Model using method, terminal and network side equipment

Abstract

The invention discloses a model using method, a terminal and network side equipment, and belongs to the field of communication, and the model using method comprises the steps that first equipment sends first indication information to second equipment, the first indication information is used for indicating at least one sub-module which needs and / or does not need to be activated by the second equipment and belongs to an AI model, and the first indication information is used for indicating at least one sub-module which needs and / or does not need to be activated by the second equipment; the AI model comprises N sub-modules, and N is an integer greater than 0.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of communication, and particularly relates to a model using method, a terminal and a network side device. BACKGROUND

[0002] Artificial intelligence (AI) is currently widely used in various fields. Integrating artificial intelligence into wireless communication networks can significantly improve throughput, reduce latency and increase user capacity. Therefore, making good use of AI models is an important task for future wireless communication networks.

[0003] Currently, when a communication device uses an AI model, the AI model is activated or deactivated as a whole, which has poor flexibility and requires relatively high hardware and computing power. SUMMARY

[0004] Embodiments of the present application provide a model using method, a terminal and a network side device, which can solve the problem of poor flexibility and high hardware and computing power requirement of existing AI model using schemes.

[0005] In a first aspect, a model using method is provided, which includes:

[0006] A first device sends first indication information to a second device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, and the AI model includes N sub-modules, N being an integer greater than 0.

[0007] In a second aspect, a model using method is provided, which includes:

[0008] A second device receives first indication information sent by a first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, and the AI model includes N sub-modules, N being an integer greater than 0.

[0009] In a third aspect, a model using apparatus is provided, which includes:

[0010] A first sending module is configured to send first indication information to a second device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, and the AI model includes N sub-modules, N being an integer greater than 0.

[0011] In a fourth aspect, a model using apparatus is provided, which includes:

[0012] The first receiving module is configured to receive first indication information sent by the first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs to be activated and / or does not need to be activated by the second device, and the AI model includes N sub-modules, where N is an integer greater than 0.

[0013] In a fifth aspect, a model using apparatus is provided, which is configured to perform the steps of the method according to the first aspect, or implement the steps of the method according to the second aspect.

[0014] In a sixth aspect, a terminal is provided, which includes a processor and a memory, the memory storing programs or instructions executable on the processor, and the programs or instructions, when executed by the processor, implement the steps of the method according to the first aspect or the second aspect.

[0015] In a seventh aspect, a terminal is provided, which includes a processor and a communication interface, wherein the communication interface is configured to send first indication information to a second device, or the communication interface is configured to receive first indication information sent by a first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs to be activated and / or does not need to be activated by the second device, and the AI model includes N sub-modules, where N is an integer greater than 0.

[0016] In an eighth aspect, a network side device is provided, which includes a processor and a memory, the memory storing programs or instructions executable on the processor, and the programs or instructions, when executed by the processor, implement the steps of the method according to the second aspect or the first aspect.

[0017] In a ninth aspect, a network side device is provided, which includes a processor and a communication interface, wherein the communication interface is configured to receive first indication information sent by a first device, or the communication interface is configured to send first indication information to a second device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs to be activated and / or does not need to be activated by the second device, and the AI model includes N sub-modules, where N is an integer greater than 0.

[0018] In a tenth aspect, a readable storage medium is provided, which stores programs or instructions, and the programs or instructions, when executed by a processor, implement the steps of the method according to the first aspect, or implement the steps of the method according to the second aspect.

[0019] In an eleventh aspect, a wireless communication system is provided, including: a terminal and a network side device, the terminal being configured to perform the steps of the method according to the first aspect, and the network side device being configured to perform the steps of the method according to the second aspect; or the terminal being configured to perform the steps of the method according to the second aspect, and the network side device being configured to perform the steps of the method according to the first aspect.

[0020] In a twelfth aspect, a chip is provided, including a processor and a communication interface, the communication interface being coupled to the processor, and the processor being configured to run programs or instructions to implement the method according to the first aspect or the method according to the second aspect.

[0021] In a thirteenth aspect, a computer program / program product is provided, stored in a storage medium, and executed by at least one processor to implement the steps of the method according to the first aspect or the method according to the second aspect.

[0022] In the embodiments of the present application, the complete AI model can be divided into multiple sub-modules, and at least one sub-module is activated, instead of directly activating the entire AI model, so that the flexibility of using the AI model can be improved, and the hardware complexity and computing power requirement can be reduced. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 FIG. 1 is a block diagram of a wireless communication system to which embodiments of the present application can be applied.

[0024] Figure 2 FIG. 2 is a flowchart of a model using method according to an embodiment of the present application.

[0025] Figure 3 FIG. 3 is a schematic diagram of a sub-module division manner of an AI model according to an embodiment of the present application.

[0026] Figure 4 FIG. 4 is a flowchart of a model using method according to another embodiment of the present application.

[0027] Figure 5 FIG. 5 is a flowchart of a model using method according to another embodiment of the present application.

[0028] Figure 6 FIG. 6 is a flowchart of a model using method according to another embodiment of the present application.

[0029] Figure 7 FIG. 7 is a flowchart of a model using method according to another embodiment of the present application.

[0030] Figure 8is a flowchart of a model using method according to an embodiment of the present application.

[0031] Figure 9 is a flowchart of a model using method according to another embodiment of the present application.

[0032] Figure 10 is a flowchart of a model using method according to another embodiment of the present application.

[0033] Figure 11 is a structural diagram of a model using apparatus according to an embodiment of the present application.

[0034] Figure 12 is a structural diagram of a model using apparatus according to another embodiment of the present application.

[0035] Figure 13 is a structural diagram of a model using apparatus according to another embodiment of the present application.

[0036] Figure 14 is a structural diagram of a model using apparatus according to an embodiment of the present application.

[0037] Figure 15 is a structural diagram of a model using apparatus according to another embodiment of the present application.

[0038] Figure 16 is a structural diagram of a model using apparatus according to another embodiment of the present application.

[0039] Figure 17 is a structural diagram of a communication device according to an embodiment of the present application.

[0040] Figure 18 is a structural diagram of a terminal according to an embodiment of the present application.

[0041] Figure 19 is a structural diagram of a network side device according to an embodiment of the present application.

[0042] Figure 20 is a structural diagram of a network side device according to another embodiment of the present application. DETAILED DESCRIPTION

[0043] The technical solutions in the embodiments of the present application will be described clearly below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by those of ordinary skill in the art belong to the scope of protection of the present application.

[0044] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, "or" in this application indicates at least one of the connected objects. For example, the scope of protection for "A or B" covers at least three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B. In addition, the terms "A and / or B," "at least one of A and B," and "at least one of A or B" also cover at least the above three scenarios. The character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0045] The term "instruction" in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction). A direct instruction can be understood as one in which the sender explicitly informs the receiver of specific information, the operation to be performed, or the requested result, etc., in the instruction sent. An indirect instruction can be understood as one in which the receiver determines the corresponding information based on the instruction sent by the sender, or makes a judgment and determines the operation to be performed or the requested result, etc., based on the judgment result.

[0046] It is worth noting that the technology described in the embodiments of the present application is not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA) or other systems. The terms "system" and "network" in the embodiments of the present application are often used interchangeably, and the described technology can be used in the above-mentioned systems and radio technologies, as well as in other systems and radio technologies. The following describes a New Radio (NR) system for the purpose of example, and NR terminology is used in most of the following description, but these technologies can also be applied to systems other than NR systems, such as 6th Generation (6G) communication systems. th

[0047] Figure 1 ​A block diagram of a wireless communication system to which embodiments of the present application can be applied is shown. The wireless communication system includes a terminal 11 and a network side device 12. The terminal 11 can be a terminal side device such as a mobile phone, a Tablet Personal Computer, a Laptop Computer, a notebook, a Personal Digital Assistant (PDA), a palm PC, a netbook, an Ultra-mobile Personal Computer (UMPC), a Mobile Internet Device (MID), an Augmented Reality (AR) device, a Virtual Reality (VR) device, a robot, a wearable device, a flight vehicle, a Vehicle User Equipment (VUE), a shipboard device, a Pedestrian User Equipment (PUE), a smart home (a home device with a wireless communication function such as a refrigerator, a television, a washing machine, or furniture, etc.), a game console, a Personal Computer (PC), a kiosk, or a self-service machine, etc. The wearable device includes a smart watch, a smart bracelet, a smart earphone, smart glasses, smart jewelry (a smart bracelet, a smart necklace, a smart ring, a smart necklace, a smart anklet, a smart necklace, etc.), a smart wristband, smart clothes, etc. The vehicle-mounted device can also be referred to as a vehicle-mounted terminal, a vehicle-mounted controller, a vehicle-mounted module, a vehicle-mounted component, a vehicle-mounted chip, or a vehicle-mounted unit, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiments of the present application. The network side device 12 can include an access network device or a core network device, wherein the access network device can also be referred to as a Radio Access Network (RAN) device, a radio access network function, or a radio access network unit. The access network device can include a base station, a Wireless Local Area Network (WLAN) Access Point (AP), or a Wireless Fidelity (WiFi) node, etc.The base station can be referred to as a Node B (NB), an evolved Node B (eNB), a next generation Node B (gNB), a New Radio Node B (NR Node B), an access point, a relay station (RBS), a serving base station (SBS), a base transceiver station (BTS), a radio base station, a radio transceiver, a basic service set (BSS), an extended service set (ESS), a home Node B (HNB), a home evolved Node B, a transmit / receive point (TRP), or some other suitable terminology in the art, and is not limited to a particular technical terminology, provided that the same technical effect is achieved. It should be noted that in the embodiments of the present application, only the base station in the NR system is taken as an example for introduction, and the specific type of the base station is not limited.

[0048] The core network device can also be referred to as a core network node, a core network function, or a core network network element, etc., which includes but is not limited to at least one of the following: a mobility management entity (MME), an access and mobility management function (AMF), a session management function (SMF), a user plane function (UPF), a policy control function (PCF), a policy and charging rules function (PCRF), an edge application server discovery function (EASDF), a unified data management (UDM), a unified data repository (UDR), a home subscriber server (HSS), a centralized network configuration (CNC), a network repository function (NRF), a network exposure function (NEF), a local NEF (L-NEF), a binding support function (BSF), an application function (AF), a location management function (LMF), a gateway mobile location center (GMLC), a network data analytics function (NWDAF), etc. It should be noted that only the core network device in the NR system is taken as an example for introduction in the embodiments of the present application, and the specific type of the core network device is not limited. If the name of the core network device mentioned in the embodiments of the present application changes in the subsequent protocol version (for example, 6G), it is also within the protection scope of the present application.

[0049] Optionally, the core network equipment can be implemented by one or more functional modules in a single device, or by multiple devices working together; this application does not specifically limit this. It is understood that the aforementioned functional modules can be network elements in hardware devices, software functional modules running on dedicated hardware, or virtualized functional modules instantiated on a platform (e.g., a cloud platform).

[0050] To address the issues of poor flexibility and high hardware and computing power requirements of existing AI model usage schemes, this application proposes a model usage method and apparatus. The following, in conjunction with the accompanying drawings, provides a detailed description of the model usage method provided by this application through some embodiments and application scenarios.

[0051] like Figure 2 As shown, an embodiment of this application proposes a method for using a model, which may include:

[0052] Step 201: The first device sends first indication information to the second device, wherein the first indication information is used to indicate at least one sub-module belonging to the AI ​​model that needs to be activated by the second device and / or does not need to be activated. The AI ​​model includes N sub-modules, where N is an integer greater than 0.

[0053] In some embodiments, the first device may be an access network device and the second device may be a UE. That is, the access network device sends a first indication message to the UE to indicate the activation and / or deactivation of at least one sub-module of the AI ​​model on the UE side.

[0054] In some embodiments, the first device may be a UE and the second device may be an access network device, that is, the UE sends a first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI ​​model on the access network device side.

[0055] In some embodiments, the first device may be an access network device and the second device may be a core network device. That is, the access network device sends a first indication message to the core network device to indicate the activation and / or deactivation of at least one sub-module of the AI ​​model on the core network device side.

[0056] In some embodiments, the first device may be a core network device and the second device may be an access network device. That is, the core network device sends a first indication message to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI ​​model on the access network device side.

[0057] The following text mainly uses the example of the first device being the access network device and the second device being the UE for illustration.

[0058] It should be noted that, if the first indication information is used to indicate that at least one sub-module belonging to the AI model does not need to be activated by the second device, in the case that the at least one sub-module has been in an activated state, the second device needs to deactivate the at least one sub-module; in the case that the at least one sub-module is in an inactivated state, the second device does not process, so that the at least one sub-module remains in the inactivated state.

