A decentralized gateway control method and system

Abstract

The present invention relates to the field of gateway control technology, and specifically to a decentralized gateway control method and system. The method allocates a mobile device node to a specific grid according to its physical location coordinates, and sends a status registration request to an edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts non-sensitive features of local original sensitive data to generate an encrypted feature vector and performs lineage dependency path verification through a device lineage relationship library; for the encrypted feature vector that passes the lineage dependency path verification, the encrypted feature vector is synchronized to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs a task scheduling operation; thereby solving the problem of dynamic and secure access to and efficient synchronization of sensitive data when mobile devices operate across regions.

Description

Technical Field

[0001] The present invention relates to the field of gateway control technology, and in particular to a decentralized gateway control method and system. Background Art

[0002] Decentralized gateways play a key role in industrial scenarios, serving as a communication bridge between mobile devices and clusters of fixed devices. They are fully responsible for data forwarding, synchronization, and access control, effectively avoiding over-reliance on a single central gateway. In traditional architectures centered around a single central gateway, a failure of that central gateway can paralyze the entire system, leading to severe consequences such as production stagnation and data loss. Decentralized gateways, on the other hand, distribute data processing and control functions to individual nodes, greatly improving system reliability and flexibility.

[0003] However, in dynamic industrial scenarios involving mobile devices, such as automated logistics and flexible manufacturing lines, these devices frequently switch workspaces and interact with different clusters of fixed equipment, posing dual challenges to both real-time and secure data access. Traditional static data synchronization mechanisms are particularly incapable of adapting to dynamic topology changes when mobile devices operate across multiple regions. Frequent full data transmission can significantly consume network bandwidth, easily causing network congestion and impacting the normal communication and data transmission of other devices. Reducing the data transmission frequency to avoid network congestion can lead to data update delays, preventing mobile devices from obtaining the latest data information in a timely manner. This can potentially cause production accidents, such as equipment operation errors and disrupted production processes.

[0004] Therefore, there is an urgent need for a decentralized gateway control method that can achieve dynamic and secure access and efficient synchronization of sensitive data by mobile devices, and effectively protect data privacy and real-time requirements. Summary of the Invention

[0005] (1) Technical problems to be solved:

[0006] The purpose of the present invention is to provide a decentralized gateway control method and system to solve the problem of dynamic and secure access and efficient synchronization of sensitive data when mobile devices operate across regions in industrial dynamic scenarios.

[0007] (2) Technical solution:

[0008] To achieve the above objectives, the present invention provides a decentralized gateway control method, which includes:

[0009] S1. Divide the physical space of the industrial site into a gridded space model, associate each grid with an equipment cluster within the workstation area, configure the gateway of the equipment cluster as an edge computing node, and store the original sensitive data in the local storage unit of the physical device that generates the original sensitive data.

[0010] S2. When a mobile device node comes online, it is dynamically assigned to a specific grid in the corresponding workstation area according to the physical location coordinates of the mobile device node, and a status registration request is sent to the edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts the non-sensitive features of the local original sensitive data and generates an encrypted feature vector bound to the grid identifier to which the device belongs; and the encrypted feature vector is marked as a data set to be synchronized.

[0011] S3. Verify the bloodline dependency path of each encrypted feature vector in the data set to be synchronized through a device bloodline relationship library; the device bloodline relationship library records the child device identifier, host device identifier, association type and the last valid timestamp.

[0012] S4. For the encrypted feature vector verified by the bloodline dependency path, calculate the final change frequency of the encrypted feature vector within the preset time window, and synchronize the encrypted feature vector to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation.

[0013] Furthermore, the method of dividing the physical space of the industrial site into a gridded space model and associating each grid with a device cluster within a workstation area includes:

[0014] The minimum outer bounding rectangle is calculated based on the actual physical boundary coordinates of the workstation area, and the grid division density is determined based on the equipment communication radius and the correlation between the workstation process; a coordinate system is established based on the minimum outer bounding rectangle, and equal-sized grid units are generated along the coordinate axis according to the division density.

[0015] Traverse the set of grid cells corresponding to each workstation area. If a single workstation area is divided into more than three grid cells, trigger the grid merging operation until each workstation area is associated with at most two grid cells; bind the merged grid cells to the workstation area. When the physical location of the equipment cluster completely falls into one of the grid cells, automatically establish the ownership relationship between the equipment cluster and the corresponding grid cell; detect the position change of the equipment cluster in real time. If the center point of the equipment cluster offsets beyond the boundary of the grid cell to which it belongs, trigger the grid cell rebinding operation.

[0016] Furthermore, when the mobile device node comes online, the method of dynamically allocating the mobile device node to a specific grid corresponding to the workstation area according to the physical location coordinates of the mobile device node and sending a status registration request to the edge computing node associated with the specific grid includes:

[0017] The physical location coordinates are collected in real time through the built-in positioning module of the mobile device node; the physical location coordinates are spatially matched through the preloaded workstation area topology map. If the physical location coordinates fall within the geometric polygon range of a certain workstation area, the workstation area is marked as a to-be-selected workstation area.

[0018] When the number of workstation areas to be selected is greater than the preset area number threshold, the workstation area where the movable device node has stayed for the longest time in history is selected as the target workstation area; based on the mapping relationship between the center point coordinates of the target workstation area and the gridded space model, the specific grid to which the movable device node belongs is determined; and a status registration request carrying a device identifier and timestamp is sent to the edge computing node associated with the specific grid.

