Model control method and device, equipment and medium

Abstract

The invention discloses a model control method and device, equipment and a medium, and relates to the field of wireless communication. The method is applied to a first communication device, the method comprising: transmitting a model control message for controlling an artificial intelligence (AI) / machine learning (ML) model, the model control message comprising an AI / ML model identifier of the AI / ML model (210). The method provides an AI / ML model control scheme based on AI / ML model identification.

Description

Model control method, device, equipment and medium Technical Field

[0001] The present application relates to the field of wireless communications, and in particular to a model control method, apparatus, device, and medium. Background Art

[0002] With the continuous development of artificial intelligence (AI) and machine learning (ML) technologies, the integration of communications technology and AI / ML technology is one of the future trends in communications. Due to the complexity and diversity of communication system scenarios, communication systems may introduce a large number of AI / ML models. However, there is still no standardized solution for managing the large number of introduced AI / ML models.

[0003] Summary of the Invention

[0004] The embodiments of the present application provide a model control method, apparatus, device, and medium, and provide an AI / ML model control solution based on AI / ML model identification. The technical solution is as follows:

[0005] According to one aspect of the present application, a model control method is provided, which is applied to a first communication device. The method includes:

[0006] A model control message is transmitted, where the model control message is used to control an artificial intelligence (AI) / machine learning (ML) model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0007] According to one aspect of the present application, a model control method is provided, which is applied to a second communication device, and the method includes:

[0008] A model control message is transmitted, where the model control message is used to control an artificial intelligence (AI) / machine learning (ML) model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0009] According to one aspect of the present application, a model control apparatus is provided, the apparatus being used to implement a first communication device, the apparatus comprising:

[0010] The first transmission module is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0011] According to one aspect of the present application, a model control apparatus is provided, the apparatus being used to implement a second communication device, the apparatus comprising:

[0012] The second transmission module is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0013] According to one aspect of the present application, a first communication device is provided, comprising: a processor and a transceiver connected to the processor; wherein,

[0014] The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0015] According to one aspect of the present application, a second communication device is provided, comprising: a processor and a transceiver connected to the processor; wherein,

[0016] The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0017] According to one aspect of the present application, a first communication device is provided, comprising: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the model control method as described in the above aspect.

[0018] According to one aspect of the present application, a second communication device is provided, comprising: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the model control method as described in the above aspect.

[0019] According to one aspect of the present application, a computer-readable storage medium is provided, in which executable instructions are stored. The executable instructions are loaded and executed by a processor to enable a communication device to implement the model control method described in the above aspect.

[0020] According to one aspect of an embodiment of the present application, a chip is provided, which includes a programmable logic circuit and / or program instructions. When the chip runs on a communication device, it is used to enable the communication device to implement the model control method described in the above aspect.

[0021] According to one aspect of the present application, a computer program product is provided. When the computer program product is executed on a processor of a communication device, the communication device executes the model control method described in the above aspect.

[0022] The technical solutions provided by the embodiments of the present application include at least the following beneficial effects:

[0023] AI / ML model identifiers are defined for each AI / ML model and are used to distinguish different AI / ML models. Model control messages carrying the AI / ML model identifier are used to control and manage specific AI / ML models. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0025] FIG1 is a schematic diagram of a communication system provided by an exemplary embodiment of the present application;

[0026] FIG2 is a schematic diagram of a network structure provided by an exemplary embodiment of the present application;

[0027] FIG3 is a flow chart of a model control method provided by an exemplary embodiment of the present application;

[0028] FIG4 is a flow chart of a model control method provided by an exemplary embodiment of the present application;

[0029] FIG5 is a schematic diagram of a model control method provided by an exemplary embodiment of the present application;

[0030] FIG6 is a flow chart of a model control method provided by an exemplary embodiment of the present application;

[0031] FIG7 is a flow chart of a model control method provided by an exemplary embodiment of the present application;

[0032] FIG8 is a schematic diagram of a model control method provided by an exemplary embodiment of the present application;

[0033] FIG9 is a schematic diagram of a model control method provided by an exemplary embodiment of the present application;

[0034] FIG10 is a structural block diagram of a model control device provided by an exemplary embodiment of the present application;

[0035] FIG11 is a structural block diagram of a model control device provided by an exemplary embodiment of the present application;

[0036] FIG12 is a schematic structural diagram of a communication device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0038] The network architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. A person skilled in the art will appreciate that, with the evolution of the network architecture and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0039] Please refer to FIG1 , which shows a schematic diagram of a network architecture 100 provided by an embodiment of the present application. The network architecture 100 may include: a terminal device 10 , an access network device 20 , and a core network device 30 .

[0040] The terminal device 10 may refer to a user equipment (UE), an access terminal device, a user unit, a user station, a mobile station, a mobile station, a remote station, a remote terminal device, a mobile device, a wireless communication device, a user agent, or a user device. Optionally, the terminal device 10 may also be a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA), a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a fifth generation mobile communication system (5GS) or a terminal device in a future evolved public land mobile communication network (PLMN), etc., and the embodiments of the present application are not limited thereto. For ease of description, the devices mentioned above are collectively referred to as terminal devices. The number of terminal devices 10 is generally multiple, and one or more terminal devices 10 may be distributed in a cell managed by each access network device 20.

[0041] The access network device 20 is a device deployed in the access network to provide wireless communication functions for the terminal device 10. The access network device 20 may include various forms of macro base stations, micro base stations, relay stations, access points, etc. In systems using different wireless access technologies, the names of devices with access network device functions may be different. For example, in the 5G NR system, it is called gNodeB or gNB (next Generation Node B, next generation node B (or new generation access network node)). With the evolution of communication technology, the name "access network device" may change. For the convenience of description, in the embodiments of the present application, the above-mentioned devices that provide wireless communication functions for the terminal device 10 are collectively referred to as access network devices. Optionally, a communication relationship can be established between the terminal device 10 and the core network device 30 through the access network device 20. For example, in a Long Term Evolution (LTE) system, the access network device 20 may be an Evolved Universal Terrestrial Radio Access Network (EUTRAN) or one or more eNodeBs in the EUTRAN. In a 5G NR system, the access network device 20 may be a Radio Access Network (RAN) or one or more gNBs in the RAN. In the embodiments of the present application, unless otherwise specified, the network device referred to herein refers to the access network device 20, such as a base station.

[0042] The core network device 30 is a device deployed in the core network. The function of the core network device 30 is mainly to provide user connection, user management, and service carrying, and to provide an interface to the external network as a bearer network. For example, the core network equipment in the 5G NR system may include an access and mobility management function (AMF) network element, an authentication server function (AUSF) network element, a user plane function (UPF) network element, a session management function (SMF) network element, a location management function (LMF) network element, a policy control function (PCF) network element, a unified data management (UDM) network element, etc.

[0043] In one example, the access network device 20 and the core network device 30 communicate with each other via an interface technology, such as the NG interface in the 5G NR system. The access network device 20 and the terminal device 10 communicate with each other via an air interface technology, such as the Uu interface.

[0044] Please refer to Figure 2, which shows a schematic diagram of a network architecture provided by another embodiment of the present application. The network architecture may include: terminal equipment, access network equipment and core network equipment.

[0045] Among them, terminal equipment (including user equipment, UE), access network supporting 3GPP technology (including Radio Access Network, RAN or Access Network, AN), user plane function (User Plane Function, UPF) network element, access and mobility management function (Access and Mobility Management Function, AMF) network element, session management function (Session Management Function, SMF) network element, policy control function (Policy Control Function, PCF) network element, application function (Application Function, AF), data network (Data Network, DN), network slice selection function (Network Slice Selection Function, NSSF), authentication service function (Authentication Server Function, AUSF), unified data management function (Unified Data Management, UDM).

[0046] Those skilled in the art will appreciate that the 5G network architecture shown in FIG2 does not constitute a limitation on the 5G network architecture. In specific implementations, the 5G network architecture may include more or fewer network elements than shown, or may combine certain network elements. It should be understood that in FIG2, the AN or RAN is represented by (R)AN.

[0047] Access network equipment is the device that connects terminal devices to the network architecture wirelessly. It is primarily responsible for radio resource management, Quality of Service (QoS) management, data compression and encryption, etc. on the air interface side. Examples include NodeBs, evolved eNodeBs, base stations in 5G mobile communication systems or next-generation wireless (NR) communication systems, and base stations in future mobile communication systems.

[0048] The core network equipment includes NSSF (Network Slice Selection Function), AUSF (Authentication Server Function), UDM (Unified Data Management), AMF (Access and Mobility Management Function), SMF (Session Management Function), PCF (Policy Control Function), and UPF (User Plane Function).

[0049] The UE establishes an access layer connection with the (R)AN (Access Network) through the Uu interface, exchanging access layer messages and wireless data transmission. The UE establishes a non-access layer (None Access Stratum, NAS) connection with the AMF through the N1 interface, exchanging NAS messages. The AMF is the mobility management function in the core network, and the SMF is the session management function in the core network. In addition to managing the mobility of the UE, the AMF is also responsible for forwarding session management-related messages between the UE and the SMF. The PCF is the policy management function in the core network, responsible for formulating policies related to the mobility management, session management, and billing of the UE. The PCF transmits data with the external application function (AF) through the N5 interface. The UPF is a user plane function in the core network, which transmits data with the external data network (DN) through the N6 interface and with the AN through the N3 interface.

[0050] The "5G NR system" in the embodiments of the present application may also be referred to as a 5G system or an NR system, but those skilled in the art will understand its meaning. The technical solutions described in the embodiments of the present application may be applicable to LTE systems, 5G NR systems, subsequent evolution systems of 5G NR systems, and other communication systems such as Narrow Band Internet of Things (NB-IoT) systems, and this application does not limit this.

[0051] Please refer to Figure 3, which shows a flowchart of a model control method provided by an embodiment of the present application. The method can be applied to a first communication device or a second communication device. For example, the first communication device can be a terminal device, an access network device, or a core network device in the system architecture shown in Figure 1, and the second communication device can be a terminal device, an access network device, or a core network device in the system architecture shown in Figure 1. The method includes the following steps.

[0052] Step 210: Transmit a model control message, where the model control message is used to control the AI / ML model. The model control message includes an AI / ML model identifier of the AI / ML model.

[0053] Optionally, the first communication device and the second communication device transmit a model control message.

[0054] The first communication device may be a first terminal device or a first network device. The second communication device may be a second terminal device or a second network device. The first network device may be a first access network device or a first core network device. The second network device may be a second access network device or a second core network device.

[0055] The first core network device is any one of the following: AMF network element, AUSF network element, UPF network element, SMF network element, LMF network element, PCF network element, and UDM network element. The second core network device is any one of the following: AMF network element, AUSF network element, UPF network element, SMF network element, LMF network element, PCF network element, and UDM network element.

[0056] Model control messages are used to manage and control AI / ML models, or to perform at least one control operation on an AI / ML model. For example, model control messages have at least one of the following functions: model control messages are used to activate, deactivate, switch, delete, or update an AI / ML model; and / or model control messages are used to transmit AI / ML model-related information about the AI / ML model; and / or model control messages are used to transmit a mapping between a globally unique identifier of an AI / ML model and a region-defined identifier.

[0057] Optionally, the model control message in the embodiment of the present application includes at least one communication message related to the AI / ML model transmitted between the first communication device and the second communication device. Alternatively, the model control message includes at least one communication message including an AI / ML model identifier transmitted between the first communication device and the second communication device.

[0058] Optionally, the present application provides ten exemplary model control messages: a first message, a second message, a third message, a fourth message, a fifth message, a sixth message, a seventh message, an eighth message, a ninth message, and a tenth message. These ten exemplary model control messages will be described separately in subsequent embodiments. Of course, model control messages are not limited to these ten examples. Model control messages can also refer to other messages used to control / manage AI / ML models.

[0059] Optionally, the model control message (any message from the first message to the tenth message) may be a unicast message, a multicast message, or a broadcast message.

[0060] Unicast message (one-to-one): The source transmits a unicast message through a unicast channel. Only communication devices that have been allocated the corresponding unicast resources can attempt to receive the unicast message. Unicast messages can also be called dedicated signaling.

[0061] Multicast message (one-to-many): The source transmits the multicast message through the multicast channel. Communication devices within the coverage area of ​​the multicast signal and that are group members can attempt to receive the multicast message. When a communication device joins a group, it obtains multicast communication-related resources.

[0062] Broadcast message (one-to-any): The source transmits the broadcast message through the broadcast channel, and any terminal device within the coverage of the broadcast signal can attempt to receive the broadcast message.

[0063] The above-mentioned communication equipment is any one of terminal equipment, access network equipment or core network equipment.

[0064] Optionally, the model control message (any one of the first to tenth messages) can be any one of the following message types: NAS (Non Access Stratum) message, RRC (Radio Resource Control) message, MAC CE (Media Access Control Control Element) message, DCI (Downlink Control Information) message, UCI (Uplink Control Information) message, PUSCH (Physical Uplink Shared Channel) message, Inter-node message, Xn message, F1 message, E1 message, NG message, core network bus message.

[0065] The model control message includes the AI / ML model identifier of the AI / ML model, which indicates the AI / ML model to be controlled / managed. Optionally, the AI / ML model identifier is associated with the AI / ML model description information associated with the AI / ML model.

[0066] In summary, the method provided in this embodiment defines an AI / ML model identifier for an AI / ML model and uses the AI / ML model identifier to distinguish different AI / ML models. This control and management of a specific AI / ML model is achieved by transmitting a model control message that carries the AI / ML model identifier.

[0067] By way of example, the definition of AI / ML model identifiers and various model control messages are described respectively.

