Management method and device

Abstract

The invention provides a management method and device, and the method comprises the steps: a first network management node determines the capability information of artificial intelligence (AI) / machine learning (ML), and the capability information of the AI / ML comprises one or more of the following: the general capability of the AI / ML, the learning capability of the AI / ML, the credible capability of the AI / ML, or the processing capability of the AI / ML; the first network management node sends the capability information of the AI / ML to a second network management node, and the capability information of the AI / ML and an acquisition mode of the capability information of the AI / ML are defined by adopting the method and the device, so that the second network management node indicates the capability demand of the AI / ML to the first network management node; therefore, the first network management node executes the capability corresponding to the AI / ML according to the capability demand of the AI / ML so as to improve the network performance.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a management method and device. Background Art

[0002] At present, artificial intelligence (AI) / machine learning (ML) technologies and related applications are being widely adopted by more and more industries. The functions of AI / ML are used in various fields, for example, they can be applied in management data analysis (MDA) and network data analytical function (NWDAF).

[0003] The 3rd Generation Partnership Project (3GPP) defines the use case of AI / ML capability discovery and mapping, but does not specifically define the capabilities of AI / ML and the capability control of AI / ML. Summary of the invention

[0004] The present application proposes a management method and device, which define the capability information of AI / ML and the method for obtaining the capability information of AI / ML, and enable the second network management node to indicate the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance.

[0005] In a first aspect, an embodiment of the present application provides a management method, the method comprising: a first network management node determines capability information of an artificial intelligence (AI) machine learning (ML), wherein the capability information of the AI / ML comprises one or more of the following: general capabilities of the AI / ML, learning capabilities of the AI / ML, trusted capabilities of the AI / ML, or processing capabilities of the AI / ML; the first network management node sends the capability information of the AI / ML to a second network management node.

[0006] In the above method, the capability information of AI / ML may refer to the capability information of ML, or the capability information of AI, or may also refer to the capability information of the AI / ML model, that is, it may refer to the capability information of the AI ​​model, or the capability information of the ML model, wherein the AI / ML model, the AI ​​model, the ML model may also be referred to as an AI / ML entity, an AI entity, an ML entity, and AI / ML may also be AIML, or other terms that can characterize AI and ML. The embodiment of the present application defines the capability information of AI / ML and the method for obtaining the AI / ML capability information, so that the second network management node can indicate the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance.

[0007] In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0008] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0009] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0010] In yet another possible implementation, the first network management node receives a request message from the second network management node, where the request message is used to request capability information of the AI / ML.

[0011] In yet another possible implementation, the first network management node sends updated AI / ML capability information to the second network management node.

[0012] In another possible implementation, the first network management node receives first indication information from the second network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML; the first network management node performs model training according to the first indication information to obtain a training result.

[0013] In the above method, the second network management node can send the first indication information to the first network management node according to the obtained learning capability of AI / ML, indicating the target learning capability. In this way, the second network management node can select suitable learning methods for different business scenarios, that is, suitable learning capabilities of AI / ML training, so that the first network management node can be trained according to the target learning capability indicated by the first indication information, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML to the first network management node (for example, different business scenarios are suitable for different learning methods), so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the learning capability of AI / ML, to improve network performance.

[0014] In another possible implementation, the first network management node receives second indication information from the second network management node, where the second indication information is used to indicate an indicator of a target trusted capability, and the target trusted capability is one of the trusted capabilities of the AI / ML; the first network management node performs a trusted assessment of the indicator of the target trusted capability according to the second indication information to obtain a trusted assessment result.

[0015] In the above method, the second network management node can send a second indication message to the first network management node based on the acquired trusted capability of AI / ML, indicating the target trusted capability indicator, so that the first network management node performs a trusted evaluation of the target trusted capability indicator to obtain a trusted evaluation result, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML (for example, the requirements of the trusted capability of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the trusted capability of AI / ML, to improve network performance.

[0016] In another possible implementation, the first network management node receives third indication information from the second network management node, where the third indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve; the first network management node performs model reasoning based on the third indication information and the reasoning capability of the AI / ML.

[0017] In the above method, by the first network management node receiving the third indication information from the second network management node, the second network management node can make a corresponding request to the first network management node according to the demand, that is, the key performance indicator KPI target or performance measurement target that the model needs to achieve, so that the first network management node performs model reasoning according to the demand of the second network management node, that is, the third indication information, and the determined AI / ML reasoning capability, thereby improving network performance. That is, the second network management node indicates the AI / ML capability requirements (for example, the AI / ML reasoning capability requirements) to the first network management node, so that the first network management node executes the corresponding AI / ML capabilities according to the AI / ML capability requirements, for example, the AI / ML reasoning capability, to improve network performance.

[0018] In another possible implementation, the first network management node receives fourth indication information from the second network management node, the fourth indication information including a correspondence between a business scenario identifier and a threshold, the threshold including one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the first network management node deploys a model based on the fourth indication information and the processing capability of the AI / ML.

[0019] In the above method, the first network management node receives the fourth indication information from the second network management node. For example, the second network management node may be an operator, so that the operator can specify the fourth indication information, that is, specify the requirements of the model in a specific scenario, so that the first network management node can deploy the model according to the requirements of the model in the specific scenario. For example, for scenarios with high latency sensitivity such as video games, it is necessary to deploy a model with high responsiveness, so as to improve network performance and user experience. That is, the second network management node indicates the capability requirements of AI / ML (for example, the processing capability requirements of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the processing capability of AI / ML, to improve network performance.

[0020] In yet another possible implementation, the first network management node sends deployment result information to the second network management node, where the deployment result information includes deployment success or deployment failure.

[0021] In a second aspect, an embodiment of the present application provides a management method, the method comprising: a second network management node receives artificial intelligence AI / machine learning ML capability information from a first network management node, wherein the AI / ML capability information comprises one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML.

[0022] In the above method, the capability information of AI / ML may refer to the capability information of ML, or the capability information of AI, or may also refer to the capability information of the AI / ML model, that is, it may refer to the capability information of the AI ​​model, or the capability information of the ML model, wherein the AI / ML model, the AI ​​model, the ML model may also be referred to as an AI / ML entity, an AI entity, an ML entity, and AI / ML may also be AIML, or other terms that can characterize AI and ML. The embodiment of the present application defines the capability information of AI / ML and the method for obtaining the AI / ML capability information, so that the second network management node can indicate the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance. In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0023] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0024] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0025] In yet another possible implementation, the second network management node sends a request message to the first network management node, where the request message is used to request capability information of the AI / ML.

[0026] In yet another possible implementation, the second network management node receives updated AI / ML capability information from the first network management node.

[0027] In another possible implementation, the second network management node sends first indication information to the first network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML, and the first indication information is used for model training.

[0028] In the above method, the second network management node can send a first indication message to the first network management node based on the acquired learning capability of AI / ML, indicating the target learning capability. In this way, the second network management node can select a suitable learning method for different business scenarios, that is, a suitable learning capability of AI / ML, so that the first network management node can be trained according to the target learning capability indicated by the first indication message, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML to the first network management node (for example, different business scenarios are suitable for different learning methods), so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the learning capability of AI / ML, to improve network performance.

[0029] In another possible implementation, the second network management node sends second indication information to the first network management node, where the second indication information is used to indicate an indicator of a target trusted capability, where the target trusted capability is one of the trusted capabilities of the AI / ML, and the target trusted capability is used for trusted assessment.

[0030] In the above method, the second network management node can send a second indication message to the first network management node based on the acquired trusted capability of AI / ML, indicating the target trusted capability, so that the first network management node performs a trusted evaluation of the target trusted capability indicator to obtain a trusted evaluation result, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML (for example, the requirements of the trusted capability of AI / ML) to the first network management node, so that the first network management node executes the corresponding capability of AI / ML according to the capability requirements of AI / ML, for example, the trusted capability of AI / ML, to improve network performance.

[0031] In another possible implementation, the second network management node sends third indication information to the first network management node, where the second indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve, and the second indication information is used for model reasoning.

[0032] In the above method, by receiving the third indication information from the second network management node by the first network management node, the second network management node can make a corresponding request to the first network management node according to the demand, that is, the key performance indicator KPI target or performance measurement target that the model needs to achieve, so that the first network management node performs model reasoning according to the demand of the second network management node, that is, the third indication information, and the determined AI / ML reasoning capability, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML (for example, the requirements of the reasoning capability of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the reasoning capability of AI / ML, to improve network performance. In another possible implementation, the second network management node sends a fourth indication information to the first network management node, and the fourth indication information includes a corresponding relationship between a business scenario identifier and a threshold, and the threshold includes one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the fourth indication information is used for model deployment.

[0033] In the above method, the first network management node receives the fourth indication information from the second network management node. For example, the second network management node may be an operator, so that the operator can specify the fourth indication information, that is, specify the requirements of the model in a specific scenario, so that the first network management node can deploy the model according to the requirements of the model in the specific scenario. For example, for scenarios with high latency sensitivity such as video games, it is necessary to deploy a model with high responsiveness, so as to improve network performance and user experience. That is, the second network management node indicates the capability requirements of AI / ML (for example, the processing capability requirements of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the processing capability of AI / ML, so as to improve network performance.

