Node networking configuration method, apparatus, device and system, and medium and product

Abstract

The present disclosure provides a node networking configuration method, apparatus, device and system, and a medium and a product. The method is applied to a dispatching center device, and comprises: receiving work order information, the work order information at least comprising first work information of a supplier and second work information of a demander; extracting the first work information and the second work information to obtain node information, wherein the node information represents privacy computing nodes of the supplier and privacy computing nodes of the demander; and on the basis of the number of nodes in the node information, determining a node networking architecture corresponding to the node information, wherein the node networking architecture is used for establishing a network connection between the privacy computing nodes of the supplier and the privacy computing nodes of the demander.

Description

Node networking configuration method, device, equipment, system, medium and product

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to Chinese Patent Application No. 202410171888.9 filed in China on February 6, 2024, the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present disclosure relates to federated learning technology, and more specifically to a node networking configuration method, apparatus, device, system, medium, and product. Background Art

[0004] In related technologies, the privacy computing node management solution adopts a point-to-point cooperation model, in which the node identity document (ID) and node information are communicated offline by telephone, email, and other means between the two parties. However, it has the disadvantage of high cost. For example, in a scenario where the first three nodes execute the first task among five nodes, and the second task is executed by two nodes, the point-to-point cost is high and the efficiency is low. Summary of the Invention

[0005] In view of this, the embodiments of the present disclosure provide a node networking configuration method, apparatus, device, system, medium and product, aiming to effectively reduce the cost of multi-node interaction and improve the efficiency of multi-node interaction.

[0006] The technical solution of the embodiment of the present disclosure is implemented as follows:

[0007] The present disclosure provides a node networking configuration method, which is applied to a dispatch center device. The method includes:

[0008] Receiving work order information, wherein the work order information includes at least first work information of a supplier and second work information of a demander;

[0009] Extracting the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander;

[0010] According to the number of nodes in the node information, determine the node networking architecture corresponding to the node information; wherein, the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

[0011] In the above solution, the node networking architecture includes a first networking architecture, a second networking architecture, and a third networking architecture; and determining the node networking architecture corresponding to the node information according to the number of nodes in the node information includes:

[0012] If the number of nodes is greater than a first number threshold, determining a first networking architecture corresponding to the node information;

[0013] If the number of nodes is greater than a second number threshold and less than the first number threshold, determining a second networking architecture corresponding to the node information; the second number threshold is less than the first number threshold;

[0014] If the number of nodes is less than the second number threshold, a third networking architecture corresponding to the node information is determined.

[0015] In the above solution, after determining the node networking architecture corresponding to the node information based on the number of nodes in the node information, the method further includes:

[0016] If it is determined that the training information of the first node meets the exchange condition, sending the first encryption parameter of the first node to the second node through the node networking architecture, and sending the second encryption parameter of the second node to the first node through the node networking architecture;

[0017] Among them, the first node is a local computing node. If the first node is the privacy computing node of the supplier, the second node is the privacy computing node of the demander corresponding to the first node. If the first node is the privacy computing node of the demander, the second node is the privacy computing node of the supplier corresponding to the first node.

[0018] In the above solution, the method further includes:

[0019] Obtaining a first time parameter corresponding to the node networking architecture; the first time parameter represents a duration for sending the second encryption parameter of the second node to the first node;

[0020] If the first time parameter is greater than or equal to the time threshold, the second encryption parameter received is processed based on the first node to obtain an updated parameter; the updated parameter is used to update the modeling model corresponding to the first node.

[0021] In the above solution, the method further includes:

[0022] Acquire interaction parameters between the first node and the second node;

[0023] If the interaction parameter is greater than the interaction threshold, detecting the second node corresponding to the interaction parameter to obtain a detection result;

[0024] If the detection result indicates that the second node is in an abnormal state, the second node is eliminated.

[0025] In the above solution, the method further includes:

[0026] If the first time parameter is greater than or equal to the time threshold, obtaining a second node corresponding to the first time parameter;

[0027] Detecting a second node corresponding to the first time parameter to obtain a detection result;

[0028] If the detection result indicates that the second node is in an abnormal state, the second node is eliminated.

[0029] In the above solution, the method further includes:

[0030] If the second time parameter corresponding to the node networking architecture meets the time condition, the privacy computing node of the supplier and the privacy computing node of the demander are controlled to disconnect the network connection; the second time parameter represents the duration of controlling the privacy computing node of the supplier and the privacy computing node of the demander to establish a network connection based on the node networking architecture.

[0031] The present disclosure provides a node networking configuration device, which is applied to a dispatch center device, including:

[0032] A receiving module, configured to receive work order information, wherein the work order information includes at least first work information of a supplier and second work information of a demander;

[0033] An extraction module, configured to extract the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander;

[0034] The first determination module is used to determine the node networking architecture corresponding to the node information based on the number of nodes in the node information; wherein, the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

[0035] An embodiment of the present disclosure provides a dispatch center device, comprising: a processor and a memory for storing a computer program that can be run on the processor, wherein:

[0036] The processor is configured to execute the steps of the method according to any one of claims 1 to 8 when running a computer program.