[0019] Furthermore, the method of performing lineage dependency path verification on each encrypted feature vector in the data set to be synchronized through a device lineage relationship library includes:

[0020] Parse the grid identifier and device identifier bound to the encrypted feature vector as the host device identifier to be verified; obtain the node identifier of the removable device that initiates the status registration request; retrieve all associated entries in the device lineage relationship library whose sub-device identifiers are equal to the node identifier of the removable device; extract the host device identifier set and association type recorded in the associated entry.

[0021] When the host device identifier to be verified exists in the host device identifier set and the association type belongs to the critical business dependency type, if the time difference between the current time and the last valid timestamp in the association entry is less than the preset validity period threshold, it is determined that the bloodline dependency path verification is passed.

[0022] Furthermore, the method for calculating the final change frequency of the encrypted feature vector within a preset time window includes:

[0023] A set of version snapshots of the original sensitive data recorded at historical time points is obtained from the local storage unit of the edge computing node; a difference measurement value is obtained by differential analysis of adjacent version snapshots in the version snapshot set according to the time sequence; when the difference measurement value exceeds a preset difference measurement threshold, it is marked as a valid change event.

[0024] The basic change frequency is calculated based on the number of occurrences of all valid change events within the preset time window and the preset time window length.

[0025] If the original sensitive data is continuous time series data, the fluctuation amplitude weight factor is calculated based on the numerical fluctuation range of adjacent version snapshots; if the original sensitive data is a discrete configuration parameter, the change density compensation coefficient is calculated based on the distribution density of valid change events; the product of the basic change frequency and the corresponding fluctuation amplitude weight factor or change density compensation coefficient is used as the final change frequency.

[0026] Furthermore, the method of synchronizing the encrypted feature vector to the removable device node according to the final change frequency includes:

[0027] The current task type, device operation status code and remaining task duration reported by the mobile device node are obtained in real time, and the real-time working condition data stream of the host device verified by the lineage dependency path is collected at the same time; the historical final change frequency sequence corresponding to the real-time working condition data stream, device operation status code and encrypted feature vector is input into the pre-trained timing prediction network, and multi-dimensional feature fusion is performed in combination with the current task type and remaining task duration to output the predicted change frequency value.

[0028] When the predicted change frequency value exceeds the preset change frequency threshold, active pre-synchronization of the encrypted feature vector is initiated to the mobile device node; a pre-synchronization data cache area is established locally in the mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp.

[0029] Furthermore, the method of inputting the real-time working condition data stream, the equipment operating status code and the historical final change frequency sequence corresponding to the encrypted feature vector into a pre-trained time series prediction network and performing multi-dimensional feature fusion in combination with the current task type and the remaining task duration includes:

[0030] The real-time working condition data stream is extracted through spatial features to obtain a first feature tensor; the historical final change frequency sequence is extracted through temporal features to obtain a second feature tensor; the first feature tensor and the second feature tensor are spliced ​​in the channel dimension to form a dynamic feature fusion tensor.

[0031] The current task type is encoded into a task type vector, and the device operation status code and the remaining task duration value are normalized and merged into a device state vector; the task type vector and the device state vector are spliced ​​to form a static parameter vector; the dynamic feature fusion tensor and the static parameter vector are subjected to nonlinear feature interaction to generate a predicted change frequency value.

[0032] Furthermore, the method of establishing a pre-synchronization data buffer area locally in the mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp includes:

[0033] An independent storage partition is created for each actively pre-synchronized encrypted feature vector and associated with a dynamic access credential, wherein the dynamic access credential is generated by a hash operation of the pre-synchronization timestamp and the identifier of the edge computing node that initiates the pre-synchronization; when the movable device node needs to call the encrypted feature vector to perform a task scheduling operation, an access request containing the current task type and the device operation status code is submitted to the pre-synchronization data cache; when the task type in the access request is consistent with the current task type recorded when the pre-synchronization is initiated, and the device operation status code is in the access-allowed state range, the dynamic access credential is activated and a temporary decryption key is generated, and the target encrypted feature vector is decrypted using the temporary decryption key.

[0034] On the other hand, based on the same inventive concept, the present invention also provides a decentralized gateway control system, which includes: a grid space modeling and gateway configuration module, an encrypted feature vector generation module, a blood dependency path verification module, and an encrypted feature vector synchronization and task scheduling module, wherein each module is sequentially connected to each other;

[0035] The gridded space modeling and gateway configuration module is used to divide the physical space of the industrial site into a gridded space model, associate each grid with the equipment cluster within the workstation area, configure the gateway of the equipment cluster as an edge computing node, and store the original sensitive data in the local storage unit of the physical device that generated the original sensitive data.

[0036] The encrypted feature vector generation module is used to dynamically allocate the mobile device node to a specific grid in the corresponding workstation area according to the physical location coordinates of the mobile device node when the node comes online, and send a status registration request to the edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts the non-sensitive features of the local original sensitive data and generates an encrypted feature vector bound to the grid identifier to which the device belongs; and marks the encrypted feature vector as a data set to be synchronized.

[0037] The lineage dependency path verification module is used to verify the lineage dependency path of each encrypted feature vector in the data set to be synchronized through the device lineage relationship library; the device lineage relationship library records the sub-device identifier, host device identifier, association type and the last valid timestamp. The sub-device is a movable device node, and the host device is a physical device fixedly deployed in the grid.

[0038] The encrypted feature vector synchronization and task scheduling module is used to calculate the final change frequency of the encrypted feature vector within a preset time window for the encrypted feature vector verified by the bloodline dependency path, and synchronize the encrypted feature vector to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation.

