Model management method, communication apparatus, storage medium, and program product

Abstract

The present application belongs to the technical field of communications. Disclosed are a model management method, a communication apparatus, a storage medium, and a program product. In the present application, a terminal receives first configuration information issued by a network device, wherein the first configuration information comprises first condition information; and on this basis, the terminal can send model information to the network device, so as to indicate at least one candidate model meeting the first condition information. In this way, on the basis of the model information, the network device can control the terminal to activate a model meeting the first condition information. It can be seen therefrom that in the embodiments of the present application, by means of configuring first condition information to a terminal, a network device controls the terminal to select and activate an appropriate model meeting a specific condition, realizing finer management of models in the terminal, thus enabling the terminal to obtain better services.

Description

Model management method, communication device, storage medium and program product

[0001] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office on March 29, 2024, with application number 202410388648.4 and application name “Model management method, communication device, storage medium and program product”, all contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the field of communication technology, and in particular to a model management method, a communication device, a storage medium, and a program product. Background Art

[0003] The Third Generation Partnership Project (3GPP) proposed applying artificial intelligence (AI) technology to wireless communication systems to improve network performance and user experience through intelligent data collection and analysis. Applying AI technology in wireless communication systems can include applying AI models or machine learning (ML) models to radio access network devices and / or terminals to implement specific functions.

[0004] Currently, when you want to apply an AI model or ML model in a terminal to implement a specific function, there may be multiple different candidate models that can achieve this function. In this case, how to manage these models so that the terminal can obtain the appropriate model is crucial. Summary of the Invention

[0005] The present application provides a model management method, a communication device, a storage medium, and a program product, which can realize refined management of the model in the terminal, so that the terminal can obtain better service.

[0006] To achieve the above objectives, this application adopts the following technical solutions:

[0007] In a first aspect, a model management method is provided, the method comprising: receiving first configuration information from a network device, the first configuration information comprising first condition information; sending model information to the network device based on the first configuration information, the model information being used to indicate at least one first candidate model that satisfies the first condition information; and receiving activation indication information from the network device, the activation indication information being used to indicate activation of a first model among the at least one first candidate model.

[0008] In the present application, a terminal receives first configuration information sent by a network device, and the first configuration information includes first condition information. Based on this, the terminal can send model information to the network device to indicate at least one candidate model that meets the first condition information. In this way, the network device can control the terminal to activate a model that meets the first condition information based on the model information. It can be seen that in the embodiment of the present application, the network device configures the first condition information to the terminal, controls the terminal to screen and activate appropriate models that meet specific conditions, and realizes more refined management of the models in the terminal, thereby enabling the terminal to obtain better services.

[0009] Optionally, the implementation process of sending model information to the network device based on the first configuration information may include: determining the at least one first candidate model that meets the first condition information from at least one model; and sending the model information to the network device based on the at least one first candidate model.

[0010] In the present application, the terminal can filter a model that meets the first condition information from at least one model based on the first condition information configured by the network device, thereby providing the network device with model information indicating the model that meets the first condition information.

[0011] Optionally, the first condition information includes a model performance threshold; determining the at least one first candidate model that satisfies the first condition information from at least one model includes: obtaining a model performance measurement value of the at least one model; determining a model from the at least one model whose model performance measurement value satisfies the model performance threshold, wherein the at least one first candidate model includes a model whose model performance measurement value satisfies the model performance threshold.

[0012] In the present application, the first condition information configured by the network device may include a model performance threshold. Based on this, the terminal can filter out models whose model performance measurement values ​​meet the model performance threshold from at least one model. It can be seen that in the present application, the terminal filters out models with better model performance through the model performance threshold configured by the network device, and reports relevant model information. In this way, the network device can control the terminal to activate models with better performance based on the model information, so that the terminal can obtain better service.

[0013] Optionally, the first configuration information also includes measurement configuration information, and the measurement configuration information is used to indicate a method for the terminal to measure the model performance measurement value of the at least one model; the implementation method of obtaining the model performance measurement value of the at least one model may include: based on the measurement configuration information, measuring the model performance of the at least one model to obtain the model performance measurement value of the at least one model.

[0014] The measurement configuration information includes the number of test samples and / or the number of model performance measurements.

[0015] Optionally, the measurement configuration information also includes first indication information, and the first indication information is used to indicate a performance measurement scenario, and the performance measurement scenario includes a first scenario or a second scenario, the first scenario is the scenario in which the terminal is currently located, and the second scenario is the scenario corresponding to the first model; on this basis, the implementation process of measuring the model performance of the at least one model based on the measurement configuration information may include: based on the first indication information, obtaining a test sample set corresponding to the first scenario or the second scenario; and using the test sample set to measure the model performance of the at least one model.

[0016] In the present application, when the network device configures a model performance threshold for the terminal, it can also configure measurement configuration information for the terminal. In this way, the terminal can measure the performance of the model according to the measurement method indicated by the measurement configuration information to obtain a model performance measurement value. It can be seen that in the present application, the network device can more finely control the way the terminal measures the model performance measurement value, and the measurement method will affect the direct model performance measurement value. For example, the number of test samples and / or the number of model performance times will directly affect the accuracy of the model performance measurement value and the model performance that can be characterized. In this way, by controlling the number of test samples and / or the number of model performance times, the terminal can be controlled to obtain a more accurate model performance measurement value, thereby improving the accuracy of model screening. For another example, the performance measurement scenario indicated by the first indication information will determine the measurement value of the model performance measurement value obtained in which scenario. In this way, through the first indication information, the terminal can be controlled to screen out models whose performance measurement values ​​in a specific scenario meet the performance threshold. In summary, by configuring the measurement configuration information, the models screened by the terminal can be more finely controlled.

[0017] Optionally, the first condition information includes scene condition information, and the scene condition information is used to indicate a first scene, where the first scene is the scene in which the terminal is currently located; the implementation process of determining the at least one first candidate model that meets the first condition information from the at least one model may include: determining a model suitable for the first scene from the at least one model, and the at least one first candidate model includes a model suitable for the first scene.

[0018] In the present application, the first condition information may include scene condition information, so that the terminal can filter and activate a model suitable for the current scene based on the scene condition information, thereby using the model to obtain better services in the current scene.

[0019] Optionally, the scenario condition information includes at least one of the antenna configuration, number of antenna ports, and carrier frequency of the network device. That is, the scenario condition information in this application may also include some configuration information of the network device. This configuration information determines the network scenario currently in which the terminal is located. Therefore, based on this configuration information, a model suitable for the current scenario can be better screened.

[0020] Optionally, the first configuration information further includes second indication information. When the second indication information includes a first indication value, if the terminal screens out at least one first candidate model that meets the first condition information, the model information is used to indicate the existence of the at least one first candidate model. If the terminal fails to screen out at least one first candidate model that meets the first condition information, the model information may also be used to indicate the absence of the at least one first candidate model.

[0021] Optionally, the first configuration information further includes second indication information. When the second indication information includes a second indication value, the model information includes a model performance measurement value and / or applicable scenario information of the at least one first candidate model.

[0022] In the present application, the terminal can flexibly report model information according to the instructions of the network device. For example, the network device can configure the second indication information including the first indication value for the terminal, and the terminal can send the model information for 1 bit to the network device based on the first indication value to indicate whether there is model information that meets the first condition information. In this way, the amount of data reported by the terminal can be reduced and the communication cost can be reduced. For another example, the network device can configure the second indication information including the second indication value for the terminal, and the terminal can send the model information including the model performance measurement value and / or applicable scenario information of at least one first candidate model to the network device based on the second indication value. In this way, the network device can perceive the performance of the model screened by the terminal and the current scenario of the terminal through the model information, thereby providing a reference for other subsequent network decisions.

[0023] Optionally, the method also includes: sending a model switching request to the network device, the model switching request including at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; receiving model switching indication information from the network device, and the model switching indication information is used to indicate switching to a second model among the at least one second candidate model.

[0024] In this application, the terminal can also actively request the network device to switch models, so that it can switch to other appropriate models in a timely manner.

[0025] In a second aspect, a model management method is provided, which includes: sending model information to a network device, wherein the model information includes a model identifier and a model performance measurement value of at least one first candidate model; and receiving activation indication information from the network device, wherein the activation indication information is used to indicate the activation of a first model among the at least one first candidate model.

[0026] The model information includes a mapping relationship between the model identifier of at least one first candidate model and a model performance measurement value, wherein in the mapping relationship, one model performance measurement value corresponds to the model identifier of one or more first candidate models, or, the model identifier of a first candidate model corresponds to one or more model performance measurement values.

[0027] In this application, a terminal can send model information to a network device. This model information includes a model identifier of at least one first candidate model and a corresponding model performance measurement value. Based on this, the network device can perceive the performance of each model through the model information reported by the terminal, and then use the model identifier in the model information to instruct the terminal to activate the model with better performance. This implements refined model management based on model performance, which can provide better service to the terminal.

[0028] Optionally, before sending model information to the network device, determine the at least one first candidate model; send a model identification request to the network device, the model identification request is used to request at least one model identification; receive N model identifications from the network device, the N model identifications including the model identification of the at least one first candidate model, and N is a positive integer greater than 0.

[0029] Optionally, the method further includes: associating at least one model identifier among the N model identifiers with the at least one first candidate model.

[0030] In the present application, the terminal can apply for a model identifier for at least one determined first candidate model, so as to associate a model identifier with at least one first candidate model. In this way, when the model information is subsequently sent to the network device, the model identifier associated with at least one first candidate model can be carried in the model information, so that the network device can manage the model in the terminal based on the model identifier.

[0031] Optionally, the model identification request includes a number M of requested model identifications, and N is not greater than the M.

[0032] In this application, the model identification request sent by the terminal to the network device may include the number of model identifications requested, and the number of model identifications allocated by the network device to the terminal may not be greater than the number of model identifications requested by the terminal. In this way, when the network device serves a large number of terminals, the situation of insufficient model identifications can be avoided as much as possible, thereby ensuring that the model management process based on model identification can proceed normally.

[0033] Optionally, the method also includes: sending a model switching request to the network device, the model switching request including at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; receiving model switching indication information from the network device, and the model switching indication information is used to indicate switching to a second model among the at least one second candidate model.

[0034] In this application, the terminal can also actively request the network device to switch models, so that it can switch to other appropriate models in a timely manner.

[0035] Optionally, the model information further includes at least one of measurement information of a model performance measurement value of the at least one first candidate model and applicable scenario information of the at least one first candidate model, and the measurement information is used to indicate a measurement method.

[0036] In the present application, the model information may also include measurement information of the model performance measurement value, so as to indicate the measurement method of the model performance measurement value. On this basis, the network device can perceive the accuracy of the model performance measurement value or the scenario in which the model performance measurement value is measured based on the measurement information. In this way, the network device can better screen the model whose performance meets the conditions based on the measurement information and control the terminal to activate the corresponding model. In addition, the model information may also include applicable scenario information of at least one first candidate model. In this way, the network device can also perceive the applicable scenario of the model, thereby controlling the terminal to activate a model that is more suitable for the current scenario and has better performance, so that the terminal can obtain better service.

[0037] Optionally, the at least one first candidate model satisfies first condition information, where the first condition information includes condition information determined by the terminal based on the current scenario or condition. That is, the at least one first candidate model may be a model that satisfies the current scenario or condition and is selected by the terminal itself. In this way, by sending the model information to the network device, the model that satisfies the current scenario or condition can be activated under the control of the network device.

[0038] In a third aspect, a model management method is provided, the method comprising: sending first configuration information to a terminal, the first configuration information including first condition information; receiving model information from the terminal, the model information being used to indicate at least one first candidate model that satisfies the first condition information; and based on the model information, sending activation indication information to the terminal, the activation indication information being used to instruct the terminal to activate a first model among the at least one first candidate model.

[0039] Optionally, the first condition information includes a model performance threshold, and the at least one first candidate model includes a model whose model performance measurement value meets the model performance threshold.

[0040] Optionally, the first configuration information further includes measurement configuration information, where the measurement configuration information is used to indicate a measurement method of the terminal for a model performance measurement value of at least one model.

[0041] Optionally, the measurement configuration information includes the number of test samples and / or the number of model performance measurements.

[0042] Optionally, the measurement configuration information also includes first indication information, where the first indication information is used to indicate a performance measurement scenario, where the performance measurement scenario includes a first scenario or a second scenario, where the first scenario is the scenario in which the terminal is currently located, and the second scenario is the scenario corresponding to the first model.

[0043] Optionally, the first condition information includes scene condition information, where the scene condition information is used to indicate a first scene, where the first scene is the scene in which the terminal is currently located, and the at least one first candidate model includes a model applicable to the first scene.

