Model registration method and communication apparatus

Abstract

A model registration method and a communication apparatus, which relate to the field of communications, and enable a node, which does not have an AI / ML model generated thereon, to implement model selection and application. In the method, a first device may learn about, by means of a model registration request, information of models generated by means of training within the management ranges of different second devices, model registration is performed, and then a model registration response is fed back to the first device, so as to implement model storage and management, such that the first device or a third-party device can find, on the basis of an inference requirement and from information of registered models, a model that meets the requirements, and thus the efficiency and real-time performance of model application can be improved.

Description

Model registration method and communication apparatus

[0001] The present application claims priority to the Chinese patent application No. 202410892891.X, filed on July 3, 2024, and entitled "Model registration method and communication apparatus", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of communication, in particular to a model registration method and a communication apparatus. BACKGROUND

[0003] Artificial intelligence (AI) / machine learning (ML) technology is increasingly being applied in various fields of wireless networks. The lifecycle management (LCM) of AI / ML models in wireless networks mainly includes training, testing, simulation, deployment and inference. In the current LCM management process of AI / ML models, after the ML model is trained and generated at a certain node (such as a single-domain management function unit in a service-oriented management architecture), the model-related information of the ML model is stored in the node, and the ML model is directly deployed to the inference function by the node. Other nodes (such as cross-domain management function units and third-party management devices in the service-oriented management architecture) cannot timely learn the ML model that meets their needs and deploy it to the inference function to perform specific inference.

[0004] Therefore, how to enable a node that does not generate an ML model to find a suitable ML model for deployment to perform specific inference has become a problem to be solved. SUMMARY

[0005] The present application provides a model registration method and a communication apparatus, which can enable a node that does not generate an ML model to select and apply a model.

[0006] To achieve the above object, the present application adopts the following technical solutions:

[0007] In a first aspect, a model registration method is provided. The method can be executed by a first device, or by a component of the first device, such as a processor, a chip, or a chip system of the first device, or by a logic module or software capable of implementing all or part of the first device. The method comprises: receiving, by the first device, a model registration request, and sending a model registration response. The model registration request is used to request registration of a first model, and the model registration response is used to indicate whether the registration of the first model is successful.

[0008] In the method, the first device can learn the information of the model generated by training in the management range of different second devices through the model registration request, perform model registration, and feed back the model registration response to the first device, so as to realize storage and management of the model information, so that the first device or a third party device can find a model meeting the demand from the registered model information according to inference demand, and the efficiency and real-time performance of model application can be improved. Thus, the model registration can record and track the trained model, so as to realize future retrieval, repeatability, sharing and loading to the target inference function in different environments.

[0009] In a possible design scheme, the model registration request can include the information of the first model, and the information of the first model can include at least one of the following: a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model. Thus, the first device can learn the information of the model provided by different second devices according to the model information in the model registration request, so that the AI / ML consumer can use the most suitable model to perform the target inference function.

[0010] In a possible design scheme, the inference type of the first model can include at least one of the following: a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type. Thus, based on the inference type of the model, the first device or the third party device can find a model meeting the inference demand from the registered model information of the first device, and the application rate of the model can be improved.

[0011] In a possible design scheme, the capability of the first model can include at least one of the following: a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0012] In a possible design scheme, the intention target of the first model can include at least one of the following: a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

[0013] In a possible design scheme, in a case where the first model is successfully registered, the model registration response can include a second identifier, which is a globally unique identifier allocated by the first device to the first model. Thus, the first device can allocate a globally unique identifier as the second identifier to the first model successfully registered, so that the first model can be uniquely identified in the management range of the first device.

[0014] In a possible design, the method in the first aspect can further include: the first device sending a model deployment request, the model deployment request including the second identifier, and the model deployment request being used to request to deploy the first model. In this way, when the first device is an AI / ML consumer, the first device can select a model that meets the inference requirement of the first device from the information of the registered models, and request to deploy the model on the inference function of the corresponding device to implement target inference.

[0015] In a possible design, the method in the first aspect can be applied to a service management architecture, and the first device is a cross-domain management function unit. The method includes: the first device receiving a model registration request; and the first device sending a model registration response. The receiving the model registration request includes: the first device receiving the model registration request from a second device. The sending the model registration response includes: the first device sending the model registration response to the second device. The second device is a single-domain management function unit. In this way, the model registration method can be applied to the registration interaction between the cross-domain management function unit and the single-domain management function unit in the service management architecture, and the cross-domain management function unit can store and manage the model information trained in the single-domain management function unit according to the model registration request of the single-domain management function unit, so that an operator or a third party can obtain the information of the target model to perform a specific operation.

[0016] In a possible design, the method in the first aspect can be applied to an open management architecture, and the first device is a management and orchestration platform. The method includes: the first device receiving a model registration request; and the first device sending a model registration response. The receiving the model registration request includes: the first device receiving the model registration request from a second device. The sending the model registration response includes: the first device sending the model registration response to the second device. The second device is a service operation application APP. In this way, the model registration method can also be applied to the registration interaction between the management and orchestration platform and the service operation application in the open management architecture, and the management and orchestration platform can store and manage the model information trained by the service operation application according to the model registration request of the service operation application, so that the management and orchestration platform as a management model requester can obtain the model information to find a suitable service operation application as a management model provider to perform target inference, and the efficiency and real-time performance of the model application are improved.

[0017] In a possible design, the first identifier of the first model in the information of the first model included in the model registration request can be a vendor identifier, and the inference type of the first model in the information of the first model can be the type of a service operation APP. In this way, in the open management architecture, the first model can be identified by the vendor identifier, and the inference type of the first model can be indicated by the type of the service operation APP.

[0018] In a possible design, the method of the first aspect can further include: receiving, by the first device, first indication information from the second device, the first indication information being used to indicate that the first model is allowed to be retrained. Retraining, by the first device, the first model according to a performance of the first model and the first indication information. Sending, by the first device, updated model information to the second device, the updated model information being used to update the first model. In this way, in the open management architecture, since the management and orchestration platform has the training function, the first device can also determine whether to retrain the model according to the indication information sent by the second device about whether the model is allowed to be retrained, in combination with the performance of the model, to improve the robustness of the model application.

[0019] In a possible design, the updated model information can include a storage address of the updated first model. That is, after the first device re-trains the model, the first device can send the storage address of the storage of the updated model to the second device, so that the second device can download the updated model according to the storage address to update the model.

[0020] In a possible design, the method of the first aspect can further include: receiving, by the first device, a model discovery request, the model discovery request being used to request to discover a model. Sending, by the first device, a model discovery response according to the information of the registered model and the model discovery request, the model discovery response including information of the model. In this way, the first device can also provide the model discovery service according to the information of the registered model, and discover the information of the corresponding model for the AI / ML consumer according to the model discovery request of the AI / ML consumer.

[0021] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0022] In a second aspect, a model registration method is provided. The method can be executed by the second device, or by a component of the second device, for example, a processor, a chip, or a chip system of the second device, or by a logic module or software that can implement all or part of the second device. The method includes: sending, by the second device, a model registration request, the model registration request being used to request to register a first model. Receiving, by the second device, a model registration response, the model registration response being used to indicate whether the first model is successfully registered.

[0023] In a possible design, the model registration request can include information of the first model, and the information of the first model can include at least one of the following: a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intended target of the first model.

[0024] In a possible design, the inference type of the first model can include at least one of the following: a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

[0025] In a possible design, the capability of the first model can include at least one of the following: a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0026] In a possible design, the intention target of the first model can include at least one of the following: a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

[0027] In a possible design, in a case where the first model is successfully registered, the model registration response can include a second identifier, which is a globally unique identifier allocated to the first model by the first device.

[0028] In a possible design, the method in the second aspect can further include: receiving, by the second device, a model deployment request, the model deployment request including the second identifier, and the model deployment request being used to request to deploy the first model.

[0029] In a possible design, the method in the second aspect is applied to a service-based management architecture, and the second device is a single-domain management function unit. The second device sending the model registration request includes: the second device sending the model registration request to the first device. The second device receiving the model registration response includes: the second device receiving the model registration response from the first device. The first device is a cross-domain management function unit.

[0030] In a possible design, the method in the second aspect is applied to an open management architecture, and the second device is a service operation application (APP). The second device sending the model registration request includes: the second device sending the model registration request to the first device. The second device receiving the model registration response includes: the second device receiving the model registration response from the first device. The first device is a management and orchestration platform.

[0031] In a possible design, the first identifier of the first model in the information of the first model included in the model registration request can be a vendor identifier, and the inference type of the first model in the information of the first model can be a type of a service operation APP.

[0032] In a possible design, the method in the second aspect can further include: sending, by the second device, first indication information to the first device, the first indication information being used to indicate that retraining of the first model is allowed. Receiving, by the second device, updated model information from the first device, the updated model information being used to update the first model.

[0033] In a possible design, the updated model information can include a storage address of the updated first model.

[0034] In a possible design, the method of the second aspect can further include that the second device sends a model discovery request, where the model discovery request is used to request discovery of a model. The second device receives a model discovery response, where the model discovery response includes information of the model.

[0035] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0036] The technical effects of the method of the second aspect can refer to the related description of the method of the first aspect, which will not be repeated here.

[0037] In a third aspect, a model registration method is provided. The method can be executed by a first device, a component of the first device, such as a processor, a chip, or a chip system of the first device, or a logic module or software that can implement all or part of the first device. The method includes: allocating a globally unique identifier for a first model by the first device after the first model is successfully registered, where the first model is obtained by offline training, and the first device is a cross-domain management function unit. The first device sends the globally unique identifier to a second device, where the second device is a single-domain management function unit.

