Capability reporting method, capability determination method, and apparatus

Abstract

The present disclosure provides a capability reporting method, a capability determining method, and an apparatus. The capability reporting method comprises: sending capability information of artificial intelligence (AI) or machine learning (ML), wherein the capability information includes at least one of the following: first capability information of a terminal, second capability information occupied by a channel state information (CSI) report, computation latency requirement information of the CSI report, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal. The capability reporting method, capability determining method, and apparatus provided in the present disclosure can measure or quantify AI- or ML-related capabilities on a terminal side, thereby effectively implementing AI or ML technology-based management on the terminal side.

Description

Method and apparatus for reporting capability and determining capability

[0001] The present disclosure claims priority to a Chinese patent application No. 202410862973.X, filed on June 28, 2024, and entitled "Method and apparatus for reporting capability and determining capability", the entire content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present disclosure relates to the field of communication technology, and in particular to a method and apparatus for reporting capability and determining capability. BACKGROUND

[0003] In current research, air interface enhancement based on artificial intelligence (English: Artificial Intelligence, AI for short) / machine learning (English: Machine Learning, ML for short) includes AI or ML based beam management, positioning, channel state information (English: Channel State Information, CSI for short) compression feedback, and CSI prediction. In these use cases, AI or ML models can be deployed on the terminal side or partially deployed on the terminal side.

[0004] In the discussion, for a single-sided model, such as a model deployed only on the terminal side, model training can be completed by the user equipment / terminal (English: User Equipment, UE for short) side, and the model structure is also determined by the UE side itself. For a double-sided model, AI or ML models need to be deployed on both the terminal side and the base station side, for example, in the CSI compression feedback use case, AI or ML models need to be deployed on both the terminal side and the base station side, the terminal side model is used for compressing CSI, and the base station side model is used for decompressing CSI.

[0005] However, the related art cannot measure or quantify the terminal side AI or ML related capability, and thus cannot effectively implement management of the terminal side based on AI or ML technology. SUMMARY

[0006] The present disclosure provides a method and apparatus for reporting capability and determining capability, which solves the technical problem that the related art cannot measure or quantify the terminal side AI or ML related capability, and thus cannot effectively implement management of the terminal side based on AI or ML technology.

[0007] In a first aspect, the present disclosure provides a method for reporting capability, applied to a terminal, the method comprising:

[0008] transmitting artificial intelligence (AI) or machine learning (ML) capability information, wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of channel state information (CSI) report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0009] In the embodiments of the present disclosure, the terminal can report at least one of: first capability information based on AI or ML, second capability information of CSI report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the receiving end can know the AI or ML related capability of the terminal side, and then the management of the terminal side based on the AI or ML technology can be realized.

[0010] In a second aspect, the present disclosure provides a capability determination method, applied to a network device, and the method comprises:

[0011] receiving AI or ML capability information of a terminal, wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0012] In the embodiments of the present disclosure, the network device receives at least one of: first capability information based on AI or ML, second capability information of CSI report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, which is reported by the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the management of the terminal side based on the AI or ML technology can be realized.

[0013] In a third aspect, the present disclosure provides a capability reporting apparatus, applied to a terminal, and the apparatus comprises a memory, a transceiver, and a processor.

[0014] The memory is configured to store a computer program; the transceiver is configured to transceive data under the control of the processor; and the processor is configured to read the computer program in the memory and perform the following operations:

[0015] transmitting artificial intelligence (AI) or machine learning (ML) capability information, wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of channel state information (CSI) report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0016] In the embodiments of the present disclosure, the terminal can report at least one of: first capability information based on AI or ML, second capability information of CSI report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the receiving end can know the AI or ML related capability of the terminal side, and then the management of the terminal side based on the AI or ML technology can be realized.

[0017] In a fourth aspect, the present disclosure provides a capability determination apparatus, which is applied to a network device, and the apparatus comprises: a memory, a transceiver, and a processor.

[0018] The memory is configured to store a computer program; the transceiver is configured to transceive data under the control of the processor; and the processor is configured to read the computer program in the memory and perform the following operations:

[0019] receiving AI or ML capability information of a terminal, wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0020] In the embodiments of the present disclosure, the network device receives at least one of: first capability information based on AI or ML, second capability information of CSI report occupation, calculation latency requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, which are reported by the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the management of the terminal side based on the AI or ML technology can be realized.

[0021] In a fifth aspect, the present disclosure provides a capability reporting apparatus, which is applied to a terminal, and the apparatus comprises:

[0022] The sending unit is configured to send AI or ML capability information, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, calculation time delay requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0023] In the embodiments of the present disclosure, the terminal can report at least one of the following: first capability information based on AI or ML, second capability information of CSI report occupation, calculation time delay requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, so as to measure or quantify the AI or ML related capability of the terminal, and then the receiving end can know the AI or ML related capability of the terminal, and the management of the terminal based on AI or ML technology can be realized.

[0024] In a sixth aspect, the present disclosure provides a capability determination apparatus, which is applied to a network device, and the apparatus comprises:

[0025] The receiving unit is configured to receive AI or ML capability information of the terminal, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0026] In the embodiments of the present disclosure, the network device can measure or quantify the AI or ML related capability of the terminal by receiving at least one of the following reported by the terminal: first capability information based on AI or ML, second capability information of CSI report occupation, calculation time delay requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, so as to realize the management of the terminal based on AI or ML technology.

[0027] In a seventh aspect, the present disclosure provides a processor readable storage medium, which stores a computer program for causing a processor to execute the method of any one of the first aspect or the second aspect.

[0028] The disclosure provides a capability reporting method, a capability determining method and device, a terminal reports at least one of the following: first capability information based on AI or ML, second capability information occupied by CSI reporting, calculation time delay requirement information of CSI reporting, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal, so as to measure or quantify AI or ML related capability on the terminal side, and then the receiving end can know the AI or ML related capability on the terminal side, and then the management of the terminal based on AI or ML technology is realized.

[0029] It should be understood that the content described in the foregoing summary section is not intended to define key or important features of the embodiments of the disclosure, nor is it intended to limit the scope of the disclosure. Other features of the disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the disclosure or the related art, the following will briefly introduce the drawings needed to be used in the embodiments or related art description. Obviously, the drawings in the following description are some embodiments of the disclosure, and other drawings can also be obtained by those skilled in the art without creative labor.

[0031] Fig. 1 is an interaction schematic diagram of the capability reporting method provided by the embodiments of the disclosure;

[0032] Fig. 2 is an interaction schematic diagram of the capability determining method provided by the embodiments of the disclosure;

[0033] Fig. 3 is a structural schematic diagram one of the capability reporting device provided by the embodiments of the disclosure;

[0034] Fig. 4 is a structural schematic diagram two of the capability reporting device provided by the embodiments of the disclosure;

[0035] Fig. 5 is a structural schematic diagram one of the capability determining device provided by the embodiments of the disclosure;

[0036] Fig. 6 is a structural schematic diagram two of the capability determining device provided by the embodiments of the disclosure. DETAILED DESCRIPTION

[0037] In the embodiments of the disclosure, the term "and / or" describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone. The character " / " generally represents a "or" relationship between the front and rear associated objects.

[0038] In the embodiments of the disclosure, the term "a plurality of" means two or more, and other quantifiers are similar.

[0039] The technical solutions in the embodiments of the present disclosure will be described clearly and completely in combination with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, and not all the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present disclosure.

[0040] The embodiments of the present disclosure provide a capability reporting method, a configuration method and apparatus, which report or receive the capability information of a terminal, such as at least one of the following: first capability information based on AI or ML, second capability information occupied by CSI reporting, calculation latency requirement information of CSI reporting, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the receiving end can learn the AI or ML related capability of the terminal side, and further implement management of the terminal side based on AI or ML technology.

[0041] The method and the apparatus are based on the same application concept. Since the principles of the method and the apparatus for solving problems are similar, the implementation of the apparatus and the method can be referred to each other, and the repeated parts will not be described again.

[0042] Embodiments of the present disclosure will be described below with reference to the accompanying drawings. In the case of no conflict between the embodiments, the embodiments and the features in the embodiments described below can be combined with each other. In addition, the sequence of steps in each method embodiment described below is only an example, and is not strictly limited.

[0043] Referring to FIG. 1, FIG. 1 is an interaction schematic diagram of a capability reporting method provided by the embodiments of the present disclosure. The execution subject of the capability reporting method provided by the present embodiment is a terminal (UE), and the capability reporting method is described in detail below.

[0044] The capability reporting method provided by the embodiments of the present disclosure includes the following steps:

[0045] The capability information of artificial intelligence (AI) or machine learning (ML) is transmitted.

[0046] The capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by channel state information (CSI) reporting, calculation latency requirement information of CSI reporting, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal.

[0047] In the embodiments of the present disclosure, the first capability information is used to represent AI or ML related capability information of the terminal, and the terminal can report its AI or ML related capability to the receiving end (such as a network device (including a base station, a core network device, etc.), which is taken as an example of a network device below), such as an absolute value of the first capability and / or a relative value of the first capability; wherein the AI or ML related capability can include a computing capability and / or a storage capability, that is, the first capability includes a first computing capability and / or a first storage capability.

[0048] The second capability information is used to represent related capability information of the terminal occupied by the AI or ML based CSI report, and the terminal can report the related capability information of the terminal occupied by the AI or ML based CSI report to the network device, such as a number of occupied CSI processing units.

[0049] The third capability information is used to represent AI or ML related capability information occupied or required by an AI or ML model (that is, an AI model or an ML model, which will not be described below) deployed by the terminal or externally connected (or jointly used by other terminals). The terminal can report AI or ML related capability occupied or required by the AI or ML model to the network device, such as an absolute value of the third capability and / or a relative value of the third capability; wherein the AI or ML related capability can include a computing capability and / or a storage capability, that is, the third capability includes a third computing capability and / or a third storage capability.

[0050] The fourth capability information is used to represent AI or ML related capability information occupied or required by an AI or ML function (that is, an AI function or an ML function, which will not be described below) deployed by the terminal or externally connected. The terminal can report AI or ML related capability occupied or required by the AI or ML function to the network device, such as an absolute value of the fourth capability and / or a relative value of the fourth capability; wherein the AI or ML related capability can include a computing capability and / or a storage capability, that is, the fourth capability includes a fourth computing capability and / or a fourth storage capability.

[0051] The computing latency requirement information of the CSI report is used to represent a computing latency requirement for completing CSI computation for CSI report reporting when the terminal performs AI or ML based CSI processing.

[0052] For example, in combination with FIG. 1, FIG. 1 is an interaction schematic diagram of the capability reporting method provided by the embodiments of the present disclosure.

[0053] Step 101: The terminal sends AI or ML capability information to the network device. The capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by the CSI report, computing latency requirement information of the CSI report, third capability information of the AI or ML model occupied or required by the terminal, or fourth capability information of the AI or ML function occupied or required by the terminal.

[0054] The AI or ML capability information can be carried by signaling or a signal. For example, the terminal sends first signaling to the network device (here, the network side, which will not be described below), and the first signaling carries the AI or ML capability information.

[0055] Since the computing resources and storage resources of the terminal side (i.e., the terminal) are limited, the AI or ML models that the terminal can activate or run are also limited. After the terminal side reports its AI or ML models or AI or ML functions to the network device, the network device can select AI or ML models or AI or ML functions to activate or deactivate. In order to ensure that the AI or ML models or AI or ML functions of the terminal side activated or running at any time do not exceed the capability (here, the first capability) of the terminal, the terminal can report to the network device its AI or ML related capability (here, the first capability), and the AI or ML related capability occupied or required by each AI or ML model or AI or ML function deployed on the terminal side.