[0037] The embodiment of the present disclosure provides a node networking configuration system, including a data sharing device, the above-mentioned dispatching center device and a node device; wherein,

[0038] The data sharing device is used for:

[0039] Generate work order information, the work order information including at least first work information of the supplier and second work information of the demander; determine the dispatch center device according to first scheduling information in the first work information and second scheduling information in the second work information; and send the work order information to the dispatch center device;

[0040] The node device is used to control the privacy computing node to perform training and obtain training information.

[0041] In the above solution, the data sharing device determines the dispatch center device according to the first scheduling information in the first work information and the second scheduling information in the second work information, including:

[0042] If it is determined that the first scheduling information and the second scheduling information are consistent, determining the first scheduling device corresponding to the first scheduling information or the second scheduling device corresponding to the second scheduling information as the scheduling center device;

[0043] If it is determined that the first scheduling information and the second scheduling information are inconsistent, a third scheduling device is determined and the third scheduling device is determined as the scheduling center device, wherein the third scheduling device is connected to the first scheduling device and the second scheduling device respectively.

[0044] An embodiment of the present disclosure provides a storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the node networking configuration method are implemented.

[0045] An embodiment of the present disclosure provides a computer program product, including a computer program, which implements the steps of the node networking configuration method when executed by a processor.

[0046] The embodiments of the present disclosure provide a node networking configuration method, device, equipment, system, medium and product. The method is applied to a dispatch center device, and includes: receiving work order information, wherein the work order information includes at least the first work information of the supplier and the second work information of the demander; extracting the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander; determining the node networking architecture corresponding to the node information according to the number of nodes in the node information; wherein the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander. By adopting the technical solution of the embodiments of the present disclosure, the node information is determined by the first work information of the supplier and the second work information of the demander included in the work order information, and the node networking architecture is determined by the number of nodes in the node information, so as to realize the network connection between the privacy computing nodes, reduce the cost of multi-node interaction, and improve the efficiency of multi-node interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] FIG1 is a schematic diagram of a process for implementing a node networking configuration method according to an embodiment of the present disclosure;

[0048] FIG2 is a schematic diagram of the structure of a node networking configuration device according to an embodiment of the present disclosure;

[0049] FIG3 is a schematic diagram of a hardware entity structure of a node networking configuration device according to an embodiment of the present disclosure;

[0050] FIG4 is a schematic diagram of the structure of a node networking configuration system according to an embodiment of the present disclosure;

[0051] FIG5 is a schematic diagram of a process flow for implementing a node networking configuration method in an application example of an embodiment of the present disclosure;

[0052] FIG6 is a schematic diagram of a hybrid topology architecture in an application example of an embodiment of the present disclosure;

[0053] FIG7 is a schematic diagram of a mesh topology architecture in an application example of an embodiment of the present disclosure. DETAILED DESCRIPTION

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the art of the present disclosure. The terms used herein in the specification of the present disclosure are only for the purpose of describing specific embodiments and are not intended to limit the present disclosure.

[0055] In the first related technology, the federated learning training process: During task execution, each node uses local data to train a local model. Once the iteration conditions for exchanging model parameters with other nodes are met, the node sends the encrypted local parameters to the partner node and simultaneously receives the encrypted parameters from the partner node. After the local and partner parameters are federated and averaged, the local model is updated.

[0056] The disadvantage of the above-mentioned related technology 1: If a node exits or the network is poor, the task cannot be executed.

[0057] The privacy computing node management solution in related technology 2 adopts a point-to-point cooperation model, and the node ID and node information are communicated offline through telephone, email, etc. between the two parties. During network connection, the network connection cannot be controlled according to the limited period of cooperation, and manual disconnection is required.

[0058] The disadvantages of the above-mentioned related technology 2 include: high cost. For example, in a scenario where the first task among five nodes is executed by the first three nodes and the second task is executed by two nodes, the point-to-point cost is high and the efficiency is low.

[0059] Understandably, when a data provider selects a data partner on a trading platform, it's not a one-time deal. It's possible that the node data used in each federated learning task may be inconsistent, the number of collaborating nodes may be inconsistent, or nodes may drop out or experience abnormal node status during the collaboration process. Related technologies utilize a point-to-point project-based model for data circulation, which prevents dynamic adjustment of network architecture based on actual needs and can cause issues such as node withdrawal during operation.

[0060] The present disclosure provides a node networking configuration method, which is applied to a dispatch center device. As shown in FIG1 , the method includes:

[0061] Step 101: receiving work order information, where the work order information includes at least first work information of the supplier and second work information of the demander.

[0062] Exemplarily, the dispatch center device may be connected to the data sharing device, and the dispatch center device may receive work order information sent by the data sharing device.

