Community-level cloud-edge collaborative micro-service architecture-based privacy computing system and method

Abstract

The application relates to the technical field of edge computing, and discloses a privacy computing system and method based on a community-level cloud-edge collaborative micro-service architecture. The system comprises a multi-source data acquisition module, which acquires multi-source heterogeneous data generated by terminal equipment in a community, such as user behavior, environmental perception and service request data; a privacy computing processing module, which dynamically encrypts the data and inputs the data into a privacy computing processing layer. The processing layer comprises a data preprocessing module and a collaborative computing module, the collaborative computing module is composed of a feature mapping layer, a collaborative scheduling layer and a result aggregation layer, can perform feature alignment, dependency relationship modeling and distributed aggregation on the data, and generates service response data. The method comprises the steps of acquiring data, encryption processing, collaborative computing and generating response data. The application realizes efficient processing and privacy protection of multi-source heterogeneous data in a community, and improves the accuracy and efficiency of community service response.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, specifically to a privacy computing system and method based on a community-level cloud-edge collaborative microservice architecture. Background Technology

[0002] With the acceleration of digitalization, community management and services are increasingly reliant on information technology. Communities contain a large number of terminal devices, such as residents' smart devices, environmental monitoring equipment, and various service facilities, which continuously generate multi-source, heterogeneous data. However, traditional data processing architectures and methods have revealed numerous problems when dealing with community-level data processing.

[0003] Regarding data acquisition, the data formats and communication protocols generated by different types of devices within the community vary significantly. User behavior data is scattered across the community cloud platform and edge nodes, making unified collection and integration difficult. Environmental perception data may originate from multiple sensors, each with different data acquisition frequencies, accuracy, and encoding methods, resulting in a complex and error-prone data acquisition and transmission process. Service request data also varies depending on the service type and business needs, making the acquisition of complete and accurate multi-source heterogeneous data a challenging task.

[0004] From a data processing perspective, community-level data processing requires real-time performance and high efficiency. Traditional centralized computing architectures transmit large amounts of data to the cloud for processing, resulting in high latency and failing to meet real-time requirements, such as handling emergency security incidents within the community. Furthermore, centralized computing is prone to resource overload and low processing efficiency when dealing with large-scale data. Moreover, community data contains a significant amount of user privacy information, such as sensitive data on user behavior habits and family situations. Traditional data encryption and privacy protection measures are often inadequate for the complex and ever-changing community environment, making it difficult to fully guarantee user privacy and security during data processing.

[0005] While microservice architecture improves system scalability and flexibility to some extent, challenges arise in the collaborative work between microservices in community-level scenarios. Different microservices may process data with dependencies, and accurately identifying and handling these dependencies to achieve efficient service scheduling and data interaction are urgent problems to be solved. For example, there is a lack of effective mechanisms for data sharing and service collaboration between the community property management microservice and the community security monitoring microservice.

[0006] Furthermore, existing privacy-preserving computation technologies have limitations in application within community settings. Traditional privacy-preserving computation methods typically focus on encrypted computation of single-type data, making it difficult to handle complex, multi-source, heterogeneous data within a community. Moreover, these methods struggle to balance computational efficiency and accuracy, failing to meet the demands of community-level services for rapid response and precise service delivery. For example, in community healthcare services, privacy-preserving computations are required on patients' personal health data, but existing privacy-preserving computation technologies cannot simultaneously guarantee data privacy, computational efficiency, and result accuracy. Summary of the Invention

[0007] The purpose of this invention is to provide a privacy computing system and method based on a community-level cloud-edge collaborative microservice architecture to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a privacy computing system based on a community-level cloud-edge collaborative microservice architecture, the system comprising:

[0009] The multi-source data acquisition module is used to acquire multi-source heterogeneous data generated by terminal devices within the target community. The multi-source heterogeneous data includes a first block of data corresponding to user behavior data, a second block of data corresponding to environmental perception data, and a third block of data corresponding to service request data. The user behavior data includes first encrypted data acquired from the community cloud platform and second encrypted data collected through edge nodes.

[0010] The privacy computing processing module is used to perform dynamic encryption strategy processing on the multi-source heterogeneous data and input it into the privacy computing processing layer for collaborative computing, and generate service response data of the target community based on the output of the privacy computing processing layer.

[0011] The privacy computing processing layer includes a data preprocessing module and a collaborative computing module. The data preprocessing module is used to perform block reassembly and noise injection on the encrypted transmission data stream. The collaborative computing module is obtained by joint training based on historical encrypted data and historical response data from multiple historical service periods. The collaborative computing module includes a feature mapping layer, a collaborative scheduling layer and a result aggregation layer connected in sequence.

