A distributed data collaboration method and device based on federal control

Abstract

The application discloses a kind of distributed data cooperation method and device based on federal control, belong to computer technology and machine learning field, it includes obtaining network configuration parameter and generates initial credit parameter for multiple entities;Local data is obtained and feature extraction is carried out to generate feature data, initial credit parameter is combined with real-time obtained network state parameter to generate dynamic mask and process feature data, and mask feature data is obtained;According to the initial credit parameter, the mask feature data received from each entity is weighted and aggregated, to generate global cooperation result and evaluate, to generate effect score and combined with the network performance parameter obtained, to generate updated cooperation strategy.The application adopts to establish initial and combined credit, combined with network state to generate dynamic mask to process data, to carry out credit weighted aggregation, and based on cooperation effect closed loop update strategy, realize the self-adaptation, high robustness and high efficiency of distributed data cooperation in complex environment.

Description

Technical Field

[0001] This invention relates to the fields of computer technology and machine learning, and in particular to a distributed data collaboration method and apparatus based on federated control. Background Technology

[0002] Distributed data collaboration is a key technology in the fields of artificial intelligence and large-scale machine learning. It allows multiple geographically dispersed entities to collaboratively complete data analysis or model training tasks without directly sharing their local raw data. This paradigm, especially within the framework of federated learning, reduces issues such as data privacy, security, and data silos by performing computations locally and uploading intermediate results to a central server for aggregation. It has broad application prospects in sensitive fields such as finance and healthcare.

[0003] Existing distributed data collaboration methods typically employ a relatively fixed collaboration pattern. For example, in federated learning, a central server periodically distributes a global model to selected participants. Each participant trains on their local data, uploads model updates, and the server then performs a simple average or data-volume-weighted aggregation of all updates. These methods can work effectively in ideal environments, but their collaboration strategies are often static, with relatively simplistic methods for selecting participants and allocating weights for data aggregation.

[0004] However, the aforementioned existing technical solutions have significant technical shortcomings in practical deployment. First, they generally lack differentiated consideration of the intrinsic attributes and reliability of participating entities, treating all entities equally. This allows entities with low computing power or unstable behavior to submit low-quality updates, thereby reducing the efficiency of global collaboration and the accuracy of the final result. Second, existing solutions are insufficiently aware of dynamically changing network environments and cannot flexibly adjust data transmission and aggregation strategies based on real-time network latency, bandwidth, and other conditions. In unstable network conditions, this can easily lead to decreased collaboration efficiency or task failure. Finally, existing technologies lack an effective closed-loop feedback mechanism, failing to dynamically optimize and adjust subsequent collaboration strategies based on historical collaboration results. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a distributed data collaboration method and apparatus based on federated control. It employs a technical solution involving establishing initial credit, generating dynamic masks based on joint credit and network state to process data, performing credit-weighted aggregation, and updating the collaboration strategy in a closed loop based on collaboration effects. This approach enables adaptive, highly robust, and highly efficient distributed data collaboration in complex environments.

[0006] The above objectives can be achieved through the following approach:

[0007] A distributed data collaboration method and apparatus based on federated control includes: acquiring network configuration parameters for defining network topology; generating initial credit parameters for multiple entities based on the network configuration parameters; acquiring local data and performing feature extraction to generate feature data; generating a dynamic mask by combining the initial credit parameters with real-time acquired network state parameters; processing the feature data using the dynamic mask to obtain masked feature data; weighting and aggregating the masked feature data received from each entity according to the initial credit parameters to generate a global collaboration result; evaluating the effectiveness of the global collaboration result to generate an effect score; and generating an updated collaboration strategy by combining the effect score with acquired network performance parameters to adjust a preset dynamic mask generation strategy and a preset credit allocation rule.

[0008] Optionally, generating initial credit parameters for multiple entities based on the network configuration parameters includes: extracting entity attribute parameters of multiple entities from the network configuration parameters; performing credit quantification processing based on the entity attribute parameters to calculate an initial credit value for each entity; and integrating the initial credit values ​​of all entities to form initial credit parameters.

[0009] Optionally, the step of generating a dynamic mask by combining the initial credit parameters with the real-time acquired network status parameters includes: real-time monitoring of the communication link status to obtain the network status parameters, which include network latency and bandwidth parameters; performing joint processing based on the network status parameters and the initial credit parameters to calculate the strength coefficient of the dynamic mask; and performing random sampling based on the strength coefficient of the dynamic mask to generate a dynamic mask.

[0010] Optionally, obtaining the mask feature data includes: scaling the feature data numerically according to the strength coefficient of the dynamic mask to generate scaled feature data; and performing a bitwise XOR operation between the scaled feature data and the dynamic mask to obtain the mask feature data.

[0011] Optionally, generating the global collaborative result includes: assigning weights to the mask feature data of each entity according to the initial credit parameters to generate a weight set; and performing weighted fusion based on the mask feature data and the weight set to calculate the global collaborative result.