[0068] 1. Definition of AI / ML model identifier

[0069] The AI / ML model-related information includes the AI / ML model identifier (AI / MLModelID) and / or AI / ML model description information. Optionally, the AI / ML model-related information also includes the algorithm data information of the AI / ML model itself (such as the structure data and weight parameters of the algorithm corresponding to the AI / ML model).

[0070] There is an association between the AI / ML model identifier and the AI / ML model description information associated with the AI / ML model.

[0071] AI / ML model description information includes at least one of the following:

[0072] Information 1: Functional characteristics associated with AI / ML models;

[0073] Information 2: Information about the types of input parameters required by the AI / ML model;

[0074] Information 3: Information about the input parameter format required by the AI / ML model;

[0075] Information 4: Input parameter preprocessing rules required by AI / ML models;

[0076] Information 5: Information about the types of output parameters required by the AI / ML model;

[0077] Information 6: Output parameter format requirements required by the AI / ML model;

[0078] Information 7: Output parameter preprocessing rules required by AI / ML models;

[0079] Information 8: Application scenario information of AI / ML models;

[0080] Information 9: Deployment location of the AI / ML model;

[0081] Information 10: Ability requirements for using AI / ML models;

[0082] Information 11: Performance monitoring indicator information of AI / ML models;

[0083] Information 12: Location of AI / ML models;

[0084] Information 13: Information on the effective use scope of AI / ML models;

[0085] Information 14: Generalization characteristics of AI / ML models;

[0086] Information 15: Version number of AI / ML model algorithm data;

[0087] Information 16: AI / ML model algorithm accuracy level information;

[0088] Information 17: Compilation or storage format information of AI / ML model algorithm data.

[0089] The following describes the descriptions of the 17 AI / ML models mentioned above:

[0090] Information 1: Functional characteristics associated with AI / ML models.

[0091] The functional characteristics associated with the AI / ML model are used to indicate the functions associated with the corresponding AI / ML model. For example, AI / ML model 1 is used for CSI (Channel State Information) compression feedback, AI / ML model 2 is used for mobility enhancement, AI / ML model 3 is used for beam prediction, and AI / ML model 4 is used for positioning.

[0092] Information 2: Information about the types of input parameters required by the AI / ML model.

[0093] The information about the types of input parameters required by the AI / ML model indicates the types of input parameters required by the corresponding AI / ML model. For example, AI / ML model 1 is used for CSI compression feedback, and the required input parameters include channel measurement results before channel estimation, channel measurement results after channel estimation, or a channel impulse response matrix obtained based on the channel measurement results.

[0094] For example, AI / ML model 3 is used for beam prediction function, and the types of input parameters it requires include the index information of the actual measured beam, the signal measurement result information of each beam obtained by actual measurement, etc.

[0095] Information 3: Information about the input parameter format required by the AI / ML model.

[0096] The input parameter format required by the AI / ML model requires information to indicate the format of the input parameters required by the corresponding AI / ML model. For example: AI / ML model 3 is used for beam prediction function. Assuming that the network device can schedule resources in 8 beam directions, but only sends measurement reference signals in 4 of the beam directions, the terminal device can only obtain the measurement results of 4 beams in each measurement process. Scenario 1: The beam prediction AI / ML model only predicts the signal measurement results of other beams that are not measured at the current moment. If the beam prediction AI / ML model used in scenario 1 only needs to input the measurement results of 4 beams, then the measurement results of the 4 beams obtained by the terminal device in each measurement can just be used as the input of the beam prediction AI / ML model in scenario 1 (the output can obtain the measurement results of the 8 beams associated with this measurement, that is: 4 true values, 4 predicted values, or the output is the predicted measurement results of 4 beams); Scenario 2: The beam prediction AI / ML model is used to predict the measurement results of 8 beams in a period of time in the future. In this scenario, the input of the beam prediction AI / ML model may be a set of measurement results of 4 beams obtained in the past N measurement processes, and the output of the beam prediction AI / ML model is a set of measurement results of 8 beams in the future. For example, the input of the beam prediction AI / ML model is a set of three sets of measurement results of 4 beams obtained at time T-3, T-2, and T-1, and the output of the beam prediction AI / ML model is a set of two sets of measurement results of 8 beams at time T and time T+1.

[0097] In summary, although the measurement results of the four beams are used for beam prediction (the model input data type is the same), different application scenarios require different input data formats. Therefore, the input parameter format requirements required by the AI / ML model are also one of the key information for describing an AI / ML model.

[0098] Information 4: Input parameter preprocessing rules required by AI / ML models.

[0099] The input parameter preprocessing rules required by the AI / ML model are used to indicate how to reprocess the input data set of the corresponding AI / ML model. The operations involved in the reprocessing process may include at least one of data dimensionality reduction, data dimensionality increase, redundancy elimination, and redistribution.

[0100] Data dimensionality reduction: If the data dimension is higher than the dimension required by the model input, the data needs to be reduced in dimension through preprocessing rules before entering the model;

[0101] Data dimensionality increase: If the data dimension is lower than the dimension required by the model input, the data needs to be increased in dimension through preprocessing rules before entering the model;

[0102] Redundancy elimination: If the distribution or information content of multiple sets of data is similar, redundancy needs to be eliminated through preprocessing rules before entering the data into the model. This is because redundant data often does not bring more performance improvements but instead increases ineffective operations.

[0103] Redistribution: In some cases, it is difficult to obtain good model generalization by training data with similar feature information together. Redistributing data with different feature information is conducive to obtaining a model with better generalization.

[0104] For example, for scenario 1 in information 3, the terminal device can measure 4 of the 8 beams in cell A, but can only measure 5 of the 8 beams in cell B and 3 of the 8 beams in cell C. If you want to use the same beam prediction AI / ML model (only 4 beam measurement results are required as input) for beam prediction in cells A, B, and C, the measurement results of cell A can be used directly, but the measurement results obtained in cell B need to be reduced in data dimension, that is, 4 measurement results are selected from the 5 available beam measurement results as the input parameters required by the AI / ML model, and the measurement results obtained in cell C need to be increased in data dimension, that is, the measurement results of 3 available beams are converted to 4 to obtain the input parameters required by the AI / ML model.

[0105] In another implementation, at least two of information 2, information 3, and information 4 may be defined in combination, and this application does not impose any restrictions on this.

[0106] Information 5: Information about the types of output parameters required by the AI / ML model.

[0107] The output parameter type information required by the AI / ML model indicates the type of output parameter required by the corresponding AI / ML model. For example, if AI / ML model 3 is used for beam prediction, the corresponding output parameter types include the index information of the actual unmeasured beam and the predicted signal measurement results of each actual unmeasured beam.

[0108] Information 6: Output parameter format requirements required by the AI / ML model.

[0109] The output parameter format requirement information required by the AI / ML model is used to indicate the format of the output parameters required by the corresponding AI / ML model. For example: AI / ML model 3 is used for beam prediction function, and the output parameter can be the beam prediction result at one future moment, or the output parameter can be a set of beam prediction results at multiple future moments. For example, the model input is a set of three sets of measurement results of 4 beams obtained at times T-3, T-2, and T-1, and the model output can be the measurement results of 8 beams at time T, or a set of two sets of measurement results of 8 beams at times T and T+1.

[0110] In summary, although the prediction of future beam measurement results is performed (the model output data type is the same), the output data format required by AI / ML model 3 in different application scenarios is different. Therefore, the output parameter format requirement information required by the AI / ML model is also one of the key information for describing an AI / ML model.

[0111] Information 7: Output parameter preprocessing rules required by AI / ML models.

[0112] The output parameter preprocessing rules required by AI / ML models specify how to reprocess the output data set of the corresponding AI / ML model. This reprocessing process may involve at least one of data dimensionality reduction, data dimensionality increase, redundant elimination, and redistribution. For specific processing methods, see the relevant instructions in Information 4.

[0113] In another implementation, at least two of information 5, information 6, and information 7 may be defined in combination, and this application does not impose any limitation on this.

[0114] Information 8: Application scenario information of AI / ML models.

[0115] The application scenario information of the AI / ML model refers to different usage scenarios under the same functional characteristics. For example, for the AI / ML model used for positioning, the application scenario can be further subdivided into a line of sight (LoS) scenario, a non-line of sight (NLoS) scenario, or a mixed scenario (including both LoS and NLoS). For another example, for the AI / ML model used for mobility control, the application scenario can be further subdivided into ground communication scenarios and non-ground communication scenarios (including satellite communication, drone communication, hot air balloon communication and other sub-scenarios). Each scenario can be further divided into high-speed mobile scenarios, medium-speed mobile scenarios, low-speed mobile scenarios and other sub-scenarios, and can also be divided into high-frequency mobility scenarios, low-frequency mobility scenarios and other sub-scenarios. It is also possible to consider a combination of two scenarios, such as a ground communication high-frequency mobility scenario.

[0116] Optionally, information 8 may have a value indicating that the application scenario of the AI / ML model is all scenarios (i.e., the AI / ML model is applicable to any scenario corresponding to its associated functional characteristics). For example, for an AI / ML model used for positioning, assuming that information 8 contains two bits, the value '00' can be used to indicate any scenario, '01' to indicate a LoS scenario, '10' to indicate an NLoS scenario, and '11' to indicate a mixed scenario.

[0117] Information 9: Deployment location information of the AI / ML model.

[0118] The deployment location information of the AI / ML model includes any of the following meanings: deployment on core network equipment, deployment on application servers (OTT servers), deployment on OAM (Operation Administration and Maintenance), deployment on access network equipment, deployment on terminal equipment, or no deployment location restrictions.

[0119] Information 10: Ability requirements for using AI / ML models.

[0120] The capability requirements for using AI / ML models include at least one of the following: the minimum floating-point computing capability required when using the AI / ML model, the minimum running memory required when using the AI / ML model, the memory size level occupied by the AI / ML model algorithm data (before compilation), the memory size level occupied by the AI / ML model algorithm data after compilation, the minimum input data rate (or bit rate) required when using the AI / ML model, the maximum input data rate (or bit rate) required when using the AI / ML model, the minimum output data rate (or bit rate) required when using the AI / ML model, and the maximum output data rate (or bit rate) required when using the AI / ML model.

[0121] Information 11: Performance monitoring indicator information of AI / ML models.

[0122] The performance monitoring indicator information of the AI / ML model includes at least one of the following: average energy consumption of a single inference process, average floating-point calculation amount of a single inference process, average latency of a single inference process, system throughput, system bit error rate, and average data packet transmission latency.

[0123] Information 12: Location information of the AI / ML model.

[0124] The location information of the AI / ML model is used to indicate the relevant information of the provider of the AI / ML model data, specifically including: the country and region information to which the AI / ML model data belongs and / or the service provider information to which the AI / ML model data belongs.

[0125] Among them, the service provider can be any of the following: OTT (Over The Top, providing application services to users through the Internet) manufacturer information, network operator information (for example: represented by PLMN (Public Land Mobile Network) identification), and public website information.

[0126] Information 13: Information about the effective use scope of AI / ML models.

[0127] The effective use scope information of the AI / ML model includes at least one of the following: geographical area information allowed for effective use, country information allowed for effective use, PLMN information allowed for effective use, tracking area code TAC (Tracking Area Code) information allowed for effective use, RANAC (RAN Area Code, access network area code) information allowed for effective use, and cell information allowed for effective use.

[0128] Information 14: Generalization characteristics of AI / ML models.

[0129] The generalization characteristic reflects the breadth of applicable scenarios of an AI / ML model. Generally speaking, the higher the generalization characteristic of an AI / ML model, the more application scenarios the model is applicable to. The generalization characteristic information of an AI / ML model can include at least one application scenario information of the AI / ML model (see the definition of Information 8), which is used to indicate that the corresponding associated AI / ML model can be used in the above-mentioned application scenarios.

[0130] Information 15: Version number information of AI / ML model algorithm data.

[0131] As different parameters in the same application scenario change over time, the algorithm data of the AI / ML model with the corresponding function may need to be fine-tuned or updated. The updated AI / ML model is associated with the same application scenario and has the same function as the AI / ML model before the update, but the model algorithm data details are slightly different. Therefore, the model algorithm version number can be used to distinguish them, which facilitates the control and management of similar AI / ML model algorithms.

[0132] Information 16: Accuracy level information of AI / ML model algorithm.

[0133] Even for AI / ML models with the same function and application scenario, the complexity of the model algorithm design is different, and the reasoning accuracy of the corresponding AI / ML models will also be different. In order to distinguish the differences in reasoning accuracy among similar AI / ML models, the accuracy level information of the AI / ML model algorithm can be used to describe the reasoning accuracy of the corresponding AI / ML model.

[0134] Information 17: Compilation or storage format information of AI / ML model algorithm data.

[0135] AI / ML model algorithms are essentially programs written in computer languages. The types of program files written in different language platforms are different. Some languages ​​are compatible with each other, while others are not. Therefore, the compilation format information of AI / ML model algorithm data is also important information for describing an AI / ML model. This information allows AI / ML model users to know the compilation environment required to use the AI / ML model. Otherwise, incorrect use of an incompatible language compilation environment will result in the inability to use the corresponding AI / ML model algorithm.

[0136] Based on the above AI / ML model description information, the following methods are used to define AI / ML model identifiers:

[0137] Definition method 1:

[0138] The AI / ML model identifier includes an information field, and the AI / ML model identifier is associated with information 1.

[0139] The AI / ML model identifier is only used to distinguish different functional features (Features). That is, the AI / ML model identifier can be used to determine the functional features associated with the corresponding AI / ML model. The relationship between the AI / ML model identifier and the functional features is agreed upon by the protocol default method (that is, the specific relationship is explicitly stated in the protocol). Under definition method 1, the model identifier is an overall information domain and cannot be divided into sub-information domains.