[0039] (3) Beneficial effects:

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. By storing original sensitive data locally, transmitting only non-sensitive feature encryption vectors, and utilizing the device lineage library for dependency verification, the risk of sensitive data leakage is effectively reduced, ensuring data privacy and security in dynamic interaction scenarios.

[0042] 2. Through a gridded spatial model and a change frequency prediction mechanism, on-demand synchronization and proactive pre-synchronization of encrypted feature vectors are achieved, avoiding network congestion caused by full transmission while ensuring real-time data updates and adapting to the dynamic topology changes of mobile devices operating across regions.

[0043] 3. By predicting the change frequency through multi-dimensional feature fusion and combining it with dynamic access control of the pre-synchronized cache area, mobile devices can efficiently decrypt data according to real-time working conditions and task requirements, reducing production accidents caused by data lag and improving overall collaborative control capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a flowchart of a decentralized gateway control method according to Example 1 of the present invention.

[0045] Figure 2 This is a schematic diagram of the module composition of a decentralized gateway control system according to Example 2 of the present invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0047] Before giving examples, it is necessary to explain the application scenarios of the present invention. The present invention is a decentralized gateway control method and system, which is used in dynamic industrial scenarios where mobile devices frequently switch work areas and interact with different fixed equipment clusters, and there are problems with the security and real-time performance of data access.

[0048] Example 1: Figure 1 As shown, this embodiment provides a decentralized gateway control method, the method comprising:

[0049] S1. Divide the physical space of the industrial site into a gridded spatial model. Associate each grid with a cluster of equipment within a workstation area. Configure the gateway of each equipment cluster as an edge computing node, and store raw sensitive data in the local storage unit of the physical device that generated the raw sensitive data. Gridding the industrial site facilitates the management of equipment and data within each workstation area. For example, the physical space of a production workshop can be gridded according to the actual workstation layout. Each grid corresponds to a specific workstation area, such as welding, assembly, and testing. Within each workstation area, several fixed devices, including robotic arms, testing equipment, and sensors, are deployed. These devices form an equipment cluster. The gateway device in each equipment cluster is configured as an edge computing node with data processing and communication capabilities. This allows for real-time processing of data generated by the equipment, enabling rapid decision-making and response. Raw sensitive data is not stored in a traditional centralized manner. For example, the robotic arm at a welding station stores sensitive process data, such as welding parameters, temperature curves, and current and voltage data, on its built-in industrial-grade solid-state drive. The torque sensor at an assembly station stores key data, such as torque values ​​and assembly sequences, in a local storage unit.

[0050] S2. When a mobile device node comes online, it is dynamically assigned to a specific grid in the corresponding workstation area according to the physical location coordinates of the mobile device node, and a status registration request is sent to the edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts non-sensitive features of the local original sensitive data, generates an encrypted feature vector bound to the grid identifier to which the device belongs, and marks the encrypted feature vector as a data set to be synchronized; after receiving the status registration request, the edge computing node immediately responds and starts the data processing process. The edge computing node extracts non-sensitive features from various types of locally stored original sensitive data. Non-sensitive features refer to information extracted from the original sensitive data that does not contain sensitive information but can reflect the characteristics and status of the data. For example, non-sensitive features such as the average value of the welding current and the range of the welding time are extracted from the original sensitive data of the welding equipment. These features can be used to describe the operating status of the welding equipment but will not leak sensitive production parameters. For numerical process parameters, statistically non-sensitive features such as mean, variance, and trend slope are extracted. For time series data, frequency domain features such as primary frequency components and power spectrum density are extracted. For configuration parameters, structural features such as parameter type, value range, and update frequency are extracted. Non-sensitive features are processed using an encryption algorithm to generate an encrypted feature vector, which is then bound to the current grid identifier.

[0051] S3. Verify the bloodline dependency path of each encrypted feature vector in the data set to be synchronized through the device bloodline relationship library; the device bloodline relationship library records the sub-device identifier, host device identifier, association type and the last valid timestamp; a sub-device refers to a device that depends on other devices to run, and a host device is a device that the sub-device depends on.

[0052] S4. For the encrypted feature vector verified by the blood dependency path, calculate the final change frequency of the encrypted feature vector within the preset time window, and synchronize the encrypted feature vector to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation. The mobile device node uses the corresponding decryption key to decrypt the received encrypted feature vector and obtain the data information therein. Then, based on this data information, the mobile device node can perform corresponding tasks, such as inspection, maintenance, data collection, etc. For example, after the mobile inspection robot decrypts the received encrypted feature vector of the welding equipment, it plans the inspection route according to the welding parameters and status information therein and inspects the welding equipment.

[0053] The method of dividing the physical space of the industrial site into a gridded space model and associating each grid with a device cluster within a workstation area includes:

