Information determination, model training methods, devices, equipment, storage media, and computer program products

By acquiring the location information and performance parameters of edge computing nodes and using a target information prediction model to optimize resource allocation, the problem of resource waste in 5G edge deployment is solved, and efficient resource utilization and traffic offloading are achieved.

CN121334759BActive Publication Date: 2026-11-17CHINA MOBILE CHENGDU INFORMATION & TELECOMM TECH CO LTD +1
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202410931736.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-11
Publication Date
2026-11-17
Estimated Expiration
2044-07-11

AI Technical Summary

Technical Problem

Existing technologies do not consider the real-time resource usage of MEC node servers during the 5G edge deployment and offloading process, resulting in resource waste.

Method used

By acquiring the location information of edge computing nodes, dividing the node set, using the target information to predict the model processing performance parameters, determining the target server, and sending data traffic adjustment instructions to optimize resource utilization.

Benefits of technology

The real-time resource usage of the server was taken into account during the traffic distribution process, which avoided resource waste and achieved rational utilization of resources and efficient traffic distribution.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121334759B_ABST
    Figure CN121334759B_ABST
Patent Text Reader

Abstract

This application discloses an information determination method, comprising: dividing multiple edge computing nodes into multiple node sets based on the obtained location information of edge computing nodes corresponding to user plane functional network elements; obtaining the performance parameters of the server corresponding to each edge computing node in each node set; processing the performance parameters using a target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; then, determining the target server from the servers corresponding to the edge computing nodes in the multiple node sets based on the fluctuation information and a target threshold, and sending an adjustment instruction including the identifier of the target server to the user plane functional network element for adjusting the data traffic carried by the target server. This application also discloses a model training method, apparatus, device, storage medium, and computer program product.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to information determination technology in the field of computers, and more particularly to an information determination, model training method, apparatus, device, storage medium, and computer program product. Background Technology

[0002] With the increasing maturity of 5G technology, traditional core network deployment models can no longer meet the low-latency service requirements. Therefore, it is necessary to move User Plane Function (UPF) network elements to the edge network and distribute user data streams to Multi-Access Edge Computing (MEC) nodes through UPF elements to meet the low latency, ultra-high bandwidth, and security requirements of vertical industries (such as healthcare and education). Currently, related technologies mainly achieve these requirements through 5G edge deployment offloading schemes, including uplink classifier schemes, IPv6 multi-homing schemes, local data network schemes, and data network identification schemes. However, these technologies do not consider the real-time resource usage of the servers corresponding to each MEC node during the offloading process, resulting in the problem of unreasonable utilization of system resources and waste. Summary of the Invention

[0003] To address the aforementioned technical problems, this application aims to provide an information determination, model training method, apparatus, device, storage medium, and computer program product, thereby resolving the issue of unreasonable utilization of system resources during traffic splitting in related technologies and avoiding resource waste.

[0004] The technical solution of this application is implemented as follows:

[0005] An information determination method, the method comprising:

[0006] Obtain the location information of the edge computing nodes corresponding to the user plane function network elements, and divide the multiple edge computing nodes into multiple node sets based on the location information;

[0007] Obtain the performance parameters of the server corresponding to each edge computing node in each node set; wherein, the performance parameters characterize the resource utilization of the server;

[0008] The performance parameters are processed using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; wherein, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the server corresponding to the sample edge computing node; the different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes.

[0009] Based on the fluctuation information and the target threshold, the target server is determined from the servers corresponding to the edge computing nodes of the multiple node sets;

[0010] Send an adjustment instruction to the user plane function network element to adjust the data traffic carried by the target server; wherein the adjustment instruction includes the identifier of the target server.

[0011] In the above scheme, obtaining the performance parameters of each server corresponding to each edge computing node in each node set includes:

[0012] Obtain the initial performance parameters of each server corresponding to each edge computing node in each node set;

[0013] The initial performance parameters are processed to obtain the performance parameters of each server.

[0014] In the above scheme, determining the target server from the servers corresponding to the edge computing nodes of the multiple node sets based on the fluctuation information and the target threshold includes:

[0015] Based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node, the target parameters of each server are determined; wherein, the target parameters characterize the resource fluctuation of each server.

[0016] Based on the target parameters and the target threshold, the target server is determined from the servers corresponding to the edge computing nodes of the plurality of node sets.

[0017] In the above scheme, determining the target parameters of each server based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node includes:

[0018] Determine the weight of each performance parameter for each server;

[0019] The target parameters for each server are determined based on the fluctuation information of each performance parameter of each server and the weight of each performance parameter.

[0020] In the above scheme, sending an adjustment instruction for adjusting the data traffic carried by the target server to the user plane function network element includes:

[0021] The adjustment command is sent to the first interface of the user plane function network element through the first target interface of the edge computing node corresponding to the target server.

[0022] The method in the above scheme further includes:

[0023] The traffic adjustment interface receives feedback information from the first interface regarding the adjustment operation of the data traffic carried by the target server; wherein the feedback information is used to indicate whether the data traffic carried by the target server has been successfully adjusted.

[0024] The method in the above scheme further includes:

[0025] The query command sent by the second interface of the user plane function network element is received through the traffic query interface of the edge computing node corresponding to the first server; wherein, the first server is any one of the servers corresponding to the edge computing node corresponding to the user plane function network element; the query command includes the identifier of the first server.

[0026] Based on the query command, the traffic information of the first server is determined, and the traffic information is sent to the second interface through the traffic query interface of the edge computing node corresponding to the first server.

[0027] A model training method, the method comprising:

[0028] Obtain the location information of the sample edge computing nodes corresponding to the user plane function network element, and divide the multiple sample edge computing nodes based on the location information to obtain multiple sample node sets;

[0029] Determine the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; wherein, the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node;

[0030] Based on the sample performance parameters, the initial information prediction model corresponding to each sample node set is trained to obtain multiple target information prediction models; wherein, the multiple target information prediction models are used to process the performance parameters of the server corresponding to the edge computing node corresponding to the user plane functional network element to obtain the fluctuation information of the performance parameters of the server corresponding to the edge computing node; the fluctuation information is used to determine the target server from the server corresponding to the edge computing node.

[0031] In the above scheme, determining the sample performance parameters of each server corresponding to each sample edge computing node in each sample node set includes:

[0032] Receive the initial sample performance parameters of the server corresponding to the sample edge computing node sent by the sample edge computing node;

[0033] Based on the initial sample performance parameters of the server corresponding to the sample edge computing node, the initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are determined.

[0034] The initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are processed to obtain the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set.

[0035] The method in the above scheme further includes:

[0036] The system receives real-time performance parameters of the server corresponding to each edge computing node sent by the edge computing node corresponding to the user plane function network element; wherein the real-time performance parameters characterize the real-time resource usage of the server.

[0037] The real-time performance parameters are processed to obtain the processed real-time performance parameters for each server.

[0038] Based on the processed real-time performance parameters of each server, the target information prediction model corresponding to the target node set is updated; wherein, the target node set is the node set corresponding to the server corresponding to the processed real-time performance parameters.

[0039] An information determining device, the device comprising:

[0040] The first processing unit is used to obtain the location information of the edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple edge computing nodes into multiple node sets based on the location information.

[0041] The acquisition unit is used to acquire the performance parameters of the server corresponding to each edge computing node in each node set; wherein the performance parameters characterize the resource utilization of the server;

[0042] The second processing unit is used to process the performance parameters using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; wherein, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the server corresponding to the sample edge computing node; the different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes.