[0056] In addition, for AI or ML based CSI reporting, the difference between the CSI processing unit occupied by AI or ML based CSI reporting and the CSI calculation delay and the traditional (non-AI or ML) CSI reporting needs to be considered, therefore, the terminal side can report the CSI report calculation delay requirement information to the network device, and then the network device can manage the AI or ML based CSI processing of the terminal side.

[0057] In the embodiments of the present disclosure, the terminal can report at least one of the following: first capability information based on AI or ML, second capability information occupied by CSI reporting, calculation delay requirement information of CSI reporting, third capability information occupied or required by AI or ML models of the terminal, or fourth capability information occupied or required by AI or ML functions of the terminal, which can measure or quantify the AI or ML related capability of the terminal side, and then the receiving end can know the AI or ML related capability of the terminal side, and then the management of the terminal side based on AI or ML technology can be realized.

[0058] In some embodiments, the first capability information includes a relative value of the first capability of the terminal.

[0059] In the embodiments of the present disclosure, the first capability information includes: an absolute value of the first capability of the terminal, and / or a relative value of the first capability of the terminal. The absolute value of the first capability can be a specific numerical value for measuring the size of the first capability (such as the size of the first computing capability and / or the size of the first storage capability) of the terminal, for example, the size of the first computing capability is measured by the number of floating point operations per second (English: Floating point operations per second, FLOPS for short), and the size of the first storage capability is measured by the occupied storage space. The AI or ML related capability reported by the terminal is one or more of the first computing capability and the first storage capability possessed by the terminal.

[0060] The relative value of the first capability can be a relative value for measuring the size of the first capability (the size of the first computing capability and / or the size of the first storage capability) of the terminal.

[0061] In some embodiments, the first capability information includes first computing capability information and / or first storage capability information.

[0062] The first computing capability information includes any of the following: the size of the first computing capability, the ratio relative to the first target computing capability information, the maximum value of the processing reference model or the reference model; and the first storage capability information includes any of the following: the size of the first storage capability, the ratio relative to the first target storage capability information, the maximum value of the storage reference model or the maximum value of the reference model.

[0063] In the embodiments of the present disclosure, the first capability information includes any of the following: the size of the first computing capability, the ratio relative to the first target computing capability information, the maximum value of the processing reference model or the reference model; and / or the first capability information includes any of the following: the size of the first storage capability, the ratio relative to the first target storage capability information, the maximum value of the storage reference model or the maximum value of the reference model.

[0064] The processing of the processing reference model or the reference model can include activating or running or using the model (such as an AI model or an ML model) for inference.

[0065] In the embodiments of the present disclosure, the method for the terminal to report the AI or ML related capability possessed by the terminal to the network device includes:

[0066] The AI or ML related capability reported by the terminal is an absolute value of the capability possessed by the terminal (in combination with mode 1-1);

[0067] The AI or ML related capability reported by the terminal is a relative value, including:

[0068] A capability unit is defined, and a relative value of the relative capability unit is reported (in combination with mode 1-2).

[0069] The terminal reports the AI or ML related capability based on the capability of processing the reference model.

[0070] Option 1-1: The AI or ML related capability reported by the terminal is one or more of the first computing capability and the first storage capability.

[0071] Option 1-2: The first target computing capability information can be a predefined or configured computing capability unit, or any predefined or configured computing capability size; and the first target storage capability information can be a predefined or configured storage capability unit, or any predefined or configured storage capability size.

[0072] The terminal reports the AI or ML related capability based on the X computing capability and / or Y bytes.

[0073] For example, X FLOPS is defined as 1 AI or ML processing unit, and the computing capability (here, the first computing capability size) of the UE is A FLOPS. The AI or ML related capability (here, the AI or ML processing capability reported by the UE, i.e., the first capability information or the first computing capability information) reported by the UE is or

[0074] For example, Y bytes is defined as 1 AI or ML storage unit, and the storage capability (here, the first storage capability size) of the UE is B bytes. The AI or ML related capability (here, the AI or ML storage capability reported by the UE, i.e., the first capability information or the first storage capability information) reported by the UE is or

[0075] The first target storage capability information can be a predefined or configured storage capability unit, or any predefined or configured storage capability size.

[0076] Option 1-3: The terminal reports the AI or ML related capability based on the capability of processing the reference model.

[0077] For example, a reference model is defined, and the complexity of the model (i.e., the reference model) is taken as the reference. The AI or ML processing capability (here, the first computing capability) reported by the terminal is the number of reference models (i.e., the maximum number of reference models that can be activated or run simultaneously) that can be activated or run simultaneously. The number or maximum value can be an integer or a decimal number, which needs to be determined according to the specific capability information, and is not limited here.

[0078] For example, a reference model or a reference model is defined, and the complexity or model size (here, the storage space occupied by the model) of the model (i.e., the reference model or the reference model) is taken as a reference. The AI or ML storage capability (here, the first storage capability) reported by the terminal is the number (i.e., the maximum number of storage reference models or the maximum number of storage reference models) of reference models or reference models that can be simultaneously stored. The number or maximum value can be an integer or a decimal number, which needs to be determined according to the specific capability information, which is not limited here.

[0079] In some embodiments, the first capability information includes first computing capability information, and the method further includes any one of the following:

[0080] According to the first computing capability size of the terminal, determine the first computing capability size in the first capability information;

[0081] According to the first computing capability size and the first target computing capability information, determine the ratio of the first capability information to the first target computing capability information;

[0082] According to the first computing capability size, and the first model information of the reference model or the first model information of the reference model, determine the maximum number of reference models or the maximum number of reference models in the first capability information.

[0083] The first model information of the reference model or the first model information of the reference model includes the number of parameters and / or the complexity of the model. The number of parameters can reflect the complexity of the model, that is, the number of parameters and the complexity are related. The greater the number of parameters, the greater the complexity, and vice versa. The complexity can be measured by floating point operations (English: Floating point operations, abbreviated: FLOPs). For example, the complexity of the reference model is defined as 10^6 FLOPs, and the first computing capability (here, the first computing capability size) of the terminal is 7*10^12 FLOPS. The computing capability (here, the first computing capability information carried in the capability information) reported by the terminal is: 7*10^6.

[0084] In the embodiments of the present disclosure, if the first capability information includes the first computing capability information, and the terminal reports the absolute value of the first capability of the terminal to the network device, the absolute value of the first capability of the terminal reported here can be the first computing capability size of the terminal. The first computing capability size of the terminal is taken as the first computing capability information, and the first computing capability size carried in the first capability information is determined. That is, according to the first computing capability size of the terminal, the first computing capability size carried in the first capability information is determined.

[0085] If the first capability information includes the first computing capability information, if the terminal reports the relative value of the first capability of the terminal to the network device, the reported relative value of the first capability of the terminal here can be the ratio of the first computing capability size of the terminal to the first target computing capability information, or the first computing capability size of the terminal is equivalent to the maximum number of simultaneously processing (including activating or running) the reference model or the maximum number of simultaneously processing (including activating or running) the reference model. The ratio of the first computing capability size of the terminal to the first target computing capability information is taken as the first computing capability information, and it is determined that the first capability information carries the ratio relative to the first target computing capability information. Alternatively, the maximum number of processing the reference model or the maximum number of processing the reference model (herein referred to as the maximum number of processing the reference model) is taken as the first computing capability information, and it is determined that the first capability information carries the maximum number of processing the reference model or the maximum number of processing the reference model. That is:

[0086] According to the first computing capability size of the terminal and the first target computing capability information, it is determined that the first capability information carries the ratio relative to the first target computing capability information; or according to the first computing capability size, and the first model information of the reference model or the first model information of the reference model, it is determined that the first capability information carries the maximum number of processing the reference model or the maximum number of processing the reference model.

[0087] In some embodiments, the first capability information includes first storage capability information, and the method further includes any one of the following:

[0088] According to the first storage capability size, it is determined that the first capability information carries the first storage capability size.

[0089] According to the first storage capability size and the first target storage capability information, it is determined that the first capability information carries the ratio relative to the first target storage capability information.

[0090] According to the first storage capability size, and the second model information of the reference model or the second model information of the reference model, it is determined that the first capability information carries the maximum number of storing the reference model or the maximum number of storing the reference model.

[0091] The second model information of the reference model or the second model information of the reference model includes the model size and / or the parameter quantity. The model size is used to represent the storage space occupied by the model, and the parameter quantity is associated with the storage space. The larger the parameter quantity is, the larger the occupied storage space is, and vice versa, the smaller the occupied storage space is.

[0092] In the embodiments of the present disclosure, if the first capability information includes the first storage capability information, when the terminal reports the absolute value of the first capability of the terminal to the network device, the reported absolute value of the first capability of the terminal can be the first storage capability size of the terminal. The first storage capability size of the terminal is taken as the first storage capability information, and it is determined that the first storage capability size is carried in the first capability information. That is, according to the first storage capability size of the terminal, it is determined that the first storage capability size is carried in the first capability information.

[0093] If the first capability information includes the first storage capability information, when the terminal reports the relative value of the first capability of the terminal to the network device, the reported relative value of the first capability of the terminal can be the ratio of the first storage capability size of the terminal to the first target storage capability information, or the first storage capability size of the terminal is the maximum value of the simultaneously stored reference model or the maximum value of the simultaneously stored reference model. The ratio of the first storage capability size of the terminal to the first target storage capability information is taken as the first storage capability information, and it is determined that the ratio of the first storage capability size to the first target storage capability information is carried in the first capability information. Alternatively, the maximum value of the reference model or the maximum value of the reference model (herein referred to as the maximum value of the reference model or the reference model) stored by the terminal is taken as the first storage capability information, and it is determined that the maximum value of the reference model or the maximum value of the reference model is carried in the first capability information. That is:

[0094] According to the first storage capability size of the terminal and the first target storage capability information, it is determined that the ratio of the first storage capability size to the first target storage capability information is carried in the first capability information. Alternatively, according to the first calculation capability storage and the second model information of the reference model or the second model information of the reference model, it is determined that the maximum value of the reference model or the maximum value of the reference model is stored in the first capability information.

[0095] In the embodiments of the present disclosure, the terminal can report the first capability information of the terminal, such as the absolute value of the first capability and / or the relative value of the first capability, so that the receiving end can measure or quantify the AI or ML related capability on the terminal side based on the first capability information, and further implement the management of the AI or ML model or AI or ML function on the terminal side.

[0096] In some embodiments, the third capability information includes a relative value of a third capability occupied or required by the AI or ML model of the terminal.

[0097] In the embodiments of the present disclosure, the third capability information includes: an absolute value of the third capability occupied or required by the AI or ML model of the terminal, and / or a relative value of the third capability occupied or required by the AI or ML model of the terminal. The absolute value of the third capability can be a specific numerical value for measuring the size of the third capability (such as the size of the third computing capability and / or the size of the third storage capability) of the terminal, for example, the size of the third computing capability is measured by using complexity (which can be measured by FLOPs) and / or the size of the third storage capability is measured by using model size and / or parameter quantity. The related capability occupied or required by the AI or ML model of the terminal reported by the terminal is one or more of the following: the computing complexity (i.e., complexity) of the AI or ML model of the terminal, the parameter quantity (i.e., parameter quantity) of the AI or ML model of the terminal, and the storage size (i.e., model size or storage space occupied by the model) of the AI or ML model of the terminal.

[0098] The relative value of the third capability can be a relative value for measuring the size of the third capability (the size of the third computing capability and / or the size of the third storage capability) occupied or required by the AI or ML model of the terminal.