[0054] The minimum bounding rectangle (MBR) is calculated based on the actual physical boundary coordinates of the workstation area. The grid density is determined based on the equipment communication radius and the process relevance of the workstation. A coordinate system is established based on the MBR, and grid cells of equal size are generated along the coordinate axes according to the density. In industrial sites, each workstation area has clear physical boundaries. For example, the area where a machine tool is located has actual spatial extents such as length, width, and height, and these boundaries can be precisely described by specific coordinates. The MBR can completely encompass the entire workstation area within a regular rectangle, providing a unified and easy-to-calculate reference range for subsequent coordinate system establishment and grid division. For example, by calculating the MBR, an irregularly shaped assembly workstation area can be simplified into a rectangular area for processing, simplifying the grid division complexity. The equipment communication radius determines the range within which effective communication between devices can occur. If the grid is too large, it may lead to increased communication delays or unstable communication between devices. The process relevance of the workstation reflects the closeness of the process connections between the equipment within the workstation area. A reasonable grid density ensures that the equipment within the grid can communicate and collaborate efficiently, while avoiding the problems of increased management costs caused by overly fine grid divisions and low communication and collaboration efficiency caused by overly coarse grid divisions. For example, for equipment clusters with a small communication radius and high process correlation, a finer grid density can be used; while for equipment clusters with a large communication radius and low process correlation, a coarser grid density can be used. The coordinate system provides a unified reference framework for grid division, so that each grid cell can have a clear coordinate position in this coordinate system, and subsequent grid division and positioning can be performed along the coordinate axis. For example, by establishing a rectangular coordinate system with the lower left corner of the minimum enclosing rectangle as the origin, the position of each grid cell within the workstation area can be accurately determined. Grid cells of equal size are easy to manage and calculate, and can ensure the uniformity and consistency of grid division.

[0055] The set of grid cells corresponding to each workstation area is traversed. If a single workstation area is divided into three or more grid cells, a grid merge operation is triggered until each workstation area is associated with at most two grid cells. The merged grid cells are then bound to the workstation area. When the physical location of a device cluster completely falls within one of the grid cells, the affiliation between the device cluster and the corresponding grid cell is automatically established. Changes in the location of the device cluster are detected in real time. If the center point of the device cluster deviates beyond the boundary of the grid cell to which it belongs, a grid cell rebinding operation is triggered. In industrial sites, some workstation areas may have complex shapes. After being divided into equal-sized grids, a single workstation area may be divided into multiple grid cells. For example, if a large assembly workstation area is divided into five grid cells, then the management and scheduling of equipment within the workstation area requires consideration of five different grid cells, increasing management complexity. The merging method can be flexibly adjusted based on the location and shape of the grid cells. For example, adjacent grid cells can be merged or the merge can be optimized based on the actual shape of the workstation area. After merging, the large assembly workstation is ultimately associated with two grid cells, which are then bound to the workstation area. When the physical location of the welding station's robotic arms, sensors, and other equipment clusters completely falls within a grid cell, an affiliation between the equipment cluster and the grid cell is automatically established. Position sensors installed on the equipment monitor changes in the equipment cluster's position in real time. If the center point of the equipment cluster deviates beyond the boundaries of its grid cell, the grid cell to which the equipment cluster belongs must be redefined. This ensures that the affiliation between the equipment cluster and the grid cell remains accurate, ensuring that the equipment cluster can correctly receive and respond to management and scheduling instructions within the grid cell.

[0056] The method of dynamically allocating a mobile device node to a specific grid corresponding to a workstation area according to the physical location coordinates of the mobile device node when the mobile device node comes online, and sending a status registration request to an edge computing node associated with the specific grid includes:

[0057] The built-in positioning module of the mobile device node collects physical location coordinates in real time. These coordinates are then spatially matched against a preloaded workstation area topology map. If the physical coordinates fall within the geometric polygon of a workstation area, the workstation area is marked as a candidate workstation area. Mobile device nodes (such as automated guided vehicles (AGVs) and mobile robots) are typically equipped with various positioning modules, such as global positioning system (GPS), ultra-wideband (UWB), laser positioning, and vision positioning. These positioning modules continuously acquire the location of the mobile device node in physical space and express it in coordinate form, such as two-dimensional (x, y) coordinates. The workstation area topology map is a pre-built map that describes the location, shape, and interrelationships of each workstation area in an industrial site. It can be a two-dimensional or three-dimensional map, depending on the complexity and management requirements of the industrial site. By calculating the positional relationship between the physical coordinates and these polygons, it can be determined whether the device node is located within a specific workstation area. For example, for a rectangular workstation area, the location of the mobile device node can be determined by determining whether its coordinates are within the range defined by the four vertices of the rectangle. For example, if spatial relationship matching finds that the physical location coordinates of the mobile device node fall within the geometric polygons of both the assembly station and the inspection station, both stations will be marked as candidate stations.

[0058] When the number of candidate workstation areas exceeds a preset threshold, the workstation area with the longest historical dwell time of the mobile device node is selected as the target workstation area. Based on the mapping relationship between the center point coordinates of the target workstation area and the gridded spatial model, the specific grid to which the mobile device node belongs is determined. A status registration request carrying a device identifier and timestamp is sent to the edge computing node associated with the specific grid. When the number of candidate workstation areas exceeds the preset threshold, this indicates that the mobile device node's current location may be near the intersection of multiple workstation areas, or there may be a certain positioning error, causing multiple workstation areas to believe that they contain the mobile device node. Querying the mobile device node's historical behavior data reveals that the mobile device node has a historical dwell time of 45 minutes at the assembly station and 20 minutes at the inspection station. Therefore, the assembly station is selected as the target workstation area. Based on the mapping relationship between the center point coordinates of the assembly station and the gridded spatial model, a coordinate transformation formula is used to determine the specific grid to which the mobile device node belongs. For example, the mapping relationship is calculated as follows: grid row number = floor((Y coordinate - Y offset) / grid length), grid column number = floor((X coordinate - X offset) / grid width), to determine the grid to which the mobile device node belongs. The device identifier is used to uniquely identify the mobile device node, making it easier for the edge computing node to identify and manage it; the timestamp records the time when the request is sent, which helps the edge computing node understand the status update time of the mobile device node and ensure the timeliness and accuracy of the data. By sending a status registration request, the mobile device node informs the edge computing node of its existence and current status, so that the edge computing node can keep abreast of the device node's information and include it in the management scope of the grid, thereby realizing decentralized gateway control and management.