[0059] AI has been widely applied in various fields at present. Integrating AI into wireless communication networks can significantly improve technical indicators such as throughput, reduce latency, and improve user capacity. There can be many kinds of basic models / basic algorithms used by AI models, such as neural networks, decision trees, support vector machines, and Bayesian classifiers.

[0060] In the embodiments of the present application, the AI model can also be referred to as an AI unit, a machine learning (ML) model, an ML unit, an AI structure, an AI function, an AI feature, a machine learning model, a neural network, a neural network function, a neural network function, or the like. Alternatively, the AI model can refer to a processing unit capable of implementing specific algorithms, formulas, processing flows, capabilities, etc. related to AI, or the AI model can be a processing method, algorithm, function, module or unit for a specific data set, or the AI model can be a processing method, algorithm, function, module or unit running on AI / ML related hardware such as GPU, NPU, TPU, ASIC, etc. The present application does not make specific limitations. The specific data set can include the input and / or output of the AI model.

[0061] In the embodiments of the present application, the identifier of the AI model can be an AI model identifier, an AI structure identifier, or an AI algorithm identifier, or the identifier of the AI model can be the identifier of a specific data set associated with the AI model, or the identifier of the AI model can be the identifier of a specific scene, environment, channel feature, or device related to AI / ML, or the identifier of the AI model can be the identifier of a function, feature, capability or module related to AI / ML. The embodiments of the present application do not make specific limitations.

[0062] It should be noted that the specific training process of the AI model is referred to the related technology, which is not the focus of the present application.

[0063] Regarding the division of sub-modules included in the AI model, in some embodiments, as shown in Figure 3 , it can be divided into input-related sub-modules, intermediate processing sub-modules, and output-related sub-modules.

[0064] In the embodiments of the present application, each of the sub-modules of the AI model has separate sub-module information, wherein the sub-module information includes at least one of the following:

[0065] a) identification information of the sub-module, including at least one of the following:

[0066] i. identification of the sub-module;

[0067] ii. position information of the sub-module in the AI model, for example, the position information can be represented by a position ID.

[0068] b) version information of the sub-module, including at least one of the following:

[0069] i. timestamp of the sub-module, which is used to indicate at least one of the following:

[0070] update time of the sub-module, for example, time when the UE updates the sub-module;

[0071] time of issuing the model parameters of the sub-module, for example, time when the access network device issues the model parameters of the sub-module to the UE;

[0072] time of receiving the model parameters of the sub-module, for example, time when the UE receives the model parameters of the sub-module issued by the access network device;

[0073] timestamp in the model parameters of the sub-module, which is used to represent the time of the sub-module.

[0074] ii. associated information (associated information / ID) of the sub-module or the AI model, which can include at least one of dataset information (dataset information / ID), data feature information and data feature identifier; optionally, the dataset information is information associated with data collection; for example, in the configuration of data collection (such as CSI reporting configuration, CSI resource configuration, reference signal resource set configuration, reference signal resource configuration, CSI-synchronization signal block resource set configuration), the dataset information is configured.

[0075] iii. at least one of base station hardware information, base station configuration information, cell information, physical cell information / ID, serving cell information / ID, area information / ID, cell group information / ID and cell list information / ID associated with the sub-module or the AI model;

[0076] iv. at least one of functionality and characteristics associated with the sub-module or the AI model.

[0077] In some embodiments, the AI model can include, but is not limited to, at least one of the following:

[0078] a) a first AI model used by the second device;

[0079] b) a reference model of the first AI model;

[0080] c) a second AI model used by the second device or the test device in the test;

[0081] d) a reference model of the second AI model;

[0082] e) a third AI model used by the second device or the test device to match the second AI model used in the test; this case is for AI models that need to be used in cooperation in a two-end communication process, for example, in the case of the second AI model being a Channel State Information (CSI) compression model, the third AI model can be a CSI decompression model; in the case of the second AI model being an encoding model, the third AI model can be a decoding model;

[0083] f) a reference model of the third AI model.

[0084] The first indication information can include identification information of at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device.

[0085] The model usage method proposed in the embodiments of the present application can divide a complete AI model into N sub-modules, and activate or deactivate at least one sub-module, instead of directly activating or deactivating the entire AI model, thereby improving the flexibility of AI model usage and reducing hardware complexity and computing power requirements.

[0086] Optionally, as shown in FIG. 2, before step 201 is performed, the model usage method proposed in the embodiments of the present application can further include: Figure 4

[0087] Step 202: The first device sends second indication information to the second device, wherein the second indication information is used to indicate pre-activation of the AI model.

[0088] The second indication information can include identification information of the AI model.

[0089] ​The purpose of the step 202 is to inform the second device of the AI model to which the to-be-activated or to-be-deactivated sub-module belongs before the first device instructs the second device to activate the at least one sub-module belonging to the AI model, so that the second device is ready to activate / deactivate the relevant sub-module.

[0090] Further, as shown in the step 202, after the first device sends the second indication information to the second device, the model using method provided in the embodiment of the present application can further include: Figure 4

[0091] The step 203 is that the first device receives the first feedback information sent by the second device for the first indication information and / or the second indication information.

[0092] Specifically, the first feedback information is only for the first indication information, or the first feedback information is only for the second indication information, or the first feedback information is for the first indication information and the second indication information.

[0093] The first feedback information can include but is not limited to at least one of the following:

[0094] The identification information of the activated sub-module belonging to the AI model supported by the second device;

[0095] The identification information of the activated sub-module belonging to the AI model not supported by the second device;

[0096] The confirmation information of whether the second device supports to activate the at least one sub-module belonging to the AI model.

[0097] The purpose of the step 203 is to make the first device know whether the second device is ready to activate or deactivate the at least one sub-module belonging to the AI model. Of course, the second device can also not feedback the first feedback information.

[0098] Optionally, as shown in the step 202, the model using method provided in the embodiment of the present application can further include: Figure 5

[0099] The step 204 is that the first device receives the second feedback information sent by the second device for the first indication information.

[0100] The second feedback information can include but is not limited to at least one of the following:

[0101] The identification information of the to-be-activated sub-module;

[0102] The identification information of the activated sub-module; ​​

[0103] identification information of the sub-module that fails to be activated;

[0104] identification information of the sub-module that fails to be activated;

[0105] signaling or information of the activation success or failure of the AI model.

[0106] signaling or information of the activation success or failure of the AI model.

[0107] The purpose of the step 204 is to make the first device aware of the activation or deactivation of the at least one sub-module belonging to the AI model by the second device. Of course, the second device can also not feedback the second feedback information.

[0108] Further, in order to improve the efficiency of the first device in activating or deactivating the at least one sub-module belonging to the AI model, and in order to further reduce the requirement for the hardware and computing power of the second device when using the AI model, in an AI model using method provided in an embodiment of the present application, the N sub-modules included in the AI model can also be activated in advance in a certain combination manner, and then when the first indication information indicates that the at least one sub-module needs or does not need to be activated by the second device, the corresponding sub-module combination activation manner is indicated, that is, the first indication information is used to indicate the at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device according to the sub-module combination activation manner. Two sub-module combination activation manners are introduced below. Optionally, the sub-module combination activation manner can be preset, or indicated or configured to the second device by the first device, or indicated or reported to the first device by the second device.

[0109] The first sub-module combination activation manner is as follows:

[0110] The N sub-modules include N1 necessary sub-modules and N2 optional sub-modules, the N1 necessary sub-modules are sub-modules that need to be activated by the second device by default, and the activation of the N1 necessary sub-modules does not need the first device to send the first indication information to the second device, wherein N1 and N2 are integers greater than 0, and N1+N2≤N.

[0111] 1) About necessary sub-modules

[0112] The necessary sub-module is a sub-module that must be activated when the AI model is used. Generally, the necessary sub-module is a basic sub-module that outputs a basic result and / or has limited result precision.

[0113] The necessary sub-module is a sub-module that needs to be activated by the second device by default when the AI model is used, and the activation of the necessary sub-module does not require the first device to send the first indication information to the second device.

[0114] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-module is activated successfully.

[0115] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0116] a) the NW does not need to indicate the activation of the necessary sub-module;

[0117] b) or, the UE does not need to report the activation of the necessary sub-module;

[0118] c) or, the UE does not need to feed back the signaling that the necessary sub-module is activated successfully.

[0119] 2) Regarding the optional sub-module (or non-necessary sub-module)

[0120] The optional sub-module is a sub-module that is selected and activated according to actual needs when the AI model is used. Generally, the optional sub-module is a sub-module that can output other auxiliary results in addition to the basic results, or a sub-module that outputs results with higher accuracy.

[0121] When the sub-modules belonging to the AI model that need to be activated include optional sub-modules, the first indication information is used to indicate at least one of the optional sub-modules that need to be activated by the second device when the AI model is used.

[0122] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-module is activated successfully.

[0123] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0124] a) the NW indicates the activation of at least one of the optional sub-modules;

[0125] b) or, the UE reports the activation of at least one of the optional sub-modules;

[0126] c) or, the UE feeds back the signaling that at least one of the optional sub-modules is activated successfully.

[0127] 3) The AI model under different activation modes meets at least one of the following conditions:

[0128] i. Performance monitoring requirements that the AI model can meet in a first activation mode A are higher than performance monitoring requirements that the AI model can meet in a second activation mode B;

[0129] The performance monitoring requirements (or Key Performance Indication, KPI) can include but are not limited to at least one of the following:

[0130] System performance indicators, such as at least one of the following indicators: throughput rate, bit error rate, block error rate, and bit error rate.

[0131] Indirect performance indicators or AI model direct indicators, such as at least one of the following indicators: mean-square error (MSE), normalized mean squared error (NMSE), squared generalized cosine similarity (SGCS), and cosine similarity.

[0132] The first activation mode A is an activation mode in which the activated sub-modules include the optional sub-module, and the second activation mode B is an activation mode in which the activated sub-modules do not include the optional sub-module (the same below).

[0133] ii. The AI model in the first activation mode A has the ability to obtain a first output and a second output, wherein,

[0134] The first output is the output of the AI model in the first activation mode A, and the second output is the output of the AI model in the second activation mode B.

[0135] Optionally, if the AI model in the first activation mode A has the ability to obtain the first output and the second output, the input of the optional sub-module in the AI model in the first activation mode B includes at least one of the following:

[0136] The second output;

[0137] The input of the AI model;

[0138] The intermediate information of the AI model.

[0139] The first output and the second output can be used by the second device and / or the first device to compare the performance of the AI model in the first activation mode A and the second activation mode B, and the purpose of the performance comparison is to determine which activation mode has better performance of the AI model, or to determine the difference between the performance indicators of the two activation modes, etc.

[0140] There are four ways to compare the performance of the AI model in the first activation mode A and the second activation mode B:

[0141] a) The second device determines the performance comparison result of the AI model in the first activation mode and the AI model in the second activation mode according to the first output and the second output;

[0142] b) The second device determines the performance comparison result according to the first output and the second output, and feeds back the performance comparison result, the first output and the second output to the first device;

[0143] c) The second device determines the performance comparison result according to the first output and the second output, and feeds back the performance comparison result, the first output and the second output to the first device, and the first device makes a final judgment information on the performance comparison result of the second device, and sends the judgment information to the second device;

[0144] d) The second device feeds back the first output and the second output to the first device, and the first device makes the performance comparison result and sends it to the second device.

[0145] For example, assuming that the first device is NW and the second device is UE, there are four ways to compare the performance of the AI model in the first activation mode A and the second activation mode B:

[0146] a) The UE determines the performance comparison result of the AI model in the first activation mode A and the AI model in the second activation mode B according to the first output and the second output;

[0147] b) The UE determines the performance comparison result according to the first output and the second output, and feeds back the performance comparison result, the first output and the second output to the NW;

[0148] c) The UE determines the performance comparison result according to the first output and the second output, and feeds back the performance comparison result, the first output and the second output to the NW, and the NW makes a final judgment information on the performance comparison result of the UE, and sends the judgment information to the UE;

[0149] d) the UE feeds the first output and the second output back to the NW, and the NW makes the performance comparison result and sends it to the UE.

[0150] In some embodiments, regardless of whether the second device makes the performance comparison result and reports it to the first device, the final determination about the performance comparison result is made at the first device.

[0151] Therefore, the model using method proposed in the embodiments of the present application can further include:

[0152] The first device receives third feedback information sent by the second device.