[0044] Optionally, the scenario condition information includes at least one of an antenna shape, a number of antenna ports, and a carrier frequency of the network device.

[0045] Optionally, the first configuration information further includes second indication information. When the second indication information includes a first indication value, the model information is used to indicate the existence of the at least one first candidate model.

[0046] Optionally, the first configuration information further includes second indication information. When the second indication information includes a second indication value, the model information includes applicable scenario information and / or model performance measurement value of the at least one first candidate model.

[0047] Optionally, the method further includes: receiving a model switching request from the terminal, the model switching request including at least one second model identifier, the at least one second model identifier being used to identify at least one second candidate model that meets the second condition information; and sending model switching indication information to the terminal, the model switching indication information being used to instruct the terminal to switch to a second model among the at least one second candidate model.

[0048] The technical effects obtained in the third aspect are similar to those obtained by the corresponding technical means in the first aspect, and will not be repeated here.

[0049] In a fourth aspect, a model management method is provided, the method comprising: receiving model information from a terminal, the model information comprising a model identifier, a model performance measurement value and applicable scenario information of at least one first candidate model; based on the model information, sending activation indication information to the terminal, the activation indication information being used to indicate activation of a first model among the at least one first candidate model.

[0050] Optionally, before receiving the model information from the terminal, a model identification request is received from the terminal, wherein the model identification request is used to request at least one model identification; N model identifications are sent to the terminal, wherein the N model identifications include the model identification of the at least one first candidate model, and N is a positive integer greater than 0.

[0051] Optionally, the model identification request includes a number M of requested model identifications, and N is not greater than the M.

[0052] Optionally, the method further includes: receiving a model switching request from the terminal, the model switching request including at least one second model identifier, the at least one second model identifier being used to identify at least one second candidate model that meets the second condition information; and sending model switching indication information to the terminal, the model switching indication information being used to instruct the terminal to switch to a second model among the at least one second candidate model.

[0053] Optionally, the model information further includes at least one of measurement information of a model performance measurement value of the at least one first candidate model and applicable scenario information of the at least one first candidate model, and the measurement information is used to indicate a measurement method.

[0054] Optionally, the at least one first candidate model satisfies first condition information, where the first condition information includes condition information determined by the terminal based on a current scenario or condition.

[0055] Optionally, the model information includes a mapping relationship between the model identifier of at least one first candidate model and a model performance measurement value, wherein, in the mapping relationship, one model performance measurement value corresponds to the model identifier of one or more first candidate models, or, the model identifier of a first candidate model corresponds to one or more model performance measurement values.

[0056] The technical effects obtained in the fourth aspect are similar to those obtained by the corresponding technical means in the second aspect, and will not be repeated here.

[0057] In a fifth aspect, a communication device is provided, comprising at least one module, wherein the at least one module is used to execute the model management method described in the first aspect, the second aspect, the third aspect, or the fourth aspect.

[0058] In a sixth aspect, a communication device is provided, comprising a processor configured to execute at least one program instruction or code stored in a memory to implement the model management method described in the first aspect, the second aspect, the third aspect, or the fourth aspect.

[0059] In a seventh aspect, a communication system is provided, comprising a first communication device and a second communication device. The first communication device is configured to execute the model management method described in the first aspect, and the second communication device is configured to execute the model management method described in the third aspect; or the first communication device is configured to execute the model management method described in the second aspect, and the second communication device is configured to execute the model management method described in the fourth aspect.

[0060] In an eighth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer device, the computer device executes the model management method described in the first aspect, the second aspect, the third aspect, or the fourth aspect.

[0061] In a ninth aspect, a computer program product comprising instructions is provided, which, when executed on a computer device, enables the computer device to execute the model management method described in the first aspect, the second aspect, the third aspect, or the fourth aspect.

[0062] The technical effects obtained in the above-mentioned fifth to ninth aspects are similar to the technical effects obtained by the corresponding technical means in the first and second aspects, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] FIG1 is an architecture diagram of a communication system provided in an embodiment of the present application;

[0064] FIG2 is a flow chart of a model management method provided in an embodiment of the present application;

[0065] FIG3 is an exemplary flow chart of a model management method provided in an embodiment of the present application;

[0066] FIG4 is a flow chart of another model management method provided in an embodiment of the present application;

[0067] FIG5 is an exemplary flow chart of another model management method provided in an embodiment of the present application;

[0068] FIG6 is a schematic diagram of an O-RAN system provided in an embodiment of the present application;

[0069] FIG7 is an exemplary flowchart of a model management method applied in an O-RAN system provided in an embodiment of the present application;

[0070] FIG8 is an exemplary flowchart of another model management method applied in an O-RAN system provided in an embodiment of the present application;

[0071] FIG9 is a schematic structural diagram of a first communication device provided in an embodiment of the present application;

[0072] FIG10 is a schematic structural diagram of a second communication device provided in an embodiment of the present application;

[0073] FIG11 is a schematic structural diagram of a third communication device provided in an embodiment of the present application;

[0074] FIG12 is a schematic structural diagram of a fourth communication device provided in an embodiment of the present application;

[0075] FIG13 is a schematic structural diagram of the fifth communication device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0076] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0077] Before explaining the embodiments of the present application in detail, the application scenarios involved in the embodiments of the present application are first introduced.

[0078] 3GPP has proposed applying AI technology to wireless communication systems, improving network performance and user experience through intelligent data collection and analysis. Currently, the application architecture for AI technology in wireless communication systems includes a data collection entity, a model training entity, a management entity, a model storage entity, and a model inference entity. The data collection entity collects data from network devices or terminals, which can be used for model training and inference. The model training entity trains models using training data provided by the data collection entity. The model storage entity stores trained or updated models. The management entity controls the model storage entity's transmission of models to the model inference entity. Furthermore, the management entity controls the model inference entity's activation, deactivation, selection, switching, or rollback of models. The model inference entity uses the activated model to perform inference based on data provided by the data collection entity, generating inference results. These inference results can be used to guide network policy adjustments, thereby improving network performance and user experience.

[0079] Currently, AI-based use cases in wireless communication systems primarily include network energy conservation, load balancing, mobility optimization, enhanced channel state information-reference signal (CSI-RS) feedback, enhanced beam scanning, and enhanced positioning. Depending on the use case, AI models or machine learning (ML) models can be deployed on different devices to implement the corresponding functions. Hereinafter, AI models or ML models are referred to as models.

[0080] For example, in some possible scenarios, the model can be deployed independently in network devices. The network devices use the deployed model to perform inference based on input data and execute corresponding network policies based on the inference results. For example, in the use case of network energy conservation, a model can be deployed in the base station. Based on this, the base station collects load, energy consumption, energy efficiency information for itself and neighboring cells, as well as terminal trajectory information and measurement results. The base station uses the model and the above collected data to infer its own load and, based on the inference results and performance indicators, executes appropriate energy conservation policies to achieve energy conservation results.

[0081] In some possible scenarios, the model can also be deployed in both the network device and the terminal. In this case, the models deployed in the network device and the terminal can be used to output different types of inference results to cooperate in implementing specific functions. For example, in the use case of CSI-RS feedback enhancement, a trained encoder can be deployed in the terminal, and a corresponding trained decoder can be deployed in the network device. Based on this, the terminal can use the encoder to compress the measured channel matrix and feed the compressed result back to the network device to reduce signaling transmission costs; the network device can use the decoder to restore the compressed result to obtain the original channel matrix to ensure the accuracy of the obtained channel matrix.

[0082] In some possible scenarios, the model can also be deployed independently in the terminal. In this case, the terminal can use the deployed model to perform inference based on the input data and feed the inference results back to the network device so that the network device can implement the corresponding policy based on the inference results. For example, in the use case of CSI-RS feedback enhancement, a model for CSI prediction can be deployed in the terminal. The terminal can use this model to perform channel prediction based on CSI measurement results over a period of time and report the prediction results to the network device so that the network device can implement subsequent downlink configuration based on the prediction results.

[0083] In AI-based use cases such as these, when deploying a model in a terminal to implement a specific function, there may be multiple models that can implement this function. For example, when a terminal needs to use a model to perform CSI prediction, there may be multiple models that can output the CSI prediction results. However, different models may achieve different results when implementing this function. In this case, how to control the terminal to select the appropriate model to implement the corresponding function is crucial. Based on this, an embodiment of the present application provides a model management method, in which a terminal receives first configuration information issued by a network device, and the first configuration information includes first condition information. Based on this, the terminal can send model information to the network device to indicate at least one candidate model that meets the first condition information. In this way, the network device can control the terminal to activate a model that meets the first condition information based on the model information. It can be seen that in the embodiment of the present application, the network device configures the first condition information to the terminal, controls the terminal to screen and activate appropriate models that meet specific conditions, and thus achieves more refined management of the models in the terminal, thereby enabling the terminal to obtain better services.

[0084] Next, the network system to which the model management method provided in the embodiment of the present application is applicable is introduced.

[0085] Figure 1 is a schematic diagram of the architecture of a communication system 10 used in an embodiment of the present application. As shown in Figure 1 , the communication system includes a radio access network (RAN) 100. RAN 100 includes at least one RAN node (e.g., 110a and 110b in Figure 1 , collectively referred to as 110) and may also include at least one terminal (e.g., 120a-120j in Figure 1 , collectively referred to as 120). RAN 100 may also include other RAN nodes, such as wireless relay devices and / or wireless backhaul devices (not shown in Figure 1 ). Terminal 120 is wirelessly connected to RAN node 110. Terminals and RAN nodes may be connected to each other via wired or wireless connections. Communication system 10 may also include a core network. RAN node 110 is wirelessly or wiredly connected to the core network. Core network devices in the core network and RAN node 110 in RAN 100 may be independent and distinct physical devices, or they may be a single physical device that integrates the logical functions of the core network devices and the logical functions of the RAN nodes. Communication system 10 may also include the Internet.

[0086] RAN100 can be an evolved universal terrestrial radio access (E-UTRA) system, a new radio (NR) system, a sixth generation (6G) communication system, or a future wireless access system defined in the 3rd Generation Partnership Project (3GPP). It can also be a WiFi system. RAN100 can also include two or more of the above different wireless access systems. RAN100 can also be an open RAN (O-RAN).

[0087] A RAN node, also known as a radio access network device, RAN entity, or access node, helps terminals access a communication system wirelessly. In one application scenario, a RAN node can be a base station, an evolved NodeB (eNodeB), a transmission reception point (TRP), a next-generation NodeB (gNB) in a fifth-generation (5G) mobile communication system, a next-generation base station in a 6G mobile communication system, a base station in a future mobile communication system, or an access node in a WiFi system. A RAN node can be a macro base station (such as 110a in Figure 1), a micro base station, an indoor station (such as 110b in Figure 1), a relay node, or a donor node.

[0088] In another application scenario, multiple RAN nodes can collaborate to help terminals achieve wireless access, with different RAN nodes implementing portions of the base station's functions. For example, a RAN node can be a centralized unit (CU), a distributed unit (DU), or a radio unit (RU). The CU implements the base station's radio resource control protocol and packet data convergence protocol (PDCP) functions, as well as the service data adaptation protocol (SDAP) functions. The DU implements the base station's radio link control layer and medium access control (MAC) layer functions, as well as some or all of the physical layer functions. For detailed descriptions of each of the above protocol layers, please refer to the relevant 3GPP technical specifications. The RU can be used to implement the transmission and reception of radio frequency signals. The CU and DU can be two independent RAN nodes, or they can be integrated into the same RAN node, such as in a baseband unit (BBU). The RU can be included in radio frequency equipment, such as a remote radio unit (RRU) or an active antenna unit (AAU). The CU can be further divided into two types of RAN nodes: CU-control plane and CU-user plane.

[0089] RAN nodes may have different names in different systems. For example, in an O-RAN system, the CU may be called an open CU (O-CU), the DU may be called an open DU (O-DU), and the RU may be called an open RU (O-RU). Furthermore, in an O-RAN system, a RAN node may also be a radio access network intelligent controller (RIC). The RIC collects network information and performs necessary optimization operations. The RIC controls the O-CU and O-DU by communicating with them via the E2 interface.

[0090] It is worth noting that the RAN node in the embodiments of the present application can be implemented as a software module, a hardware module, or a combination of software and hardware modules. For example, the RAN node can be a server loaded with the corresponding software module. The embodiments of the present application do not limit the specific technology and device form used by the RAN node. For ease of description, the following description uses a base station as an example of a RAN node.