[0012] Optionally, the step of evaluating the effectiveness of the global collaboration results, generating an effectiveness score, and combining the effectiveness score with the acquired network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and preset credit allocation rules includes: continuously monitoring the communication status of the federated control network and acquiring network performance parameters including latency, packet loss rate, and throughput; performing consistency verification and error analysis on the global collaboration results, and generating an effectiveness score reflecting the effectiveness of this collaboration by combining historical collaboration effectiveness data; calculating the strategy adjustment amount through a preset multi-objective optimization function based on the effectiveness score and the network performance parameters; modifying the complexity control parameters in the dynamic mask generation strategy according to the strategy adjustment amount, and synchronously updating the reward and penalty magnitudes in the credit allocation rules to generate the updated collaboration strategy.

[0013] Optionally, the calculation of the strength coefficient of the dynamic mask includes: performing quantitative analysis based on the initial credit parameters to determine the priority level of the entity; performing joint mapping between the priority level and the network state parameters to obtain a mask reference value; and dynamically adjusting the mask reference value according to the network state parameters to generate the strength coefficient of the dynamic mask.

[0014] Optionally, generating the weight set includes: calculating the credit weight of each entity based on the initial credit parameters; obtaining network state parameters for quantifying communication quality, and adjusting the credit weight of each entity according to the network state parameters to generate an adaptive weight value; and aggregating the adaptive weight values ​​of all entities to form a weight set.

[0015] Optionally, the step of performing quantitative analysis based on the initial credit parameters to determine the priority level of an entity includes: inputting the initial credit parameters into a credit rating unit and dividing the entity into multiple credit intervals based on the credit value; matching the corresponding priority label according to the credit interval to which each entity belongs; obtaining historical collaborative behavior data and fine-tuning the priority label based on the historical collaborative behavior data to form a priority level.

[0016] Based on the same inventive concept, this invention also provides a distributed data collaboration device based on federated control. The device includes: a network credit initialization module, used to acquire network configuration parameters for defining network topology, and generate initial credit parameters for multiple entities based on the network configuration parameters; a dynamic mask generation module, used to acquire local data and perform feature extraction to generate feature data, combine the initial credit parameters with real-time acquired network state parameters to generate a dynamic mask, and use the dynamic mask to process the feature data to obtain mask feature data; a weighted aggregation collaboration module, used to perform weighted aggregation of the mask feature data received from each entity according to the initial credit parameters to generate a global collaboration result; and a policy evaluation and update module, used to evaluate the effectiveness of the global collaboration result, generate an effect score, and combine the effect score with acquired network performance parameters to generate an updated collaboration strategy for adjusting a preset dynamic mask generation strategy and a preset credit allocation rule.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] 1. This invention improves the system's high adaptability and robustness by constructing a closed-loop feedback collaborative strategy update mechanism. This method can continuously evaluate the effectiveness of global collaborative results and monitor network performance. Based on the evaluation results, it dynamically adjusts the dynamic mask generation strategy and credit allocation rules, enabling the system to proactively adapt to dynamic changes in the behavior of participating entities and the network environment, maintaining the stability and efficiency of collaborative tasks in complex and ever-changing environments.

[0019] 2. This invention improves the accuracy and reliability of global collaborative results by introducing a dynamic weighting mechanism based on entity attributes and network status. This method does not treat all participating entities equally; instead, it weights and aggregates the data contributed by each entity based on initial credit parameters and dynamically adjusts the weights according to real-time network status. This ensures that entities with high credit and high communication quality play a leading role in the collaboration, reducing the negative impact of low-quality data and network jitter on the final result.

[0020] 3. This invention achieves refined and secure control over data contribution by designing a dynamic mask generation and application method that combines initial credit and network state. This dynamic mask can adaptively scale and perturb uploaded feature data based on the credibility of each entity and its network environment. This not only ensures collaborative effectiveness while dynamically adjusting information transmission volume according to network bandwidth, but also provides a degree of privacy protection during data sharing, enhancing the security of the entire distributed collaborative system.

[0021] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a distributed data collaboration method based on federated control, according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram illustrating the relationship between the dynamic mask strength coefficient and the influence factor in an embodiment of the present invention.

[0025] Figure 3 This is a diagram of the strategy evaluation mechanism in an embodiment of the present invention.

[0026] Figure 4 This is a schematic diagram illustrating the determination of entity priority levels in an embodiment of the present invention.

[0027] Figure 5 This is a schematic diagram of the structure of a distributed data collaboration device based on federated control according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0029] Reference Figure 1 One embodiment of the present invention proposes a distributed data collaboration method based on federated control. It adopts a technical solution of establishing initial credit, generating dynamic mask processing data by combining credit and network state, performing credit weighted aggregation, and updating the collaboration strategy in a closed loop based on collaboration effect. This method can achieve adaptive, highly robust and efficient distributed data collaboration in complex environments.

[0030] The method described in this embodiment specifically includes:

[0031] Obtain network configuration parameters for defining the network topology, and generate initial credit parameters for multiple entities based on the network configuration parameters;

[0032] Local data is acquired and features are extracted to generate feature data. The initial credit parameters and real-time network state parameters are combined to generate a dynamic mask. The feature data is then processed using the dynamic mask to obtain mask feature data.

[0033] The mask feature data received from each entity is weighted and aggregated according to the initial credit parameters to generate a global collaborative result;

[0034] The effectiveness of the global collaboration results is evaluated to generate an effect score. The effect score is then combined with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules.