[0043] The first determining unit is used to determine the target server from the servers corresponding to the edge computing nodes of the plurality of node sets based on the fluctuation information and the target threshold.

[0044] The sending unit is configured to send an adjustment instruction for adjusting the data traffic carried by the target server to the user plane function network element; wherein the adjustment instruction includes the identifier of the target server.

[0045] A model training device, the device comprising:

[0046] The third processing unit is used to obtain the location information of the sample edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple sample edge computing nodes based on the location information to obtain multiple sample node sets.

[0047] The second determining unit is used to obtain the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; wherein, the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node.

[0048] The training unit is used to train the initial information prediction model corresponding to each sample node set based on the sample performance parameters to obtain multiple target information prediction models; wherein, the multiple target information prediction models are used to process the performance parameters of the server corresponding to the edge computing node corresponding to the user plane functional network element to obtain the fluctuation information of the performance parameters of the server corresponding to the edge computing node; the fluctuation information is used to determine the target server from the server corresponding to the edge computing node.

[0049] An edge computing node, the edge computing node comprising: a first processor, a first memory, and a first communication bus;

[0050] The first communication bus is used to establish a communication connection between the first processor and the first memory;

[0051] The first processor is used to execute the information determination program in the first memory to implement the steps of the above-described information determination method.

[0052] A central node, the central node comprising: a second processor, a second memory, and a second communication bus;

[0053] The second communication bus is used to establish a communication connection between the second processor and the second memory;

[0054] The second processor is used to execute the model training program in the second memory to implement the steps of the model training method described above.

[0055] A computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the above-described information determination method or model training method.

[0056] A computer program product comprising a computer program that, when executed by a processor, implements the steps of the aforementioned information determination method or model training method.

[0057] The information determination, model training method, apparatus, device, storage medium, and computer program product provided in the embodiments of this application can divide multiple edge computing nodes into multiple node sets based on the location information of the edge computing nodes corresponding to the user plane functional network element. Then, a target information prediction model corresponding to each node set is used to process the performance parameters of the servers corresponding to each edge computing node to obtain the fluctuation information of the performance parameters of each server. Afterwards, a target server is determined from the servers corresponding to the edge computing nodes based on the fluctuation information and a target threshold. Finally, an adjustment instruction for adjusting the data traffic carried by the target server is sent to the user plane functional network element. In this way, the performance parameters of each server can be analyzed using target information prediction models corresponding to different node sets to obtain the fluctuation information of the performance parameters. Furthermore, the target server whose data traffic needs to be adjusted can be determined by combining the fluctuation information and the target threshold, and the adjustment instruction can be sent to the user plane functional network element. Thus, the user plane functional network element can adjust the data traffic carried by the target server. This means that the real-time resource usage of the servers corresponding to each edge computing node is considered during the traffic splitting process, solving the problem of unreasonable utilization of system resources in related technologies and thus avoiding resource waste. Attached Figure Description

[0058] Figure 1 A flowchart illustrating an information determination method provided in an embodiment of this application;

[0059] Figure 2 A flowchart illustrating another information determination method provided in an embodiment of this application;

[0060] Figure 3 A schematic flowchart illustrating a model training method provided in an embodiment of this application;

[0061] Figure 4 A schematic diagram of a single-hidden-layer autoencoder neural network in a model training method provided in an embodiment of this application;

[0062] Figure 5 This is a schematic diagram of the structure of an information determination device provided in an embodiment of this application;

[0063] Figure 6 This is a schematic diagram of the structure of a model training device provided in an embodiment of this application;

[0064] Figure 7 This is a schematic diagram of the structure of an edge computing node provided in an embodiment of this application;

[0065] Figure 8 This is a schematic diagram of the structure of a central node provided in an embodiment of this application. Detailed Implementation

[0066] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0067] It should be understood that the phrases "embodiments of this application" or "foreign embodiments" throughout the specification mean that a specific feature, structure, or characteristic related to an embodiment is included in at least one embodiment of this application. Therefore, "embodiments of this application" or "in the foreign embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0068] Unless otherwise specified, any step in the embodiments of this application performed by the electronic device may be executed by the processor of the electronic device. It is also worth noting that the embodiments of this application do not limit the order in which the electronic device performs the following steps. Furthermore, the methods used to process data in different embodiments may be the same or different methods. It should also be noted that any step in the embodiments of this application can be executed independently by the electronic device; that is, when the electronic device performs any step in the following embodiments, it may not depend on the execution of other steps.

[0069] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application.

[0070] This application provides an information determination method, which can be applied to edge computing nodes, referring to... Figure 1 As shown, the method may include the following steps:

[0071] Step 101: Obtain the location information of the edge computing nodes corresponding to the user plane function network elements, and divide the multiple edge computing nodes into multiple node sets based on the location information.

[0072] Among them, user plane function network elements can refer to UPF network elements; edge computing nodes can refer to MEC nodes; each node set can include multiple edge computing nodes (i.e., MEC nodes).

[0073] In this embodiment, a clustering algorithm can be used to cluster multiple edge computing nodes based on the location information (i.e., geographical location) of each edge computing node corresponding to the user plane functional network element, resulting in multiple node sets (multiple clusters). In one possible implementation, the clustering algorithm can refer to the adaptive K-means algorithm.

[0074] Step 102: Obtain the performance parameters of the server corresponding to each edge computing node in each node set.

[0075] Among them, performance parameters characterize the server's resource utilization.

[0076] In this embodiment, one edge computing node can correspond to multiple servers; one server can have multiple performance parameters. These performance parameters may include parameters such as Central Processing Unit (CPU) utilization, CPU load, memory load, and disk utilization.

[0077] In this embodiment of the application, the initial performance parameters of each server can be collected first through each edge computing node of each node set, and then the initial performance parameters can be processed to obtain the performance parameters of each server.

[0078] Step 103: Use the target information prediction model corresponding to each node set to process the performance parameters and obtain the fluctuation information of the server's performance parameters.

[0079] Among them, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the server corresponding to the sample edge computing node; different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes.

[0080] It should be noted that each node set corresponds to a target information prediction model; that is, different node sets have different target information prediction models.

[0081] In this embodiment, the performance parameters of each server can be used as input parameters and input into the target information prediction model corresponding to the node set corresponding to each server. After the performance parameters of each server are processed by the target information prediction model, the output of the target information prediction model is the fluctuation information of the performance parameters.

[0082] Step 104: Based on fluctuation information and target threshold, determine the target server from the servers corresponding to the edge computing nodes of multiple node sets.

[0083] The target threshold can be preset based on the specific application scenario and actual needs.

[0084] In this embodiment, the target parameters for each server can be determined first based on the fluctuation information of each performance parameter of each server. Then, the target server is determined from all servers based on the target parameters and target thresholds of each server. It should be noted that the target parameters characterize the resource fluctuation of the server.

[0085] Step 105: Send an adjustment command to the user plane function element to adjust the data traffic carried by the target server.

[0086] The adjustment instructions include the identifier of the target server.