[0034] In yet another possible implementation, the second network management node receives deployment result information from the first network management node, where the deployment result information includes deployment success or deployment failure.

[0035] In a third aspect, an embodiment of the present application provides a management method, the method comprising: a first network management node determines the general capabilities of artificial intelligence AI / machine learning ML, the general capabilities of AI / ML comprising one or more of the following: training capabilities of AI / ML, testing capabilities of AI / ML, reasoning capabilities of AI / ML, or simulation capabilities of AI / ML; the first network management node sends the general capabilities of AI / ML to a second network management node.

[0036] In a fourth aspect, an embodiment of the present application provides a management method, the method comprising: a second network management node receives general capabilities of artificial intelligence AI / machine learning ML from a first network management node, wherein the general capabilities of AI / ML include one or more of the following: training capabilities of AI / ML, testing capabilities of AI / ML, reasoning capabilities of AI / ML, or simulation capabilities of AI / ML.

[0037] In a fifth aspect, an embodiment of the present application provides a management method, which includes: a first network management node determines the learning capability of artificial intelligence AI / machine learning ML, and the learning capability of AI / ML includes one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the first network management node sends the learning capability of AI / ML to a second network management node.

[0038] In a sixth aspect, an embodiment of the present application provides a management method, the method comprising: a second network management node receives the learning capability of artificial intelligence AI / machine learning ML from a first network management node, wherein the learning capability of AI / ML comprises one or more of the following: federated learning, transfer learning, generative learning or reinforcement learning.

[0039] In the seventh aspect, an embodiment of the present application provides a management method, which includes: a first network management node determines the trusted capabilities of artificial intelligence AI / machine learning ML, and the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the first network management node sends the trusted capabilities of the AI / ML to a second network management node.

[0040] In an eighth aspect, an embodiment of the present application provides a management method, the method comprising: a second network management node receives trusted capabilities of artificial intelligence AI / machine learning ML from a first network management node, wherein the trusted capabilities of AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability.

[0041] In a ninth aspect, an embodiment of the present application provides a management method, the method comprising: a first network management node determines the processing capability of artificial intelligence AI / machine learning ML, the processing capability of AI / ML comprising one or more of the following: training processing capability of AI / ML, testing processing capability of AI / ML, reasoning processing capability of AI / ML or simulation processing capability of AI / ML; the first network management node sends the processing capability of AI / ML to a second network management node.

[0042] In the tenth aspect, an embodiment of the present application provides a management method, the method comprising: a second network management node receives the processing capability of artificial intelligence AI / machine learning ML from a first network management node, the processing capability of AI / ML comprising one or more of the following: training processing capability of AI / ML, testing processing capability of AI / ML, reasoning processing capability of AI / ML or simulation processing capability of AI / ML.

[0043] In the eleventh aspect, an embodiment of the present application provides a management device, which is a first network management node, comprising: a processing unit and a transceiver unit, the processing unit is used to determine the capability information of artificial intelligence AI / machine learning ML, and the capability information of AI / ML includes one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML; the transceiver unit is used to send the capability information of AI / ML to the second network management node.

[0044] In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0045] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0046] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0047] In yet another possible implementation, the transceiver unit is used to receive a request message from the second network management node, where the request message is used to request capability information of the AI / ML.

[0048] In yet another possible implementation, the transceiver unit is configured to send updated AI / ML capability information to the second network management node.

[0049] In another possible implementation, the transceiver unit is used to receive first indication information from the second network management node, where the first indication information is used to indicate a target learning capability, and the target learning capability is one of the learning capabilities of the AI / ML; and the processing unit is used to perform model training according to the first indication information to obtain a training result.

[0050] In another possible implementation, the transceiver unit is used to receive second indication information from the second network management node, where the second indication information is used to indicate an indicator of a target trusted capability, and the target trusted capability is one of the trusted capabilities of the AI / ML; the processing unit is used to perform a trusted assessment of the indicator of the target trusted capability to obtain a trusted assessment result.

[0051] In another possible implementation, the transceiver unit is used to receive third indication information from the second network management node, where the third indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve; and the processing unit is used to perform model reasoning based on the third indication information and the reasoning capability of the AI / ML.

[0052] In another possible implementation, the transceiver unit is used to receive fourth indication information from the second network management node, the fourth indication information including a correspondence between a business scenario identifier and a threshold, the threshold including one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the processing unit is used to deploy a model according to the fourth indication information and the processing capability of the AI / ML.

[0053] In yet another possible implementation, the transceiver unit is configured to send deployment result information to the second network management node, where the deployment result information includes deployment success or deployment failure.

[0054] Regarding the technical effects brought about by the eleventh aspect or possible implementation methods, reference may be made to the introduction to the technical effects of the first aspect or corresponding implementation methods.

[0055] In a twelfth aspect, an embodiment of the present application provides a management device, which may be a second network management node, comprising: a processing unit and a transceiver unit, the transceiver unit being used to receive capability information of artificial intelligence AI / machine learning ML from the first network management node, wherein the capability information of AI / ML comprises one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML.

[0056] In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0057] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0058] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0059] In yet another possible implementation, the transceiver unit is used to send a request message to the first network management node, where the request message is used to request capability information of the AI / ML.

[0060] In yet another possible implementation, the transceiver unit is configured to receive updated AI / ML capability information from the first network management node.

[0061] In another possible implementation, the transceiver unit is used to send first indication information to the first network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML, and the first indication information is used for model training.

[0062] In another possible implementation, the transceiver unit is used to send second indication information to the first network management node, where the second indication information is used to indicate an indicator of a target trusted capability, where the target trusted capability is one of the trusted capabilities of the AI / ML, and where the target trusted capability is used for trusted assessment.

[0063] In another possible implementation, the transceiver unit is used to send third indication information to the first network management node, the second indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve, and the second indication information is used for model reasoning.

[0064] In another possible implementation, the transceiver unit is used to send fourth indication information to the first network management node, and the fourth indication information includes a correspondence between a business scenario identifier and a threshold, and the threshold includes one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the fourth indication information is used for model deployment.

[0065] In yet another possible implementation, the transceiver unit is configured to receive deployment result information from the first network management node, where the deployment result information includes deployment success or deployment failure.

[0066] Regarding the technical effects brought about by the twelfth aspect or possible implementation methods, reference may be made to the introduction to the technical effects of the second aspect or corresponding implementation methods.

[0067] In the thirteenth aspect, an embodiment of the present application provides a management device, which may be a first network management node. The management device includes at least one processor and a communication interface. The at least one processor calls a computer program or instruction stored in a memory to execute the method described in the first aspect or a possible implementation method in the first aspect.

[0068] In the fourteenth aspect, an embodiment of the present application provides a management device, which may be a second network management node. The management device includes at least one processor and a communication interface. The at least one processor calls a computer program or instruction stored in a memory to execute the method described in the above-mentioned second aspect or a possible implementation method in the second aspect.

[0069] In a fifteenth aspect, an embodiment of the present application provides a chip device, wherein the chip device includes at least one processor, and the at least one processor is used to execute computer programs or instructions to implement the method described in any one of the above aspects.

[0070] In the sixteenth aspect, an embodiment of the present application provides a computer-readable storage medium, in which a computer program or instructions are stored. When the computer program or instructions are executed on a processor, the method described in any one of the above aspects is implemented.

[0071] In the seventeenth aspect, an embodiment of the present application provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed on a computer, the method described in any one of the above aspects is implemented.

[0072] In the eighteenth aspect, an embodiment of the present application provides a communication system, which includes: the device as described in the eleventh aspect and the device as described in the twelfth aspect.

[0073] In the above method, the capability information of AI / ML and the method of obtaining the AI / ML capability information are defined through the above method, so that the second network management node indicates the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 It is a schematic diagram of a service-oriented management architecture provided in an embodiment of the present application;

[0075] Figure 2 is a schematic diagram of a management function entity provided in an embodiment of the present application;

[0076] Figure 3 is a schematic diagram of a producer and a consumer provided in an embodiment of the present application;

[0077] Figure 4 It is a schematic diagram of a MnS Consumer triggering AI / ML model training provided in an embodiment of the present application;

[0078] Figure 5 is a schematic diagram of an AI / ML capability discovery provided by an embodiment of the present application;

[0079] Figure 6 It is a flowchart of a management method provided by an embodiment of the present application;

[0080] Figure 7 It is a flowchart of another management method provided in an embodiment of the present application;

[0081] Figure 8 It is a structural diagram of a management device provided in an embodiment of the present application;

[0082] Fig. 9 It is a structural diagram of another management device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0083] The technical solutions in the embodiments of the present application are described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0084] References to "one embodiment" or "some embodiments" etc. described in this application mean that a particular feature, structure or characteristic described in conjunction with the embodiment is included in one or more embodiments of the present application. Thus, the phrases "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear at different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0085] In the description of this application, unless otherwise specified, " / " means "or", for example, A / B can mean A or B. "And / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, "at least one" means one or more, and "plurality" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a, b, and c. Among them, a, b, and c can be single or multiple.