[0038] In the method, the first device can register a model obtained by offline training, and allocate a globally unique identifier for the registered offline model, so as to send the globally unique identifier to a second device that can deploy the model, to implement registration and management of the offline model, and to realize traceability and query functions of the model, to track and recall the most suitable model version.

[0039] In a possible design, the method of the third aspect can further include: obtaining information of the first model, where the information of the first model can include at least one of the following: a third identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

[0040] In a possible design, the inference type of the first model can include at least one of the following: a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

[0041] In a possible design, the capability of the first model can include at least one of the following: a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0042] In a possible design, the intention target of the first model can include at least one of the following: a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

[0043] In a possible design, the globally unique identifier can be carried in a model deployment request.

[0044] In a possible design, the method in the third aspect can further include: receiving, by the first device, a model discovery request, the model discovery request being used to request to discover a model; and sending, by the first device, a model discovery response according to the information of the registered model and the model discovery request, the model discovery response including information of the model.

[0045] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0046] The technical effect of the method in the third aspect can refer to the related description of the method in the first aspect, which will not be repeated here.

[0047] In a fourth aspect, a model registration method is provided. The method can be executed by a second device, or by a component of the second device, for example, a processor, a chip, or a chip system of the second device, or by a logic module or software that can implement all or part of the second device. The method includes: receiving, by the second device, a globally unique identifier of a first model from a first device, the first device being a cross-domain management function unit, and the second device being a single-domain management function unit; and determining, by the second device, the first model according to the globally unique identifier.

[0048] In a possible design, the globally unique identifier can be carried in a model deployment request.

[0049] In a possible design, the method in the fourth aspect can further include: sending, by the second device, a model discovery request to the first device, the model discovery request being used to request to discover a model; and receiving, by the second device, a model discovery response from the first device, the model discovery response including information of the model.

[0050] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0051] The technical effect of the method in the fourth aspect can refer to the related description of the method in the third aspect, which will not be repeated here.

[0052] In a fifth aspect, a communication apparatus is provided for implementing the various methods described above. The communication apparatus can be the first device in the first aspect, or an apparatus including the first device, or an apparatus included in the first device, such as a chip. The communication apparatus includes corresponding modules, units, or means for implementing the methods of the first aspect, which can be implemented by hardware, software, or by a combination of hardware and software. The hardware or software includes one or more modules or units corresponding to the functions described above.

[0053] In some possible designs, the communication apparatus includes a processing module and a communication module. The processing module is configured to control the communication module to receive a model registration request and send a model registration response. The model registration request is used to request registration of a first model, and the model registration response is used to indicate whether the first model is successfully registered.

[0054] In one possible design, the model registration request can include information of the first model, and the information of the first model can include at least one of a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intent target of the first model.

[0055] In one possible design, the inference type of the first model can include at least one of a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

[0056] In one possible design, the capability of the first model can include at least one of a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0057] In one possible design, the intent target of the first model can include at least one of a coverage intent target, an energy saving intent target, or a wireless resource optimization intent target.

[0058] In one possible design, in a case where the first model is successfully registered, the model registration response can include a second identifier, which is a globally unique identifier allocated by the first device for the first model.

[0059] In one possible design, the communication module is configured to send a model deployment request, and the model deployment request includes the second identifier. The model deployment request is used to request deployment of the first model.

[0060] In a possible design, the communication device is a cross-domain management function unit applied to a service management architecture. The processing module is configured to control the communication model to receive a model registration request, including: the processing module is configured to control the communication model to receive a model registration request from a second device; and the processing module is further configured to control the communication model to send a model registration response, including: the processing module is configured to control the communication model to send the model registration response to the second device. The second device is a single-domain management function unit.

[0061] In a possible design, the communication device is a management and orchestration platform applied to an open management architecture. The processing module is configured to control the communication model to receive a model registration request, including: the processing module is configured to control the communication model to receive a model registration request from a second device; and the processing module is further configured to control the communication model to send a model registration response, including: the processing module is configured to control the communication model to send the model registration response to the second device. The second device is a service operation application APP.

[0062] In a possible design, the first identifier of the first model in the information of the first model included in the model registration request can be a vendor identifier, and the inference type of the first model in the information of the first model can be a type of a service operation APP.

[0063] In a possible design, the communication module is configured to receive first indication information from a second device, the first indication information being used to indicate that the first model is allowed to be retrained. The processing module is configured to retrain the first model according to the performance of the first model and the first indication information. The communication module is further configured to send updated model information to the second device, the updated model information being used to update the first model.

[0064] In a possible design, the updated model information can include a storage address of the updated first model.

[0065] In a possible design, the communication module is further configured to receive a model discovery request, the model discovery request being used to request to discover a model. The processing module is configured to control the communication module to send a model discovery response according to the information of the registered model and the model discovery request, the model discovery response including information of the model.

[0066] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0067] In a possible design, the communication module can include a receiving module and a sending module. The sending module is configured to implement the sending function of the communication device in the fifth aspect, and the receiving module is configured to implement the receiving function of the communication device in the fifth aspect.

[0068] In a possible design, the communication apparatus in the fifth aspect can further include a storage module, which stores programs or instructions. When the processing module executes the programs or instructions, the communication apparatus in the fifth aspect can execute the method in the first aspect.

[0069] In a sixth aspect, a communication apparatus is provided for implementing the methods described above. The communication apparatus can be the second device in the second aspect, or an apparatus including the second device, or an apparatus included in the second device, such as a chip. The communication apparatus includes corresponding modules, units, or means for implementing the methods of the second aspect, which can be implemented by hardware, software, or by executing corresponding software by hardware. The hardware or software includes one or more modules or units corresponding to the functions described above.

[0070] In some possible designs, the communication apparatus includes a processing module and a communication module. The processing module is configured to control the communication module to send a model registration request, where the model registration request is used to request registration of a first model. The processing module is further configured to control the communication module to receive a model registration response, where the model registration response is used to indicate whether the first model is successfully registered.

[0071] In a possible design, the model registration request can include information of the first model, and the information of the first model can include at least one of the following: a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

[0072] In a possible design, the inference type of the first model can include at least one of the following: a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

[0073] In a possible design, the capability of the first model can include at least one of the following: a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0074] In a possible design, the intention target of the first model can include at least one of the following: a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

[0075] In a possible design, in a case where the first model is successfully registered, the model registration response can include a second identifier, which is a globally unique identifier allocated by the first device for the first model.

[0076] In a possible design, the communication module is configured to receive a model deployment request, where the model deployment request includes the second identifier, and the model deployment request is used to request deployment of the first model.

[0077] In a possible design, the communication device is a single-domain management function unit applied to a service-oriented management architecture. The processing module is configured to control the communication module to send a model registration request, including: the processing module is configured to control the communication module to send the model registration request to a first device; and the processing module is further configured to control the communication module to receive a model registration response, including: the processing module is configured to control the communication module to receive the model registration response from the first device. The first device is a cross-domain management function unit.

[0078] In a possible design, the communication device is a service operation application (APP) applied to an open management architecture. The processing module is configured to control the communication module to send a model registration request, including: the processing module is configured to control the communication module to send the model registration request to a first device; and the processing module is further configured to control the communication module to receive a model registration response, including: the processing module is configured to control the communication module to receive the model registration response from the first device. The first device is a management and orchestration platform.

[0079] In a possible design, the first identifier of the first model in the information of the first model included in the model registration request can be a vendor identifier, and the inference type of the first model in the information of the first model can be a type of a service operation APP.

[0080] In a possible design, the communication module is configured to send first indication information to the first device, where the first indication information is used to indicate that the first model is allowed to be retrained. The communication module is further configured to receive updated model information from the first device, where the updated model information is used to update the first model.

[0081] In a possible design, the updated model information can include a storage address of the updated first model.

[0082] In a possible design, the communication module is further configured to send a model discovery request, where the model discovery request is used to request to discover a model. The communication module is further configured to receive a model discovery response, where the model discovery response includes information of the model.

[0083] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0084] In a possible design, the communication module can include a receiving module and a sending module. The sending module is configured to implement the sending function of the communication device in the sixth aspect, and the receiving module is configured to implement the receiving function of the communication device in the sixth aspect.

[0085] In a possible design, the communication apparatus in the sixth aspect can further include a storage module, which stores programs or instructions. When the processing module executes the programs or instructions, the communication apparatus in the sixth aspect can execute the method in the second aspect.

[0086] In a seventh aspect, a communication apparatus is provided for implementing the methods described above. The communication apparatus can be the first device in the third aspect, or a device including the first device, or a device included in the first device, such as a chip. The communication apparatus includes corresponding modules, units, or means for implementing the methods in the third aspect, which can be implemented by hardware, software, or by executing corresponding software by hardware. The hardware or software includes one or more modules or units corresponding to the functions described above.

[0087] In some possible designs, the communication apparatus includes a processing module and a communication module. The processing module is configured to assign a globally unique identifier to a first model after the first model is successfully registered, the first model being obtained through offline training, and the communication apparatus being a cross-domain management function unit. The communication module is configured to send the globally unique identifier to a second device, the second device being a single-domain management function unit.

[0088] In a possible design, the processing module is further configured to obtain information of the first model, the information of the first model including at least one of the following: a third identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

[0089] In a possible design, the inference type of the first model includes at least one of the following: a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

[0090] In a possible design, the capability of the first model includes at least one of the following: a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

[0091] In a possible design, the intention target of the first model includes at least one of the following: a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

[0092] In a possible design, the globally unique identifier can be carried in a model deployment request and sent.

[0093] In a possible design, the communication module is further configured to receive a model discovery request, where the model discovery request is used to request discovery of a model. The communication module is further configured to send a model discovery response according to the information of the registered model and the model discovery request, where the model discovery response includes information of the model.