[0063] Exemplarily, the first work information may include the supplier's scheduling center name information, the supplier's scheduling center address information, the privacy computing node ID information, node address information, node port information, and node number information participating in the task.

[0064] Exemplarily, the second work information can be work order name information, demander dispatch center name information, demander dispatch center address information, demander privacy computing node ID information, node address information, node port information, node number information, computing type information (federated learning, hidden query, joint statistics, privacy intersection), resource name information, cooperation time information and other information.

[0065] Step 102: Extract the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander.

[0066] Exemplarily, the dispatch center device may include a delivery center unit configured to receive work order information and extract node information from the work order information. The node information includes information such as the node name, node address, node port, and node number of the privacy computing node.

[0067] In the disclosed embodiment, a networked approach may be used to manage nodes, and the cooperative nodes for each task may be confirmed through work order information, and then a networking approach may be selected based on the number of local nodes and cooperative nodes.

[0068] Step 103: Determine the node networking architecture corresponding to the node information based on the number of nodes in the node information; wherein the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

[0069] Exemplarily, the number interval of the node quantity can be determined, and the node networking architecture corresponding to the node information can be determined based on the preset relationship and the number interval; it can be understood that the preset relationship represents different number intervals corresponding to different node networking architectures.

[0070] In an application example, the node networking architecture includes a first networking architecture, a second networking architecture, and a third networking architecture; determining the node networking architecture corresponding to the node information based on the number of nodes in the node information includes:

[0071] If the number of nodes is greater than a first number threshold, determining a first networking architecture corresponding to the node information;

[0072] If the number of nodes is greater than a second number threshold and less than the first number threshold, determining a second networking architecture corresponding to the node information; the second number threshold is less than the first number threshold;

[0073] If the number of nodes is less than the second number threshold, a third networking architecture corresponding to the node information is determined.

[0074] Exemplarily, the first quantity threshold may be a numerically larger threshold value b, and the first networking architecture may be determined based on actual conditions and is not limited herein. As an example, the first networking architecture may be a hybrid topology architecture. If the number of nodes is greater than the threshold value b, the networking architecture for determining the node information is a hybrid topology architecture. For a hybrid topology architecture, the nodes are grouped, a heap node is selected from each group, and the dispatch center device serves as the central coordination node. During each iteration, the heap node merges the sum of the parameters of the heap and sends it to the central coordination node. The central coordination node averages the sum and feeds it back to the heap node, and then feeds it back to other nodes through the heap node.

[0075] Exemplarily, the second number threshold may be threshold a, which is less than threshold b. The second network architecture may be determined based on actual circumstances and is not limited herein. As an example, the second network architecture may be a star network architecture. If the number of nodes is greater than threshold a and less than threshold b, the network architecture for determining the node information is a star network architecture. For a star architecture, a dispatch center device is selected as a coordinating node.

[0076] For example, the third networking architecture can be determined based on actual circumstances and is not limited herein. As an example, the third networking architecture can be a mesh topology architecture. If the number of nodes is less than a threshold value a, the node information-based networking architecture is determined to be a mesh topology architecture. For a mesh topology architecture, during each iteration, the coordinating node for the next round can be determined based on the collected node network quality information.

[0077] In the disclosed embodiment, the network architecture can be selected according to the number of nodes cooperating in each task, and the central node of each modeling task can also be determined according to the network architecture.

[0078] In an application example, after determining the node network architecture corresponding to the node information based on the number of nodes in the node information, the method further includes:

[0079] If it is determined that the training information of the first node meets the exchange condition, sending the first encryption parameter of the first node to the second node through the node networking architecture, and sending the second encryption parameter of the second node to the first node through the node networking architecture;

[0080] Among them, the first node is a local computing node. If the first node is the privacy computing node of the supplier, the second node is the privacy computing node of the demander corresponding to the first node. If the first node is the privacy computing node of the demander, the second node is the privacy computing node of the supplier corresponding to the first node.

[0081] Exemplarily, the second node can be a cooperative computing node corresponding to the local computing node. It is understood that if the provider's privacy computing node is a local computing node, then the demander's privacy computing node corresponding to the provider's privacy computing node is a cooperative computing node; if the demander's privacy computing node is a local computing node, then the provider's privacy computing node corresponding to the demander's privacy computing node is a cooperative computing node.

[0082] For example, the training information of the first node may be training parameters obtained by training a local model using local data during the first node's modeling task. The exchange condition may be an iterative condition for the first node to exchange model parameters with other nodes. The first encrypted parameter may be a parameter obtained by encrypting the training parameter.

[0083] In an application example, the method further includes:

[0084] Obtaining a first time parameter corresponding to the node networking architecture; the first time parameter represents a duration for sending a second encryption parameter of the second node to the first node;

[0085] If the first time parameter is greater than or equal to the time threshold, the received second encryption parameter is processed based on the first node to obtain an updated parameter; the updated parameter is used to update the modeling model corresponding to the first node.