[0035] Specifically, this method first establishes an initial credit system by analyzing the inherent configuration parameters of network entities, assigning a basic initial credit parameter to each participant. During the collaboration process, each entity does not directly share the original features when processing data locally, but instead applies a dynamically generated mask. This dynamic mask depends on both the entity's static initial credit and dynamically changing network state parameters, thereby achieving intelligent adjustment of data contribution. In the data aggregation phase, the mask feature data submitted by each party is weighted and fused using the initial credit parameter, ensuring that high-credit entities have a greater influence on the final result. Most importantly, this method introduces a feedback control loop. After each round of collaboration, it evaluates the effectiveness of the global collaboration result and combines the effect score with network performance parameters to jointly drive a policy update mechanism. This dynamically adjusts the future dynamic mask generation strategy and the credit system allocation rules, enabling the entire collaboration system to continuously self-optimize and evolve. By introducing a federated control mechanism, this method improves the intelligence, security, and efficiency of the distributed data collaboration process. Dynamic masks generated based on initial credit and real-time network status, along with subsequent weighted aggregation, improve the accuracy and robustness of global collaboration results, effectively reducing interference from low-quality nodes and poor network environments on collaborative tasks. More importantly, through policy module updates, the entire collaboration framework is endowed with the ability to learn and continuously optimize itself. It can dynamically adapt to complex and ever-changing network environments and participant behaviors, automatically finding and maintaining the optimal balance between collaboration effectiveness and data security, thereby ensuring the stability, efficiency, and reliability of distributed data collaboration in long-term operation.

[0036] Optionally, generating initial credit parameters for multiple entities based on the network configuration parameters includes:

[0037] Extract entity attribute parameters of multiple entities from the network configuration parameters;

[0038] Credit quantification is performed based on the entity attribute parameters to calculate an initial credit value for each entity.

[0039] The initial credit values ​​of all entities are combined to form initial credit parameters.

[0040] Specifically, firstly, entity attribute parameters related to each entity are extracted from the network configuration parameters used to define the network topology. These entity attribute parameters are a collection describing the entity's static capabilities, including but not limited to hardware computing capabilities such as CPU performance and memory capacity, network connectivity capabilities such as nominal bandwidth, and security authentication levels and historical operational stability records. Next, credit quantification processing is performed on the extracted entity attribute parameters to calculate an initial credit value for each entity. This quantification process aims to transform multi-dimensional, heterogeneous attributes into a unified credit scalar. For calculating the first... Initial credit value of an entity ,have:

[0041] ;

[0042] in, Representing the The entity in the first The quantized values ​​of the entity attribute parameters are obtained by extracting the original attribute values ​​of the entity from the network configuration parameters and converting these heterogeneous attributes into values ​​with uniform dimensions by performing max-min normalization. This represents the total number of entity attribute parameters. Representing the The weight coefficients corresponding to the attribute parameters of each entity are set based on the analysis of the contribution of different attributes to the collaborative effect in historical collaborative tasks, such as regression analysis or feature importance assessment. This reflects the importance of different attributes in the credit assessment system and satisfies the condition that the sum of all weight coefficients is 1. Through this calculation, each entity obtains an initial credit value that reflects its comprehensive capabilities. Finally, the initial credit values ​​of all entities are integrated to form a structured set of initial credit parameters, such as a credit vector or mapping table. These initial credit parameters will serve as the objective basis for subsequent steps such as dynamic mask generation and weighted aggregation collaboration. The initial credit parameters generated in this way provide an objective and differentiated starting point for distributed data collaboration in federated control. By conducting initial trust assessments on the inherent and verifiable attributes of entities, the rationality of resource allocation and collaborative weight distribution is improved. Entities with strong computing power, good network conditions, and high security levels can be identified and preferentially selected from the outset, laying the foundation for subsequent collaborative tasks and improving the quality of the overall collaborative results.

[0043] Optionally, generating a dynamic mask by combining the initial credit parameters with the real-time acquired network state parameters includes:

[0044] Real-time monitoring of communication link status to obtain network status parameters including network latency and bandwidth parameters;

[0045] The strength coefficient of the dynamic mask is calculated by jointly processing the network state parameters and the initial credit parameters.

[0046] A dynamic mask is generated by randomly sampling based on the strength coefficients of the dynamic mask;