[0086] It can be understood that in this application, "indication" can include direct indication, indirect indication, explicit indication, implicit indication. When describing that a certain indication information is used to indicate A, it can be understood that the indication information carries A, directly indicates A, or indirectly indicates A.

[0087] In this application, the information indicated by the indication information is referred to as the information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, the information to be indicated can be directly indicated, such as the information to be indicated itself or the index of the information to be indicated, etc., or the information to be indicated can be indirectly indicated by indicating other information, wherein there is an association relationship between the other information and the information to be indicated. It is also possible to indicate only a part of the information to be indicated, while the other parts of the information to be indicated are known or agreed in advance. For example, the indication of specific information can also be achieved with the help of the arrangement order of each information agreed in advance (such as specified by the protocol), thereby reducing the indication overhead to a certain extent.

[0088] The information to be indicated can be sent as a whole or divided into multiple sub-information and sent separately, and the sending period and / or sending time of these sub-information can be the same or different. The specific sending method is not limited in this application. Among them, the sending period and / or sending time of these sub-information can be pre-defined, for example, pre-defined according to a protocol, or can be configured by the transmitting end device by sending configuration information to the receiving end device.

[0089] It can be understood that "sending" and "receiving" in this application indicate the direction of signal transmission. For example, "sending information to XX" can be understood as the destination of the information is XX, which can include direct sending through the air interface, and also includes indirect sending through the air interface by other units or modules. "Receiving information from YY" can be understood as the source of the information is YY, which can include directly receiving from YY through the air interface, and also includes indirectly receiving from YY through the air interface from other units or modules. "Sending" can also be understood as the "output" of the chip interface, and "receiving" can also be understood as the "input" of the chip interface.

[0090] In other words, sending and receiving can be performed between devices, for example, between a network device and a terminal device, or can be performed within a device, for example, sending or receiving between components, modules, chips, software modules, or hardware modules within the device through a bus, wiring, or interface.

[0091] The communication method provided in the embodiment of the present application can be applied to cellular communication systems related to the third generation partnership project (3GPP), for example, fourth generation (4G) communication systems, such as long term evolution (LTE) communication systems, and can also be applied to fifth generation (5G) communication systems, such as 5G new radio (NR) communication systems, or to various future communication systems, such as sixth generation (6G) communication systems. The method provided in the embodiment of the present application can also be applied to Bluetooth systems, wireless fidelity (WiFi) systems, LoRa systems or Internet of Vehicles systems, communication systems that support the integration of multiple wireless technologies, and device-to-device (D2D) systems. The method provided in the embodiment of the present application can also be applied to satellite communication systems, wherein the satellite communication system can be integrated with the above-mentioned communication system. The wireless communication systems involved in this application also include but are not limited to: narrowband Internet of Things system (NB-IoT), global system for mobile communications (GSM), enhanced data rate for GSM evolution (EDGE), wideband code division multiple access system (WCDMA), code division multiple access 2000 system (CDMA2000), or time division-synchronization code division multiple access system (TD-SCDMA).

[0092] See also Figure 1 , Figure 1It is a schematic diagram of a service-oriented management architecture provided in an embodiment of the present application. Among them, the service-oriented management architecture includes a business support system (BSS), a cross domain management function unit (CD-MnF), a domain management function unit (Domain-MnF) and a network element (element). It should be understood that the cross-domain management function unit can be a network management system (NMS), a management service producer (MnS Producer), a management service consumer (MnS Consumer) and other nodes, and the domain management function unit can be an element management system (EMS), a wireless automation engine (MAE), a MnS Producer, a MnS Consumer and other nodes.

[0093] If the management service is a management service provided by a cross-domain management functional unit, then the cross-domain management functional unit is the management service producer, and the business support system is the management service consumer.

[0094] If the management service is a management service provided by a domain management functional unit, then the domain management functional unit is a management service producer, and the cross-domain management functional unit is a management service consumer.

[0095] When the management service is a management service provided by a network element, the network element is a management service producer, and the domain management functional unit is a management service consumer.

[0096] The business support system is oriented to communication services and is used to provide functions and management services such as billing, settlement, accounting, customer service, business, network monitoring, communication service life cycle management, and business intent translation. The business support system can be an operator's operating system or a vertical industry operating system (vertical OTsystem).

[0097] The cross-domain management function unit, also called the network management function unit (network management function, NMF), can be, for example, a network management entity such as NMS, network function management service consumer (network function management service consumer, NFMS_C), etc. The cross-domain management function unit provides one or more of the following management functions or management services: network life cycle management, network deployment, network fault management, network performance management, network configuration management, network assurance, network optimization function, and translation of the network intent (intent from communication service provider, Intent-CSP) of the service producer, etc.

[0098] Among them, the network referred to in the above management function or management service may include one or more network elements or subnetworks, or it may be a network slice. That is to say, the network management function unit may be a network slice management function unit (network slice management function, NSMF), or a cross-domain management data analytical function unit (management data analytical function, MDAF), or a cross-domain self-organization network function (self-organization network function, SON Function), or a cross-domain AIML management function unit (AI / ML management service, AI / ML function) or a cross-domain intent driven management function unit (intent driven management service, IDMS).

[0099] Optionally, in some deployment scenarios, the cross-domain management functional unit can also provide sub-network lifecycle management, sub-network deployment, sub-network fault management, sub-network performance management, sub-network configuration management, sub-network assurance, sub-network optimization functions, sub-network service producer network intent (Intent-CSP) or sub-network service consumer network intent (intent from communication service consumer, Intent-CSC) translation, etc. The sub-network here is composed of multiple small sub-networks, which can be a network slice sub-network.

[0100] Domain management function (Domain-MnF), also called network management function (NMF) or network element management function. For example, the domain management function can be a network element management entity such as MAE, EMS, network function management service provider (NFMS_P), etc.

[0101] The domain management functional unit provides one or more of the following functions or management services: lifecycle management of sub-networks or network elements, deployment of sub-networks or network elements, fault management of sub-networks or network elements, performance management of sub-networks or network elements, assurance of sub-networks or network elements, optimization functions of sub-networks or network elements, and translation of intent (intent from network operator, Intent-NOP) of sub-networks or network elements. The sub-network here includes one or more network elements. A sub-network may also include a sub-network, that is, one or more sub-networks form a larger sub-network.

[0102] Optionally, the subnetwork here can also be a network slice subnetwork. The domain management system can be a network slice subnet management function unit (network slice subnet management function, NSSMF), a domain management data analytical function unit (management data analytical function, Domain MDAF), a domain self-organization network function (self-organization network function, SON Function), a domain intent management function unit IntentDriven MnS, etc.

[0103] The domain management functional units can be classified as follows:

[0104] According to the network type, it can be divided into: access network domain management function unit (radio access network domain management function, RAN-Domain-MnF), core network domain management function unit (core network domain management function, CN-Domain-MnF), transport network domain management function unit (transport network domain management function, TN-Domain-MnF), etc. It should be noted that the domain management function unit can also be a domain network management system, which can manage one or more of the access network, core network or transport network;

[0105] According to the administrative region classification, it can be divided into: domain management functional units of a certain area, such as Shanghai domain management functional unit, Beijing domain management functional unit, etc.

[0106] Network elements are entities that provide network services, including core network elements and access network elements. Among them, core network elements include: access and mobility management function (AMF), session management function (SMF), policy control function (PCF), network data analytical function (NWDAF), network repository unit (NFRepository Function, NRF) and gateways. Access network elements include: base stations (such as gNB, eNB), central unit control plane (CUCP), central unit (CU), distributed unit (DU), central unit user plane (CUUP), etc.

[0107] Among them, the network element can provide one or more of the following management functions or management services: network element lifecycle management, network element deployment, network element fault management, network element performance management, network element assurance, network element optimization function and network element intent translation, etc.

[0108] See also Figure 2 , Figure 2 It is a schematic diagram of a management function entity provided by this application; Figure 2 is Figure 1Further explanation of the service-oriented management architecture based on the above. Management function (MnF) is a management entity defined by the 3rd generation partnership project (3GPP), and its externally visible behavior and interface are defined as management services (MnS). In the service-oriented management architecture, MnF can play the role of MnS Producer or MnS Consumer. For example, Figure 2 As shown, the management service produced by the management service producer played by MnF may have multiple consumers. MnF can consume multiple management services from one or more management service producers. In other words, the management service producer played by MnF can manage multiple management service consumers and can also obtain management information from multiple management service consumers.

[0109] Exemplarily, the management service producer may be a network element management entity such as MAE or EMS.

[0110] Exemplarily, the management service consumer may specifically be an NMS or the like.