[0099] In some embodiments, the third capability information includes third computing capability information and / or third storage capability information.

[0100] The third computing capability information includes any one of the following: first model information of the AI or ML model of the terminal, a ratio relative to third target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and the third storage capability information includes any one of the following: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

[0101] The first model information of the AI or ML model of the terminal is used to indicate the third computing capability, and the second model information of the AI or ML model of the terminal is used to indicate the third storage capability.

[0102] In the embodiments of the present disclosure, the third capability information includes any one of the following: first model information of the AI or ML model of the terminal, a ratio relative to third target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and / or, the third capability information includes any one of the following: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

[0103] The first model information of the AI or ML model of the terminal is used to indicate the third computing capability size, and the second model information of the AI or ML model of the terminal is used to indicate the third storage capability size. The first model information of the AI or ML model of the terminal includes a parameter quantity and / or complexity of the model, and the second model information of the AI or ML model of the terminal includes a parameter quantity and / or model size of the model.

[0104] In the embodiments of the present disclosure, the method for the terminal to report AI or ML model of the terminal to the network device includes:

[0105] The terminal reports an absolute value of complexity or storage space of the AI or ML model, such as complexity, parameter quantity, and storage size (in combination with mode 2-1).

[0106] The AI or ML related capability occupied or required by the AI or ML model reported by the terminal is a relative value, including:

[0107] The terminal reports a relative value of a relative complexity unit (here, the complexity unit can be used as the third target computing capability information) or a storage unit (here, the complexity unit can be used as the third target storage capability information), such as defining N floating point operations FLOPs as a complexity unit, and the complexity of the AI or ML model reported by the terminal is a multiple of the relative N FLOPs (in combination with mode 2-2).

[0108] The AI or ML related capability occupied by the reference model is used as a reference, and the AI or ML related capability occupied or required by the AI or ML model reported by the terminal is a multiple of the AI or ML capability occupied by the reference model (in combination with mode 2-3).

[0109] Mode 2-1: The terminal reports the AI or ML related capability occupied or required by the AI or ML model of the terminal as one or more of the computing complexity of the model, the parameter quantity of the model, and the storage size of the model.

[0110] Mode 2-2: The third target computing capability information can be a predefined or configured computing capability unit (or a computing unit, which can be defined based on complexity and / or parameter quantity), or a predefined or configured any computing capability (which can be defined based on model size and / or parameter quantity). The third target storage capability information can be a predefined or configured storage capability unit (or a storage unit, which can be defined based on model size and / or parameter quantity), or a predefined or configured any storage capability size. For example, the third target computing capability information includes a first complexity and / or a first parameter quantity, and the third target storage information includes a first model size and / or a second parameter quantity.

[0111] One or more of the following is taken as a unit: O operations (such as FLOPs, multiply-accumulate operation times (English: Multiply-Accumulate Operations, for short: MACs)), P bytes, and Q parameter quantities, and the terminal reports the related capabilities occupied by the AI or ML model to the network device.

[0112] For example, define 1 computing unit as 1 FLOP, if the complexity of a certain AI or ML model is K FLOPs, then the complexity or computing resources occupied by the model reported by the UE is or

[0113] For example, define 1 storage unit as P bytes, if the size of a certain AI or ML model is M bytes, then the storage space or storage resources occupied by the model reported by the UE is or

[0114] For example, define 1 computing unit as Q parameter quantities (i.e., the parameter quantity of the model), if the parameter quantity of a certain AI or ML model is N, then the complexity or parameter quantity of the model reported by the UE is or

[0115] Method 2-3: Taking the AI or ML related capabilities occupied or required by the reference model as a reference, the AI or ML related capabilities (such as computing capabilities, storage capabilities) occupied or required by the AI or ML model reported by the terminal is taken as a reference.

[0116] For example, if the FLOPs of the reference model is W, and the complexity of a certain AI or ML model is K FLOPs, then the complexity or computing resources occupied by the model reported by the UE is or

[0117] It should be noted that when the terminal reports the related capabilities occupied or required by the AI or ML function of the terminal to the network device, a similar method as described above can be used, and details are not repeated here.

[0118] In some embodiments, the third capability information includes third computing capability information; the method further includes any one of the following:

[0119] According to the first model information of the AI or ML model of the terminal, determining the first model information of the AI or ML model of the terminal in the third capability information;

[0120] determine, according to the first model information of the AI or ML model of the terminal and the third target computing capability information, a ratio of the third capability information to the third target computing capability information;

[0121] determine, according to the first model information of the AI or ML model of the terminal and the first model information of the reference model, a ratio of the third capability information to the first model information of the reference model.

[0122] In the embodiments of the present disclosure, if the third capability information includes the third computing capability information, and the terminal reports the absolute value of the third capability of the terminal to the network device, the absolute value of the third capability of the terminal reported here can be the first model information of the AI or ML model of the terminal. The first model information of the AI or ML model of the terminal is taken as the third computing capability information, and the first model information of the AI or ML model of the terminal is determined to be carried in the third capability information. That is, according to the first model information of the AI or ML model of the terminal, it is determined that the first model information of the AI or ML model of the terminal is carried in the third capability information.

[0123] If the third capability information includes the third computing capability information, and the terminal reports the relative value of the third capability of the terminal to the network device, the relative value of the third capability of the terminal reported here can be a ratio of the first model information of the AI or ML model of the terminal to the third target computing capability information, or a ratio of the first model information of the AI or ML model of the terminal to the first model information of the reference model (herein referred to as a ratio of the first model information of the reference model). The ratio of the first model information of the reference model or the ratio of the first model information of the reference model is taken as the third computing capability information, and the ratio of the third target computing capability information is determined to be carried in the third capability information. Alternatively, the ratio of the first model information of the reference model or the ratio of the first model information of the reference model is taken as the third computing capability information, and the ratio of the first model information of the reference model or the ratio of the first model information of the reference model is determined to be carried in the third capability information. That is:

[0124] determine, according to the first model information of the AI or ML model of the terminal and the third target computing capability information, a ratio of the third capability information to the third target computing capability information;

[0125] determine, according to the first model information of the AI or ML model of the terminal and the first model information of the reference model (herein referred to as the first model information of the reference model), that the ratio of the first model information of the reference model is carried in the third capability information.

[0126] In some embodiments, the third capability information comprises third storage capability information; and the method further comprises any one of the following:

[0127] determining the second model information of the AI or ML model of the terminal in the third capability information according to the second model information of the AI or ML model of the terminal;

[0128] determining the ratio of the third target storage capability information in the third capability information according to the second model information of the AI or ML model of the terminal and the third target storage capability information;

[0129] determining the ratio of the second model information of the reference model or the ratio of the second model information of the reference model in the third capability information according to the second model information of the AI or ML model of the terminal and the second model information of the reference model or the second model information of the reference model.

[0130] In the embodiments of the present disclosure, if the third capability information comprises the third storage capability information, when the terminal reports the absolute value of the third capability of the terminal to the network device, the absolute value of the third capability of the terminal reported here can be the second model information of the AI or ML model of the terminal. The second model information of the AI or ML model of the terminal is taken as the third storage capability information, and the second model information of the AI or ML model of the terminal carried in the third capability information is determined. That is, the second model information of the AI or ML model of the terminal carried in the third capability information is determined according to the second model information of the AI or ML model of the terminal.

[0131] If the third capability information comprises the third storage capability information, when the terminal reports the relative value of the third capability of the terminal to the network device, the relative value of the third capability of the terminal reported here can be the ratio of the second model information of the AI or ML model of the terminal to the third target storage capability information, or the ratio of the second model information of the AI or ML model of the terminal to the second model information of the reference model or the second model information of the reference model (herein referred to as the ratio of the second model information of the reference model or the second model information of the reference model). The ratio of the second model information of the reference model or the second model information of the reference model is taken as the third storage capability information, and the ratio of the third target storage capability information carried in the third capability information is determined. Alternatively, the ratio of the second model information of the reference model or the second model information of the reference model is taken as the third storage capability information, and the ratio of the second model information of the reference model or the second model information of the reference model carried in the third capability information is determined. That is:

[0132] According to the second model information of the AI or ML model of the terminal and the third target storage capability information, a ratio carried in the third capability information relative to the third target storage capability information is determined.

[0133] According to the second model information of the AI or ML model of the terminal and the second model information of the reference model or the second model information of the reference model (herein referred to as the second model information of the reference model), a ratio carried in the third capability information relative to the second model information of the reference model or the second model information of the reference model is determined.

[0134] In the embodiments of the present disclosure, the terminal can report the third capability information of the terminal, such as the absolute value of the third capability and / or the relative value of the third capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML model on the terminal side based on the third capability information, and further implement the management of the AI or ML model on the terminal side.

[0135] In some embodiments, the fourth capability information includes a relative value of the fourth capability occupied or required by the AI or ML function of the terminal.

[0136] The AI or ML function of the terminal includes at least one AI or ML model. Herein, the AI or ML function can be deployed on the terminal side, or can be an AI or ML function used by the terminal in conjunction with other terminals. The deployment manner of the AI or ML function is not limited herein.

[0137] In the embodiments of the present disclosure, the fourth capability information includes an absolute value of the fourth capability occupied or required by the AI or ML function of the terminal, and / or a relative value of the fourth capability occupied or required by the AI or ML function of the terminal. The absolute value of the fourth capability can be a specific numerical value for measuring the size of the fourth capability (such as the size of the fourth computing capability and / or the size of the fourth storage capability) of the terminal, such as using complexity (which can use FLOPs to measure complexity) and / or parameter quantity to measure the size of the fourth computing capability, and using model size and / or parameter quantity to measure the size of the fourth storage capability. The related capability reported by the terminal and occupied or required by the AI or ML function of the terminal is one or more of the computing complexity (i.e., complexity) of the AI or ML function, the parameter quantity (i.e., parameter quantity) of the AI or ML function, and the storage size (i.e., model size or storage space occupied by the model).

[0138] The relative value of the fourth capability can be a relative value for measuring the size of the fourth capability (the size of the fourth computing capability and / or the size of the fourth storage capability) occupied or required by the AI or ML function of the terminal.

[0139] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0140] The fourth capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model; or,

[0141] The fourth capability information is a sum of capability information occupied or required by activated AI or ML models in the AI or ML function of the terminal; or,

[0142] The fourth capability information is a sum of capability information of AI or ML models running or performing inference operations in the AI or ML function of the terminal.

[0143] In the embodiments of the present disclosure, the fourth capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and / or, the fourth capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

[0144] The first model information of at least one AI or ML model in the AI or ML function of the terminal is used to indicate the fourth computing capability, and the second model information of at least one AI or ML model in the AI or ML function of the terminal is used to indicate the fourth storage capability.

[0145] The first model information of the AI or ML model in the AI or ML function of the terminal is used to indicate the fourth computing capability, and the second model information of the AI or ML model (which is at least one) in the AI or ML function of the terminal is used to indicate the fourth storage capability. The first model information of the AI or ML model in the AI or ML function of the terminal includes a parameter amount and / or complexity of the model (here, a parameter amount and / or complexity of the AI or ML function), and the second model information of the AI or ML model in the AI or ML function of the terminal includes a parameter amount and / or model size of the model (here, a parameter amount and / or model size of the AI or ML function).