[0091] A terminal is a device with wireless transceiver capabilities that can send signals to a base station or receive signals from a base station. A terminal may also be referred to as a terminal device, user equipment (UE), mobile station, mobile terminal, etc. Terminals can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), Internet of Things (IoT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grid, smart furniture, smart office, smart wearable, smart transportation, smart city, etc. A terminal can be a mobile phone, tablet computer, computer with wireless transceiver capabilities, wearable device, vehicle, airplane, ship, robot, robotic arm, smart home device, etc. The embodiments of this application do not limit the specific technology and specific device form adopted by the terminal.

[0092] Base stations and terminals can be fixed or mobile. They can be deployed on land, indoors or outdoors, handheld or vehicle-mounted; on water; or on aircraft, balloons, and satellites. The embodiments of this application do not limit the application scenarios of base stations and terminals.

[0093] The roles of base stations and terminals can be relative. For example, the helicopter or drone 120i in Figure 1 can be configured as a mobile base station. To terminals 120j accessing the wireless access network 100 via 120i, terminal 120i is a base station. However, to base station 110a, 120i is a terminal, meaning that communication between 110a and 120i occurs via a wireless air interface protocol. Of course, communication between 110a and 120i can also occur via a base station-to-base station interface protocol. In this case, 120i is also a base station relative to 110a. Therefore, base stations and terminals can be collectively referred to as communication devices. 110a and 110b in Figure 1 can be referred to as communication devices with base station functionality, while 120a-120j in Figure 1 can be referred to as communication devices with terminal functionality.

[0094] Communication between base stations and terminals, between base stations, and between terminals can be carried out through authorized spectrum, unauthorized spectrum, or both; communication can be carried out through spectrum below 6 gigahertz (GHz), spectrum above 6 GHz, or spectrum below 6 GHz and spectrum above 6 GHz. The embodiments of the present application do not limit the spectrum resources used for wireless communication.

[0095] In the embodiments of the present application, the functions of the base station may also be performed by a module (such as a chip) in the base station, or by a control subsystem that includes the base station functions. The control subsystem that includes the base station functions here may be a control center in the above-mentioned application scenarios such as smart grid, industrial control, smart transportation, and smart city. The functions of the terminal may also be performed by a module (such as a chip or modem) in the terminal, or by a device that includes the terminal functions.

[0096] Next, the model management method provided in the embodiment of the present application is explained in detail.

[0097] Figure 2 is a flow chart of a model management method provided in an embodiment of the present application. This method can be applied to a wireless communication system including a terminal and a network device, wherein the network device can be a RAN node described in the aforementioned embodiment, for example, a base station or a RIC in an O-RAN architecture, and the terminal can be any of the various types of terminals described in the aforementioned embodiment. Referring to Figure 2, the process may include the following steps:

[0098] S201: A network device sends first configuration information to a terminal, where the first configuration information includes first condition information.

[0099] In an embodiment of the present application, a terminal may need to use a model to implement one or more specific functions, and for any specific function, there may be one or more models that can implement the function. In this case, the network device can send first configuration information to the terminal to enable the terminal to determine a candidate model that meets the conditions from one or more models. The first configuration information can be determined by the network device based on the expected model effect. Optionally, the first configuration information can also be determined by the network device based on the currently detected scene of the terminal.

[0100] Specifically, the first configuration information may include first condition information, where the first condition information is used to indicate specific conditions or scenarios that the model needs to meet when the terminal screens the model.

[0101] In some embodiments, the first condition information may include a model performance threshold, which indicates the model performance value that the network expects the model reported by the terminal to achieve, and is used to compare with the model performance measurement value to achieve model screening.

[0102] As an example, the first condition information may include model performance thresholds corresponding to one or more use cases. The model performance thresholds corresponding to any two different use cases may be different or the same. The use case here may refer to the AI-based use case introduced above, and the model performance threshold corresponding to the use case refers to the performance threshold of the AI ​​model or ML model applicable to the use case. In addition, the different model performance thresholds corresponding to any two different use cases may mean that the two model performance thresholds are thresholds of different types of performance indicators, or may mean that the two model performance thresholds are thresholds of the same type of performance indicators, but with different numerical values.

[0103] For example, the first condition information may include a model performance threshold corresponding to the CSI-RS feedback enhancement use case, a model performance threshold corresponding to the beam scanning enhancement use case, and a model performance threshold corresponding to the positioning enhancement use case. The model performance threshold corresponding to the CSI-RS feedback enhancement use case may be a squared generalized cosine similarity (SGCS) threshold for evaluating the similarity between the predicted result output by the model and the true result, or it may be a normalized mean square error (NMSE) threshold for evaluating the difference between the predicted result output by the model and the true result. The model performance threshold corresponding to the beam scanning enhancement use case may be a top-1 accuracy threshold, where the so-called top-1 accuracy refers to the accuracy of the corresponding predicted probability ranking first and the actual category. In the beam scanning enhancement use case, the top-1 accuracy may specifically refer to the ratio of the transmitting beam that results in the maximum layer-1 reference signal receiving power (L1-RSRP) among all transmitting beams and receiving beams, that is, the proportion of the transmitting beam with the maximum predicted probability leading to the maximum L1-RSRP. The model performance threshold corresponding to the positioning enhancement use case can be a threshold used to evaluate the accuracy of the positioning results output by the model, for example, it can be a threshold corresponding to the percentile of the cumulative distribution function (CDF) of the horizontal accuracy of the positioning results output by the model.

[0104] As another example, the first condition information may also include a model performance threshold, which may be a performance threshold applicable to various use cases.

[0105] Optionally, since the model performance threshold is used to compare with the model performance measurement value to implement model screening, and the model performance measurement value may be affected by the model performance measurement method, based on this, when the first condition information includes the model performance threshold, the first configuration information may also include measurement configuration information, which is used to indicate the terminal's measurement method for the model performance measurement value of at least one model. In other words, the network device may also configure the model performance measurement method for the terminal.

[0106] Exemplarily, the measurement configuration information may include the number of test samples and / or the number of model performance measurements.

[0107] Among them, the number of test samples refers to the number of samples used to measure the performance of the model. Specifically, the number of test samples here may refer to the number of samples used to perform a performance measurement of the model once, or it may also refer to the total number of samples used to perform multiple performance measurements of the model. The number of model performance measurements refers to the number of times the performance of the model is measured. It should be noted that configuring the number of test samples and / or the number of model performance measurements so that the terminal uses the same measurement method when measuring the performance of different models can improve the accuracy of the model performance measurement value.

[0108] Optionally, the measurement configuration information may further include first indication information, where the first indication information may be used to indicate a performance measurement scenario, wherein the performance measurement scenario may refer to a scenario in which the terminal measures model performance.

[0109] Exemplarily, the performance measurement scenario indicated by the first indication information may include a first scenario, where the first scenario is the scenario the terminal is currently in. In this case, the first indication information may instruct the terminal to measure the performance of the model in the current scenario.

[0110] Alternatively, the performance measurement scenario indicated by the first indication information may include a second scenario, and the second scenario may include scenarios corresponding to one or more models. For example, the second scenario may include a scenario corresponding to the first model. In this case, the first indication information may instruct the terminal to measure the performance of the first model in the scenario corresponding to the first model. When a model corresponds to a scenario, it can be understood that the model is trained for the scenario, or it can also be understood that the model is trained using a training data set for the scenario.

[0111] Alternatively, the performance measurement scenario indicated by the first indication information may also be a non-specific scenario or a multiple scenario, where the non-specific scenario refers to not distinguishing the measurement scenario, and the multiple scenario may refer to various different scenarios. In this case, the first indication information may indicate the performance of the terminal measurement model in a non-specific scenario or in multiple scenarios.

[0112] It should be noted that when the network device indicates a performance measurement scenario by configuring the first indication information, the model performance measurement value obtained by the terminal subsequently measuring the model performance based on the performance measurement scenario will also be the measurement value under the performance measurement scenario. Based on this, in order to ensure the accuracy of the screening, the model performance threshold used for comparison with the measurement value under the performance measurement scenario can also be the model performance threshold corresponding to the performance measurement scenario. In this case, the performance measurement scenario not only refers to the scenario in which the terminal measures the model performance, but also indirectly indicates the scenario to which the aforementioned configured model performance threshold applies.

[0113] For example, still taking the example where the performance test scenario indicated by the first indication information includes the first scenario, the first indication information may instruct the terminal to measure the performance of the model in the scenario currently located by the terminal, thereby obtaining a model performance measurement value in the current scenario. Correspondingly, the model performance threshold configured by the network device is also the model performance threshold in the scenario currently located by the terminal, and is used for comparison with the model performance measurement value in the current scenario.

[0114] In some other embodiments, the first condition information may include scenario condition information, where the scenario condition information is used to instruct the terminal to filter models that are applicable to the scenario indicated by the scenario condition information.

[0115] Exemplarily, the scene condition information may be used to indicate a first scene, that is, to indicate the scene the terminal is currently in. In this case, the scene condition information may instruct the terminal to filter a model that is applicable to the scene in which the terminal is currently located.

[0116] Specifically, in a possible implementation, the scene condition information may include a third indication value, where the third indication value is used to indicate the current scene. For example, the third indication value may be 1.

[0117] It should be noted that the scene in which the terminal is currently located may include the terminal-side scene. Among them, the terminal-side scene of the terminal may refer to the scene represented by the current terminal-side data of the terminal, and the terminal-side data may include the terminal's moving speed, moving direction, geographical location, software and hardware operation information, measured cell signal quality, channel state quality, etc. For example, the terminal-side scene of the terminal may be a high-speed scene or a low-speed scene represented by the current moving speed of the terminal. For another example, the terminal-side scene of the terminal may be a high-load scene or a low-load scene represented by the current hardware load information of the terminal. Since the terminal-side scene of the terminal is represented by the terminal-side data, the network device may not be able to obtain the terminal-side data of the terminal. Therefore, the third indication value can be used to instruct the terminal to filter the model applicable to the terminal-side scene represented by its own terminal-side data.

[0118] Optionally, the scenario in which the terminal is currently located may further include the network scenario in which the terminal is located. The network scenario in which the terminal is located may refer to the network environment in which the terminal is currently located, and the network scenario may be characterized by network data obtained by the network device. Based on this, the scenario condition information may also include specific network data obtained by the network device that can characterize the network scenario in which the terminal is located. For example, the scenario condition information may include at least one of the antenna shape, number of antenna ports, and carrier frequency of the network device. On this basis, through the network data included in the scenario condition information and the third indication value, the terminal can be instructed to screen a model suitable for the terminal-side scenario and network scenario in which it is currently located.

[0119] It is worth noting that the first condition information may include one of the model performance threshold and scene condition information introduced above, or may also include the model performance threshold and scene condition information at the same time.

[0120] Optionally, the first configuration information may further include second indication information, where the second indication information may be used to indicate the content of the model information sent by the terminal.

[0121] Exemplarily, the second indication information may include a first indication value, where the first indication value instructs the terminal to send model information indicating whether there is at least one first candidate model that meets the first condition information. For example, the first indication value may be 0.

[0122] Alternatively, the second indication information may include a second indication value, which indicates that the terminal sends model information including applicable scenario information and / or model performance measurement value of at least one first candidate model that meets the first condition information. For example, the second indication value may be 1.

[0123] S202: The terminal sends model information to the network device based on the first configuration information, where the model information is used to indicate at least one first candidate model that meets the first condition information.

[0124] After receiving the first configuration information, the terminal may determine at least one first candidate model that meets the first condition information from at least one model available to the terminal based on the first configuration information, and then send model information to the network device based on the at least one first candidate model.

[0125] According to different first configuration information, the terminal may adopt different implementation methods to determine at least one first candidate model that meets the first condition information.

[0126] Method 1: If the first condition information includes a model performance threshold, the terminal can obtain the model performance measurement value of at least one model; determine the model whose model performance measurement value meets the model performance threshold from the at least one model, and these models that meet the model performance threshold are the first candidate models screened out.

[0127] In the first example, if the first condition information includes model performance thresholds corresponding to one or more use cases, taking any use case as an example, for the convenience of description, the use case is referred to as the first use case. The terminal can obtain the model performance measurement value of the model corresponding to the first use case in at least one model, and compare the obtained model performance measurement value with the model performance threshold corresponding to the first use case to determine whether there is a first candidate model that meets the model performance threshold corresponding to the first use case.

[0128] In one possible implementation, the terminal may obtain the model performance measurement value of the model corresponding to the first use case from another device. For example, an over-the-top (OTT) server may store model performance measurement values ​​of models corresponding to different use cases. Based on this, the terminal may obtain the model performance measurement value of the model corresponding to the first use case from the OTT server.

[0129] In another possible implementation, the terminal may measure the performance of the model corresponding to the first use case, thereby obtaining a model performance measurement value.

[0130] If the first configuration information also includes measurement configuration information, the terminal may measure the model performance of the model corresponding to the first use case based on the measurement configuration information to obtain a model performance measurement value of the model corresponding to the first use case.