[0153] The third feedback information includes any of the following:

[0154] The performance comparison result made by the second device about the AI model in the first activation mode A and the AI model in the second activation mode B;

[0155] The performance comparison result made by the second device, the first output and the second output;

[0156] The first output and the second output.

[0157] Correspondingly, the model using method proposed in the embodiments of the present application can further include:

[0158] The first device determines, according to the third feedback information, determination information about the performance comparison result;

[0159] The first device sends the determination information to the second device.

[0160] The determination information can be a determination result or a determination indication, for example, the determination result can be that the performance comparison result given by the second device is correct or incorrect, and the determination indication can be to agree or disagree with the performance comparison result given by the second device.

[0161] Correspondingly, on the side of the second device, if it is determined according to the determination information that the performance of the AI model in the first activation mode A is lower than or similar to the performance of the AI model in the second activation mode B, the optional sub-module is deactivated. It can be understood that if the performance of the AI model in the first activation mode A is lower than or similar to the performance of the AI model in the second activation mode B, it means that the activation of the optional sub-module does not help to improve the performance of the AI model, and therefore the optional sub-module can not be activated to save computing power.

[0162] iii. The switching delay of the AI model in the first activation mode A is greater than the switching delay of the AI model in the second activation mode B. It can be understood that since the AI model in the first activation mode A is larger than the AI model in the second activation mode B, the switching delay is longer.

[0163] The second sub-module combination activation mode is:

[0164] The second device has at least two functions (functionality), and different functions correspond to activation of different sub-module sets.

[0165] Wherein, the function (functionality) can refer to the function, use, feature, feature group, etc. that the AI model can realize. For example, spatial beam prediction is one function, and time domain beam prediction is another function; or, time domain beam prediction suitable for low speed is one function, and time domain beam prediction suitable for high speed is another function; or, time domain beam prediction predicting future 4 beams is one function, and time domain beam prediction predicting future 8 beams is another function. Optionally, the function belongs to the capability of the second device, and is reported to the first device in the capability report.

[0166] For example, the first function corresponds to the first sub-module set activated, and the first sub-module set contains N3 sub-modules; the second function corresponds to the second sub-module set activated, and the second sub-module set contains N4 sub-modules. Wherein, N3 and N4 are integers greater than 0, and N3≤N, N4≤N.

[0167] At this time, the first indication information can be used to instruct the second device to start the first function, or the first indication is used to instruct the second device to switch from the first function to the second function.

[0168] Optionally, in the case that there are same sub-modules in the first sub-module set and the second sub-module set, if the first indication information is used to instruct the second device to switch from the first function to the second function, then:

[0169] a) The sub-modules that need to be activated or deactivated indicated in the first indication information do not include the same sub-modules.

[0170] Taking the first device as the NW and the second device as the UE as an example, from the first function to the second function, the NW does not need to indicate the activation and deactivation of the same sub-modules, and / or the NW needs to indicate the deactivation of the non-same sub-modules corresponding to the first function and the activation of the non-same sub-modules corresponding to the second function.

[0171] For example, the sub-module set corresponding to the first function includes sub-module 1, sub-module 2 and sub-module 3, and the sub-module set corresponding to the second function includes sub-module 3, sub-module 4 and sub-module 5, wherein the sub-module 3 is the same sub-module or shared sub-module corresponding to the first function and the second function. Then, when switching from the first function to the second function, the NW does not need to indicate the activation and deactivation of the sub-module 3, and / or the NW needs to indicate the deactivation of the sub-modules 1 and 2 corresponding to the first function, and the NW needs to indicate the activation of the sub-modules 4 and 5 corresponding to the second function.

[0172] b) The second feedback information for the first indication information does not include the activation information or the deactivation information of the same sub-module.

[0173] Similarly, taking the first device as the NW and the second device as the UE as an example, when switching from the first function to the second function, the UE does not need to report the activation and deactivation of the same sub-module, and / or the UE reports the deactivation of the non-same sub-module corresponding to the first function and the activation of the non-same sub-module corresponding to the second function.

[0174] For example, the sub-module set corresponding to the first function includes sub-module 1, sub-module 2 and sub-module 3, and the sub-module set corresponding to the second function includes sub-module 3, sub-module 4 and sub-module 5, wherein the sub-module 3 is the same sub-module or shared sub-module corresponding to the first function and the second function. Then, when switching from the first function to the second function, the UE does not need to report the activation and deactivation of the sub-module 3, and / or the UE reports the deactivation of the sub-modules 1 and 2 corresponding to the first function, the activation of the sub-modules 4 and 5 corresponding to the second function.

[0175] The following will be described in combination with Figure 3 The division of the sub-module set of the AI model is described.

[0176] As Figure 3 shown, an AI model can be divided into three parts of sub-modules: input-related (corresponding to an input layer or an input side or input processing) sub-modules, intermediate processing sub-modules and output-related (corresponding to an output layer or an output side or output processing) sub-modules.

[0177] For the case that the sub-modules activated corresponding to the first function and the second function are completely different, it can include:

[0178] Case 1: There is no shared input or output sub-module between different inputs or outputs, that is, corresponding to completely different input or output sub-modules. For example, input type 1 corresponds to input sub-module 1, and input type 2 corresponds to input sub-module 2; output type 1 corresponds to output sub-module 1 and sub-module 2, and output type 2 corresponds to output sub-module 3 and sub-module 4.

[0179] Case 2: There are shared input or output sub-modules between different inputs or outputs, i.e., there are some same and some different corresponding input or output sub-modules. For example, input type 1 corresponds to input sub-module 1, sub-module 2 and sub-module 3, input type 2 corresponds to input sub-module 2, sub-module 3 and sub-module 4, wherein input sub-module 2 and sub-module 3 are the same; output type 1 corresponds to input sub-module 1, sub-module 2 and sub-module 3, output type 2 corresponds to input sub-module 1, sub-module 2, sub-module 3, sub-module 4 and sub-module 5, wherein input sub-module 2 and sub-module 3 are the same.

[0180] Case 3: For the case of input dimension change: the input dimensions of at least part of the input sub-modules are the same, and different numbers of input sub-modules are combined to meet the input dimension requirement. For example, the input is 52 RBs, and the input dimension of each input sub-module is 13 RBs, so 4 input sub-modules are needed; when the input becomes 104 RBs, 8 input sub-modules are needed.

[0181] Optionally, the parameters of each input sub-module are the same.

[0182] Optionally, the first device can indicate the number of activated input sub-modules through the first indication information (without indicating the sub-module ID).

[0183] Case 4: For the case of output dimension change: the output dimensions of at least part of the output sub-modules are the same, and different numbers of output sub-modules are combined to meet the output dimension requirement. Similar to the "input" in case 3, it will not be repeated here.

[0184] Case 5: Part of the intermediate sub-modules are connected one-to-one with the input sub-modules, or part of the intermediate sub-modules are connected one-to-one with the output sub-modules, or part of the intermediate sub-modules change synchronously with the input sub-modules or the output sub-modules. For example, if the input sub-modules change from sub-module 1 and sub-module 2 to sub-module 2, sub-module 3 and sub-module 4, then the intermediate sub-modules also change from sub-module 1 and sub-module 2 to sub-module 2, sub-module 3 and sub-module 4; or, for example, if the output sub-modules change from sub-module 1 and sub-module 2 to sub-module 2, sub-module 3 and sub-module 4, then the intermediate sub-modules also change from sub-module 1 and sub-module 2 to sub-module 2, sub-module 3 and sub-module 4. Optionally, other intermediate sub-modules that are not connected one-to-one with the input sub-modules or the output sub-modules remain unchanged.

[0185] The following describes a model using method proposed in the present application through two specific embodiments. In the following two embodiments, the first device is a NW, the second device is a UE, and the following two embodiments are applicable to both of the above two sub-module combination activation modes.

[0186] Embodiment 1

[0187] For example,Figure 6 As shown, the model using method provided by an embodiment of the present application can include the following steps:

[0188] In step 601, the NW sends second indication information to the UE, wherein the second indication information is used to indicate pre-activation of the AI model.

[0189] Here, "pre-activation" is intended to inform the UE that the sub-modules of the AI model are about to be activated, so as to facilitate the UE to make preparations.

[0190] Optionally, the NW receives first feedback information sent by the UE for the second indication information, wherein the first feedback information includes at least one of the following:

[0191] identification information of the sub-modules supported by the UE to be activated;

[0192] identification information of the sub-modules not supported by the UE to be activated;

[0193] confirmation information of whether the UE supports activation of the at least one sub-module.

[0194] In step 602, the NW sends first indication information to the UE, wherein the first indication information is used to indicate at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the UE, and the AI model includes N sub-modules, N being an integer greater than 0.

[0195] In a specific implementation, the first indication information can include identification information of at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the UE.

[0196] In step 603 (optionally), the NW receives second feedback information sent by the UE for the first indication information.

[0197] The second feedback information includes at least one of the following:

[0198] identification information of the sub-modules to be activated, i.e., the UE feeds back identification information of the sub-modules to be activated, specifically, the UE can feed back before activating the sub-modules after receiving the first indication information;

[0199] identification information of the sub-modules activated successfully, i.e., the UE feeds back identification information of the sub-modules activated successfully, specifically, the UE can feed back after activating the sub-modules;

[0200] identification information of the sub-modules activated unsuccessfully, i.e., the UE feeds back identification information of the sub-modules activated unsuccessfully, specifically, the UE can feed back after performing the operation of activating the sub-modules;

[0201] The identification information of the unactivated sub-module, i.e., the UE feeds back the identification information of the unactivated sub-module, specifically, the UE can feed back after performing the operation of activating the sub-module;

[0202] The signaling or information of activation success or failure, specifically, the UE can feed back after updating the sub-module;

[0203] The signaling or information of activation success or failure of the AI model, specifically, the UE can feed back after updating the sub-module.

[0204] Of course, the UE can also not send the second feedback information to the NW.

[0205] Figure 6 The model usage method proposed in the embodiment can divide the complete AI model on the UE side into multiple sub-modules, and activate at least one sub-module, instead of directly activating the entire AI model, thereby improving the flexibility of AI model usage and reducing the hardware complexity and computing power requirements of the UE.

[0206] Embodiment two

[0207] As Figure 7 indicated, another embodiment of the application proposes a model usage method, which can include:

[0208] Step 701, the NW sends second indication information to the UE, wherein the second indication information is used to indicate pre-activation of the AI model.

[0209] Here, "pre-activation" aims to inform the UE that the sub-module of the AI model will be activated, so as to prepare the UE.

[0210] The second indication information can include the identification information of the AI model.

[0211] Step 702, the NW sends third indication information to the UE, wherein the third indication information is used to indicate pre-activation of at least one sub-module belonging to the AI model, and the AI model includes N sub-modules, N is an integer greater than 0.

[0212] The third indication information can include the identification information of the at least one sub-module.

[0213] The above step 701 and the above step 702 can be understood as an activation preparation phase.

[0214] Step 703 (optionally), the NW receives fourth feedback information sent by the UE for the third indication information, wherein the fourth feedback information includes at least one of the following:

[0215] The identification information of the sub-modules supported by the UE to be activated, the sub-modules supported by the UE to be activated can be in the set of sub-modules indicated by the NW to be activated, or can be outside the set of sub-modules indicated by the NW to be not activated;

[0216] The identification information of the sub-modules not supported by the UE to be activated, the sub-modules not supported by the UE to be activated can be outside the set of sub-modules indicated by the NW to be activated, or can be in the set of sub-modules indicated by the NW to be not activated;

[0217] The confirmation information of whether the UE supports the at least one sub-module to be activated.

[0218] Step 704, the NW sends the first indication information to the UE, wherein the first indication information is used to indicate the at least one sub-module of the AI model which needs and / or does not need to be activated by the UE (formal activation stage).

[0219] In specific implementation, the first indication information can contain the identification information of the at least one sub-module of the AI model which needs and / or does not need to be activated by the UE.

[0220] At this time, the sub-modules indicated by the NW to be activated in step 704 are contained in the sub-modules indicated by the NW to be pre-activated in step 702, and / or the sub-modules indicated by the NW to be activated in step 704 are contained in the sub-modules supported by the UE to be activated in step 703.

[0221] Or, in step 704, the NW indicates the confirmation information of the third indication information sent in step 702, and / or the NW indicates the confirmation information of the third feedback information fed back by the UE in step 703.

[0222] Specifically, in step 704, if the confirmation information of the third indication information sent in step 702, it means that the UE needs to activate all or at least one of the sub-modules indicated by the NW to be pre-activated in step 702, or the UE needs to activate all or at least one of the sub-modules outside the sub-modules indicated by the NW to be not activated in step 702.