[0140] For example, the AI / ML model used for CSI compression feedback has a model identifier of ID1, the AI / ML model used for beam prediction has a model identifier of ID2, and the AI / ML model used for positioning has a model identifier of ID3. Based on an AI / ML model identifier, its associated functional characteristics can be obtained. For example, if an AI / ML model has an AI / ML model identifier of ID1, then the AI / ML model is used for CSI compression feedback.

[0141] Definition 2:

[0142] The AI / ML model identifier includes an information field. The AI / ML model identifier is associated with information 1, and the AI / ML model identifier is associated with at least one of information 2 to information 17.

[0143] The AI / ML model identifier is associated not only with the functional characteristics (Information 1) but also with at least one of the aforementioned Information 2 through Information 17. Specifically, a single AI / ML model identifier reveals not only the functional characteristics associated with the AI / ML model, but also the corresponding sub-information describing the AI / ML model (i.e., at least one of Information 2 through Information 17). This association is defined by default in the protocol. Under Definition Method 2, the model identifier is a single, integrated information domain, and cannot be divided into sub-information domains.

[0144] For example, assuming the AI / ML model identifier is associated with both information 1 and information 8, the protocol stipulates that the AI / ML model used for positioning in LoS scenarios has model identifier ID31, the AI / ML model used for positioning in NLoS scenarios has model identifier ID32, and the AI / ML model used for positioning in a mixed LoS and NLoS scenario has model identifier ID33. If the model identifier for an AI / ML model is ID32, then the AI / ML model is used for positioning and its application scenario is NLoS.

[0145] Definition 3:

[0146] The AI / ML model identifier includes at least two information fields, and each of the at least two information fields is associated with at least one of information 1 to information 17.

[0147] The AI / ML model identifier is not only associated with functional characteristics, but can also be clearly divided into at least two AI / ML model identifier sub-information domains. Each information domain is associated with at least one of the above-mentioned information 1 to information 17, and each information domain (or sub-information domain) contains at least one bit.

[0148] Definition method 3 can be further divided into the following two sub-definition methods:

[0149] Definition method 3-1: The at least two information domains identified by the AI / ML model include a first information domain and at least one additional information domain; the first information domain is associated with information 1; and each additional information domain in the at least one additional information domain is associated with at least one of information 2 to information 17.

[0150] The AI / ML model identifier includes information domain 1 (or the first information domain), which is only associated with the functional characteristics of the AI / ML model. The AI / ML model identifier also includes at least one additional information domain, each of which is associated with at least one of the above-mentioned information 2 to information 17. The association between different additional information domains and at least one of the above-mentioned information 2 to information 17 is agreed upon by default in the protocol.

[0151] For example, the AI / ML model identifier is shown in Table 1 and includes Information Field 1, Information Field 2, and Information Field 3. Information Field 1 is the first information field, while Information Field 2 and Information Field 3 are additional information fields. Information Field 1 is associated with functional characteristics, while Information Field 2 and Information Field 3 can each be associated with at least one of the aforementioned information fields 2 through 17. For example, Information Field 2 is associated with Information 8, and Information Field 3 is associated with Information 9 and Information 10.

[0152] Table 1 AI / ML model identification definition method 3-1

[0153]

[0154] Definition method 3-2: The at least two information domains identified by the AI / ML model include a first information domain and at least one additional information domain; the first information domain is associated with information 1, and the first information domain is associated with at least one of information 2 to information 17; each additional information domain in the at least one additional information domain is associated with at least one of information 2 to information 17.

[0155] The AI / ML model identifier includes information domain 1 (or the first information domain). Information domain 1 is associated not only with the functional characteristics associated with the AI / ML model, but also with at least one of the above-mentioned information 2 to information 17. At the same time, the AI / ML model identifier also includes at least one additional information domain. Each additional information domain is associated with at least one of the above-mentioned information 2 to information 17. The association relationship between different information domains and at least one of the above-mentioned information 2 to information 17 is agreed upon by default in the protocol.

[0156] For example, the AI / ML model identifier is shown in Table 2 and includes Information Field 1, Information Field 2, and Information Field 3. Information Field 1 is the first information field described above, while Information Field 2 and Information Field 3 are additional information fields. Information Field 1 is associated with the AI / ML model's functional characteristics and at least one of the aforementioned information fields 2 through 17. Information Field 2 and Information Field 3 can each be associated with at least one of the aforementioned information fields 2 through 17. For example, Information Field 1 is associated with Information 1 and Information 8, Information Field 2 is associated with Information 9 and Information 10, and Information Field 3 is associated with Information 11.

[0157] Table 2 AI / ML model identification definition method 3-2

[0158]

[0159] Of course, AI / ML model identification is not limited to the examples in Table 1 and Table 2 above. There are many other examples of AI / ML model identification, which are not listed here one by one.

[0160] Optionally, regardless of the above-mentioned definition method of the AI / ML model identifier, the AI / ML model identifier can be a globally unique identifier or a regionally defined identifier.

[0161] Globally unique identifier: The AI / ML model identifier has a universal meaning. For example, if the AI / ML model identifier for positioning is ID3, then AI / ML model identifier 3 can only be used to indicate the AI / ML model corresponding to positioning in any operator network worldwide.

[0162] Region-defined identifier: A region-defined identifier can also be called an operator-defined identifier. A region-defined identifier is only valid for a specific operator, and its meaning is usually defined by the operator.

[0163] Optionally, the region definition identifier value range is generally strictly distinguished from the global unique identifier value range. A simple way to distinguish them is to define a dedicated value field for the region definition identifier value, so as to avoid misunderstandings caused by the same values ​​of the two identifiers.

[0164] In summary, the method provided in this embodiment divides the definition of the AI / ML model identifier into several sub-information domains, and can use different message formats (long and short formats) according to different scenarios to achieve flexible control of operations such as activation, deactivation, switching, and deletion of the AI / ML model.

[0165] Exemplarily, based on the above definition of AI / ML model identification, function control based on AI / ML model identification can be implemented.

[0166] 2. Functional control based on AI / ML model identification.

[0167] The present application embodiment exemplarily provides six types of function control based on AI / ML model identification:

[0168] Function 1: Activation;

[0169] Function 2: Deactivation;

[0170] Function 3: Switch;

[0171] Function 4: delete;

[0172] Function 5: Obtain information related to the AI / ML model associated with the second communication device;

[0173] Function 6: Obtain information related to the AI / ML model after local mapping on the second communication device.

[0174] Function 1: Activation.

[0175] Please refer to Figure 4, which shows a flowchart of a model control method provided by one embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0176] Step 301: A first communication device sends a first message, where the first message is used for activation of an AI / ML model.

[0177] The second communication device receives the first message sent by the first communication device. Optionally, the first message can be any one of a unicast message, a multicast message, and a broadcast message.

[0178] Implementation method 1: If the AI / ML model identifier adopts the above-mentioned AI / ML model identifier definition method 1 or definition method 2, the AI / ML model identifier field contained in the AI / ML model activation information field included in the first message has only one information format, namely: the definition format of the AI / ML model identifier.

[0179] That is, the first message includes an AI / ML model activation information field and the AI / ML model activation information field includes an AI / ML model identification field; when the AI / ML model identification includes an information field, the information format of the AI / ML model identification field contained in the AI / ML model activation information field is the defined format of the AI / ML model identification.

[0180] Implementation method 2: If the AI / ML model identifier adopts the above-mentioned definition method 3 of the AI / ML model identifier (including definition method 3-1 or definition method 3-2), then the information format of the AI / ML model identification field contained in the AI / ML model activation information field contained in the first message has at most 2 to the power of n minus 1, where n represents the number of information fields in the AI / ML model identifier definition.

[0181] That is, the first message includes an AI / ML model activation information field and the AI / ML model activation information field includes an AI / ML model identification field; when the AI / ML model identification field includes at least two information fields, the information format of the AI / ML model identification field included in the AI / ML model activation information field includes at most (2 n -1) types, where n is the number of information domains identified by the AI / ML model.

[0182] Optionally, regardless of which definition method (any one of definition methods 1 to 3) is used for the AI / ML model identifier, the AI / ML model activation information field included in the first message also includes at least one other information field (information field other than the AI / ML model identifier information field), and each other information field contains at least one of the aforementioned information 2 to information 17; optionally, each other information field may also include at least one of cell identification information, frequency identification information, and BWP (Bandwidth Part, partial bandwidth) identification information.

[0183] Optionally, each of the at least one other information field included in the above-mentioned AI / ML model activation information field is an optional field or a mandatory field. Assuming that the AI / ML model activation information field includes m (m is a natural number) other information fields, the information format of the AI / ML model activation information field included in the first message has at most (2 n+m -1) types, where n is the number of information domains identified by the AI / ML model.

[0184] Exemplarily, the first communication device can flexibly choose which information format of the AI / ML model activation information domain to use to activate the corresponding AI / ML model in the second communication device according to actual needs. For example: the AI / ML model identifier is defined as in Table 1 above. If there is only one AI / ML model associated with information domain 1 in the second communication device, then the AI / ML model activation information domain sent by the first communication device only needs to include AI / ML model identification information domain 1, and information domain 2, information domain 3 and other information domains may not be included; if there are at least two AI / ML models associated with information domain 1 in the second communication device, then the AI / ML model activation information domain sent by the first communication device, in addition to including AI / ML model identification information domain 1, also needs to include at least one of information domain 2, information domain 3 and other information domains according to the adaptability of the model differentiation requirements.

[0185] Exemplarily, the information format of the AI / ML model activation information domain includes: x information domains in at least one information domain (including other information domain cases) are mandatory domains of the AI / ML model activation information domain, and information domains other than the above x information domains in at least one information domain are optional domains of the AI / ML model activation information domain, where x is a positive integer.

[0186] For example, Table 3 illustrates the format definition of an AI / ML model activation information field included in a first message. The first message itself may include a message header. Table 3 only describes the AI / ML model activation information field included in the first message. Field 1 is mandatory, while the remaining fields are optional. "O" indicates "Optional." The "O" field occupies one bit and indicates whether the corresponding information field is present.

[0187] Table 3 Schematic diagram of the format definition of the AI / ML model activation information field contained in the first message 1

[0188]

[0189] For another example, Table 4 shows another format definition of the AI / ML model activation information field included in the first message. In Table 4, information field 1 and information field 2 must be present, while the remaining information fields are optional.

[0190] Table 4 Schematic 2 of the format definition of the AI / ML model activation information field contained in the first message

[0191]

[0192] That is, in the format definition of the AI / ML model activation information field included in the first message in the second implementation method, at least one field is a mandatory field, and the remaining fields are optional fields.

[0193] It should be noted that, when the definition of the AI / ML model identifier includes n information fields and the AI / ML model activation information field also includes m other information fields, the information format of the AI / ML model activation information field included in the first message of the above implementation method 2 in theory is at most (2 n+m -1) types, but considering the effectiveness of practical applications, some formats are meaningless. For example, the information format lacking Information Field 1 in Table 3 cannot be associated with functional characteristics, and the second communication device cannot determine which specific AI / ML model to activate. Obviously, this format needs to be excluded. The specific format definition of the information format of the AI / ML model activation information field contained in the first message can be given one by one through protocol default, but at least one format must be defined.

[0194] Optionally, the first message is used to activate at least one AI / ML model. That is, the type (number) of AI / ML models activated by one first message can be one or more.

[0195] Optionally, the first message is a dedicated message for AI / ML model activation. For example, the first message can be called a model activation message.

[0196] Alternatively, the first message is a general AI / ML model control message, and the first message includes an AI / ML model activation information field for an activation operation. For example, the first message may be called a model control message, and the model control message includes multiple AI / ML model control information fields corresponding to different control operations, such as an AI / ML model activation information field corresponding to an activation operation.

[0197] In summary, the method provided in this embodiment implements the activation operation of the specified AI / ML model by transmitting a first message carrying the AI / ML model identifier.

[0198] Function 2: Deactivate.

[0199] Please refer to Figure 5, which shows a flowchart of a model control method provided by an embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0200] Step 302: The first communication device sends a second message, where the second message is used for deactivation of the AI / ML model.

[0201] The second communication device receives the second message sent by the first communication device. Optionally, the second message can be any one of a unicast message, a multicast message, and a broadcast message.

[0202] Implementation method 1: If the AI / ML model identifier adopts the above-mentioned AI / ML model identifier definition method 1 or definition method 2, the AI / ML model identifier field contained in the AI / ML model deactivation information field included in the second message has only one information format, namely: the definition format of the AI / ML model identifier.

[0203] That is, the second message includes an AI / ML model deactivation information field and the AI / ML model deactivation information field includes an AI / ML model identification field; when the AI / ML model identification includes an information field, the information format of the AI / ML model identification field contained in the AI / ML model deactivation information field is the definition format of the AI / ML model identification.

[0204] Implementation method 2: If the AI / ML model identifier adopts the above-mentioned definition method 3 of the AI / ML model identifier (including definition method 3-1 or definition method 3-2), then the information format of the AI / ML model identifier field contained in the AI / ML model deactivation information field contained in the second message has at most 2 to the power of n minus 1, where n represents the number of information fields in the AI / ML model identifier definition.

[0205] That is, the second message includes an AI / ML model deactivation information field and the AI / ML model deactivation information field includes an AI / ML model identification field; when the AI / ML model identification field includes at least two information fields, the information format of the AI / ML model identification field included in the AI / ML model deactivation information field includes at most (2 n -1) types, where n is the number of information domains identified by the AI / ML model.