[0131] Specifically, when the measurement configuration information includes the number of test samples, the terminal may obtain a corresponding number of test samples and use the obtained test samples to measure the performance of the model corresponding to the first use case. For example, if the number of test samples included in the measurement configuration information is K, the terminal may obtain K test samples and use the K test samples to measure the performance of the model corresponding to the first use case.

[0132] Optionally, if the measurement configuration information includes the number of model performance measurements, the terminal may use test samples to perform a corresponding number of performance measurements on the model corresponding to the first use case. For example, if the number of model performance measurements is T, the terminal may use test samples to perform T performance measurements on the model corresponding to the first use case. It should be noted that the test samples used for each performance measurement may be different. Furthermore, if the measurement configuration information also includes the number of test samples, the number of test samples used for each performance measurement is determined by the number of test samples in the measurement configuration information. For example, if the number of test samples K in the measurement configuration information is the number of test samples required for one performance measurement, the terminal may use K test samples each time to perform performance measurements on the model corresponding to the first use case, for a total of T measurements. If the number of test samples K in the measurement configuration information is the total number of test samples required to measure model performance, then if the number of model performance measurements is T, the number of test samples required for each performance measurement is K / T. Furthermore, when the number of model performance measurements is greater than 1, that is, when the terminal performs multiple performance measurements on the model, the terminal may determine the model performance measurement value for the model based on the results of the multiple measurements. For example, the terminal may use the average of multiple measurement results as the model performance measurement value, or use the best measurement result among the multiple measurement results as the model performance measurement value, or use the worst measurement result among the multiple measurement results as the model performance measurement value.

[0133] Optionally, when the measurement configuration information further includes first indication information, the terminal may further measure the model performance of the model corresponding to the first use case in the performance measurement scenario based on the performance measurement scenario indicated by the first indication information.

[0134] Specifically, as described in S201 above, the performance measurement scenario indicated by the first indication information may include a first scenario, a second scenario, or multiple scenarios. Based on this, the terminal may obtain a test sample set corresponding to the first scenario, the second scenario, or the multiple scenarios indicated by the first indication information; and use the obtained test sample set to measure the model performance of the model corresponding to the first use case.

[0135] It should be noted that the performance measurement scenario refers to the scenario in which the terminal measures the model performance of the model corresponding to the first use case. In other words, the first indication information indicates the scenario in which the terminal measures the model performance of the model. Based on this, to ensure measurement accuracy, when measuring the model performance of the model in a certain scenario, the terminal can obtain a test sample set corresponding to the corresponding scenario to measure the model.

[0136] Exemplarily, when the performance measurement scenario indicated by the first indication information is the first scenario, the terminal may obtain a first test sample set corresponding to the scenario in which it is currently located, and use the test samples in the first test sample set to perform performance measurement on the model.

[0137] The terminal may collect data of test samples that can serve as a model corresponding to the first use case in the current scenario, thereby obtaining a first test sample set.

[0138] When the performance measurement scenario indicated by the first indication information is the second scenario, the terminal may obtain a second test sample set corresponding to the second scenario, and use the second test sample set to measure the performance of the model corresponding to the second scenario.

[0139] For example, if the second scenario includes a high-speed scenario, the terminal may obtain a test sample set corresponding to the high-speed scenario, and use the test sample set to perform performance measurement on the model corresponding to the first use case under the high-speed scenario.

[0140] When the performance measurement scenario indicated by the first indication information is a non-specific scenario or multiple scenarios, the terminal can obtain a comprehensive test sample set containing test samples in different scenarios, and use the comprehensive test sample set to perform performance measurement on the model corresponding to the first use case. Alternatively, the terminal can also obtain its own currently stored test sample set to perform performance measurement on the model corresponding to the first use case. Alternatively, the terminal can obtain test sample sets corresponding to different scenarios, and use each test sample set to measure the performance of the model corresponding to the first use case, respectively, to obtain the model performance measurement value corresponding to each test sample set, and then use the average of the model performance measurement values ​​corresponding to the multiple test sample sets as the model performance measurement value of the model corresponding to the first use case.

[0141] Optionally, if the first configuration information does not include measurement configuration information, the first terminal may also measure the model performance of the model corresponding to the first use case based on a preset measurement method.

[0142] It should be noted that when the terminal obtains the model performance measurement value of the model corresponding to the first use case through any of the methods introduced above, the obtained model performance measurement value and the corresponding model performance threshold are both values ​​of the same performance indicator.

[0143] For example, the first use case is a CSI compression use case in a CSI-RS feedback enhancement use case, and the corresponding model performance threshold is an SGCS threshold. Accordingly, the model performance measurement value of the model corresponding to the first use case obtained by the terminal is an SGCS measurement value.

[0144] For each model corresponding to each use case included in at least one model, the terminal can refer to the method introduced above to obtain the model performance measurement value of each model corresponding to the corresponding use case, and then compare the model performance measurement value of each model corresponding to each use case with the model performance threshold corresponding to the corresponding use case, so as to determine at least one first candidate model that meets the model performance threshold.

[0145] It should be noted that, depending on the indicator type of the model performance threshold, the model performance measurement value meeting the model performance threshold may mean that the model performance measurement value is greater than or not less than the model performance threshold, or it may also mean that the model performance measurement value is less than or not greater than the model performance threshold.

[0146] For example, if the first use case is a CSI compression use case in a CSI-RS feedback enhancement use case, and the model performance threshold corresponding to the first use case is the NMSE threshold, then the NMSE measurement value satisfying the NMSE threshold may mean that the NMSE measurement value is not greater than the NMSE threshold. Alternatively, if the first use case is a beam management use case, and the model performance threshold corresponding to the first use case is the Top-1 accuracy threshold, then the Top-1 accuracy measurement value satisfying the Top-1 accuracy threshold may mean that the Top-1 accuracy measurement value is not less than the Top-1 accuracy threshold.

[0147] In the second example, if the first condition information includes a model performance threshold applicable to models corresponding to various different use cases, the terminal can refer to the method for obtaining model performance measurement values ​​described above to obtain the model performance measurement value of at least one model available to itself, and compare the obtained model performance measurement value with the model performance threshold to determine at least one first candidate model that meets the model performance threshold.

[0148] Mode 2: If the first condition information includes scene condition information, the terminal may determine a model suitable for the scene indicated by the scene condition information from at least one model available to the terminal, and these models are the first candidate models.

[0149] Exemplarily, the scene condition information may be used to indicate a first scene, where the first scene is the scene that the terminal is currently in. In this case, the terminal may determine a model suitable for the current scene from at least one model.

[0150] As described in S201 above, the scenario condition information may include a third indicator value indicating the first scenario, and may also include network data sent by the network device. Therefore, if the scenario condition information includes the third indicator value, the terminal may obtain its own terminal-side data and / or network data under the guidance of the third indicator value, and then obtain a model applicable to the terminal-side scenario indicated by the terminal-side data and / or the network scenario indicated by the network data.

[0151] Specifically, the mapping relationship between the model and the applicable scenario information can be stored in the terminal. The applicable scenario information corresponding to each model can indicate the end-side scenario and / or network scenario to which the model is applicable. Based on this, in a possible implementation method, the applicable scenario information may include an end-side scenario label and / or a network scenario label. In this case, the terminal can collect the current end-side data. If the current end-side data is collected, the current end-side scene label is determined based on the collected end-side data, and the model corresponding to the current end-side scene label is determined from the above mapping relationship. At this time, the model corresponding to the current end-side scene label is the model applicable to the current scene.

[0152] Optionally, if the scene condition information also includes network data sent by the network device, the terminal can also determine the current network scene label based on the network data sent by the network device, and determine the model that corresponds to both the current end-side scene label and the network scene label from the above mapping relationship. At this time, the model that corresponds to both the current end-side scene label and the network scene label is the model suitable for the current scene.

[0153] In another possible implementation, the applicable scenario information corresponding to the model may include an end-side scenario label and / or network data. Based on this, after the terminal collects the current end-side data, it can determine the current end-side scenario label based on the end-side data. Afterwards, the model corresponding to the current end-side scenario label is determined based on the mapping relationship between the stored model and the applicable scenario information. In the case where the scene condition information also includes network data sent by the network device, the terminal can determine the model corresponding to the network data and the current end-side scenario label based on the mapping relationship, thereby obtaining a first candidate model.

[0154] The above mainly introduces the implementation method for the terminal to determine at least one first candidate model when the first condition information includes a model performance threshold or scenario condition information. Optionally, when the first condition information includes both a model performance threshold and scenario condition information, the terminal can refer to the above-mentioned method 1 and method 2 to determine a model that satisfies both the model performance threshold and the scenario condition information from its available models, thereby obtaining at least one first candidate model.

[0155] Of course, in some possible situations, there may not be a model that meets the first condition information among the at least one model available to the terminal. In this case, the terminal will not be able to screen out at least one first candidate model through the method introduced above.

[0156] After performing model screening using the above method, the terminal may send model information to the network device based on the model screening result.

[0157] In one possible scenario, the first configuration information sent by the network device includes second instruction information, and the second instruction information is used to instruct the terminal to send the content of the model information. Based on this, the terminal can send the model information to the network device based on the model screening result and the second instruction information.

[0158] In one example, the second indication information includes a first indication value, which is used to indicate whether the terminal reports whether there is at least one first candidate model that meets the first condition information. Based on this, the terminal can send model information indicating whether there is at least one first candidate model to the network device based on the model screening result.

[0159] For example, the model information may be one-bit indication information. If the model screening result is that at least one first candidate model is not screened out, the value of the indication information is 0, thereby indicating that there is no at least one first candidate model that meets the first condition information. If the model screening result is that at least one first candidate model is screened out, the value of the indication information is 1, indicating that there is at least one first candidate model that meets the first condition information.

[0160] For another example, when the model screening result is that at least one first candidate model is screened out, the model information may include the model structure of each first candidate model, the type of input data, the type of output result, or the size of the training data set used. When the model screening result is that at least one first candidate model is not screened out, the model information may include indication information indicating that at least one first candidate model does not exist. Alternatively, in this case, the terminal may not report the model information to the network device.

[0161] In another example, the second indication information includes a second indication value, which is used to instruct the terminal to report the model performance measurement value and / or applicable scenario information of at least one first candidate model. Based on this, if the terminal determines at least one first candidate model, the terminal can send model information containing the model performance measurement value and / or applicable scenario information of the at least one candidate model to the network device. If the terminal fails to screen out at least one first candidate model, the terminal may not send the model information.

[0162] For example, the model information sent by the terminal may be as shown in Table 1, which may include the model performance measurement value and applicable scenario information corresponding to each first candidate model. As shown in Table 1, the applicable scenario information of the first model indicates that the model is a generalized model applicable to the public land mobile network (PLMN), and the corresponding model performance measurement value is a performance index value of 1, wherein the generalized model applicable to the PLMN refers to a model applicable to all base stations within the PLMN. The applicable scenario information of the second model indicates that the model is a generalized model applicable within the station, and the corresponding model performance measurement value is a performance index value of 2, wherein the generalized model applicable within the station refers to a model applicable within all cells provided by the base station. The applicable scenario information of the third model indicates that the model is a model applicable to the high-speed scenario corresponding to use case 1, and the corresponding model performance measurement value is a performance index value of 3. The applicable scenario information of the fourth model indicates that the model is a model applicable to the network scenario indicated by network data A corresponding to use case 2, and the corresponding model performance measurement value is a performance index value of 4. It should be noted that use case 1 and use case 2 may refer to AI-based use cases. In addition, the various applicable scenario information in Table 1 below are all examples and do not constitute a specific limitation on the applicable scenario information.

[0163] Table 1 Model information table

[0164] In another possible scenario, if the first configuration information sent by the network device does not include second indication information for indicating the content of the model information sent by the terminal, the terminal can directly send the model performance measurement value and / or applicable scenario information of the at least one first candidate model to the network device after screening out at least one second candidate model.

[0165] S203: The network device sends activation indication information to the terminal based on the model information, where the activation indication information is used to instruct activation of a first model among the at least one first candidate model.

[0166] After receiving model information from a terminal, the network device may assign at least one model identifier to the terminal based on the model information, so that the terminal associates the at least one model identifier with at least one first candidate model. Subsequently, the network device may send activation indication information to the terminal, which may include a first model identifier among the at least one model identifier, thereby instructing the terminal to activate the first model associated with the first model identifier. Associating a model with a model identifier may also be referred to as corresponding or mapping the model with the model identifier. Once a model is associated with a model identifier, the model identifier associated with the model can be used to identify the model.