[0059] The method of verifying the lineage dependency path of each encrypted feature vector in the to-be-synchronized data set through a device lineage relationship library includes:

[0060] The encrypted feature vector is parsed as the host device identifier to be verified. The node identifier of the mobile device initiating the status registration request is obtained. The device lineage database is searched for all associated entries whose sub-device identifiers are equal to the mobile device node identifier. The host device identifier set and association type recorded in the associated entries are extracted. During the generation and transmission of the encrypted feature vector, key information is bound to it. The grid identifier indicates the grid area to which the encrypted feature vector belongs, and the device identifier identifies the specific device from which it originates. These two identifiers are parsed as the host device identifier to be verified in order to subsequently compare them with information in the device lineage database to confirm whether the encrypted feature vector's source device meets lineage dependency requirements. For example, an encrypted feature vector generated by a sensor device is bound to its grid number and its own device number. Parsing this information can reveal the approximate source range and specific device of the encrypted feature vector. When a mobile device node comes online, it sends a status registration request to the edge computing node associated with the specific grid, including the unique identifier of the mobile device node. This identifier is obtained to search the device lineage database for lineage information related to the mobile device and determine its dependencies with other devices. For example, when an automated guided vehicle (AGV) enters a new work area and goes online, it sends a status registration request, which includes its device ID. This ID can be used to further query its associations with surrounding devices. The device lineage database is a database that stores the lineage dependencies between devices, recording information such as parent-child relationships and association types. Using the mobile device node identifier as a search criterion, all association entries in the device lineage database with that device as a child are searched to find the host device information on which the mobile device node depends. For example, a robotic arm (mobile device node) may depend on a specific control host (host device). By searching the device lineage database, the association entries between them can be found. From the retrieved association entries, a set of host device identifiers is extracted, which contains the identifiers of all host devices that have a lineage dependency with the mobile device node. The association type is also extracted. The association type describes the nature of the dependency between devices, such as whether it is a critical business dependency or a general data interaction dependency. Different association types may be handled differently during data validation and interaction. For example, when data is exchanged between critical business-dependent devices, stricter verification and authorization are required.

[0061] When the host device identifier to be verified exists in the host device identifier set and the association type belongs to the critical business dependency type, if the time difference between the current time and the last valid timestamp in the association entry is less than the preset validity period threshold, it is determined that the lineage dependency path verification has been passed. If the host device identifier to be verified is in the extracted host device identifier set, it means that there is a potential lineage dependency relationship between the device from which the encrypted feature vector comes and the removable device node. For example, if the host device identifier to be verified is a control host, and the control host is in the host device identifier set, the identifier matching condition is met. Then determine whether the association type belongs to the critical business dependency type. The data interaction between devices of the critical business dependency type is usually crucial to the production process and requires more stringent verification. Only when the association type meets the requirements will the next step of verification be carried out. Finally, check whether the time difference between the current time and the last valid timestamp in the association entry is less than the preset validity period threshold. This is to ensure that the lineage dependency relationship between devices is still valid at the current moment. If the time difference is less than the preset validity period threshold, it means that the lineage dependency relationship is within the validity period and can be determined to have passed the lineage dependency path verification; otherwise, the lineage dependency relationship may no longer be valid due to changes in device relationships, time expiration, etc., and the verification fails.

[0062] The method for calculating the final change frequency of the encrypted feature vector within a preset time window includes:

[0063] A set of version snapshots of the original sensitive data recorded at historical points in time is retrieved from the local storage unit of the edge computing node. Adjacent version snapshots in the set are differentially analyzed according to chronological order to obtain a difference metric. When the difference metric exceeds a preset difference metric threshold, a valid change event is marked. During operation, the edge computing node continuously records the state of the original sensitive data at different historical points in time, forming a series of version snapshots and storing them locally. These snapshots act like "historical photos" of the data, fully preserving the characteristics of the data at each moment. For example, data on the welding parameters of a robotic arm is extracted from 288 version snapshots recorded over the past 24 hours. Each version snapshot contains multi-dimensional parameter data such as welding current, voltage, temperature, and speed. The differential analysis process is performed by calculating the Euclidean distance between adjacent version snapshots. The difference metric reflects the degree of change in the data between adjacent time points. The preset difference metric threshold is a standard value set based on the actual application scenario and data characteristics. Only when the calculated difference metric exceeds the preset difference metric threshold is the data considered to have undergone a meaningful change and marked as a valid change event. This is to avoid mistaking minor, insignificant data fluctuations for significant changes.

[0064] The basic change frequency is calculated based on the number of occurrences of all valid change events within the preset time window and the preset time window length.

[0065] If the original sensitive data is continuous time-series data, a fluctuation weighting factor is calculated based on the numerical fluctuation range of adjacent version snapshots. If the original sensitive data is a discrete configuration parameter, a change density compensation coefficient is calculated based on the distribution density of valid change events. The final change frequency is the product of the base change frequency and the corresponding fluctuation weighting factor or change density compensation coefficient. For example, if the temperature fluctuation range between adjacent version snapshots is large, indicating drastic temperature fluctuations, the calculated fluctuation weighting factor will be larger, allowing the final change frequency to better reflect these drastic changes. Discrete configuration parameters typically have a finite number of possible values, such as the on / off state of a device or the operating mode of a system. For example, in a device's operating mode configuration, if there are multiple mode switches (valid change events) within a certain number of hours, but few changes at other times, the calculated change density compensation coefficient will be larger, allowing the final change frequency to better reflect the concentration of these changes.