[0012] Preferably, the feature mapping layer is used to perform feature alignment processing on multiple blocks of data contained in the encrypted transmission data stream to obtain block feature association data; the collaborative scheduling layer is used to model the dynamic dependency relationship between the block feature association data corresponding to each block of data to obtain collaborative scheduling feature data; and the result aggregation layer is used to perform distributed aggregation based on the collaborative scheduling feature data and the block feature association data to obtain service response data.

[0013] Preferably, the step of modeling the dynamic dependency relationships between the feature-related data corresponding to each data block to obtain collaborative scheduling feature data includes:

[0014] The key dependency nodes in the block feature association data are identified by using a dynamic sequence partitioning algorithm, and the dependency relationship sequence corresponding to each block of data is determined based on the service type corresponding to each key dependency node.

[0015] Calculate the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks, and determine the collaborative scheduling feature data between the two data blocks based on the path similarity.

[0016] Preferably, calculating the path similarity between dependency nodes with the same service type in the dependency relationship sequences corresponding to any two data blocks includes:

[0017] When the number of dependent nodes in the dependency relationship sequence corresponding to any two data blocks is inconsistent, virtual nodes are added to the last dependent node in the one with fewer dependent nodes, based on the service type corresponding to it, and the path similarity between dependent nodes with the same service type is calculated based on the added data.

[0018] Preferably, the data preprocessing module is specifically used for:

[0019] The information contained in the first block data, the second block data, and the third block data are standardized and divided into blocks according to the preset block rules to obtain standardized first block data, standardized second block data, and standardized third block data.

[0020] The standardized first block data and the standardized second block data are encrypted in real time using a dynamic key generation method, and the standardized third block data is statically encrypted using a fixed key encryption method to generate first encrypted block data, second encrypted block data and third encrypted block data; wherein, the first encrypted block data contains dynamically encrypted first user behavior data and dynamically encrypted second user behavior data.

[0021] Preferably, the data preprocessing module is further used for:

[0022] Calculate the correlation coefficient between the dynamically encrypted first user behavior data and the dynamically encrypted second user behavior data within a historical time period;

[0023] Based on the correlation coefficient and the service demand of the dynamically encrypted first user behavior data in the real-time time period, predict the predicted service demand of the dynamically encrypted second user behavior data in the real-time time period.

[0024] Target user behavior data is generated based on the dynamically encrypted second user behavior data and the predicted service demand of the dynamically encrypted second user behavior data within the real-time time period, and the block data corresponding to the target user behavior data is used as the first encrypted block data.

[0025] Preferably, the collaborative scheduling layer specifically includes:

[0026] The dependency analysis unit is used to perform service link analysis on each block of data contained in the block feature association data, so as to extract the corresponding service dependency chain from each block of data.

[0027] The scheduling optimization unit is used to dynamically match the service dependency chain extracted from each data block with the corresponding data block feature data to generate collaborative scheduling feature data.

[0028] Preferably, the collaborative scheduling layer further includes:

[0029] The redundancy elimination unit is used to remove redundant nodes from the cooperative scheduling feature data.

[0030] Preferably, the result aggregation layer specifically includes:

[0031] The distributed fusion unit contains multiple sub-aggregation nodes, and each sub-aggregation node is connected to each block in the collaborative scheduling feature data and the block feature association data through weight configuration.

[0032] The dynamic weight adjustment unit is used to optimize the weight configuration through a dynamic weight adjustment algorithm to minimize the deviation between service response data and actual service demand.

[0033] The service priority division unit is used to perform service priority division and resource allocation prediction based on the collaborative scheduling feature data and the block feature association data, and generate service response data.

[0034] Preferably, the present invention also includes a privacy computing method based on a community-level cloud-edge collaborative microservice architecture, the method comprising:

[0035] Acquire multi-source heterogeneous data generated by terminal devices within the target community. The multi-source heterogeneous data includes a first block of data corresponding to user behavior data, a second block of data corresponding to environmental perception data, and a third block of data corresponding to service request data. The user behavior data includes first encrypted data obtained from the community cloud platform and second encrypted data collected through edge nodes.

[0036] The multi-source heterogeneous data is processed by a dynamic encryption strategy to generate an encrypted transmission data stream;

[0037] The encrypted data stream is input into the privacy computing processing layer for collaborative computation.

[0038] The privacy computing processing layer generates service response data for the target community based on its output. The privacy computing processing layer includes a data preprocessing module and a collaborative computing module. The data preprocessing module performs block reassembly and noise injection on the encrypted transmission data stream. The collaborative computing module is jointly trained based on historical encrypted data and historical response data from multiple historical service periods. The collaborative computing module includes a feature mapping layer, a collaborative scheduling layer, and a result aggregation layer connected sequentially. The feature mapping layer performs feature alignment processing on multiple blocks of data contained in the encrypted transmission data stream to obtain block feature association data. The collaborative scheduling layer models the dynamic dependencies between the block feature association data corresponding to each block of data to obtain collaborative scheduling feature data. The result aggregation layer performs distributed aggregation based on the collaborative scheduling feature data and the block feature association data to obtain service response data.