[0206] Optionally, regardless of which definition method (any one of definition method 1 to definition method 3) is used for the AI / ML model identifier, the AI / ML model deactivation information field included in the second message also includes at least one other information field (information field other than the AI / ML model identifier information field), and each other information field contains at least one of the aforementioned information 2 to information 17; optionally, each other information field may also include at least one of cell identification information, frequency identification information, and BWP identification information.

[0207] Optionally, each of the at least one other information field included in the above-mentioned AI / ML model deactivation information field is an optional field or a mandatory field. Assuming that the AI / ML model deactivation information field includes m (m is a natural number) other information fields, the information format of the AI / ML model deactivation information field included in the second message has at most (2 n+m -1) types, where n is the number of information domains identified by the AI / ML model.

[0208] Exemplarily, the first communication device can flexibly choose which AI / ML model deactivation information domain information format to use to activate the corresponding AI / ML model in the second communication device according to actual needs. For example: the AI / ML model identifier is defined in the above Table 1. If there is only one AI / ML model associated with information domain 1 in the second communication device, then the AI / ML model deactivation information domain sent by the first communication device only needs to include AI / ML model identification information domain 1, and information domain 2, information domain 3 and other information domains may not be included; if there are at least two AI / ML models associated with information domain 1 in the second communication device, then the AI / ML model deactivation information domain sent by the first communication device, in addition to including AI / ML model identification information domain 1, also needs to include at least one of information domain 2, information domain 3 and other information domains according to the model differentiation requirement adaptability.

[0209] Exemplarily, the information format of the AI / ML model deactivation information domain includes: y information domains in at least one information domain (including other information domain situations) are mandatory domains of the AI / ML model deactivation information domain, and information domains other than the above y information domains in at least one information domain are optional domains of the AI / ML model deactivation information domain, and y is a positive integer.

[0210] That is, in the format definition of the AI / ML model deactivation information field included in the second message in the second implementation method, at least one field is a mandatory field, and the remaining fields are optional fields.

[0211] It should be noted that, when the definition of the AI / ML model identifier includes n information fields and the AI / ML model deactivation information field also includes m other information fields, the information format of the AI / ML model deactivation information field included in the second message of the above implementation method 2 is theoretically at most (2 n+m -1) types, but considering the effectiveness of practical applications, some formats are meaningless. For example, the information format lacking Information Field 1 in Table 3 cannot be associated with functional characteristics, and the second communication device cannot determine which specific AI / ML model needs to be deactivated. Obviously, this format needs to be excluded. The specific format definition of the information format of the AI / ML model deactivation information field contained in the second message can be given one by one through protocol default, but at least one format must be defined.

[0212] Optionally, the second message is used to deactivate at least one AI / ML model. That is, the type (number) of AI / ML models deactivated by one second message may be one or more.

[0213] Optionally, the second message is a dedicated message for deactivating the AI / ML model. For example, the second message may be called a model deactivation message.

[0214] Alternatively, the second message is a general AI / ML model control message, and includes an AI / ML model deactivation information field for a deactivation operation. For example, the second message may be a model control message, and the model control message includes multiple AI / ML model control information fields corresponding to different control operations, such as an AI / ML model deactivation information field corresponding to a deactivation operation.

[0215] It should be noted that the second message is different from the first message (if they are different messages, the different messages themselves can distinguish whether they are model activation operations or deactivation operations).

[0216] Alternatively, the second message is the same as the first message (if they are the same message, it is necessary to distinguish between model activation and deactivation operations through different information fields of the same message). Under the condition that the first message and the second message are the same message, the activation operation and the deactivation operation correspond to different information fields. In this case, the first message or the second message is used to activate at least one AI / ML model and / or deactivate at least one AI / ML model. Table 5 gives an example of a first message:

[0217] Table 5: Schematic diagram of model activation and deactivation operations implemented through different information fields in the same message

[0218]

[0219] In Table 5, the AI / ML model identifier can adopt any of the above-mentioned definition methods, such as the definition method of Table 3 or Table 4, which will not be repeated here. Specifically, the first information field contained in the first message (i.e., the AI / ML model identifier of at least one AI / ML model to be activated) is used to activate at least one AI / ML model, and the second information field contained in the first message (i.e., the AI / ML model identifier of at least one AI / ML model to be deactivated) is used to deactivate at least one AI / ML model. Optionally, the above-mentioned first information field and second information field are both optional information fields of the first message (AI / ML model control general message).

[0220] In summary, the method provided in this embodiment implements the deactivation operation of the specified AI / ML model by transmitting the second message carrying the AI / ML model identifier.

[0221] Function 3: Switch.

[0222] Please refer to Figure 6, which shows a flowchart of a model control method provided by one embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0223] Step 303: The first communication device sends a third message, where the third message is used for a switching operation (model switch) of an AI / ML model.

[0224] The second communication device receives a third message sent by the first communication device. Optionally, the third message can be any one of a unicast message, a multicast message, and a broadcast message.

[0225] Model switching usually involves activating one model while deactivating another model. There are two specific implementation methods:

[0226] Implementation method 1: Through AI / ML model activation and deactivation functions.

[0227] In this way, the model activation and deactivation functions are implemented through different information fields of the same message, that is, in the implementation method corresponding to Table 5, model switching is indirectly implemented through the model activation and deactivation functions without the need for additional messages or information fields.

[0228] That is, the third message includes an AI / ML model activation information field and an AI / ML model deactivation information field; the AI / ML model activation information field includes an AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation; the AI / ML model deactivation information field includes an AI / ML model identifier of at least one AI / ML model that needs to perform a deactivation operation.

[0229] Implementation method 2: Activate the function through AI / ML model.

[0230] In this way, model switching is achieved through the AI / ML model activation function, and the second communication device itself determines which AI / ML model needs to be deactivated at the same time. For example: the protocol stipulates that the AI / ML model used for the positioning function in the LoS scenario corresponds to the model identifier ID31, the AI / ML model used for the positioning function in the NLoS scenario corresponds to the model identifier ID32, and the AI / ML model used for the positioning function in the mixed LoS and NLoS scenario corresponds to the model identifier ID33. Only one positioning model can be activated at a time. At the beginning, the second communication device uses the positioning model with the model identifier ID32, and then receives the AI / ML model activation information field sent by the third message requesting the activation of the positioning model with the model identifier ID33. Since there is already a positioning model activated on the second communication device, the second communication device will actively deactivate the previously used positioning model while activating another positioning model.

[0231] That is, the third message includes an AI / ML model activation information field; the AI / ML model activation information field includes an AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation.

[0232] In summary, the method provided in this embodiment implements the switching operation of the specified AI / ML model by transmitting the third message carrying the AI / ML model identifier.

[0233] Function 4: Delete.

[0234] Please refer to Figure 7, which shows a flowchart of a model control method provided by one embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0235] Step 304: The first communication device sends a fourth message, where the fourth message is used for deleting the AI / ML model.

[0236] The second communication device receives the fourth message sent by the first communication device. Optionally, the fourth message can be any one of a unicast message, a multicast message, and a broadcast message.

[0237] Implementation method 1: If the AI / ML model identifier adopts the above-mentioned AI / ML model identifier definition method 1 or definition method 2, the AI / ML model identifier field contained in the AI / ML model deletion information field included in the fourth message has only one information format, namely: the definition format of the AI / ML model identifier.

[0238] That is, the fourth message includes an AI / ML model deletion information field and the AI / ML model deletion information field includes an AI / ML model identification field; when the AI / ML model identification includes an information field, the information format of the AI / ML model identification field contained in the AI / ML model deletion information field is the definition format of the AI / ML model identification.

[0239] Implementation method 2: If the AI / ML model identifier adopts the above-mentioned definition method 3 of the AI / ML model identifier (including definition method 3-1 or definition method 3-2), then the information format of the AI / ML model identifier field contained in the AI / ML model deletion information field contained in the fourth message has at most 2 to the power of n minus 1, where n represents the number of information fields in the AI / ML model identifier definition.

[0240] That is, the fourth message includes an AI / ML model deletion information field and the AI / ML model deletion information field includes an AI / ML model identification field; in the case where the AI / ML model identification field includes at least two information fields, the information format of the AI / ML model identification field included in the AI / ML model deletion information field includes at most (2 n -1) types, where n is the number of information domains identified by the AI / ML model.

[0241] Optionally, regardless of which definition method (any one of definition methods 1 to 3) is used for the AI / ML model identifier, the AI / ML model deletion information field included in the fourth message also includes at least one other information field (information field other than the AI / ML model identifier information field), and each other information field contains at least one of the aforementioned information 2 to information 17; optionally, each other information field may also include at least one of cell identification information, frequency identification information, and BWP identification information.

[0242] Optionally, each of the at least one other information field included in the above-mentioned AI / ML model deletion information field is an optional field or a mandatory field. Assuming that the AI / ML model deletion information field includes m (m is a natural number) other information fields, the information format of the AI / ML model deletion information field included in the fourth message has at most (2 n+m-1) types, where n is the number of information domains identified by the AI / ML model.

[0243] Exemplarily, the first communication device can flexibly choose which AI / ML model deletion information format to use to delete the corresponding AI / ML model in the second communication device according to actual needs. For example: the AI / ML model identifier is defined in the above Table 1. If there is only one AI / ML model associated with information domain 1 in the second communication device, then the AI / ML model deletion information domain sent by the first communication device only needs to include AI / ML model identification information domain 1, and information domain 2, information domain 3 and other information domains may not be included; if there are at least two AI / ML models associated with information domain 1 in the second communication device, then the AI / ML model deletion information domain sent by the first communication device, in addition to including AI / ML model identification information domain 1, also needs to include at least one of information domain 2, information domain 3 and other information domains according to the adaptability of the model differentiation requirements.

[0244] Exemplarily, the information format of the AI / ML model deletion information domain includes: z information domains in at least one information domain (including other information domain cases) are mandatory domains of the AI / ML model deletion information domain, and information domains other than the above z information domains in at least one information domain are optional domains of the AI / ML model deletion information domain, and z is a positive integer.

[0245] That is, in the format definition of the AI / ML model deletion information field included in the fourth message in the second implementation method, at least one field is a mandatory field, and the remaining fields are optional fields.

[0246] It should be noted that, when the definition of the AI / ML model identifier includes n information fields and the AI / ML model deletion information field also includes m other information fields, the information format of the AI / ML model deletion information field included in the fourth message of the above implementation method 2 in theory is at most (2 n+m -1) types, but considering the effectiveness of practical applications, some formats are meaningless. For example, the information format lacking Information Field 1 in Table 3 cannot be associated with functional characteristics, and the second communication device cannot determine which specific AI / ML model to delete. Obviously, this format needs to be excluded. The specific format definition of the information format of the AI / ML model deletion information field contained in the fourth message can be given one by one through protocol default, but at least one format must be defined.

[0247] Optionally, the fourth message is used to delete at least one AI / ML model. That is, the type (number) of AI / ML models deleted by one fourth message may be one or more.

[0248] Optionally, the fourth message is a dedicated message for deleting the AI / ML model. For example, the fourth message can be called a model deletion message.

[0249] Alternatively, the fourth message is a general AI / ML model control message, and includes an AI / ML model deletion information field for a deletion operation. For example, the fourth message may be a model control message, and the model control message includes multiple AI / ML model control information fields corresponding to different control operations, such as an AI / ML model deletion information field corresponding to a deletion operation.

[0250] It should be noted that the fourth message is different from the first message (if they are different messages, the different messages themselves can distinguish whether they are model activation operations or deletion operations).

[0251] Alternatively, the fourth message is the same as the first message (if they are the same message, model activation and deletion operations need to be distinguished by different information fields of the same message). If the first message and the fourth message are the same message, the activation operation and the deletion operation correspond to different information fields. In this case, the first message or the fourth message is used to activate at least one AI / ML model and / or delete at least one AI / ML model. Table 6 provides an example of a first message:

[0252] Table 6: Schematic diagram of model activation and deletion operations implemented through different information fields in the same message

[0253]

[0254] In Table 6, the AI / ML model identifier can adopt any of the above-mentioned definition methods, such as the definition method of Table 3 or Table 4, which will not be repeated here. Specifically, the first information field contained in the first message (i.e., the AI / ML model identifier of at least one AI / ML model to be activated) is used to activate at least one AI / ML model, and the third information field contained in the first message (i.e., the AI / ML model identifier of at least one AI / ML model to be deleted) is used to delete at least one AI / ML model. Optionally, the above-mentioned first information field and third information field are both optional information fields of the first message (AI / ML model control general message).

[0255] Optionally, the deactivation operation of the model can also be implemented through different information fields of the same message as the activation operation and the deletion operation, that is, Table 5 and Table 6 are merged, and the first message includes the first information field, the second information field and the third information field, which are not repeated here.

[0256] In summary, the method provided in this embodiment implements the deletion operation of the specified AI / ML model by transmitting the fourth message carrying the AI / ML model identifier.

[0257] Function 5: Obtain information related to the AI / ML model associated with the second communication device.

[0258] Please refer to Figure 8, which shows a flowchart of a model control method provided by one embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0259] Step 306: The second communication device sends a fifth message, where the fifth message includes information related to the AI / ML model associated with the second communication device. The AI / ML model related information includes an AI / ML model identifier and / or AI / ML model description information.

[0260] The first communication device receives a fifth message sent by the second communication device. Optionally, the fifth message may be any one of a unicast message, a multicast message, and a broadcast message.