[0047] Specifically, the built-in network monitoring unit first monitors the communication link status between each entity and other nodes in real time. This is typically achieved by sending probe packets or analyzing feedback information from data transmission protocols, thereby obtaining real-time network status parameters. Key indicators include network latency reflecting data transmission delay and network bandwidth parameters characterizing link transmission capacity. After obtaining these parameters, the network status parameters are jointly processed with initial credit parameters to calculate a dynamic mask strength coefficient. This dynamic mask strength coefficient is a quantitative indicator that integrates the entity's credibility with the current network communication quality. After calculating the dynamic mask strength coefficient, random sampling is performed to generate the final dynamic mask. The dynamic mask is a data structure with the same dimension as the feature data, and the generation probability of its elements is related to the dynamic mask strength coefficient. For example, each bit or element value in the dynamic mask can be drawn from a Bernoulli distribution, and the probability of its value being 1 is equal to the dynamic mask strength coefficient. The generation of the dynamic mask is no longer static or immutable, but can respond in real time to changes in the intrinsic attributes of the participating entities and the external network environment. Figure 2 As shown in the diagram, this 3D scatter plot illustrates how the dynamic mask strength coefficient is jointly influenced by an entity's initial credit parameter and the real-time network quality index. Each point in the plot represents the state of an entity at a given moment, with darker colors generally indicating a higher mask strength coefficient. The plot shows that when both the initial credit parameter and the network quality index are high (i.e., the area in the front right of the plot), the dynamic mask strength coefficient tends to be lower. Conversely, when the credit is low or the network quality is poor (i.e., the area in the back left of the plot), the dynamic mask strength coefficient is relatively higher. When a high-credit entity is in a favorable network environment, a lower-strength mask is generated, allowing it to contribute higher-fidelity feature data, thereby improving the accuracy of the global collaborative results. Conversely, when an entity has low credit or deteriorating network conditions, a stronger mask is generated. This reduces the negative impact of low-quality data on the global results and adapts to limited bandwidth by reducing the amount of effective information, improving the timeliness of data transmission and enhancing the adaptability and robustness of the entire distributed data collaboration in complex and changing environments.

[0048] Optionally, the obtained mask feature data includes:

[0049] The feature data is numerically scaled based on the strength coefficient of the dynamic mask to generate scaled feature data;

[0050] Perform a bitwise XOR operation between the scaling feature data and the dynamic mask to obtain the mask feature data;

[0051] Specifically, firstly, based on the intensity coefficients of the dynamic mask obtained above, the locally extracted feature data is numerically scaled. Local data is obtained from the entity's local storage unit or real-time data acquisition interface. Local data includes, but is not limited to, sensor time-series data, image data, text logs, or model update gradients. Feature extraction algorithms are used to process the local data, extracting feature vectors that represent the essential characteristics of the data as feature data. For example, principal component analysis can be used for dimensionality reduction, or convolutional neural networks can be used for deep feature extraction of image data. The numerical scaling operation involves multiplying the intensity coefficients of the dynamic mask by each element in the feature data to generate scaled feature data. This operation adjusts the overall magnitude of the feature data based on real-time trust and network evaluation. Subsequently, this scaled feature data is XORed with the dynamic mask generated in the previous step. Since the bitwise XOR operation requires the operands to be in binary format, the floating-point or integer scaled feature data must be converted into a binary bit string through quantization or encoding before the operation. The dynamic mask itself is a binary bit string or matrix. Finally, the transformed scaled feature data is XORed bit by bit with the dynamic mask to obtain the final mask feature data. By combining numerical scaling and bitwise XOR operations, dual processing of the original feature data is achieved. The numerical scaling step adjusts the contribution weight of the feature data according to the strength coefficient of the dynamic mask, which is equivalent to a pre-weighting before data upload. This makes the numerical influence of data contributed by entities with high credit and good network conditions greater, and vice versa, thereby optimizing the quality of data fusion. The subsequent bitwise XOR operation, without completely destroying the data structure, performs controlled perturbation on the data, playing a role in privacy protection and data obfuscation. This dual processing mechanism not only improves the accuracy and robustness of the collaborative results, but also ensures the security of data in a distributed environment.

[0052] Optionally, generating global collaborative results includes:

[0053] A weight set is generated by assigning weights to the mask feature data of each entity based on the initial credit parameters.

[0054] The global collaborative result is calculated by weighting and fusing the mask feature data with the weight set.

[0055] Specifically, firstly, based on the initial credit parameters, a corresponding weight is assigned to each piece of mask feature data. All these weights together form a weight set. The logic for weight allocation is that the higher the initial credit parameters of an entity, the greater the proportion of its contributed data in the aggregation process. For example, the... The weights of each entity can be calculated by normalizing its initial credit value to the total credit value of all participating entities. Next, a weighted fusion operation is performed, multiplying the masked feature data of each entity by its corresponding weight, and then summing all the products to calculate the final global collaborative result. (The calculation of the global collaborative result is described in the original text.) and weight set assigned to the first Weight of each entity ,have:

[0056] ;

[0057] in, Then from the first The mask feature data received by each entity; This is the sum of the initial credit scores of all participating entities. This calculation allows data scattered across various entities to be organically combined according to their credibility, forming a unified and representative collaborative output. By introducing a weighting mechanism based on initial credit parameters, the data aggregation process is no longer a simple, undifferentiated averaging, but an intelligent and biased fusion. This plays a more decisive role in the formation of the global collaborative result, improving its accuracy and reliability. Simultaneously, it suppresses the negative interference of low-quality or unreliable data from low-credit entities on the final result, enhancing the robustness of the entire distributed collaboration and the value of the final output.

[0058] Optionally, the evaluation of the effectiveness based on the global collaboration results, the generation of an effectiveness score, and the combination of the effectiveness score with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules include:

[0059] Continuously monitor the communication status of the federal control network and obtain network performance parameters including latency, packet loss rate, and throughput;

[0060] The global collaboration results are subjected to consistency verification and error analysis. Combined with historical collaboration effect data, an effect score reflecting the effectiveness of this collaboration is generated.