[0167] In some embodiments, if the model information sent by the terminal indicates that there is at least one first candidate model, the network device may assign a model identifier to the terminal. In this case, the model identifier is the first model identifier. After receiving the first model identifier, the terminal associates the first model identifier with a model in at least one first candidate model, that is, uses the first model identifier as the model identifier of a first candidate model. In this case, the model associated with the first model identifier is the first model. After assigning the first model identifier to the terminal, the network device may send activation indication information containing the first model identifier to the terminal. In this way, after receiving the activation indication information, since the first model identifier in the activation indication information is associated with the first model, the terminal can activate the first model associated with the first model identifier based on the first model identifier in the activation indication information.

[0168] In other embodiments, if the model information sent by the terminal includes a model performance measurement value and / or applicable scenario information of at least one first candidate model, the network device may determine the number n of first candidate models based on the model information. Thereafter, the network device may allocate m model identifiers to the terminal based on the determined number n of first candidate models, where m may not be greater than n and m is an integer greater than 0. After receiving the m model identifiers allocated by the network device, the terminal may associate the m model identifiers with the m first candidate models in a one-to-one correspondence. After allocating the m model identifiers to the terminal, the network device may select a model identifier from the m model identifiers as the first model identifier, and then send activation indication information including the first model identifier to the terminal. Accordingly, since the terminal associates the first model identifier with a first candidate model, based on the first model identifier in the activation indication information, the terminal may activate the first candidate model associated with the first model identifier. At this time, the first candidate model associated with the first model identifier is the first model.

[0169] After the terminal activates the first model based on the activation instruction information of the network device, the terminal may then switch models through the following steps. It should be noted that the following steps S204 and S205 are optional steps.

[0170] S204: The terminal sends a model switching request to the network device, where the model switching request is used to request switching to a second model that meets the second condition information.

[0171] In the first possible scenario, after the terminal activates the first model, if the first condition information has not changed and the first condition information includes a model performance threshold, if the scenario in which the terminal is currently located has not changed, the terminal can monitor whether the model performance measurement value of the first model meets the model performance threshold. If it is monitored that the first model no longer meets the model performance threshold in the first condition information, the terminal can send a model switching request to the network device to request switching to a model that meets the second condition information. At this time, the second condition information is the same as the first condition information.

[0172] Among them, the terminal can measure the model performance of the first model within the preset time interval at intervals, thereby obtaining a model performance measurement value. Thereafter, the model performance measurement value is compared with the corresponding model performance threshold in the first condition information. If the model performance measurement value of the first model does not meet the corresponding model performance threshold, and the first candidate model screened out in the aforementioned S202 includes other models in addition to the first model, the terminal can select at least one second candidate model from the other first candidate models other than the first model, and then send a model switching request to the network device based on the at least one second candidate model to request switching to the second model of the at least one second candidate model.

[0173] Specifically, the terminal may select a model with the best model performance indicated by the model performance measurement value from other first candidate models other than the first model as the second candidate model. Alternatively, the terminal may select all other remaining first candidate models as second candidate models. Alternatively, the terminal may select a model with an associated model identifier from the remaining first candidate models as the second candidate model.

[0174] After determining at least one second candidate model, if the at least one second candidate model already has an associated model identifier, the at least one second model identifier associated with the at least one second candidate model is carried in the model switching request and sent to the network device.

[0175] Optionally, if the at least one second candidate model does not have an associated model identifier, the terminal may send a model switching request to the network device, and the model switching request may not carry the model identifier.

[0176] In a second possible scenario, after the terminal activates the first model, if the first condition information changes, for example, if the network device reconfigures condition information different from the first condition information, or if the scenario in which the terminal is currently located changes, the terminal can determine at least one second candidate model that meets the second condition information from the at least one available model. Then, based on the at least one determined second candidate model, a model switching request is sent to the network device. The second condition information can be condition information reconfigured by the network device, or it can be the same as the first condition information.

[0177] Among them, the implementation method of the terminal determining at least one second candidate model that meets the second condition information can refer to the implementation method of determining at least one first candidate model that meets the first condition information introduced above, and will not be repeated here.

[0178] In addition, after determining the second candidate model, if there is at least one second candidate model with an associated second model identifier, the terminal can carry at least one second model identifier associated with the at least one second candidate model in the model switching request and send it to the network device.

[0179] Optionally, if the determined second candidate model does not have an associated second model identifier, the model switching request sent by the terminal to the network device does not carry the model identifier. In this case, the model switching request may carry a model performance measurement value and / or applicable scenario information of at least one second candidate model.

[0180] In the third possible situation, after the terminal activates the first model, when the current scene of the terminal has not changed, the first condition information has not changed, and the first condition information includes the model performance threshold, if there are multiple first candidate models determined as mentioned above, the terminal can monitor the model performance measurement values ​​of multiple first candidate models including the first model. If it is monitored that there is at least one second candidate model among the multiple first candidate models whose model performance measurement value is better than the model performance measurement value of the first model, the terminal can send a model switching request to the network device to request switching to a model that meets the second condition information. At this time, the second condition information is the same as the first condition information.

[0181] In this case, after determining at least one second candidate model whose model performance measurement value is better than the first model, the terminal can select one with an associated model identifier from the at least one second candidate model as the second model, and then carry the second model identifier associated with the second model in the model switching request and send it to the network device.

[0182] Alternatively, in the case that the at least one second candidate model does not have an associated model identifier, the terminal may also send a model switching request without carrying a model identifier to the network device.

[0183] S205: The network device sends model switching indication information to the terminal, where the model switching indication information is used to indicate switching to the second model.

[0184] After receiving the model switching request from the terminal, the network device may send model switching indication information to the terminal based on the model switching request.

[0185] In one possible scenario, if the model switching request includes at least one second model identifier, the network device may select a model identifier from the at least one second model identifier and send model switching indication information to the terminal device, where the model switching indication information includes the selected model identifier. Accordingly, after receiving the model switching indication information, the terminal may switch from the first model to the model associated with the model identifier, since the model identifier in the model switching indication information is already associated with a model in at least one second candidate model. In this case, the model associated with the model identifier is the second model. The terminal switching from the first model to the second model may mean that the terminal deactivates the first model and activates the second model.

[0186] In another possible scenario, if the model switching request does not include a model identifier, the network device may allocate at least one second model identifier to the terminal. After receiving the at least one second model identifier, the terminal may associate the at least one second model identifier with at least one second candidate model. After sending the at least one second model identifier to the terminal, the network device may send model switching indication information to the terminal, and the model switching indication information may include one model identifier in the at least one second model identifier. Accordingly, after receiving the model switching indication information, the terminal may switch from the first model to the second candidate model associated with the model identifier because the model identifier in the model switching indication information is associated with a second candidate model. In this case, the second candidate model associated with the model identifier is the second model.

[0187] It should be noted that if the model switching request does not include a model identifier, but includes the model performance measurement value and / or applicable scenario information of at least one second candidate model, the network device can determine the number of second candidate models based on the model performance measurement value and / or applicable scenario information of the at least one second candidate model, and then allocate at least one second model identifier to the terminal based on the number of second candidate models.

[0188] The above S204 and S205 mainly introduce the process of the terminal requesting the network device to switch the model. In some possible situations, the network device may also monitor whether the model performance measurement value of the first model in the terminal meets the corresponding model performance threshold, or the network device may monitor whether there are other models with model performance measurement values ​​better than the first model. If it is monitored that the model performance measurement value of the first model does not meet the corresponding model performance threshold, or there are other models with model performance measurement values ​​better than the first model, the network device may actively send a model switching indication message to the terminal to instruct the terminal to switch to the second model.

[0189] In an embodiment of the present application, a terminal receives first configuration information sent by a network device, and the first configuration information includes first condition information. Based on this, the terminal can send model information to the network device to indicate at least one candidate model that meets the first condition information. In this way, the network device can control the terminal to activate a model that meets the first condition information based on the model information. It can be seen that in an embodiment of the present application, the network device configures the first condition information to the terminal, controls the terminal to screen and activate appropriate models that meet specific conditions, and thus realizes more refined management of the models in the terminal, thereby enabling the terminal to obtain better services.

[0190] In addition, in an embodiment of the present application, the first condition information sent by the network device to the terminal may include a model performance threshold and / or scenario condition information, and the scenario condition information may be used to instruct the terminal to filter models that match the current scenario. Based on this, the terminal can filter models that meet the model performance threshold and / or match the current scenario, and report the corresponding model information to the network device. On this basis, the network device can control the terminal to activate a model with better performance and / or that matches the current scenario based on the model information reported by the terminal, thereby enabling the terminal to obtain better service.

[0191] Finally, in an embodiment of the present application, the network device can also flexibly configure the content of the model information sent to the terminal. For example, the network device can configure the second indication information including the first indication value for the terminal. Based on the first indication value, the terminal can send 1-bit model information to the network device to indicate whether there is model information that meets the first condition information. In this way, the amount of data reported by the terminal can be reduced, and the communication cost can be reduced. For another example, the network device can configure the second indication information including the second indication value for the terminal. Based on the second indication value, the terminal can send model information including the model performance measurement value and / or applicable scenario information of at least one first candidate model to the network device. In this way, the network device can perceive the performance of the model screened by the terminal and the current scenario of the terminal through the model information, thereby providing a reference for other subsequent network decisions.

[0192] Based on the model management method described in the embodiment shown in FIG2 , the present application also provides an exemplary process of the model management method. Referring to FIG3 , the exemplary process may include the following steps:

[0193] S301: A network device sends first configuration information to a terminal, where the first configuration information includes a model performance threshold, measurement configuration information, and scenario condition information.

[0194] The implementation of this step may refer to the implementation of S201.

[0195] S302: The terminal measures the model performance of at least one model based on the measurement configuration information to obtain a model performance measurement value of the at least one model.

[0196] The implementation of this step may refer to the implementation of measuring model performance based on measurement configuration information introduced in S202.

[0197] S303: The terminal determines at least one first candidate model that meets the model performance threshold and scenario condition information from at least one model.

[0198] The implementation of this step may refer to the method 1 and the method 2 introduced in S202 to determine at least one first candidate model that satisfies both the model performance threshold and the scene condition information.

[0199] S304: The terminal sends model information to the network device, where the model information is used to indicate at least one first candidate model.

[0200] The implementation of this step may refer to the implementation of the terminal sending model information introduced in S202.

[0201] S305: The network device sends at least one model identifier to the terminal.

[0202] S306: The terminal associates at least one model identifier with at least one first candidate model.

[0203] S307: The network device sends activation indication information to the terminal, where the activation indication information includes a first model identifier among the at least one model identifier.

[0204] S308: The terminal activates the first model associated with the first model identifier.

[0205] The implementation of the above S305 to S308 can refer to the relevant implementation introduced in S203.

[0206] S309: The terminal sends a model switching request to the network device, where the model switching request includes a second model identifier.

[0207] S310: The network device sends model switching indication information to the terminal, where the model switching indication information includes a second model identifier.

[0208] S311: The terminal switches from the first model to the second model associated with the second model identifier.

[0209] The implementation of the above S309 to S311 may be the related implementation of S204 and S205, and will not be repeated here.

[0210] The present application also provides another model management method, which can be applied to a wireless communication system including a terminal and a network device. The network device can be a RAN node as described in the above embodiments, for example, a base station or a RIC in an O-RAN architecture, and the terminal can be any of the various types of terminals described in the above embodiments. Referring to FIG. 4 , the model management method can include the following steps:

[0211] S401: The terminal sends model information to a network device, where the model information includes a model identifier of at least one first candidate model and a corresponding model performance measurement value.

[0212] In an embodiment of the present application, the terminal can determine at least one first candidate model that meets specific conditions from at least one model available to itself, and then send model information to the network device based on the at least one first candidate model.

[0213] Exemplarily, the terminal may determine at least one first candidate model that satisfies the first condition information from at least one model. The first condition information may be determined by the terminal or pre-configured by the network device. Furthermore, the first condition information may include a model performance threshold and / or scenario condition information. Accordingly, the at least one first candidate model that satisfies the first condition information may refer to a model that satisfies the model performance threshold and / or scenario condition information.

[0214] In the first possible scenario, the first condition information may include scene condition information determined by the terminal based on the current scene or condition. For example, the first condition information may include information about the scene in which the terminal is currently located, determined by the terminal based on the currently collected end-side data, wherein the implementation method for the terminal to determine the scene in which the terminal is currently located based on the currently collected end-side data and / or network data can refer to the relevant introduction in the aforementioned S202 and will not be repeated here. For example, if the terminal determines that it is currently in a high-speed motion scene based on the currently collected speed data, the first condition information may include a high-speed scene label. Alternatively, the first condition information may also include a first model performance threshold determined by the terminal based on its own needs.