[0261] Optionally, when the first communication device is a terminal device or an access network device (first access network device), and the second communication device is an access network device (second access network device), the AI / ML model-related information includes AI / ML model-related information associated with the serving cell and / or AI / ML model-related information associated with at least one neighboring cell of the serving cell, wherein the cell that sends the fifth message to the first communication device is generally referred to as the serving cell or the source cell, and each cell can provide AI / ML model-related information associated with its own cell and / or AI / ML model-related information associated with at least one neighboring cell of its own cell. Optionally, the AI / ML model-related information associated with the at least one neighboring cell can be configured according to frequency granularity or cell granularity, wherein the frequency granularity configuration indicates that one or a group of frequency identifiers are associated with a set of AI / ML model-related information, and the cell granularity configuration indicates that one or a group of cell identifiers are associated with a set of AI / ML model-related information, and the cell identifier is any one of the following: CGI (Cell Global Identity), serving cell index, or (frequency + PCI (Physical Cell Identity)) combination. Optionally, the message for sending information related to the AI / ML model associated with the serving cell and the message for sending information related to the AI / ML model associated with at least one neighboring cell of the serving cell are the same message or different messages.

[0262] That is, the sender of the fifth message is the second communication device; when the first communication device is a terminal device or a first access network device and the second communication device is a second access network device, the AI / ML model-related information includes the AI / ML model-related information associated with the serving cell and / or the AI / ML model-related information associated with at least one neighboring cell of the service message. Optionally, the AI / ML model-related information associated with at least one neighboring cell is configured according to frequency granularity or cell granularity.

[0263] Optionally, when the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or according to the TAC granularity, wherein the configuration according to the access network granularity indicates that one or a group of access network device identifiers are associated with a set of AI / ML model related information (for example, a gNB identifier is associated with a set of AI / ML model related information), and the TAC granularity configuration indicates that one or a group of TAC identifiers are associated with a set of AI / ML model related information.

[0264] That is, the sender of the fifth message is the second communication device; when the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or according to the TAC granularity.

[0265] Optionally, when the first communication device is a terminal device or an access network device or a core network device (first core network device), and the second communication device is a core network device (second core network device), the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity, wherein the configuration according to the core network granularity indicates that a core network device is associated with a set of AI / ML model related information (for example, an AMF device is associated with a set of AI / ML model related information), and the TAC granularity configuration indicates that one or a group of TAC identifiers are associated with a set of AI / ML model related information.

[0266] That is, the sender of the fifth message is the second communication device; when the first communication device is a terminal device or an access network device or a first core network device, and the second communication device is a second core network device, the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity.

[0267] Optionally, when the first communication device is a terminal device (first terminal device) or an access network device or a core network device, and the second communication device is a terminal device (second terminal device), the AI / ML model-related information is configured according to the terminal granularity. The terminal granularity configuration indicates that a terminal device is associated with a set of AI / ML model-related information.

[0268] That is, the sender of the fifth message is the second communication device; when the first communication device is a first terminal device or an access network device or a core network device, and the second communication device is a second terminal device, the AI / ML model related information is configured according to the terminal device granularity.

[0269] In an exemplary embodiment, as shown in FIG8 , before step 306, the process further includes step 305:

[0270] Optionally, step 305: the first communication device sends a sixth message, where the sixth message is used to request the second communication device to provide AI / ML model related information.

[0271] The second communication device receives the sixth message sent by the first communication device. Optionally, the sixth message can be any one of a unicast message, a multicast message, and a broadcast message.

[0272] Optionally, the sixth message includes at least one AI / ML model identifier. Optionally, the fifth message is a response message to the sixth message. Optionally, the second communication device may also reject the sixth message sent by the first communication device, that is, refuse to provide AI / ML model-related information to the first communication device.

[0273] In another exemplary embodiment, as shown in FIG8 , after step 306, the process further includes step 307:

[0274] Optionally, step 307: the first communication device sends a seventh message, where the seventh message includes information related to the AI / ML model associated with the first communication device, where the information related to the AI / ML model includes an AI / ML model identifier and / or AI / ML model description information.

[0275] The second communication device receives the seventh message sent by the first communication device. Optionally, the seventh message can be any one of a unicast message, a multicast message, and a broadcast message.

[0276] Optionally, the seventh message is a response message to the fifth message, or the seventh message is a message actively sent by the first communication device to the second communication device.

[0277] The definition method of the AI / ML model identifier included in the fifth message or the seventh message can adopt any of the definition methods provided in the above embodiments, which will not be repeated here.

[0278] In summary, the method provided in this embodiment provides information related to the AI / ML model associated with the second communication device to the first communication device by transmitting the fifth message. Obtaining information related to the AI / ML model associated with the second communication device through the AI / ML model identifier facilitates mobility management based on the AI / ML model and improves the user switching experience.

[0279] Function 6: Obtain information related to the AI / ML model after local mapping on the second communication device.

[0280] Since the AI / ML model identifier includes two usage modes: a globally unique identifier and a regionally defined identifier, in order to realize the localized mapping of the globally unique identifier in any operator network (establishing a mapping relationship between the globally unique identifier and the regionally defined identifier), an embodiment of the present application provides a model control method as shown in Figure 9.

[0281] Please refer to Figure 9, which shows a flowchart of a model control method provided by one embodiment of the present application. This method can be applied to a communication system consisting of a first communication device and a second communication device. The first communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The second communication device can be one of the terminal device, access network device, and core network device of the system architecture shown in Figure 1. The method includes the following steps.

[0282] Step 309: The second communication device sends an eighth message, which includes at least one AI / ML model identifier defined in a globally unique identification manner and at least one AI / ML model identifier defined in a regional definition identification manner; wherein, there is a one-to-one association relationship, a one-to-many association relationship, or a many-to-one association relationship between the at least one AI / ML model identifier defined in a globally unique identification manner and the at least one AI / ML model identifier defined in a regional definition identification manner.

[0283] That is: an AI / ML model identifier defined as a globally unique identifier is associated with an AI / ML model identifier defined as a regionally defined identifier; or, an AI / ML model identifier defined as a globally unique identifier is associated with a group of AI / ML model identifiers defined as regionally defined identifiers; or, a group of AI / ML model identifiers defined as globally unique identifiers is associated with an AI / ML model identifier defined as a regionally defined identifier. The two AI / ML model identifiers with the above associations can be treated equivalently.

[0284] The first communication device receives an eighth message sent by the second communication device. Optionally, the eighth message may be any one of a unicast message, a multicast message, and a broadcast message.

[0285] Optionally, the eighth message further includes AI / ML model related information associated with each AI / ML model identifier or each group of AI / ML model identifiers defined in the area definition identification manner.

[0286] Optionally, when the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, the first communication device requests the second communication device to update the above mapping relationship, or another second communication device actively triggers a process to update the above mapping relationship.

[0287] That is, when the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, the second communication device proactively triggers the update process and sends the eighth message to the first communication device. The first communication device receives the eighth message proactively triggered by the second communication device. Exemplarily, the second communication device may be a different device of the same type as the second communication device in step 309. For example, the second communication device that initially sends the eighth message to the first communication device is a second access network device, and the second communication device that triggers the update process to send the eighth message to the first communication device again is a third access network device.

[0288] Alternatively, when the first communication device leaves a valid area of ​​a region-defined identifier supported by the second communication device, the first communication device sends a tenth message requesting an update of a mapping relationship between at least one AI / ML model identifier defined in a globally unique identifier manner and at least one AI / ML model identifier defined in a region-defined identifier manner. The second communication device receives the tenth message and, in response to the tenth message, sends an eighth message to the first communication device to update the mapping relationship to the first communication device.

[0289] In another exemplary embodiment, as shown in FIG9 , the process further includes step 308 before step 309:

[0290] Optionally, step 308: the first communication device sends a ninth message, where the ninth message includes at least one AI / ML model identifier defined in a globally unique identification manner.

[0291] The second communication device receives the ninth message sent by the first communication device. Optionally, the ninth message can be any one of a unicast message, a multicast message, and a broadcast message.

[0292] Optionally, the ninth message also includes AI / ML model-related information associated with each AI / ML model identifier defined in a globally unique identifier manner.

[0293] Optionally, the eighth message is a response message to the ninth message, or the eighth message is a message actively triggered by the message sender (the second communication device).

[0294] In summary, the method provided in this embodiment provides the first communication device with information related to the locally mapped AI / ML model of the second communication device by transmitting the eighth message. Obtaining information related to the locally mapped AI / ML model of the second communication device through the AI / ML model identifier facilitates the regional application of AI / ML functions and realizes regional roaming based on AI / ML features, allowing users to continue using AI / ML features to optimize communication performance after regional roaming.

[0295] It should be noted that the method steps in the above embodiments can be arbitrarily combined to obtain new embodiments, and this application does not impose any restrictions on this.

[0296] FIG10 shows a structural block diagram of a model control apparatus provided by an exemplary embodiment of the present application. The apparatus may be implemented as a first communication device, or as a part of the first communication device. The apparatus includes:

[0297] The first transmission module 401 is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0298] In an optional embodiment, there is an association relationship between the AI / ML model identifier and the AI / ML model description information associated with the AI / ML model.

[0299] In an optional embodiment, the AI / ML model description information includes at least one of the following information:

[0300] Information 1: functional characteristics associated with the AI / ML model;

[0301] Information 2: Information about the types of input parameters required by the AI / ML model;

[0302] Information 3: The input parameter format requirements of the AI / ML model;

[0303] Information 4: Input parameter preprocessing rules required by the AI / ML model;

[0304] Information 5: Information about the types of output parameters required by the AI / ML model;

[0305] Information 6: Output parameter format requirements of the AI / ML model;

[0306] Information 7: Output parameter preprocessing rules required by the AI / ML model;

[0307] Information 8: Application scenario information of the AI / ML model;

[0308] Information 9: Deployment location information of the AI / ML model;

[0309] Information 10: Capability requirements for using the AI / ML model;

[0310] Information 11: performance monitoring indicator information of the AI / ML model;

[0311] Information 12: Location information of the AI / ML model;

[0312] Information 13: information about the effective usage scope of the AI / ML model;

[0313] Information 14: generalization characteristics of the AI / ML model;

[0314] Information 15: Version number of the AI / ML model algorithm data;

[0315] Information 16: accuracy level information of the AI / ML model algorithm;

[0316] Information 17: compilation or storage format information of the AI / ML model algorithm data.

[0317] In an optional embodiment, the AI / ML model identifier includes an information field, and the AI / ML model identifier is associated with the information 1.

[0318] In an optional embodiment, the AI / ML model identifier includes an information field, the AI / ML model identifier is associated with the information 1, and the AI / ML model identifier is associated with at least one of the information 2 to the information 17.

[0319] In an optional embodiment, the AI / ML model identifier includes at least two information domains, and each of the at least two information domains is associated with at least one of the information 1 to the information 17.

[0320] In an optional embodiment, the at least two information fields include a first information field and at least one additional information field;

[0321] The first information domain is associated with the information 1;

[0322] Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

[0323] In an optional embodiment, the at least two information fields include a first information field and at least one additional information field;

[0324] The first information field is associated with the information 1, and the first information field is associated with at least one of the information 2 to the information 17;

[0325] Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

[0326] In an optional embodiment, the AI / ML model identifier is a globally unique identifier or a region-defined identifier.

[0327] In an optional embodiment, the first transmission module 401 includes:

[0328] The first sending unit 402 is configured to send a first message, where the first message is used for an activation operation of an AI / ML model.

[0329] In an optional embodiment, the first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field;

[0330] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model activation information field is the definition format of the AI / ML model identifier.

[0331] In an optional embodiment, the first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field;

[0332] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model activation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0333] In an optional embodiment, the AI / ML model activation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0334] In an optional embodiment, the AI / ML model activation information domain includes at least one information domain, and the information format of the AI / ML model activation information domain includes:

[0335] The x information fields in the at least one information field are mandatory fields of the AI / ML model activation information field, and the information fields other than the x information fields in the at least one information field are optional fields of the AI / ML model activation information field, where x is a positive integer.

[0336] In an optional embodiment, the first message is used to activate at least one AI / ML model.

[0337] In an optional embodiment, the first message is an AI / ML model activation dedicated message;

[0338] Alternatively, the first message is a general AI / ML model control message, and the first message includes an AI / ML model activation information field for activation operation.

[0339] In an optional embodiment, the first transmission module 401 includes:

[0340] The first sending unit 402 is configured to send a second message, where the second message is used for deactivating the AI / ML model.

[0341] In an optional embodiment, the second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field;

[0342] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field is the definition format of the AI / ML model identifier.

[0343] In an optional embodiment, the second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field;

[0344] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0345] In an optional embodiment, the AI / ML model deactivation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0346] In an optional embodiment, the AI / ML model deactivation information field includes at least one information field, and the information format of the AI / ML model deactivation information field includes:

[0347] Y information fields in the at least one information field are mandatory fields of the AI / ML model deactivation information field, and information fields other than the y information fields in the at least one information field are optional fields of the AI / ML model deactivation information field, where y is a positive integer.

[0348] In an optional embodiment, the second message is used to deactivate at least one AI / ML model.

[0349] In an optional embodiment, the second message is a dedicated message for deactivating the AI / ML model;

[0350] Alternatively, the second message is an AI / ML model control general message, and the second message includes an AI / ML model deactivation information field for a deactivation operation.

[0351] In an optional embodiment, the first transmission module 401 includes:

[0352] The first sending unit 402 is configured to send a third message, where the third message is used for a switching operation of an AI / ML model.

[0353] In an optional embodiment, the third message includes an AI / ML model activation information field and an AI / ML model deactivation information field;

[0354] The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be activated;

[0355] The AI / ML model deactivation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be deactivated.