[0039] The step of modeling the dynamic dependencies between the feature-related data corresponding to each data block to obtain collaborative scheduling feature data includes:

[0040] The key dependency nodes in the block feature association data are identified by using a dynamic sequence partitioning algorithm, and the dependency relationship sequence corresponding to each block of data is determined based on the service type corresponding to each key dependency node.

[0041] Calculate the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks, and determine the collaborative scheduling feature data between the two data blocks based on the path similarity; the calculation of the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks includes:

[0042] When the number of dependent nodes in the dependency relationship sequence corresponding to any two data blocks is inconsistent, virtual nodes are added to the last dependent node in the one with fewer dependent nodes, based on the service type corresponding to it, and the path similarity between dependent nodes with the same service type is calculated based on the added data.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] In terms of data processing, the multi-source data acquisition module effectively integrates heterogeneous data generated by terminal devices within the target community, covering user behavior data, environmental perception data, and service request data. This comprehensive data acquisition approach ensures the integrity and accuracy of community data, providing a solid data foundation for subsequent analysis and services. By dividing user behavior data into first encrypted data acquired from the community cloud platform and second encrypted data collected through edge nodes, the data storage advantages of the cloud platform are utilized, while leveraging the real-time acquisition capabilities of edge nodes, thus optimizing the data acquisition process.

[0045] The privacy-preserving computation processing module applies dynamic encryption strategies to multi-source heterogeneous data and performs collaborative computation through the privacy-preserving computation processing layer. Specifically, the data preprocessing module performs block reassembly and noise injection on the encrypted transmission data stream, effectively enhancing data security during transmission and processing and preventing data leakage and malicious attacks. A flexible encryption approach is used, employing dynamic key generation for real-time encryption of some data and fixed-key encryption for static encryption of other data. This approach ensures data security while balancing encryption efficiency and system performance. By calculating the correlation coefficient of dynamically encrypted user behavior data, service demands based on user behavior data can be predicted, enabling a more accurate understanding of user needs and improving the targeting and quality of services.

[0046] The collaborative computing module performs joint training based on historical encrypted data and historical response data from multiple historical service cycles, enabling the system to continuously learn and optimize its computing model. The feature mapping layer performs feature alignment processing on the segmented data in the encrypted transmission data stream, ensuring effective matching of data from different sources at the feature level, providing an accurate data foundation for subsequent collaborative computing. The collaborative scheduling layer models the dynamic dependencies between segmented feature-related data, identifies key dependency nodes, determines dependency sequences, and calculates path similarity, achieving precise control over the dependencies between various segmented data. This not only improves service scheduling efficiency but also avoids service conflicts and delays caused by improper handling of data dependencies. The redundancy elimination unit removes redundant nodes from the collaborative scheduling feature data, further optimizing the scheduling process and improving system performance. The result aggregation layer, through the collaborative work of the distributed fusion unit, dynamic weight adjustment unit, and service priority partitioning unit, can perform distributed aggregation based on collaborative scheduling feature data and segmented feature-related data to generate high-quality service response data. The dynamic weight adjustment algorithm can minimize the deviation between service response data and actual service demand. The service priority division unit divides service priorities and predicts resource allocation based on data characteristics, ensuring that critical services and urgent needs can be responded to in a timely manner, thereby improving resource utilization efficiency and service satisfaction.

[0047] From a system architecture perspective, the cloud-edge collaborative microservice architecture fully leverages the advantages of cloud computing and edge computing. Edge nodes are responsible for real-time data collection and preliminary processing, reducing data transmission volume and latency, and improving system real-time performance. The cloud computing platform provides powerful computing and storage resources for handling complex computing tasks and large-scale data storage. This architectural pattern satisfies the community service's real-time requirements while also possessing the ability to handle massive amounts of data, enhancing the system's scalability and flexibility. The various microservices interact with each other through clear interfaces and collaboration mechanisms, facilitating data exchange and service calls, making the system easy to maintain and upgrade.

[0048] Regarding privacy protection, the privacy computing system and method of this invention employ strict privacy protection measures throughout the entire process from data collection to processing. Encrypted data stream transmission, block reassembly, noise injection, and a combination of multiple encryption methods comprehensively safeguard the privacy and security of user data. Even if data is intercepted during transmission and processing, attackers will find it difficult to obtain genuine user data, effectively preventing user privacy leaks and enhancing user trust in the community service system.