[0111] Among them, management service producers (such as MAE, EMS, etc.) and management service consumers (such as NMS) can be collectively referred to as network management entities. In other words, the network management entity can refer to management service producers, management service consumers, NMS, MAE, or EMS.

[0112] See also Figure 3 , Figure 3 This is a schematic diagram of a producer and consumer provided in an embodiment of the present application. The training capability of the AI / ML model is provided by the MnS Producer to the authorized MnS Consumer through the model training management service. For example, the MnS Producer sends the training capability of the AI / ML model to the MnS Consumer through the AI / ML model training management service service interface. The AI / ML model training process can include two triggering methods:

[0113] In the first method, AI / ML model training is triggered by MnS Consumer. AI / ML model training can be triggered by requests from one or more MnS Consumers. MnS Consumers may provide source data as input for training data. MnS Producer returns a response message to notify MnS Consumer whether the training request is accepted. Figure 4 , Figure 4This is a schematic diagram of an MnS Consumer triggering AI / ML model training provided in an embodiment of the present application, which is as follows:

[0114] Step 1: MnS Consumer sends an AI / ML model training request message to MnS Producer.

[0115] Among them, the interface between MnS Consumer and MnS Producer can be a machine learning model training management service (machine learning model training, MLMT MnS) service interface, and MnS Consumer sending an AI / ML model training request message to MnSProducer can be MnS Consumer sending an AI / ML model training request message to MnSProducer through the MLMT MnS service interface.

[0116] Optionally, the MnS Consumer may indicate the model function or purpose of the AI / ML model, such as coverage problem analysis; the MnS Consumer may also send source data to the MnS Producer, which may be input for candidate training data.

[0117] Step 2: MnS Producer sends a response message to MnS Consumer.

[0118] The response message is a response to the training request message.

[0119] Step 3: MnS Producer sends the training results to MnS Consumer.

[0120] Before sending the training result to the MnS Consumer, the MnS Producer selects the training data. Optionally, the MnS Producer may select the source data provided by the MnS Consumer as the training data. Optionally, the MnS Producer selects other available data as the training data. The MnS Producer uses the training data to train the AI / ML model.

[0121] Among them, the training results may include the location of the trained AI / ML model, etc.

[0122] In the second way, AI / ML model training is triggered by MnS Producer, for example, based on feedback received from consumers, as a result of AI / ML model performance evaluation, to trigger AI / ML model training, or when new training data describing new network states / events is available. Then, MnS Producer selects training data, and optionally, MnS Producer can select source data provided by MnS Consumer as training data. Optionally, MnS Producer selects other available data as training data. MnS Producer trains AI / ML model using training data. Then, MnS Producer sends training results to MnS Consumer that subscribes to receive training results. Optionally, the training results may include the location of the trained AI / ML model, etc.

[0123] See also Figure 5 , Figure 5 This is a schematic diagram of an AI / ML capability discovery provided by an embodiment of the present application, which is as follows:

[0124] Step 1: MnS Consumer sends a request message to MnS Producer.

[0125] The request message is used to request AI / ML capabilities.

[0126] Step 2: MnS Producer sends AI / ML capabilities to MnS Consumer.

[0127] Depend on Figure 5It can be seen that the above is a use case for the discovery and mapping of AI / ML capabilities, but the capabilities of AI / ML are not specifically defined in detail. For example, for the processing capabilities of the reasoning function of AI / ML, different entities that apply ML models to the AI / ML reasoning function may have different reasoning requirements and capabilities, so the processing capabilities of the target reasoning function are also of great significance to model selection and deployment. For example, when producers perform model selection and deployment, they also need to consider other needs of consumers, such as the requirements for the processing capabilities of the reasoning entities applicable to different models (latency sensitivity, available resources) and energy consumption requirements. The producer comprehensively selects the appropriate model deployment based on the processing capabilities and requirements. For example, the processing capabilities of the reasoning entity also have an impact on the consumer's execution model selection and deployment. For example, for reasoning scenarios with high latency sensitivity such as video games, it is necessary to select a model with high deployment responsiveness. For example, the available resources for reasoning, such as memory size and CPU conditions, also have an impact on model selection. Therefore, how to obtain the processing capabilities of these reasonings in this scenario to select a suitable model deployment to meet the reasoning requirements is a technical problem that people in this field are solving. In order to solve the above problems, the embodiments of this application propose the following solutions.

[0128] based on Figure 1 and Figure 2 The service-oriented management architecture provided by this application provides a management method. Figure 6 , Figure 6 It is a flow chart of a management method provided in an embodiment of the present application. The method may include the following steps S601-S603.

[0129] Step S601: The first network management node determines the capability information of AI / ML.

[0130] The first network management node is a producer, for example, it can be an AI / ML producer (AI / MLproducer), an AI / ML training producer, an AI / ML reasoning producer, an AI / ML test producer, an AI / ML simulation producer, etc. The first network management node can be a Domain-MnF, for example, it can be an EMS or a MAE.

[0131] It should be noted that the first network management node determines the AI / ML capability information, which can also be understood as the AI / ML capability information possessed by the first network management node.

[0132] It should also be noted that the AI / ML capability information here may refer to the capability information of ML, or the capability information of AI, or it may also refer to the capability information of the AI / ML model, that is, it may refer to the capability information of the AI ​​model, or the capability information of the ML model, and this application does not limit this.

[0133] It should also be noted that the AI / ML model, AI model, and ML model referred to in this application may also be referred to as AI / ML entity, AI entity, and ML entity.

[0134] It should also be noted that the AI / ML referred to in this application may also be AIML, or other terms that can represent AI and ML, and this application is not limited thereto.

[0135] It should also be noted that the model mentioned in this application refers to the AI / ML model.

[0136] The capability information of AI / ML includes one or more of the following: general capability of AI / ML, learning capability of AI / ML, trust capability of AI / ML, or processing capability of AI / ML. It should be noted that the capability information of AI / ML defined in the embodiment of the present application can be applied to Figure 4 The AI / ML model training triggered by MnS Consumer in the process can also be applied to the AI / ML model training triggered by MnSProducer, which is not limited in the embodiments of the present application.

[0137] Among them, the general capabilities of AI / ML include one or more of the following: AI / ML training capabilities, AI / ML testing capabilities, AI / ML reasoning capabilities, or AI / ML simulation capabilities. The training capabilities of AI / ML include joint training capabilities, which include the ability to jointly train between models. Optionally, it can be the ability to jointly train between two models, or it can refer to the ability to jointly train between multiple models, which is not limited in the embodiments of the present application. In one example, the joint training capability may include the ability to jointly train between model 1 and model 2, for example, the output result of model 1 is the input data of model 2, and joint training is performed between model 1 and model 2. The testing capabilities of AI / ML include joint testing capabilities, which include the ability to jointly test between models. Optionally, it can be the ability to jointly test between two models, or it can refer to the ability to jointly test between multiple models, which is not limited in the embodiments of the present application. In one example, the joint testing capability may include the ability to jointly test between model A and model B. The reasoning capabilities of AI / ML include analysis capabilities and decision-making capabilities. For example, the analysis capability may include network data analysis capabilities and / or management data analysis capabilities. In one example, the management data analysis capability may include coverage analysis, mobility analysis, etc. (specifically, reference may be made to the use case scenario of management data analysis defined in 3GPP TS28.104, which is not described in detail in this application). In another example, the network data analysis capability may include network data analysis of slice load, network data analysis of quality of service (QoS), etc. (specifically, reference may be made to the use case scenario of network data analysis defined in 3GPP TS23.288, which is not described in detail in this application). The decision-making capability may include one or more of the following: self-organization network (SON) optimization capability, radio access network intelligence (RAN intelligence) optimization capability (specifically, reference may be made to the use case scenario of RAN intelligence defined in 3GPP TR 37.340, which is not described in detail in this application), or centralized energy saving (centralized ES) optimization capability. Among them, self-organizing network optimization may also be referred to as self-optimization function optimization. In an example, the SON optimization capability may include coverage optimization, mobility optimization or energy saving optimization (for details, reference may be made to the use case scenario of SON optimization defined in 3GPP TS 28.313, which will not be described in detail in this application).The simulation capability of AI / ML refers to the ability to simulate AI / ML models in specific simulation environments, such as the simulation environment of twin networks and the simulation environment of simulated existing networks, and to perform reasoning of AI / ML models to evaluate the results of reasoning and the impact of the reasoning results on network performance.

[0138] Among them, the learning capability of AI / ML includes one or more of the following: federated learning, transfer learning, generative learning, or enhanced learning. Optionally, the learning capability of AI / ML can be included in the training capability of AI / ML, that is, the training capability of AI / ML can also include the learning capability of AI / ML, and it can also be understood that the learning capability of AI / ML can be a subset of the training capability of AI / ML. For example, AI / ML models can be trained through federated learning.