[0356] In an optional embodiment, the third message includes an AI / ML model activation information field;

[0357] The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation.

[0358] In an optional embodiment, the first transmission module 401 includes:

[0359] The first sending unit 402 is configured to send a fourth message, where the fourth message is used for deleting the AI / ML model.

[0360] In an optional embodiment, the fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field;

[0361] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field is the definition format of the AI / ML model identifier.

[0362] In an optional embodiment, the fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field;

[0363] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0364] In an optional embodiment, the AI / ML model deletion information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0365] In an optional embodiment, the AI / ML model deletion information field includes at least one information field, and the information format of the AI / ML model deletion information field includes:

[0366] The z information fields in the at least one information field are mandatory fields of the AI / ML model deletion information field, and the information fields other than the z information fields in the at least one information field are optional fields of the AI / ML model deletion information field, where z is a positive integer.

[0367] In an optional embodiment, the fourth message is used to delete at least one AI / ML model.

[0368] In an optional embodiment, the fourth message is a dedicated message for AI / ML model deletion;

[0369] Alternatively, the fourth message is an AI / ML model control general message, and the fourth message includes an AI / ML model deletion information field for a deletion operation.

[0370] In an optional embodiment, the first transmission module 401 includes:

[0371] The first receiving unit 403 is configured to receive a fifth message, where the fifth message includes information related to the AI / ML model associated with the second communication device, where the information related to the AI / ML model includes the AI / ML model identifier and / or AI / ML model description information.

[0372] In an optional embodiment, the first transmission module 401 includes:

[0373] The first sending unit 402 is configured to send a sixth message, where the sixth message is used to request the second communication device to provide information related to the AI / ML model.

[0374] In an optional embodiment, the first transmission module 401 includes:

[0375] The first sending unit 402 is configured to send a seventh message, where the seventh message includes information related to the AI / ML model associated with the first communication device.

[0376] In an optional embodiment, the sender of the fifth message is the second communication device;

[0377] When the first communication device is a terminal device or a first access network device and the second communication device is a second access network device, the AI / ML model related information includes the AI / ML model related information associated with the service cell and / or the AI / ML model related information associated with at least one neighboring area of ​​the service message.

[0378] In an optional embodiment, the AI / ML model related information associated with the at least one neighboring area is configured according to frequency granularity or according to cell granularity.

[0379] In an optional embodiment, the sender of the fifth message is the second communication device;

[0380] When the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or the tracking area TAC granularity.

[0381] In an optional embodiment, the sender of the fifth message is the second communication device;

[0382] When the first communication device is a terminal device or an access network device or a first core network device, and the second communication device is a second core network device, the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity.

[0383] In an optional embodiment, the sender of the fifth message is the second communication device;

[0384] When the first communication device is a first terminal device or an access network device or a core network device, and the second communication device is a second terminal device, the AI / ML model related information is configured according to the terminal granularity.

[0385] In an optional embodiment, the first transmission module 401 includes:

[0386] A first receiving unit 403 is configured to receive an eighth message, where the eighth message includes at least one AI / ML model identifier defined in a globally unique identifier manner and at least one AI / ML model identifier defined in a regionally defined identifier manner;

[0387] There is a one-to-one association relationship, a one-to-many association relationship, or a many-to-one association relationship between the at least one AI / ML model identifier defined in a globally unique identification manner and the at least one AI / ML model identifier defined in a regional definition identification manner.

[0388] In an optional embodiment, the eighth message further includes AI / ML model related information associated with each AI / ML model identifier or each group of AI / ML model identifiers defined in the area definition identifier manner.

[0389] In an optional embodiment, the first transmission module 401 includes:

[0390] The first sending unit 402 is configured to send a ninth message, where the ninth message includes at least one AI / ML model identifier defined in a globally unique identifier manner.

[0391] In an optional embodiment, the ninth message further includes AI / ML model-related information associated with each AI / ML model identifier defined in a globally unique identifier manner.

[0392] In an optional embodiment, the eighth message is a response message to the ninth message, or the eighth message is a message actively triggered by the message sender.

[0393] In an optional embodiment, the sender of the eighth message is a second communication device, and the first transmission module 401 includes:

[0394] The first receiving unit 403 is configured to receive the eighth message actively triggered by the second communication device when the first communication device leaves a valid area of ​​an area definition identifier supported by the second communication device.

[0395] In an optional embodiment, the sender of the eighth message is a second communication device, and the first transmission module 401 includes:

[0396] The first sending unit 402 is used to send a tenth message when the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, and the tenth message is used to request to update the mapping relationship between the at least one AI / ML model identifier defined in a globally unique identifier manner and the at least one AI / ML model identifier defined in an area definition identifier manner.

[0397] In an optional embodiment, the model control message is a unicast message, a multicast message or a broadcast message.

[0398] In an optional embodiment, the model control message is any one of the following message types:

[0399] Non-access layer NAS messages, radio resource control RRC messages, media access control control unit MAC CE messages, downlink control information DCI messages, uplink control information UCI messages, physical uplink shared channel PUSCH messages, inter-node messages, Xn messages, F1 messages, E1 messages, NG messages, core network bus messages.

[0400] In an optional embodiment, the first transmission module 401 is configured to transmit the model control message to the second communication device;

[0401] Wherein, the first communication device is a terminal device, and the second communication device is a network device;

[0402] Alternatively, the first communication device is a network device, and the second communication device is a terminal device;

[0403] Or, the first communication device is a first terminal device, and the second communication device is a second terminal device;

[0404] Alternatively, the first communication device is a first network device, and the second communication device is a second network device.

[0405] FIG11 shows a structural block diagram of a model control device provided by an exemplary embodiment of the present application. The device can be implemented as a second communication device, or as a part of a second communication device. The device includes:

[0406] The second transmission module 404 is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0407] In an optional embodiment, there is an association relationship between the AI / ML model identifier and the AI / ML model description information associated with the AI / ML model.

[0408] In an optional embodiment, the AI / ML model description information includes at least one of the following information:

[0409] Information 1: functional characteristics associated with the AI / ML model;

[0410] Information 2: Information about the types of input parameters required by the AI / ML model;

[0411] Information 3: The input parameter format requirements of the AI / ML model;

[0412] Information 4: Input parameter preprocessing rules required by the AI / ML model;

[0413] Information 5: Information about the types of output parameters required by the AI / ML model;

[0414] Information 6: Output parameter format requirements of the AI / ML model;

[0415] Information 7: Output parameter preprocessing rules required by the AI / ML model;

[0416] Information 8: Application scenario information of the AI / ML model;

[0417] Information 9: Deployment location information of the AI / ML model;

[0418] Information 10: Capability requirements for using the AI / ML model;

[0419] Information 11: performance monitoring indicator information of the AI / ML model;

[0420] Information 12: Location information of the AI / ML model;

[0421] Information 13: information about the effective usage scope of the AI / ML model;

[0422] Information 14: generalization characteristics of the AI / ML model;

[0423] Information 15: Version number of the AI / ML model algorithm data;

[0424] Information 16: accuracy level information of the AI / ML model algorithm;

[0425] Information 17: compilation or storage format information of the AI / ML model algorithm data.

[0426] In an optional embodiment, the AI / ML model identifier includes an information field, and the AI / ML model identifier is associated with the information 1.

[0427] In an optional embodiment, the AI / ML model identifier includes an information field, the AI / ML model identifier is associated with the information 1, and the AI / ML model identifier is associated with at least one of the information 2 to the information 17.

[0428] In an optional embodiment, the AI / ML model identifier includes at least two information domains, and each of the at least two information domains is associated with at least one of the information 1 to the information 17.

[0429] In an optional embodiment, the at least two information fields include a first information field and at least one additional information field;

[0430] The first information domain is associated with the information 1;

[0431] Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

[0432] In an optional embodiment, the at least two information fields include a first information field and at least one additional information field;

[0433] The first information field is associated with the information 1, and the first information field is associated with at least one of the information 2 to the information 17;

[0434] Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

[0435] In an optional embodiment, the AI / ML model identifier is a globally unique identifier or a region-defined identifier.

[0436] In an optional embodiment, the second transmission module 404 includes:

[0437] The second receiving unit 406 is configured to receive a first message, where the first message is used for an activation operation of an AI / ML model.

[0438] In an optional embodiment, the first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field;

[0439] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model activation information field is the definition format of the AI / ML model identifier.

[0440] In an optional embodiment, the first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field;

[0441] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model activation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0442] In an optional embodiment, the AI / ML model activation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0443] In an optional embodiment, the AI / ML model activation information domain includes at least one information domain, and the information format of the AI / ML model activation information domain includes:

[0444] The x information fields of the at least one information field are mandatory fields of the AI / ML model activation information field, and the information fields other than the x information fields of the at least one information field are optional fields of the AI / ML model activation information field, where x is a positive integer. In an optional embodiment, the first message is used to activate at least one AI / ML model.

[0445] In an optional embodiment, the first message is an AI / ML model activation dedicated message;

[0446] Alternatively, the first message is a general AI / ML model control message, and the first message includes an AI / ML model activation information field for activation operation.

[0447] In an optional embodiment, the second transmission module 404 includes:

[0448] The second receiving unit 406 is configured to receive a second message, where the second message is used for deactivating the AI / ML model.

[0449] In an optional embodiment, the second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field;

[0450] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field is the definition format of the AI / ML model identifier.

[0451] In an optional embodiment, the second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field;

[0452] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0453] In an optional embodiment, the AI / ML model deactivation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0454] In an optional embodiment, the AI / ML model deactivation information field includes at least one information field, and the information format of the AI / ML model deactivation information field includes:

[0455] Y information fields in the at least one information field are mandatory fields of the AI / ML model deactivation information field, and information fields other than the y information fields in the at least one information field are optional fields of the AI / ML model deactivation information field, where y is a positive integer. In an optional embodiment, the second message is used to deactivate at least one AI / ML model.

[0456] In an optional embodiment, the second message is a dedicated message for deactivating the AI / ML model;

[0457] Alternatively, the second message is an AI / ML model control general message, and the second message includes an AI / ML model deactivation information field for a deactivation operation.

[0458] In an optional embodiment, the second transmission module 404 includes:

[0459] The second receiving unit 406 is configured to receive a third message, where the third message is used for a switching operation of an AI / ML model.

[0460] In an optional embodiment, the third message includes an AI / ML model activation information field and an AI / ML model deactivation information field;

[0461] The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be activated;

[0462] The AI / ML model deactivation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be deactivated.

[0463] In an optional embodiment, the third message includes an AI / ML model activation information field;

[0464] The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation.

[0465] In an optional embodiment, the second transmission module 404 includes:

[0466] The second receiving unit 406 is configured to receive a fourth message, where the fourth message is used for a deletion operation of the AI / ML model.

[0467] In an optional embodiment, the fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field;

[0468] In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field is the definition format of the AI / ML model identifier.

[0469] In an optional embodiment, the fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field;

[0470] In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

[0471] In an optional embodiment, the AI / ML model deletion information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

[0472] In an optional embodiment, the AI / ML model deletion information field includes at least one information field, and the information format of the AI / ML model deletion information field includes:

[0473] The z information fields in the at least one information field are mandatory fields of the AI / ML model deletion information field, and the information fields other than the z information fields in the at least one information field are optional fields of the AI / ML model deletion information field, where z is a positive integer.

[0474] In an optional embodiment, the fourth message is used to delete at least one AI / ML model.

[0475] In an optional embodiment, the fourth message is a dedicated message for AI / ML model deletion;

[0476] Alternatively, the fourth message is an AI / ML model control general message, and the fourth message includes an AI / ML model deletion information field for a deletion operation.

[0477] In an optional embodiment, the second transmission module 404 includes:

[0478] The second sending unit 405 is configured to send a fifth message, where the fifth message includes information related to the AI / ML model associated with the second communication device, where the information related to the AI / ML model includes the AI / ML model identifier and / or AI / ML model description information.

[0479] In an optional embodiment, the second transmission module 404 includes:

[0480] The second receiving unit 406 is configured to receive a sixth message, where the sixth message is used to request the second communication device to provide information related to the AI / ML model.

[0481] In an optional embodiment, the second transmission module 404 includes:

[0482] The second receiving unit 406 is configured to receive a seventh message, where the seventh message includes information related to the AI / ML model associated with the first communication device.

[0483] In an optional embodiment, the recipient of the fifth message is the first communication device;

[0484] When the first communication device is a terminal device or a first access network device and the second communication device is a second access network device, the AI / ML model related information includes the AI / ML model related information associated with the service cell and / or the AI / ML model related information associated with at least one neighboring area of ​​the service message.

[0485] In an optional embodiment, the AI / ML model related information associated with the at least one neighboring area is configured according to frequency granularity or according to cell granularity.

[0486] In an optional embodiment, the recipient of the fifth message is the first communication device;

[0487] When the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or the tracking area TAC granularity.

[0488] In an optional embodiment, the recipient of the fifth message is the first communication device;

[0489] When the first communication device is a terminal device or an access network device or a first core network device, and the second communication device is a second core network device, the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity.

[0490] In an optional embodiment, the recipient of the fifth message is the first communication device;

[0491] When the first communication device is a first terminal device or an access network device or a core network device, and the second communication device is a second terminal device, the AI / ML model related information is configured according to the terminal granularity.

[0492] In an optional embodiment, the second transmission module 404 includes:

[0493] The second sending unit 405 is configured to send an eighth message, where the eighth message includes at least one AI / ML model identifier defined in a globally unique identifier manner and at least one AI / ML model identifier defined in a regionally defined identifier manner;

[0494] There is a one-to-one association relationship, a one-to-many association relationship, or a many-to-one association relationship between the at least one AI / ML model identifier defined in a globally unique identification manner and the at least one AI / ML model identifier defined in a regional definition identification manner.