[0094] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0095] In a possible design, the communication module can include a receiving module and a sending module. The sending module is configured to implement the sending function of the communication apparatus in the seventh aspect, and the receiving module is configured to implement the receiving function of the communication apparatus in the seventh aspect.

[0096] In a possible design, the communication apparatus in the seventh aspect can further include a storage module that stores programs or instructions. When the processing module executes the programs or instructions, the communication apparatus in the seventh aspect can execute the method in the third aspect.

[0097] In an eighth aspect, a communication apparatus is provided for implementing the methods described above. The communication apparatus can be the second device in the fourth aspect, or an apparatus including the second device, or an apparatus included in the second device, such as a chip. The communication apparatus includes corresponding modules, units, or means for implementing the methods in the fourth aspect, which can be implemented by hardware, software, or by executing corresponding software by hardware. The hardware or software includes one or more modules or units corresponding to the functions described above.

[0098] In some possible designs, the communication apparatus includes a processing module and a transceiver module. The communication module is configured to receive a globally unique identifier of a first model from a first device, where the first device is a cross-domain management function unit, and the communication apparatus is a single-domain management function unit. The processing module is configured to determine the first model according to the globally unique identifier.

[0099] In a possible design, the globally unique identifier can be carried in a model deployment request and sent.

[0100] In a possible design, the communication module is further configured to send a model discovery request to the first device, where the model discovery request is used to request discovery of a model. The communication module is further configured to receive a model discovery response from the first device, where the model discovery response includes information of the model.

[0101] In a possible design, the model discovery request can include an identifier of the model and / or a version of the model.

[0102] In a possible design, the communication module can include a receiving module and a sending module. The sending module is configured to implement the sending function of the communication apparatus in the eighth aspect, and the receiving module is configured to implement the receiving function of the communication apparatus in the eighth aspect.

[0103] In a possible design, the communication apparatus in the eighth aspect can further include a storage module, which stores programs or instructions. When the processing module executes the programs or instructions, the communication apparatus in the eighth aspect can execute the method in the fourth aspect.

[0104] In the ninth aspect, a communication apparatus (for example, the communication apparatus can be a chip or a chip system) is provided. The communication apparatus includes a processor configured to implement the functions involved in any of the aspects.

[0105] In a possible design, the communication apparatus can further include a memory configured to store necessary programs, instructions and data. The processor is coupled to the memory, and is configured to execute the computer programs or instructions stored in the memory, so that the communication apparatus executes the method in any of the possible implementation manners of the first aspect to the fourth aspect.

[0106] In a possible design, the communication apparatus in the ninth aspect can further include a transceiver. The transceiver can be a transceiver circuit or an interface circuit. The transceiver can be configured to enable the communication apparatus in the ninth aspect to communicate with other communication apparatuses.

[0107] In a possible design, the processor can be integrated with the memory.

[0108] In some possible designs, when the apparatus is a chip system, the apparatus can be composed of a chip, or can include a chip and other discrete devices.

[0109] In the tenth aspect, a communication apparatus is provided. The communication apparatus includes a processor and an interface circuit. The interface circuit is configured to receive a signal from another communication apparatus outside the communication apparatus and transmit the signal to the processor, or send a signal from the processor to another communication apparatus outside the communication apparatus. The processor is configured to implement the method in any of the possible implementation manners of the first aspect to the fourth aspect by means of a logic circuit or by executing code instructions.

[0110] It can be understood that, when the communication apparatus in any of the ninth aspect or the tenth aspect is a chip, the sending action / function described above can be understood as output, and the receiving action / function described above can be understood as input.

[0111] In an eleventh aspect, a computer-readable storage medium is provided, which stores a computer program or instructions, when executed on a communication device, cause the communication device to perform the method of any one of the first aspect to the fourth aspect.

[0112] In a twelfth aspect, a computer program product containing instructions, which, when executed on a communication device, cause the communication device to perform the method of any one of the first aspect to the fourth aspect.

[0113] In a thirteenth aspect, a communication system is provided, comprising: a first device for implementing the method of the first aspect, and a second device for implementing the method of the second aspect.

[0114] In a fourteenth aspect, a communication system is provided, comprising: a first device for implementing the method of the third aspect, and a second device for implementing the method of the fourth aspect. BRIEF DESCRIPTION OF DRAWINGS

[0115] FIG. 1 is a schematic diagram of an AI / ML management workflow;

[0116] FIG. 2 is a schematic diagram of an architecture of a communication system provided by an embodiment of the present application;

[0117] FIG. 3 is a schematic diagram of an architecture of a management service provided by an embodiment of the present application;

[0118] FIG. 4 is a schematic diagram of a service-based management architecture provided by an embodiment of the present application;

[0119] FIG. 5 is a schematic diagram of an open management architecture provided by an embodiment of the present application;

[0120] FIG. 6 is a schematic diagram of a model registration method provided by an embodiment of the present application;

[0121] FIG. 7 is a schematic diagram of another model registration method provided by an embodiment of the present application;

[0122] FIG. 8 is a schematic diagram of yet another model registration method provided by an embodiment of the present application;

[0123] FIG. 9 is a schematic diagram of a structure of a communication device provided by an embodiment of the present application;

[0124] FIG. 10 is a schematic diagram of a structure of another communication device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0125] The various aspects, embodiments or features presented can be presented with respect to a system that can include multiple devices, components, modules, etc. It should be appreciated that various systems can include additional devices, components, modules, etc. and / or can not include all of the devices, components, modules etc. discussed in connection with the figures. Additionally, a combination of these approaches can be used.

[0126] The technical solutions of the embodiments of the present application can be applied to various communication systems, such as a wireless fidelity (Wi-Fi) system, a vehicle to everything (V2X) communication system, a device-to-device (D2D) communication system, a vehicle networking communication system, a worldwide interoperability for microwave access (WiMAX) communication system, a 4th generation (4G) mobile communication system such as a long term evolution (LTE) system, a worldwide interoperability for microwave access (WiMAX) communication system, a 5th generation (5G) mobile communication system such as a new radio (NR) system, and a future communication system, etc.

[0127] In order to better understand the embodiments of the present application, the following points are explained before introducing the embodiments of the present application.

[0128] First, in the embodiments of the present application, "for indicating" can include for directly indicating and for indirectly indicating. When describing that certain "indication information" is for indicating A, it can include that the indication information directly indicates A or indirectly indicates A, and does not mean that A must be carried in the indication information.

[0129] The information indicated by the indication information is referred to as to-be-indicated information. In a specific implementation process, there are many ways to indicate the to-be-indicated information, for example, but not limited to, the to-be-indicated information can be directly indicated, such as the to-be-indicated information itself or an index of the to-be-indicated information. The to-be-indicated information can also be indirectly indicated by indicating other information, where the other information and the to-be-indicated information have an association relationship. The to-be-indicated information can also be only indicated in part, and the other part of the to-be-indicated information is known or agreed in advance. For example, the indication of specific information can also be achieved by means of the arrangement order of each information agreed in advance (for example, specified by a protocol), thereby reducing the indication overhead to a certain extent. Meanwhile, the common part of each information can be identified and uniformly indicated, so as to reduce the indication overhead caused by separately indicating the same information.

[0130] In addition, the specific indication manner can also be various existing indication manners, for example, but not limited to, the above indication manners and various combinations thereof. Specific details of various indication manners can be referred to the prior art, and will not be described herein. As known from the above, for example, when multiple information of the same type needs to be indicated, the indication manners of different information can be different. In a specific implementation process, the required indication manner can be selected according to specific needs, and the selected indication manner is not limited by the embodiments of the present application. In this way, the indication manner involved in the embodiments of the present application should be understood as covering various methods that can enable the to-be-indicated party to know the to-be-indicated information.

[0131] The to-be-indicated information can be sent as a whole, or can be divided into multiple sub-information and sent separately, and the sending period and / or sending time of the sub-information can be the same or different. The specific sending method is not limited by the present application. The sending period and / or sending time of the sub-information can be predefined, for example, predefined according to a protocol, or configured by the transmitting end device by sending configuration information to the receiving end device.

[0132] Secondly, in the embodiments of the present application, the first, second and various numerical numbers are only for differentiation for convenience of description, and do not limit the scope of the embodiments of the present application. For example, different indication information is differentiated. For another example, the first indication information and the second indication information are only for differentiating different indication information, and the sequence thereof is not limited. Those skilled in the art can understand that the words of "first", "second" and the like do not limit the quantity and execution sequence, and the words of "first", "second" and the like do not necessarily mean different.

[0133] Third, in the embodiments of the present application, the descriptions such as "when", "in the case of", "if" and "whether" all refer to the device making corresponding processing under certain objective circumstances, and are not limited in time, and do not require the device to have a judgment action when implemented, nor does it mean that there are other limitations.

[0134] Meanwhile, in the embodiments of the present application, the words such as "exemplary" or "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the words such as "exemplary" or "for example" are intended to present the relevant concept in a specific manner for understanding.

[0135] Finally, the network architecture and service scenarios described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. It can be known by those skilled in the art that, with the evolution of network architecture and the appearance of new service scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0136] The related terms, concepts or technologies that may be involved in the embodiments of the present application are introduced as follows:

[0137] 1. ML model (may be referred to as model)

[0138] Learned from data, a function that can realize a specific function / mapping. The model can be obtained based on AI or ML technology, and therefore, can also be referred to as an artificial intelligence / AI model, a machine learning / ML model, etc. Commonly used algorithms for generating AI / ML models include supervised learning, unsupervised learning, reinforcement learning, and the corresponding models can be referred to as supervised learning models, unsupervised learning models, reinforcement learning models. For example, a supervised learning model can be a classification model, a prediction model, a regression model, etc., and an unsupervised learning model can be a clustering model. In addition, the model can also be obtained based on neural network (NN) technology, and such a model can also be referred to as a neural network model, a deep learning model, etc.