[0066] The method for synchronizing the encrypted feature vector to the removable device node according to the final change frequency includes:

[0067] The system acquires the current task type, device status code, and remaining task duration reported by the mobile device node in real time. It also collects the host device's real-time operating data stream, verified by a lineage dependency path. The real-time operating data stream, device status code, and historical final change frequency sequence corresponding to the encrypted feature vector are input into a pre-trained time series prediction network. Multi-dimensional feature fusion is performed on the network, combining the current task type and remaining task duration to output a predicted change frequency value. The current task type reflects the task being performed by the device, such as handling or assembly. The device status code indicates the device's specific operating status, such as normal operation, fault, or standby. The remaining task duration indicates the estimated time required for the task to complete. The host device is a device with a lineage dependency relationship to the mobile device node. Its operating data stream contains various real-time data during the device's operation, such as temperature, pressure, and speed. Lineage dependency path verification ensures that the collected host device data has a legitimate association with the mobile device node, ensuring data accuracy and reliability.

[0068] When the predicted change frequency exceeds the preset change frequency threshold, an active pre-synchronization of the encrypted feature vector is initiated for the mobile device node. A pre-synchronization data cache is established locally on the mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp. When the predicted change frequency exceeds the preset change frequency threshold, it indicates that the encrypted feature vector is likely to change in the future. To ensure that the mobile device node can obtain the latest data in a timely manner, active pre-synchronization is required.

[0069] The method of inputting the real-time working condition data stream, the equipment operating status code, and the historical final change frequency sequence corresponding to the encrypted feature vector into a pre-trained time series prediction network and performing multi-dimensional feature fusion in combination with the current task type and the remaining task duration includes:

[0070] The real-time operating condition data stream is extracted through spatial features to obtain a first feature tensor. The historical final change frequency sequence is extracted through temporal features to obtain a second feature tensor. The first and second feature tensors are concatenated in the channel dimension to form a dynamic feature fusion tensor. Spatial feature extraction utilizes a convolutional neural network structure to convert the real-time data stream, which contains multi-dimensional parameters such as current, voltage, temperature, and speed, into the first feature tensor. Temporal feature extraction aims to capture temporal trends and periodicity from historical data. Common methods include recurrent neural networks (RNNs) and long short-term memory (LSTM) networks. Through temporal feature extraction, the historical final change frequency sequence is converted into a second feature tensor that reflects the dynamic changes in the encrypted feature vector change frequency in the temporal dimension. Channel-dimensional concatenation is a common feature fusion method that preserves all information in the original feature tensor while fusing spatial and temporal features to form a more comprehensive and richer dynamic feature fusion tensor.

[0071] The current task type is encoded as a task type vector. The device operation status code and remaining task duration are numerically normalized and then merged into a device state vector. The task type vector and device state vector are concatenated to form a static parameter vector. The dynamic feature fusion tensor is subjected to nonlinear feature interaction with the static parameter vector to generate a predicted change frequency value. The current task type is discrete categorical information. To facilitate fusion and processing with numerical features, it must be encoded as a task type vector. Common encoding methods include one-hot encoding. For example, if there are five different task types, one-hot encoding will represent each task type as a five-dimensional vector with only one dimension set to 1 and the remaining dimensions set to 0. The device operation status code and remaining task duration are numerical data, but their value ranges may differ. To ensure comparability in subsequent calculations, they must be numerically normalized to map their value ranges to a uniform interval (e.g., [0, 1]). The static parameter vector combines the current task characteristics and device operation status information. It changes little over a relatively short period of time and is therefore called a static parameter. Nonlinear feature interaction can be achieved through structures such as fully connected neural networks. By introducing nonlinear factors into the network through nonlinear activation functions (such as ReLU and Sigmoid), the model can learn more complex feature relationships. Through this interaction, the model can comprehensively consider dynamic and static information and more accurately predict the change frequency of encrypted feature vectors.

[0072] The method of establishing a pre-synchronization data buffer area locally on a mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp includes:

[0073] For each actively pre-synchronized encrypted feature vector, a separate storage partition is created and associated with a dynamic access credential. The dynamic access credential is generated by hashing the pre-synchronization timestamp and the edge computing node identifier that initiated the pre-synchronization. When a mobile device node needs to invoke an encrypted feature vector to perform a task scheduling operation, it submits an access request containing the current task type and the device's operating status code to the pre-synchronization data cache. If the task type in the access request matches the current task type recorded when pre-synchronization was initiated, and the device's operating status code is within the permitted access range, the dynamic access credential is activated and a temporary decryption key is generated. This temporary decryption key is used to decrypt the target encrypted feature vector. Creating a separate storage partition for each actively pre-synchronized encrypted feature vector enables isolated data storage. Different encrypted feature vectors may have different uses, security levels, or update frequencies. Separate storage partitions prevent interference between them, improving data management flexibility and security. The irreversible and unique nature of the hash operation ensures the uniqueness and security of the generated dynamic access credential. The pre-synchronization timestamp records the time of data synchronization, and the edge computing node identifier identifies the node that initiated the pre-synchronization. Combining these two elements to generate dynamic access credentials ensures that each pre-synchronization operation has a unique access identifier, enhancing data access security. When performing task scheduling operations, mobile device nodes may need to invoke corresponding encrypted feature vectors based on different task requirements. To ensure the legitimacy and security of the invocation, the node submits an access request to the pre-synchronization data cache, including the current task type and the device's operating status code. The current task type reflects the nature of the task the mobile device node is currently executing or about to execute, while the device's operating status code indicates the node's current operating state. This information helps the cache determine whether to grant the access request. The access request checks whether the task type in the access request matches the current task type recorded when pre-synchronization was initiated. This ensures that the invoked encrypted feature vector is relevant to the current task and prevents improper data use. Secondly, the device's operating status code is checked to ensure that access is permitted. Different device operating states may correspond to different data access permissions. For example, a device in a faulty state may deny access to certain critical data. Only when both conditions are met will the subsequent operation proceed. When the access request is verified, the dynamic access credential is activated. The activated dynamic access credential is used to generate a temporary decryption key. The temporary decryption key is time-limited and can only be used within a specific time period.