[0495] In an optional embodiment, the eighth message further includes AI / ML model related information associated with each AI / ML model identifier or each group of AI / ML model identifiers defined in the area definition identifier manner.

[0496] In an optional embodiment, the second transmission module 404 includes:

[0497] The second receiving unit 406 is configured to receive a ninth message, where the ninth message includes at least one AI / ML model identifier defined in a globally unique identifier manner.

[0498] In an optional embodiment, the ninth message further includes AI / ML model-related information associated with each AI / ML model identifier defined in a globally unique identifier manner.

[0499] In an optional embodiment, the eighth message is a response message to the ninth message, or the eighth message is a message actively triggered by the second communication device.

[0500] In an optional embodiment, the recipient of the eighth message is a first communication device, and the second transmission module 404 includes:

[0501] The second sending unit 405 is configured to actively trigger the sending of the eighth message when the first communication device leaves a valid area of ​​the area definition identifier supported by the second communication device.

[0502] In an optional embodiment, the recipient of the eighth message is a first communication device, and the second transmission module 404 includes:

[0503] The second receiving unit 406 is used to receive a tenth message when the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, and the tenth message is used to request to update the mapping relationship between the at least one AI / ML model identifier defined in a globally unique identifier manner and the at least one AI / ML model identifier defined in an area definition identifier manner.

[0504] In an optional embodiment, the model control message is a unicast message, a multicast message or a broadcast message.

[0505] In an optional embodiment, the model control message is any one of the following message types:

[0506] Non-access layer NAS messages, radio resource control RRC messages, media access control control unit MAC CE messages, downlink control information DCI messages, uplink control information UCI messages, physical uplink shared channel PUSCH messages, inter-node messages, Xn messages, F1 messages, E1 messages, NG messages, core network bus messages.

[0507] In an optional embodiment, the second transmission module 404 is configured to transmit the model control message to the first communication device;

[0508] Wherein, the first communication device is a terminal device, and the second communication device is a network device;

[0509] Alternatively, the first communication device is a network device, and the second communication device is a terminal device;

[0510] Or, the first communication device is a first terminal device, and the second communication device is a second terminal device;

[0511] Alternatively, the first communication device is a first network device, and the second communication device is a second network device.

[0512] FIG12 shows a schematic structural diagram of a communication device (terminal device or network device) provided by an exemplary embodiment of the present application. The communication device includes: a processor 101 , a receiver 102 , a transmitter 103 , a memory 104 and a bus 105 .

[0513] The processor 101 includes one or more processing cores. The processor 101 executes various functional applications and information processing by running software programs and modules.

[0514] The receiver 102 and the transmitter 103 may be implemented as a communication component, which may be a communication chip, and may be referred to as a transceiver.

[0515] The memory 104 is connected to the processor 101 via a bus 105 .

[0516] The memory 104 may be used to store at least one instruction, and the processor 101 may be used to execute the at least one instruction to implement each step in the above method embodiment.

[0517] In addition, the memory 104 can be implemented by any type of volatile or non-volatile storage device or a combination thereof. Volatile or non-volatile storage devices include but are not limited to: magnetic disks or optical disks, electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), read-only memory (ROM), magnetic memory, flash memory, and programmable read-only memory (PROM).

[0518] Among them, when the communication device is implemented as a first communication device, the processor and transceiver in the communication device involved in the embodiment of the present application can execute the steps performed by the first communication device in any of the methods shown above, which will not be repeated here.

[0519] In a possible implementation, when the communication device is implemented as a first communication device,

[0520] The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0521] Among them, when the communication device is implemented as a second communication device, the processor and transceiver in the communication device involved in the embodiment of the present application can execute the steps performed by the second communication device in any of the methods shown above, which will not be repeated here.

[0522] In a possible implementation, when the communication device is implemented as a second communication device,

[0523] The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

[0524] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one instruction, at least one program, code set or instruction set is stored. The at least one instruction, the at least one program, the code set or instruction set is loaded and executed by a processor to implement the model control method performed by a communication device provided in the above-mentioned various method embodiments.

[0525] In an exemplary embodiment, a chip is further provided. The chip includes a programmable logic circuit and / or program instructions. When the chip runs on a communication device, it is used to enable the communication device to implement the model control method described in the above aspects.

[0526] In an exemplary embodiment, a computer program product is further provided. When the computer program product is executed on a processor of a communication device, the communication device executes the model control method described in the above aspects.

[0527] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0528] The above description is merely an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A model control method, characterized in that: The method is performed by a first communication device, and the method includes: A model control message is transmitted, where the model control message is used to control an artificial intelligence (AI) / machine learning (ML) model, and the model control message includes an AI / ML model identifier of the AI / ML model.

2. The method according to claim 1, characterized in that There is an association relationship between the AI / ML model identifier and the AI / ML model description information associated with the AI / ML model.

3. The method according to claim 2, characterized in that The AI / ML model description information includes at least one of the following information: Information 1: functional characteristics associated with the AI / ML model; Information 2: Information about the types of input parameters required by the AI / ML model; Information 3: The required input parameter format of the AI / ML model; Information 4: Input parameter preprocessing rules required by the AI / ML model; Information 5: Information about the types of output parameters required by the AI / ML model; Information 6: Output parameter format requirements of the AI / ML model; Information 7: Output parameter preprocessing rules required by the AI / ML model; Information 8: Application scenario information of the AI / ML model; Information 9: Deployment location information of the AI / ML model; Information 10: Capability requirements for using the AI / ML model; Information 11: performance monitoring indicator information of the AI / ML model; Information 12: Location information of the AI / ML model; Information 13: information about the effective usage scope of the AI / ML model; Information 14: generalization characteristics of the AI / ML model; Information 15: Version number of the AI / ML model algorithm data; Information 16: accuracy level information of the AI / ML model algorithm; Information 17: compilation or storage format information of the AI / ML model algorithm data.

4. The method according to claim 3, characterized in that The AI / ML model identifier includes an information field, and the AI / ML model identifier is associated with the information 1.

5. The method according to claim 3, characterized in that The AI / ML model identifier includes an information field, the AI / ML model identifier is associated with the information 1, and the AI / ML model identifier is associated with at least one of the information 2 to the information 17.

6. The method according to claim 3, characterized in that The AI / ML model identifier includes at least two information fields, and each of the at least two information fields is associated with at least one of the information 1 to the information 17.

7. The method according to claim 6, characterized in that The at least two information fields include a first information field and at least one additional information field; The first information domain is associated with the information 1; Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

8. The method according to claim 6, characterized in that The at least two information fields include a first information field and at least one additional information field; The first information field is associated with the information 1, and the first information field is associated with at least one of the information 2 to the information 17; Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

9. The method according to any one of claims 1 to 8, characterized in that: The AI / ML model identifier is a globally unique identifier or a region-defined identifier.

10. The method according to any one of claims 1 to 9, characterized in that: The transmission model control message includes: A first message is sent, where the first message is used for an activation operation of an AI / ML model.

11. The method according to claim 10, characterized in that The first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model activation information field is the definition format of the AI / ML model identifier.

12. The method according to claim 10, characterized in that The first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model activation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

13. The method according to claim 11 or 12, characterized in that The AI / ML model activation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

14. The method according to any one of claims 11 to 13, characterized in that: The AI / ML model activation information domain includes at least one information domain, and the information format of the AI / ML model activation information domain includes: The x information fields in the at least one information field are mandatory fields of the AI / ML model activation information field, and the information fields other than the x information fields in the at least one information field are optional fields of the AI / ML model activation information field, where x is a positive integer.

15. The method according to any one of claims 10 to 14, characterized in that: The first message is used to activate at least one AI / ML model.

16. The method according to any one of claims 10 to 15, characterized in that: The first message is a dedicated message for AI / ML model activation; Alternatively, the first message is a general AI / ML model control message, and the first message includes an AI / ML model activation information field for activation operation.

17. The method according to claims 1 to 16, characterized in that The transmission model control message includes: Send a second message, where the second message is used to deactivate the AI / ML model.

18. The method according to claim 17, characterized in that The second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field is the definition format of the AI / ML model identifier.

19. The method according to claim 17, wherein The second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

20. The method according to claim 18 or 19, characterized in that The AI / ML model deactivation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

21. The method according to any one of claims 18 to 20, characterized in that The AI / ML model deactivation information domain includes at least one information domain, and the information format of the AI / ML model deactivation information domain includes: Y information fields in the at least one information field are mandatory fields of the AI / ML model deactivation information field, and information fields other than the y information fields in the at least one information field are optional fields of the AI / ML model deactivation information field, where y is a positive integer.

22. The method according to any one of claims 17 to 21, characterized in that The second message is used to deactivate at least one AI / ML model.

23. The method according to any one of claims 17 to 22, characterized in that The second message is a dedicated message for deactivating the AI / ML model; Alternatively, the second message is an AI / ML model control general message, and the second message includes an AI / ML model deactivation information field for a deactivation operation.

24. The method according to any one of claims 1 to 23, characterized in that The transmission model control message includes: A third message is sent, where the third message is used for a switching operation of the AI / ML model.

25. The method according to claim 24, characterized in that The third message includes an AI / ML model activation information field and an AI / ML model deactivation information field; The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be activated; The AI / ML model deactivation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be deactivated.

26. The method according to claim 24, characterized in that The third message includes an AI / ML model activation information field; The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation.

27. The method according to any one of claims 1 to 26, characterized in that The transmission model control message includes: A fourth message is sent, where the fourth message is used for a deletion operation of the AI / ML model.

28. The method according to claim 27, characterized in that The fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field is the definition format of the AI / ML model identifier.

29. The method according to claim 27, characterized in that The fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

30. The method according to claim 28 or 29, characterized in that The AI / ML model deletion information domain also includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

31. The method according to any one of claims 28 to 30, characterized in that The AI / ML model deletion information domain includes at least one information domain, and the information format of the AI / ML model deletion information domain includes: The z information fields in the at least one information field are mandatory fields of the AI / ML model deletion information field, and the information fields other than the z information fields in the at least one information field are optional fields of the AI / ML model deletion information field, where z is a positive integer.

32. The method according to any one of claims 27 to 31, characterized in that The fourth message is used to delete at least one AI / ML model.

33. The method according to any one of claims 27 to 32, characterized in that The fourth message is a message dedicated to AI / ML model deletion; Alternatively, the fourth message is an AI / ML model control general message, and the fourth message includes an AI / ML model deletion information field for a deletion operation.

34. The method according to any one of claims 1 to 33, characterized in that The transmission model control message includes: A fifth message is received, where the fifth message includes information related to the AI / ML model associated with the second communication device, where the information related to the AI / ML model includes the AI / ML model identifier and / or AI / ML model description information.

35. The method according to claim 34, wherein The method further comprises: Send a sixth message, where the sixth message is used to request the second communication device to provide information related to the AI / ML model.

36. The method according to claim 34 or 35, characterized in that The method further comprises: Send a seventh message, where the seventh message includes information related to the AI / ML model associated with the first communication device.

37. The method according to any one of claims 34 to 36, characterized in that The sender of the fifth message is the second communication device; When the first communication device is a terminal device or a first access network device and the second communication device is a second access network device, the AI / ML model related information includes the AI / ML model related information associated with the service cell and / or the AI / ML model related information associated with at least one neighboring area of ​​the service message.

38. The method according to claim 37, wherein The AI / ML model related information associated with the at least one neighboring cell is configured according to frequency granularity or cell granularity.

39. The method according to any one of claims 34 to 36, characterized in that The sender of the fifth message is the second communication device; When the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or the tracking area code TAC granularity.

40. The method according to any one of claims 34 to 36, characterized in that The sender of the fifth message is the second communication device; When the first communication device is a terminal device or an access network device or a first core network device, and the second communication device is a second core network device, the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity.

41. The method according to any one of claims 34 to 36, characterized in that The sender of the fifth message is the second communication device; When the first communication device is a first terminal device or an access network device or a core network device, and the second communication device is a second terminal device, the AI / ML model related information is configured according to the terminal device granularity.

42. The method according to any one of claims 1 to 41, characterized in that The transmission model control message includes: receiving an eighth message, the eighth message including at least one AI / ML model identifier defined in a globally unique identifier manner and at least one AI / ML model identifier defined in a regionally defined identifier manner; There is a one-to-one association relationship, a one-to-many association relationship, or a many-to-one association relationship between the at least one AI / ML model identifier defined in a globally unique identification manner and the at least one AI / ML model identifier defined in a regional definition identification manner.

43. The method according to claim 42, characterized in that The eighth message further includes AI / ML model related information associated with each AI / ML model identifier or each group of AI / ML model identifiers defined in the area definition identifier manner.

44. The method according to claim 42 or 43, characterized in that The method further comprises: A ninth message is sent, where the ninth message includes at least one AI / ML model identifier defined in a globally unique identifier manner.

45. The method according to claim 44, wherein The ninth message also includes AI / ML model related information associated with each AI / ML model identifier defined in a globally unique identifier manner.

46. ​​The method according to any one of claims 42 to 45, characterized in that The eighth message is a response message to the ninth message, or the eighth message is a message actively triggered by the message sender.

47. The method according to any one of claims 42 to 46, characterized in that The sender of the eighth message is the second communication device, and the receiving of the eighth message includes: When the first communication device leaves a valid area of ​​an area definition identifier supported by the second communication device, the eighth message actively triggered by the second communication device is received.