[0139] The main process of building a model includes problem analysis (determining the data to be collected and the type of model to be used, etc.), data collection, model training and model inference, etc. Among them, model training is a process of determining the parameters of the model using data, and model inference is a process of predicting the output result according to the new input using the trained model.

[0140] The ML model defined in the 3rd generation partnership project (3GPP) technical specification (TS) is manageable for the ML model algorithm. Among them,

[0141] (1) The ML model algorithm is a mathematical algorithm, through which a set of inference outputs can be generated by running a set of input data.

[0142] (2) The ML model algorithm is proprietary and not within the scope of standardization, so it is not handled in this specification.

[0143] (3) The ML model can include metadata, which can include, for example, information related to the trained model, and the applicable running is the context.

[0144] It should be understood that in the embodiments of the present application, AI refers to a technology that presents human intelligence through a computer program, and ML focuses on developing computer programs that can access data and use these data for self-learning. In the subsequent description, artificial intelligence and machine learning are no longer distinguished, for example, the ML model can also be referred to as the AI model.

[0145] 2. AI / ML model application in wireless network

[0146] In order to improve the intelligent and automated level of the network, AI and ML technologies are being applied in more and more fields, including the management domain, the core network (CN) domain and the radio access network (RAN) domain. Multiple fields are studying how to apply AI / ML technology to enable network intelligence.

[0147] The application of AI / ML functions in 5G system includes intelligent optimization use cases in RAN, such as mobility load balancing (MLB) or load balancing (LB), mobility robustness optimization (MRO) or mobility optimization (MO), energy saving (ES) or network energy saving (NES), etc., management data analysis services in network management, such as management data analytic function (MDAF), and network data analysis services in core network, such as network data analytics function (NWDAF).

[0148] The life cycle management (LCM) capability of AI / ML is defined in 3GPP TS28.105, as shown in FIG. 1, including five main management capabilities: training, testing, simulation, deployment and inference, which are described as follows:

[0149] Training: training one or a group of ML models, including initial training and retraining (referred to as retraining). It also includes the validation of ML models to evaluate the performance of ML models on training data and validation data.

[0150] Testing: if the validation result is not as expected (for example, the variance is not acceptable), the ML model needs to be retrained. ML model training is the initial stage of the life cycle process.

[0151] Simulation: running the ML model for inference in a simulation environment. The purpose is to evaluate the inference performance of the AI / ML entity in a simulation environment before it is applied to the target network or system.

[0152] Deployment: the process of making the trained ML model available for the target AI / ML inference function.

[0153] Inference: the process of using ML models for inference through AI / ML inference functions.

[0154] Further, 3GPP TS28.105 also defines the application deployment architecture of the life cycle management capability of AI / ML based on the service-oriented management architecture, taking the training function and the inference function as an example. For example, the training function and the inference function can be deployed in a single-domain management function unit, or the training function is deployed in a single-domain management function unit and the inference function is deployed in a base station.

[0155] Under the service management architecture, the ML model is generated in a single-domain management function unit, and the information of the model is stored in the single-domain management function unit. The operator and the third-party application cannot learn the model information that meets their needs in time, which leads to the fact that the operator or the third-party application cannot select to perform a specific operation, for example, selecting a suitable model to be deployed into an inference function to perform a specific inference function.

[0156] In addition, the AI / ML life cycle management function can also be deployed based on an open management architecture, under which model inference is performed to generate in each application (APP), and the number of model inference providers in the network increases. The management model requester cannot find a suitable model inference provider to perform a specific operation, which affects the efficiency and real-time performance of the model application.

[0157] That is, in the current LCM of the ML model, after the ML model is generated at a node, the model-related information of the ML model is stored in the node, and the ML model is directly deployed to the inference function by the node. Other nodes cannot obtain the ML model that meets the demand and deploy it to the inference function to perform a specific inference.

[0158] In addition, after the ML model is applied, how to search, reproduce the target inference function in different application environments to track the recall of the most suitable model version, so as to handle the sudden change of the deployment environment of the target inference function, is not clear at present.

[0159] To solve the above problems, the embodiment of the present application provides a model registration method, which can record and track the trained ML model to ensure the traceability, repeatability and maintainability of the model, thereby improving the efficiency and quality of model development and deployment, so as to realize future search, repeatability, sharing and loading to the target inference function in different environments.

[0160] Please refer to FIG. 2, which is a schematic diagram of an architecture of a communication system to which embodiments of the present application can be applied. As an example, as shown in FIG. 2, the communication system includes a first device and a second device, which can indirectly or directly communicate with each other.

[0161] In some possible cases, the first device can serve as a management service (MnS) consumer, which can provide functions such as model registration, management and discovery, while the second device can serve as a management service producer (MnS Producer), which can provide model information and use the model for network management.

[0162] In this case, in one possible implementation, as for the model registration of the second device online training (which can be referred to as online model registration), the first device can receive a model registration request from the second device, register the model according to the model registration request, and send a model registration response to the second device to inform the second device whether the model registration is successful, so as to realize the management of the model generated by the first device. Therefore, the first device or the first device according to the request of the third party device can select a model meeting its demand based on the information of the registered model to deploy, so as to execute corresponding inference.

[0163] In another possible implementation, as for the model registration of offline training (which can be referred to as offline model registration), the first device can assign a globally unique identifier to the offline model after the offline model registration is successful, and send the globally unique identifier to the second device, so as to facilitate the second device to know the model supported by its capability.

[0164] In some other possible cases, the first device can also act as a management service producer, and the second device acts as a management service consumer, which is not limited.

[0165] The first device and the second device can be management entities / management devices defined in 3GPP, such as the management function (MnF) shown in FIG. 3, the externally visible behavior and interface of which are defined as MnS. In the service-giving management architecture, the MnF plays the role of management service producer (MnS Producer) or management service consumer (MnS Consumer). The management service produced by the MnF as the management service producer can have multiple management service consumers, and the MnF as the management service consumer can consume multiple management services from one or more management service producers. In some implementations, one MnF can simultaneously act as a management service producer and a management service consumer.

[0166] Optionally, the communication system can further include a third party device, such as an operator device, a third party management device, a third party application, etc. The third party device can obtain a model meeting its demand from the registered model stored by the first device for deployment.

[0167] Further, the embodiment of the present application further provides a schematic diagram of a service management architecture applicable to the communication system shown in Fig. 2. As shown in Fig. 4, the service management architecture includes a business support system (BSS), a cross-domain management function (CD-MnF) unit / system, a (single) domain management function (D-MnF) unit / system and network elements. The interface between the CD-MnF and the D-MnF can be referred to as a northbound interface, the interface between the D-MnF and the network elements can be referred to as a southbound interface, the CD-MnF can be the first device, the D-MnF can be the second device, the first device can also be referred to as a first management device, and the second device can also be referred to as a second management device, which is not limited.

[0168] It should be understood that the cross-domain management function unit can be a northbound management system (NMS), a MnS Producer, a MnS Consumer or the like, and the domain management function unit can be an element management system (EMS), a mobile broadband automation engine (MAE), a MnS Producer, a MnS Consumer or the like.

[0169] If the management service is a management service provided by the cross-domain management function unit, the cross-domain management function unit is a management service producer, and the business support system is a management service consumer. If the management service is a management service provided by the domain management function unit, the domain management function unit is a management service producer, and the cross-domain management function unit is a management service consumer.

[0170] If the management service is a management service provided by the network element, the network element is a management service producer, and the domain management function unit is a management service consumer.

[0171] The business support system is for communication services, and is used to provide charging, settlement, accounting, customer service, business, network monitoring, communication service life cycle management, service intention translation and the like.

[0172] The cross-domain management function unit, which can also be referred to as a network management function (NMF) unit, can be a network management system (NMS), a network function management service consumer (NFMS_C), or another network management entity. The cross-domain management function unit provides one or more of the following management functions or services: network lifecycle management, network deployment, network fault management, network performance management, network configuration management, network assurance, network optimization, and translation of an intent from a communication service provider (Intent-CSP) into a network.

[0173] The network referred to in the management functions or services can include one or more network elements or sub-networks, or a network slice. In other words, the network management function unit can be a network slice management function (NSMF) unit, a cross-domain MDAF, a cross-domain self-organization network function (SON Function), or a cross-domain intent driven management service (MnS) unit.

[0174] Optionally, in some deployment scenarios, the cross-domain management function unit can also provide lifecycle management of a sub-network, deployment of a sub-network, fault management of a sub-network, performance management of a sub-network, configuration management of a sub-network, assurance of a sub-network, optimization of a sub-network, translation of an intent from a communication service provider (Intent-CSP) into a network, or translation of an intent from a communication service consumer (Intent-CSC) into a network. The sub-network can be composed of multiple small sub-networks, and can be a network slice sub-network.

[0175] Domain management function (Domain-MnF), also known as network management function (NMF) or network element management function. For example, the domain management function can be a MAE, EMS, network function management service provider (NFMS_P) and other network management entities.

[0176] The domain management function provides one or more of the following functions or management services: lifecycle management of subnetwork or network element, deployment of subnetwork or network element, fault management of subnetwork or network element, performance management of subnetwork or network element, assurance of subnetwork or network element, optimization function of subnetwork or network element, and translation of intent from network operator (Intent-NOP) of subnetwork or network element, etc. The subnetwork here includes one or more network elements. The subnetwork can also include a subnetwork, i.e. one or more subnetworks form a larger subnetwork.