[0223] In step 704, if the NW indicates the confirmation information of the third feedback information fed back by the UE in step 703, it means that the UE needs to activate all or at least one of the sub-modules indicated by the UE to be supported to be activated in step 702, or the UE needs to activate all or at least one of the sub-modules outside the sub-modules indicated by the UE to be not supported to be activated in step 703.

[0224] Step 705 (optionally), the NW receives the second feedback information sent by the UE for the first indication information.

[0225] The second feedback information includes at least one of the following:

[0226] The identification information of the sub-module to be activated, i.e., the UE feeds back the identification information of the sub-module to be activated, specifically, the UE can feed back before activating the sub-module after receiving the first indication information.

[0227] The identification information of the sub-module activated successfully, i.e., the UE feeds back the identification information of the sub-module activated successfully, specifically, the UE can feed back after activating the sub-module.

[0228] The identification information of the sub-module activated unsuccessfully, i.e., the UE feeds back the identification information of the sub-module activated unsuccessfully, specifically, the UE can feed back after performing the operation of activating the sub-module.

[0229] The identification information of the sub-module not activated, i.e., the UE feeds back the identification information of the sub-module not activated, specifically, the UE can feed back after performing the operation of activating the sub-module.

[0230] The signaling or information of the sub-module activated successfully or activated unsuccessfully, specifically, the UE can feed back after updating the sub-module.

[0231] The signaling or information of the AI model activated successfully or activated unsuccessfully, specifically, the UE can feed back after updating the sub-module.

[0232] Of course, the UE can also not send the second feedback information to the NW.

[0233] Figure 7 The model usage method proposed in the embodiment can divide the complete AI model on the UE side into multiple sub-modules, and activate at least one sub-module, instead of directly activating the entire AI model, thereby improving the flexibility of AI model usage and reducing the hardware complexity and computing power requirements of the UE.

[0234] As Figure 8 The application embodiment also proposes a model usage method, which includes:

[0235] Step 801, the second device receives the first indication information sent by the first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, the AI model includes N sub-modules, and N is an integer greater than 0.

[0236] In some embodiments, the first device can be an access network device, and the second device can be a UE, i.e., the access network device sends the first indication information to the UE to indicate the activation and / or deactivation of at least one sub-module of the AI model on the UE side.

[0237] In some embodiments, the first device can be a UE, and the second device can be an access network device, i.e., the UE sends the first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the side of the access network device.

[0238] In some embodiments, the first device can be an access network device, and the second device can be a core network device, i.e., the access network device sends the first indication information to the core network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the side of the core network device.

[0239] In some embodiments, the first device can be a core network device, and the second device can be an access network device, i.e., the core network device sends the first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the side of the access network device.

[0240] Hereinafter, the first device is taken as an access network device, and the second device is taken as a UE as an example.

[0241] It should be noted that if the first indication information is used to indicate that at least one sub-module belonging to the AI model does not need to be activated by the second device, in the case that the at least one sub-module is already in an activated state, the second device needs to deactivate the at least one sub-module; in the case that the at least one sub-module is in an unactivated state, the second device does not need to process, and the at least one sub-module remains in the unactivated state.

[0242] Regarding the division of the sub-modules contained in the AI model, in some embodiments, as shown in FIG. 1, the sub-modules can be divided into input-related sub-modules, intermediate processing sub-modules, and output-related sub-modules. Figure 3

[0243] In the embodiments of the present application, there is separate sub-module information for each sub-module of the AI model, wherein the sub-module information includes at least one of the following:

[0244] a) identification information of the sub-module, including at least one of the following:

[0245] i. identification of the sub-module;

[0246] ii. position information of the sub-module in the AI model, for example, the position information can be represented by a position ID.

[0247] b) version information of the sub-module, including at least one of the following:

[0248] i. timestamp of the sub-module, the timestamp of the sub-module is used to indicate at least one of the following: ​

[0249] an update time of the sub-module, e.g., a time when the UE updates the sub-module;

[0250] a time of model parameter delivery of the sub-module, e.g., a time when the access network device delivers the model parameter of the sub-module to the UE;

[0251] a receiving time of the model parameter of the sub-module, e.g., a time when the UE receives the model parameter of the sub-module delivered by the access network device;

[0252] a timestamp in the model parameter of the sub-module, used to represent the time of the sub-module.

[0253] ii. associated information (associated information / ID) of the sub-module or the AI model, the associated information can include at least one of dataset information (dataset information / ID), data feature information, and data feature identifier; optionally, the dataset information is information associated with data collection; for example, in the configuration of data collection (such as CSI reporting configuration, CSI resource configuration, reference signal resource set configuration, reference signal resource configuration, CSI-synchronization signal block resource set configuration), the dataset information is configured.

[0254] iii. at least one of base station hardware information, base station configuration information, cell information, physical cell information / ID, serving cell information / ID, area information / ID, cell group information / ID, and cell list information / ID associated with the sub-module or the AI model;

[0255] iv. at least one of function and characteristic associated with the sub-module or the AI model.

[0256] In some embodiments, the AI model can include but is not limited to at least one of the following:

[0257] a) a first AI model used by the second device;

[0258] b) a reference model of the first AI model;

[0259] c) a second AI model used by the second device or test device in testing;

[0260] d) a reference model of the second AI model;

[0261] e) The second device or test device is used to match the third AI model of the second AI model used in the test; this case applies to AI models that need to be used together in the process of two-end communication. For example, if the second AI model is a Channel State Information (CSI) compression model, the third AI model can be a CSI decompression model; if the second AI model is an encoding model, the third AI model can be a decoding model.

[0262] f) The reference model of the third AI model.

[0263] The first instruction information may include identification information of at least one sub-module belonging to the AI ​​model that needs and / or does not need to be activated by the second device.

[0264] The model usage method proposed in this application can divide a complete AI model into N sub-modules and activate or deactivate at least one of the sub-modules instead of directly activating or deactivating the entire AI model. Therefore, it can improve the flexibility of AI model usage and reduce hardware complexity and computing power requirements.

[0265] Optionally, such as Figure 9 As shown, before performing step 801, the model usage method proposed in this application embodiment may further include:

[0266] Step 802: The second device receives a second indication message sent by the first device, wherein the second indication message is used to indicate pre-activation of the AI ​​model.

[0267] The second instruction information may include the identification information of the AI ​​model.

[0268] The purpose of adding step 802 is that, before instructing the second device to activate at least one sub-module belonging to the AI ​​model, the first device informs the second device of the AI ​​model to which the sub-module to be activated or deactivated belongs, so that the second device can prepare to activate / deactivate the relevant sub-module.

[0269] Furthermore, such as Figure 9 As shown, after step 802, the model usage method proposed in this application embodiment may further include:

[0270] Step 803: The second device sends a first feedback message to the first device regarding the first indication message and / or the second indication message.

[0271] Specifically, the first feedback information is only for the first indication information, or the first feedback information is only for the second indication information, or the first feedback information is for the first indication information and the second indication information.

[0272] The first feedback information can include, but is not limited to, at least one of the following:

[0273] The second device supports the identification information of the activated sub-module belonging to the AI model;

[0274] The second device does not support the identification information of the activated sub-module belonging to the AI model;

[0275] The second device does not support the identification information of the activated sub-module belonging to the AI model;

[0276] The second device does not support the identification information of the activated sub-module belonging to the AI model;

[0277] Optionally, as shown in the method for using the model according to an embodiment of the present application can further include: Figure 10

[0278] Step 804, the second device sends second feedback information for the first indication information to the first device.

[0279] The second feedback information can include, but is not limited to, at least one of the following:

[0280] The identification information of the sub-module to be activated;

[0281] The identification information of the sub-module activated successfully;

[0282] The identification information of the sub-module activated unsuccessfully;

[0283] The identification information of the sub-module not activated;

[0284] Signaling or information of successful activation or unsuccessful activation;

[0285] Signaling or information of successful activation or unsuccessful activation of the AI model.

[0286] The purpose of adding step 804 is to make the first device know the activation or deactivation of the at least one sub-module belonging to the AI model by the second device. Of course, the second device can also not feedback the second feedback information.

[0287] ​Further, in order to improve the efficiency of the first device activating or deactivating at least one sub-module belonging to the AI model, and in order to further reduce the requirement for hardware and computing power of the second device when using the AI model, in an AI model use method provided in an embodiment of the present application, the N sub-modules included in the AI model can also be activated in advance in a certain combination manner, and then when the first indication information indicates that at least one sub-module needs / not to be activated by the second device, the corresponding sub-module combination activation manner is indicated, that is, the first indication information is used to indicate at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device according to the sub-module combination activation manner. Two sub-module combination activation manners are introduced below. Optionally, the sub-module combination activation manner can be preset, or indicated or configured to the second device by the first device, or indicated or reported to the first device by the second device.

[0288] The first sub-module combination activation manner is:

[0289] The N sub-modules include N1 necessary sub-modules and N2 optional sub-modules, the N1 necessary sub-modules are sub-modules that need to be activated by the second device by default, and the activation of the N1 necessary sub-modules does not require the first device to send the first indication information to the second device, wherein N1 and N2 are integers greater than 0, and N1+N2≤N.

[0290] The first indication information is used to indicate at least one optional sub-module that needs to be activated by the second device.

[0291] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-modules are activated successfully.

[0292] 1) About necessary sub-modules

[0293] The necessary sub-module is a sub-module that must be activated when the AI model is used. Generally, the necessary sub-module is a basic sub-module that outputs a basic result and / or has limited output result accuracy.

[0294] When the AI model is used, the necessary sub-module is a sub-module that needs to be activated by the second device by default, and the activation of the necessary sub-module does not require the first device to send the first indication information to the second device.

[0295] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-modules are activated successfully.

[0296] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0297] d) the NW does not indicate the activation of the necessary sub-module;

[0298] e) or, the UE does not report the activation of the necessary sub-module;

[0299] f) or, the UE does not feed back the signaling of the successful activation of the necessary sub-module.

[0300] 2) Regarding the optional sub-module (or non-necessary sub-module)

[0301] The optional sub-module is a sub-module that is selected to be activated according to actual needs when the AI model is used. Generally, the optional sub-module is a sub-module that can output other auxiliary results in addition to the basic results, or a sub-module that outputs results with higher accuracy.

[0302] When the sub-module belonging to the AI model that needs to be activated includes an optional sub-module during use of the AI model, the first indication information is used to indicate at least one optional sub-module that needs to be activated by the second device.

[0303] Optionally, the second feedback information for the first indication information does not include information about the successful activation of the optional sub-module.

[0304] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0305] d) the NW indicates the activation of at least one optional sub-module in the optional sub-module;

[0306] e) or, the UE reports the activation of at least one optional sub-module;

[0307] f) or, the UE feeds back the signaling of the successful activation of at least one optional sub-module.

[0308] 3) The AI model in different activation modes meets at least one of the following:

[0309] The performance detection requirement that can be met by the AI model in the first activation mode is higher than the performance detection requirement that can be met by the AI model in the second activation mode;

[0310] The AI model in the first activation mode has the ability to obtain a first output and a second output, wherein the first output is the output of the AI model in the first activation mode, and the second output is the output of the AI model in the second activation mode;

[0311] The switching delay of the AI model in the first activation mode is greater than the switching delay of the AI model in the second activation mode;

[0312] The first activation mode is an activation mode in which the activated sub-modules include the optional sub-module, and the second activation mode is an activation mode in which the activated sub-modules do not include the optional sub-module.

[0313] i. The performance monitoring requirement that the AI model can meet in the first activation mode A is higher than the performance monitoring requirement that the AI model can meet in the second activation mode B.

[0314] The performance monitoring requirement (or Key Performance Indication, KPI) can include but is not limited to at least one of the following:

[0315] System performance indicators, such as at least one of the following indicators: throughput rate, bit error rate, block error rate, and bit error rate.

[0316] Indirect performance indicators or AI model direct indicators, such as at least one of the following indicators: mean-square error (MSE), normalized mean squared error (NMSE), squared generalized cosine similarity (SGCS), and cosine similarity.

[0317] The first activation mode A is an activation mode in which the activated sub-modules include the optional sub-module, and the second activation mode B is an activation mode in which the activated sub-modules do not include the optional sub-module (the same below).

[0318] ii. The AI model has the ability to obtain a first output and a second output in the first activation mode A, wherein the first output is the output of the AI model in the first activation mode A, and the second output is the output of the AI model in the second activation mode B.