[0087] In this embodiment, after the target server is determined, an adjustment instruction for adjusting the data traffic carried by the target server can be sent from the first target interface of the edge computing node corresponding to the target server to the first interface of the user plane function network element. This allows the first interface of the user plane function network element to adjust the data traffic carried by the target server upon receiving the adjustment instruction. It should be noted that the first target interface can refer to an adjustment request interface, used to send a traffic adjustment request from the target server to the first interface of the user plane function network element; the first interface can refer to the routing configuration and correction interface of the user plane function network element, used to receive the adjustment instruction sent by the edge computing node, so that the user plane function network element can adjust the data traffic carried by the target server.

[0088] The information determination method provided in the embodiments of this application can analyze the performance parameters of each server using a target information prediction model corresponding to different node sets to obtain the fluctuation information of the performance parameters. It can also combine the fluctuation information and the target threshold to determine the target server that needs to adjust the data traffic it carries and send adjustment instructions to the user plane function network element. Thus, the user plane function network element can adjust the data traffic carried on the target server. That is, the real-time resource usage of the servers corresponding to each edge computing node is considered during the traffic splitting process, which solves the problem of unreasonable utilization of system resources in related technologies, thereby avoiding resource waste.

[0089] Based on the foregoing embodiments, embodiments of this application provide an information determination method, referring to... Figure 2 As shown, the method may include the following steps:

[0090] Step 201: The edge computing node obtains the initial performance parameters of each server corresponding to each edge computing node in each node set.

[0091] In this embodiment, the initial performance parameters of the server can refer to the initial performance parameters of the server at different times. That is, the edge computing node can obtain the initial performance parameters of each server corresponding to each edge computing node at different times and store the obtained initial performance parameters of the server at different times. Specifically, the initial performance parameter of the server corresponding to any edge computing node at a certain time can be represented by x. (t) It means, and Where p represents the number of servers corresponding to the i-th edge computing node. Let represent the initial performance parameters of the first server corresponding to the i-th edge computing node at time t. Let represent the initial performance parameters of the second server corresponding to the i-th edge computing node at time t. This represents the initial performance parameters of the p-th server corresponding to the i-th edge computing node at time t.

[0092] Step 202: The edge computing nodes process the initial performance parameters to obtain the performance parameters of each server.

[0093] In this embodiment, the edge computing node can obtain the initial performance parameters of a server at different times within a certain time period based on the stored initial performance parameters of the server at different times, and normalize the initial performance parameters of each server at different times within a certain time period to obtain the performance parameters of each server. Specifically, the initial performance parameters of any server at different times within a certain time period can be represented by x. i It means that x i =[xi (1) ,x i (2) ,...,x i (t) ] T Among them, x i (1) x i (2) and x i (t) This represents the initial performance parameters of a server corresponding to the i-th edge computing node at different times within the time period T. Similarly, the initial performance parameters of each server corresponding to each edge computing node at different times within the time period T can be obtained.

[0094] In one feasible approach, if it is necessary to obtain the initial performance parameters of the server at different times within 30 seconds, and the initial performance parameters are obtained every 2 seconds, then the initial performance parameter x of the server within 30 seconds is... i =[x i (2) ,x i (4) ,...,x i (30) ] 30s Among them, x i (2) Let x represent the initial performance parameters of a server corresponding to the i-th edge computing node at time 2s. i (4) Let x represent the initial performance parameters of a server corresponding to the i-th edge computing node at time 4s. i (30) This represents the initial performance parameters of a server corresponding to the i-th edge computing node at time 30s.

[0095] Then, the initial performance parameters of the server at different times within a certain period of time can be normalized according to the following formula (1) to obtain the performance parameters of the server:

[0096]

[0097] Where t represents any point in time within a certain time period. This represents the performance parameters of the j-th server corresponding to the i-th edge computing node at time t. Let max(x) represent the initial performance parameters of the j-th server corresponding to the i-th edge computing node at time t. i,j ) is the largest initial performance parameter among the initial performance parameters of the j-th server at different times within a certain time period, min(x) i,j) is the smallest initial performance parameter among the initial performance parameters of the j-th server at different times within a certain time period.

[0098] Step 203: The edge computing nodes use the target information prediction model corresponding to each node set to process the performance parameters and obtain the fluctuation information of the server's performance parameters.

[0099] The target information prediction model is obtained by training the initial information prediction model for different sample node sets based on the sample performance parameters of the servers corresponding to the sample edge computing nodes. The different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes. It should be noted that the initial information prediction model can refer to a single hidden layer autoencoder neural network model.

[0100] In this embodiment of the application, the fluctuation information can characterize the degree of change of the server's performance parameters. It should be noted that the greater the degree of change, the greater the fluctuation of the server's performance parameters, and the worse the stability of the server's operation. In this case, service interruption or delay may occur.

[0101] Step 204: The data determination device determines the target parameters for each server based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node.

[0102] The target parameter represents the resource fluctuation of each server.

[0103] In this embodiment of the application, after obtaining the fluctuation information of each performance parameter of each server, the weight of each performance parameter of each server can be determined, and then the fluctuation value of each performance parameter can be obtained by calculating the weight of each performance parameter and the fluctuation information of each performance parameter. After that, the target parameter of each server can be obtained by calculating the fluctuation value of each performance parameter of each server.

[0104] In this embodiment of the application, step 204 can be implemented through steps 204a to 204b:

[0105] Step 204a: The edge computing node determines the weight of each performance parameter for each server.

[0106] In this embodiment, the weights represent the importance of the performance parameters. It should be noted that the sum of the weights of all performance parameters for each server is 1.

[0107] Step 204b: The edge computing node determines the target parameters for each server based on the fluctuation information of each performance parameter of each server and the weight of each performance parameter.

[0108] In this embodiment, the weight and fluctuation information of each performance parameter of each server can be multiplied to obtain the fluctuation value of each performance parameter. Then, the fluctuation values ​​of each performance parameter are summed to obtain the resource fluctuation of each server (i.e., the target parameter). Specifically, the calculation formula of the target parameter can be shown in the following formula (2):

[0109]

[0110] Where, ξ i Let α represent the target parameters of the i-th server, m represent the number of performance parameters of the i-th server, and α represent the target parameters of the i-th server. i,j ε represents the weight of the j-th performance parameter of the i-th server. i,j This represents the fluctuation information of the j-th performance parameter of the i-th server.

[0111] Step 205: Based on the target parameters and target threshold, the edge computing nodes determine the target server from the servers corresponding to the edge computing nodes in multiple node sets.

[0112] It should be noted that the target server is the server whose data traffic needs to be adjusted.

[0113] In this embodiment of the application, the target parameters and target thresholds of each server can be compared, and the target parameters that are greater than the target threshold can be selected from multiple target parameters based on the comparison results. Then, the server corresponding to the target parameter that is greater than the target threshold is taken as the target server.

[0114] It should be noted that if any of the target parameters exceeds the target threshold, it indicates that the resource usage of the server corresponding to that target parameter deviates from the theoretical value of the server's resource usage. In other words, the server corresponding to that target parameter is currently overloaded. In this case, the server needs to be selected as the target server, and the data traffic carried by the target server needs to be adjusted.

[0115] In this embodiment, an initial information prediction model can be constructed using an autoencoder neural network model. The initial information prediction model is then trained using real-time performance parameters of the server resources corresponding to each MEC node to obtain a target information prediction model. Subsequently, the target information prediction model is used to predict the resource fluctuations of the server, thereby adaptively adjusting the data traffic carried on the server. This not only ensures efficient processing of data traffic and reasonable allocation of resources, achieving true intelligent traffic distribution, but also maximizes the rational use of limited resources and achieves optimal performance and the highest stability in complex network environments.

