level risk alerts include: F1: When a risk event is detected, a graded alarm is triggered according to the threat level; when it is a medium-risk marker or a suspicious node, an orange warning icon is displayed and an alarm sound is emitted; when it is a high-risk marker or a malicious node, a red warning icon is displayed and a modal alarm pops up. F2: Sends visualized data and tiered alarm information to the data management module.
9. The distributed data security collaborative computing platform system for privacy-oriented computing according to claim 2, characterized in that, The steps for storing system-based basic data and sending tiered alarm information to security management personnel include: Tiered alarm information is sent to the receiving device of safety management personnel via email, and staff are reminded to check their emails via SMS.
10. A distributed data security collaborative computing method for privacy-preserving computation, implemented according to any one of claims 1-9, characterized in that, The work includes the following steps: S1: Collect node security dataset; S2: Collect known dangerous datasets from the network; S3: Preprocess the node security dataset and the dangerous dataset to obtain a joint feature vector; S4: Analyze joint feature vectors and malicious node information based on machine learning models to detect high-risk data and behaviors, and implement proactive defense measures; S5: Based on the integration of federated learning, secure multi-party computation and lightweight homomorphic encryption technology, it enables secure collaborative computation among distributed nodes while protecting the original data from leakage. S6: Based on privacy-preserving computation results, quantitatively assess the security risks of distributed nodes and monitor data flow; S7: Dynamically optimizes collaboration schemes based on threat level and node credit score; S8: Provides managers with global situational awareness and real-time control capabilities through a visual panel that integrates real-time status monitoring and multi-level risk alarms; S9: Stores system basic data and sends tiered alarm information to security management personnel.