[0049] Regarding the improvement of community service quality, the service response data generated by the system and method of this invention is more accurate and timely. Whether in community safety management, environmental monitoring, or resident life services, higher-quality services can be provided based on accurate data analysis. For example, in community safety monitoring, abnormal behavior can be quickly identified and alerts issued promptly; in community environmental management, environmental governance measures can be optimized based on environmental perception data; and in community life services, personalized service recommendations can be provided based on user behavior data and service request data. This significantly improves the quality of life and satisfaction of community residents, promoting the intelligent development and efficient management of the community. Attached Figure Description

[0050] Figure 1 This is a schematic diagram illustrating the working principle of the privacy computing system and method based on a community-level cloud-edge collaborative microservice architecture as described in this invention.

[0051] Figure 2 A flowchart for calculating the path similarity of dependency sequences;

[0052] Figure 3 This is a flowchart for dependency analysis and scheduling optimization at the collaborative scheduling layer. Detailed Implementation

[0053] 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.

[0054] Please see Figures 1-3 This invention provides a privacy computing system and method based on a community-level cloud-edge collaborative microservice architecture, and its specific implementation scheme is described in detail below.

[0055] The system comprises a multi-source data acquisition module and a privacy computing processing module. The multi-source data acquisition module is responsible for acquiring heterogeneous data from various sources generated by terminal devices within the target community. This data includes a first segment of data corresponding to user behavior data, a second segment of data corresponding to environmental perception data, and a third segment of data corresponding to service request data. User behavior data comes from a wide range of sources, including both first-encrypted data acquired from the community cloud platform and second-encrypted data collected through edge nodes. For example, in a smart community, exercise data generated by residents using smart fitness equipment is collected by edge nodes and becomes second-encrypted data; reservation data generated by residents booking community activities on the community cloud platform is first-encrypted data acquired from the community cloud platform. Environmental perception data can be data such as temperature, humidity, and light intensity collected by various sensors within the community; this data is organized into segments and becomes the second segment of data. Service request data consists of requests from residents to the community service system, such as repair requests and property service inquiries, which are processed to become the third segment of data.

[0056] The privacy-preserving computation processing module applies dynamic encryption strategies to multi-source heterogeneous data and inputs the processed data into the privacy-preserving computation processing layer for collaborative computation. Finally, it generates service response data for the target community based on the output of the privacy-preserving computation processing layer. The privacy-preserving computation processing layer comprises a data preprocessing module and a collaborative computation module. The data preprocessing module primarily performs block reassembly and noise injection on the encrypted transmission data stream. The collaborative computation module is jointly trained based on historical encrypted data and historical response data from multiple historical service periods. This collaborative computation module includes a feature mapping layer, a collaborative scheduling layer, and a result aggregation layer connected in sequence.

[0057] The technical solution of the present invention will be further described in detail below with reference to specific embodiments.

[0058] Example 1:

[0059] This embodiment details the data preprocessing module's handling of multi-source heterogeneous data. After the multi-source data acquisition module obtains the first, second, and third data blocks, the data preprocessing module begins its work. According to preset segmentation rules, the information contained in the first, second, and third data blocks is standardized and segmented. For example, for user behavior data in the first data block, assuming it includes information such as user activity trajectories and the time spent using community facilities, the preset segmentation rule might be to divide the data into blocks every half hour in chronological order, organizing the relevant information into the corresponding time blocks to obtain standardized first data blocks. Similarly, for environmental perception data in the second data block, if the environmental perception data includes data collected by multiple sensors at different times, it is standardized and segmented according to factors such as sensor type and collection time interval to obtain standardized second data blocks; for service request data in the third data block, it is standardized and segmented according to dimensions such as request type and request time to obtain standardized third data blocks.

[0060] Next, a dynamic key generation method is used to encrypt the standardized first and second data blocks in real time. This method generates constantly changing keys based on real-time conditions within the community, such as the current network security status and data access frequency. For example, when abnormal access behavior occurs within the community network, the dynamic key generation algorithm adjusts the key generation rules accordingly, generating a new key to encrypt the data. A fixed-key encryption method is then used to statically encrypt the standardized third data block. This method uses a pre-set key because service request data is relatively stable; using a fixed key ensures a certain level of security while reducing the computational load. Finally, three encrypted data blocks are generated: the first, second, and third encrypted data blocks. The first encrypted data block contains dynamically encrypted first and second user behavior data.