[0215] In a second possible scenario, the first condition information may include a second model performance threshold value pre-configured by the network device or scenario condition information. The manner in which the second model performance threshold value and scenario condition information are pre-configured by the network device can be referred to the introduction of S201 in the aforementioned embodiment, and will not be further described in this embodiment of the present application.

[0216] In the case where the first condition information includes a model performance threshold and / or scenario condition information, the implementation method for the terminal to determine at least one first candidate model that meets the first condition information from at least one model can refer to the relevant introduction of S202 in the aforementioned embodiment and will not be repeated here.

[0217] After determining the first candidate model, the terminal may send model information to the network device based on the determined first candidate model.

[0218] In one possible scenario, if there is at least one first candidate model with an associated model identifier, the terminal may directly send model information to the network device, where the model information may include the model identifier associated with the at least one first candidate model and the corresponding model performance measurement value.

[0219] Among them, the terminal can obtain the model performance measurement value of the at least one first candidate model from the OTT server. Alternatively, the terminal can measure the model performance of the at least one first candidate model to obtain the model performance measurement value. Among them, when the terminal measures the model performance of the at least one first candidate model, it can be to determine the measurement value of the specified performance indicator according to the measurement method of the default configuration, or, when the first condition information includes a model performance threshold, it can also be to determine the measurement value of the performance indicator indicated by the model performance threshold. Of course, if the network device has pre-configured measurement configuration information for the terminal, the terminal can also measure the measurement value of the performance indicator indicated by the model performance threshold according to the measurement method indicated by the measurement configuration information. The specific implementation method of measuring model performance to obtain the model performance measurement value can refer to the introduction in the aforementioned S202 and will not be repeated here.

[0220] After obtaining the model performance measurement value of at least one first candidate model, the terminal can associate the model identifier of each first candidate model with the corresponding model performance measurement value, thereby obtaining a mapping relationship between the model identifier of at least one first candidate model and the corresponding model performance measurement value, and then send the mapping relationship as model information to the network device.

[0221] In this mapping relationship, a model identifier may correspond to one model performance measurement value; or, a model identifier may also correspond to multiple model performance measurement values. In this case, the multiple model performance measurement values ​​corresponding to the model identifier may refer to the measurement values ​​of different performance indicators of the model identified by the model identifier, or may also be the model performance measurement values ​​of the model identified by the model identifier in different scenarios. In this case, the model performance measurement values ​​of the model in different scenarios may refer to the results obtained by performance testing the model using test sample sets corresponding to different scenarios.

[0222] In addition, in the mapping relationship, one model performance measurement value may correspond to one model identifier; or, one model performance measurement value may correspond to multiple model identifiers. In this case, it means that the model performance measurement values ​​of the multiple models identified by the multiple model identifiers are the same.

[0223] Optionally, in some embodiments, in the above mapping relationship, the model performance measurement value of at least one first candidate model may further correspond to measurement information.

[0224] The measurement information corresponding to the model performance measurement value is used to indicate the measurement method of the corresponding model performance measurement value. For example, the measurement information may include the number of test samples and / or the number of measurements used to measure the model performance measurement value.

[0225] Optionally, the measurement information may further include first indication information for indicating a measurement scenario for the model performance measurement value. For example, the first indication information may indicate a first scenario, a second scenario, or multiple scenarios. The first scenario refers to the scenario in which the terminal is currently located, and the second scenario refers to the scenario corresponding to the model. Based on this, when the first indication information indicates the first scenario, it indicates that the corresponding model performance measurement value is the performance measurement value of the model in the scenario in which the terminal is currently located, or it can also be understood that the corresponding model performance measurement value is the performance value measured using a test sample set corresponding to the scenario in which the terminal is currently located. When the first indication information indicates the second scenario, it indicates that the corresponding model performance measurement value is the performance measurement value of the model in the scenario corresponding to the model, or it can also be understood that the corresponding model performance measurement value is the performance value measured using a test sample set corresponding to the scenario corresponding to the model. When the first indication information indicates multiple scenarios, it indicates that the corresponding model performance measurement value is the performance measurement value of the model in multiple scenarios, or it can also be understood that the corresponding model performance measurement value is the performance value measured using a sample set containing test samples from multiple scenarios.

[0226] Optionally, the measurement information may further include a sample set identifier, which is used to indicate the test sample set used when measuring the corresponding model performance measurement value. It should be noted that, in some possible cases, when the measurement information includes first indication information, and the first indication information indicates a first scenario, the sample set identifier may be the identifier of the test sample set corresponding to the current scenario; when the second indication information indicates a second scenario, the sample set identifier may be the identifier of the test sample set corresponding to the scenario corresponding to the model; when the second indication information indicates multiple scenarios, the sample set identifier may be the identifier of a test sample set containing test samples corresponding to multiple scenarios.

[0227] Optionally, in other embodiments, in the above-mentioned mapping relationship, the model identifier may further correspond to applicable scenario information, where the applicable scenario information is used to indicate the applicable scenario of the model identified by the corresponding model identifier. The applicable scenario information of the model identified by each model identifier may be obtained from the OTT server, and each model identifier may have one or more applicable scenarios.

[0228] In another possible scenario, if each first candidate model determined by the terminal does not have an associated model identifier, the terminal may send a model identifier request to the network device, requesting at least one model identifier. After receiving the model identifier request, the network device may send N model identifiers to the terminal, where N is a positive integer greater than 0. The terminal may then associate at least one of the N received model identifiers with at least one first candidate model, and send model information to the network device based on the model identifier associated with the at least one first candidate model.

[0229] The terminal may determine the number M of requested model identifiers based on the number of determined first candidate models, and carry M in a model identification request sent to the network device. M may be equal to the number of first candidate models. Alternatively, the model identification request sent by the terminal may not include the requested number of model identifiers.

[0230] After receiving the model identification request, if the model identification request does not include the requested number of model identifications, the network device may send N model identifications to the terminal according to the default configured number N. For example, the network device may send one model identification to the terminal.

[0231] If the model identification request includes the requested number M of model identifications, the network device may send N model identifications to the terminal based on M, where N is not greater than M. For example, when M is greater than 1, the network device may send fewer than M model identifications to the terminal.

[0232] If the number of model identifiers received by the terminal is less than or equal to the number of first candidate models, the terminal may associate the N received model identifiers with the N first candidate models. In this case, each of the N first candidate models will be associated with a model identifier, and different models will be associated with different model identifiers. If the number of model identifiers received by the terminal is greater than the number of first candidate models, for example, if the number of first candidate models is R, the terminal may associate R of the N model identifiers with the R first candidate models. In this case, each first candidate model will be associated with a model identifier.

[0233] After associating at least one first candidate model with at least one model identifier, the terminal may send model information to the network device based on the at least one first candidate model associated with the model identifier. The model information may include the model identifier and corresponding model performance measurement value of the at least one first candidate model. In addition, it may also include applicable scenario information corresponding to the at least one first candidate model and measurement information corresponding to the model performance measurement value. For specific implementation methods, please refer to the relevant introduction above and will not be repeated here.

[0234] S402: The network device sends activation indication information to the terminal based on the model information, where the activation indication information is used to instruct activation of a first model among at least one first candidate model.

[0235] After receiving the model information from the terminal, the network device may send activation indication information to the terminal based on the model identifier of at least one first candidate model and the corresponding model performance measurement value in the model information.

[0236] In a first implementation, the network device may determine, based on the model identifier of at least one first candidate model included in the model information and the corresponding model performance measurement value, a first model identifier whose corresponding model performance measurement value satisfies a third model performance threshold. The network device may then send activation indication information to the terminal, where the activation indication information includes the first model identifier. After receiving the activation indication information, the terminal may activate the model associated with the first model identifier. In this case, the model associated with the first model identifier is the first model.

[0237] It should be noted that the third model performance threshold may be a model performance threshold determined by the network device based on its own needs. Optionally, if at least one first candidate model is a model selected by the first model performance threshold or the second model performance threshold, the model performance represented by the third model performance threshold may be superior to the model performance represented by the first model performance threshold or the second model performance threshold.

[0238] Optionally, if the model information also includes measurement information corresponding to the model performance measurement value, the network device can also determine the model performance measurement value that conforms to the measurement method required by the network device based on the measurement information, and determine the first model performance measurement value that meets the third model performance threshold from these model performance measurement values, and use the model identifier corresponding to the first model performance measurement value as the first model identifier.

[0239] Optionally, if the model information also includes applicable scenario information corresponding to the model identifier, the network device can also determine that the applicable scenario indicated by the corresponding applicable scenario information includes the scenario required by the network device, and the corresponding model performance measurement value meets the first model identifier of the third model performance threshold.

[0240] If there are multiple first model identifiers determined by any of the above methods, the network device may select one first model identifier from them and send activation indication information to the terminal, where the activation indication information includes the selected first model identifier.

[0241] It should be noted that if the first model identifier that meets the conditions cannot be determined through any of the above methods, the network device will not send activation indication information to the terminal, or the network device may send other types of notification messages to the terminal to notify the terminal that there is no activatable model that meets the conditions.

[0242] In a second implementation, if the model information includes a model identifier and a corresponding model performance measurement value, the network device may also directly send activation instruction information to the terminal based on the model identifier. In this case, the activation instruction information includes the model identifier. Accordingly, upon receiving the activation instruction information, the terminal activates the model associated with the model identifier, i.e., the first model.

[0243] After the terminal activates the first model based on the activation instruction information sent by the network device, the terminal may subsequently switch models. The implementation method for the terminal to switch models may refer to the switching method described in S204 and S205 of the aforementioned embodiment, or may refer to the implementation method in which the network device actively instructs the terminal to switch models as described in the aforementioned embodiment, and will not be repeated here.

[0244] In an embodiment of the present application, a terminal can report model information to a network device. The model information includes a model identifier of at least one first candidate model and a corresponding model performance measurement value. Based on this, the network device can perceive the performance of each model through the model information reported by the terminal, and then use the model identifier in the model information to instruct the terminal to activate a model with better performance. This implements refined model management based on model performance, enabling the terminal to obtain better service.

[0245] In addition, in an embodiment of the present application, the model information sent by the terminal may also include applicable scenario information corresponding to at least one first candidate model. Based on this, the network device can also perceive the applicable scenario of the model and control the terminal to activate a model that is more suitable for the current scenario and has better performance, thereby enabling the terminal to obtain better service.

[0246] Finally, in this embodiment of the present application, the model identification request sent by the terminal to the network device may include the number of model identifications requested. The network device may then control the number of model identifications allocated to the terminal based on the number of model identifications requested by the terminal. This allows for minimizing the number of model identifications required when the network device serves a large number of terminals, thereby ensuring that the model management process based on model identifications can proceed normally.

[0247] Based on the model management method provided in the embodiment shown in FIG4 , the embodiment of the present application further provides an exemplary process of the model management method, as shown in FIG5 , which includes the following steps:

[0248] S501: The terminal determines at least one first candidate model.

[0249] S502: The terminal sends a model identification request to the network device, where the model identification request includes the requested number M of model identifications.

[0250] S503: The network device sends N model identifiers to the terminal, where N is not greater than M.

[0251] S504: The terminal associates at least one model identifier among the N model identifiers with at least one first candidate model.

[0252] S505: The terminal sends model information to the network device, where the model information includes a model identifier of at least one first candidate model and a corresponding model performance measurement value.

[0253] S506: The network device sends activation indication information to the terminal, where the activation indication information includes a first model identifier among the model identifiers of at least one first candidate model.

[0254] S507: The terminal activates the first model associated with the first model identifier.

[0255] S508: The terminal sends a model switching request to the network device, where the model switching request includes the second model identifier.

[0256] S509: The network device sends model switching indication information to the terminal, where the model switching indication information includes a second model identifier.

[0257] S510: The terminal switches from the first model to the second model associated with the second model identifier.

[0258] The implementation of S501 to S505 can refer to the implementation of S401 in the above embodiment. The implementation of S506 to S507 can refer to the implementation of S402 in the above embodiment. The implementation of S508 to S510 can refer to the implementation of S204 and S205 in the above embodiment.

[0259] It is worth noting that each of the above-mentioned embodiments can be applied to the O-RAN system. Referring to Figure 6, the O-RAN system includes RIC, O-CU and O-DU. Among them, RIC can be used to collect network information and perform necessary optimization operations. RIC can communicate with O-CU and O-DU respectively through the E2 interface to control O-CU and O-DU. O-CU can be used to implement the functions of radio resource control protocol and PDCP, and can also implement the functions of SDAP; O-DU can be used to implement the functions of radio link control layer and MAC layer, and can also complete the functions of part or all of the physical layer. Among them, O-CU can communicate with O-DU through F1 interface.