[0116] Step 206: The edge computing node sends an adjustment command to the first interface of the user plane function network element through the first target interface of the edge computing node corresponding to the target server.

[0117] The adjustment instruction includes the identifier of the target server; the identifier of the target server may refer to the device ID and access address of the target server; the first interface may refer to the routing configuration and correction interface; and the first target interface may refer to the adjustment request interface.

[0118] In this embodiment of the application, after the edge computing node determines the target server, it can send an adjustment instruction carrying the identifier of the target server to the routing configuration and correction interface of the user plane function network element through its own adjustment request interface. After receiving the corresponding adjustment instruction, the routing configuration and correction interface of the user plane function network element can adjust the data traffic carried by the target server according to the identifier of the target server carried in the adjustment instruction.

[0119] It should be noted that the first target program code is pre-programmed into both the first interface (i.e., the route configuration and correction interface) and the first target interface (i.e., the adjustment request interface). Specifically, the first target program code is used to implement communication between the first interface (i.e., the route configuration and correction interface) and the first target interface (i.e., the adjustment request interface).

[0120] In one possible implementation, the first target program code burned into the first interface (i.e., the routing configuration and correction interface) includes first request information. This first request information may refer to the adjustment instructions received by the first interface. Specifically, a detailed description of the first request information is shown in Table 1.

[0121]

[0122] Table 1

[0123] The first request information may be sent to the first interface in the form of a Uniform Resource Locator (URL) through the first target interface, and the parameters carried in the first request information may be the server's device ID, access address, and data traffic carried by the server.

[0124] Step 207: The edge computing node receives feedback information from the first target interface regarding the adjustment operation of the data traffic carried by the target server.

[0125] The feedback information is used to indicate whether the adjustment operation on the data traffic carried by the target server was successful.

[0126] In this embodiment of the application, after sending an adjustment instruction to the routing configuration and correction interface of the user plane function network element, the edge computing node will receive feedback information sent by the routing configuration and correction interface through the adjustment request interface by executing the specific program code it has burned, in order to determine whether the adjustment operation for the data traffic carried by the target server is successful.

[0127] In one feasible implementation, the first target program code burned into the first interface also includes first response information. This first response information can refer to information sent from the first interface to the second interface, indicating whether the adjustment operation of the data traffic carried by the target server was successful. Specifically, the details of the first response information are shown in Table 1 above. Specifically, regarding the first response information to be transmitted, if no errors are reported during the adjustment of the data traffic carried by the target server, the adjustment operation is considered successful, and the first target interface of the edge computing node receives a "200 OK" response code sent by the first interface. If errors are reported during the adjustment of the data traffic carried by the target server, the adjustment operation is considered to have failed, and the first target interface of the edge computing node receives a "400 Bad Request" response code and a detailed description of the error reason sent by the first interface. It should be noted that the first target interface also pre-burns the first target program code to receive the feedback information sent by the first interface.

[0128] In other embodiments of this application, the method may further include the following steps.

[0129] Step 208: The edge computing node receives the query command sent by the second interface of the user plane function network element through the second target interface of the edge computing node corresponding to the first server.

[0130] Wherein, the first server is any server among the servers of the edge computing node corresponding to the user plane function network element; the second interface may refer to the routing query interface of the user plane function network element; the second target interface may refer to the traffic feedback interface of the edge computing node used to communicate with the second interface; the query instruction includes the identifier of the first server; the traffic information may include the data traffic currently carried by the server corresponding to the identifier information.

[0131] In this embodiment, the traffic feedback interface of the edge computing node corresponding to the first server can receive a query command sent by the routing query interface of the user plane function network element by executing specific program code burned into itself, so that the traffic feedback interface can query the data traffic currently carried by the first server according to the query command.

[0132] It should be noted that the second interface (i.e., the routing query interface) and the second target interface (i.e., the traffic feedback interface) have pre-programmed second target program code. This second target program code is used to implement communication between the second interface and the second target interface.

[0133] In one possible implementation, the second target program code of the second interface (i.e., the routing query interface) includes second request information. This second request information can refer to the information in the query instruction sent by the second interface to the second target interface. Specifically, a detailed description of the second request information is shown in Table 2.

[0134]

[0135] Table 2

[0136] As shown in Table 2 above, the second request information can be sent from the second interface to the second target interface in the form of a URL, and the parameter carried in the second request information can be the server's device ID.

[0137] Step 209: The edge computing node determines the traffic information of the first server based on the query command, and sends the traffic information to the second interface through the second target interface of the edge computing node corresponding to the first server.

[0138] It should be noted that traffic information may also include the device identifier and access address of the server corresponding to the identification information; the traffic feedback interface (i.e., the second target interface) and the routing query interface (i.e., the second interface) are implemented by burning specific program code to enable communication between edge computing nodes and user plane functional network elements.

[0139] In this embodiment, the traffic feedback interface parses the query command using its own programmed code to obtain the identification information of a server. It then queries the data traffic currently carried by the server corresponding to that identification information (i.e., the first server). Afterward, it sends the traffic information to the routing query interface through the traffic feedback interface of the edge computing node corresponding to the first server. This allows the user plane function network element to analyze the received traffic information based on the code in the routing query interface to determine the data traffic currently carried by the first server. In one possible implementation, the second target program code programmed in the second interface also includes second response information. This second response information can refer to information sent by the second target interface to the second interface to provide feedback on the data traffic currently carried by the target server. Specifically, the detailed description of the second response information is shown in Table 2 above. Specifically, regarding the second response information to be received, if the edge computing node does not encounter any errors during the process of determining traffic information, the edge computing node will send a response code of "200 OK" and traffic information including the server's device ID, access address, and the data traffic currently being carried by the server to the second interface through the second target interface; if the edge computing node encounters any errors during the process of determining traffic information, the edge computing node will send a response code of "400 Bad Request" and a second response information including a description of the specific error reason to the routing query interface through the traffic feedback interface.

[0140] In this embodiment, it can be determined whether the user data traffic allocated to each MEC node needs to be adjusted based on the resource load of the server corresponding to each MEC node. During the traffic splitting process, the routing setting and correction interface of the UPF network element is called to adjust the data traffic carried by the server corresponding to each MEC node in real time. In this way, the performance of the server corresponding to the MEC node can be guaranteed, the server load can be avoided, and the system stability can be guaranteed to a certain extent, reducing the risk of user request errors.

[0141] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0142] The information determination method provided in the embodiments of this application can analyze the performance parameters of each server using a target information prediction model corresponding to different node sets to obtain the fluctuation information of the performance parameters. It can also combine the fluctuation information and the target threshold to determine the target server that needs to adjust the data traffic it carries and send adjustment instructions to the user plane function network element. Thus, the user plane function network element can adjust the data traffic carried on the target server. That is, the real-time resource usage of the servers corresponding to each edge computing node is considered during the traffic splitting process, which solves the problem of unreasonable utilization of system resources in related technologies, thereby avoiding resource waste.

[0143] This application provides a model training method, which can be applied to a central node, as shown in the following embodiments. Figure 3 As shown, the method may include the following steps:

[0144] Step 301: The central node obtains the location information of the sample edge computing nodes corresponding to the user plane function network elements, and divides multiple sample edge computing nodes based on the location information to obtain multiple sample node sets.