[0086] For example, the time threshold can be determined according to actual conditions and is not limited here. As an example, the time threshold can be a preset reception duration for the first node to receive the second encryption parameter sent by the second node, and the time threshold can be less than or equal to the preset reception duration t1.

[0087] Exemplarily, the first node processes the received second encryption parameter to obtain the updated parameter, which may be obtained by federating the average of the local training parameter and the partner's second encryption parameter. The modeling model may be the local model corresponding to the first node. It should be noted that if the first time parameter is greater than or equal to the time threshold, the first node no longer receives the second encryption parameter and only processes the already received second encryption parameter.

[0088] In an application example, the method further includes:

[0089] Obtaining interaction parameters between the first node and the second node;

[0090] If the interaction parameter is greater than the interaction threshold, the second node corresponding to the interaction parameter is detected to obtain a detection result;

[0091] If the detection result indicates that the second node is in an abnormal state, the second node is removed.

[0092] Exemplarily, the process for obtaining the interaction parameters can be determined based on actual circumstances and is not limited herein. As an example, the interaction parameters can be obtained by processing the interaction information between the first node and the second node, where the interaction parameters can be a distance parameter. Exemplarily, the interaction threshold can be a distance threshold T. Exemplarily, the process for obtaining the interaction parameters based on the interaction information can include obtaining first interaction information during a first round of interaction between the first node and the second node; obtaining second interaction information during a second round of interaction between the first node and the second node; and determining the distance parameter between the first interaction information and the second interaction information.

[0093] Exemplarily, the interaction information includes node name information, central processing unit (CPU) information, memory information, bandwidth information between nodes, delay information between nodes, jitter information between nodes, packet loss information between nodes, local node parameter sending time information, parameter receiving cooperative node time information, and other information.

[0094] Exemplarily, the interaction information also includes the name information of the cooperative node that did not receive the parameters; the second node corresponding to the interaction parameter can be the above-mentioned cooperative node that did not receive the parameters; the second node corresponding to the interaction parameter is detected to obtain the detection result, which can be to obtain the heartbeat status of the second node corresponding to the interaction parameter, wherein the heartbeat status is the node interconnection status, and determine whether there is any abnormality in the heartbeat status of the second node.

[0095] In an application example, the method further includes:

[0096] If the first time parameter is greater than or equal to the time threshold, obtaining the second node corresponding to the first time parameter;

[0097] Detecting the second node corresponding to the first time parameter to obtain a detection result;

[0098] If the detection result indicates that the second node is in an abnormal state, the second node is removed.

[0099] It can be understood that if the first time parameter is greater than or equal to the time threshold, and the second node fails to successfully send its own second encryption parameter to the first node, the heartbeat status of the second node is obtained, where the heartbeat status is the node interconnection status, to determine whether there is any abnormality in the heartbeat status of the second node.

[0100] Compared to the federated modeling method in the related art, the disclosed embodiment adds a node anomaly monitoring strategy, specifically adding a timer tasker unit and a task monitor unit. During each round of aggregation of local and cooperative parameters, the node status and abnormal node handling are uniformly managed based on receiving only the parameters of the specified partner nodes, and based on the network status of the data received in this round and the names of the unmerged nodes.

[0101] In an application example, the method further includes:

[0102] If the second time parameter corresponding to the node networking architecture meets the time condition, the privacy computing node of the supplier and the privacy computing node of the demander are controlled to disconnect the network connection; the second time parameter represents the duration of time for establishing a network connection between the privacy computing node of the supplier and the privacy computing node of the demander based on the node networking architecture.

[0103] For example, the time condition indicates that the duration of a privacy computing node establishing a network connection in a node networking architecture exceeds the cooperation duration. The cooperation duration can be determined based on the cooperation time information of the second work information in the work order information, and the cooperation time information can represent the cooperation duration. In the disclosed embodiment, after confirming the networking architecture, a timed node connection policy is configured. After the cooperation duration expires, the node is automatically disconnected.

[0104] The embodiment of the present disclosure provides a node networking configuration device, which is applied to a dispatch center device. As shown in FIG2 , the node networking configuration device 200 includes: a receiving module 201, an extraction module 202, and a first determination module 203; wherein,

[0105] A receiving module 201 is configured to receive work order information, wherein the work order information includes at least first work information of a supplier and second work information of a buyer;

[0106] An extraction module 202 is configured to extract the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander;

[0107] The first determination module 203 is used to determine the node networking architecture corresponding to the node information based on the number of nodes in the node information; wherein the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

[0108] In some embodiments, the node networking architecture includes a first networking architecture, a second networking architecture and a third networking architecture; the first determination module 203 is also used to determine the first networking architecture corresponding to the node information if the number of nodes is greater than the first number threshold; if the number of nodes is greater than the second number threshold and less than the first number threshold, determine the second networking architecture corresponding to the node information; the second number threshold is less than the first number threshold; if the number of nodes is less than the second number threshold, determine the third networking architecture corresponding to the node information.