[0061] Based on the performance score and the network performance parameters, the strategy adjustment amount is calculated using a preset multi-objective optimization function;

[0062] Based on the policy adjustment amount, modify the complexity control parameters in the dynamic mask generation strategy, and synchronously update the reward and penalty ranges in the credit allocation rules to generate an updated collaborative strategy.

[0063] Specifically, firstly, monitoring agents deployed in the federal control network continuously monitor the communication status, thereby acquiring a set of network performance parameters in real time. These parameters are multi-dimensional, including latency (measuring data round-trip time), packet loss rate (reflecting data transmission reliability), and throughput (characterizing effective data transmission rate). Simultaneously, the effectiveness of the global collaborative results generated in the previous round is deeply evaluated. This evaluation includes two levels: first, through consistency verification and error analysis, the global collaborative results are compared with a preset benchmark or theoretical optimal solution to quantify their accuracy; second, by comparing with historical collaborative effect data longitudinally, performance fluctuation trends are analyzed. Combining these two levels of analysis, a quantitative performance score is generated for this collaboration. Next, the acquired network performance parameters and the generated performance score are fed into a preset multi-objective optimization function to calculate the policy adjustment amount. This multi-objective optimization function aims to find a balance point that maximizes the collaborative performance score while minimizing network resource consumption. The specific form of the multi-objective optimization function can be a trained reinforcement learning model whose goal is to output the optimal policy adjustment instruction based on the current performance and network state. Using performance scores and network performance parameters as state inputs and policy parameter adjustments as action outputs, a model such as a deep Q-network is trained using historical collaborative data. This allows the model to learn policy adjustment decisions that maximize long-term cumulative rewards in complex environments. Finally, based on the calculated policy adjustment amounts, two core policies are modified simultaneously. Firstly, the complexity control parameters in the dynamic mask generation policy are modified, such as adjusting the weights of credit and network state in the dynamic mask strength calculation, or directly adjusting the sparsity of the dynamic mask to adapt to the current network capacity. Secondly, the reward and penalty magnitudes in the credit allocation rules are updated simultaneously; that is, based on the effect of this collaboration, the credit reward for entities contributing high-quality data is increased, or the credit penalty for entities performing poorly is increased. After these two adjustments, an updated collaborative policy adapted to the current state is generated to guide the next round of data collaboration. Figure 3 As shown, the heatmap visually displays the adjustment of different strategy parameters in multiple rounds of collaboration, with color intensity indicating the magnitude of adjustment and positive or negative values ​​representing enhancement or weakening. It can adjust its core operating strategy in real time and automatically based on the actual collaborative effect and network environment. This closed-loop feedback mechanism continuously seeks a balance between collaborative effect and resource consumption under changing network conditions, thus maintaining stability and robustness in long-term operation and ultimately improving the continuously optimized global collaborative performance.

[0064] Optionally, the calculation of the strength coefficient of the dynamic mask includes:

[0065] Based on the initial credit parameters, a quantitative analysis is performed to determine the priority level of the entity;

[0066] A mask reference value is obtained by jointly mapping the priority level with the network state parameters.

[0067] The mask reference value is dynamically adjusted based on the network state parameters to generate the strength coefficient of the dynamic mask.

[0068] Specifically, the process begins with quantitative analysis based on an entity's initial credit parameters to determine its priority level. This involves comparing each entity's initial credit value with a set of preset credit thresholds, classifying it into different priority levels such as "high," "medium," or "low" according to its credit range. This classification provides a foundational, macro-level categorization for subsequent processing. After determining the priority level, a joint mapping is performed between this priority level and real-time acquired network state parameters to obtain a mask baseline value. This joint mapping can be understood as a two-dimensional lookup table or a decision function, with discrete priority levels and quantified network state parameters as inputs, and an initial mask strength baseline as the output. For example, a "high-priority" entity will be mapped to a lower mask baseline value when the network state is good; conversely, a "low-priority" entity may be mapped to a relatively high mask baseline value even when the network state is good. This mask baseline value establishes the basic level of mask strength under specific trust and macro-level network conditions. Finally, the mask baseline value is dynamically fine-tuned based on real-time, continuously changing network state parameters, ultimately generating the strength coefficients of the dynamic mask. For calculating the... Dynamic mask strength coefficients generated by each entity ,have:

[0069] ;

[0070] in, It is the mask reference value obtained in the previous step through joint mapping; It is based on real-time network performance parameters The calculated adjustment factor It includes continuously changing metrics such as network latency and bandwidth. (Function) The design ensures that when network conditions deteriorate, such as increased latency or reduced bandwidth, the output value is positive, thus increasing the final strength coefficient; conversely, when network conditions improve, the output value is negative or approaches zero, thus decreasing or maintaining the strength coefficient. The functional form can be determined through multi-objective optimization modeling and simulation parameter tuning. Through this series of steps, a dynamic mask strength coefficient is obtained that reflects both the inherent trustworthiness of entities and responds instantly to fluctuations in the network environment. By introducing two intermediate layers—priority levels and mask baseline values—this method clearly separates static, trust-based policy setting from dynamic, real-time adjustments based on the network environment. This hierarchical control not only makes policy formulation and adjustment clearer and more interpretable but also reduces the nonlinear effects and parameter sensitivity problems that may result from simply mixing credit and network state parameters in calculations.