[0061] The data preprocessing module also calculates the correlation coefficient between dynamically encrypted first user behavior data and dynamically encrypted second user behavior data over a historical period. Taking the behavior data of residents using smart locks in a community as an example, the first user behavior data may be the lock usage permission data set by residents on the community cloud platform, and the second user behavior data is the record data of residents actually using the locks. By analyzing information such as the time and frequency of residents setting permissions and actually using the locks over a historical period, the correlation coefficient between the two is calculated. Based on the correlation coefficient and the service demand of the dynamically encrypted first user behavior data in the real-time period, the predicted service demand of the dynamically encrypted second user behavior data in the real-time period is predicted. Assuming that the correlation coefficient is high and the first user behavior data shows that residents frequently modify lock usage permissions during the real-time period, it can be predicted that the frequency of residents using the locks may also increase during the real-time period. Target user behavior data is generated based on the dynamically encrypted second user behavior data and the predicted service demand of the dynamically encrypted second user behavior data in the real-time period, and the block data corresponding to the target user behavior data is used as the first encrypted block data.

[0062] Example 2:

[0063] In the collaborative computing module, the feature mapping layer performs feature alignment processing on multiple data blocks contained in the encrypted transmission data stream. After obtaining the block feature association data, the collaborative scheduling layer begins to function. The collaborative scheduling layer first includes a dependency analysis unit, which performs service link analysis on each data block contained in the block feature association data. For example, in a block feature association data containing community security monitoring data, resident health data, and community service appointment data, for the community security monitoring data block, the dependency analysis unit analyzes the entire service link from the data acquisition device to the storage system, and then to the community security personnel's viewing and use. From this service link, the corresponding service dependency chain is extracted, such as the normal operation of the acquisition device depending on the device's power supply, and data transmission depending on network stability. Similarly, a similar service link analysis is performed on the resident health data block and the community service appointment data block to extract their respective service dependency chains.

[0064] The scheduling optimization unit dynamically matches the service dependency chains extracted from each data block with the corresponding block feature data. For example, in the community service reservation data block, the success of a reservation depends on the sufficiency of service resources, and the block feature data contains real-time status information of the current community service resources. Based on this information, the scheduling optimization unit matches the demands in the service dependency chains with the resource status in the block feature data. If a certain community service resource is currently scarce, the scheduling optimization unit will adjust the priority of service reservations or recommend other alternative services, generating collaborative scheduling feature data.

[0065] The collaborative scheduling layer also includes a redundancy elimination unit, which removes redundant nodes from the collaborative scheduling feature data. The generated collaborative scheduling feature data may contain duplicate or unnecessary node information. For example, when integrating service dependency chains of multiple data blocks, some intermediate nodes may appear repeatedly in different dependency chains without substantially affecting the final scheduling decision. The redundancy elimination unit identifies these redundant nodes and removes them from the collaborative scheduling feature data, improving data processing efficiency and scheduling accuracy.

[0066] Example 3:

[0067] The result aggregation layer includes a distributed fusion unit, a dynamic weight adjustment unit, and a service priority partitioning unit. This embodiment focuses on describing the specific working mechanism of the result aggregation layer. The result aggregation layer includes a distributed fusion unit, a dynamic weight adjustment unit, and a service priority partitioning unit.

[0068] The distributed fusion unit comprises multiple sub-aggregation nodes. Each sub-aggregation node is connected to various blocks in the collaborative scheduling feature data and the block feature association data through weight configuration. Taking a community resource allocation scenario as an example, suppose the collaborative scheduling feature data involves information such as parking space allocation and community activity space allocation, while the block feature association data contains residents' demand data for these resources and the current status data of the resources. The sub-aggregation nodes in the distributed fusion unit are responsible for the aggregation calculation of different types of resources. For example, one sub-aggregation node specifically processes parking space-related data. Based on the weight configuration, it retrieves information related to parking space scheduling from the collaborative scheduling feature data and information such as residents' demand for parking spaces and the real-time status of parking spaces from the block feature association data.

[0069] The dynamic weight adjustment unit optimizes the weight configuration using a dynamic weight adjustment algorithm to minimize the deviation between service response data and actual service demand. In a parking space allocation scenario, if residents' demand for parking spaces in a specific area suddenly increases over a period of time, and the parking space allocation scheme calculated based on the current weight configuration cannot meet the actual demand, the dynamic weight adjustment unit will adjust the weights related to parking spaces in that area based on real-time data. Assuming... Indicates weight, The deviation between service response data and actual service demand is represented by the following formula, which is used for weight adjustment: ,in, These are the adjusted new weights. It is the old weight before the adjustment. It is the learning rate, used to control the step size of weight adjustments. Indicates deviation For the old weights The partial derivatives of the formula can be used to guide the weights towards reducing the bias. The direction was adjusted.