[0109] In some embodiments, the first determination module 203 is also used to send the first encryption parameter of the first node to the second node through the node networking architecture if it is determined that the training information of the first node meets the exchange condition, and send the second encryption parameter of the second node to the first node through the node networking architecture; wherein, the first node is a local computing node, if the first node is the privacy computing node of the supplier, the second node is the privacy computing node of the demander corresponding to the first node, if the first node is the privacy computing node of the demander, the second node is the privacy computing node of the supplier corresponding to the first node.

[0110] In some embodiments, the node networking configuration device 200 further includes: a first acquisition module and a processing module; wherein,

[0111] A first acquisition module is configured to acquire a first time parameter corresponding to the node networking architecture; the first time parameter represents a duration for sending the second encryption parameter of the second node to the first node;

[0112] A processing module is used to process the second encryption parameter received based on the first node to obtain an updated parameter if the first time parameter is greater than or equal to a time threshold; the updated parameter is used to update the modeling model corresponding to the first node.

[0113] In some embodiments, the node networking configuration device 200 further includes: a second acquisition module, a first detection module and a first elimination module; wherein,

[0114] A second acquisition module, configured to acquire interaction parameters between the first node and the second node;

[0115] A first detection module is configured to detect a second node corresponding to the interaction parameter to obtain a detection result if the interaction parameter is greater than an interaction threshold;

[0116] A removal module is configured to remove the second node if the detection result indicates that the second node is in an abnormal state.

[0117] In some embodiments, the node networking configuration device 200 further includes: a third acquisition module, a second detection module, and a second elimination module; wherein,

[0118] A third acquisition module, configured to acquire a second node corresponding to the first time parameter if the first time parameter is greater than or equal to the time threshold;

[0119] A second detection module is used to detect the second node corresponding to the first time parameter to obtain a detection result;

[0120] The second elimination module is configured to eliminate the second node if the detection result indicates that the second node is in an abnormal state.

[0121] In some embodiments, the node networking configuration device 200 also includes: a control module, which is used to control the privacy computing node of the supplier and the privacy computing node of the demander to disconnect the network connection if the second time parameter corresponding to the node networking architecture meets the time condition; the second time parameter represents the duration of controlling the privacy computing node of the supplier and the privacy computing node of the demander to establish a network connection based on the node networking architecture.

[0122] It should be noted that the node networking configuration device provided in the above embodiment only uses the division of the above-mentioned program modules as an example when performing control. In actual applications, the above-mentioned processing can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program modules to complete all or part of the above-mentioned processing. In addition, the node networking configuration device provided in the above embodiment and the aforementioned node networking configuration method embodiment are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0123] Based on the hardware implementation of the program modules described above, and in order to implement the methods of the embodiments of the present disclosure, the embodiments of the present disclosure further provide a node networking configuration device. Figure 3 illustrates only an exemplary structure of the node networking configuration device, not its entire structure. Part or all of the structure shown in Figure 3 may be implemented as needed.

[0124] As shown in Figure 3, a node networking configuration device 300 provided in an embodiment of the present disclosure includes at least one processor 301, a memory 302, and a user interface 303. The various components in the node networking configuration device 300 are coupled together via a bus system 304. It will be appreciated that the bus system 304 is used to enable connectivity and communication between these components. In addition to a data bus, the bus system 304 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in Figure 3, all of these buses are labeled as the bus system 304.

[0125] The user interface 303 may include a display, a keyboard, a mouse, a trackball, a click wheel, keys, buttons, a touch pad or a touch screen.

[0126] The memory 302 in the embodiment of the present disclosure is used to store various types of data to support the operation of the control device. Examples of such data include: any computer program used to operate on the control device.

[0127] The node networking configuration method disclosed in the embodiments of the present disclosure can be applied to or implemented by the processor 301. The processor 301 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the node networking configuration method can be completed by hardware integrated logic circuits or software instructions in the processor 301. The processor 301 described above can be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 301 can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present disclosure can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium located in the memory 302. The processor 301 reads the information in the memory 302 and, in conjunction with its hardware, completes the steps of the node networking configuration method provided in the embodiments of the present disclosure.

[0128] In an exemplary embodiment, the node networking configuration device can be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0129] It is understood that memory 302 can be volatile memory or non-volatile memory, or can include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), magnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disk, or compact disc read-only memory (CD-ROM); magnetic surface memory can be magnetic disk memory or magnetic tape memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memories described in the embodiments of the present disclosure are intended to include, but are not limited to, these and any other suitable types of memories.