[0260] Based on this, an embodiment of the present application further provides an exemplary process of a model management method applied in an O-RAN system. Referring to FIG7 , the process may include the following steps:

[0261] S701: The RIC sends first configuration information to the O-DU, where the first configuration information includes a model performance threshold and / or scenario condition information.

[0262] The RIC may send the first configuration information to the O-DU via the E2 interface. For the description of the model performance threshold and / or scenario condition information, please refer to the introduction in the aforementioned S201.

[0263] S702: The O-DU sends first configuration information to the terminal.

[0264] S703: The terminal sends model information to the O-DU, where the model information is used to indicate at least one first candidate model that meets the model performance threshold and / or scenario condition information.

[0265] The implementation method of the terminal sending the model information may refer to the introduction in the aforementioned S202.

[0266] S704: The O-DU sends the model information to the RIC.

[0267] The O-DU sends the model information from the terminal to the RIC via the E2 interface.

[0268] S705: The RIC sends at least one model identifier to the O-DU.

[0269] The at least one model identifier is a model identifier assigned by the RIC to at least one first candidate model screened for the terminal.

[0270] S706: The O-DU sends the at least one model identifier to the terminal.

[0271] S707: The terminal associates at least one model identifier with at least one first candidate model.

[0272] S708: The RIC sends activation indication information to the O-DU, where the activation indication information includes a first model identifier among the at least one model identifier.

[0273] S709: The O-DU sends activation instruction information to the terminal.

[0274] S710: The terminal activates the first model associated with the first model identifier.

[0275] FIG8 is an exemplary flow chart of another model management method applied to an O-RAN system provided in an embodiment of the present application. Referring to FIG8 , the process includes the following steps:

[0276] S801: The RIC sends first configuration information to the O-DU through the O-CU. The first configuration information includes a model performance threshold and / or scenario condition information.

[0277] The RIC may send the first configuration information to the O-CU via the E2 interface, and the O-CU may send the first configuration information to the O-DU via the F1 interface. For a description of the model performance threshold and / or scenario condition information, please refer to the introduction in S201 above.

[0278] S802: The O-DU sends first configuration information to the terminal.

[0279] S803: The terminal sends model information to the O-DU, where the model information is used to indicate at least one first candidate model that meets the model performance threshold and / or scenario condition information.

[0280] The implementation method of the terminal sending the model information may refer to the introduction in the aforementioned S202.

[0281] S804: The O-DU sends the model information to the RIC via the O-CU.

[0282] Among them, O-DU sends the model information from the terminal to O-CU through the F1 interface, and O-CU then sends the model information to RIC through the E2 interface.

[0283] S805: The RIC sends at least one model identifier to the O-DU through the O-CU.

[0284] The at least one model identifier is a model identifier assigned by the RIC to at least one first candidate model screened for the terminal.

[0285] S806: The O-DU sends the at least one model identifier to the terminal.

[0286] S807: The terminal associates at least one model identifier with at least one first candidate model.

[0287] S808: The RIC sends activation indication information to the O-DU through the O-CU. The activation indication information includes a first model identifier among the at least one model identifier.

[0288] S809: The O-DU sends activation instruction information to the terminal.

[0289] S810: The terminal activates the first model associated with the first model identifier.

[0290] Figures 7 and 8 above illustrate exemplary processes when the model management method in the embodiment shown in Figure 2 is applied to an O-RAN system. It should be noted that the model management method in the embodiment shown in Figure 4 can also be applied to an O-RAN system. In this case, the functions of the network device in the embodiment shown in Figure 4 can be implemented by the RIC, and the RIC can refer to the above-mentioned Figures 7 or 8 to exchange information with the terminal through the O-DU or through the O-CU and O-DU. The embodiments of this application will not be repeated here.

[0291] Next, the communication device provided in the embodiment of the present application is introduced.

[0292] Figure 9 is a structural diagram of a communication device provided in an embodiment of the present application. The functions of the terminal or network device in the aforementioned embodiment can be implemented by one or more communication devices shown in Figure 9. For example, referring to Figure 9, the communication device 900 includes at least one processor 901, a communication bus 902, a memory 903 and at least one communication interface 904. It should be noted that the device structure shown in Figure 9 does not constitute a limitation on the communication device 900. The communication device 900 may include more or fewer components than shown in the figure, or combine certain components, or arrange components differently, which is not limited in this embodiment of the present application. The following is a detailed introduction to the various components of the communication device 900 in conjunction with Figure 9:

[0293] The processor 901 is the control center of the communication device 900, which can be a processor or a general term for multiple processing elements. For example, the processor 901 can be a general-purpose central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs). Among them, the processor 901 can perform various functions of the communication device 900 by running or executing software programs stored in the memory 903, and calling data stored in the memory 903. For example, the actions of the cloud server in the embodiments below can be executed by calling the data in the memory by the processor of the corresponding device.

[0294] As an embodiment, the processor 901 may include one or more CPUs, such as CPU0 and CPU1 shown in FIG. 9 .

[0295] As an embodiment, the communication device 900 may include multiple processors, such as the processor 901 and the processor 905 shown in FIG9 . Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. The processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0296] Communication bus 902 may include a path for transmitting information between the aforementioned components. Communication bus 902 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Such buses may be classified as address buses, data buses, or control buses. For ease of illustration, FIG9 shows a single thick line, but this does not necessarily indicate that there is only one bus or only one type of bus.

[0297] The memory 903 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program codes in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory 903 may be independent and connected to the processor 901 via a communication bus 902. The memory 903 may also be integrated with the processor 901. Among them, the memory 903 is used to store the software program that executes the solution provided in the embodiment of the present application, and is controlled and executed by the processor 901.

[0298] The communication interface 904 is used to communicate with other devices or communication networks, such as Ethernet, RAN, wireless local area networks (WLAN), etc. The communication interface 904 may include a receiving unit to implement a receiving function, and a sending unit to implement a sending function.

[0299] As an embodiment, the communication device 900 may further include an output device 906 and an input device 907. The output device 906 communicates with the processor 901 and can display information in a variety of ways. For example, the output device 906 can be a liquid crystal display (LCD), a light emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector. The input device 907 communicates with the processor 901 and can receive input in a variety of ways. For example, the input device 907 can be a mouse, a keyboard, a touch screen device, or a sensor device.

[0300] FIG10 is a schematic diagram of the structure of another communication device provided in an embodiment of the present application. The communication device can be deployed in a terminal. As shown in FIG10 , the communication device 1000 includes: a first receiving module 1001 and a first sending module 1002 .

[0301] The first receiving module 1001 is used to receive the first configuration information sent by the network device in S301 in the aforementioned embodiment, and receive the activation instruction information sent by the network device in S303; the first sending module 1002 is used to execute S302 in the aforementioned embodiment.

[0302] Optionally, the communication device 1000 also includes a first processing module 1003, which is used to determine at least one first candidate model that meets the first condition information from at least one model; the first sending module 1002 is specifically used to send model information to the network device based on at least one first candidate model.

[0303] Optionally, the first condition information includes a model performance threshold; the first processing module 1003 is specifically used to: obtain a model performance measurement value of at least one model; determine a model from at least one model whose model performance measurement value meets the model performance threshold, and at least one first candidate model includes a model whose model performance measurement value meets the model performance threshold.

[0304] Optionally, the first configuration information also includes measurement configuration information, which is used to indicate the terminal's measurement method for the model performance measurement value of at least one model; the first processing module 1003 is specifically used to: based on the measurement configuration information, measure the model performance of at least one model to obtain the model performance measurement value of at least one model.

[0305] Optionally, the measurement configuration information includes the number of test samples and / or the number of model performance measurements.

[0306] Optionally, the measurement configuration information also includes first indication information, and the first indication information is used to indicate a performance measurement scenario, and the performance measurement scenario includes a first scenario or a second scenario, the first scenario is the scenario in which the terminal is currently located, and the second scenario is the scenario corresponding to the first model; the first processing module 1003 is specifically used to: based on the first indication information, obtain a test sample set corresponding to the first scenario or the second scenario; and use the test sample set to measure the model performance of at least one model.

[0307] Optionally, the first condition information includes scene condition information, and the scene condition information is used to indicate a first scene, where the first scene is the scene in which the terminal is currently located; the first processing module 1003 is specifically used to: determine a model suitable for the first scene from at least one model, and at least one first candidate model includes a model suitable for the first scene.

[0308] Optionally, the scenario condition information includes at least one of an antenna shape, a number of antenna ports, and a carrier frequency of the network device.

[0309] Optionally, the first configuration information further includes second indication information. When the second indication information includes the first indication value, the model information is used to indicate that there is at least one first candidate model.

[0310] Optionally, the first configuration information further includes second indication information. When the second indication information includes a second indication value, the model information includes a model performance measurement value and / or applicable scenario information of at least one first candidate model.

[0311] Optionally, the first sending module 1002 is also used to: send a model switching request to the network device, the model switching request includes at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; the first receiving module 1001 is also used to: receive model switching indication information from the network device, and the model switching indication information is used to indicate switching to a second model in at least one second candidate model.

[0312] In an embodiment of the present application, a terminal receives first configuration information sent by a network device, and the first configuration information includes first condition information. Based on this, the terminal can send model information to the network device to indicate at least one candidate model that meets the first condition information. In this way, the network device can control the terminal to activate a model that meets the first condition information based on the model information. It can be seen that in an embodiment of the present application, the network device configures the first condition information to the terminal, controls the terminal to screen and activate appropriate models that meet specific conditions, and thus realizes more refined management of the models in the terminal, thereby enabling the terminal to obtain better services.

[0313] Figure 11 is a structural diagram of another communication device provided in an embodiment of the present application. The communication device can be deployed in the terminal in the aforementioned embodiment. Referring to Figure 11, the communication device 1100 may include a second sending module 1101 and a second receiving module 1102.

[0314] The second sending module 1101 is used to execute S501 in the aforementioned embodiment, and the second receiving module 1102 is used to receive the activation instruction information sent by the network device in S502 in the aforementioned embodiment.

[0315] Optionally, the communication device 1100 also includes a second processing module 1103, which is used to determine at least one first candidate model; the second sending module 1101 is also used to send a model identification request to the network device, and the model identification request is used to request at least one model identification; the second receiving module 1102 is also used to receive N model identifications from the network device, and the N model identifications include the model identification of at least one first candidate model, and N is a positive integer greater than 0.

[0316] Optionally, the second processing module 1103 is further configured to: associate at least one model identifier among the N model identifiers with at least one first candidate model.

[0317] Optionally, the model identification request includes a number M of requested model identifications, where N is not greater than M.

[0318] Optionally, the second sending module 1101 is also used to send a model switching request to the network device, the model switching request includes at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; the second receiving module 1102 is also used to receive model switching indication information from the network device, and the model switching indication information is used to indicate switching to a second model in at least one second candidate model.

[0319] Optionally, the model information further includes at least one of measurement information of a model performance measurement value of at least one first candidate model and applicable scenario information of at least one first candidate model, and the measurement information is used to indicate a measurement method.

[0320] Optionally, at least one first candidate model satisfies first condition information, where the first condition information includes condition information determined by the terminal based on the current scenario or condition.

[0321] Optionally, the model information includes a mapping relationship between a model identifier of at least one first candidate model and a model performance measurement value, wherein, in the mapping relationship, one model performance measurement value corresponds to one or more model identifiers of the first candidate models, or, a model identifier of the first candidate model corresponds to one or more model performance measurement values.

[0322] In an embodiment of the present application, a terminal may send model information to a network device, the model information including a model identifier of at least one first candidate model and a corresponding model performance measurement value. Based on this, the network device may perceive the performance of each model through the model information reported by the terminal, and then use the model identifier in the model information to instruct the terminal to activate a model with better performance. This implements refined model management based on model performance, enabling the terminal to obtain better service.

[0323] FIG12 is a schematic diagram of the structure of another communication device provided in an embodiment of the present application. The communication device can be deployed in the network device of the aforementioned embodiment. Referring to FIG12 , the communication device 1200 includes: a third sending module 1201 and a third receiving module 1202 .

[0324] The third sending module 1201 is used to execute S301 and S303 in the above embodiment. The third receiving module 1202 is used to receive the model information sent by the terminal in S302 in the above embodiment.

[0325] Optionally, the first condition information includes a model performance threshold, and the at least one first candidate model includes a model whose model performance measurement value satisfies the model performance threshold.

[0326] Optionally, the first configuration information further includes measurement configuration information, where the measurement configuration information is used to indicate a measurement method of the terminal for a model performance measurement value of at least one model.

[0327] Optionally, the measurement configuration information includes the number of test samples and / or the number of model performance measurements.