[0145] It should be noted that sample edge computing nodes can refer to sample MEC nodes; each sample node set can include multiple sample edge computing nodes (i.e., sample MEC nodes).

[0146] In this embodiment, a clustering algorithm can be used to cluster multiple sample edge computing nodes based on the location information (i.e., geographical location) of each sample edge computing node corresponding to the user plane functional network element, resulting in multiple sample node sets (i.e., multiple clusters). It should be noted that the criteria for clustering the sample edge computing nodes are the same as those for clustering edge computing nodes. In one possible implementation, the clustering algorithm can refer to the adaptive K-means algorithm.

[0147] Step 302: The central node determines the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set.

[0148] Among them, the sample performance parameters characterize the resource usage of the server corresponding to the sample edge computing node.

[0149] In this embodiment, one sample edge computing node can correspond to multiple servers; one server can have multiple sample performance parameters. These sample performance parameters may include parameters such as CPU utilization, CPU load, memory load, and disk utilization.

[0150] In this embodiment of the application, the central node receives the initial sample performance parameters of each server corresponding to each sample edge computing node sent by the edge computing node, and processes the received initial sample performance parameters to obtain the sample performance parameters of each server corresponding to each sample edge computing node.

[0151] In the embodiments of this application, step 302 can be implemented by steps 302a to 302c.

[0152] Step 302a: The central node receives the initial sample performance parameters of the server corresponding to the sample edge computing node sent by the sample edge computing node.

[0153] In this embodiment, the initial performance parameters of the server corresponding to the sample edge computing node can refer to the initial sample performance parameters of the server at different times. That is, the central node can receive and store the initial sample performance parameters of each server corresponding to each sample edge computing node in each sample node set at different times. Specifically, the initial sample performance parameter of the server corresponding to any sample edge computing node at a certain time can be represented by y. (t) It means, and Where q represents the number of servers corresponding to the edge computing node of the nth sample. Let represent the initial sample performance parameters of the first server corresponding to the nth sample edge computing node at time t. Let represent the initial sample performance parameters of the second server corresponding to the nth sample edge computing node at time t. This represents the initial sample performance parameters of the q-th server corresponding to the n-th sample edge computing node at time t.

[0154] Step 302b: The central node determines the initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set based on the initial sample performance parameters of the server corresponding to the sample edge computing node.

[0155] In this embodiment of the application, after receiving the initial sample performance parameters of all servers sent by all sample edge nodes, the central node will determine the initial sample performance parameters of the server corresponding to each sample edge node in each sample node set.

[0156] Step 302c: The central node processes the initial sample performance parameters to obtain the sample performance parameters of each server corresponding to each sample edge computing node.

[0157] In this embodiment, the central node can obtain the initial sample performance parameters of the server at different times within a certain time period based on the stored initial sample performance parameters of the server at different times, and normalize the initial sample performance parameters of each server at different times within a certain time period to obtain the sample performance parameters of each server corresponding to each sample edge computing node. It should be noted that the process of normalizing the initial sample performance parameters is the same as the process of normalizing the initial performance parameters in step 202.

[0158] Step 303: The central node trains the initial information prediction model corresponding to each sample node set based on the sample performance parameters to obtain multiple target information prediction models.

[0159] Among them, multiple target information prediction models are used to process the performance parameters of the servers corresponding to the edge computing nodes of the user plane functional network elements to obtain the fluctuation information of the performance parameters of the servers corresponding to the edge computing nodes; the fluctuation information is used to determine the target server from the servers corresponding to the edge computing nodes.

[0160] In this embodiment of the application, each set of sample nodes corresponds to an initial information prediction model; the initial information prediction model may refer to an autoencoder neural network model.

[0161] In this embodiment, the autoencoder neural network model can refer to a single-hidden-layer autoencoder neural network. If the single-hidden-layer autoencoder neural network consists of an input layer, a hidden layer, and an output layer, then the structural diagram of the single-hidden-layer autoencoder neural network is as follows. Figure 4 As shown. Specifically, the mapping from the input layer to the hidden layer is called the encoder, and the mapping from the hidden layer to the output layer is called the decoder. Further, the encoder y = f(W·x + b) f ), where x is an n-dimensional input layer vector (i.e., the server's CPU utilization, CPU load, memory load, and disk utilization), y is a k-dimensional hidden layer vector, and W [k×n] Let b be the weighting matrix of the encoder. f[k×1] Let f(·) be the encoder's bias vector, and f(·) be the encoder's nonlinear mapping function. Correspondingly, the decoder z = g(V·y + b) g ), where y is the hidden layer vector, z is the n-dimensional output vector, and V [k×n] Let b be the weighting matrix of the decoder. g[k×1] Let W be the bias vector of the decoder, and g(·) be the mapping function of the decoder. In one possible implementation, the nonlinear mapping function of the encoder can refer to the Logistic function; the mapping function of the decoder can refer to the Logistic function and the identity function. It should be noted that W... [k×n] bf[k×1] V [k×n] and b g[k×1] All of these settings are based on historical data and specific application scenarios. Next, the squared error between the input layer vector and the output layer vector can be defined as the objective function. Specifically, the objective function can be represented by J, and the formula for calculating the objective function J is shown in formula (4):

[0162]

[0163] Where m is the effective amount of data used to train the model, and θ = {W, b} f} represents the optimization parameters of the decoder, θ' = {V, b} g} represents the encoder's optimization parameters.

[0164] In this embodiment, the fluctuation information (i.e., the objective function) of the server's sample performance parameters can be predicted based on the initial information prediction model. When the predicted fluctuation information of the server's sample performance parameters differs significantly from the actual fluctuation information of the sample performance parameters, the parameters of the initial information prediction model are adjusted, and the adjusted initial information prediction model is trained again based on the server's sample performance parameters until a target information prediction model corresponding to a certain sample node set is obtained. This process is repeated, training the initial information prediction model for each node set to obtain a target information prediction model for each node set. Each node set corresponds to one target information prediction model.

[0165] It should be noted that after the central node obtains the target information prediction model corresponding to each node set, it will send multiple target information prediction models to the edge computing nodes. This allows the edge computing nodes to use the target information prediction models to process the performance parameters of the servers corresponding to each edge computing node for the user plane functional network element, obtain the fluctuation information of the performance parameters of each server, and then determine the target server from the servers corresponding to the edge computing nodes based on the fluctuation information.

[0166] In other embodiments of this application, the method may further include the following steps.

[0167] Step 304: The central node receives the real-time performance parameters of the server corresponding to each edge computing node sent by the edge computing node corresponding to the user plane function network element.

[0168] Among them, real-time performance parameters characterize the server's real-time resource usage.

[0169] In this embodiment of the application, the real-time performance parameters of the server corresponding to the edge computing node can refer to the real-time performance parameters of the server at different times. That is, the central node can receive and store the real-time performance parameters of each server corresponding to each edge computing node in each node set at different times.

[0170] Step 305: The central node processes the real-time performance parameters to obtain the processed real-time performance parameters for each server.

[0171] In this embodiment, the central node can obtain the real-time performance parameters of the server at different times within a certain time period based on the stored real-time performance parameters of the server at different times, and then normalize the real-time performance parameters of the server at different times within the certain time period to obtain the processed real-time performance parameters of each server. It should be noted that the process of normalizing the real-time performance parameters is the same as the process of normalizing the initial performance parameters in step 202.