[0146] In the embodiments of the present disclosure, the method for the terminal to report the AI or ML function occupation or required related capability of the terminal to the network device is similar to the method for the terminal to report the AI or ML model occupation or required AI or ML related capability of the AI or ML model in the AI or ML function of the terminal to the network device, that is, the method for the terminal to report the AI or ML function occupation or required related capability of the terminal to the network device comprises:

[0147] The terminal reports the absolute value of the complexity (that is, the complexity of the AI or ML function) or storage space of the AI or ML model in the AI or ML function, such as the complexity, the number of parameters, and the storage size (similar to mode 2-1, which will not be repeated here);

[0148] The AI or ML related capability occupied or required by the AI or ML function reported by the terminal is a relative value, which comprises:

[0149] The terminal reports the relative value of the relative complexity unit (here, the complexity unit can be used as the fourth target computing capability information) and the storage unit (here, the complexity unit can be used as the fourth target storage capability information), such as defining N floating point operations FLOPs as a complexity unit, and the complexity of the AI or ML function reported by the terminal is a multiple of the relative N FLOPs (similar to mode 2-2, which will not be repeated here);

[0150] The AI or ML related capability occupied by the benchmark model or the reference model is taken as a benchmark, and the AI or ML related capability occupied or required by the AI or ML function reported by the terminal is a multiple of the AI or ML capability occupied by the benchmark model or the reference model (similar to mode 2-3, which will not be repeated here).

[0151] In some embodiments, the fourth capability information comprises fourth computing capability information; and the method further comprises any one of the following:

[0152] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, the first model information of at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information is determined;

[0153] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target computing capability information, the ratio of the fourth capability information to the fourth target computing capability information is determined;

[0154] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, and the first model information of the benchmark model or the first model information of the reference model, the ratio of the fourth capability information to the first model information of the benchmark model or the ratio of the fourth capability information to the first model information of the reference model is determined.

[0155] In the embodiments of the present disclosure, if the fourth capability information includes the fourth computing capability information, when the terminal reports the absolute value of the fourth capability of the terminal to the network device, the reported absolute value of the fourth capability of the terminal can be the first model information of the AI or ML model in the AI or ML function of the terminal (hereinafter referred to as the first model information of the AI or ML function of the terminal, which will not be described below). The first model information of the AI or ML function of the terminal is taken as the fourth computing capability information, and it is determined that the fourth capability information carries the first model information of the AI or ML function of the terminal. That is, according to the first model information of the AI or ML function of the terminal, it is determined that the fourth capability information carries the first model information of the AI or ML function of the terminal.

[0156] If the fourth capability information includes the fourth computing capability information, when the terminal reports the relative value of the fourth capability of the terminal to the network device, the reported relative value of the fourth capability of the terminal can be the ratio of the first model information of the AI or ML function of the terminal to the fourth target computing capability information, or the ratio of the first model information of the AI or ML function of the terminal to the first model information of the reference model or the first model information of the reference model (hereinafter referred to as the ratio of the first model information of the reference model or the first model information of the reference model). The ratio of the first model information of the reference model or the first model information of the reference model is taken as the fourth computing capability information, and it is determined that the fourth capability information carries the ratio of the fourth target computing capability information. Alternatively, the ratio of the first model information of the reference model or the first model information of the reference model is taken as the fourth computing capability information, and it is determined that the fourth capability information carries the ratio of the first model information of the reference model or the first model information of the reference model. That is:

[0157] According to the first model information of the AI or ML function of the terminal and the fourth target computing capability information, it is determined that the fourth capability information carries the ratio of the fourth target computing capability information.

[0158] According to the first model information of the AI or ML function of the terminal, and the first model information of the reference model or the first model information of the reference model (hereinafter referred to as the first model information of the reference model or the first model information of the reference model), it is determined that the fourth capability information carries the ratio of the first model information of the reference model or the first model information of the reference model.

[0159] In some embodiments, the fourth capability information includes fourth storage capability information; the method further includes any one of the following:

[0160] determine the second model information of at least one AI or ML model in the AI or ML function of the terminal according to the second model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability information;

[0161] determine the ratio of the fourth capability information to the fourth target storage capability information according to the second model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability information;

[0162] determine the ratio of the fourth capability information to the second model information of the reference model according to the second model information of at least one AI or ML model in the AI or ML function of the terminal and the second model information of the reference model.

[0163] In the embodiments of the present disclosure, if the fourth capability information includes the fourth storage capability information, when the terminal reports the absolute value of the fourth capability of the terminal to the network device, the absolute value of the fourth capability of the terminal reported here can be the second model information of the AI or ML model in the AI or ML function of the terminal (herein referred to as the second model information of the AI or ML function of the terminal, which will not be described below). The second model information of the AI or ML function of the terminal is taken as the fourth storage capability information, and the second model information of the AI or ML model in the terminal is determined to be carried in the fourth capability information. That is, the second model information of the AI or ML model in the terminal is determined to be carried in the fourth capability information according to the second model information of the AI or ML function of the terminal.

[0164] If the fourth capability information includes the fourth storage capability information, when the terminal reports the relative value of the fourth capability of the terminal to the network device, the relative value of the fourth capability of the terminal reported here can be the ratio of the second model information of the AI or ML function of the terminal to the fourth target storage capability information, or the ratio of the second model information of the AI or ML function of the terminal to the second model information of the reference model (herein referred to as the ratio of the second model information of the reference model). The ratio of the second model information of the reference model is taken as the fourth storage capability information, and the ratio of the fourth target storage capability information is determined to be carried in the fourth capability information. Alternatively, the ratio of the second model information of the reference model is taken as the fourth storage capability information, and the ratio of the second model information of the reference model is determined to be carried in the fourth capability information. That is:

[0165] According to the second model information of the AI or ML function of the terminal and the fourth target storage capability information, a ratio of the fourth capability information to the fourth target storage capability information is determined.

[0166] According to the second model information of the AI or ML function of the terminal and the second model information of the reference model or the second model information of the reference model (herein referred to as the second model information of the reference model or the second model information of the reference model), a ratio of the fourth capability information to the second model information of the reference model or the second model information of the reference model is determined.

[0167] In the embodiments of the present disclosure, the terminal can report the fourth capability information of the terminal, such as the absolute value of the fourth capability and / or the relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function of the terminal based on the fourth capability information, and further implement the management of the AI or ML function of the terminal.

[0168] In some embodiments, the fourth capability information is the capability information occupied or required by a third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is an AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0169] In the embodiments of the present disclosure, the occupation mode of the AI or ML model or the AI or ML function of the terminal to the AI or ML related capability is the occupation mode one or the occupation mode two; wherein the occupation mode one is that the AI or ML model or the AI or ML function occupies the AI or ML related capability when the AI or ML model or the AI or ML function is activated; and the occupation mode two is that the AI or ML model or the AI or ML function occupies the AI or ML related capability when the AI or ML model or the AI or ML function is activated and when the AI or ML model or the AI or ML function runs.

[0170] There are potentially two ways for the AI or ML model or the AI or ML function to occupy or require the terminal side AI or ML related capability. The first way is that the AI or ML model or the AI or ML function occupies the AI or ML related capability when the AI or ML model or the AI or ML function is activated (occupation mode one), and the second way is that the AI or ML model or the AI or ML function does not occupy the AI or ML related processing capability after being activated, and only occupies the AI or ML related processing capability when running, i.e., reasoning (occupation mode two).

[0171] Among them, one terminal can deploy at least one AI or ML function, at least one AI or ML model is included in one AI or ML function, and multiple AI or ML models can be included in one AI or ML function. For example, the AI or ML related capability (here, the fourth capability) occupied by the AI or ML function reported by the terminal can be the AI or ML related capability of the AI or ML model with the largest AI or ML related capability in the AI or ML function (here, the AI or ML model with the largest AI or ML related capability is the third AI or ML model), or the AI or ML related capability of the activated AI or ML model in the AI or ML function (for the first occupation mode), or the AI or ML related capability of the AI or ML model running or reasoning in the AI or ML function (for the second occupation mode).

[0172] In some embodiments, the method further comprises:

[0173] According to the time range or symbol of the third capability information occupied or required by each AI or ML model of the terminal when running or reasoning, or according to the time range or symbol of the fourth capability information occupied or required by the AI or ML function of the terminal when running or reasoning, the following operations are performed:

[0174] For any moment or symbol, if it is determined that the third capability information occupied or required by the AI or ML model running or performing reasoning operation of the terminal or the fourth capability information occupied or required by the AI or ML function running or performing reasoning operation of the terminal is greater than the first capability information of the terminal, the first AI or ML model and the second AI or ML model are determined from the activated AI or ML model or AI or ML function of the terminal.

[0175] Among them, the first AI or ML model is an AI or ML model running or performing reasoning operation, and the second AI or ML model is an AI or ML model not running or performing reasoning operation.

[0176] In the embodiments of the present disclosure, for the first occupation mode, it is necessary to ensure that the total AI or ML related capability occupied or required by the currently activated AI or ML model or AI or ML function of the terminal does not exceed the AI or ML related capability of the terminal. There are as follows:

[0177] If activating a certain AI or ML model or AI or ML function will exceed the capability of the UE (the first capability of the UE, which will not be described below), the AI or ML model or AI or ML function is not allowed to be activated (combined with mode 3-1);

[0178] If the UE's capability is exceeded after activating a certain AI or ML model or AI or ML function, the AI or ML model or AI or ML function activated first in time is deactivated until the AI or ML related capability occupied by the remaining activated AI or ML model or AI or ML function is less than or equal to the AI or ML related capability of the terminal (in combination with Mode 3-2);

[0179] If the UE's capability is exceeded after activating a certain AI or ML model or AI or ML function, the AI or ML model or AI or ML function with lower priority is deactivated (in combination with Mode 3-3).

[0180] According to the AI or ML related capability (i.e., the first capability or the first capability information) possessed by the terminal and the AI or ML related capability occupied or required by each AI or ML model or AI or ML function (i.e., the AI or ML related capability occupied or required by each AI or ML model (herein referred to as the third capability or the third capability information), or the AI or ML related capability occupied or required by each AI or ML function (herein referred to as the fourth capability or the fourth capability information)), it is ensured that the AI or ML model currently activated or running or performing inference on the terminal side does not exceed the capability of the UE. This is achieved by at least three methods as follows:

[0181] Mode 3-1: The UE does not expect the network device to activate a certain AI or ML model or AI or ML function, and the sum of the at least one AI related capability occupied by these activated AI or ML models or AI or ML functions is greater than the AI related capability of the UE. That is, if at least one of the AI or ML related capabilities on the terminal side does not support the activation of a certain AI or ML model or AI or ML function, the AI or ML model or AI or ML function cannot be activated.

[0182] Mode 3-2: If the network device activates a certain AI or ML model or AI or ML function, the at least one AI or ML related capability occupied by these activated AI or ML models or AI or ML functions is greater than the AI or ML related capability of the UE, then the AI or ML model or AI or ML function activated first in time is deactivated until the AI or ML related capability occupied or required by the remaining activated AI or ML model or AI or ML function is less than or equal to the AI or ML related capability of the terminal.

[0183] Manner 3-3: If the network device activates a certain AI or ML model or AI or ML function, and the activated AI or ML model or AI or ML function occupies at least one AI-related capability in total, which is greater than the AI-related capability of the UE, then the S AI or ML models or AI or ML functions with lower priority are deactivated, S being the minimum integer value that satisfies that the AI or ML-related capability occupied or required by the remaining activated AI or ML function or AI or ML model after deactivation is less than or equal to the relevant AI or ML capability of the terminal. The priority of the AI or ML model or AI or ML function (here, the priority of the AI or ML model, or the priority of the AI or ML function) is indicated by the network device or reported by the terminal.