[0177] Optionally, the subnetwork here can also be a network slice subnetwork. The domain management system can be a network slice subnetwork management function (NSSMF), a domain management data analysis function (Domain MDAF), a domain SON function, a domain intent MnS, etc.

[0178] The domain management function can be classified in the following ways, including:

[0179] According to the network type, it can be classified into: radio access network domain management function (RAN-Domain-MnF), core network domain management function (CN-Domain-MnF), transport network domain management function (TN-Domain-MnF), etc. It should be noted that the domain management function can also be a domain network management system, which can manage one or more of the access network, core network or transport network;

[0180] According to the administrative region classification, the domain management function unit of a certain region can be classified, such as a Shanghai domain management function unit, a Beijing domain management function unit, and the like.

[0181] The network element is an entity providing network services, including core network elements, access network devices, and the like. Among them, the core network elements include: access and mobility management function (AMF), session management function (SMF), policy control function (PCF), NWDAF, network repository function (NRF), and gateways, and the like. The access network devices include: base stations (such as gNB, eNB), central unit control plane (CUCP), central unit (CU), distribution unit (DU), central unit user plane (CU-UP), and the like.

[0182] Among them, the network element can provide one or more of the following management functions or management services: lifecycle management of the network element, deployment of the network element, fault management of the network element, performance management of the network element, guarantee of the network element, optimization function of the network element, and translation of the network element intention, and the like.

[0183] In addition to the service management architecture described above, the communication system provided by the embodiments of the present application can also be applied to an open management architecture, as shown in FIG. 5, which includes network elements and an operation and maintenance system, and the operation and maintenance system includes a management and orchestration platform and an operation and maintenance APP. Among them, the management and orchestration platform includes a training function and a model repository, provides management and orchestration of ML models, and the management and orchestration platform provides a management interface for each operation and maintenance APP (including registration and discovery of ML models, loading of management ML models, and the like), and the operation and maintenance APP includes two types of network operation APP and service operation APP, which can provide network operation specific management functions, including network construction management functions, network optimization management functions, network maintenance management functions, service provisioning management functions, and service guarantee management functions, and the like.

[0184] Among them, the management and orchestration platform can be a first device, such as a server, a computer, and the like, at this time, the first device can also be referred to as a first management device, and the operation and maintenance APP can be a second device, and usually the operation and maintenance APP corresponding to the second device is a service operation APP.

[0185] It should be understood that the names of nodes, modules, devices or network elements in different scenarios or architectures or systems, and the names of communication interfaces between two nodes, modules, devices or network elements in the embodiments of the present application are exemplarily given, and the possibility of name changes in future communication systems or scenarios or architectures is not excluded.

[0186] The model registration method provided by the embodiments of the present application will be described in detail below with reference to FIGS. 6-8.

[0187] Exemplarily, FIG. 6 is a flowchart of a model registration method provided by the embodiments of the present application, which is exemplarily described by taking the communication between the first device and the second device shown in FIG. 2 as an example. Of course, the subject performing the actions of the first device in the method can also be a device / module in the first device, such as a chip, a processor, a processing unit, etc. in the first device, and the subject performing the actions of the second device in the method can also be a device / module in the second device, such as a chip, a processor, a processing unit, etc. in the second device, and the like, without limitation.

[0188] As shown in FIG. 6, the model registration method includes:

[0189] S601, the second device sends a model registration request to the first device. Correspondingly, the first device receives the model registration request from the second device.

[0190] After training the generated model within the management range of the second device, the second device can register the information of the trained model to the first device through the model registration request, so that the first device completes the registration and management of the model. Taking the trained model as a first model as an example, the model registration request is used to request registration of the first model, or in other words, the model registration request is used to request the first device to learn the information of the first model. Therefore, the first device can register, store and manage the information of multiple models based on the model registration requests sent by different second devices.

[0191] The model registration request can include the information of the model to be registered, such as the information of the first model, which can include at least one of the following: a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

[0192] The first identifier of the first model can refer to the information used to identify the first model within the management range of the second device, which can be an identifier allocated by the second device for the first model.

[0193] The inference type of the first model can refer to a network optimization type applicable using inference of the first model, and the inference type can be identified by a differentiating name (DN). For example, the inference type of the first model can include at least one of a RAN intelligent type, a management data analysis type, or a network data analysis type. The RAN intelligent type, such as MRO, MLB, and ES, can refer to use cases defined in Chapter 7 of 3GPP TS 28.313 V17.9.0 or use cases defined in Chapter 5.1 of 3GPP TS 28.310 V18.4.0, the management data analysis (MDA) type, such as coverage analysis and mobility analysis, can refer to types of MDA defined in Chapter 8.4 of 3GPP TS 28.104 V18.2.0, and the network data analysis type, such as NWDAF application optimization, such as slice load analysis and user data congestion analysis, can refer to use cases defined in Chapter 6 of 3GPP TS 23.288 V18.4.0, which will not be described in embodiments of the present application.

[0194] The version of the first model can refer to a version number of the current first model, and can be used to feed back an update of the first model.

[0195] The capability of the first model can refer to a function or capability possessed or owned by the trained first model, and the capability of the first model can be indicated by a string. For example, the capability of the first model can include at least one of traffic analysis capability, coverage analysis capability, mobility analysis capability, load analysis capability, fault analysis capability, network slice throughput rate analysis capability, slice load analysis capability, network slice traffic prediction analysis capability, service experience analysis capability, energy efficiency analysis capability, or energy consumption analysis capability. It should be understood that a model implementing a specific inference type can correspond to one or more capabilities, for example, a model of the MLB inference type can have traffic analysis capability and mobility analysis capability. In some implementations, the capability of the first model can also be customized by a manufacturer, which will not be limited in embodiments of the present application.

[0196] The intent target of the first model can refer to an intent implementation that the AI / ML consumer can achieve according to the first model. After receiving the intent, the AI / ML producer (such as the second device) can implement a simple intent by calling a single model, or can also need to call multiple models for orchestration to implement a complex intent. For a simple intent, it can be easy to map the execution logic to one or more models. For a complex intent, it can be necessary to use multiple models and the corresponding functions to manage their interrelated execution. Therefore, the AI / ML consumer (such as the first device) needs to obtain the information about the capabilities and interrelationships of the models to be used, which will help the AI / ML consumer to select a suitable intent target according to the intent target information that the models can support. Exemplarily, the intent target of the first model can include at least one of a coverage intent target, an energy saving intent target, or a wireless resource optimization intent target, wherein the coverage intent target can include a signal quality and signal strength intent target, the energy saving intent target can include an energy efficiency and / or energy consumption intent target, and the wireless resource optimization intent target can include a user rate intent target. The intent target can also be identified by a DN, which can be defined in the chapter 6.2.2.1.1.3 of 3GPP TS 28.321.

[0197] In the embodiments of the present application, the AI / ML consumer can also be referred to as an AI / ML MnS consumer or a MnS consumer, and correspondingly, the AI / ML producer can also be referred to as an AI / ML MnS producer or a MnS producer, which is not limited.

[0198] It should be understood that the above exemplary lists several pieces of information of the first model, in addition to which, the information of the first model can also include other information that can represent the characteristics or purposes of the model, such as type information of the first model, which is used to indicate whether the first model is a single-sided model or a double-sided model, and / or input information and / or output information of the first model, wherein the input information can be a type or format of a data set of the model, and the output information can be a type or format of an output of the model, and / or a storage address of the first model, which is not limited.

[0199] It should also be understood that in addition to the model information, the model registration request can also include information of the second device, such as an identifier of the second device, a type of the second device, and the like, which is not limited.

[0200] Optionally, the information of the first model can be carried in a newly defined model instance object, for example, by defining an MLEntityRegister IOC in the model registration message to carry the information of the first model, or the information of the first model can also be carried by reusing an existing model instance object, for example, by defining an MLModel IOC in the model registration request to carry the information of the first model, which is not limited.

[0201] S602, the first device sends a model registration response to the second device. Correspondingly, the second device receives the model registration response from the first device.

[0202] After receiving the model registration request, the first device can perform model registration on the first model, determine whether the first model meets the registration requirement according to the information of the first model, and feed back a model registration response to the second device, which is used to indicate / notify whether the first model is successfully registered or the registration is completed.

[0203] In the case that the first model is successfully registered or the registration is completed, that is, the model registration response is used to indicate that the first model is successfully registered or the registration is completed, the first device can store the information of the first model. Optionally, the first device can also assign a globally unique identifier to the successfully registered first model as a second identifier of the first model, which can enable the first model to be identified within the management range of the first device, and the assigned second identifier can be sent to the second device in the model registration response, that is, the model registration response includes the second identifier, so as to be used for subsequent deployment and inference of the model.

[0204] Model registration can be understood as enabling the first device to learn and store the information of the model, so as to assign an identifier to the model, which can be a unique identifier within the management range of the first device (such as the second identifier described above). When the first device assigns a unique identifier to the model, it can be considered that the registration is successful, otherwise it can be considered that the registration fails. In some implementations, model registration can be performed in the model registration function in the first device, and optionally, model registration can also be performed in the model storage library in the first device, which is not limited in the present application.

[0205] As an AI / ML consumer, the first device can request a device with inference function to deploy a model supported by the device according to the information of the registered model and inference requirements, so that the device can perform target inference function according to the deployed model.