[0139] Among them, the trusted capabilities of AI / ML include one or more of the following: interpretability capability, robustness capability or fairness capability. Optionally, the trusted capabilities of AI / ML can be divided into the following four categories: trusted capabilities of training, trusted capabilities of testing, trusted capabilities of simulation, and trusted capabilities of reasoning. Among them, taking the trusted capabilities of training as an example, the trusted capabilities of training can include one or more of the following: interpretability capability of data, robustness capability of data, fairness capability of data, interpretability capability of training, robustness capability of training, or fairness capability of training. The trusted capabilities of testing can include one or more of the following: interpretability capability of data, robustness capability of data, fairness capability of data, interpretability capability of testing, robustness capability of testing, or fairness capability of testing. It should be noted that the data here refers to data used for model training, model testing, model reasoning, or model simulation. By analogy, the specific contents of the trusted capabilities of simulation and the trusted capabilities of reasoning are similar to those of the trusted capabilities of training, and will not be repeated here.

[0140] Among them, the processing capability of AI / ML may include: training processing capability of AI / ML, testing processing capability of AI / ML, reasoning processing capability of AI / ML or simulation processing capability of AI / ML. It should be noted that the training processing capability of AI / ML here refers to the processing capability of the module or entity where the training function of the AI / ML model is located, the testing processing capability of AI / ML refers to the processing capability of the module or entity where the testing function of the AI / ML model is located, the reasoning processing capability of AI / ML refers to the processing capability of the module or entity where the reasoning function of the AI / ML model is located, and the simulation processing capability of AI / ML refers to the processing capability of the module or entity where the simulation function of the AI / ML model is located. Among them, the processing capability may include one or more of the following: dedicated resources, latency sensitivity or execution time. Optionally, dedicated resources may include available memory and central processing unit (CPU). In one example, the training processing capability of AI / ML may include the dedicated resources available for training, for example, available memory for training.

[0141] Optionally, the first network management node may also create and configure an AI / ML capability (AIMLCapabilites) instance. For example, it may be an AI / ML capability information object class (AIMLcapabilites information object class, AIMLCapabilities IOC), and the specific object class attribute information is as shown in Table 1 and Table 2, as shown below:

[0142] Table 1

[0143]

[0144] Table 2

[0145]

[0146]

[0147] Step S602: The first network management node sends AI / ML capability information to the second network management node.

[0148] The second network management node is a consumer, for example, an AI / ML consumer, an AI / ML training consumer, an AI / ML reasoning consumer, an AI / ML testing consumer, an AI / ML simulation consumer, etc. The second network management node may be a CD-MnF, for example, an NMS.

[0149] In a possible implementation, before the first network management node sends the capability information of the AI / ML to the second network management node, the first network management node receives a request message from the second network management node, where the request message is used to request the capability information of the AI / ML. That is, the second network management node may send a capability query message to the first network management node to query the capability information of the AI / ML. Correspondingly, after receiving the capability query message, the first network management node sends a capability query result to the second network management node, where the capability query result includes the capability information of the AI / ML. It should be noted that the first network management node may send the AI / ML capability information to the second network management node after receiving a request message or a capability query message from the second network management node, or the first network management node may send the AI / ML capability information to the second network management node after receiving the control parameters of the AI / ML capability report of the second network management node. Optionally, the control parameter may be a reporting period, a reporting event (the event may be reported when a certain threshold is met, and the threshold may be a performance threshold of the existing network), or the first network management node may actively report, that is, the first network management node may send the AI / ML capability information to the second network management node without receiving a request message or a capability query message.

[0150] In another possible implementation, after the first network management node sends the capability information of the AI / ML to the second network management node, the first network management node may send the updated capability information of the AI / ML to the second network management node. That is, it can be understood that after the first network management node updates the capability information of the AI / ML, it sends the updated capability information of the AI / ML to the second network management node.

[0151] In another possible implementation, the first network management node receives first indication information from the second network management node, where the first indication information is used to indicate the target learning capability; the first network management node performs model training according to the first indication information to obtain a training result.

[0152] Among them, the target learning capability is one of the learning capabilities of AI / ML. The purpose of model training is to use a specific learning method through a given data set to allow the model to learn the characteristics and rules of the data, so that it can predict or classify unknown data. The training results may include one or more of the following: model identification, model performance, or model location. The learning method refers to the method used in model training. The process in which the first network management node receives the first indication information from the second network management node and performs model training according to the first indication information to obtain the training result can be understood as the second network management node specifies the target learning capability to complete a specific business scenario, and accordingly, the first network management node performs training according to the target learning capability.

[0153] In one example, when the business scenario of the second network management node is an energy-saving optimization scenario, the second network management node sends a first indication message to the first network management node, and the target learning capability indicated by the first indication message is transfer learning. After receiving the transfer learning indicated by the first indication message, the first network management node performs model training according to the transfer learning algorithm to obtain a training result. Optionally, the training result may include whether a specified learning method is used. For example, the training result includes the use of a specified learning method, that is, transfer learning. For example, the training result includes not using a specified learning method. When the training result includes not using a specified learning method, other learning methods used may be optionally reported. In another example, when the business scenario is a mobility optimization scenario, the second network management node sends a first indication message to the first network management node, and the target learning capability indicated by the first indication message is generative learning. After receiving the generative learning indicated by the first indication message, the first network management node performs model training according to the generative learning algorithm to obtain a training result. Optionally, the training result may include whether a specified learning method is used. For example, the training result includes the use of a specified learning method, i.e., generative learning. For example, the training result includes not using a specified learning method. When the training result includes not using a specified learning method, other learning methods used may be optionally reported.

[0154] In the above method, the second network management node can send the first indication information to the first network management node according to the obtained learning capability of AI / ML, indicating the target learning capability. In this way, the second network management node can select suitable learning methods for different business scenarios, that is, suitable learning capabilities of AI / ML training, so that the first network management node can be trained according to the target learning capability indicated by the first indication information, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML to the first network management node (for example, different business scenarios are suitable for different learning methods), so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the learning capability of AI / ML, to improve network performance.

[0155] In yet another possible implementation, the first network management node receives second indication information from the second network management node, and the first network management node performs a trusted evaluation of an indicator of the target trusted capability to obtain a trusted evaluation result.

[0156] Wherein, the second indication information is used to indicate the index of the target trusted capability, and the target trusted capability is one of the trusted capabilities of AI / ML. Accordingly, the first network management node performs a trusted assessment of the index of the target trusted capability to report the trusted assessment result. Wherein, when the target trusted capability indicated by the second indication information is one of the trusted capabilities of training, the first network management node obtains a trusted assessment result during the execution of model training or after the model training is completed, and the trusted assessment result includes the degree of trust of the training process or the degree of trust of the training result. When the target trusted capability indicated by the second indication information is one of the trusted capabilities of testing, the first network management node obtains a trusted assessment result during the execution of model testing or after the model testing is completed, and the trusted assessment result includes the degree of trust of the model testing process or the degree of trust of the test result. When the target trusted capability indicated by the second indication information is one of the trusted capabilities of simulation, the first network management node obtains a trusted assessment result during the execution of model simulation or after the model simulation is completed, and the trusted assessment result includes the degree of trust of the model simulation process or the degree of trust of the simulation result. When the target trusted capability indicated by the second indication information is one of the trusted capabilities of reasoning, the first network management node obtains a trusted assessment result during the execution of the model reasoning process or after the model reasoning is completed. The trusted assessment result includes the credibility of the model reasoning process or the credibility of the reasoning result.

[0157] Among them, optionally, the credibility of the training process or the credibility of the training results can be understood as the interpretability of the training process or the training results, the robustness of the training process or the training results, or the fairness of the training process or the training results. By analogy, the credibility of the test process or the credibility of the test results, the credibility of the simulation process or the credibility of the simulation results, or the credibility of the reasoning process or the credibility of the reasoning results can be similar to the credibility of the training results, which will not be repeated here. Optionally, the credible evaluation result can be embodied in the form of an indicator, and there is a corresponding relationship between the indicator and the interpretability capability, the robustness capability or the fairness capability. For example, the indicator corresponding to the interpretability capability can be a specific score value, the indicator corresponding to the robustness capability can be a numerical percentage, and the indicator corresponding to the fairness capability can be a grade, which is not limited in the embodiments of the present application.

[0158] In one example, the second network management node sends a second indication message to the first network management node, where the second indication message is used to indicate an indicator of a target trusted capability, where the indicator is a score value, and the target trusted capability is the trained interpretability capability, that is, the second indication message indicates the score of the trained interpretability capability. The first network management node receives the target trusted capability and obtains a trusted assessment result during or after the model training, where the trusted assessment result includes that the interpretability capability of the training result is 90 points (out of 100 points). Optionally, the first network management node sends the trusted assessment result to the second network management node.

[0159] In the above method, the second network management node can send a second indication message to the first network management node based on the acquired trusted capability of AI / ML, indicating the target trusted capability, so that the first network management node performs a trusted evaluation of the relevant indicators of the target trusted capability, and reports the trusted evaluation result, thereby improving network performance. That is, the second network management node indicates the capability requirements of AI / ML (for example, the requirements of the trusted capability of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the trusted capability of AI / ML, to improve network performance.