[0071] Optionally, the generated weight set includes:

[0072] The credit weight of each entity is calculated based on the initial credit parameters.

[0073] Obtain network state parameters for quantifying communication quality, and adjust the credit weight of each entity based on the network state parameters to generate an adaptive weight value;

[0074] The adaptive weight values ​​of all entities are aggregated to form a weight set.

[0075] Specifically, firstly, based on predetermined initial credit parameters, a credit weight is calculated for each participating entity. This calculation is typically performed by normalizing the initial credit parameters of all entities, ensuring that high-credit entities receive relatively high base weights, with the sum of each entity's credit weights being 1. This step establishes a static benchmark for weight allocation, reflecting the long-term, inherent reliability of the entities. Subsequently, network state parameters used to quantify communication quality are acquired in real time. These network state parameters typically include link bandwidth, latency, and packet loss rate, and these heterogeneous network state parameters are merged into a unified, dimensionless communication quality adjustment factor. This adjustment factor is then used to dynamically adjust the credit weight of each entity, generating an adaptive weight value. This adjustment process multiplies each entity's credit weight by its corresponding communication quality adjustment factor to obtain the entity's adaptive weight value. Specifically, each entity's credit weight is derived from a normalized calculation of its initial credit parameters, reflecting its long-term credibility. The communication quality adjustment factor, on the other hand, is a positive number dynamically calculated based on real-time network state parameters, reflecting the entity's current transmission conditions. When network quality is better than a set benchmark, the communication quality adjustment factor is greater than 1, increasing the weight; when network quality deteriorates, the communication quality adjustment factor is less than 1, decreasing the weight accordingly. The calculation of the communication quality adjustment factor is based on normalized network parameters, allowing for a unified and comprehensive evaluation of indicators with different dimensions, such as bandwidth and latency. After calculating the adaptive weight values ​​for all entities, normalization is performed again to ensure that the final sum of adaptive weight values ​​is 1. Finally, the final normalized adaptive weight values ​​of all entities are aggregated to form a complete weight set that can be directly used for weighted aggregation. By upgrading the weight allocation for data aggregation from a single-dimensional, static evaluation to a multi-dimensional, dynamically perceptive decision-making process, it not only reflects the entity's historical credibility but also instantly reflects its current data transmission capabilities. When a high-credit entity encounters temporary network congestion or connection instability, its weight is dynamically reduced to minimize the negative impact of potential delays or incomplete data on the overall collaborative outcome. Conversely, entities with good network conditions receive higher weights, ensuring their high-quality data contributions are fully utilized. This adaptive weight adjustment enhances the robustness of distributed data collaboration to network environment fluctuations, improving the overall timeliness and accuracy of the final results.

[0076] Optionally, the step of performing quantitative analysis based on the initial credit parameters to determine the entity's priority level includes:

[0077] The initial credit parameters are input into the credit rating unit, and multiple credit intervals are divided according to the credit value.

[0078] Match the corresponding priority label based on the credit range to which each entity belongs;

[0079] Historical collaborative behavior data is obtained, and the priority labels are fine-tuned based on the historical collaborative behavior data to form a priority level.

[0080] Specifically, the initial credit parameter of an entity, i.e., a quantified credit score, is first input into a dedicated credit rating unit. This unit has multiple preset credit intervals, dividing the complete credit score range through a series of thresholds, such as 0 to 60, 61 to 85, and 86 to 100. The entity's credit score is matched against these credit intervals, and based on the specific interval it falls into, an initial priority label is assigned, such as "low," "medium," or "high." This step completes the preliminary classification based on static attributes. However, this priority label is not the final result; further analysis of the entity's historical collaborative behavior data is conducted. This data records the entity's actual performance in past collaborative tasks, including but not limited to the timeliness of data contributions, the quality assessment of submitted data, and records of compliance with collaborative protocols. Based on this historical collaborative behavior data, the initially assigned priority label is dynamically fine-tuned. For example, an entity initially labeled "high" may have its label temporarily downgraded if its recent behavior data shows frequent late submissions or a decline in data quality. Conversely, an entity labeled "medium" but with consistently excellent performance may have its label upgraded. Through this fine-tuning combined with historical behavior, the initial, static priority label is calibrated into a final priority level that more comprehensively and dynamically reflects the entity's current overall state. For example... Figure 4 As shown in the figure, the horizontal axis represents the entity ID, and the vertical axis represents the initial credit value. The initial credit parameters are input into a credit rating unit. Using preset credit thresholds, namely the "low / medium threshold" and "medium / high threshold" shown in the figure, the initial credit value is divided into multiple credit intervals such as "low," "medium," and "high." Each entity is assigned a priority based on the interval in which its initial credit value falls. By combining static, attribute-based initial credit assessment with dynamic, contribution-based behavioral history, this system reduces the lag and bias that may result from relying solely on static credit. This dual assessment mechanism ensures that the priority level not only reflects an entity's basic capabilities but also its true attitude and long-term value within the collaborative network. It can promptly identify and reward entities that consistently make high-quality contributions, while also identifying and restricting poorly performing or potentially malicious entities, thus improving the security and collaborative efficiency of the entire federated control system.