[0070] The service priority allocation unit performs service priority allocation and resource allocation prediction based on collaborative scheduling feature data and block feature correlation data, generating service response data. In the community resource allocation scenario, the service priority allocation unit comprehensively considers various factors. For services involving residents' urgent needs, such as the demand for community medical resources during sudden illnesses, their priority is increased; for routine community service needs, such as registration for community activities, priority is allocated based on resource availability and the urgency of the need. Then, based on the priority allocation results, resource allocation prediction is performed to generate the final service response data, such as prioritizing the allocation of medical resources for emergency medical needs and rationally arranging venues and times for routine community activities. Assuming... Indicates service priority. Indicates the degree of resource availability. The urgency of the need is indicated by a formula. To calculate service priority, where It is a weighting coefficient, and its value ranges from... arrive Between, used to balance the degree of resource availability and the urgency of the need Service Priority The impact.

[0071] Example 4:

[0072] This embodiment details the specific processes of the feature mapping layer, collaborative scheduling layer, and result aggregation layer in collaborative computing. After the encrypted data stream is input into the collaborative computing module, it first reaches the feature mapping layer. The feature mapping layer performs feature alignment processing on the multiple data blocks contained in the encrypted data stream. For example, in a community smart service system, the data blocks may include resident shopping preference data, community surrounding merchant discount information data, and community delivery service data. The feature mapping layer analyzes the features of this data, aligning the product category features in the resident shopping preference data with the product promotion category features in the merchant discount information data, and aligning the delivery area features in the delivery service data with the resident's location features, thereby obtaining the feature-associative data for the data blocks.

[0073] After receiving the segmented feature association data, the collaborative scheduling layer models the dynamic dependencies between the segmented feature association data corresponding to each segment. A dynamic sequence partitioning algorithm is used to identify key dependency nodes in the segmented feature association data. For example, in the above data, if a resident's purchase of a certain type of product depends on a merchant's promotional activities, then the nodes related to the merchant's promotional activities are key dependency nodes. The dependency sequence for each segment of data is determined based on the service type corresponding to each key dependency node. Assume the dependency sequence for resident shopping preference data is: focus on product category - view merchant promotions - select to purchase product; the dependency sequence for merchant promotion information data is: publish promotional activities - attract resident attention - facilitate purchase.

[0074] Calculate the path similarity between dependency nodes with the same service type in the dependency relationship sequences corresponding to any two data blocks. When the number of dependency nodes in the dependency relationship sequences corresponding to any two data blocks is inconsistent, a virtual node is added to the sequence based on the service type of the last dependency node in the data block with fewer dependency nodes. For example, if the dependency relationship sequence of resident shopping preference data has 3 nodes and the dependency relationship sequence of merchant discount information data has 2 nodes, a virtual node "Complete Purchase Transaction" is added after the last dependency node "Facilitate Purchase" in the merchant discount information data sequence to make its number of nodes consistent with that of the dependency relationship sequence of resident shopping preference data. Then, the path similarity between dependency nodes with the same service type is calculated based on the supplemented data, and the collaborative scheduling feature data between any two data blocks is determined based on the path similarity.

[0075] The aggregation layer performs distributed aggregation based on collaborative scheduling feature data and block feature association data. Sub-aggregation nodes in the distributed fusion unit perform initial aggregation of data according to weight configurations, the dynamic weight adjustment unit optimizes the weight configurations, and the service priority partitioning unit performs service priority partitioning and resource allocation prediction, ultimately obtaining service response data. For example, in community delivery services, based on collaborative scheduling feature data and block feature association data, and considering factors such as the urgency of residents' orders and delivery distance, the delivery priority and resource allocation scheme are determined, generating delivery service response data and informing residents of the estimated delivery time and other information.

[0076] Example 5:

[0077] This embodiment further illustrates how the privacy-preserving computation processing module performs collaborative computation based on historical data and generates service response data. The collaborative computation module within the privacy-preserving computation processing module is obtained through joint training of historical encrypted data and historical response data from multiple historical service periods. In a community property service scenario, a historical service period can be measured in weeks. Over multiple historical service periods, a large amount of historical encrypted data has been accumulated, such as weekly resident repair request data, community facility usage data, and corresponding historical response data, such as repair service completion times and facility maintenance records.

[0078] Upon receiving the current encrypted data stream, the collaborative computing module utilizes this historical data for collaborative computation. The feature mapping layer, collaborative scheduling layer, and result aggregation layer all reference patterns and regularities in the historical data during the computation process. For example, when performing feature alignment, the feature mapping layer considers the correlations between different types of data features in the historical data to improve the accuracy of feature alignment. When analyzing the dynamic dependencies between data chunks, the collaborative scheduling layer compares the dependencies with those in similar situations during historical service cycles to optimize the modeling of dynamic dependencies. When performing distributed aggregation, the result aggregation layer adjusts the weight configuration and service priority allocation rules based on the service requirements and resource allocation in the historical data.