[0074] Example 2: Based on the same inventive concept, Figure 2As shown, this embodiment also provides a decentralized gateway control system, which includes: a grid space modeling and gateway configuration module, an encrypted feature vector generation module, a blood dependency path verification module, an encrypted feature vector synchronization and task scheduling module, and each module is sequentially connected in communication;

[0075] The gridded space modeling and gateway configuration module is used to divide the physical space of the industrial site into a gridded space model, associate each grid with the equipment cluster within the workstation area, configure the gateway of the equipment cluster as an edge computing node, and store the original sensitive data in the local storage unit of the physical device that generated the original sensitive data.

[0076] The encrypted feature vector generation module is used to dynamically allocate the mobile device node to a specific grid in the corresponding workstation area according to the physical location coordinates of the mobile device node when the node comes online, and send a status registration request to the edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts the non-sensitive features of the local original sensitive data and generates an encrypted feature vector bound to the grid identifier to which the device belongs; and marks the encrypted feature vector as a data set to be synchronized.

[0077] The lineage dependency path verification module is used to verify the lineage dependency path of each encrypted feature vector in the data set to be synchronized through the device lineage relationship library; the device lineage relationship library records the sub-device identifier, host device identifier, association type and the last valid timestamp. The sub-device is a movable device node, and the host device is a physical device fixedly deployed in the grid.

[0078] The encrypted feature vector synchronization and task scheduling module is used to calculate the final change frequency of the encrypted feature vector within a preset time window for the encrypted feature vector verified by the bloodline dependency path, and synchronize the encrypted feature vector to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation.

[0079] It should be noted that, regarding the system in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0080] Finally, it should be noted that although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A decentralized gateway control method, characterized in that: The method comprises: Dividing the physical space of the industrial site into a gridded spatial model, associating each grid with a device cluster within a workstation area, configuring the gateway of the device cluster as an edge computing node, and storing the original sensitive data in a local storage unit of the physical device that generated the original sensitive data; When a mobile device node comes online, it is dynamically assigned to a specific grid in the corresponding workstation area according to the physical location coordinates of the mobile device node, and a status registration request is sent to the edge computing node associated with the specific grid. In response to the status registration request, the edge computing node extracts non-sensitive features of the local original sensitive data and generates an encrypted feature vector bound to the grid identifier to which the device belongs; the encrypted feature vector is marked as a data set to be synchronized. Verify the lineage dependency path of each encrypted feature vector in the data set to be synchronized through a device lineage relationship library; the device lineage relationship library records the sub-device identifier, host device identifier, association type and last valid timestamp; For the encrypted feature vector verified by the bloodline dependency path, the final change frequency of the encrypted feature vector within the preset time window is calculated, and the encrypted feature vector is synchronized to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation.

2. A decentralized gateway control method according to claim 1, characterized in that: The method of dividing the physical space of the industrial site into a gridded space model and associating each grid with a device cluster within a workstation area includes: Calculate the minimum outer bounding rectangle based on the actual physical boundary coordinates of the workstation area, and determine the grid division density based on the equipment communication radius and the correlation between the workstation process; establish a coordinate system based on the minimum outer bounding rectangle, and generate equal-sized grid cells along the coordinate axis according to the division density; Traverse the set of grid cells corresponding to each workstation area. If a single workstation area is divided into more than three grid cells, trigger the grid merging operation until each workstation area is associated with at most two grid cells; bind the merged grid cells to the workstation area. When the physical location of the equipment cluster completely falls into one of the grid cells, automatically establish the ownership relationship between the equipment cluster and the corresponding grid cell; detect the position change of the equipment cluster in real time. If the center point of the equipment cluster offsets beyond the boundary of the grid cell to which it belongs, trigger the grid cell rebinding operation.

3. A decentralized gateway control method according to claim 2, characterized in that: The method of dynamically allocating a mobile device node to a specific grid corresponding to a workstation area according to the physical location coordinates of the mobile device node when the mobile device node comes online, and sending a status registration request to an edge computing node associated with the specific grid includes: The physical location coordinates are collected in real time by the built-in positioning module of the mobile device node; the physical location coordinates are spatially matched with the preloaded workstation area topology map; if the physical location coordinates fall within the geometric polygon range of a certain workstation area, the workstation area is marked as a candidate workstation area; When the number of workstation areas to be selected is greater than the preset area number threshold, the workstation area where the movable device node has stayed for the longest time in history is selected as the target workstation area; based on the mapping relationship between the center point coordinates of the target workstation area and the gridded space model, the specific grid to which the movable device node belongs is determined; and a status registration request carrying a device identifier and timestamp is sent to the edge computing node associated with the specific grid.