[0184] For the second occupation manner, it is necessary to ensure that the total AI-related capability occupied by the currently running AI or ML model or AI or ML function does not exceed the AI or ML-related capability of the terminal. It is necessary to define the time range of the AI or ML-related capability occupied by each AI or ML model or AI or ML function during inference (here, it refers to defining the AI or ML-related capability occupied by each AI or ML model during inference, or defining the AI or ML-related capability occupied by each AI or ML function during inference). The start symbol and end symbol of each AI or ML model or AI or ML function (i.e., during inference operation or execution of inference operation). For any moment or symbol, the AI or ML model or AI or ML function running or performing inference occupies or requires AI or ML-related capability that cannot exceed the AI or ML-related capability of the terminal. If a certain AI or ML model or AI or ML function is running, and the sum of the at least one AI-related capability occupied by all running AI or ML models or AI or ML functions at a certain symbol is greater than the AI or ML-related capability of the UE, then according to the priority of the AI or ML model or AI or ML function, it is determined which AI or ML model or AI or ML function does not perform inference (in combination with manner 3-4).

[0185] Manner 3-4: defining a time range of occupying AI or ML related capability when each AI or ML model or AI or ML function is reasoning, such as defining a start symbol and a stop symbol of running each AI or ML model or AI or ML function (i.e., when performing reasoning operation). For any time instant or symbol, the AI or ML related capability occupied by running AI or ML model or AI or ML function cannot exceed the AI or ML related capability of the terminal. If after running a certain AI or ML model or AI or ML function, the sum of at least one AI related capability occupied by all running AI or ML models or AI or ML functions at a symbol is greater than the AI related capability of the UE, then according to the priority of the AI or ML model or AI or ML function, it is determined that S AI or ML models or AI or ML functions with lower priority do not perform reasoning or do not report or do not update the results obtained according to the S AI or ML models or AI or ML functions. S is the minimum integer value that satisfies that after not running the S AI or ML models or AI or ML functions, the sum of AI related capability when running the remaining AI or ML functions or AI or ML models is less than or equal to the AI related capability of the terminal. Wherein, the priority of the AI or ML model or AI or ML function is indicated by the network device or reported by the terminal.

[0186] In the embodiments of the present disclosure, the terminal can guarantee that the total AI or ML related capability occupied by the currently running AI or ML model or AI or ML function does not exceed the AI or ML related capability of the terminal by managing the running or reasoning of each AI or ML model of the terminal, or by managing the running or reasoning of each AI or ML function of the terminal.

[0187] In some embodiments, the second capability information includes: a number of occupied CSI processing units.

[0188] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information and the calculation latency requirement information of the CSI report have a mapping relationship.

[0189] In some embodiments, the calculation latency requirement information of the CSI report is associated with at least one of the following:

[0190] a complexity of an AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0191] a parameter amount of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0192] first capability information;

[0193] third capability information or fourth capability information;

[0194] content of the CSI report;

[0195] an output of an AI or ML model, or an output of an AI or ML function.

[0196] In the embodiments of the present disclosure, after the AI or ML based CSI processing method, there can be an impact on the CSI processing unit and the CSI processing time. The terminal can report related information of the CSI report, including at least one of the following: second capability information of the CSI report, calculation delay requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the terminal side based on AI or ML based CSI processing based on the related information of the CSI report, and further realize the management of the terminal side based on AI or ML based CSI processing.

[0197] In the embodiments of the present disclosure, for the AI or ML based CSI reporting, the number of CSI processing units occupied and the CSI calculation delay can be different from the traditional (non-AI or ML) CSI reporting. For the CSI report content, the number of processing units (English: CSI processing unit, abbreviated as CPU) of the AI or ML based CSI reporting (i.e. the number of CPUs of the CSI reporting, here referring to the number of CPUs occupied by the CSI report) and the CSI calculation delay (here referring to the calculation delay requirement information of the CSI report; wherein the number of processing units and the calculation delay based on the non-AI or ML based CSI reporting method can adopt related technologies) are defined, including:

[0198] When the CSI report content is obtained based on the traditional (non-AI or ML) method, the number of CSI processing units occupied is N1, and when the CSI report content is obtained based on AI or ML, the number of CSI processing units occupied is N2;

[0199] For the AI or ML based CSI report, the CSI calculation time is related to the complexity of the AI or ML model adopted by the terminal, the AI or ML processing capability of the terminal, the content of the CSI report, the output of the AI or ML model, or the output of the AI or ML function (here referring to the calculation delay requirement information of the CSI report, which is related to at least one of the complexity of the AI or ML model in the AI or ML model or AI or ML function, the parameter amount of the AI or ML model in the AI or ML model or AI or ML function, the first capability information, the third capability information or the fourth capability information, the content of the CSI report, the output of the AI or ML model, or the output of the AI or ML function):

[0200] The UE reports the capability level of AI or ML based CSI reporting. The computing time of CSI (i.e. the computing latency requirement information of the CSI reporting) is determined according to the capability level of AI or ML based CSI reporting reported by the UE (i.e. there is a mapping relationship between the capability level information and the computing latency requirement information of the CSI reporting).

[0201] The CSI reporting includes traditional (non-AI or ML) CSI reporting and AI or ML based CSI reporting. The number of processing units occupied and the CSI computing time are different for the two kinds of CSI reporting.

[0202] 1) When the CSI reporting content is obtained based on the traditional (non-AI or ML) method, the number of CSI processing units occupied is N1, and when the CSI reporting content is obtained based on AI or ML, the number of CSI processing units occupied is N2.

[0203] 2) The CSI computing time is different when the CSI reporting content is obtained based on the traditional (non-AI or ML) method and when the CSI reporting content is obtained based on AI or ML. For the AI or ML based CSI computing time, different levels of time are defined, and the AI or ML based CSI computing time is determined according to the capability level reported by the UE.

[0204] For example, embodiment one (the terminal reports the AI or ML related capability, the AI or ML method occupied by each AI or ML model deployed is reported as an absolute value)

[0205] The terminal reports its AI or ML related capability (i.e. the terminal reports its first capability or first capability information, which will not be described below)

[0206] Computing capability (herein referred to as first computing capability size or first computing capability information): 7*10^12 FLOPS;

[0207] Storage capability (herein referred to as first storage capability size or first storage capability information): 8G.

[0208] The terminal reports the AI or ML related capability occupied by the AI or ML model deployed (i.e. the terminal reports its third capability or third capability information) as:

[0209] Model 1: complexity: 8M FLOPs, size (i.e. model size or storage size of occupied storage space, which will not be described below): 4M;

[0210] Model 2: complexity: 40M FLOPs, size: 30M.

[0211] Exemplarily, embodiment two (terminal reports AI or ML related capability it has, each AI or ML function deployed occupies or AI or ML way required is reported as relative value relative to defined unit)

[0212] Define 10^12 FLOPS as 1 computing capability unit, 2G as 1 storage capability unit. Assuming that the computing capability of the terminal (herein referred to as the first computing capability or the first computing capability information) is 7*10^12 FLOPS, and the storage capability (herein referred to as the first storage capability or the first storage capability size) is 8G, the terminal reports its AI or ML related capability as:

[0213] Computing capability (herein referred to as the relative value of the first capability or the first computing capability information): 7;

[0214] Storage capability (herein referred to as the relative value of the first capability or the first storage capability information): 4.

[0215] Assuming that 1 AI or ML function deployed on the terminal side includes the following 2 models

[0216] Model 1: complexity: 8M FLOPs, size: 4M;

[0217] Model 2: complexity: 40M FLOPs, size: 30M.

[0218] Define 1M FLOPs occupies 1 computing unit, and 5M occupies 1 storage unit. When the terminal reports the AI or ML capability occupied by the AI or ML function, the computing capability and the storage capability of the model occupying the largest AI or ML capability in the AI or ML function are reported, that is, the AI related capability (herein referred to as the fourth capability or the fourth capability information) occupied by the AI or ML function reported by the terminal is:

[0219] Computing unit (herein referred to as the relative value of the fourth capability or the fourth computing capability information): 40; storage unit (herein referred to as the relative value of the fourth capability or the fourth computing capability information): 6.

[0220] Exemplarily, embodiment three (terminal reports AI or ML related capability it has, each AI or ML model occupies or AI or ML way required is reported as relative value relative to reference model)

[0221] Define the complexity of the reference model (herein referred to as the first model information of the reference model) as 10^6 FLOPS, and the model size (herein referred to as the second model information of the reference model) as 50M.

[0222] Assuming that the computing capability of the terminal is 7*10^12 FLOPS, and the storage capability is 8G, the terminal reports its AI or ML related capability as:

[0223] Computing capability: 7*10^6;

[0224] Storage capability: 160.

[0225] Suppose the complexity and size of 2 AI or ML models deployed at the terminal side are:

[0226] Model 1: complexity: 8M FLOPs, size: 40M;

[0227] Model 2: complexity: 40M FLOPs, size: 300M.

[0228] The terminal reports the complexity and size of the above 2 models relative to the reference model, and the reporting content of the terminal (here, the third capability or third capability information) is:

[0229] Model 1: complexity (here, the relative value of the third capability or the third computing capability information, which will not be described below): 8, size (here, the relative value of the third capability or the third storage capability information): 0.8;

[0230] Model 2: complexity: 40, size: 6.

[0231] Exemplary, embodiment four (corresponding to mode 3-1)

[0232] Suppose the AI or ML related capability at the terminal side is:

[0233] Computing capability: 7;

[0234] Storage capability: 4.

[0235] The following 2 AI or ML functions at the terminal side are activated AI or ML functions, and the computing capability / unit (i.e., computing capability or computing unit) and storage capability / unit (i.e., storage capability or storage unit) required by the 2 AI or ML functions are:

[0236] AI or ML function 1: computing capability 3, storage capability 1;

[0237] AI or ML function 2: computing capability 2, storage capability 3.

[0238] Suppose the AI or ML function 3 occupies AI or ML capability: computing capability 1, storage capability 1.

[0239] If the network side activates AI or ML function 3, it will exceed the storage capability at the terminal side, and AI or ML function 3 is not activated.

[0240] Exemplary, embodiment five (corresponding to mode 3-2)

[0241] Suppose the AI or ML related capability at the terminal side is:

[0242] Computing capability: 7;

[0243] Storage capability: 4.

[0244] The following two AI or ML functions on the terminal side are activated AI or ML functions, and the computing capability / unit and storage capability / unit occupied by the two AI or ML functions are:

[0245] AI or ML function 1: computing capability 3, storage capability 1;

[0246] AI or ML function 2: computing capability 2, storage capability 3.

[0247] Suppose that the AI or ML capability occupied by AI or ML function 3 is: computing capability 1, storage capability 1.

[0248] Suppose that the network side activates AI or ML function 3, and since the simultaneous activation of the three AI or ML functions will exceed the storage capability on the terminal side, AI or ML function 1 is deactivated according to the rule (suppose that the activation time of AI or ML function 1 is earlier than that of AI or ML function 2), and AI or ML function 2 and AI or ML function 3 are in the activated state.

[0249] Exemplarily, embodiment six (corresponding to mode 3-3)

[0250] Suppose that the AI or ML related capability on the terminal side is:

[0251] Computing capability: 7;

[0252] Storage capability: 4.

[0253] The following two AI or ML functions on the terminal side are activated AI or ML functions, and the computing capability / unit and storage capability / unit occupied by the two AI or ML functions are:

[0254] AI or ML function 1: computing capability 3, storage capability 1;

[0255] AI or ML function 2: computing capability 2, storage capability 3.

[0256] Suppose that the AI or ML capability occupied by AI or ML function 3 is: computing capability 1, storage capability 1.