[0079] Through collaborative computing, calculation results are obtained, and then service response data for the target community is generated based on these results. In property service scenarios, if a resident submits a new repair request, the collaborative computing module will combine historical repair data and current repair resource availability to calculate the estimated completion time of the repair service, the required repair personnel and materials, and generate service response data to provide feedback to the resident, informing them of the repair service arrangements.

[0080] Example 6:

[0081] This embodiment combines a real-world community application scenario to comprehensively demonstrate the operation process of the entire privacy computing system and method based on a community-level cloud-edge collaborative microservice architecture. Taking a large smart community as an example, a large number of terminal devices are distributed within the community. The multi-source data acquisition module continuously acquires various types of data, such as property service requests issued by residents through smart terminals (e.g., garbage collection appointments, water and electricity bill payment inquiries, etc.) as the third block of data corresponding to service request data; air quality, noise levels, and other data collected by environmental monitoring equipment within the community as the second block of data corresponding to environmental perception data; and behavioral data generated by residents using various facilities within the community (e.g., smart gyms, library self-service borrowing and returning systems) and operation data on the community cloud platform (e.g., community activity registration, neighborhood communication information release) as the first block of data corresponding to user behavior data, part of which is acquired from the community cloud platform and part is collected through edge nodes.

[0082] These multi-source heterogeneous data are transmitted to the privacy-preserving computation processing module. The data preprocessing module performs standardization, encryption, and noise injection on the data according to preset rules, generating an encrypted transmission data stream. This encrypted data stream enters the privacy-preserving computation processing layer. In the collaborative computation module, the feature mapping layer performs feature alignment on the data, the collaborative scheduling layer analyzes the dynamic dependencies between data and optimizes scheduling, and the result aggregation layer performs distributed aggregation, weight adjustment, and service priority allocation, ultimately generating service response data.

[0083] For example, when a large-scale event is held in the community, the multi-source data acquisition module obtains a large amount of event-related data, including residents' event registration data, environmental perception data around the event venue, and service request data from the event organizers. The privacy-preserving computation processing module processes this data, while the collaborative computation module performs collaborative calculations based on historical event data and the current situation. The result aggregation layer generates service response data such as resource allocation plans for the event site (e.g., personnel guidance, material allocation) and suggestions for optimizing the surrounding environment (e.g., noise control, traffic management) based on the calculation results, and feeds this data back to the event organizers and relevant management departments to ensure the smooth running of the event.

[0084] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0085] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A privacy computing system based on a community-level cloud-edge collaborative microservice architecture, characterized in that, include: The multi-source data acquisition module is used to acquire multi-source heterogeneous data generated by terminal devices within the target community. The multi-source heterogeneous data includes a first block of data corresponding to user behavior data, a second block of data corresponding to environmental perception data, and a third block of data corresponding to service request data. The user behavior data includes first encrypted data acquired from the community cloud platform and second encrypted data collected through edge nodes. The privacy computing processing module is used to perform dynamic encryption strategy processing on the multi-source heterogeneous data and input it into the privacy computing processing layer for collaborative computing, and generate service response data of the target community based on the output of the privacy computing processing layer. The privacy computing processing layer includes a data preprocessing module and a collaborative computing module. The data preprocessing module is used to perform block reassembly and noise injection on the encrypted transmission data stream. The collaborative computing module is obtained by joint training based on historical encrypted data and historical response data from multiple historical service periods. The collaborative computing module includes a feature mapping layer, a collaborative scheduling layer, and a result aggregation layer connected in sequence. The feature mapping layer is used to perform feature alignment processing on multiple blocks of data contained in the encrypted transmission data stream to obtain block feature association data; the collaborative scheduling layer is used to model the dynamic dependency relationship between the block feature association data corresponding to each block of data to obtain collaborative scheduling feature data; the result aggregation layer is used to perform distributed aggregation based on the collaborative scheduling feature data and the block feature association data to obtain service response data. The dynamic dependency relationships between the feature association data corresponding to each data block are modeled to obtain collaborative scheduling feature data, including: The key dependency nodes in the block feature association data are identified by using a dynamic sequence partitioning algorithm, and the dependency relationship sequence corresponding to each block of data is determined based on the service type corresponding to each key dependency node. Calculate the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks, and determine the collaborative scheduling feature data between the two data blocks based on the path similarity; The calculation of path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks includes: When the number of dependent nodes in the dependency relationship sequence corresponding to any two data blocks is inconsistent, virtual nodes are added to the last dependent node in the one with fewer dependent nodes, based on the service type corresponding to it, and the path similarity between dependent nodes with the same service type is calculated based on the added data.