[0319] Optionally, if the AI model has the ability to obtain the first output and the second output in the first activation mode A, the input of the optional sub-module in the AI model in the first activation mode B includes at least one of the following:

[0320] The second output;

[0321] The input of the AI model;

[0322] The intermediate information of the AI model.

[0323] The first output and the second output can be used by the second device and / or the first device to compare the performance of the AI model in the first activation mode A and the second activation mode B, and the performance comparison aims to determine which activation mode has better performance of the AI model, or determine the difference between the performance indicators in the two activation modes, etc.

[0324] In some embodiments, whether the second device makes the performance comparison result and reports it to the first device, the final determination about the performance comparison result is in the first device. Therefore, the model usage method proposed in the embodiments of the present application can further include: the second device sends third feedback information to the first device;

[0325] The third feedback information includes any of the following:

[0326] The performance comparison result made by the second device about the AI model in the first activation mode and the AI model in the second activation mode;

[0327] The performance comparison result made by the second device, the first output, and the second output;

[0328] The first output and the second output.

[0329] Further, the model usage method proposed in the embodiments of the present application can further include:

[0330] The second device receives determination information sent by the first device, wherein the determination information is about the performance comparison result, and the determination information is determined by the first device according to the third feedback information;

[0331] In a case where the second device determines according to the determination information that the performance of the AI model in the first activation mode is lower than or approximately equal to the performance of the AI model in the second activation mode, the second device deactivates the optional sub-module.

[0332] It can be understood that if the performance of the AI model in the first activation mode A is lower than or approximately equal to the performance of the AI model in the second activation mode B, it means that the activation of the optional sub-module is not helpful for the performance improvement of the AI model, and the optional sub-module can be deactivated to save computing power.

[0333] ⅲ. The switching delay of the AI model in the first activation mode A is greater than the switching delay of the AI model in the second activation mode B. It can be understood that since the AI model in the first activation mode A is larger than the AI model in the second activation mode B, the switching delay is longer.

[0334] The second sub-module combination activation mode is:

[0335] There are at least two functionalities in the second device, and different functionalities correspond to activation of different sub-module sets.

[0336] The functionality can refer to a function, use, feature, feature group, etc. that the AI model can implement. For example, spatial beam prediction is one function, and time domain beam prediction is another function; or, time domain beam prediction suitable for low speed is one function, and time domain beam prediction suitable for high speed is another function; or, time domain beam prediction predicting future 4 beams is one function, and time domain beam prediction predicting future 8 beams is another function. Optionally, the functionality belongs to the capability of the second device, and is reported to the first device in the capability report.

[0337] For example, the first functionality corresponds to a first sub-module set, and the first sub-module set includes N3 sub-modules; the second functionality corresponds to a second sub-module set, and the second sub-module set includes N4 sub-modules. Wherein, N3 and N4 are integers greater than 0, and N3≤N, N4≤N.

[0338] At this time, the first indication information can be used to instruct the second device to start the first function, or the first indication can be used to instruct the second device to switch from the first function to the second function.

[0339] Optionally, in the case where there are same sub-modules in the first sub-module set and the second sub-module set, if the first indication information is used to instruct the second device to switch from the first function to the second function, then:

[0340] a) The sub-modules that need to be activated or deactivated indicated in the first indication information do not include the same sub-modules.

[0341] Taking the first device as the NW and the second device as the UE as an example, from the first function to the second function, the NW does not need to indicate the activation and deactivation of the same sub-modules, and / or the NW needs to indicate the deactivation of the non-same sub-modules corresponding to the first function and the activation of the non-same sub-modules corresponding to the second function.

[0342] For example, the set of submodules corresponding to the first function includes submodule 1, submodule 2, and submodule 3, and the set of submodules corresponding to the second function includes submodule 3, submodule 4, and submodule 5. Submodule 3 is the same or shared submodule corresponding to both the first and second functions. Therefore, when switching from the first function to the second function, the NW does not need to indicate the activation or deactivation of submodule 3, and / or, the NW needs to indicate the deactivation of submodules 1 and 2 corresponding to the first function, and the NW needs to indicate the activation of submodules 4 and 5 corresponding to the second function.

[0343] b) The second feedback information for the first indication information does not include the activation or deactivation information of the same sub-module.

[0344] Similarly, taking the first device as NW and the second device as UE as an example, when switching from the first function to the second function, the UE does not need to report the activation and deactivation of the same sub-module, and / or the UE needs to report the deactivation of the different sub-modules corresponding to the first function and the activation of the different sub-modules corresponding to the second function.

[0345] For example, the set of submodules corresponding to the first function includes submodule 1, submodule 2, and submodule 3, and the set of submodules corresponding to the second function includes submodule 3, submodule 4, and submodule 5. Submodule 3 is the same or shared submodule corresponding to both the first and second functions. Therefore, when switching from the first function to the second function, the UE does not need to report the activation and deactivation of submodule 3, and / or the UE needs to report the deactivation of submodules 1 and 2 corresponding to the first function, and the activation of submodules 4 and 5 corresponding to the second function.

[0346] This application provides a model usage method, in which the executing entity can be a virtual device. This application uses a virtual device executing the model usage method as an example to illustrate the model usage apparatus provided in this application.

[0347] like Figure 11 As shown, one embodiment of this application proposes a model usage device 1100, which can be used in a first device. The device 1400 may include: a first sending module 1101, used to send first indication information to a second device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs to be activated by the second device and / or does not need to be activated by the second device. The AI ​​model includes N sub-modules, where N is an integer greater than 0.

[0348] In some embodiments, the first device may be an access network device and the second device may be a UE. That is, the access network device sends a first indication message to the UE to indicate the activation and / or deactivation of at least one sub-module of the AI ​​model on the UE side.

[0349] In some embodiments, the first device can be a UE, and the second device can be an access network device, i.e., the UE sends the first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the access network device side.

[0350] In some embodiments, the first device can be an access network device, and the second device can be a core network device, i.e., the access network device sends the first indication information to the core network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the core network device side.

[0351] In some embodiments, the first device can be a core network device, and the second device can be an access network device, i.e., the core network device sends the first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the access network device side.

[0352] Hereinafter, the first device is mainly taken as an access network device, and the second device is mainly taken as a UE for example.

[0353] In some embodiments, the first device can be a core network device, and the second device can be an access network device, i.e., the core network device sends the first indication information to the access network device to indicate the activation and / or deactivation of at least one sub-module of the AI model on the access network device side.

[0354] Hereinafter, the first device is mainly taken as an access network device, and the second device is mainly taken as a UE for example.

[0355] In the embodiments of the present application, there is separate sub-module information for each sub-module of the AI model, wherein the sub-module information includes at least one of the following:

[0356] a) identification information of the sub-module, including at least one of the following:

[0357] i. identification of the sub-module;

[0358] ii. position information of the sub-module in the AI model, for example, the position information can be represented by a position ID.

[0359] b) version information of the sub-module, including at least one of the following:

[0360] i. timestamp of the sub-module, which is used to indicate at least one of the following:

[0361] update time of the sub-module, for example, the time when the UE updates the sub-module;

[0362] time of issuing the model parameters of the sub-module, for example, the time when the access network device issues the model parameters of the sub-module to the UE;

[0363] a time when the UE receives the model parameters of the sub-module from the access network device;

[0364] a timestamp in the model parameters of the sub-module, used to represent the time of the sub-module.

[0365] ii. associated information (associated information / ID) of the sub-module or the AI model, which can include at least one of dataset information (dataset information / ID), data feature information, and data feature identifier; optionally, the dataset information is information associated with data collection; for example, in the configuration of data collection (such as CSI reporting configuration, CSI resource configuration, reference signal resource set configuration, reference signal resource configuration, CSI-synchronization signal block resource set configuration), the dataset information is configured.

[0366] iii. at least one of base station hardware information, base station configuration information, cell information, physical cell information / ID, serving cell information / ID, area information / ID, cell group information / ID, and cell list information / ID associated with the sub-module or the AI model;

[0367] iv. at least one of functions and characteristics associated with the sub-module or the AI model.

[0368] In some embodiments, the AI model can include but is not limited to at least one of the following:

[0369] a) a first AI model used by the second device;

[0370] b) a reference model of the first AI model;

[0371] c) a second AI model used by the second device or test device in testing;

[0372] d) a reference model of the second AI model;

[0373] e) a third AI model used by the second device or test device to match the second AI model used in testing; this case is for AI models that need to be used together in a two-end communication process, for example, in the case of the second AI model being a channel state information (CSI) compression model, the third AI model can be a CSI decompression model; in the case of the second AI model being an encoding model, the third AI model can be a decoding model;

[0374] f) a reference model of the third AI model.

[0375] The first indication information can include identification information of at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device.

[0376] The model using device 1100 proposed in the embodiments of the present application can divide a complete AI model into N sub-modules, and activate or deactivate at least one sub-module, instead of directly activating or deactivating the entire AI model, so that the flexibility of AI model use can be improved, and the hardware complexity and computing power requirement can be reduced.

[0377] Optionally, as shown in the model using device 1100 proposed in the embodiments of the present application can further include: Figure 12

[0378] The second sending module 1102 is configured to send second indication information to the second device before sending the first indication information to the second device, wherein the second indication information is used to indicate pre-activation of the AI model.

[0379] The second indication information can include identification information of the AI model.

[0380] Further, as shown in the model using method proposed in the embodiments of the present application can further include: Figure 12

[0381] The first receiving module 1103 is configured to receive first feedback information sent by the second device for the first indication information and / or the second indication information.

[0382] Specifically, the first feedback information is only for the first indication information, or the first feedback information is only for the second indication information, or the first feedback information is for the first indication information and the second indication information.

[0383] The first feedback information can include but is not limited to at least one of the following:

[0384] Identification information of a sub-module belonging to the AI model that the second device supports to activate;

[0385] Identification information of a sub-module belonging to the AI model that the second device does not support to activate;

[0386] Confirmation information of whether the second device supports to activate the at least one sub-module belonging to the AI model.

[0387] Optionally, as shown in the model using device 1100 proposed in the embodiments of the present application can further include: Figure 13 ​​As shown, the model using device 1100 provided in the embodiments of the present application can further include:

[0388] The second receiving module 1104 is configured to receive second feedback information sent by the second device for the first indication information.

[0389] The second feedback information can include, but is not limited to, at least one of the following:

[0390] The identification information of the sub-module to be activated;

[0391] The identification information of the sub-module activated successfully;

[0392] The identification information of the sub-module activated unsuccessfully;

[0393] The identification information of the sub-module not activated;

[0394] The signaling or information of the successful or unsuccessful activation;

[0395] The signaling or information of the successful or unsuccessful activation of the AI model.

[0396] Further, in order to improve the efficiency of the first device activating or deactivating at least one sub-module belonging to the AI model, and in order to further reduce the requirement for the hardware and computing power of the second device when using the AI model, in the AI model using method provided in the embodiments of the present application, the N sub-modules included in the AI model can be activated in advance in a certain combination manner, and then when the first indication information indicates that at least one sub-module needs or does not need to be activated by the second device, the corresponding sub-module combination activation manner is indicated, that is, the first indication information is used to indicate at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device according to the sub-module combination activation manner. Two sub-module combination activation manners are introduced below. Optionally, the sub-module combination activation manner can be preset, or indicated or configured to the second device by the first device, or indicated or reported to the first device by the second device.

[0397] The first sub-module combination activation manner:

[0398] The N sub-modules include N1 necessary sub-modules and N2 optional sub-modules, the N1 necessary sub-modules are sub-modules that need to be activated by the second device by default, and the activation of the N1 necessary sub-modules does not require the first device to send the first indication information to the second device, wherein N1 and N2 are integers greater than 0, and N1+N2≤N.

[0399] Optionally, the second feedback information for the first indication information does not include information about the successful activation of the necessary sub-modules.

[0400] The optional submodule is a submodule that is selected and activated according to actual needs when the AI model is used. Generally, the optional submodule is a submodule that can output other auxiliary results in addition to basic results, or a submodule that outputs results with higher accuracy.

[0401] When the AI model is used, if the submodule that needs to be activated and belongs to the AI model includes an optional submodule, the first indication information is used to indicate at least one optional submodule that needs to be activated by the second device.

[0402] Optionally, the second feedback information for the first indication information does not include information that the optional submodule is activated successfully.