[0257] If the network side activates AI or ML function 3, and since the simultaneous activation of the three AI or ML functions will exceed the storage capability on the terminal side, AI or ML function 2 is deactivated according to the priority (suppose that the priority of AI or ML function 2 is lower than that of AI or ML function 1), and AI or ML function 1 and AI or ML function 3 are in the activated state.

[0258] Exemplarily, embodiment seven (corresponding to mode 3-4)

[0259] Assume the AI or ML related capability at the terminal side is:

[0260] Computing capability: 7.

[0261] Storage capability: 4.

[0262] The following 3 AI or ML functions at the terminal side are activated AI or ML functions, and the computing capability / unit and storage capability / unit occupied by the 3 AI or ML functions are:

[0263] AI or ML function 1: computing capability 3, storage capability 1;

[0264] AI or ML function 2: computing capability 2, storage capability 3;

[0265] AI or ML function 3: computing capability 1, storage capability 1.

[0266] The priority of the above AI or ML functions is AI or ML function 2 > AI or ML function 1 > AI or ML function 3.

[0267] Assume that the terminal determines according to the symbol occupied by each AI or ML function reasoning that running the 3 AI or ML functions for reasoning will exceed the terminal capability at a certain symbol, then the terminal does not run AI or ML function 3, and AI or ML function 1 and AI or ML function 2 normally run for reasoning.

[0268] Example Eight

[0269] It is defined that when the reporting content in the CSI report is cri-RI-PMI-CQI, the codebook configuration is typeII-Doppler-r18, and the CPU number occupied when obtaining the reporting content based on the traditional (non-AI or ML) method is 2K; the CPU number occupied when obtaining the reporting content based on AI or ML is K, where K is the number of reference signals associated with the CSI report.

[0270] The CSI calculation time Z is:

[0271] Time 1 represents that when the configuration is 1, the CSI calculation time when the UE reports the CSI reporting capability based on non-AI or ML is capability 1 is 22 symbols;

[0272] Time 2 represents that when the configuration is 1, the CSI calculation time when the UE reports the CSI reporting capability based on non-AI or ML is capability 2 is 40 symbols;

[0273] Time 3 represents that when the configuration is 1, the CSI calculation time when the UE reports the CSI reporting capability based on AI or ML is capability 1 is 30 symbols;

[0274] Time 4 represents that when the UE reports the AI or ML based CSI reporting capability as capability 2, the CSI calculation time is 45 symbols.

[0275] Therefore, the capability reporting method provided by the present disclosure can ensure that the AI or ML model or AI or ML function activated or running at any time on the terminal side does not exceed the terminal capability (i.e., the capability of the UE), and enables the network side to know the activation or running of the AI or ML model or AI or ML function on the terminal side. At the same time, for AI or ML based CSI reporting, the difference between the CSI processing unit occupied by AI or ML based CSI reporting, the CSI calculation delay and the traditional (non-AI or ML) CSI reporting is considered, so that the CSI processing unit and calculation delay occupied by the UE based on AI or ML based CSI reporting can be determined according to the reported capability.

[0276] It should be noted that the capability reporting method provided by the present disclosure with the terminal as the execution subject can refer to the implementation process of the embodiments shown in FIG. 1, embodiments 1 to 8, and the specific implementation process will not be described here.

[0277] Referring to FIG. 2, FIG. 2 is an interaction schematic diagram of the capability determination method provided by the embodiment of the present disclosure. The execution subject of the capability reporting method provided by the present embodiment is a network device (such as a base station, a core network, etc., which is not limited here), and the capability reporting method is described in detail below.

[0278] The capability determination method provided by the embodiment of the present disclosure includes the following steps:

[0279] Receiving AI or ML capability information of the terminal, wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0280] In the embodiment of the present disclosure, the network device can measure or quantify the AI or ML related capability on the terminal side by receiving at least one of the following reported by the terminal: AI or ML based first capability information, CSI report occupation second capability information, CSI report calculation delay requirement information, AI or ML model occupation or requirement third capability information of the terminal, or AI or ML function occupation or requirement fourth capability information of the terminal, thereby realizing the management of the terminal side based on AI or ML technology.

[0281] For example, in combination with FIG. 2:

[0282] Step 201: The network device receives AI or ML capability information sent by the terminal. The capability information includes at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, CSI report calculation latency requirement information, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0283] The AI or ML capability information can be carried based on signaling or a signal. For example, the terminal sends first signaling to the network device (herein referred to as the network side, which will not be described below), and the first signaling carries the AI or ML capability information.

[0284] Step 202: According to the received capability information, it is determined whether to activate or deactivate the target AI or ML model and / or the target AI or ML function.

[0285] It should be noted that the implementation of the network device receiving the AI or ML capability information sent by the terminal can refer to the embodiments of the above capability reporting method, which will not be described here.

[0286] In some embodiments, the first capability information of the terminal includes a relative value of the first capability information of the terminal.

[0287] In some embodiments, the first capability information is used to determine the first capability information, and the first capability information includes first calculation capability information and / or first storage capability information.

[0288] The first capability information includes any one of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum number of processing reference models or reference models; and / or, the first capability information includes any one of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum number of stored reference models or a maximum number of reference models.

[0289] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes first capability information, such as an absolute value of the first capability and / or a relative value of the first capability. The AI or ML related capability on the terminal side can be measured or quantified based on the first capability information, and the management of the AI or ML model or AI or ML function on the terminal side can be realized.

[0290] In some embodiments, the third capability information includes a relative value of the third capability information of the AI or ML model occupation or requirement of the terminal.

[0291] In some embodiments, the third capability information is used to determine the third capability information, and the third capability information includes third calculation capability information and / or third storage capability information.

[0292] The third capability information includes any one of the following: first model information of an AI or ML model of the terminal, a ratio to third target computing capability information, a ratio to first model information of a reference model, or a ratio to first model information of a reference model; and the third capability information includes any one of the following: second model information of an AI or ML model of the terminal, a ratio to fourth target storage capability information, a ratio to second model information of a reference model, or a ratio to second model information of a reference model.

[0293] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes third capability information, such as an absolute value of the third capability and / or a relative value of the third capability, and the related capability occupied by the AI or ML model on the terminal side can be measured or quantified based on the third capability information, thereby realizing the management of the AI or ML model on the terminal side.

[0294] In some embodiments, the fourth capability information includes a relative value of the fourth capability information occupied or required by the AI or ML function of the terminal.

[0295] In some embodiments, the fourth capability information is used to determine fourth capability information, and the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0296] The fourth computing capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio to fourth target computing capability information, a ratio to first model information of a reference model, or a ratio to first model information of a reference model; and the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio to fourth target storage capability information, a ratio to second model information of a reference model, or a ratio to second model information of a reference model; or

[0297] The fourth capability information is the sum of capability information occupied or required by the activated AI or ML model in the AI or ML function of the terminal; or

[0298] The fourth capability information is the sum of capability information of the AI or ML model running or performing inference operation in the AI or ML function of the terminal.

[0299] In some embodiments, the fourth capability information is capability information occupied or required by a third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is an AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0300] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes fourth capability information of the terminal, such as an absolute value of the fourth capability and / or a relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function on the terminal side based on the fourth capability information, thereby realizing the management of the AI or ML function on the terminal side.

[0301] In some embodiments, the method further comprises:

[0302] determining, according to the capability information, whether to activate the first target AI or ML model and / or the first target AI or ML function; and / or,

[0303] determining, according to the capability information, whether to deactivate the second target AI or ML model and / or the second target AI or ML function.

[0304] The first target AI or ML model and the second target AI or ML model are AI or ML models in at least one AI or ML model of the terminal, and the first target AI or ML function and the second target AI or ML function are AI or ML functions in at least one AI or ML function of the terminal.

[0305] In the embodiments of the present disclosure, in combination with FIG. 2, the network device determines whether to activate or deactivate the target AI or ML model and / or the target AI or ML function based on the received capability information, thereby realizing the management of the AI or ML technology on the terminal side. The specific implementation manner can be referred to the above manners 3-1 to 3-4, embodiments four to seven, which will not be described here.

[0306] In some embodiments, the determining, according to the capability information, whether to activate the first target AI or ML model and / or the first target AI or ML function comprises:

[0307] In the scenario that the third capability information of the terminal is occupied when the AI or ML model of the terminal is activated or the fourth capability information of the terminal is occupied when the AI or ML function of the terminal is activated, according to the first capability information, if the sum of the third capability information occupied or required by the activated AI or ML model in the terminal and the fourth capability information occupied or required by the activated AI or ML function is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated, it is determined that the first target AI or ML model and / or the first target AI or ML function is not activated.

[0308] In some embodiments, the determining, according to the capability information, whether to deactivate the second target AI or ML model and / or the second target AI or ML function comprises:

[0309] In a scenario where the third capability information occupied or required by the AI or ML model of the terminal when activated or the fourth capability information occupied or required by the AI or ML function of the terminal when activated, according to the capability information, if the sum of the third capability information occupied or required by the activated AI or ML model in the terminal and the fourth capability information occupied or required by the activated AI or ML function is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated, it is determined to deactivate the second target AI or ML model and / or the second target AI or ML function.

[0310] wherein, after the second target AI or ML model and / or the second target AI or ML function is deactivated and the first target AI or ML model and / or the first target AI or ML function is activated, the sum of the third capability information occupied or required by the activated AI or ML model in the terminal and the fourth capability information occupied by the activated AI or ML function is less than or equal to the first capability information of the terminal.

[0311] In some embodiments, the second target AI or ML model and / or the second target AI or ML function is at least one.

[0312] wherein, the second target AI or ML model and / or the second target AI or ML function is the AI or ML model and / or AI or ML function activated first in time; or, the second target AI or ML model and / or the second target AI or ML function is the AI or ML model and / or AI or ML function with low activation priority.

[0313] In some embodiments, the second capability information includes: the number of CSI processing units occupied.

[0314] In some embodiments, the calculation latency requirement information of the CSI report is associated with at least one of:

[0315] the complexity of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0316] the number of parameters of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0317] the first capability information;

[0318] the third capability information or the fourth capability information;

[0319] the content of the CSI report;

[0320] An output of an AI or ML model, or an output of an AI or ML function.

[0321] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, the capability level information being in a mapping relationship with the calculation time delay requirement information of the CSI report.

[0322] In the embodiments of the present disclosure, after the AI or ML based CSI processing method, there may be an impact on the CSI processing unit and the CSI processing time. The terminal can report related information of the CSI report, including at least one of the following: second capability information occupied by the CSI report, calculation time delay requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the terminal side AI or ML based CSI processing based on the related information of the CSI report, and further realize the management of the terminal side AI or ML based CSI processing.

[0323] In a third aspect, the present disclosure provides a capability reporting apparatus, which is applied to a terminal and includes a memory, a transceiver, and a processor.

[0324] The memory is configured to store a computer program; the transceiver is configured to transceive data under the control of the processor; and the processor is configured to read the computer program in the memory and perform the following operations:

[0325] transmitting artificial intelligence (AI) or machine learning (ML) capability information, wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by channel state information (CSI) report, calculation time delay requirement information of the CSI report, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal.

[0326] In the embodiments of the present disclosure, the terminal can report at least one of the following: AI or ML based first capability information, second capability information occupied by the CSI report, calculation time delay requirement information of the CSI report, third capability information occupied or required by the AI or ML model of the terminal, or fourth capability information occupied or required by the AI or ML function of the terminal, which can measure or quantify the AI or ML related capability of the terminal side, and further enable the receiving end to know the AI or ML related capability of the terminal side, and further realize the management of the terminal side based on the AI or ML technology.