[0172] Step 306: The central node updates the target information prediction model corresponding to the target node set based on the real-time performance parameters processed by each server.

[0173] The target node set is the set of nodes corresponding to the servers whose real-time performance parameters have been processed.

[0174] In this embodiment, the central node can determine the target node set corresponding to the server based on the processed real-time performance parameters of each server, and update the target information prediction model corresponding to the target node set to realize an autoencoded unsupervised learning neural network, thereby improving the accuracy of the fluctuation information of the determined server performance parameters.

[0175] It should be noted that the descriptions of the same steps and contents as in other embodiments in this embodiment can be found in the descriptions in other embodiments, and will not be repeated here.

[0176] The model training method provided in this application can divide multiple sample edge computing nodes into multiple sample node sets based on the location information of sample edge computing nodes corresponding to user plane functional network elements. Then, based on the sample performance parameters of each sample edge computing node in each sample node set, an initial information prediction model corresponding to each sample node set is trained to obtain multiple target information prediction models. The target information prediction models are then used to process the performance parameters of the servers corresponding to the edge computing nodes of the user plane functional network elements to obtain fluctuation information of the performance parameters of the target servers used to determine the target servers from the servers corresponding to the edge computing nodes. That is, the real-time resource usage of the servers corresponding to each sample edge computing node is considered when training the target information prediction model, making the fluctuation information of the performance parameters of the servers corresponding to the edge computing nodes determined by the target information prediction model more accurate. This allows for accurate determination of the target servers that need to adjust the data traffic they carry based on the fluctuation information, solving the problem of unreasonable utilization of system resources during traffic splitting in related technologies and thus avoiding resource waste.

[0177] Based on the foregoing embodiments, this application provides an information determining device, which can be applied to... Figures 1-2 In the information determination method provided in the corresponding embodiment, refer to Figure 5 As shown, the information determining device 4 may include: a first processing unit 41, an acquisition unit 42, a second processing unit 43, a first determining unit 44, and a sending unit 45, wherein:

[0178] The first processing unit 41 is used to obtain the location information of the edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple edge computing nodes into multiple node sets based on the location information.

[0179] The acquisition unit 42 is used to acquire the performance parameters of the server corresponding to each edge computing node in each node set; wherein, the performance parameters characterize the resource utilization of the server;

[0180] The second processing unit 43 is used to process the performance parameters using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; wherein, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the server corresponding to the sample edge computing node; different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes.

[0181] The first determining unit 44 is used to determine the target server from the servers corresponding to the edge computing nodes of multiple node sets based on fluctuation information and target threshold.

[0182] The sending unit 45 is used to send an adjustment instruction for adjusting the data traffic carried by the target server to the user plane function network element; wherein the adjustment instruction includes the identifier of the target server.

[0183] In other embodiments of this application, the acquisition unit 42 is further configured to perform the following steps:

[0184] Obtain the initial performance parameters of each server corresponding to each edge computing node in each node set;

[0185] The initial performance parameters are processed to obtain the performance parameters for each server.

[0186] In other embodiments of this application, the first determining unit 44 is further configured to perform the following steps:

[0187] Based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node, the target parameters of each server are determined; whereby the target parameters characterize the resource fluctuation of each server.

[0188] Based on target parameters and target thresholds, the target server is determined from the servers corresponding to edge computing nodes in multiple node sets.

[0189] In other embodiments of this application, the first determining unit 44 is further configured to perform the following steps:

[0190] Determine the weight of each performance parameter for each server;

[0191] Based on the fluctuation information of each performance parameter of each server and the weight of each performance parameter, the target parameters of each server are determined.

[0192] In other embodiments of this application, the sending unit 45 is further configured to perform the following steps:

[0193] The adjustment command is sent from the first target interface of the edge computing node corresponding to the target server to the first interface of the user plane function network element.

[0194] Correspondingly, the sending unit 45 is also used to receive feedback information on the adjustment operation of the data traffic carried by the target server sent by the first interface through the first target interface; wherein, the feedback information is used to characterize whether the adjustment operation of the data traffic carried by the target server is successful.

[0195] In other embodiments of this application, the sending unit 45 is further configured to perform the following steps:

[0196] The system receives query commands sent by the second interface of the user plane function network element through the second target interface, and determines the traffic information of the server corresponding to the identifier information based on the identifier information carried in the query command.

[0197] Traffic information is sent to the second interface through the second target interface of the edge computing node corresponding to the server with the identification information.

[0198] It should be noted that a detailed explanation of the steps performed by each unit can be found in [reference needed]. Figures 1-2 The description of the information determination method provided in the corresponding embodiments will not be repeated here.

[0199] The information determination device provided in the embodiments of this application can analyze the performance parameters of each server using a target information prediction model corresponding to different node sets to obtain the fluctuation information of the performance parameters. It can also combine the fluctuation information and the target threshold to determine the target server that needs to adjust the data traffic it carries and send the adjustment instruction to the user plane function network element. Thus, the user plane function network element can adjust the data traffic carried on the target server. That is, the real-time resource usage of the servers corresponding to each edge computing node is considered during the traffic splitting process, which solves the problem of unreasonable utilization of system resources in the related technologies, thereby avoiding resource waste.

[0200] Based on the foregoing embodiments, this application provides a model training device that can be applied to... Figure 3 In the model training method provided in the corresponding embodiment, refer to Figure 6 As shown, the model training device 5 may include: a third processing unit 51, a second determining unit 52, and a training unit 53, wherein:

[0201] The third processing unit 51 is used to obtain the location information of the sample edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple sample edge computing nodes based on the location information to obtain multiple sample node sets.

[0202] The second determining unit 52 is used to determine the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; wherein, the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node.

[0203] Training unit 53 is used to train the initial information prediction model corresponding to each sample node set based on the sample performance parameters to obtain multiple target information prediction models. Among them, the multiple target information prediction models are used to process the performance parameters of the server corresponding to the edge computing node corresponding to the user plane functional network element to obtain the fluctuation information of the performance parameters of the server corresponding to the edge computing node. The fluctuation information is used to determine the target server from the server corresponding to the edge computing node.

[0204] In other embodiments of this application, the second determining unit 52 is further configured to perform the following steps:

[0205] Receive the initial sample performance parameters of the server corresponding to the sample edge computing node sent by the sample edge computing node;

[0206] Based on the initial sample performance parameters of the servers corresponding to the sample edge computing nodes, determine the initial sample performance parameters of the servers corresponding to each sample edge computing node in each sample node set.

[0207] The initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are processed to obtain the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set.

[0208] In other embodiments of this application, the training unit 53 is further configured to perform the following steps:

[0209] The system receives real-time performance parameters of the server corresponding to each edge computing node from the edge computing node corresponding to the user plane function network element; where the real-time performance parameters characterize the server's real-time resource usage.

[0210] The real-time performance parameters are processed to obtain the processed real-time performance parameters for each server.

[0211] Based on the processed real-time performance parameters of each server, the target information prediction model corresponding to the target node set is updated; where the target node set is the node set corresponding to the server with the processed real-time performance parameters.

[0212] It should be noted that a detailed explanation of the steps performed by each unit can be found in [reference needed]. Figure 3 The model training method provided in the corresponding embodiments will not be described in detail here.