[0403] Optionally, the AI model in different activation modes meets at least one of the following conditions:

[0404] i. The performance monitoring requirement that can be met by the AI model in the first activation mode A is higher than the performance monitoring requirement that can be met by the AI model in the second activation mode B;

[0405] ii. The AI model in the first activation mode A has the ability to obtain a first output and a second output, wherein the first output is the output of the AI model in the first activation mode A, and the second output is the output of the AI model in the second activation mode B;

[0406] Optionally, if the AI model in the first activation mode A has the ability to obtain the first output and the second output, the input of the optional submodule in the AI model in the first activation mode B includes at least one of the following:

[0407] The second output;

[0408] Input of the AI model;

[0409] Intermediate information of the AI model.

[0410] The first output and the second output can be used for the second device and / or the first device to compare the performance of the AI model in the first activation mode A and the second activation mode B, and the purpose of the performance comparison is to determine which activation mode has better performance of the AI model, or to determine the difference between the performance indicators in the two activation modes, etc.

[0411] In some embodiments, whether the second device makes the performance comparison result and reports it to the first device, the final determination of the performance comparison result is in the first device.

[0412] Therefore, the model using device 1100 proposed in the embodiments of the present application can further include:

[0413] The third receiving module is configured to receive third feedback information sent by the second device.

[0414] The third feedback information includes any of the following:

[0415] a performance comparison result made by the second device on the AI model in the first activation mode A and the AI model in the second activation mode B;

[0416] the performance comparison result made by the second device, the first output, and the second output;

[0417] the first output and the second output.

[0418] Correspondingly, the model using device 1100 proposed in the embodiments of the present application can further include:

[0419] The determination information determining module is configured to determine determination information on the performance comparison result according to the third feedback information.

[0420] The determination information sending module is configured to send the determination information to the second device.

[0421] The determination information can be a determination result or a determination indication, for example, the determination result can be that the performance comparison result given by the second device is correct or incorrect, and the determination indication can be to agree or disagree with the performance comparison result given by the second device.

[0422] Correspondingly, on the side of the second device, if it is determined according to the determination information that the performance of the AI model in the first activation mode A is lower than or similar to the performance of the AI model in the second activation mode B, the optional sub-module is deactivated. It can be understood that if the performance of the AI model in the first activation mode A is lower than or similar to the performance of the AI model in the second activation mode B, it means that the activation of the optional sub-module is not helpful for the performance improvement of the AI model, and therefore the optional sub-module can not be activated to save computing power.

[0423] ⅲ. The switching delay of the AI model in the first activation mode A is greater than the switching delay of the AI model in the second activation mode B. It can be understood that since the AI model in the first activation mode A is larger than the AI model in the second activation mode B, the switching delay is longer.

[0424] The second sub-module combination activation mode:

[0425] The second device has at least two functionalities, and different functionalities correspond to different sets of activated sub-modules.

[0426] The functionality can refer to a function, a use, a feature, a feature group, etc. that the AI model can implement. For example, spatial beam prediction is one function, and time domain beam prediction is another function; or, time domain beam prediction suitable for low speed is one function, and time domain beam prediction suitable for high speed is another function; or, time domain beam prediction predicting future 4 beams is one function, and time domain beam prediction predicting future 8 beams is another function.

[0427] At this time, the first indication information can be used to instruct the second device to start the first function, or the first indication can be used to instruct the second device to switch from the first function to the second function.

[0428] Optionally, in the case that there is a same sub-module in the first set of sub-modules and the second set of sub-modules, if the first indication information is used to instruct the second device to switch from the first function to the second function, then:

[0429] a) The sub-module that needs to be activated or deactivated indicated in the first indication information does not include the same sub-module.

[0430] b) The second feedback information for the first indication information does not include activation information or deactivation information of the same sub-module.

[0431] As Figure 14 shown, the embodiments of the present application also propose a model using device 1400 applied to a second device. The device 1400 can include a first receiving module 1401 for receiving first indication information sent by a first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, and the AI model includes N sub-modules, and N is an integer greater than 0.

[0432] In the embodiments of the present application, there is separate sub-module information for each sub-module of the AI model, wherein the sub-module information includes at least one of the following:

[0433] a) Identification information of the sub-module, including at least one of the following:

[0434] i. The identification of the sub-module;

[0435] ii. Position information of the sub-module in the AI model, which can be represented by a position ID.

[0436] b) version information of the sub-module, including at least one of:

[0437] i. a timestamp of the sub-module, the timestamp of the sub-module being used to indicate at least one of:

[0438] an update time of the sub-module, for example, a time when the UE updates the sub-module;

[0439] a time of issuing the model parameters of the sub-module, for example, a time when the access network device issues the model parameters of the sub-module to the UE;

[0440] a time of receiving the model parameters of the sub-module, for example, a time when the UE receives the model parameters of the sub-module issued by the access network device;

[0441] a timestamp in the model parameters of the sub-module, used to represent the time of the sub-module.

[0442] ii. associated information (associated information / ID) of the sub-module or the AI model, the associated information can include at least one of dataset information (dataset information / ID), data feature information and data feature identifier;

[0443] iii. at least one of base station hardware information, base station configuration information, cell information, physical cell information / ID, serving cell information / ID, area information / ID, cell group information / ID and cell list information / ID associated with the sub-module or the AI model;

[0444] iv. at least one of functions and characteristics associated with the sub-module or the AI model.

[0445] In some embodiments, the AI model can include but is not limited to at least one of:

[0446] a) a first AI model used by the second device;

[0447] b) a reference model of the first AI model;

[0448] c) a second AI model used by the second device or test device in testing;

[0449] d) a reference model of the second AI model;

[0450] e) the second device or test device is configured to match a third AI model of the second AI model used in the test; this case is for AI models that need to be used in cooperation in a two-end communication process, for example, in the case of the second AI model being a channel state information (CSI) compression model, the third AI model can be a CSI decompression model; in the case of the second AI model being an encoding model, the third AI model can be a decoding model;

[0451] f) a reference model of the third AI model.

[0452] The first indication information can include identification information of at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device.

[0453] The model using device 1400 proposed in the embodiments of the present application can divide a complete AI model into N sub-modules, and activate or deactivate at least one sub-module, instead of directly activating or deactivating the entire AI model, so that the flexibility of AI model use can be improved, and the hardware complexity and computing power requirements can be reduced.

[0454] Optionally, as shown in the Figure 15 embodiments of the present application, the model using device 1400 can further include a second receiving module 1402 configured to receive second indication information sent by the first device before receiving the first indication information, wherein the second indication information is used to indicate pre-activation of the AI model.

[0455] The second indication information can include identification information of the AI model.

[0456] Further, as shown in the Figure 15 embodiments of the present application, the device 1400 can further include a first sending module 1403 configured to send first feedback information for the first indication information and / or the second indication information to the first device.

[0457] Specifically, the first feedback information is only for the first indication information, or the first feedback information is only for the second indication information, or the first feedback information is for the first indication information and the second indication information.

[0458] The first feedback information can include but is not limited to at least one of the following:

[0459] identification information of the sub-module belonging to the AI model that the second device supports to activate;

[0460] identification information of a sub-module belonging to the AI model that is to be activated;

[0461] confirmation information of whether the second device supports activation of the at least one sub-module belonging to the AI model.

[0462] Optionally, as shown in the figure, the model using apparatus 1400 proposed in the embodiments of the present application can further include a second sending module 1404 configured to send second feedback information for the first indication information to the first device. Figure 16

[0463] The second feedback information can include, but is not limited to, at least one of the following:

[0464] identification information of the sub-module that is to be activated;

[0465] identification information of the sub-module that is activated successfully;

[0466] identification information of the sub-module that is activated unsuccessfully;

[0467] identification information of the sub-module that is not activated;

[0468] signaling or information of activation success or activation failure;

[0469] signaling or information of activation success or activation failure of the AI model.

[0470] Further, in order to improve the efficiency of the first device in activating or deactivating at least one sub-module belonging to the AI model, and in order to further reduce the requirement of hardware and computing power of the second device when using the AI model, in an AI model using method provided in the embodiments of the present application, N sub-modules included in the AI model can be activated in advance in a certain combination manner, and then when the first indication information indicates that at least one sub-module that needs / does not need to be activated by the second device, the corresponding sub-module combination activation manner is indicated, that is, the first indication information is used to indicate at least one sub-module belonging to the AI model that needs and / or does not need to be activated by the second device according to the sub-module combination activation manner. Two sub-module combination activation manners are introduced below. Optionally, the sub-module combination activation manner can be preset, or indicated or configured to the second device by the first device, or indicated or reported to the first device by the second device.

[0471] The first sub-module combination activation manner is:

[0472] ​The N sub-modules include N1 necessary sub-modules and N2 optional sub-modules, the N1 necessary sub-modules are sub-modules that are required to be activated by the second device by default, and activation of the N1 necessary sub-modules does not require the first device to send the first indication information to the second device, wherein N1 and N2 are integers greater than 0, and N1+N2≤N.

[0473] The first indication information is used to indicate at least one optional sub-module that needs to be activated by the second device.

[0474] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-modules are activated successfully.

[0475] 1) About necessary sub-modules

[0476] The necessary sub-module is a sub-module that must be activated when the AI model is used. Generally, the necessary sub-module is a basic sub-module that outputs a basic result and / or has limited output result accuracy.

[0477] When the AI model is used, the necessary sub-module is a sub-module that is required to be activated by the second device by default, and activation of the necessary sub-module does not require the first device to send the first indication information to the second device.

[0478] Optionally, the second feedback information for the first indication information does not include information that the necessary sub-modules are activated successfully.

[0479] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0480] g) The NW does not need to indicate activation of the necessary sub-module;

[0481] h) or, the UE does not need to report activation of the necessary sub-module;

[0482] i) or, the UE does not need to feedback signaling that the necessary sub-module is activated successfully.

[0483] 2) About optional sub-modules (or non-necessary sub-modules)

[0484] The optional sub-module is a sub-module that is selected and activated according to actual needs when the AI model is used. Generally, the optional sub-module is a sub-module that can output other auxiliary results in addition to basic results, or has higher output result accuracy.

[0485] When the sub-modules belonging to the AI model that need to be activated include optional sub-modules, the first indication information is used to indicate at least one optional sub-module that needs to be activated by the second device when the AI model is used.

[0486] Optionally, the second feedback information for the first indication information does not include information that the optional sub-module is activated successfully.

[0487] For example, assuming that the first device is a network device (NetWork, NW) and the second device is a UE, then:

[0488] g) The NW indicates the activation of at least one of the optional sub-modules;

[0489] h) Or, the UE reports the activation of at least one optional sub-module;

[0490] i) Or, the UE feeds back signaling that at least one optional sub-module is activated successfully.

[0491] 3) The AI model in different activation modes meets at least one of the following:

[0492] The performance detection requirement that the AI model can meet in the first activation mode is higher than the performance detection requirement that the AI model can meet in the second activation mode;

[0493] The AI model in the first activation mode has the ability to obtain a first output and a second output, wherein the first output is the output of the AI model in the first activation mode, and the second output is the output of the AI model in the second activation mode;

[0494] The switching delay of the AI model in the first activation mode is greater than the switching delay of the AI model in the second activation mode;

[0495] Wherein, the first activation mode is an activation mode in which the activated sub-modules include the optional sub-module, and the second activation mode is an activation mode in which the activated sub-modules do not include the optional sub-module.

[0496] ⅰ. The performance detection requirement that the AI model can meet in the first activation mode A is higher than the performance detection requirement that the AI model can meet in the second activation mode B;

[0497] Wherein, the performance detection requirement (or Key Performance Indication, KPI) can include but is not limited to at least one of the following:

[0498] System performance indicators, such as at least one of the following indicators: throughput rate, bit error rate, block error rate, and bit error rate;

[0499] at least one of indirect performance indicators or AI model direct indicators, such as mean-square error (MSE), normalized mean squared error (NMSE), Squared Generalized Cosine Similarity (SGCS), and Cosine Similarity.

[0500] The first activation mode A is an activation mode in which the activated sub-modules include the optional sub-module, and the second activation mode B is an activation mode in which the activated sub-modules do not include the optional sub-module.

[0501] ii. The AI model has the capability of obtaining a first output and a second output in the first activation mode A, wherein

[0502] The first output is the output of the AI model in the first activation mode A, and the second output is the output of the AI model in the second activation mode B.

[0503] Optionally, if the AI model has the capability of obtaining the first output and the second output in the first activation mode A, the input of the optional sub-module in the AI model in the first activation mode B includes at least one of the following:

[0504] The second output;

[0505] The input of the AI model;

[0506] Intermediate information of the AI model.

[0507] The first output and the second output can be used by the second device and / or the first device to compare the performance of the AI model in the first activation mode A and the second activation mode B, and the purpose of the performance comparison is to determine which activation mode has better performance of the AI model, or to determine the difference between the performance indicators in the two activation modes.