[0130] The embodiment of the present disclosure provides a node networking configuration system, as shown in FIG4 , the node networking configuration system includes a data sharing device 401 , the above-mentioned dispatch center device 402 and a node device 403 ; wherein,

[0131] The data sharing device 401 is used for:

[0132] Generate work order information, the work order information including at least first work information of the supplier and second work information of the demander; determine a dispatch center device based on first scheduling information in the first work information and second scheduling information in the second work information; and send the work order information to the dispatch center device;

[0133] Node device 403 is used to control the privacy computing node to perform training and obtain training information.

[0134] Exemplarily, data sharing device 401 includes a work order management unit configured to generate a work order based on first work information and second work information. The first scheduling information may be information about a first scheduling device of a supplier, which may be a first scheduling center; the second scheduling information may be information about a second scheduling device of a demander, which may be a second scheduling center.

[0135] For example, the process of generating work order information can be determined based on actual circumstances and is not limited herein. As an example, a first terminal device of a demander can create initial work order information based on the first work information, push the initial work order information to a second terminal device of a supplier, and the second terminal device determines the work order information based on the second work information and the initial work order information.

[0136] In some embodiments, before generating the work order information, the method also includes: the data sharing device 401 stores a product resource catalog of the first terminal device; receives selection information based on the product resource catalog sent by the second terminal device; generates contract information according to the selection information; determines order information based on the contract information, wherein the order information includes order number information, resource name information, supplier name information, demander name information, order time information, calculation type information (federated learning, hidden query, joint statistics, privacy intersection), cooperation time information and other information.

[0137] In some embodiments, the data sharing device 401 determines the dispatch center device according to the first dispatch information in the first work information and the second dispatch information in the second work information, including:

[0138] If it is determined that the first scheduling information and the second scheduling information are consistent, determining the first scheduling device corresponding to the first scheduling information or the second scheduling device corresponding to the second scheduling information as the scheduling center device;

[0139] If it is determined that the first scheduling information and the second scheduling information are inconsistent, a third scheduling device is determined and the third scheduling device is determined as the scheduling center device, wherein the third scheduling device is connected to the first scheduling device and the second scheduling device respectively.

[0140] For example, the first dispatch information may be the name of a first dispatch device; the second dispatch information may be the name of a second dispatch device; and the third dispatch device may be a dispatch center connected to the first and second dispatch devices, respectively. Specifically, if the name of the first dispatch device matches the name of the second dispatch device, the first or second dispatch device is identified as the dispatch center device, thereby interconnecting the supplier and demander through the dispatch center network.

[0141] The node networking configuration method of the embodiment of the present disclosure is described below with reference to an application example, which can be a method for joint modeling of networked privacy computing nodes, applied to a joint modeling system of networked privacy computing nodes.

[0142] As shown in Figure 5, this networked privacy-preserving node-based joint modeling system consists of three devices: a data sharing device, a dispatch center device, and a node device. The data sharing device includes a resource directory unit, an order management unit, a work order management unit, and a dispatch center network configuration unit. The dispatch center device includes a delivery center unit, a privacy-preserving node management unit, a node network configuration unit, and a node monitoring center unit. The node devices primarily provide privacy-preserving computing functions, including a task monitor unit and a timer unit. It should be noted that nodes must be deployed before any tasks begin.

[0143] As shown in Figure 5, the joint modeling method based on networked privacy computing nodes includes the following steps:

[0144] Step 501: Product ordering and contract signing.

[0145] Step 502: Generate an order.

[0146] For example, in steps 501 and 502, after the supply and demand parties complete resource selection and contract signing based on the resource catalog unit of the product in the data sharing device, an order is generated in the order management unit. The order information includes the order number, resource name, supplier name, demander name, order time, calculation type (federated learning, hidden query, joint statistics, privacy intersection), cooperation time, etc.

[0147] Step 503: Generate a work order.

[0148] For example, the buyer creates a work order in the work order management unit. The work order information includes the work order name, the name and address of the buyer's dispatch center, the buyer's privacy computing node ID, node address, node port, number of nodes, computing type (federated learning, hidden query, joint statistics, privacy intersection), resource name, cooperation time, etc. After the work order is created, it is pushed to the supplier in sequence.

[0149] In the work order management unit, the supplier edits the pushed work order information, adds the supplier's dispatch center name, the supplier's dispatch center address, the privacy computing node ID, node address, node port, number of nodes involved in the task, etc. After completion, the work order information is pushed to the dispatch center equipment for network configuration.

[0150] Step 504: The dispatch center network is interconnected.

[0151] For example, the specific method for configuring the network interconnection between the supplier and demand dispatch centers is as follows: First, determine whether the dispatch center names of the supplier and demand sides are consistent. If so, push the work order to the delivery center unit of the dispatch center equipment. If not, identify a third dispatch center, activate the dispatch center network policies of both the supply and demand sides for this third dispatch center, complete the network interconnection, and simultaneously configure the task monitor unit based on the cooperation time. If the cooperation time expires, the network policies of both parties will automatically expire. After the network configuration is completed, push the work order to the delivery center that pushed it to the dispatch center.