[0206] Taking the first model as an example, the model deployment of the first device can be implemented in the following manner: the first device sends a model deployment request, the model deployment request comprising the second identifier, the model deployment request being used to request deployment of the first model into a specific inference function. For example, the first device can send a model deployment request to the second device, the model deployment request indicating the model identifier to be deployed. Optionally, the model deployment request can further comprise an inference type. Correspondingly, the second device can receive the model deployment request from the first device, so that the second device can deploy the first model into the inference function according to the model identifier and the inference type, and then the inference function performs inference according to the first model. Thus, after receiving the model deployment request, the second device can determine whether it can deploy the first model according to the inference capability, the computing power size, the memory size, and the like.

[0207] Since the first model is generated by training within the management range of the second device, the first model can be stored in the second device. If the first device needs to request a third-party device (such as a third device) other than the second device to deploy the first model, the third device can obtain the first model from the second device for deployment after receiving the model deployment request. In this case, the model deployment request can comprise a storage address of the first model.

[0208] Optionally, the device receiving the model deployment request can further feed back a model deployment response to the first device, the model deployment response being used to indicate whether the model is successfully deployed and the model deployment process state (for example, “running”, “cancelling”, “paused”, “completed”). For example, the first device receives a model deployment response from the second device, the model deployment response being used to indicate whether the first model is successfully deployed. In the case that the model is not successfully deployed, the model deployment response can further carry the reason why the model is not successfully deployed, such as insufficient computing power or insufficient memory.

[0209] Optionally, the first device can further send a model deployment requirement to the second device or the third device when requesting the model deployment, the model deployment requirement can refer to a requirement for enabling the second device or the third device to implement the deployment of the model to the target inference function, i.e., the second device or the third device can trigger the model deployment (or loading) according to the model deployment requirement, for example, the model deployment requirement can include at least one of the following: a first threshold, a second threshold, wherein the first threshold is related to the model to be loaded, and the second threshold is related to an existing model used for inference, and the first threshold and the second threshold can be performance thresholds of the model. Optionally, the model deployment requirement can further include an identifier of the model, an inference type of the model, a version of the model, a capability of the model, or an intended target of the model, etc., which are not limited herein. Thus, the model deployment requirement can be understood as a control strategy for triggering the model deployment / loading by the second device or the third device, and the model deployment requirement is configured by the first device to the second device or the third device.

[0210] Optionally, the model deployment requirement can be sent in the model deployment request, or can be sent separately from the model deployment request, which is not limited herein.

[0211] In addition, the first device can further provide a model discovery service, so that the second device or the third device can discover available models. In one possible implementation, the first device can receive a model discovery request, and send a model discovery response according to the information of the registered model and the model discovery request. The model discovery request can be used to query or discover the information of the model, for example, the model discovery request is used to request to discover / query the model, the model discovery request can include an identifier and / or a version of the model to be discovered, and the model discovery response includes the information of the model. In some scenarios, the model discovery request can also be referred to as a model query request, and correspondingly, the model discovery response can also be referred to as a model query response, which are not limited herein.

[0212] Optionally, the device discovering the model can further request the corresponding device to deploy the model (on the inference function of the device) according to the storage address of the model and / or the identifier of the model, so as to execute the target inference function and complete the target task.

[0213] In the model registration method, the first device can obtain the information of the model generated by the different second devices within the management range of the second device through the model registration request, register the model, and feed back the model registration response to the first device, so as to implement the storage and management of the model, thereby enabling the first device or the third device to find a model meeting the requirement from the registered model information according to the inference requirement, and improving the efficiency and real-time performance of the model application.

[0214] The model registration method between the first device and the second device in the above embodiment can be applied to the service-oriented management architecture as shown in FIG. 4 and the open management architecture as shown in FIG. 5 respectively. The following takes the two architectures as examples to explain S601 and S602 in detail.

[0215] Case 1: The model registration method can be applied to the service-oriented management architecture as shown in FIG. 4. At this time, the first device can be a cross-domain management function unit, and the second device can be a single-domain management function unit. The second device has a training function or has both a training function and an inference function.

[0216] In this case, the first model can be trained by the training function in the single-domain management function unit, or can be trained by the training function of the network element managed by the single-domain management function unit, or can be trained by the training function of a third-party device. The model training can be online training or offline training.

[0217] That is, the model trained by the training function within the management range of the single-domain management function unit can be registered in the cross-domain management function unit by the single-domain management function unit through a model registration request. At this time, S601 is specifically: the single-domain management function unit sends a model registration request to the cross-domain management function unit. Correspondingly, the cross-domain management function unit receives the model registration request sent by the single-domain management function unit.

[0218] After the cross-domain management function unit completes the registration of the first model according to the information of the first model in the model registration request, it can assign a globally unique identifier, i.e., a second identifier, to the registered first model. The second identifier can be sent in the model registration response. At this time, S602 is specifically: the cross-domain management function unit sends a model registration response to the single-domain management function unit. Correspondingly, the single-domain management function unit receives the model registration response from the cross-domain management function unit.

[0219] Therefore, the cross-domain management function unit can register and manage the information of the models from different single-domain management function units.

[0220] When the cross-domain management function unit acts as an AI / ML consumer, it can discover or select a model that meets its needs according to the information of the registered models stored by it, to perform model deployment and inference. For example, the cross-domain management function unit sends a model deployment request to the single-domain management function unit, the model deployment request being used to request the single-domain management function unit to deploy the first model. Correspondingly, the single-domain management function unit can deploy the first model according to the model deployment request, to perform corresponding inference based on the deployed first model.

[0221] Or when the third-party management function unit / single-domain management function unit is an AI / ML consumer, the third-party management function unit / single-domain management function unit can send a model discovery request to the cross-domain management function unit, the model discovery request being used to request to discover a model, and correspondingly, the cross-domain management function unit can determine information of a model to be discovered according to information of the registered model, and feed back the information to the third-party management function unit / single-domain management function unit through a model discovery response. For the third-party management function unit, after obtaining the information of the model, the third-party management function unit can request the single-domain management function unit to deploy the model according to the information of the model through a model deployment request, so as to execute target reasoning.

[0222] In case 1, the single-domain management function unit registers the model into the registration entity in the cross-domain management function unit, so that the cross-domain management function unit can learn the model information in each single-domain management function unit, thereby enabling the cross-domain management function unit or the third-party management function or the operator and the like as an AI / ML consumer to use the most suitable model version to execute target reasoning functions.

[0223] Case 2, the above model registration method can be applied to the open management architecture as shown in the above FIG. 5, at this time, the first device can be a management and orchestration platform, and the second device can be a service operation APP, the first device has a training function, and the second device has a reasoning function, and optionally, the second device can also have a training function.

[0224] In this case, the first model can be pre-stored in the service operation APP, and the service operation APP can execute corresponding network operation and management according to the first model, for example, the service operation APP is a coverage capability optimization (CCO) APP, and coverage capability optimization can be executed according to the first model, that is, the first model is used to execute coverage capability optimization reasoning.

[0225] The first model of the service operation APP can be registered into the management and orchestration platform by the service operation APP through a model registration request. At this time, S601 is specifically: the service operation APP sends a model registration request to the management and orchestration platform, and correspondingly, the management and orchestration platform receives the model registration request from the service operation APP.

[0226] Optionally, the first identifier of the first model in the model registration request can be a vendor ID, and the reasoning type of the first model can be the type of the service operation APP. That is, the first model can be identified by the vendor ID, and the application type of the service operation APP is the reasoning type of the corresponding first model. Optionally, the first identifier can also be indicated by the vendor ID and the vendor certificate identifier.

[0227] It should be understood that the model registration request can also include the identification of the service operation APP.

[0228] After the management and orchestration platform completes the registration of the first model according to the information of the first model in the model registration request, S602 specifically comprises: the management and orchestration platform sends a model registration response to the service operation APP, and correspondingly, the service operation APP receives the model registration response from the management and orchestration platform.

[0229] Therefore, the management and orchestration platform can register and manage the information of models from different service operation APPs.

[0230] When the management and orchestration platform is an AI / ML consumer, it discovers or selects a model that meets its needs according to the information of the registered models it stores, deploys it to a suitable inference function corresponding to the service operation APP, and performs target inference.

[0231] When the third-party device / service operation APP is an AI / ML consumer, similar to the third-party management function unit / single-domain management function unit in scenario 1, it can also obtain the information of available models through a model discovery request, which will not be described here. For example, the service operation APP can send a model discovery request to the management and orchestration platform to request the historical model information of the first model, where the model discovery request can include the identification and / or version information of the first model to be queried, which is not limited.

[0232] In the scenario of scenario 2, since the management and orchestration platform has a training function, the service operation APP (i.e., the second device) can also send an indication to the management and orchestration platform (i.e., the first device) whether the management and orchestration platform is allowed to perform model retraining on the model, so that the management and orchestration platform can determine whether the registered model needs to be retrained.

[0233] In a possible design, the second device can send first indication information to the first device, and correspondingly, the first device receives the first indication information from the second device. The first indication information is used to indicate whether the first model is allowed to be retrained, for example, the first indication information can be indicated by 1 bit, and the bit value of 1 indicates that the first model is allowed to be retrained, and the bit value of 0 indicates that the first model is not allowed to be retrained.

[0234] In the case where the first indication information is used to indicate that the first model is allowed to be retrained, the first device can retrain the first model according to the performance of the first model and the first indication information, and send updated model information to the second device, the updated model information being used to update the first model.

[0235] That is, the management and orchestration platform has a model performance monitoring management function, which can monitor the performance of the first model when used in the service operation APP, such as the performance of the trained model, the performance of the tested model, and the performance of the inference, etc. According to the performance of the first model, it can be judged whether to retrain the first model. In the case that the performance does not meet the requirements, the first model can be retrained, and the retrained first model is sent to the service operation APP through the updated model information, so as to update the first model of the service operation APP.