[0508] In some embodiments, whether the second device makes the performance comparison result and reports it to the first device, the final determination about the performance comparison result is in the first device. Therefore, the model using apparatus 1400 proposed in the embodiments of the present application can further include a third sending module configured to send third feedback information to the first device;

[0509] The third feedback information includes any one of the following:

[0510] the performance comparison result made by the second device about the AI model in the first activation mode and the AI model in the second activation mode;

[0511] the performance comparison result made by the second device, the first output, and the second output;

[0512] the first output and the second output.

[0513] Further, the model using device 1400 proposed in an embodiment of the present application can further include:

[0514] a third receiving module, configured to receive determination information sent by the first device, wherein the determination information is about the performance comparison result, and the determination information is determined by the first device according to the third feedback information;

[0515] a deactivation module, configured to deactivate the optional sub-module in a case where it is determined according to the determination information that the performance of the AI model in the first activation mode is lower than or similar to the performance of the AI model in the second activation mode.

[0516] It can be understood that if the performance of the AI model in the first activation mode A is lower than or similar to the performance of the AI model in the second activation mode B, it means that the activation of the optional sub-module is not helpful for the performance improvement of the AI model, and therefore the optional sub-module can not be activated to save computing power.

[0517] ⅲ. The switching delay of the AI model in the first activation mode A is greater than the switching delay of the AI model in the second activation mode B. It can be understood that since the AI model in the first activation mode A is larger than the AI model in the second activation mode B, the switching delay is longer.

[0518] The second sub-module combination activation mode:

[0519] The second device has at least two functionalities, and different functionalities correspond to activation of different sub-module sets.

[0520] The functionality can refer to a function, a use, a feature, a feature group, and the like that can be implemented by the AI model. For example, spatial beam prediction is one function, and time domain beam prediction is another function; or, time domain beam prediction suitable for low speed is one function, and time domain beam prediction suitable for high speed is another function; or, time domain beam prediction predicting future 4 beams is one function, and time domain beam prediction predicting future 8 beams is another function. Alternatively, the function belongs to the capability of the second device, and is reported to the first device in capability reporting.

[0521] For example, the first function corresponds to a first set of activated sub-modules, and the first set of activated sub-modules includes N3 sub-modules; and the second function corresponds to a second set of activated sub-modules, and the second set of activated sub-modules includes N4 sub-modules. N3 and N4 are integers greater than 0, and N3≤N and N4≤N.

[0522] At this time, the first indication information can be used to instruct the second device to start the first function, or the first indication is used to instruct the second device to switch from the first function to the second function.

[0523] Alternatively, in the case where there is a same sub-module in the first set of sub-modules and the second set of sub-modules, if the first indication information is used to instruct the second device to switch from the first function to the second function, then:

[0524] a) The sub-module that needs to be activated or deactivated indicated in the first indication information does not include the same sub-module.

[0525] b) The second feedback information for the first indication information does not include activation information or deactivation information of the same sub-module.

[0526] The model using apparatus 1100 provided in the embodiments of the present application can implement each process of the method embodiments Figure 2 , and achieve the same technical effects. To avoid repetition, details are not described herein. The model using apparatus 1400 provided in the embodiments of the present application can implement each process of the method embodiments Figure 8 , and achieve the same technical effects. To avoid repetition, details are not described herein.

[0527] As Figure 17As shown, the embodiments of the present application further provide a communication device 1700, which comprises a processor 1701 and a memory 1702, and the memory 1702 stores programs or instructions executable on the processor 1701. For example, when the communication device 1700 is a terminal, the programs or instructions are executed by the processor 1701 to implement the steps of the above model using method embodiments and achieve the same technical effects. When the communication device 1700 is a network side device, the programs or instructions are executed by the processor 1701 to implement the steps of the above model using method embodiments and achieve the same technical effects. To avoid repetition, details are not described here.

[0528] The embodiments of the present application further provide a terminal, which comprises a processor and a communication interface, the communication interface and the processor are coupled, and the processor is configured to run programs or instructions to implement the steps in the above method embodiments. The terminal embodiments correspond to the above terminal side method embodiments, and each implementation process and implementation manner of the above method embodiments can be applied to the terminal embodiments and achieve the same technical effects. The terminal can be the model using apparatus shown in Figure 2 or Figure 8 The terminal can be the model using apparatus shown in Figure 11 or Figure 14 Specifically, Figure 18 A hardware structure diagram of a terminal for implementing the embodiments of the present application.

[0529] The terminal 1800 comprises at least part of the components shown in the drawing, such as a radio frequency unit 1801, a network module 1802, an audio output unit 1803, an input unit 1804, a sensor 1805, a display unit 1806, a user input unit 1807, an interface unit 1808, a memory 1809, and a processor 1810.

[0530] Those skilled in the art can understand that the terminal 1800 can further comprise a power supply (such as a battery) for supplying power to each component, and the power supply can be logically connected to the processor 1810 through a power management system, so as to realize the functions of power management, discharge management, and power consumption management through the power management system. Figure 18 The terminal structure shown in the drawing does not constitute a limitation on the terminal, and the terminal can comprise more or fewer components than shown in the drawing, or combine certain components, or different component arrangements, which are not described here.

[0531] It should be understood that in the embodiments of the present application, the input unit 1804 can include a graphics processor 18041 and a microphone 18042, and the graphics processor 18041 processes image data of a still picture or a video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 1806 can include a display panel 18061, which can be configured in the form of a liquid crystal display, an organic light-emitting diode, etc. The user input unit 1807 includes at least one of a touch panel 18071 and other input devices 18072. The touch panel 18071 is also called a touch screen. The touch panel 18071 can include two parts of a touch detection device and a touch controller. The other input devices 18072 can include, but are not limited to, a physical keyboard, function keys (such as volume control keys, on-off keys, etc.), a trackball, a mouse, a joystick, and the like, which will not be described here.

[0532] In the embodiments of the present application, after the radio frequency unit 1801 receives the downlink data from the network side device, it can be transmitted to the processor 1810 for processing. In addition, the radio frequency unit 1801 can send uplink data to the network side device. Generally, the radio frequency unit 1801 includes, but is not limited to, an antenna, an amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc.

[0533] The memory 1809 can be used to store software programs or instructions and various data. The memory 1809 can mainly include a first storage area storing programs or instructions and a second storage area storing data, wherein the first storage area can store an operating system, application programs or instructions required by at least one function (such as a sound playing function, an image playing function, etc.), etc. In addition, the memory 1809 can include a volatile memory or a non-volatile memory. The non-volatile memory can be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a Random Access Memory (RAM), a Static RAM (SRAM), a Dynamic RAM (DRAM), a Synchronous DRAM (SDRAM), a Double Data Rate SDRAM (DDR SDRAM), an Enhanced SDRAM (ESDRAM), a Synch link DRAM (SLDRAM), and a Direct Rambus RAM (DRRAM). The memory 1809 in the embodiments of the present application includes but is not limited to these and any other suitable types of memory.

[0534] The processor 1810 can include one or more processing units; optionally, the processor 1810 integrates an application processor and a modem processor, wherein the application processor mainly processes operations related to an operating system, a user interface, and an application program, and the modem processor mainly processes wireless communication signals, such as a baseband processor. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 1810.

[0535] The radio frequency unit 1801 is configured to send first indication information to a second device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs and / or does not need to be activated by the second device, and the AI model includes N sub-modules, N is an integer greater than 0.

[0536] Alternatively, the radio frequency unit 1801 is configured to receive first indication information sent by the first device, wherein the first indication information is used to indicate at least one sub-module belonging to an AI model that needs to be activated by the second device or does not need to be activated by the second device, the AI model includes N sub-modules, and N is an integer greater than 0.

[0537] The terminal provided in the embodiments of the present application can divide a complete AI model into multiple sub-modules, and activate at least one sub-module, instead of directly activating the entire AI model, so that the flexibility of AI model use can be improved, and the hardware complexity and computing power requirement can be reduced.

[0538] It can be understood that the implementation processes of the implementation manners mentioned in the embodiments can refer to the related descriptions of the method embodiments, and achieve the same or corresponding technical effects. To avoid repetition, they will not be described here again.

[0539] The embodiments of the present application also provide a network side device, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to realize the steps of the method embodiments as shown in Figure 2 or Figure 8 The network side device embodiments correspond to the network side device method embodiments described above. The implementation processes and implementation manners of the method embodiments described above can be applied to the network side device embodiments, and the same technical effects can be achieved.

[0540] Specifically, the embodiments of the present application also provide a network side device, which can be the model use apparatus as shown in Figure 11 or Figure 14 The network side device 1900 includes an antenna 191, a radio frequency device 192, a baseband device 193, a processor 194, and a memory 195 as shown in Figure 19 The antenna 191 is connected with the radio frequency device 192. In the uplink direction, the radio frequency device 192 receives information through the antenna 191, and sends the received information to the baseband device 193 for processing. In the downlink direction, the baseband device 193 processes the information to be sent and sends it to the radio frequency device 192. The radio frequency device 192 processes the received information and sends it out through the antenna 191.

[0541] The method performed by the network side device in the above embodiments can be implemented in the baseband device 193, which includes a baseband processor.

[0542] The baseband device 193 may, for example, include at least one baseband board on which a plurality of chips are arranged, such as Figure 19As shown in the figure, one of the chips, for example, is a baseband processor, which is connected with the memory 195 through a bus interface to invoke a program in the memory 195 to perform the network side device operation shown in the above method embodiment.

[0543] The network side device can further include a network interface 196, which is, for example, a common public radio interface (CPRI).

[0544] Specifically, the network side device 1900 of the embodiment of the present application further includes instructions or programs stored in the memory 195 and executable on the processor 194, and the processor 194 invokes the instructions or programs in the memory 195 to perform the method executed by each module shown in the figure and achieve the same technical effects, and thus the details are not described herein. Figure 11 Or Figure 14 The method executed by each module shown in the figure and achieve the same technical effects, and thus the details are not described herein.

[0545] Specifically, the embodiment of the present application further provides a network side device. As shown in the figure, Figure 20 The network side device 2000 includes a processor 2001, a network interface 2002, and a memory 2003. The network side device can be Figure 11 Or Figure 14 The model using device shown in the figure. The network interface 2002 is, for example, a common public radio interface (CPRI).

[0546] Specifically, the network side device 2000 of the embodiment of the present application further includes instructions or programs stored in the memory 2003 and executable on the processor 2001, and the processor 2001 invokes the instructions or programs in the memory 2003 to perform the method executed by each module shown in the figure and achieve the same technical effects, and thus the details are not described herein. Figure 11 Or Figure 14 The method executed by each module shown in the figure and achieve the same technical effects, and thus the details are not described herein.

[0547] The embodiment of the present application further provides a readable storage medium, and the readable storage medium stores programs or instructions, which are executed by a processor to implement each process of the above model using method embodiment and achieve the same technical effects. To avoid repetition, the details are not described herein.

[0548] The processor is the processor in the terminal described in the above embodiments. The readable storage medium includes a computer readable storage medium, such as a computer readable only memory (ROM), a random access memory (RAM), a magnetic disc or an optical disc, etc. In some examples, the readable storage medium can be a non-transitory readable storage medium.

[0549] The chip provided by the embodiment of the present application also can be called a system chip, a chip system, a system on chip, or the like.

[0550] It should be understood that the chip mentioned in the embodiment of the present application can also be called a system chip, a chip system, a system on chip, or the like.

[0551] The embodiment of the present application further provides a computer program / program product stored in a storage medium, which is executed by at least one processor to implement the processes of the model using method embodiments and achieve the same technical effects. To avoid repetition, details are not repeated here.

[0552] The embodiment of the present application further provides a communication system, which comprises a terminal and a network side device. Figure 2 The terminal can be used to execute the steps of the model using method, and the network side device can be used to execute the steps of the model using method. Figure 13 The terminal can be used to execute the steps of the model using method, and the network side device can be used to execute the steps of the model using method.

[0553] It should be noted that in this document, the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, articles, or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or further include elements inherent to such processes, methods, articles, or devices. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article, or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to the order of performing the functions shown or discussed, and can also include performing the functions in a substantially simultaneous manner or in a reverse order, for example, the described method can be performed in an order different from that described, and various steps can also be added, omitted, or combined. In addition, the features described with reference to certain examples can be combined in other examples.

[0554] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of computer software products and general hardware platforms, of course, they can also be realized by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, magnetic disc, optical disc, etc.), which includes a plurality of instructions for making the terminal or network side device execute the method described in each embodiment of the present application.