[0327] In some embodiments, the first capability information includes a relative value of the first capability of the terminal.

[0328] In some embodiments, the first capability information includes first calculation capability information and / or first storage capability information.

[0329] The first computing capability information includes any one of a first computing capability size, a ratio relative to first target computing capability information, a maximum value of a processing reference model, or a maximum value of a reference model; and the first storage capability information includes any one of a first storage capability size, a ratio relative to first target storage capability information, a maximum value of a storage reference model, or a maximum value of a reference model.

[0330] In some embodiments, the first capability information includes first computing capability information, and the processor is further configured to perform any one of:

[0331] determining, according to a first computing capability size of the terminal, a first computing capability size in the first capability information;

[0332] determining, according to the first computing capability size and first target computing capability information, a ratio relative to the first target computing capability information in the first capability information;

[0333] determining, according to the first computing capability size and first model information of a reference model or first model information of a reference model, a maximum value of a processing reference model or a maximum value of a reference model in the first capability information.

[0334] In some embodiments, the first capability information includes first storage capability information, and the processor is further configured to perform any one of:

[0335] determining, according to a first storage capability size of the terminal, a first storage capability size in the first capability information;

[0336] determining, according to the first storage capability size and first target storage capability information, a ratio relative to the first target storage capability information in the first capability information;

[0337] determining, according to the first storage capability size and second model information of a reference model or second model information of a reference model, a maximum value of a storage reference model or a maximum value of a reference model in the first capability information.

[0338] In the embodiments of the present disclosure, the terminal can report the first capability information of the terminal, such as an absolute value of the first capability and / or a relative value of the first capability, so that the receiving end can measure or quantify the AI or ML related capability on the terminal side based on the first capability information, and further implement management of the AI or ML model or AI or ML function on the terminal side.

[0339] In some embodiments, the third capability information includes a relative value of AI or ML model occupancy or required third capability of the terminal.

[0340] In some embodiments, the third capability information comprises third computing capability information and / or third storage capability information.

[0341] The third computing capability information comprises any one of the following: first model information of an AI or ML model of the terminal, a ratio to third target computing capability information, a ratio to first model information of a reference model, or a ratio to first model information of a reference model; and the third storage capability information comprises any one of the following: second model information of an AI or ML model of the terminal, a ratio to third target storage capability information, a ratio to second model information of a reference model, or a ratio to second model information of a reference model.

[0342] In some embodiments, the third capability information comprises third computing capability information; and the processor is further configured to perform any one of the following:

[0343] determine, according to the first model information of an AI or ML model of the terminal, the first model information of an AI or ML model of the terminal in the third capability information;

[0344] determine, according to the first model information of an AI or ML model of the terminal and third target computing capability information, a ratio to the third target computing capability information in the third capability information;

[0345] determine, according to the first model information of an AI or ML model of the terminal and first model information of a reference model or first model information of a reference model, a ratio to the first model information of a reference model or a ratio to the first model information of a reference model in the third capability information.

[0346] In some embodiments, the third capability information comprises third storage capability information; and the processor is further configured to perform any one of the following:

[0347] determine, according to the second model information of an AI or ML model of the terminal, the second model information of an AI or ML model of the terminal in the third capability information;

[0348] determine, according to the second model information of an AI or ML model of the terminal and third target storage capability information, a ratio to the third target storage capability information in the third capability information;

[0349] determine, according to the second model information of an AI or ML model of the terminal and second model information of a reference model or second model information of a reference model, a ratio to the second model information of a reference model or a ratio to the second model information of a reference model in the third capability information.

[0350] In the embodiments of the present disclosure, the terminal can report third capability information of the terminal, such as an absolute value of the third capability and / or a relative value of the third capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML model on the terminal side based on the third capability information, and further implement management of the AI or ML model on the terminal side.

[0351] In some embodiments, the fourth capability information includes a relative value of a fourth capability occupied or required by the AI or ML function of the terminal.

[0352] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0353] The fourth computing capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model; or,

[0354] The fourth capability information is a sum of capability information occupied or required by an activated AI or ML model in the AI or ML function of the terminal; or,

[0355] The fourth capability information is a sum of capability information of an AI or ML model running or performing inference operation in the AI or ML function of the terminal.

[0356] In some embodiments, the fourth capability information includes fourth computing capability information; and the processor further performs any one of the following:

[0357] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, the first model information of at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information is determined.

[0358] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target computing capability information, a ratio relative to the fourth target computing capability information in the fourth capability information is determined.

[0359] determine, according to the first model information of the at least one AI or ML model in the AI or ML function of the terminal and the first model information of the reference model or the first model information of the reference model, a ratio of the fourth capability information to the first model information of the reference model or the first model information of the reference model.

[0360] In some embodiments, the fourth capability information includes fourth storage capability information; and the processor is further configured to perform any one of the following:

[0361] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal, the second model information of the at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information;

[0362] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability information, a ratio of the fourth capability information to the fourth target storage capability information;

[0363] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal and the second model information of the reference model or the second model information of the reference model, a ratio of the fourth capability information to the second model information of the reference model or the second model information of the reference model.

[0364] In the embodiments of the present disclosure, the terminal can report the fourth capability information of the terminal, such as the absolute value of the fourth capability and / or the relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function of the terminal based on the fourth capability information, and further implement the management of the AI or ML function of the terminal.

[0365] In some embodiments, the processor is further configured to perform the following operations:

[0366] perform the following operations according to the time range or symbol of the third capability information occupied or required by each AI or ML model of the terminal when running or reasoning, or according to the time range or symbol of the fourth capability information occupied or required by the AI or ML function of the terminal when running or reasoning:

[0367] For any time or symbol, if it is determined that the third capability information occupied or required by the AI or ML model running or performing reasoning operation of the terminal or the fourth capability information occupied or required by the AI or ML function running or performing reasoning operation of the terminal is greater than the first capability information of the terminal, determine a first AI or ML model and a second AI or ML model from the activated AI or ML model or AI or ML function of the terminal.

[0368] The first AI or ML model is an AI or ML model running or performing inference operation, and the second AI or ML model is an AI or ML model not running or performing inference operation.

[0369] In the embodiments of the present disclosure, the terminal can manage the running or inference of each AI or ML model of the terminal, or manage the running or inference of each AI or ML function of the terminal, to ensure that the total AI or ML related capability occupied by the currently running AI or ML model or AI or ML function does not exceed the AI or ML related capability of the terminal.

[0370] In some embodiments, the fourth capability information is capability information occupied or required by a third AI or ML model in the AI or ML functions of the terminal, and the third AI or ML model is an AI or ML model with the largest occupied or required capability in the AI or ML functions of the terminal.

[0371] In some embodiments, the second capability information includes the number of occupied CSI processing units.

[0372] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information and the calculation latency requirement information of the CSI report have a mapping relationship.

[0373] In some embodiments, the calculation latency requirement information of the CSI report is associated with at least one of the following:

[0374] The complexity of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0375] The parameter amount of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0376] The first capability information;

[0377] The third capability information or the fourth capability information;

[0378] The content of the CSI report;

[0379] The output of the AI or ML model, or the output of the AI or ML function.

[0380] In the embodiments of the present disclosure, after the CSI processing method based on AI or ML, there may be an impact on the CSI processing unit and the CSI processing time. The terminal can report related information of the CSI report, including at least one of the following: second capability information occupied by the CSI report, calculation delay requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the terminal side based on AI or ML CSI processing based on the related information of the CSI report, and further realize the management of the terminal side based on AI or ML CSI processing.

[0381] It should be noted that the second capability information occupied by the CSI report, the calculation delay requirement information of the CSI report, and the capability level information of the CSI report received by the network device can be used to realize the management of the terminal side based on AI or ML CSI processing. Please refer to the embodiments of the above capability reporting method, which will not be repeated here.

[0382] Therefore, the capability determination method provided by the present disclosure can ensure that the AI or ML model or AI or ML function of the terminal side activated or running at any time does not exceed the terminal capability (i.e. the capability of the UE), and the network side knows the activation or running of the AI or ML model or AI or ML function of the terminal side. At the same time, for the AI or ML based CSI reporting, the difference between the CSI processing unit occupied by the AI or ML based CSI reporting and the calculation delay of the traditional (non-AI or ML) CSI reporting is considered, so that the CSI processing unit and the calculation delay occupied by the AI or ML based CSI reporting can be determined according to the capability reported by the UE.

[0383] It should be noted that the capability determination method provided by the present disclosure with the network device as the execution subject can refer to the embodiments shown in FIG. 2, embodiments one to eight for specific implementation process, which will not be repeated here.

[0384] Based on the same technical concept, the present disclosure also provides a capability reporting device, which can realize the functions of the terminal side in the above embodiments.

[0385] Referring to FIG. 3, FIG. 3 is a structural schematic diagram one of the capability reporting device provided by the embodiments of the present disclosure. As shown in FIG. 3, the capability reporting device provided by the present embodiment is applied to a terminal, and the capability reporting device comprises: a transceiver 300, configured to receive and send data under the control of a processor 310.

[0386] In FIG. 3, the bus architecture can include any number of interconnected buses and bridges, specifically, various circuitry of one or more processors represented by the processor 310 and memory represented by the memory 320 linked together. The bus architecture can also link various other circuitry such as peripheral devices, voltage regulators, and power management circuitry, which are well known in the art, and thus, are not further described herein. The bus interface provides an interface. The transceiver 300 can be a plurality of elements, i.e., including a transmitter and a receiver, providing a unit for communicating with various other apparatuses over transmission media, including wireless channels, wired channels, optical cables, and the like transmission media. The user interface 330 can also be an interface capable of externally connecting the required devices for different user equipment, including but not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.

[0387] The processor 310 is responsible for managing the bus architecture and general processing, and the memory 320 can store data used by the processor 310 in performing operations.

[0388] In some embodiments, the processor 310 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD), and the processor can also adopt a multi-core architecture.

[0389] The processor 310 is used to execute any of the methods provided by the embodiments of the present disclosure according to the obtained executable instructions by calling the programs stored in the memory 320. The processor 310 and the memory 320 can also be physically arranged separately.

[0390] In the present embodiment, the memory 320 is configured to store a computer program; the transceiver 300 is configured to transceive data under the control of the processor; and the processor 310 is configured to read the computer program in the memory and perform the following operations:

[0391] transmitting capability information of artificial intelligence (AI) or machine learning (ML), wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of a channel state information (CSI) report occupation, calculation latency requirement information of the CSI report, third capability information of an AI or ML model occupation or requirement of the terminal, or fourth capability information of an AI or ML function occupation or requirement of the terminal.

[0392] In the embodiments of the present disclosure, the terminal can report at least one of the following: first capability information based on AI or ML, second capability information occupied by CSI reporting, calculation delay requirement information of CSI reporting, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal, so as to measure or quantify the AI or ML related capability of the terminal side, and then the receiving end can learn the AI or ML related capability of the terminal side, and then the management of the terminal side based on the AI or ML technology is realized.

[0393] In some embodiments, the first capability information includes a relative value of the first capability of the terminal.

[0394] In some embodiments, the first capability information includes first calculation capability information and / or first storage capability information.

[0395] The first calculation capability information includes any one of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum value of a processing reference model or a reference model; and the first storage capability information includes any one of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum value of a storage reference model or a maximum value of a reference model.

[0396] In some embodiments, the first capability information includes first calculation capability information, and the processor 310 is further configured to perform any one of the following:

[0397] According to the first calculation capability size of the terminal, determine the first calculation capability size in the first capability information;

[0398] According to the first calculation capability size and the first target calculation capability information, determine the ratio relative to the first target calculation capability information in the first capability information;

[0399] According to the first calculation capability size, and the first model information of the reference model or the first model information of the reference model, determine the maximum value of the processing reference model or the maximum value of the reference model in the first capability information.