[0160] In another possible implementation, the first network management node receives third indication information from the second network management node, and the first network management node performs model reasoning according to the third indication information and the reasoning capability of AI / ML.

[0161] Among them, the third indication information is used to indicate the key performance indicator (KPI) target and / or performance measurement (PM) target that the model needs to achieve (specifically, refer to the relevant indicators defined in TS28.554 and TS28.552 in 3GPP, which will not be repeated in this application), or the target of a specific business scenario. Optionally, the KPI can be a category, such as coverage, capacity, voice, switching, etc. For example, coverage may include the ratio of coverage holes, the ratio of weak coverage, the ratio of over-coverage, the ratio of cross-area coverage, or the ratio of uplink and downlink imbalance; capacity includes user throughput, the number of users in the cell, etc.; voice includes QoS flow-related measurements; switching includes switching success rate, switching failure probability, switching too early ratio, or switching too late ratio, etc. Optionally, the target of a specific business scenario may include one or more of the following: energy saving target, load target or coverage optimization target. For example, the energy saving target may include the power consumption of the base station. The load target may include the utilization rate of the uplink and downlink physical resource blocks (physical resource blocks, PRBs), the number of radio resource connections (radio resource control, RRCs), and the coverage optimization target may include the satisfaction rate of the reference signal received power (reference signal received power, RSRP), the satisfaction rate of the reference signal received quality (reference signal receiving quality, RSRQ), the satisfaction rate of the signal to interference plus noise ratio (signal to interference plus noise ratio, SINR), switching performance, or the average throughput rate of the cell. Optionally, the first network management node performs model reasoning according to the third indication information and the reasoning capability of the AI / ML to obtain a reasoning result, and the reasoning result may include the reasoning capability of the AI / ML, and may also include the correspondence between the third indication information and the reasoning capability of the AI / ML. Among them, the reasoning capability of the AI / ML here may be the reasoning capability of the AI / ML selected by the first network management node according to the third indication information. For example, the third indication information indicates that the KPI that the model needs to achieve is the coverage threshold 1 and the switching threshold 2, and the first network management node determines that the selected AI / ML reasoning capability includes coverage optimization and mobility optimization according to the third indication information.

[0162] In one example, the third indication information is KPI1, KPI1 refers to the coverage threshold 1 reached, and the reasoning capabilities of AI / ML selected by the first network management node are coverage optimization and load optimization, wherein coverage optimization and load optimization belong to SON optimization in the reasoning capabilities of AI / ML. The first network management node performs model reasoning based on the third indication information KPI1 and the reasoning capabilities of AI / ML to obtain a reasoning result. Optionally, the reasoning result can be as shown in Table 3, where the reasoning result includes the correspondence between the third indication information and the reasoning capabilities of AI / ML, that is, the reasoning result includes KPI1, coverage optimization and load optimization.

[0163] Table 3

[0164]

[0165] In the above method, by the first network management node receiving the third indication information from the second network management node, the second network management node can make a corresponding request to the first network management node according to the demand, that is, the key performance indicator KPI target or performance measurement target that the model needs to achieve, so that the first network management node performs model reasoning according to the demand of the second network management node, that is, the third indication information, and the determined AI / ML reasoning capability, thereby improving network performance. That is, the second network management node indicates the AI / ML capability requirements (for example, the AI / ML reasoning capability requirements) to the first network management node, so that the first network management node executes the corresponding AI / ML capabilities according to the AI / ML capability requirements, for example, the AI / ML reasoning capability, to improve network performance.

[0166] In another possible implementation, the first network management node receives fourth indication information from the second network management node, and the first network management node deploys the model according to the fourth indication information and the processing capability of AI / ML.

[0167] Among them, the fourth indication information includes the correspondence between the business scenario identifier and the threshold, that is, the fourth indication information includes the business scenario identifier and the threshold, and the threshold includes one or more of the following: delay sensitivity threshold, execution time threshold or threshold of available dedicated resources. Optionally, the fourth indication information can also directly indicate the correspondence between the business scenario and the threshold, that is, the fourth indication information includes the business scenario and the threshold, for example, the business scenario can be indicated in a string format. For details, please refer to Table 4.

[0168] Table 4

[0169] Business scenario identification Business scenario description Delay Sensitivity Execution time Dedicated resources available Logo 1 Energy-saving optimization scenarios Threshold 1 Threshold 2 Threshold 3 Logo 2 Coverage optimization scenario Threshold 4 Threshold 5 Threshold 6

[0170] Optionally, the threshold of the same indicator in different business scenarios may be different. In one example, for example, when the business scenario is an energy-saving optimization scenario, the threshold of delay sensitivity is threshold 1, and when the business scenario is a coverage optimization scenario, the threshold of delay sensitivity is threshold 4, where threshold 1 and threshold 4 may be different.

[0171] Among them, after the first network management node deploys the model according to the fourth indication information and the processing capability of AI / ML, the first network management node can send deployment result information to the second network management node.

[0172] The deployment result information includes deployment success or deployment failure. When the deployment is successful, the first network management node can determine the energy consumption information and resource usage information of the model, and optionally, can send the energy consumption information and resource usage information of the model to the second network management node. When the deployment fails, the first network management node determines the reason for the deployment failure, and optionally, the reason for the deployment failure can be insufficient computing energy, insufficient energy storage, energy consumption problems, etc. Optionally, the first network management node sends the reason for the deployment failure to the second network management node.

[0173] In one example, the fourth indication information is a business scenario identifier as identifier 1, and its demand for training processing capabilities is: delay sensitivity is threshold 1, execution time is threshold 2, and available dedicated resources are threshold 3. The processing capability of AI / ML includes the processing capability of training, and the processing capability of training includes: delay sensitivity is threshold 7, execution time is threshold 8, and the threshold of dedicated resources for training is threshold 9, wherein threshold 1 is greater than threshold 7, threshold 2 is greater than threshold 8, and threshold 3 is less than threshold 9, that is, the processing capability of AI / ML training can meet the needs of the business scenario, that is, the business scenario corresponding to business scenario identifier 1 has the demand for training processing capabilities. The first network management node deploys the model according to the fourth indication information and the processing capability of AI / ML, and sends deployment result information to the second network management node, and the deployment result information includes deployment success.

[0174] In the above method, the first network management node receives the fourth indication information from the second network management node. For example, the second network management node may be an operator, so that the operator can specify the fourth indication information, that is, specify the requirements of the model in a specific scenario, so that the first network management node can deploy the model according to the requirements of the model in the specific scenario. For example, for scenarios with high latency sensitivity such as video games, it is necessary to deploy a model with high responsiveness. For example, the requirements for the processing capabilities of the reasoning entities applicable to different models (latency sensitivity, available resources) and energy consumption requirements, the first network management node comprehensively selects the appropriate model deployment based on the processing capabilities and requirements, thereby improving network performance and user experience. That is, the second network management node indicates the capability requirements of AI / ML (for example, the processing capability requirements of AI / ML) to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML, for example, the processing capability of AI / ML, to improve network performance.

[0175] Optionally, before the second network management node sends the first indication information, the second indication information, the third indication information, or the fourth indication information to the first network management node, the second network management node may define an AI / ML capability control (AIMLCapabilitesControl) object instance, such as an AI / ML capability control object class (AIMLcapabilitescontrolinformation object class, AIMLCapabilitesControl IOC), wherein AIMLCapabilitesControl may include one or more of the following: the first indication information, the second indication information, the third indication information, or the fourth indication information. Accordingly, after the first network management node receives the first indication information, the second indication information, the third indication information, or the fourth indication information, it may create an AIMLCapabilitesControl object instance according to the first indication information, the second indication information, the third indication information, or the fourth indication information, and configure the first indication information, the second indication information, the third indication information, or the fourth indication information to the AIMLCapabilitesControl object instance.

[0176] Step S603: The second network management node receives the AI / ML capability information from the first network management node.

[0177] Among them, the second network management node has a readable permission on the capability information of AI / ML in the capability information object class of AI / ML, but cannot write it, that is, the second network management node cannot modify the capability information of AI / ML.

[0178] In the above method, the capability information of AI / ML and the method of obtaining the AI / ML capability information are defined through the above method, so that the second network management node indicates the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance.

[0179] Optional, see Figure 7 , Figure 7 This is a flow chart of another management method provided by the embodiment of the present application, which is as follows:

[0180] Step S701: the second network management node sends a request message to the first network management node.

[0181] The request message is used to request the capability information of the AI / ML. The capability information of the AI / ML can refer to the relevant description in step S601, which will not be repeated here.

[0182] Step S702: The first network management node receives a request message from the second network management node.

[0183] Before the first network management node receives the request message from the second network management node, the first network management node may create and configure an AI / ML capability (AIMLCapabilites) instance. For example, it may be an AI / ML capability information object class (AIMLcapabilites information object class, AIMLCapabilities IOC), and the details may refer to the above description.