[0236] The updated model information can include the storage address of the updated first model. Thus, the service operation APP can download the updated first model according to the storage address and redeploy the model. In some implementations, the management and orchestration platform can send the updated first model to the service operation APP in the form of a file, so that the service operation APP updates the first model according to the file, or the updated model information can include the parameters of the updated first model.

[0237] Optionally, the first indication information can be sent in the model registration request or separately from the model registration request, such as being sent in other defined or newly defined messages, which is not limited.

[0238] In case 2, the service operation APP registers the model in the management and orchestration platform, which can make the management model requester such as the management and orchestration platform or the third party device know the model information, so as to find a suitable management model provider such as the service operation APP to perform a specific operation, which improves the efficiency and real-time performance of the model application.

[0239] As shown in FIG. 5 in the open management architecture, the management and orchestration platform includes a training function, a registration function and a model storage function. The above S601 and S602 can be implemented by the registration function, that is, the registration function receives the model registration request from the service operation APP, stores the information of the registered model in the model storage library, and sends the model registration response to the service operation APP. For example, one possible implementation can be as shown in FIG. 7, which is a specific implementation process of the model registration method in the open management architecture. The method includes:

[0240] S701, the service operation APP sends a model registration request to the registration function. Correspondingly, the registration function receives the model registration request from the service operation APP.

[0241] The model registration request is used to request registration of the first model, the first model is pre-installed in the service operation APP, and the model registration request can include one or more first model information in the manufacturer identifier of the APP, the manufacturer certificate identifier of the APP, the identifier of the APP, the application type of the APP, the first identifier of the first model, the capability of the first model, the intention target of the first model, and the first indication information. The manufacturer identifier of the APP and the manufacturer certificate identifier of the APP are used to identify the manufacturer of the APP, and the first identifier of the first model can be an identifier assigned by the APP and unique within the APP application. The specific description of other model information can be referred to the description above, and will not be repeated here.

[0242] S702, the registration function registers the first model.

[0243] The registration function registers the first model according to the information of the first model in the model registration request, and assigns a registration identifier to the registered first model, which is a globally unique identifier of the first model corresponding to the second identifier described above.

[0244] Further, after the registration function completes the registration of the first model, the information of the registered first model can be stored in the model storage.

[0245] S703, the registration function sends a model registration response to the service operation APP. Correspondingly, the service operation APP receives the model registration response from the registration function.

[0246] The model registration response is used to indicate whether the first model is successfully registered, and in the case that the first model is successfully registered, the model registration response can include the registration identifier of the first model and the identifier of the APP. The specific description of the model registration response can be referred to the description above, and will not be repeated here.

[0247] In the case that the model registration request includes the first indication information and the first indication information is used to indicate that the first model is allowed to be retrained, the model registration method can further include the following steps:

[0248] S704, the registration function and the training function perform model retraining.

[0249] In one possible implementation, the registration function can send the first indication information to the training function. Correspondingly, the training function can receive the first indication information from the registration function, so as to determine to retrain the first model according to the performance of the first model and the first indication information, and obtain the information of the updated first model after completing the model retraining.

[0250] In another possible implementation, the registration function can determine whether to retrain the first model according to the first indication information and the performance of the first model, and in the case of retraining, the registration function can trigger the training function to retrain the first model to obtain information of the updated first model.

[0251] It should be understood that after the first model is retrained, the information of the updated first model is also stored in the model repository.

[0252] S705, the registration function sends the updated model information to the service operation APP. Correspondingly, the service operation APP receives the updated model information from the registration function.

[0253] The updated model information is used to update the first model, and the updated model information can include a storage address of the updated first model or the updated first model, which is not limited.

[0254] The description of the model retraining process in S704 and S705 can be referred to the above implementation, which is not repeated here.

[0255] When the model is discovered, the registration function can also receive the model discovery request of the service operation APP to send the model discovery response to the service operation APP, as follows S706 and S707:

[0256] S706, the service operation APP sends a model discovery request to the registration function. Correspondingly, the registration function receives the model discovery request from the service operation APP.

[0257] The model discovery request is used to request to discover available models.

[0258] S707, the registration function sends a model discovery response to the service operation APP. Correspondingly, the service operation APP receives the model discovery response from the registration function.

[0259] The model discovery response includes information of the model.

[0260] The description of the model discovery process in S706 and S707 can also be referred to the above implementation, which is not repeated here.

[0261] In the open management architecture shown in FIG. 5, on-demand and fast access to network performance, configuration, alarm and MDA analysis data between APP and platform can be supported without implementing complex file and stream management.

[0262] In addition to the model registration method described above, the embodiment of the present application also provides a model registration method, which is applicable to the service-oriented management architecture shown in FIG. 4, and is registration of a model trained offline. Taking a first model trained offline, a first device as a cross-domain management function unit, and a second device as a single-domain management function unit as examples, the model registration method is shown in FIG. 8, and includes the following steps.

[0263] S801. The first device allocates a globally unique identifier for the first model after the first model is successfully registered.

[0264] The first model is obtained offline, for example, the first model can be obtained offline by an operator device, or can be obtained offline by a third-party device, and the present application is not limited in this regard.

[0265] The first device registers the first model, which can complete the registration according to the information of the first model. For example, the first device obtains the information of the offline-trained first model, which can be input by a man-machine, can be downloaded into the first device in the form of a file, and the like, and the present application is not limited in this regard.

[0266] For example, the information of the first model can include at least one of the third identifier of the first model, the inference type of the first model, the version of the first model, the capability of the first model, or the intention target of the first model. The third identifier of the first model can be an identifier allocated by the device that obtains the first model offline, and the specific description of other model information can refer to the related description of the model information in S601 described above, which is not repeated here.

[0267] The process of registering the first model according to the information of the first model by the first device can refer to the related description in S602 described above, which is not repeated here.

[0268] S802. The first device sends the globally unique identifier to the second device. Correspondingly, the second device receives the globally unique identifier from the first device.

[0269] After the first device completes the registration of the offline first model, the first device can allocate a globally unique identifier, such as the second identifier described above, for the first model, which is used to uniquely identify the first model within the management range of the first device, and sends the globally unique identifier to the second device within the management range of the first device, so that the second device knows the identifier of the model that can be deployed on the inference function thereof.

[0270] In a possible implementation, the globally unique identifier can be carried in a model deployment request. That is, the first device can send a model deployment request to the second device, and the second device can receive the model deployment request from the first device. The model deployment request includes the globally unique identifier of the first model, and the model deployment request is used to request deployment of the first model. It should be understood that, since the first model is trained offline, if the second device needs to deploy the first model, the model deployment request can further include the generation parameters of the first model (or the first model directly).

[0271] In this implementation, after the second device performs the deployment of the first model, the second device can optionally send a model deployment response to the first device, and the first device can receive the model deployment response from the second device. The model deployment response is used to indicate whether the deployment of the first model is successful. For example, the cross-domain management function unit can request the single-domain management function unit to create an instance object of the model deployment request, and add the model information (such as model identifier and model parameters) to be deployed in the object. The instance object of the model deployment request can reuse the MLEntityLoadingRequest defined in the existing protocol.

[0272] S803. The second device determines the first model according to the globally unique identifier.

[0273] After receiving the globally unique identifier, the second device can know that it can deploy the model corresponding to the globally unique identifier according to the globally unique identifier.

[0274] In addition, the first device can still provide a model discovery service. Based on a model discovery request of an operator device, the second device, or a third-party device, the first device can determine a model meeting the requirements of an AI / ML consumer from the registered model information, and feed back the information of the corresponding model to the AI / ML consumer. For example, the first device receives a model discovery request, and sends a model discovery response according to the registered model information and the model discovery request. The model discovery request is used to request discovery of a model, and the model discovery response includes the information of the model. The specific implementation of the model discovery can be referred to the description of the model discovery process between the first device and the second device shown in FIG. 6, which will not be described here.

[0275] The model registration method shown in FIG. 8 can implement the traceability and queryability of the model for the registration, management, and discovery of the offline trained model, so as to track and recall the most suitable model version.

[0276] In the above various embodiments, the method and / or steps implemented by the first device can also be implemented by components (e.g., a processor, a chip, a chip system, a circuit, a logic module, or software) available to the first device; the method and / or steps implemented by the second device can also be implemented by components (e.g., a processor, a chip, a chip system, a circuit, a logic module, or software) available to the second device.

[0277] The above mainly introduces the schemes provided in the application. Correspondingly, the application further provides a communication apparatus, which is used to implement various methods in the above method embodiments. The communication apparatus can be the first device in the above method embodiments, or an apparatus containing the first device, or a component (e.g., a chip or a chip system) available to the first device. Alternatively, the communication apparatus can be the second device in the above method embodiments, or an apparatus containing the second device, or a component (e.g., a chip or a chip system) available to the second device.

[0278] In some embodiments, the communication apparatus contains hardware structures and / or software modules corresponding to the implementation of various functions in order to implement the above functions. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of the examples described in the embodiments disclosed herein, the application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is implemented in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0279] The embodiments of the application can divide the functional modules of the communication apparatus according to the above method embodiments, for example, each functional module can be divided according to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or software functional module. It should be noted that the division of modules in the embodiments of the application is illustrative, and is only a logical functional division. There can be another division manner when actually implemented.

[0280] Taking the communication apparatus as the first device or the second device in the above method embodiments, Fig. 9 is a structural schematic diagram of a communication apparatus provided in an embodiment of the application. As shown in Fig. 9, the communication apparatus 900 includes a processing module 901 and a communication module 902. The processing module 901 is used to perform the processing functions of the first device or the second device in the above method embodiments. The communication module 902 is used to perform the communication functions of the first device or the second device in the above method embodiments.