[0213] The model training apparatus provided in the embodiments of this application can consider the real-time resource usage of the servers corresponding to each sample edge computing node when training the target information prediction model. This makes the fluctuation information of the performance parameters of the servers corresponding to the edge computing nodes determined by the target information prediction model more accurate. In turn, the target server that needs to adjust the data traffic it carries can be accurately determined based on the fluctuation information. This solves the problem of unreasonable utilization of system resources when performing traffic splitting in related technologies, thereby avoiding resource waste.

[0214] Based on the foregoing embodiments, embodiments of this application provide an edge computing node that can be applied to... Figures 1-2 In the information determination method provided in the corresponding embodiment, refer to Figure 7 As shown, the edge computing node 6 may include: a first processor 61, a first memory 62, and a first communication bus 63, wherein:

[0215] The first communication bus 63 is used to realize the communication connection between the first processor 61 and the first memory 62;

[0216] The first processor 61 is used to execute the information determination program in the first memory 62 to perform the following steps:

[0217] Obtain the location information of the edge computing nodes corresponding to the user plane function network elements, and divide the multiple edge computing nodes into multiple node sets based on the location information;

[0218] Obtain the performance parameters of the server corresponding to each edge computing node in each node set; where the performance parameters characterize the server's resource utilization.

[0219] The performance parameters are processed using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters. The target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the server corresponding to the sample edge computing node. Different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes.

[0220] Based on fluctuation information and target threshold, the target server is determined from the servers corresponding to edge computing nodes in multiple node sets;

[0221] Send adjustment instructions to the user plane function network element to adjust the data traffic carried by the target server; wherein the adjustment instructions include the identifier of the target server.

[0222] In other embodiments of this application, the first processor 61 is used to execute the information determination program in the first memory 62 to obtain the performance parameters of each server corresponding to each edge computing node in each node set, so as to implement the following steps:

[0223] Obtain the initial performance parameters of each server corresponding to each edge computing node in each node set;

[0224] The initial performance parameters are processed to obtain the performance parameters for each server.

[0225] In other embodiments of this application, the first processor 61 is used to execute an information determination program in the first memory 62 to determine a target server from the servers corresponding to edge computing nodes of a plurality of node sets based on fluctuation information and a target threshold, in order to implement the following steps:

[0226] Based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node, the target parameters of each server are determined; whereby the target parameters characterize the resource fluctuation of each server.

[0227] Based on target parameters and target thresholds, the target server is determined from the servers corresponding to edge computing nodes in multiple node sets.

[0228] In other embodiments of this application, the first processor 61 is used to execute the information determination program in the first memory 62 to determine the target parameters of each server based on the fluctuation information of each performance parameter corresponding to each edge computing node, in order to implement the following steps:

[0229] Determine the weight of each performance parameter for each server;

[0230] Based on the fluctuation information of each performance parameter of each server and the weight of each performance parameter, the target parameters of each server are determined.

[0231] In other embodiments of this application, the first processor 61 is used to execute the information determination program in the first memory 62 to send adjustment instructions for adjusting the data traffic carried by the target server to the user plane function network element, so as to achieve the following steps:

[0232] The adjustment command is sent from the first target interface of the edge computing node corresponding to the target server to the first interface of the user plane function network element.

[0233] The system receives feedback information from the first interface regarding the adjustment operation of the data traffic carried by the target server; wherein the feedback information is used to indicate whether the adjustment operation of the data traffic carried by the target server was successful.

[0234] In other embodiments of this application, the first processor 61 is used to execute an information determination program in the first memory 62 to perform the following steps:

[0235] The system receives query commands sent by the second interface of the user plane function network element through the second target interface, and determines the traffic information of the server corresponding to the identifier information based on the identifier information carried in the query command.

[0236] Traffic information is sent to the second interface through the second target interface of the edge computing node corresponding to the server with the identification information.

[0237] It should be noted that a detailed description of the steps performed by the first processor can be found in [reference needed]. Figures 1-2 The information determination method provided in the corresponding embodiments will not be described again here.

[0238] The edge computing nodes provided in the embodiments of this application can analyze the performance parameters of each server using target information prediction models corresponding to different node sets to obtain performance parameter fluctuation information. Furthermore, they can combine the fluctuation information and target thresholds to determine the target server whose data traffic needs to be adjusted and send adjustment instructions to the user plane function network element. Thus, the user plane function network element can adjust the data traffic carried on the target server. In other words, the real-time resource usage of the servers corresponding to each edge computing node is considered during the traffic splitting process, solving the problem of unreasonable utilization of system resources during traffic splitting in related technologies, thereby avoiding resource waste.

[0239] Based on the foregoing embodiments, embodiments of this application provide a central node that can be applied to... Figure 3 In the model training method provided in the corresponding embodiment, refer to Figure 8 As shown, the central node 7 may include: a second processor 71, a second memory 72, and a second communication bus 73, wherein:

[0240] The second communication bus 73 is used to realize the communication connection between the second processor 71 and the second memory 72;

[0241] The second processor 71 is used to execute the model training program in the second memory 72 to perform the following steps:

[0242] Obtain the location information of the sample edge computing nodes corresponding to the user plane function network elements, and divide the multiple sample edge computing nodes based on the location information to obtain multiple sample node sets;

[0243] Determine the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; where the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node.

[0244] The initial information prediction model corresponding to each sample node set is trained based on the sample performance parameters to obtain multiple target information prediction models. Among them, the multiple target information prediction models are used to process the performance parameters of the servers corresponding to the edge computing nodes of the user plane functional network elements to obtain the fluctuation information of the performance parameters of the servers corresponding to the edge computing nodes. The fluctuation information is used to determine the target server from the servers corresponding to the edge computing nodes.

[0245] In other embodiments of this application, the second processor 71 is used to execute the model training program in the second memory 72 to determine the sample performance parameters of each server corresponding to each sample edge computing node in each sample node set, so as to achieve the following steps:

[0246] Receive the initial sample performance parameters of the server corresponding to the sample edge computing node sent by the sample edge computing node;

[0247] Based on the initial sample performance parameters of the servers corresponding to the sample edge computing nodes, determine the initial sample performance parameters of the servers corresponding to each sample edge computing node in each sample node set.

[0248] The initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are processed to obtain the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set.

[0249] In other embodiments of this application, the second processor 71 is used to execute the model training program in the second memory 72 to perform the following steps:

[0250] The system receives real-time performance parameters of the server corresponding to each edge computing node from the edge computing node corresponding to the user plane function network element; where the real-time performance parameters characterize the server's real-time resource usage.

[0251] The real-time performance parameters are processed to obtain the processed real-time performance parameters for each server.

[0252] Based on the processed real-time performance parameters of each server, the target information prediction model corresponding to the target node set is updated; where the target node set is the node set corresponding to the server with the processed real-time performance parameters.

[0253] It should be noted that a detailed description of the steps performed by the second processor can be found in [reference needed]. Figure 3 The model training method provided in the corresponding embodiments will not be described in detail here.

[0254] The central node provided in the embodiments of this application can take into account the real-time resource usage of the servers corresponding to each sample edge computing node when training the target information prediction model. This makes the fluctuation information of the performance parameters of the servers corresponding to the edge computing nodes determined by the target information prediction model more accurate. In turn, the target server that needs to adjust the data traffic it carries can be accurately determined based on the fluctuation information. This solves the problem of unreasonable utilization of system resources when performing traffic splitting in related technologies, thereby avoiding resource waste.