[0081] Based on the same inventive concept, such as Figure 5As shown, the present invention also provides a distributed data collaboration device based on federated control, the device comprising:

[0082] The network credit initialization module is used to obtain network configuration parameters for defining network topology and generate initial credit parameters for multiple entities based on the network configuration parameters.

[0083] The dynamic mask generation module is used to acquire local data and perform feature extraction to generate feature data, combine the initial credit parameters with the real-time acquired network state parameters to generate a dynamic mask, and use the dynamic mask to process the feature data to obtain mask feature data.

[0084] The weighted aggregation and collaboration module is used to perform weighted aggregation on the mask feature data received from each entity according to the initial credit parameters, and generate a global collaboration result.

[0085] The strategy evaluation and update module is used to evaluate the effectiveness of the global collaboration results, generate an effect score, and combine the effect score with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules.

[0086] To verify the feasibility of this invention in practice, it was applied to a distributed industrial IoT predictive maintenance system in a manufacturing enterprise. The enterprise aims to construct a high-precision global fault prediction model by collaboratively analyzing the operating data of multiple CNC machine tools distributed throughout the factory, in order to detect potential equipment failures in advance.

[0087] Traditional collaborative methods typically perform simple average aggregation of data submitted by all edge computing nodes, ignoring the significant differences in hardware capabilities, network environment, and historical data quality among nodes. This results in low global model accuracy and low collaborative efficiency when the network is congested, making it difficult to meet the high-precision and high-timeliness requirements of industrial production.

[0088] In this embodiment, the enterprise deploys the distributed data collaboration system based on federated control proposed in this invention. The system first uses a network credit initialization module to generate an initial credit parameter for each edge computing node based on its hardware configuration (e.g., CPU performance, memory capacity) and network access capabilities (e.g., nominal bandwidth). During data collaboration, each node uses a dynamic mask generation module to extract features from the collected machine tool vibration and temperature data locally. This feature is then combined with its initial credit parameter and real-time monitored network status parameters (e.g., network latency, bandwidth) to generate a dynamic mask, which in turn generates mask feature data. Subsequently, the weighted aggregation collaboration module of the central collaboration server performs weighted aggregation of all received mask feature data based on the initial credit parameters of each node, generating a global fault prediction model. Finally, the policy evaluation and update module evaluates the prediction accuracy of the model, generates an effect score, and, combined with real-time acquired network performance parameters (e.g., latency, packet loss rate), performs closed-loop feedback and adaptive updates to the dynamic mask generation strategy and credit allocation rules.

[0089] To verify the beneficial effects of this invention, we conducted a six-month comparative test. The experimental group used the method of this invention and selected three representative edge computing nodes, namely nodes A, B, and C, for focused observation; the control group used the traditional indiscriminate average aggregation method. During the test, we recorded in detail the credit rating of each node, the changes in dynamic mask strength, the prediction accuracy of the global model, and the network resource consumption.

[0090] During the system initialization phase, the network credit initialization module evaluates each node. For example, node A, deployed on a high-performance server and accessed via a gigabit wired network, has an initial credit parameter of 95; node B, deployed on a standard industrial control computer and accessed via industrial WiFi, has an initial credit parameter of 70; while node C, being an older device with unstable performance and network connectivity, has an initial credit parameter of only 55.

[0091] The dynamic mask generation module plays a crucial role in collaborative task execution. During a collaborative task at 10:30 AM on April 15, 2025, severe interference was detected in the wireless network environment where Node B was located. Network latency surged to 180ms, and bandwidth dropped to 30% of usable value. The system acquired the network status parameters in real time and, combined with Node B's initial credit parameters, calculated a dynamic mask strength coefficient as high as 0.88. This coefficient was used to strongly mask the feature data uploaded by Node B, reducing the negative impact of its low-quality data on the overall result. Meanwhile, Node A, with a stable network environment and a latency of 12ms, had a dynamic mask strength coefficient of only 0.25, thus enhancing the contribution of its high-quality data.

[0092] In the weighted aggregation and collaboration phase, the system assigns weights to each node based on initial credit parameters. For example, the data weight contributed by node A is set to 0.22, node B to 0.11, and node C to only 0.06. Through this trust-based weighted aggregation, the global model generated by this invention achieved a fault prediction accuracy of 91.5% in the first round of testing, while the control group using simple average aggregation only achieved an accuracy of 82.3%.

[0093] The system's strategy evaluation and update module demonstrated its closed-loop adaptive capability. After the first month of operation, the system evaluated the effectiveness of the global collaborative results at a score of 92, while also detecting a slight increase in the overall packet loss rate of the factory network due to busy production tasks. Based on this effectiveness score and network performance parameters, the system calculated the strategy adjustment amount through a multi-objective optimization function and generated an updated collaborative strategy. This strategy increased the complexity control parameter in the dynamic mask generation strategy to adapt to the current poor network environment; simultaneously, because node A consistently contributed high-quality data over the past month, its credit score was rewarded and increased to 98, while the poorly performing node C was penalized and its score was reduced to 51. After this round of strategy updates, the system's model accuracy further improved in the following month's collaborative tasks, and network bandwidth usage decreased compared to the previous month.