[0328] Optionally, the measurement configuration information also includes first indication information, where the first indication information is used to indicate a performance measurement scenario. The performance measurement scenario includes a first scenario or a second scenario. The first scenario is the scenario in which the terminal is currently located, and the second scenario is the scenario corresponding to the first model.

[0329] Optionally, the first condition information includes scene condition information, where the scene condition information is used to indicate a first scene, where the first scene is the scene in which the terminal is currently located, and the at least one first candidate model includes a model applicable to the first scene.

[0330] Optionally, the scenario condition information includes at least one of an antenna shape, a number of antenna ports, and a carrier frequency of the network device.

[0331] Optionally, the first configuration information further includes second indication information. When the second indication information includes the first indication value, the model information is used to indicate that there is at least one first candidate model.

[0332] Optionally, the first configuration information further includes second indication information. When the second indication information includes a second indication value, the model information includes applicable scenario information and / or a model performance measurement value of at least one first candidate model.

[0333] Optionally, the third receiving module 1202 is also used to receive a model switching request from the terminal, the model switching request includes at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; the third sending module 1201 is also used to send model switching indication information to the terminal, and the model switching indication information is used to instruct the terminal to switch to the second model in at least one second candidate model.

[0334] In an embodiment of the present application, the network device sends first configuration information including first condition information to the terminal to control the terminal to screen at least one first candidate model that meets the first condition information, and sends model information for indicating the at least one first candidate model. In this way, the network device can control the terminal to activate one of the at least one first candidate model based on the model information. It can be seen that in an embodiment of the present application, the network device configures the first condition information to the terminal, controls the terminal to screen and activate suitable models that meet specific conditions, thereby achieving more refined management of the models in the terminal, thereby enabling the terminal to obtain better services.

[0335] Figure 13 is a structural diagram of another communication device provided in an embodiment of the present application. The communication device can be deployed in the network equipment in the aforementioned embodiment. Referring to Figure 13, the communication device 1300 includes: a fourth receiving module 1301 and a fourth sending module 1302.

[0336] The fourth receiving module 1301 is used to receive the model information sent by the terminal in S501 in the above embodiment; the fourth sending module 1302 is used to execute S502 in the above embodiment.

[0337] Optionally, the fourth receiving module 1301 is also used to receive a model identification request from the terminal, and the model identification request is used to request at least one model identification; the fourth sending module 1302 is also used to send N model identifications to the terminal, and the N model identifications include the model identification of at least one first candidate model, and N is a positive integer greater than 0.

[0338] Optionally, the model identification request includes a number M of requested model identifications, where N is not greater than M.

[0339] Optionally, the fourth receiving module 1301 is also used to: receive a model switching request from the terminal, the model switching request includes at least one second model identifier, and the at least one second model identifier is used to identify at least one second candidate model that meets the second condition information; the fourth sending module 1302 is also used to: send model switching indication information to the terminal, and the model switching indication information is used to instruct the terminal to switch to the second model in at least one second candidate model.

[0340] Optionally, the model information further includes at least one of measurement information of a model performance measurement value of at least one first candidate model and applicable scenario information of at least one first candidate model, and the measurement information is used to indicate a measurement method.

[0341] Optionally, at least one first candidate model satisfies first condition information, where the first condition information includes condition information determined by the terminal based on the current scenario or condition.

[0342] Optionally, the model information includes a mapping relationship between a model identifier of at least one first candidate model and a model performance measurement value, wherein, in the mapping relationship, one model performance measurement value corresponds to one or more model identifiers of the first candidate models, or, a model identifier of the first candidate model corresponds to one or more model performance measurement values.

[0343] In an embodiment of the present application, a network device receives model information sent by a terminal, and the model information includes a model identifier and a model performance measurement value of at least one first candidate model. In this way, through the model information, the network device can perceive the performance of the model, and then use the model identifier in the model information to instruct the terminal to activate a model with better performance, thereby realizing refined model management based on model performance, so that the terminal can obtain better services.

[0344] It should be noted that the division of modules in the communication devices provided in the above embodiments is schematic and is only a logical functional division. In actual implementation, there may be other division methods. In addition, the functional modules in the various embodiments of the present application may be integrated into a processor, or may exist separately physically, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or software functional modules.

[0345] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be an electronic device or a server, etc.) or a processor to execute all or part of the steps of the method of each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0346] In addition, the communication device and model management method embodiments provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0347] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware, or any combination thereof. When implemented using 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 instructions are loaded and executed on a computer, all or part of the processes or functions described in 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 device. 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 via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0348] In the various embodiments of the present application, unless otherwise specified or logically conflicting, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other. The technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships. In the embodiments of the present application, "at least one" refers to one or more, and "more" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the textual description of the embodiments of the present application, the character " / " generally indicates that the related objects before and after are in an "or" relationship. In the present application, "first", "second" and various numerical numbers are only used to distinguish for the convenience of description and are not used to limit the scope of the embodiments of the present application. For example, to distinguish different messages, etc., rather than to describe a specific order or sequence.

[0349] It is understood that the various numbers used in the embodiments of the present application are only used for ease of description and are not intended to limit the scope of the embodiments of the present application. The order of the sequence numbers of the above-mentioned processes does not necessarily indicate the order in which they are executed. The order in which the processes are executed should be determined by their functions and internal logic.

[0350] Finally, it should be noted that the above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A model management method, characterized in that: The method comprises: receiving first configuration information from a network device, wherein the first configuration information includes first condition information; Based on the first configuration information, sending model information to the network device, where the model information is used to indicate at least one first candidate model that meets the first condition information; Activation indication information is received from the network device, where the activation indication information is used to instruct activation of a first model among the at least one first candidate model.

2. The method according to claim 1, characterized in that The sending model information to the network device based on the first configuration information includes: Determine the at least one first candidate model that satisfies the first condition information from at least one model; Based on the at least one first candidate model, the model information is sent to the network device.

3. The method according to claim 2, characterized in that The first condition information includes a model performance threshold; The determining, from at least one model, the at least one first candidate model that satisfies the first condition information comprises: obtaining a model performance measure of the at least one model; A model whose model performance measurement value satisfies the model performance threshold is determined from the at least one model, wherein the at least one first candidate model includes a model whose model performance measurement value satisfies the model performance threshold.

4. The method according to claim 3, characterized in that The first configuration information further includes measurement configuration information, where the measurement configuration information is used to indicate a measurement method of the terminal for a model performance measurement value of the at least one model; The obtaining of a model performance measurement value of the at least one model includes: Based on the measurement configuration information, the model performance of the at least one model is measured to obtain a model performance measurement value of the at least one model.

5. The method according to claim 4, characterized in that The measurement configuration information includes the number of test samples and / or the number of model performance measurements.

6. The method according to claim 4 or 5, characterized in that The measurement configuration information further includes first indication information, where the first indication information is used to indicate a performance measurement scenario, where the performance measurement scenario includes a first scenario or a second scenario, where the first scenario is a scenario in which the terminal is currently located, and the second scenario is a scenario corresponding to the first model; The measuring the model performance of the at least one model based on the measurement configuration information includes: Based on the first indication information, obtaining a test sample set corresponding to the first scenario or the second scenario; The model performance of the at least one model is measured using the test sample set.

7. The method according to any one of claims 2 to 6, characterized in that: The first condition information includes scene condition information, where the scene condition information is used to indicate a first scene, where the first scene is the scene the terminal is currently in; The determining, from at least one model, the at least one first candidate model that satisfies the first condition information comprises: A model suitable for the first scenario is determined from the at least one model, the at least one first candidate model including a model suitable for the first scenario.

8. The method according to claim 7, characterized in that The scenario condition information includes at least one of an antenna shape, a number of antenna ports, and a carrier frequency of the network device.

9. The method according to any one of claims 1 to 8, characterized in that: The first configuration information further includes second indication information. When the second indication information includes a first indication value, the model information is used to indicate the existence of the at least one first candidate model.

10. The method according to any one of claims 1 to 8, characterized in that: The first configuration information also includes second indication information. When the second indication information includes a second indication value, the model information includes a model performance measurement value and / or applicable scenario information of the at least one first candidate model.

11. The method according to any one of claims 1 to 10, characterized in that: The method further comprises: Sending a model switching request to the network device, where the model switching request includes at least one second model identifier, where the at least one second model identifier is used to identify at least one second candidate model that meets second condition information; Model switching indication information is received from the network device, where the model switching indication information is used to indicate switching to a second model among the at least one second candidate model.

12. A model management method, characterized in that: The method comprises: Sending model information to the network device, the model information including a model identifier of at least one first candidate model and a corresponding model performance measurement value; Activation indication information is received from the network device, where the activation indication information is used to instruct activation of a first model among the at least one first candidate model.

13. The method according to claim 12, characterized in that Before sending the model information to the network device, the method further includes: determining the at least one first candidate model; Sending a model identification request to the network device, wherein the model identification request is used to request at least one model identification; N model identifiers are received from the network device, where the N model identifiers include the model identifier of the at least one first candidate model, and N is a positive integer greater than 0.

14. The method according to claim 12 or 13, characterized in that The model information further includes at least one of measurement information of a model performance measurement value of the at least one first candidate model and applicable scenario information of the at least one first candidate model, and the measurement information is used to indicate a measurement method.

15. A model management method, characterized in that: The method comprises: Sending first configuration information to the terminal, where the first configuration information includes first condition information; receiving model information from the terminal, where the model information is used to indicate at least one first candidate model that meets the first condition information; Based on the model information, activation indication information is sent to the terminal, where the activation indication information is used to instruct activation of a first model in the at least one first candidate model.

16. The method according to claim 15, characterized in that The first condition information includes a model performance threshold, and the at least one first candidate model includes a model whose model performance measurement value satisfies the model performance threshold.

17. The method according to claim 16, characterized in that The first configuration information further includes measurement configuration information, where the measurement configuration information is used to indicate a measurement method for the terminal of a model performance measurement value of at least one model.

18. The method according to claim 17, characterized in that The measurement configuration information includes the number of test samples and / or the number of model performance measurements.

19. The method according to claim 17 or 18, characterized in that The measurement configuration information also includes first indication information, where the first indication information is used to indicate a performance measurement scenario. The performance measurement scenario includes a first scenario or a second scenario. The first scenario is the scenario in which the terminal is currently located, and the second scenario is the scenario corresponding to the first model.

20. The method according to any one of claims 15 to 19, characterized in that The first condition information includes scene condition information, where the scene condition information is used to indicate a first scene, where the first scene is the scene in which the terminal is currently located, and the at least one first candidate model includes a model applicable to the first scene.

21. The method according to claim 20, characterized in that The scenario condition information includes at least one of an antenna configuration, a number of antenna ports, and a carrier frequency of a network device.

22. The method according to any one of claims 15 to 21, characterized in that The first configuration information also includes second indication information. When the second indication information includes a first indication value, the model information is used to indicate the existence of at least one first candidate model; when the second indication information includes a second indication value, the model information includes applicable scenario information and / or model performance measurement value of the at least one first candidate model.

23. The method according to any one of claims 15 to 22, characterized in that The method further comprises: receiving a model switching request from the terminal, the model switching request including at least one second model identifier, the at least one second model identifier being used to identify at least one second candidate model that meets second condition information; Model switching indication information is sent to the terminal, where the model switching indication information is used to instruct the terminal to switch to a second model among the at least one second candidate model.

24. A model management method, characterized in that: The method comprises: receiving model information from a terminal, the model information including a model identifier of at least one first candidate model and a corresponding model performance measurement value; Based on the model information, activation indication information is sent to the terminal, where the activation indication information is used to instruct activation of a first model in the at least one first candidate model.

25. The method according to claim 24, characterized in that Before receiving the model information from the terminal, the method further includes: receiving a model identification request from the terminal, wherein the model identification request is used to request at least one model identification; N model identifiers are sent to the terminal, where the N model identifiers include the model identifier of the at least one first candidate model, and N is a positive integer greater than 0.

26. The method according to claim 24 or 25, characterized in that The model information further includes at least one of measurement information of a model performance measurement value of the at least one first candidate model and applicable scenario information of the at least one first candidate model, and the measurement information is used to indicate a measurement method.

27. A communication device, characterized in that: The communication device includes at least one module, and the at least one module is used to execute the model management method according to any one of claims 1 to 26.

28. A communication device, characterized in that: The communication device includes a processor, which is used to execute at least one program instruction or code stored in the memory to implement the model management method according to any one of claims 1 to 26.

29. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on a computer device, the computer device executes the model management method according to any one of claims 1 to 26.

30. A computer program product comprising instructions, characterized in that When the instructions are executed by a computer device, the computer device is caused to execute the model management method according to any one of claims 1 to 26.

Images