[0255] Based on the foregoing embodiments, embodiments of this application provide a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement... Figures 1-2 The corresponding embodiments provide information determination methods, and Figure 3The corresponding implementation provides the steps of the model training method.

[0256] Based on the foregoing embodiments, embodiments of this application provide a computer program product, which includes a computer program that is implemented when executed by a first processor 61 and a second processor 71. Figures 1-2 The corresponding embodiments provide information determination methods, and Figure 3 The corresponding implementation provides the steps of the model training method.

[0257] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0258] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0259] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0260] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0261] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for determining information, characterized in that, The method includes: Obtain the location information of the edge computing nodes corresponding to the user plane function network elements, and divide the multiple edge computing nodes based on the location information to obtain multiple node sets; Obtain the performance parameters of the server corresponding to each edge computing node in each node set; wherein, the performance parameters characterize the resource utilization of the server; The performance parameters are processed using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; wherein, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the servers corresponding to the sample edge computing nodes; the different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes; Based on the fluctuation information and the target threshold, the target server is determined from the servers corresponding to the edge computing nodes of the multiple node sets; Send an adjustment instruction to the user plane function network element to adjust the data traffic carried by the target server; wherein the adjustment instruction includes the identifier of the target server.

2. The method according to claim 1, characterized in that, The process of obtaining the performance parameters of the server corresponding to each edge computing node in each node set includes: Obtain the initial performance parameters of each server corresponding to each edge computing node in each node set; The initial performance parameters are processed to obtain the performance parameters of each server.

3. The method according to claim 1, characterized in that, The step of determining the target server from the servers corresponding to the edge computing nodes of the multiple node sets based on the fluctuation information and the target threshold includes: Based on the fluctuation information of each performance parameter of each server corresponding to the edge computing node, the target parameters of each server are determined; wherein, the target parameters characterize the resource fluctuation of the server; Based on the target parameters and the target threshold, the target server is determined from the servers corresponding to the edge computing nodes of the plurality of node sets.

4. The method according to claim 3, characterized in that, The determination of target parameters for each server based on fluctuation information of each performance parameter corresponding to the edge computing node includes: Determine the weight of each performance parameter for each server; The target parameters for each server are determined based on the fluctuation information of each performance parameter of each server and the weight of each performance parameter.

5. The method according to claim 1, characterized in that, Sending the adjustment instruction for adjusting the data traffic carried by the target server to the user plane function network element includes: The adjustment instruction is sent to the first interface of the user plane function network element through the first target interface of the edge computing node corresponding to the target server. Accordingly, the method further includes: The system receives feedback information from the first target interface regarding the adjustment operation of the data traffic carried by the target server; wherein the feedback information is used to indicate whether the data traffic carried by the target server has been successfully adjusted.

6. The method according to claim 5, characterized in that, The method further includes: The query command is received through the second target interface of the edge computing node corresponding to the first server, which sends a query instruction sent by the second interface of the user plane function network element; wherein, the first server is any one of the servers corresponding to the edge computing node corresponding to the user plane function network element; the query instruction includes the identifier of the first server; Based on the query command, the traffic information of the first server is determined, and the traffic information is sent to the second interface through the second target interface of the edge computing node corresponding to the first server.

7. A model training method, characterized in that, The method includes: Obtain the location information of the sample edge computing nodes corresponding to the user plane function network element, and divide the multiple sample edge computing nodes based on the location information to obtain multiple sample node sets; Determine the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; wherein, the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node; Based on the sample performance parameters, the initial information prediction model corresponding to each sample node set is trained to obtain multiple target information prediction models; wherein, the multiple target information prediction models are used to process the performance parameters of the server corresponding to the edge computing node corresponding to the user plane functional network element to obtain the fluctuation information of the performance parameters of the server corresponding to the edge computing node; the fluctuation information is used to determine the target server from the server corresponding to the edge computing node.

8. The method according to claim 7, characterized in that, The process of determining the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set includes: Receive the initial sample performance parameters of the server corresponding to the sample edge computing node sent by the sample edge computing node; Based on the initial sample performance parameters of the server corresponding to the sample edge computing node, the initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are determined. The initial sample performance parameters of the server corresponding to each sample edge computing node in each sample node set are processed to obtain the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set.

9. The method according to claim 7, characterized in that, The method further includes: The system receives real-time performance parameters of the server corresponding to each edge computing node sent by the edge computing node corresponding to the user plane function network element; wherein the real-time performance parameters characterize the real-time resource usage of the server. The real-time performance parameters are processed to obtain the processed real-time performance parameters for each server. Based on the processed real-time performance parameters of each server, the target information prediction model corresponding to the target node set is updated; wherein, the target node set is the node set corresponding to the server corresponding to the processed real-time performance parameters.

10. An information determining device, characterized in that, The device includes: The first processing unit is used to obtain the location information of the edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple edge computing nodes into multiple node sets based on the location information. The acquisition unit is used to acquire the performance parameters of the server corresponding to each edge computing node in each node set; wherein the performance parameters characterize the resource utilization of the server; The second processing unit is used to process the performance parameters using the target information prediction model corresponding to each node set to obtain the fluctuation information of the server's performance parameters; wherein, the target information prediction model is obtained by training the initial information prediction model of different sample node sets based on the sample performance parameters of the servers corresponding to the sample edge computing nodes; the different sample node sets are obtained by dividing multiple sample edge computing nodes based on the location information of the sample edge computing nodes. The first determining unit is used to determine the target server from the servers corresponding to the edge computing nodes of the plurality of node sets based on the fluctuation information and the target threshold. The sending unit is configured to send an adjustment instruction for adjusting the data traffic carried by the target server to the user plane function network element; wherein the adjustment instruction includes the identifier of the target server.

11. A model training device, characterized in that, The device includes: The third processing unit is used to obtain the location information of the sample edge computing nodes corresponding to the user plane functional network elements, and to divide the multiple sample edge computing nodes into multiple sample node sets based on the location information. The second determining unit is used to determine the sample performance parameters of the server corresponding to each sample edge computing node in each sample node set; wherein, the sample performance parameters characterize the resource utilization of the server corresponding to the sample edge computing node. The training unit is used to train the initial information prediction model corresponding to each sample node set based on the sample performance parameters to obtain multiple target information prediction models; wherein, the multiple target information prediction models are used to process the performance parameters of the server corresponding to the edge computing node corresponding to the user plane functional network element to obtain the fluctuation information of the performance parameters of the server corresponding to the edge computing node; the fluctuation information is used to determine the target server from the server corresponding to the edge computing node.

12. An edge computing node, characterized in that, The edge computing node includes: a first processor, a first memory, and a first communication bus; The first communication bus is used to establish a communication connection between the first processor and the first memory; The first processor is configured to execute the information determination program in the first memory to implement the steps of the information determination method as described in any one of claims 1-6.

13. A central node, characterized in that, The central node includes: a second processor, a second memory, and a second communication bus; The second communication bus is used to establish a communication connection between the second processor and the second memory; The second processor is used to execute the model training program in the second memory to implement the steps of the model training method as described in any one of claims 7-9.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the information determination method as described in any one of claims 1 to 6, or the model training method as described in any one of claims 7 to 9.

15. A computer program product, the computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6 or 7 to 9.

Citation Information

Patent Citations

  • Method and device for detecting abnormal node

    CN108228442A

  • Network performance determination method and device, server and storage medium

    CN116634453A