[0184] Step S703: The first network management node sends AI / ML capability information to the second network management node.

[0185] For details, please refer to step S602, which will not be repeated here.

[0186] Step S704: The second network management node receives the AI / ML capability information from the first network management node.

[0187] Step S705: the second network management node sends the first indication information, the second indication information, the third indication information or the fourth indication information to the first network management node.

[0188] It should be noted that this step is optional.

[0189] Among them, the relevant explanation of the first indication information, the second indication information, the third indication information or the fourth indication information can refer to the relevant description in step S602. Before the second network management node sends the first indication information, the second indication information, the third indication information or the fourth indication information to the first network management node, the second network management node can define an AI / ML capability control (AIMLCapabilitesControl) object instance, and the specific explanation can refer to the relevant description in step S602.

[0190] Step S706: the first network management node receives the first indication information, the second indication information, the third indication information or the fourth indication information from the second network management node.

[0191] It should be noted that this step is optional.

[0192] Step S707: the first network management node performs relevant processing according to the first indication information, the second indication information, the third indication information or the fourth indication information.

[0193] It should be noted that this step is an optional step. For example, the first network management node may perform model training according to the first indication information to obtain a training result, and the details may refer to the relevant description in step S602.

[0194] For example, the first network management node performs a trustworthy evaluation of the indicator of the target trustworthy capability to obtain a trustworthy evaluation result, and details may refer to the relevant description in step S602.

[0195] For example, the first network management node performs model reasoning according to the third indication information and the reasoning capability of AI / ML. For details, please refer to the relevant description in step S602.

[0196] For example, the first network management node deploys the model according to the fourth indication information and the processing capability of AI / ML. For details, please refer to the relevant description in step S602.

[0197] In the above method, the capability information of AI / ML and the method for obtaining the capability information of AI / ML are defined in the above manner, so that the second network management node can indicate the capability requirements of AI / ML to the first network management node, so that the first network management node executes the corresponding capabilities of AI / ML according to the capability requirements of AI / ML to improve network performance.

[0198] The method of the embodiment of the present application is described in detail above, and the device of the embodiment of the present application is provided below.

[0199] See also Figure 8 , Figure 8800 is a schematic diagram of the structure of a management device 800 provided in an embodiment of the present application. The management device 800 may be a first network management node or a chip in the first network management node. The management device 800 may include a processing unit 801 and a transceiver unit 802. The specific units are as follows:

[0200] The processing unit 801 is used to determine the capability information of artificial intelligence AI / machine learning ML, where the capability information of AI / ML includes one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML; the transceiver unit 802 is used to send the capability information of AI / ML to the second network management node.

[0201] In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0202] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0203] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0204] In yet another possible implementation, the transceiver unit 802 is configured to receive a request message from the second network management node, where the request message is used to request capability information of the AI / ML.

[0205] In yet another possible implementation, the transceiver unit 802 is configured to send updated AI / ML capability information to the second network management node.

[0206] In another possible implementation, the transceiver unit 802 is used to receive first indication information from the second network management node, where the first indication information is used to indicate a target learning capability, and the target learning capability is one of the learning capabilities of the AI / ML; the processing unit 801 is used to perform model training according to the first indication information to obtain a training result.

[0207] In another possible implementation, the transceiver unit 802 is used to receive second indication information from the second network management node, where the second indication information is used to indicate an indicator of a target trusted capability, and the target trusted capability is one of the trusted capabilities of the AI / ML; the processing unit 801 is used to perform a trusted assessment of the indicator of the target trusted capability to obtain a trusted assessment result.

[0208] In another possible implementation, the transceiver unit 802 is used to receive third indication information from the second network management node, where the third indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve; and the processing unit 801 is used to perform model reasoning based on the third indication information and the reasoning capability of the AI / ML.

[0209] In another possible implementation, the transceiver unit 802 is used to receive fourth indication information from the second network management node, the fourth indication information including a correspondence between a business scenario identifier and a threshold, the threshold including one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the processing unit 801 is used to deploy a model according to the fourth indication information and the processing capability of the AI / ML.

[0210] In yet another possible implementation, the transceiver unit 802 is configured to send deployment result information to the second network management node, where the deployment result information includes deployment success or deployment failure.

[0211] Optionally, the management device 800 may be a second network management node or a chip in the second network management node, and each unit is specifically as follows:

[0212] The transceiver unit 802 is used to receive artificial intelligence AI / machine learning ML capability information from the first network management node, where the AI / ML capability information includes one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML.

[0213] In one possible implementation, the general capabilities of the AI / ML include one or more of the following: AI / ML training capability, AI / ML testing capability, AI / ML reasoning capability, or AI / ML simulation capability; the learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; the trusted capabilities of the AI / ML include one or more of the following: explainability capability, robustness capability, or fairness capability; the processing capabilities of the AI / ML include one or more of the following: AI / ML training processing capability, AI / ML testing processing capability, AI / ML reasoning processing capability, or AI / ML simulation processing capability.

[0214] In another possible implementation, the training capability of the AI / ML includes a joint training capability, wherein the joint training capability includes the capability of joint training between models; the testing capability of the AI / ML includes a joint testing capability, wherein the joint testing capability includes the capability of joint testing between models; the reasoning capability of the AI / ML includes an analysis capability and a decision-making capability, wherein the analysis capability includes a management data analysis capability and / or a network data analysis capability, and the decision-making capability includes a self-optimization function SON optimization capability and / or a radio access network intelligence RANintellgence optimization capability; the simulation capability of the AI / ML includes support for a simulation environment.

[0215] In another possible implementation, the explainability capability includes one or more of the following: the explainability capability of training, the explainability capability of testing, the explainability capability of reasoning, or the explainability capability of simulation; the robustness capability includes one or more of the following: the robustness capability of training, the robustness capability of testing, the robustness capability of reasoning, or the robustness capability of simulation; the fairness capability includes one or more of the following: the fairness capability of training, the fairness capability of testing, the fairness capability of reasoning, or the fairness capability of simulation.

[0216] In yet another possible implementation, the transceiver unit 802 is configured to send a request message to the first network management node, where the request message is used to request capability information of the AI / ML.

[0217] In yet another possible implementation, the transceiver unit 802 is configured to receive updated AI / ML capability information from the first network management node.

[0218] In another possible implementation, the transceiver unit 802 is used to send first indication information to the first network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML, and the first indication information is used for model training.

[0219] In another possible implementation, the transceiver unit 802 is used to send second indication information to the first network management node, where the second indication information is used to indicate an indicator of a target trusted capability, where the target trusted capability is one of the trusted capabilities of the AI / ML, and the target trusted capability is used for trusted assessment.

[0220] In another possible implementation, the transceiver unit 802 is used to send third indication information to the first network management node, where the second indication information is used to indicate a key performance indicator KPI target or performance measurement target that the model needs to achieve, and the second indication information is used for model reasoning.

[0221] In another possible implementation, the transceiver unit 802 is used to send fourth indication information to the first network management node, and the fourth indication information includes a correspondence between a business scenario identifier and a threshold, and the threshold includes one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the fourth indication information is used for model deployment.

[0222] In yet another possible implementation, the transceiver unit 802 is configured to receive deployment result information from the first network management node, where the deployment result information includes deployment success or deployment failure.

[0223] It should be noted that the implementation and beneficial effects of each unit can also refer to Figure 6 or Figure 7 The corresponding description of the method embodiment shown.

[0224] See also Fig. 9 , Fig. 9 9 is a structural diagram of another management device 900 provided in an embodiment of the present application, the management device 900 includes at least one processor 901 and a communication interface 903, and optionally, also includes a memory 902, the processor 901, the memory 902 and the communication interface 903 are interconnected via a bus 904. Optionally, the processor 901 and the memory 902 may be integrated together.

[0225] The memory 902 includes, but is not limited to, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), or a portable read-only memory (CD-ROM), and the memory 902 is used for related computer programs and data. The communication interface 903 is used to receive and send data.

[0226] The processor 901 may be one or more central processing units (CPUs). When the processor 901 is a CPU, the CPU may be a single-core CPU or a multi-core CPU.

[0227] The processor 901 in the management device 900 is used to read the computer program or instructions stored in the memory 902 to implement the functions of the above-mentioned processing unit, and the communication interface 903 in the management device 900 is used to implement the functions of the above-mentioned transceiver unit.

[0228] The embodiment of the present application also provides a chip device, the chip device includes at least one processor, the at least one processor is used to call a computer program or instruction stored in a memory, so that the processor executes the above Figure 6 or Figure 7 The illustrated embodiments provide methods.

[0229] In a possible implementation, the input of the chip device corresponds to the above Figure 6 or Figure 7 In the receiving operation of any one of the embodiments shown in the embodiment, the output of the chip device corresponds to the above Figure 6 or Figure 7 The sending operation in any of the embodiments shown.

[0230] Optionally, the processor is coupled to the memory via an interface.