[0555] The embodiments of the present application are described above with reference to the accompanying drawings, but the present application is not limited to the above-described specific embodiments, and the above-described specific embodiments are merely illustrative, but not restrictive, and a person of ordinary skill in the art can make many forms of embodiments under the inspiration of the present application without departing from the purpose of the present application and the scope protected by the claims, and these embodiments all belong to the protection of the present application.

Claims

1. A method for using a model, characterized in that, The method includes: The first device sends a first instruction message to the second device, wherein the first instruction message is used to indicate at least one sub-module belonging to the AI ​​model that needs to be activated by the second device and / or does not need to be activated. The AI ​​model includes N sub-modules, where N is an integer greater than 0.

2. The method according to claim 1, characterized in that, Before the first device sends the first indication information to the second device, the method further includes: The first device sends a second instruction message to the second device, wherein the second instruction message is used to instruct the pre-activation of the AI ​​model.

3. The method according to claim 2, characterized in that, The method further includes: The first device receives a first feedback message from the second device in response to the first indication information and / or the second indication information.

4. The method according to claim 3, characterized in that, The first feedback information includes at least one of the following: The second device supports the identification information of the activated sub-module; The second device does not support the identification information of the activated submodule; Confirmation information regarding whether the second device supports activating the at least one sub-module.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: The first device receives second feedback information sent by the second device in response to the first indication information.

6. The method according to claim 5, characterized in that, The second feedback information includes at least one of the following: The identification information of the sub-module to be activated; The identification information of the successfully activated submodule; Identification information of the submodule that failed to activate; Identification information of the inactive submodule; Signals or messages indicating successful or failed activation; The signal or information indicating whether the AI ​​model was successfully activated or failed to activate.

7. The method according to claim 1, characterized in that, The first indication information is used to indicate, based on the submodule combination activation method, at least one submodule belonging to the AI ​​model that needs and / or does not need to be activated by the second device.

8. The method according to claim 7, characterized in that, The submodule combination activation method is as follows: the N submodules include N1 necessary submodules and N2 optional submodules. The N1 necessary submodules are submodules that need to be activated by the second device by default, and the activation of the N1 necessary submodules does not require the first device to send the first indication information to the second device. Here, N1 and N2 are integers greater than 0, and N1+N2≤N. The first indication information is used to indicate at least one of the optional sub-modules that require activation by the second device.

9. The method according to claim 8, characterized in that, The second feedback information in response to the first indication does not include information indicating that the necessary submodule has been successfully activated.

10. The method according to claim 8, characterized in that, The AI ​​model meets at least one of the following criteria: The performance testing requirements that the AI ​​model can meet in the first activation mode are higher than those that the AI ​​model can meet in the second activation mode. The AI ​​model has the ability to obtain a first output and a second output in the first activation mode, wherein the first output is the output of the AI ​​model in the first activation mode and the second output is the output of the AI ​​model in the second activation mode. The switching latency of the AI ​​model in the first activation mode is greater than the switching latency of the AI ​​model in the second activation mode; Wherein, the submodule activated by the first activation mode includes the activation mode of the optional submodule, and the submodule activated by the second activation mode does not include the activation mode of the optional submodule.

11. The method according to claim 10, characterized in that, If the AI ​​model has the ability to obtain the first output and the second output in the first activation mode, then the input of the optional sub-module in the AI ​​model in the first activation mode includes at least one of the following: The second output; The input to the AI ​​model; The intermediate information of the AI ​​model.

12. The method according to claim 10, characterized in that, The method further includes: The first device receives the third feedback information sent by the second device; The third feedback information includes any one of the following: The performance comparison results of the AI ​​model in the first activation mode and the AI ​​model in the second activation mode made by the second device; The performance comparison result made by the second device, the first output, and the second output; The first output and the second output.

13. The method according to claim 12, characterized in that, The method further includes: The first device determines judgment information regarding the performance comparison result based on the third feedback information; The first device sends the determination information to the second device.

14. The method according to claim 7, characterized in that, The sub-module combination activation method is as follows: the second device has at least two functions, and different functions correspond to the activation of different sub-module sets; The first indication information is used to instruct the second device to enable the first function, or the first indication is used to instruct the second device to switch from the first function to the second function; Wherein, the set of sub-modules activated corresponding to the first function is the first set of sub-modules, and the set of sub-modules activated corresponding to the second function is the second set of sub-modules.

15. The method according to claim 14, characterized in that, If the same submodule exists in both the first submodule set and the second submodule set, and if the first indication information is used to instruct the second device to switch from the first function to the second function, then The sub-modules that need to be activated or deactivated as indicated in the first indication information do not include the same sub-modules; and / or The second feedback information in response to the first indication information does not include activation or deactivation information of the same sub-module.

16. The method according to any one of claims 1 to 15, characterized in that, The submodule has corresponding submodule information, wherein the submodule information includes at least one of the following: The identification information of the submodule; Version information of the submodule.

17. The method according to claim 16, characterized in that, in, The identification information of the submodule includes at least one of the following: The identifier of the submodule; The position information of the submodule in the AI ​​model.

18. The method according to claim 17, characterized in that, in, The version information of the submodule includes at least one of the following: The timestamp of the submodule; The relevant information associated with the submodule or the AI ​​model may include at least one of dataset information, data feature information, and data feature identifier; The submodule or the AI ​​model may be associated with at least one of the following: base station hardware information, base station configuration information, cell information, physical cell information, serving cell information, region information, cell group information, and cell list information; At least one of the functions and features associated with the submodule or the AI ​​model.

19. The method according to any one of claims 1 to 18, characterized in that, The AI ​​model includes at least one of the following: The second device uses the first AI model; The reference model for the first AI model; The second AI model used in the test by the second device or test device; The reference model for the second AI model; The second device or testing device is used to match the third AI model of the second AI model used in the test; The reference model of the third AI model.

20. A method for using a model, characterized in that, The method includes: The second device receives a first indication message sent by the first device, wherein the first indication message is used to indicate at least one sub-module belonging to the AI ​​model that needs to be activated by the second device and / or does not need to be activated, and the AI ​​model includes N sub-modules, where N is an integer greater than 0.

21. The method according to claim 20, characterized in that, Before the second device receives the first indication information sent by the first device, the method further includes: The second device receives a second indication message sent by the first device, wherein the second indication message is used to indicate the pre-activation of the AI ​​model.

22. The method according to claim 21, characterized in that, The method further includes: The second device sends a first feedback message to the first device in response to the first indication message and / or the second indication message.

23. The method according to claim 22, characterized in that, The first feedback information includes at least one of the following: The second device supports the identification information of the activated sub-module; The second device does not support the identification information of the activated submodule; Confirmation information regarding whether the second device supports activating the at least one sub-module.

24. The method according to any one of claims 20 to 23, characterized in that, The method further includes: The second device sends a second feedback message to the first device in response to the first instruction message.

25. The method according to claim 24, characterized in that, The second feedback information includes at least one of the following: The identification information of the sub-module to be activated; The identification information of the successfully activated submodule; Identification information of the submodule that failed to activate; Identification information of the inactive submodule; Signals or messages indicating successful or failed activation; The signal or information indicating whether the AI ​​model was successfully activated or failed to activate.

26. The method according to claim 20, characterized in that, The first indication information is used to indicate, based on the submodule combination activation method, at least one submodule belonging to the AI ​​model that needs and / or does not need to be activated by the second device.

27. The method according to claim 26, characterized in that, The submodule combination activation method is as follows: the N submodules include N1 necessary submodules and N2 optional submodules. The N1 necessary submodules are submodules that need to be activated by the second device by default, and the activation of the N1 necessary submodules does not require the first device to send the first indication information to the second device. Here, N1 and N2 are integers greater than 0, and N1+N2≤N. The first indication information is used to indicate at least one of the optional sub-modules that require activation by the second device.

28. The method according to claim 27, characterized in that, The second feedback information in response to the first indication does not include information indicating that the necessary submodule has been successfully activated.

29. The method according to claim 27, characterized in that, The AI ​​model meets at least one of the following criteria: The performance testing requirements that the AI ​​model can meet in the first activation mode are higher than those that the AI ​​model can meet in the second activation mode. The AI ​​model has the ability to obtain a first output and a second output in the first activation mode, wherein the first output is the output of the AI ​​model in the first activation mode and the second output is the output of the AI ​​model in the second activation mode. The switching latency of the AI ​​model in the first activation mode is greater than the switching latency of the AI ​​model in the second activation mode; Wherein, the submodule activated by the first activation mode includes the activation mode of the optional submodule, and the submodule activated by the second activation mode does not include the activation mode of the optional submodule.

30. The method according to claim 29, characterized in that, If the AI ​​model has the ability to obtain the first output and the second output in the first activation mode, then the input of the optional sub-module in the AI ​​model in the first activation mode includes at least one of the following: The second output; The input to the AI ​​model; The intermediate information of the AI ​​model.

31. The method according to claim 29, characterized in that, The method further includes: The second device sends a third feedback message to the first device; The third feedback information includes any one of the following: The performance comparison results of the AI ​​model in the first activation mode and the AI ​​model in the second activation mode made by the second device; The performance comparison result made by the second device, the first output, and the second output; The first output and the second output.

32. The method according to claim 31, characterized in that, The method further includes: The second device receives the determination information sent by the first device, wherein the determination information is about the performance comparison result, and the determination information is determined by the first device based on the third feedback information; If the second device determines, based on the determination information, that the performance of the AI ​​model in the first activation mode is lower than or approximately the performance of the AI ​​model in the second activation mode, it will deactivate the optional sub-module.

33. The method according to claim 26, characterized in that, The sub-module combination activation method is as follows: the second device has at least two functions, and different functions correspond to the activation of different sub-module sets; The first indication information is used to instruct the second device to enable the first function, or the first indication is used to instruct the second device to switch from the first function to the second function; Wherein, the set of sub-modules activated corresponding to the first function is the first set of sub-modules, and the set of sub-modules activated corresponding to the second function is the second set of sub-modules.

34. The method according to claim 33, characterized in that, If the same submodule exists in both the first submodule set and the second submodule set, and if the first indication information is used to instruct the second device to switch from the first function to the second function, then The sub-modules that need to be activated or deactivated as indicated in the first indication information do not include the same sub-modules; and / or The second feedback information in response to the first indication information does not include activation or deactivation information of the same sub-module.

35. The method according to any one of claims 20 to 34, characterized in that, The submodule has corresponding submodule information, wherein the submodule information includes at least one of the following: The identification information of the submodule; Version information of the submodule.

36. The method according to claim 35, characterized in that, in, The identification information of the submodule includes at least one of the following: The identifier of the submodule; The position information of the submodule in the AI ​​model.

37. The method according to claim 35, characterized in that, in, The version information of the submodule includes at least one of the following: The timestamp of the submodule; The relevant information associated with the submodule or the AI ​​model may include at least one of dataset information, data feature information, and data feature identifier; The submodule or the AI ​​model may be associated with at least one of the following: base station hardware information, base station configuration information, cell information, physical cell information, serving cell information, region information, cell group information, and cell list information; The AI ​​functions and / or AI features associated with the submodule or the AI ​​model.

38. The method according to any one of claims 20 to 37, characterized in that, The AI ​​model includes at least one of the following: The second device uses the first AI model; The reference model for the first AI model; The second AI model used in the test by the second device or test device; The reference model for the second AI model; The second device or testing device is used to match the third AI model of the second AI model used in the test; The reference model of the third AI model.

39. A model-using device, characterized in that, include: A first sending module is configured to send first indication information to a second device, wherein the first indication information is configured to indicate at least one sub-module belonging to an AI model that needs to be activated by the second device and / or does not need to be activated, the AI ​​model comprising N sub-modules, where N is an integer greater than 0.

40. A model-using device, characterized in that, include: The first receiving module is configured to receive first indication information sent by the first device, wherein the first indication information is used to indicate at least one sub-module belonging to the AI ​​model that needs to be activated by the second device and / or does not need to be activated, and the AI ​​model includes N sub-modules, where N is an integer greater than 0.

41. A terminal, characterized in that, It includes a processor and a memory, the memory storing a program or instructions executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the model-using method as described in any one of claims 1 to 19, or, when executed by the processor, the program or instructions implement the steps of the model-using method as described in any one of claims 20 to 38.

42. A network-side device, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the model using method as described in any one of claims 20 to 38, or the program or instructions being executed by the processor to implement the steps of the model using method as described in any one of claims 1 to 19.

43. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the model usage method as described in any one of claims 1-19, or implement the steps of the model usage method as described in any one of claims 20-38.

Images