[0152] Step 505: Receive the work order and extract node information.

[0153] Exemplarily, after receiving the work order, the delivery center unit of the dispatch center device extracts the node information, which includes the node name, node address, node port, node number, etc.

[0154] Step 506: Confirm the node network architecture.

[0155] Exemplarily, the node information is pushed to the privacy computing node management unit. After receiving the node information, the privacy computing node management unit first determines the number of nodes and determines the corresponding networking architecture based on the number of nodes.

[0156] In the first step, if the value is greater than the threshold b, a hybrid topology architecture is adopted. This hybrid topology, shown in Figure 6, randomly divides the nodes into k heaps. Within each heap, a node is randomly selected as the heap center node. Computing nodes within each heap are interconnected with heap nodes, and heap nodes are interconnected with each other. k can be determined based on the number of cooperating nodes; specifically, every 10 nodes can be grouped into a heap. Heap nodes can also select the optimal node based on metrics such as node memory, CPU, and network. Each time a modeling task is executed, the heap nodes first perform a summation, which is then sent to the dispatch center for average summation and fed back to the heap nodes. The heap nodes then synchronize the data with other nodes.

[0157] In the second step, if the number is greater than the threshold a and less than the threshold b, a star network architecture is adopted, with the dispatch center device acting as the central node, and the computing nodes only need to communicate with the central node network: each time the central node sums and averages the data and feeds it back to the computing node.

[0158] In the third step, if the value is below threshold a, a mesh topology is adopted. Figure 7 shows a mesh topology, where a can be 3-5 and b can be 10 or greater. In a mesh topology, computing nodes are interconnected. During task execution, each round receives feedback from the node monitoring center regarding node network quality, and the computing node with the best network quality information is selected as the coordinating node for the next round.

[0159] Step 507: Nodes are connected regularly.

[0160] For example, after determining the network architecture, the privacy computing node management unit pushes the network architecture to the node network configuration unit. The node network configuration unit then establishes a virtual private network between the nodes based on the collaboration time, completing the network connection between the nodes. A timer task unit is also configured to automatically disconnect the node connection if the collaboration time expires. After completing the node network connection, the node network configuration unit pushes the work order to the privacy computing node.

[0161] Step 508: Start the modeling task.

[0162] For example, the privacy computing node starts the modeling task, and the engine local node exchanges encrypted parameters with the cooperating node when executing the task. In order to prevent the node from temporarily exiting or the network from being unavailable during the modeling process, each privacy computing node adds a timer unit and a task monitor unit to execute the task.

[0163] Step 509: Swap parameters in the tth round.

[0164] For example, during a task, each node trains its local model using local data. Once the iteration conditions for exchanging model parameters with other nodes are met, the node sends the encrypted local parameters to the partner node and simultaneously receives the encrypted parameters from the partner node. The local and partner parameters are federated and averaged to update the local model.

[0165] Step 510: Update the t-th round local model.

[0166] For example, before starting the task, the maximum waiting time t1 for each round of parameter interaction is set in the timer task unit. During the task running process, after the local node reaches the iteration condition for exchanging model parameters with other nodes, when other cooperative nodes transmit parameters to the local node, after the parameter receiving time t1 is exceeded, no more parameters will be received, and the received parameters will be directly merged and aggregated for federal averaging to update the local model.

[0167] Step 511: Collect node network information.

[0168] Exemplarily, when monitoring each round of parameter interaction through the task monitor, information such as the node name, CPU, memory, bandwidth between nodes, delay between nodes, jitter between nodes, packet loss between nodes, time when the local node sends parameters, name of the cooperative node that did not receive the parameters, time when the cooperative node received the parameters, etc. are collected, and the above information is transmitted to the node monitoring center unit.

[0169] Step 512: Compare the differences with the previous round.

[0170] Step 513: Check the nodes and remove abnormal nodes.

[0171] For example, in steps 512 and 513, before starting the task, a threshold T is set in the task monitor unit. The node monitoring center unit compares the information from the previous round and calculates the distance. Nodes with distances greater than T and nodes that have not timely fed back network information are fed back to the node network configuration unit of the dispatch center device to detect the node interconnection status (for example, heartbeat detection). If the status is abnormal, the abnormal node is eliminated in the next round of node interaction, and only normal nodes are allowed to participate in task execution. Multiple rounds of iterations are performed until the node local model is stable, and the task is stopped.

[0172] Step 514: Swap parameters in the t+1th round.

[0173] Step 515: Update the local model of round t+1.

[0174] In an exemplary embodiment, the present disclosure further provides a storage medium, namely, a computer storage medium, which may be a computer-readable storage medium. For example, the storage medium 302 may store a computer program. The computer program may be executed by the processor 301 of the node's networking configuration device 300 to complete the steps of the method described in the embodiment of the present disclosure. The computer-readable storage medium may be a memory such as a ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface storage, optical disk, or CD-ROM.