[0094] After six months of continuous operation and optimization, the experimental group using this invention achieved a stable global model fault prediction accuracy of over 96%, which is superior to the control group's accuracy of around 85%. Simultaneously, the system's adaptive mechanism effectively balanced collaborative effects with resource overhead, and the overall network bandwidth consumption was reduced compared to transmission methods without policy control.

[0095] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0096] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A distributed data collaboration method based on federated control, characterized in that: Obtain network configuration parameters for defining the network topology, and generate initial credit parameters for multiple entities based on the network configuration parameters; Local data is acquired and features are extracted to generate feature data. The initial credit parameters and real-time network state parameters are combined to generate a dynamic mask. The feature data is then processed using the dynamic mask to obtain mask feature data. The mask feature data received from each entity is weighted and aggregated according to the initial credit parameters to generate a global collaborative result; The effectiveness of the global collaboration results is evaluated to generate an effect score. The effect score is then combined with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules.

2. The distributed data collaboration method based on federated control according to claim 1, characterized in that, The process of generating initial credit parameters for multiple entities based on the network configuration parameters includes: Extract entity attribute parameters of multiple entities from the network configuration parameters; Credit quantification is performed based on the entity attribute parameters to calculate an initial credit value for each entity. The initial credit values ​​of all entities are combined to form initial credit parameters.

3. The distributed data collaboration method based on federated control according to claim 1, characterized in that, The process of generating a dynamic mask by combining the initial credit parameters with the real-time acquired network state parameters includes: Real-time monitoring of communication link status to obtain network status parameters including network latency and bandwidth parameters; The strength coefficient of the dynamic mask is calculated by jointly processing the network state parameters and the initial credit parameters. A dynamic mask is generated by randomly sampling based on the strength coefficient of the dynamic mask.

4. The distributed data collaboration method based on federated control according to claim 3, characterized in that, The obtained mask feature data includes: The feature data is numerically scaled based on the strength coefficient of the dynamic mask to generate scaled feature data; The scaling feature data and the dynamic mask are XORed bitwise to obtain the mask feature data.

5. The distributed data collaboration method based on federated control according to claim 1, characterized in that, The generation of global collaborative results includes: A weight set is generated by assigning weights to the mask feature data of each entity based on the initial credit parameters. The global collaborative result is calculated by weighting and fusing the mask feature data with the weight set.

6. The distributed data collaboration method based on federated control according to claim 1, characterized in that, The evaluation of the effectiveness of the global collaboration results, the generation of an effectiveness score, and the combination of the effectiveness score with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules include: Continuously monitor the communication status of the federal control network and obtain network performance parameters including latency, packet loss rate, and throughput; The global collaboration results are subjected to consistency verification and error analysis. Combined with historical collaboration effect data, an effect score reflecting the effectiveness of this collaboration is generated. Based on the performance score and the network performance parameters, the strategy adjustment amount is calculated using a preset multi-objective optimization function; Based on the policy adjustment amount, modify the complexity control parameters in the dynamic mask generation strategy, and synchronously update the reward and penalty ranges in the credit allocation rules to generate an updated collaborative strategy.

7. The distributed data collaboration method based on federated control according to claim 3, characterized in that, The calculation of the strength coefficient of the dynamic mask includes: Based on the initial credit parameters, a quantitative analysis is performed to determine the priority level of the entity; A mask reference value is obtained by jointly mapping the priority level with the network state parameters. The mask reference value is dynamically adjusted based on the network state parameters to generate the strength coefficient of the dynamic mask.

8. A distributed data collaboration method based on federated control according to claim 5, characterized in that, The generated weight set includes: The credit weight of each entity is calculated based on the initial credit parameters. Obtain network state parameters for quantifying communication quality, and adjust the credit weight of each entity based on the network state parameters to generate an adaptive weight value; The adaptive weight values ​​of all entities are aggregated to form a weight set.

9. A distributed data collaboration method based on federated control according to claim 7, characterized in that, The quantitative analysis based on the initial credit parameters to determine the entity's priority level includes: The initial credit parameters are input into the credit rating unit, and multiple credit intervals are divided according to the credit value. Match the corresponding priority label based on the credit range to which each entity belongs; Historical collaborative behavior data is obtained, and the priority labels are fine-tuned based on the historical collaborative behavior data to form a priority level.

10. A distributed data collaboration device based on federated control, applied to a distributed data collaboration method based on federated control as described in any one of claims 1-9, characterized in that, The device includes: The network credit initialization module is used to obtain network configuration parameters for defining network topology and generate initial credit parameters for multiple entities based on the network configuration parameters. The dynamic mask generation module is used to acquire local data and perform feature extraction to generate feature data, combine the initial credit parameters with the real-time acquired network state parameters to generate a dynamic mask, and use the dynamic mask to process the feature data to obtain mask feature data. The weighted aggregation and collaboration module is used to perform weighted aggregation on the mask feature data received from each entity according to the initial credit parameters, and generate a global collaboration result. The strategy evaluation and update module is used to evaluate the effectiveness of the global collaboration results, generate an effect score, and combine the effect score with the obtained network performance parameters to generate an updated collaboration strategy for adjusting the preset dynamic mask generation strategy and the preset credit allocation rules.

Images