2. The privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in claim 1, characterized in that, The data preprocessing module is specifically used for: The information contained in the first block data, the second block data, and the third block data are standardized and divided into blocks according to the preset block rules to obtain standardized first block data, standardized second block data, and standardized third block data. The standardized first block data and the standardized second block data are encrypted in real time using a dynamic key generation method, and the standardized third block data is statically encrypted using a fixed key encryption method to generate first encrypted block data, second encrypted block data and third encrypted block data; wherein, the first encrypted block data contains dynamically encrypted first user behavior data and dynamically encrypted second user behavior data.

3. The privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in claim 2, characterized in that, The data preprocessing module is also used for: Calculate the correlation coefficient between the dynamically encrypted first user behavior data and the dynamically encrypted second user behavior data within a historical time period; Based on the correlation coefficient and the service demand of the dynamically encrypted first user behavior data in the real-time time period, predict the predicted service demand of the dynamically encrypted second user behavior data in the real-time time period. Target user behavior data is generated based on the dynamically encrypted second user behavior data and the predicted service demand of the dynamically encrypted second user behavior data within the real-time time period, and the block data corresponding to the target user behavior data is used as the first encrypted block data.

4. The privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in claim 1, characterized in that, The coordinated scheduling layer specifically includes: The dependency analysis unit is used to perform service link analysis on each block of data contained in the block feature association data, so as to extract the corresponding service dependency chain from each block of data. The scheduling optimization unit is used to dynamically match the service dependency chain extracted from each data block with the corresponding data block feature data to generate collaborative scheduling feature data.

5. The privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in claim 4, characterized in that, The collaborative scheduling layer also includes: The redundancy elimination unit is used to remove redundant nodes from the cooperative scheduling feature data.

6. The privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in claim 1, characterized in that, The result aggregation layer specifically includes: The distributed fusion unit contains multiple sub-aggregation nodes, and each sub-aggregation node is connected to each block in the collaborative scheduling feature data and the block feature association data through weight configuration. The dynamic weight adjustment unit is used to optimize the weight configuration through a dynamic weight adjustment algorithm to minimize the deviation between service response data and actual service demand. The service priority division unit is used to perform service priority division and resource allocation prediction based on the collaborative scheduling feature data and the block feature association data, and generate service response data.

7. A privacy computing method based on a community-level cloud-edge collaborative microservice architecture, implementing the privacy computing system based on a community-level cloud-edge collaborative microservice architecture as described in any one of claims 1 to 6, characterized in that, include: Acquire multi-source heterogeneous data generated by terminal devices within the target community. The multi-source heterogeneous data includes a first block of data corresponding to user behavior data, a second block of data corresponding to environmental perception data, and a third block of data corresponding to service request data. The user behavior data includes first encrypted data obtained from the community cloud platform and second encrypted data collected through edge nodes. The multi-source heterogeneous data is processed by a dynamic encryption strategy to generate an encrypted transmission data stream; The encrypted data stream is input into the privacy computing processing layer for collaborative computation. The privacy computing processing layer generates service response data for the target community based on its output. The privacy computing processing layer includes a data preprocessing module and a collaborative computing module. The data preprocessing module performs block reassembly and noise injection on the encrypted transmission data stream. The collaborative computing module is jointly trained based on historical encrypted data and historical response data from multiple historical service periods. The collaborative computing module includes a feature mapping layer, a collaborative scheduling layer, and a result aggregation layer connected sequentially. The feature mapping layer performs feature alignment processing on multiple blocks of data contained in the encrypted transmission data stream to obtain block feature association data. The collaborative scheduling layer models the dynamic dependencies between the block feature association data corresponding to each block of data to obtain collaborative scheduling feature data. The result aggregation layer performs distributed aggregation based on the collaborative scheduling feature data and the block feature association data to obtain service response data. The step of modeling the dynamic dependencies between the feature-related data corresponding to each data block to obtain collaborative scheduling feature data includes: The key dependency nodes in the block feature association data are identified by using a dynamic sequence partitioning algorithm, and the dependency relationship sequence corresponding to each block of data is determined based on the service type corresponding to each key dependency node. Calculate the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks, and determine the collaborative scheduling feature data between the two data blocks based on the path similarity; the calculation of the path similarity between dependency nodes with the same service type in the dependency relationship sequence corresponding to any two data blocks includes: When the number of dependent nodes in the dependency relationship sequence corresponding to any two data blocks is inconsistent, virtual nodes are added to the last dependent node in the one with fewer dependent nodes, based on the service type corresponding to it, and the path similarity between dependent nodes with the same service type is calculated based on the added data.

Images