[0281] All the related content of each step involved in the method embodiments can be cited to the function description of the corresponding function module, and will not be repeated here.

[0282] In a possible design, the communication module 902 can include a receiving module and a sending module (not shown in FIG. 9). The sending module and the receiving module are respectively used for implementing the sending function and the receiving function of the communication apparatus 900.

[0283] In a possible design, the communication apparatus 900 can further include a storage module (not shown in FIG. 9), which stores programs or instructions. When the processing module 901 executes the programs or instructions, the communication apparatus 900 can perform the function of the first device or the second device in any of the methods shown in FIGS. 6-8.

[0284] In some embodiments, the processing module 901 involved in the communication apparatus 900 can be implemented by a processor or a processor-related circuit component, and can be a processor or a processing unit; the communication module 902 can be implemented by a transceiver or a transceiver-related circuit component, and can be a transceiver or a receiving unit.

[0285] Exemplarily, FIG. 10 is a structural schematic diagram of another communication apparatus provided by an embodiment of the present application. The communication apparatus can be the first device or the second device in the method embodiments, or can be a chip (system) or other components or assemblies that can be arranged in the first device or the second device. As shown in FIG. 10, the communication apparatus 1000 can include a processor 1001, a bus 1002, a communication interface 1003, and a memory 1004. The processor 1001, the memory 1004, and the communication interface 1003 communicate through the bus 1002. It should be understood that the number of processors and memories in the communication apparatus 1000 is not limited by the present application.

[0286] The bus 1002 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one line is shown in FIG. 10, but it does not mean that there is only one bus or only one type of bus. The bus 1002 can include a path for transmitting information between various components (for example, the memory 1004, the processor 1001, and the communication interface 1003) of the communication apparatus 1000.

[0287] The processor 1001 can include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP), among other processors.

[0288] The memory 1004 can include volatile memory, such as random access memory (RAM), and non-volatile memory, such as read-only memory (ROM), a floppy disk drive, a hard disk drive, or a solid state drive, among others.

[0289] The communication interface 1003 uses a transceiver module, such as but not limited to a network interface card, a transceiver, among others, to enable communications between the communication apparatus 1000 and other devices or communication networks.

[0290] The memory 1004 stores executable program code that the processor 1001 executes to implement the functionality of the first device or the second device, respectively, as described in the foregoing method embodiments. That is, the memory 1004 has instructions stored thereon for performing the above-described methods.

[0291] In yet another aspect, the embodiments of the present application further provide a computer program product containing instructions, which, when executed on a communication apparatus, cause the communication apparatus to perform the method according to any of the embodiments described above.

[0292] In yet another aspect, the embodiments of the present application further provide a computer readable storage medium. The computer readable storage medium stores computer programs or instructions, which, when executed on a communication apparatus, cause the communication apparatus to perform the method according to any of the embodiments described above.

[0293] In yet another aspect, the embodiments of the present application further provide a communication system, which includes a first device and a second device for implementing the method embodiments described above.

[0294] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions according to the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions can 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 instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired (for example, coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.) mode. The computer readable storage medium can be any available medium that can be accessed by a computer or data storage device such as one or more servers, data centers, etc. integrated with one or more media. The available media can be magnetic media (for example, floppy disk, hard disk, magnetic tape), optical media (for example, digital video disc (DVD)), or semiconductor media (for example, SSD), etc.

[0295] Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0296] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0297] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the described device embodiments are merely schematic. The division of the units is merely logical function division. There can be other division manners in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0298] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0299] In addition, each functional unit in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0300] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that make contributions to the prior art, or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or an access network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a random access memory RAM, a magnetic disk or an optical disk, and various media that can store program codes.

[0301] Although the present application is described herein in conjunction with various embodiments, other variations of the disclosed embodiments can be understood and implemented by those skilled in the art with reference to the attached drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "one" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. Some measures described in mutually different dependent claims can be combined and produce good results.

[0302] Although the present application has been described in connection with certain specific features and embodiments thereof, it is to be understood that it is provided as an example to the best of the applicant's knowledge and that various modifications and combinations of the described features and embodiments are possible and are within the spirit and scope of the application. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense, and all such modifications and variations are considered within the scope of the present application as defined by the following claims and their equivalents. Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the claims and their equivalents, the present application can be practiced otherwise than as specifically described.

Claims

1. A model registration method characterized by comprising: The method comprises: The first device receives a model registration request, the model registration request being used for requesting registration of a first model; The first device sends a model registration response, the model registration response being used for indicating whether the first model is successfully registered.

2. The method of claim 1, wherein, The model registration request comprises information of the first model, the information of the first model comprising at least one of a first identifier of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

3. The method of claim 2, wherein, The inference type of the first model comprises at least one of a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

4. The method according to claim 2 or 3, characterized in that, The capability of the first model comprises at least one of a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

5. The method according to any one of claims 2-4, characterized in that, The intention target of the first model comprises at least one of a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

6. The method according to any one of claims 1-5, characterized in that, In a case where the first model is successfully registered, the model registration response comprises a second identifier, the second identifier being a globally unique identifier allocated by the first device for the first model.

7. The method of claim 6, wherein, The method further comprises: The first device sends a model deployment request, the model deployment request comprising the second identifier, the model deployment request being used for requesting deployment of the first model.

8. The method according to any one of claims 1-7, characterized in that, The method is applied to a service-based management architecture, and the first device is a cross-domain management function unit. The first device receives a model registration request, comprising: The first device receives the model registration request from a second device, the second device being a single-domain management function unit. The first device sends a model registration response, comprising: The first device sends the model registration response to the second device.

9. The method according to any one of claims 1-7, characterized in that, The method is applied to an open management architecture, and the first device is a management and orchestration platform. The first device receives a model registration request, comprising: The first device receives the model registration request from a second device, the second device being a service operation application (APP). The first device sends a model registration response, comprising: The first device sends the model registration response to the second device.

10. The method of claim 9, wherein, The first identifier of the first model in the information of the first model comprised in the model registration request is a vendor identifier, and the inference type of the first model in the information of the first model is a type of the service operation APP.

11. The method according to claim 9 or 10, characterized in that, The method further comprises: The first device receives first indication information from the second device, the first indication information being used for indicating that retraining of the first model is allowed; The first device re-trains the first model according to a performance of the first model and the first indication information; The first device sends updated model information to the second device, the updated model information being used for updating the first model.

12. The method of claim 11, wherein, The updated model information comprises an updated storage address of the first model.

13. The method according to any one of claims 1-12, characterized in that, The method further comprises: The first device receives a model discovery request, the model discovery request being used for requesting discovery of a model; The first device sends a model discovery response according to the information of the registered model and the model discovery request, the model discovery response including information of the model.

14. The method of claim 13, wherein, The model discovery request includes an identification of the model and / or a version of the model.

15. A model registration method characterized by comprising: The method includes: The second device sends a model registration request, the model registration request being used to request registration of a first model The second device receives a model registration response, the model registration response being used to indicate whether the first model is successfully registered.

16. The method of claim 15, wherein, The model registration request includes information of the first model, the information of the first model including at least one of a first identification of the first model, an inference type of the first model, a version of the first model, a capability of the first model, or an intention target of the first model.

17. The method of claim 16, wherein, The inference type of the first model includes at least one of a radio access network (RAN) intelligence type, a management data analysis type, or a network data analysis type.

18. The method according to claim 16 or 17, characterized in that The capability of the first model includes at least one of a traffic analysis capability, a coverage analysis capability, or a mobility analysis capability.

19. The method of any one of claims 16-18, wherein, The intention target of the first model includes at least one of a coverage intention target, an energy saving intention target, or a wireless resource optimization intention target.

20. The method of any one of claims 15-19, wherein, In a case where the first model is successfully registered, the model registration response includes a second identification, the second identification being a globally unique identification allocated to the first model by a first device.

21. The method of claim 20, wherein, The method further includes: The second device receives a model deployment request, the model deployment request including the second identification, the model deployment request being used to request deployment of the first model.

22. The method of any one of claims 15-21, wherein, The method is applied to a service-based management architecture, and the second device is a single-domain management function unit. The second device sends a model registration request, including: The second device sends the model registration request to a first device, the first device being a cross-domain management function unit. The second device receives a model registration response, including: The second device receives the model registration response from the first device.

23. The method of any one of claims 15-21, wherein, The method is applied to an open management architecture, and the second device is a service operation application (APP). The second device sends a model registration request, including: The second device sends the model registration request to a first device, the first device being a management and orchestration platform. The second device receives a model registration response, including: The second device receives the model registration response from the first device.

24. The method of claim 23, wherein, The first identification of the first model in the information of the first model included in the model registration request is a vendor identification, and the inference type of the first model in the information of the first model is a type of the service operation APP.

25. The method of claim 23 or 24, wherein, The method further includes: The second device sends first indication information to the first device, the first indication information being used to indicate that retraining of the first model is allowed; The second device receives updated model information from the first device, the updated model information being used to update the first model.

26. The method of claim 25, wherein, The updated model information includes an updated storage address of the first model.

27. The method of any one of claims 15-26, wherein, The method further includes: The second device sends a model discovery request, the model discovery request being used to request discovery of a model; The second device receives a model discovery response, the model discovery response including information of a model.

28. A communications device, characterized by Comprising: a processor; The processor is configured to run a computer program or instructions, so that the method in any one of claims 1-14, or 15-27 is implemented.

29. A communication system, characterized by Comprising: a first device configured to perform the method in any one of claims 1-14, and a second device configured to perform the method in any one of claims 15-27.

30. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program or instructions, when the computer program or instructions are executed by a communication device, the method in any one of claims 1-14, or 15-27 is implemented.

Images