[0175] In an exemplary embodiment, the present disclosure further provides a computer program product, including a computer program, which can be executed by the processor 301 of the node networking configuration device 300 to complete the steps of any of the aforementioned methods.

[0176] It should be noted that: "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0177] In addition, the technical solutions described in the embodiments of the present disclosure can be arbitrarily combined without conflict.

[0178] The above description is merely a specific embodiment of the present disclosure, but the scope of protection of the present disclosure is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present disclosure should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A node networking configuration method, applied to a dispatch center device, comprising: Receiving work order information, wherein the work order information includes at least first work information of a supplier and second work information of a demander; Extracting the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander; According to the number of nodes in the node information, determine the node networking architecture corresponding to the node information; wherein, the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

2. The method according to claim 1, wherein The node networking architecture includes a first networking architecture, a second networking architecture, and a third networking architecture; and determining the node networking architecture corresponding to the node information according to the number of nodes in the node information includes: If the number of nodes is greater than a first number threshold, determining a first networking architecture corresponding to the node information; If the number of nodes is greater than a second number threshold and less than the first number threshold, determining a second networking architecture corresponding to the node information; the second number threshold is less than the first number threshold; If the number of nodes is less than the second number threshold, a third networking architecture corresponding to the node information is determined.

3. The method according to claim 1, wherein After determining the node network architecture corresponding to the node information based on the number of nodes in the node information, the method further includes: If it is determined that the training information of the first node meets the exchange condition, sending the first encryption parameter of the first node to the second node through the node networking architecture, and sending the second encryption parameter of the second node to the first node through the node networking architecture; Among them, the first node is a local computing node. If the first node is the privacy computing node of the supplier, the second node is the privacy computing node of the demander corresponding to the first node. If the first node is the privacy computing node of the demander, the second node is the privacy computing node of the supplier corresponding to the first node.

4. The method according to claim 3, further comprising: Obtaining a first time parameter corresponding to the node networking architecture; The first time parameter represents the duration of sending the second encryption parameter of the second node to the first node; If the first time parameter is greater than or equal to the time threshold, the second encryption parameter received is processed based on the first node to obtain an updated parameter; the updated parameter is used to update the modeling model corresponding to the first node.

5. The method according to claim 3, further comprising: Acquire interaction parameters between the first node and the second node; If the interaction parameter is greater than the interaction threshold, detecting the second node corresponding to the interaction parameter to obtain a detection result; If the detection result indicates that the second node is in an abnormal state, the second node is eliminated.

6. The method according to claim 4, further comprising: If the first time parameter is greater than or equal to the time threshold, obtaining a second node corresponding to the first time parameter; Detecting a second node corresponding to the first time parameter to obtain a detection result; If the detection result indicates that the second node is in an abnormal state, the second node is eliminated.

7. The method according to claim 1, further comprising: If the second time parameter corresponding to the node networking architecture meets the time condition, the privacy computing node of the supplier and the privacy computing node of the demander are controlled to disconnect from the network; The second time parameter represents the duration of controlling the establishment of a network connection between the privacy computing node of the supplier and the privacy computing node of the demander based on the node networking architecture.

8. A node networking configuration device, applied to a dispatch center device, comprising: A receiving module, configured to receive work order information, wherein the work order information includes at least first work information of a supplier and second work information of a demander; An extraction module, configured to extract the first work information and the second work information to obtain node information; wherein the node information represents the privacy computing node of the supplier and the privacy computing node of the demander; The first determination module is used to determine the node networking architecture corresponding to the node information based on the number of nodes in the node information; wherein, the node networking architecture is used to establish a network connection between the privacy computing node of the supplier and the privacy computing node of the demander.

9. A dispatch center device comprising: A processor and a memory for storing a computer program capable of being executed on the processor, wherein The processor is configured to execute the steps of the method according to any one of claims 1 to 7 when running a computer program.

10. A node networking configuration system, comprising a data sharing device, a dispatch center device as claimed in claim 9, and a node device; wherein: The data sharing device is used for: Generate work order information, the work order information including at least first work information of the supplier and second work information of the demander; determine the dispatch center device according to first scheduling information in the first work information and second scheduling information in the second work information; Sending the work order information to the dispatch center device; The node device is used to control the privacy computing node to perform training and obtain training information.

11. The system according to claim 10, wherein: The data sharing device determines the dispatch center device according to the first dispatch information in the first work information and the second dispatch information in the second work information, including: If it is determined that the first scheduling information and the second scheduling information are consistent, determining the first scheduling device corresponding to the first scheduling information or the second scheduling device corresponding to the second scheduling information as the scheduling center device; If it is determined that the first scheduling information and the second scheduling information are inconsistent, a third scheduling device is determined and the third scheduling device is determined as the scheduling center device, wherein the third scheduling device is connected to the first scheduling device and the second scheduling device respectively.

12. A storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

13. A computer program product comprising a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 7.

Images