4. A decentralized gateway control method according to claim 3, characterized in that: The method of verifying the lineage dependency path of each encrypted feature vector in the to-be-synchronized data set through a device lineage relationship library includes: Parsing the grid identifier and device identifier bound to the encrypted feature vector as the host device identifier to be verified; obtaining the node identifier of the removable device that initiated the status registration request; retrieving all associated entries in the device lineage relationship database whose sub-device identifiers are equal to the node identifier of the removable device; extracting the host device identifier set and association type recorded in the associated entry; When the host device identifier to be verified exists in the host device identifier set and the association type belongs to the critical business dependency type, if the time difference between the current time and the last valid timestamp in the association entry is less than the preset validity period threshold, it is determined that the bloodline dependency path verification is passed.

5. A decentralized gateway control method according to claim 1, characterized in that: The method for calculating the final change frequency of the encrypted feature vector within a preset time window includes: Obtaining a set of version snapshots of the original sensitive data recorded at historical time points from a local storage unit of an edge computing node; performing differential analysis on adjacent version snapshots in the set of version snapshots in chronological order to obtain a difference metric value; and marking a valid change event when the difference metric value exceeds a preset difference metric threshold; The basic change frequency is calculated based on the number of occurrences of all valid change events within the preset time window and the preset time window length; If the original sensitive data is continuous time series data, the fluctuation amplitude weight factor is calculated based on the numerical fluctuation range of adjacent version snapshots; if the original sensitive data is a discrete configuration parameter, the change density compensation coefficient is calculated based on the distribution density of valid change events; the product of the basic change frequency and the corresponding fluctuation amplitude weight factor or change density compensation coefficient is used as the final change frequency.

6. A decentralized gateway control method according to claim 5, characterized in that: The method for synchronizing the encrypted feature vector to the removable device node according to the final change frequency includes: The system obtains the current task type, device operation status code, and remaining task duration reported by the mobile device node in real time, and simultaneously collects the real-time operating condition data stream of the host device verified by the lineage dependency path; the historical final change frequency sequence corresponding to the real-time operating condition data stream, device operation status code, and encrypted feature vector is input into the pre-trained time series prediction network, and multi-dimensional feature fusion is performed in combination with the current task type and remaining task duration to output the predicted change frequency value; When the predicted change frequency value exceeds the preset change frequency threshold, active pre-synchronization of the encrypted feature vector is initiated to the mobile device node; a pre-synchronization data cache area is established locally in the mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp.

7. A decentralized gateway control method according to claim 6, characterized in that: The method of inputting the real-time working condition data stream, the equipment operating status code, and the historical final change frequency sequence corresponding to the encrypted feature vector into a pre-trained time series prediction network and performing multi-dimensional feature fusion in combination with the current task type and the remaining task duration includes: Extracting spatial features from the real-time working condition data stream to obtain a first feature tensor; extracting temporal features from the historical final change frequency sequence to obtain a second feature tensor; and concatenating the first feature tensor and the second feature tensor in the channel dimension to form a dynamic feature fusion tensor. The current task type is encoded into a task type vector, and the device operation status code and the remaining task duration value are normalized and merged into a device state vector; the task type vector and the device state vector are spliced ​​to form a static parameter vector; the dynamic feature fusion tensor and the static parameter vector are subjected to nonlinear feature interaction to generate a predicted change frequency value.

8. A decentralized gateway control method according to claim 6, characterized in that: The method of establishing a pre-synchronization data buffer area locally on a mobile device node to store the received encrypted feature vector and record the pre-synchronization timestamp includes: An independent storage partition is created for each actively pre-synchronized encrypted feature vector and associated with a dynamic access credential, wherein the dynamic access credential is generated by a hash operation of the pre-synchronization timestamp and the identifier of the edge computing node that initiates the pre-synchronization; when the movable device node needs to call the encrypted feature vector to perform a task scheduling operation, an access request containing the current task type and the device operation status code is submitted to the pre-synchronization data cache; when the task type in the access request is consistent with the current task type recorded when the pre-synchronization is initiated, and the device operation status code is in the access-allowed state range, the dynamic access credential is activated and a temporary decryption key is generated, and the target encrypted feature vector is decrypted using the temporary decryption key.

9. A decentralized gateway control system, characterized in that: The system includes: a gridded space modeling and gateway configuration module, an encrypted feature vector generation module, a lineage dependency path verification module, an encrypted feature vector synchronization and task scheduling module, and each module is sequentially connected to communicate with each other; A gridded space modeling and gateway configuration module, configured to divide the physical space of the industrial site into a gridded space model, associate each grid with a device cluster within a workstation area, configure the gateway of the device cluster as an edge computing node, and store the original sensitive data in the local storage unit of the physical device that generated the original sensitive data; The encrypted feature vector generation module is configured to dynamically assign a mobile device node to a specific grid in the corresponding workstation area based on the physical location coordinates of the mobile device node when the node comes online, and send a status registration request to the edge computing node associated with the specific grid; in response to the status registration request, the edge computing node extracts non-sensitive features of the local original sensitive data and generates an encrypted feature vector bound to the grid identifier to which the device belongs; and marks the encrypted feature vector as a data set to be synchronized; A lineage dependency path verification module is used to verify the lineage dependency path of each encrypted feature vector in the data set to be synchronized through a device lineage relationship library; the device lineage relationship library records the sub-device identifier, host device identifier, association type and last valid timestamp, the sub-device is a movable device node, and the host device is a physical device fixedly deployed in the grid; The encrypted feature vector synchronization and task scheduling module is used to calculate the final change frequency of the encrypted feature vector within a preset time window for the encrypted feature vector verified by the bloodline dependency path, and synchronize the encrypted feature vector to the mobile device node according to the final change frequency; the mobile device node decrypts the received encrypted feature vector and then performs the task scheduling operation.

Images