48. The method according to any one of claims 42 to 47, characterized in that The sender of the eighth message is the second communication device, and the method further includes: When the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, a tenth message is sent, where the tenth message is used to request an update of the mapping relationship between the at least one AI / ML model identifier defined in a globally unique identifier manner and the at least one AI / ML model identifier defined in an area definition identifier manner.

49. The method according to any one of claims 1 to 48, characterized in that The model control message is a unicast message, a multicast message or a broadcast message.

50. The method according to any one of claims 1 to 49, characterized in that The model control message is any one of the following message types: Non-access layer NAS messages, radio resource control RRC messages, media access control control unit MAC CE messages, downlink control information DCI messages, uplink control information UCI messages, physical uplink shared channel PUSCH messages, inter-node messages, Xn messages, F1 messages, E1 messages, NG messages, core network bus messages.

51. The method according to any one of claims 1 to 50, characterized in that The transmission model control message includes: transmitting the model control message to a second communication device; Wherein, the first communication device is a terminal device, and the second communication device is a network device; Alternatively, the first communication device is a network device, and the second communication device is a terminal device; Or, the first communication device is a first terminal device, and the second communication device is a second terminal device; Alternatively, the first communication device is a first network device, and the second communication device is a second network device.

52. A model control method, characterized in that: The method is performed by a second communication device, and the method includes: A model control message is transmitted, where the model control message is used to control an artificial intelligence (AI) / machine learning (ML) model, and the model control message includes an AI / ML model identifier of the AI / ML model.

53. The method according to claim 52, characterized in that There is an association relationship between the AI / ML model identifier and the AI / ML model description information associated with the AI / ML model.

54. The method according to claim 53, wherein The AI / ML model description information includes at least one of the following information: Information 1: functional characteristics associated with the AI / ML model; Information 2: Information about the types of input parameters required by the AI / ML model; Information 3: The required input parameter format of the AI / ML model; Information 4: Input parameter preprocessing rules required by the AI / ML model; Information 5: Information about the types of output parameters required by the AI / ML model; Information 6: Output parameter format requirements of the AI / ML model; Information 7: Output parameter preprocessing rules required by the AI / ML model; Information 8: Application scenario information of the AI / ML model; Information 9: Deployment location information of the AI / ML model; Information 10: Capability requirements for using the AI / ML model; Information 11: performance monitoring indicator information of the AI / ML model; Information 12: Location information of the AI / ML model; Information 13: information about the effective usage scope of the AI / ML model; Information 14: generalization characteristics of the AI / ML model; Information 15: Version number of the AI / ML model algorithm data; Information 16: accuracy level information of the AI / ML model algorithm; Information 17: compilation or storage format information of the AI / ML model algorithm data.

55. The method according to claim 54, characterized in that The AI / ML model identifier includes an information field, and the AI / ML model identifier is associated with the information 1.

56. The method according to claim 54, wherein The AI / ML model identifier includes an information field, the AI / ML model identifier is associated with the information 1, and the AI / ML model identifier is associated with at least one of the information 2 to the information 17.

57. The method according to claim 54, characterized in that The AI / ML model identifier includes at least two information fields, and each of the at least two information fields is associated with at least one of the information 1 to the information 17.

58. The method according to claim 57, wherein The at least two information fields include a first information field and at least one additional information field; The first information domain is associated with the information 1; Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

59. The method according to claim 57, wherein The at least two information fields include a first information field and at least one additional information field; The first information field is associated with the information 1, and the first information field is associated with at least one of the information 2 to the information 17; Each of the at least one additional information field is associated with at least one of the information 2 to the information 17 .

60. The method according to any one of claims 52 to 59, characterized in that The AI / ML model identifier is a globally unique identifier or a region-defined identifier.

61. The method according to any one of claims 52 to 60, characterized in that The transmission model control message includes: A first message is received, where the first message is used for an activation operation of an AI / ML model.

62. The method according to claim 61, characterized in that The first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model activation information field is the definition format of the AI / ML model identifier.

63. The method according to claim 61, characterized in that The first message includes an AI / ML model activation information field, and the AI / ML model activation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model activation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

64. The method according to claim 62 or 63, characterized in that The AI / ML model activation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

65. The method according to any one of claims 62 to 64, characterized in that The AI / ML model activation information domain includes at least one information domain, and the information format of the AI / ML model activation information domain includes: The x information fields in the at least one information field are mandatory fields of the AI / ML model activation information field, and the information fields other than the x information fields in the at least one information field are optional fields of the AI / ML model activation information field, where x is a positive integer.

66. The method according to any one of claims 61 to 65, characterized in that The first message is used to activate at least one AI / ML model.

67. The method according to any one of claims 61 to 66, characterized in that The first message is a dedicated message for AI / ML model activation; Alternatively, the first message is a general AI / ML model control message, and the first message includes an AI / ML model activation information field for activation operation.

68. The method according to claims 52 to 67, characterized in that The transmission model control message includes: A second message is received, where the second message is used for a deactivation operation of the AI / ML model.

69. The method according to claim 68, characterized in that The second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field is the definition format of the AI / ML model identifier.

70. The method according to claim 68, wherein The second message includes an AI / ML model deactivation information field, and the AI / ML model deactivation information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deactivation information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

71. The method according to claim 69 or 70, characterized in that The AI / ML model deactivation information domain further includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

72. The method according to any one of claims 69 to 71, characterized in that The AI / ML model deactivation information domain includes at least one information domain, and the information format of the AI / ML model deactivation information domain includes: Y information fields in the at least one information field are mandatory fields of the AI / ML model deactivation information field, and information fields other than the y information fields in the at least one information field are optional fields of the AI / ML model deactivation information field, where y is a positive integer.

73. The method according to any one of claims 68 to 72, characterized in that The second message is used to deactivate at least one AI / ML model.

74. The method according to any one of claims 68 to 73, characterized in that The second message is a dedicated message for deactivating the AI / ML model; Alternatively, the second message is an AI / ML model control general message, and the second message includes an AI / ML model deactivation information field for a deactivation operation.

75. The method according to any one of claims 52 to 74, characterized in that The transmission model control message includes: A third message is received, where the third message is used for a switching operation of the AI / ML model.

76. The method according to claim 75, characterized in that The third message includes an AI / ML model activation information field and an AI / ML model deactivation information field; The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be activated; The AI / ML model deactivation information field includes the AI / ML model identifier of at least one AI / ML model that needs to be deactivated.

77. The method according to claim 75, characterized in that The third message includes an AI / ML model activation information field; The AI / ML model activation information field includes the AI / ML model identifier of at least one AI / ML model that needs to perform an activation operation.

78. The method according to any one of claims 52 to 77, characterized in that The transmission model control message includes: A fourth message is received, where the fourth message is used for a deletion operation of the AI / ML model.

79. The method according to claim 78, characterized in that The fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes an information field, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field is the definition format of the AI / ML model identifier.

80. The method of claim 78, wherein: The fourth message includes an AI / ML model deletion information field, and the AI / ML model deletion information field includes an AI / ML model identification field; In the case where the AI / ML model identifier includes at least two information fields, the information format of the AI / ML model identifier field included in the AI / ML model deletion information field includes at most (2 n -1), where n is the number of information domains identified by the AI / ML model.

81. The method according to claim 79 or 80, characterized in that The AI / ML model deletion information domain also includes at least one other information domain, and each of the at least one other information domain contains at least one of the information 2 to the information 17.

82. The method according to any one of claims 79 to 81, characterized in that The AI / ML model deletion information domain includes at least one information domain, and the information format of the AI / ML model deletion information domain includes: The z information fields in the at least one information field are mandatory fields of the AI / ML model deletion information field, and the information fields other than the z information fields in the at least one information field are optional fields of the AI / ML model deletion information field, where z is a positive integer.

83. The method according to any one of claims 78 to 82, characterized in that The fourth message is used to delete at least one AI / ML model.

84. The method according to any one of claims 78 to 83, characterized in that The fourth message is a message dedicated to AI / ML model deletion; Alternatively, the fourth message is an AI / ML model control general message, and the fourth message includes an AI / ML model deletion information field for a deletion operation.

85. The method according to any one of claims 52 to 84, characterized in that The transmission model control message includes: Send a fifth message, where the fifth message includes information related to the AI / ML model associated with the second communication device, where the information related to the AI / ML model includes the AI / ML model identifier and / or AI / ML model description information.

86. The method according to claim 85, characterized in that The method further comprises: A sixth message is received, where the sixth message is used to request the second communication device to provide information related to the AI / ML model.

87. The method according to claim 85 or 86, characterized in that The method further comprises: A seventh message is received, where the seventh message includes information related to the AI / ML model associated with the first communication device.

88. The method according to any one of claims 85 to 87, characterized in that The recipient of the fifth message is the first communication device; When the first communication device is a terminal device or a first access network device and the second communication device is a second access network device, the AI / ML model related information includes the AI / ML model related information associated with the service cell and / or the AI / ML model related information associated with at least one neighboring area of ​​the service message.

89. The method according to claim 88, characterized in that The AI / ML model related information associated with the at least one neighboring cell is configured according to frequency granularity or cell granularity.

90. The method according to any one of claims 85 to 87, characterized in that The recipient of the fifth message is the first communication device; When the first communication device is a core network device and the second communication device is an access network device, the AI / ML model related information is configured according to the access network granularity or the tracking area code TAC granularity.

91. The method according to any one of claims 85 to 87, characterized in that The recipient of the fifth message is the first communication device; When the first communication device is a terminal device or an access network device or a first core network device, and the second communication device is a second core network device, the AI / ML model related information is configured according to the core network granularity or according to the TAC granularity.

92. The method according to any one of claims 85 to 87, characterized in that The recipient of the fifth message is the first communication device; When the first communication device is a first terminal device or an access network device or a core network device, and the second communication device is a second terminal device, the AI / ML model related information is configured according to the terminal device granularity.

93. The method according to any one of claims 52 to 92, characterized in that The transmission model control message includes: Sending an eighth message, the eighth message including at least one AI / ML model identifier defined in a globally unique identifier manner and at least one AI / ML model identifier defined in a regionally defined identifier manner; There is a one-to-one association relationship, a one-to-many association relationship, or a many-to-one association relationship between the at least one AI / ML model identifier defined in a globally unique identification manner and the at least one AI / ML model identifier defined in a regional definition identification manner.

94. The method according to claim 93, wherein The eighth message further includes AI / ML model related information associated with each AI / ML model identifier or each group of AI / ML model identifiers defined in the area definition identifier manner.

95. The method according to claim 93 or 94, characterized in that The method further comprises: A ninth message is received, where the ninth message includes at least one AI / ML model identifier defined in a globally unique identifier manner.

96. The method according to claim 95, characterized in that The ninth message also includes AI / ML model related information associated with each AI / ML model identifier defined in a globally unique identifier manner.

97. The method according to any one of claims 93 to 96, characterized in that The eighth message is a response message to the ninth message, or the eighth message is a message actively triggered by the second communication device.

98. The method according to any one of claims 93 to 97, characterized in that The recipient of the eighth message is the first communication device, and the receiving of the eighth message includes: When the first communication device leaves a valid area of ​​an area definition identifier supported by the second communication device, the sending of the eighth message is actively triggered.

99. The method according to any one of claims 93 to 98, characterized in that The recipient of the eighth message is the first communication device, and the method further includes: When the first communication device leaves the valid area of ​​the area definition identifier supported by the second communication device, a tenth message is received, where the tenth message is used to request an update of the mapping relationship between the at least one AI / ML model identifier defined in a globally unique identifier manner and the at least one AI / ML model identifier defined in an area definition identifier manner.

100. The method according to any one of claims 52 to 99, characterized in that The model control message is a unicast message, a multicast message or a broadcast message.

101. The method according to any one of claims 52 to 100, characterized in that The model control message is any one of the following message types: Non-access layer NAS messages, radio resource control RRC messages, media access control control unit MAC CE messages, downlink control information DCI messages, uplink control information UCI messages, physical uplink shared channel PUSCH messages, inter-node messages, Xn messages, F1 messages, E1 messages, NG messages, core network bus messages.

102. The method according to any one of claims 52 to 101, characterized in that The transmission model control message includes: transmitting the model control message with the first communication device; Wherein, the first communication device is a terminal device, and the second communication device is a network device; Alternatively, the first communication device is a network device, and the second communication device is a terminal device; Or, the first communication device is a first terminal device, and the second communication device is a second terminal device; Alternatively, the first communication device is a first network device, and the second communication device is a second network device.

103. A model control device, characterized in that: The apparatus is used to implement a first communication device, and the apparatus includes: The first transmission module is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

104. A model control device, characterized in that: The apparatus is used to implement a second communication device, and the apparatus includes: The second transmission module is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

105. A first communication device, characterized in that: The first communication device includes: a processor and a transceiver connected to the processor; wherein, The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

106. A second communication device, characterized in that: The second communication device includes: a processor and a transceiver connected to the processor; wherein, The transceiver is used to transmit a model control message, where the model control message is used to control an artificial intelligence AI / machine learning ML model, and the model control message includes an AI / ML model identifier of the AI / ML model.

107. A first communication device, characterized in that: The first communication device includes: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the model control method as described in any one of claims 1 to 51.

108. A second communication device, characterized in that: The second communication device includes: a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the model control method as described in any one of claims 52 to 102.

109. A computer-readable storage medium, characterized in that The readable storage medium stores executable instructions, which are loaded and executed by a processor to enable the communication device to implement the model control method according to any one of claims 1 to 102.

110. A chip, characterized in that: The chip includes a programmable logic circuit or a program, and a communication device equipped with the chip is used to implement the model control method according to any one of claims 1 to 102.

Images