[0231] Optionally, the chip device further comprises a memory in which computer program instructions are stored.

[0232] An embodiment of the present application also provides a computer-readable storage medium, in which a computer program or instruction is stored. When the computer program or instruction is executed on a processor, the method executed by the first network management node or the second network management node in the above method embodiment is implemented.

[0233] An embodiment of the present application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed on a processor, the method performed by the first network management node or the second network management node in the above method embodiment is implemented.

[0234] The embodiment of the present application also provides a communication system, which includes the first network management node in the above embodiment and the second network management node in the above embodiment. The first network management node is used to execute some or all operations executed by the first network management node in the above method embodiment, and the second network management node is used to execute some or all operations of the second network management node in the above method embodiment.

[0235] It is understandable that the processor in the embodiments of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0236] The method steps in the embodiments of the present application can be implemented by hardware, or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in a random access memory, a flash memory, a read-only memory, a programmable read-only memory, an erasable programmable read-only memory, an electrically erasable programmable read-only memory, a register, a hard disk, a mobile hard disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC. In addition, the ASIC can be located in a base station or a terminal. Of course, the processor and the storage medium can also be present in a base station or a terminal as discrete components.

[0237] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented by software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instruction is loaded and executed on a computer, the process or function described in the embodiment of the present application is executed in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device or other programmable device. The computer program or instruction may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer program or instruction may be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server, data center, etc. that integrates one or more available media. The available medium may be a magnetic medium, for example, a floppy disk, a hard disk, a tape; it may also be an optical medium, for example, a digital video disc; it may also be a semiconductor medium, for example, a solid-state hard disk. The computer-readable storage medium may be a volatile or nonvolatile storage medium, or may include both volatile and nonvolatile types of storage media.

[0238] In the various embodiments of the present application, unless otherwise specified or provided for in any logical conflict, the terms and / or descriptions between the different embodiments are consistent and may be referenced to each other, and the technical features in the different embodiments may be combined to form new embodiments according to their inherent logical relationships.

[0239] In the description of this application, words such as "first", "second", "S601", or "S602" are only used to distinguish the description and facilitate the context. Different sequence numbers themselves do not have specific technical meanings and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying the order of execution of operations. The execution order of each process should be determined by its function and internal logic.

[0240] The term "and / or" in this application is only a description of the association relationship of the associated objects, indicating that there can be three kinds of relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article indicates that the associated objects before and after are in an "or" relationship.

[0241] In this application, "transmission" may include the following three situations: sending of data, receiving of data, or sending of data and receiving of data. In this application, "data" may include service data and / or signaling data.

[0242] In this application, the terms "comprises" or "has" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process / method comprising a series of steps, or a system / product / apparatus comprising a series of units, is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes / methods / products / apparatus.

[0243] In the description of this application, the number of nouns, unless otherwise specified, means "singular noun or plural noun", that is, "one or more". "At least one" means one or more. "Including at least one of the following: A, B, C." means that it may include A, or include B, or include C, or include A and B, or include A and C, or include B and C, or include A, B and C. A, B, and C can be single or plural.

Claims

1. A management method, characterized in that: include: The first network management node determines capability information of the artificial intelligence AI / machine learning ML, where the capability information of the AI / ML includes one or more of the following: general capability of the AI / ML, learning capability of the AI / ML, trust capability of the AI / ML, or processing capability of the AI / ML; The first network management node sends the AI / ML capability information to the second network management node.

2. The method according to claim 1, characterized in that The general capabilities of AI / ML include one or more of the following: training capabilities of AI / ML, testing capabilities of AI / ML, reasoning capabilities of AI / ML, or simulation capabilities of AI / ML; The learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; The trustworthy capabilities of the AI / ML include one or more of the following: explainability, robustness, or fairness; The processing capability of the AI / ML includes one or more of the following: the training processing capability of the AI / ML, the testing processing capability of the AI / ML, the reasoning processing capability of the AI / ML, or the simulation processing capability of the AI / ML.

3. The method according to claim 2, characterized in that The training capability of the AI / ML includes a joint training capability, and the joint training capability includes a capability of joint training between models; The AI / ML testing capability includes a joint testing capability, and the joint testing capability includes a capability for joint testing between models; The reasoning capability of the AI / ML includes analysis capability and decision-making capability. The analysis capability includes management data analysis capability and / or network data analysis capability. The decision-making capability includes self-optimization function SON optimization capability and / or radio access network intelligence RANintellgence optimization capability. The simulation capability of the AI / ML includes support for a simulation environment.

4. The method according to any one of claims 1 to 3, characterized in that: The method further comprises: The first network management node receives a request message from the second network management node, where the request message is used to request capability information of the AI / ML.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: The first network management node sends updated AI / ML capability information to the second network management node.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: The first network management node receives first indication information from the second network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML; The first network management node performs model training according to the first indication information to obtain a training result.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: The first network management node receives second indication information from the second network management node, where the second indication information is used to indicate an indicator of a target trusted capability, where the target trusted capability is one of the trusted capabilities of the AI / ML; The first network management node performs a trustworthy evaluation of an indicator of the target trustworthy capability to obtain a trustworthy evaluation result.

8. The method according to any one of claims 1 to 7, characterized in that: The method further comprises: The first network management node receives third indication information from the second network management node, where the third indication information is used to indicate a key performance indicator KPI target or a performance measurement target that the model needs to achieve; The first network management node performs model reasoning according to the third indication information and the reasoning capability of the AI / ML.

9. The method according to any one of claims 1 to 8, characterized in that: The method further comprises: The first network management node receives fourth indication information from the second network management node, where the fourth indication information includes a correspondence between a service scenario identifier and a threshold, where the threshold includes one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; The first network management node deploys a model according to the fourth indication information and the processing capability of the AI / ML.

10. A management method, characterized in that: include: The second network management node receives artificial intelligence AI / machine learning ML capability information from the first network management node, where the AI / ML capability information includes one or more of the following: general capabilities of AI / ML, learning capabilities of AI / ML, trusted capabilities of AI / ML, or processing capabilities of AI / ML.

11. The method according to claim 10, characterized in that The general capabilities of AI / ML include one or more of the following: training capabilities of AI / ML, testing capabilities of AI / ML, reasoning capabilities of AI / ML, or simulation capabilities of AI / ML; The learning capabilities of the AI / ML include one or more of the following: federated learning, transfer learning, generative learning, or reinforcement learning; The trustworthy capabilities of the AI / ML include one or more of the following: explainability, robustness, or fairness; The processing capability of the AI / ML includes one or more of the following: the training processing capability of the AI / ML, the testing processing capability of the AI / ML, the reasoning processing capability of the AI / ML, or the simulation processing capability of the AI / ML.

12. The method according to claim 10 or 11, characterized in that: The training capability of the AI / ML includes a joint training capability, and the joint training capability includes a capability of joint training between models; The AI / ML testing capability includes a joint testing capability, and the joint testing capability includes a capability for joint testing between models; The reasoning capability of the AI / ML includes analysis capability and decision-making capability. The analysis capability includes management data analysis capability and / or network data analysis capability. The decision-making capability includes self-optimization function SON optimization capability and / or radio access network intelligence RANintellgence optimization capability. The simulation capability of the AI / ML includes support for a simulation environment.

13. The method according to any one of claims 10 to 12, characterized in that: The method further comprises: The second network management node sends a request message to the first network management node, where the request message is used to request capability information of the AI / ML.

14. The method according to any one of claims 10 to 13, characterized in that: The method further comprises: The second network management node receives updated AI / ML capability information from the first network management node.

15. The method according to any one of claims 10 to 14, characterized in that: The method further comprises: The second network management node sends first indication information to the first network management node, where the first indication information is used to indicate a target learning capability, where the target learning capability is one of the learning capabilities of the AI / ML, and the first indication information is used for model training.

16. The method according to any one of claims 10 to 15, characterized in that: The method further comprises: The second network management node sends second indication information to the first network management node, where the second indication information is used to indicate an indicator of a target trusted capability, where the target trusted capability is one of the trusted capabilities of the AI / ML, and the target trusted capability is used for trusted assessment.

17. The method according to any one of claims 10 to 16, characterized in that: The method further comprises: The second network management node sends third indication information to the first network management node, where the third indication information is used to indicate a key performance indicator KPI target or a performance measurement target that the model needs to achieve, and the second indication information is used for model reasoning.

18. The method according to any one of claims 10 to 17, characterized in that: The method further comprises: The second network management node sends fourth indication information to the first network management node, and the fourth indication information includes the correspondence between the business scenario identifier and the threshold, and the threshold includes one or more of the following: a delay sensitivity threshold, an execution time threshold, or a threshold of available dedicated resources; the fourth indication information is used for model deployment.

19. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program or instruction, which, when executed on a processor, implements the method according to any one of claims 1 to 9.

20. A computer program product, characterized in that The computer program product includes a computer program or instructions, and when the computer program or instructions are executed on a computer, the method according to any one of claims 1 to 9 is implemented.

Images