[0400] In some embodiments, the first capability information includes first storage capability information, and the processor 310 is further configured to perform any one of the following:

[0401] According to the first storage capability size, determine the first storage capability size in the first capability information;

[0402] According to the first storage capability size and the first target storage capability information, determine the ratio relative to the first target storage capability information in the first capability information;

[0403] According to the first storage capability size, and the second model information of the reference model or the second model information of the reference model, a maximum value of the reference model or a maximum value of the reference model stored in the first capability information is determined.

[0404] In the embodiments of the present disclosure, the terminal can report the first capability information of the terminal, such as the absolute value of the first capability and / or the relative value of the first capability, so that the receiving end can measure or quantify the AI or ML related capability of the terminal side based on the first capability information, and further realize the management of the AI or ML model or AI or ML function of the terminal side.

[0405] In some embodiments, the third capability information includes a relative value of the third capability occupied or required by the AI or ML model of the terminal.

[0406] In some embodiments, the third capability information includes third computing capability information and / or third storage capability information.

[0407] The third computing capability information includes any one of the following: first model information of the AI or ML model of the terminal, a ratio relative to the third target computing capability information, a ratio relative to the first model information of the reference model, or a ratio relative to the first model information of the reference model; and the third storage capability information includes any one of the following: second model information of the AI or ML model of the terminal, a ratio relative to the third target storage capability information, a ratio relative to the second model information of the reference model, or a ratio relative to the second model information of the reference model.

[0408] In some embodiments, the third capability information includes third computing capability information; and the processor 310 is further configured to perform any one of the following:

[0409] According to the first model information of the AI or ML model of the terminal, the first model information of the AI or ML model of the terminal in the third capability information is determined.

[0410] According to the first model information of the AI or ML model of the terminal and the third target computing capability information, the ratio relative to the third target computing capability information in the third capability information is determined.

[0411] According to the first model information of the AI or ML model of the terminal, and the first model information of the reference model or the first model information of the reference model, the ratio relative to the first model information of the reference model or the ratio relative to the first model information of the reference model in the third capability information is determined.

[0412] In some embodiments, the third capability information includes third storage capability information; and the processor 310 is further configured to perform any one of the following:

[0413] determining the second model information of the AI or ML model of the terminal in the third capability information according to the second model information of the AI or ML model of the terminal;

[0414] determining a ratio of the third target storage capability information in the third capability information according to the second model information of the AI or ML model of the terminal and the third target storage capability information;

[0415] determining a ratio of the second model information of the reference model or a ratio of the second model information of the reference model in the third capability information according to the second model information of the AI or ML model of the terminal and the second model information of the reference model or the second model information of the reference model.

[0416] In the embodiments of the present disclosure, the terminal can report the third capability information of the terminal, such as the absolute value of the third capability and / or the relative value of the third capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML model on the terminal side based on the third capability information, and further implement the management of the AI or ML model on the terminal side.

[0417] In some embodiments, the fourth capability information includes a relative value of the fourth capability occupied or required by the AI or ML function of the terminal.

[0418] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0419] The fourth computing capability information includes any one of the following: the first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio of the fourth target computing capability information, a ratio of the first model information of the reference model, or a ratio of the first model information of the reference model; the fourth storage capability information includes any one of the following: the second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio of the fourth target storage capability information, a ratio of the second model information of the reference model, or a ratio of the second model information of the reference model; or,

[0420] The fourth capability information is the sum of the capability information occupied or required by the activated AI or ML model in the AI or ML function of the terminal; or,

[0421] The fourth capability information is the sum of the capability information of the AI or ML model running or performing inference operation in the AI or ML function of the terminal.

[0422] In some embodiments, the fourth capability information includes fourth computing capability information; and the processor 310 further performs any one of the following:

[0423] determine, according to the first model information of the at least one AI or ML model in the AI or ML function of the terminal, the first model information of the at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information;

[0424] determine, according to the first model information of the at least one AI or ML model in the AI or ML function of the terminal and the fourth target computing capability information, a ratio of the fourth capability information to the fourth target computing capability information;

[0425] determine, according to the first model information of the at least one AI or ML model in the AI or ML function of the terminal and the first model information of the reference model or the first model information of the reference model, a ratio of the fourth capability information to the first model information of the reference model or the ratio of the fourth capability information to the first model information of the reference model.

[0426] In some embodiments, the fourth capability information includes fourth storage capability information; and the processor 310 is further configured to perform any one of the following:

[0427] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal, the second model information of the at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information;

[0428] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability information, a ratio of the fourth capability information to the fourth target storage capability information;

[0429] determine, according to the second model information of the at least one AI or ML model in the AI or ML function of the terminal and the second model information of the reference model or the second model information of the reference model, a ratio of the fourth capability information to the second model information of the reference model or the ratio of the fourth capability information to the second model information of the reference model.

[0430] In the embodiments of the present disclosure, the terminal can report the fourth capability information of the terminal, such as the absolute value of the fourth capability and / or the relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function on the terminal side based on the fourth capability information, and further implement the management of the AI or ML function on the terminal side.

[0431] In some embodiments, the processor 310 is further configured to perform the following operations:

[0432] According to a time range or a symbol of the third capability information occupied or required by each AI or ML model of the terminal during running or inference, or according to a time range or a symbol of the fourth capability information occupied or required by an AI or ML function of the terminal during running or inference, the following operation is performed:

[0433] For any moment or symbol, if it is determined that the third capability information occupied or required by the AI or ML model of the terminal during running or inference, or the fourth capability information occupied or required by the AI or ML function of the terminal during running or inference, is greater than the first capability information of the terminal, a first AI or ML model and a second AI or ML model are determined from the activated AI or ML models or AI or ML functions of the terminal.

[0434] The first AI or ML model is an AI or ML model that runs or performs inference operation, and the second AI or ML model is an AI or ML model that does not run or perform inference operation.

[0435] In the embodiments of the present disclosure, the terminal can ensure that the total AI or ML related capability occupied by the currently running AI or ML model or AI or ML function does not exceed the AI or ML related capability of the terminal by managing the running or inference of each AI or ML model of the terminal, or by managing the running or inference of each AI or ML function of the terminal.

[0436] In some embodiments, the fourth capability information is capability information occupied or required by a third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is an AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0437] In some embodiments, the second capability information includes the number of CSI processing units occupied.

[0438] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information and the calculation latency requirement information of the CSI report have a mapping relationship.

[0439] In some embodiments, the calculation latency requirement information of the CSI report is associated with at least one of the following:

[0440] The complexity of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0441] The number of parameters of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0442] The first capability information;

[0443] third capability information or fourth capability information;

[0444] content of the CSI report;

[0445] output of the AI or ML model, or output of the AI or ML function.

[0446] In the embodiments of the present disclosure, after the AI or ML based CSI processing method, there can be an impact on the CSI processing unit and the CSI processing time, and the terminal can report related information of the CSI report, including at least one of the following: second capability information occupied by the CSI report, calculation delay requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the terminal side based on AI or ML based CSI processing based on the related information of the CSI report, and further realize the management of the terminal side based on AI or ML based CSI processing.

[0447] Therefore, the capability reporting device provided by the present disclosure can ensure that the AI or ML model or AI or ML function activated or running at any time on the terminal side does not exceed the terminal capability (i.e., the capability of the UE), and the network side can know the activation or running situation of the AI or ML model or AI or ML function on the terminal side. At the same time, for AI or ML based CSI reporting, the difference between the CSI processing unit occupied by AI or ML based CSI reporting and the calculation delay of traditional (non-AI or ML) CSI reporting is considered, so that the CSI processing unit and calculation delay occupied by the AI or ML based CSI reporting of the UE can be determined according to the reported capability.

[0448] It should be noted that the capability reporting device provided by the present disclosure can realize all method steps realized by the terminal side performed capability reporting method embodiment, and can achieve the same technical effects. The same parts and beneficial effects in this embodiment as the method embodiment will not be described here.

[0449] Based on the same technical concept, the present disclosure also provides a capability reporting device, which can realize the functions of the terminal side in the foregoing embodiments.

[0450] Referring to FIG. 4, FIG. 4 is a structural schematic diagram of a capability reporting device provided by an embodiment of the present disclosure. As shown in FIG. 4, the capability reporting device provided by the present embodiment is applied to a terminal, and the capability reporting device comprises: a sending unit 401, and can further comprise a processing unit 402.

[0451] The sending unit is configured to send artificial intelligence (AI) or machine learning (ML) capability information, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of channel state information (CSI) report occupation, calculation time delay requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0452] In the embodiments of the present disclosure, the terminal can report at least one of the following: first capability information based on AI or ML, second capability information of CSI report occupation, calculation time delay requirement information of the CSI report, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal, which can measure or quantify the AI or ML related capability of the terminal side, so that the receiving end can know the AI or ML related capability of the terminal side, and further implement the management of the terminal side based on the AI or ML technology.

[0453] In some embodiments, the first capability information comprises a relative value of the first capability of the terminal.

[0454] In some embodiments, the first capability information comprises first calculation capability information and / or first storage capability information.

[0455] The first calculation capability information comprises any one of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum number of processing reference models or reference models; and the first storage capability information comprises any one of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum number of storage reference models or a maximum number of reference models.

[0456] In some embodiments, the first capability information comprises first calculation capability information, and the method further comprises any one of the following:

[0457] According to the first calculation capability size of the terminal, a first calculation capability size in the first capability information is determined.

[0458] According to the first calculation capability size and first target calculation capability information, a ratio relative to the first target calculation capability information in the first capability information is determined.

[0459] According to the first calculation capability size, and first model information of a reference model or first model information of a reference model, a maximum number of processing the reference model or a maximum number of the reference model in the first capability information is determined.

[0460] In some embodiments, the first capability information comprises first storage capability information, and the processing unit 402 is further configured to perform any one of the following:

[0461] determine, according to the first storage capability size, a first storage capability size in the first capability information;

[0462] determine, according to the first storage capability size and first target storage capability information, a ratio of the first capability information to the first target storage capability information;

[0463] determine, according to the first storage capability size and second model information of a reference model or second model information of a benchmark model, a maximum value of the reference model or a maximum value of the benchmark model in the first capability information.

[0464] In the embodiments of the present disclosure, the terminal can report the first capability information of the terminal, such as the absolute value of the first capability and / or the relative value of the first capability, so that the receiving end can measure or quantify the AI or ML related capability of the terminal side based on the first capability information, and further implement the management of the AI or ML model or AI or ML function of the terminal side.

[0465] In some embodiments, the third capability information comprises a relative value of AI or ML model of the terminal.

[0466] In some embodiments, the third capability information comprises third computing capability information and / or third storage capability information.

[0467] The third computing capability information comprises any one of the following: first model information of the AI or ML model of the terminal, a ratio to third target computing capability information, a ratio to first model information of a benchmark model, or a ratio to first model information of a reference model; and the third storage capability information comprises any one of the following: second model information of the AI or ML model of the terminal, a ratio to third target storage capability information, a ratio to second model information of a benchmark model, or a ratio to second model information of a reference model.

[0468] In some embodiments, the third capability information comprises third computing capability information, and the processing unit 402 is further configured to perform any one of the following:

[0469] determine, according to the first model information of the AI or ML model of the terminal, the first model information of the AI or ML model of the terminal in the third capability information;

[0470] determine, according to the first model information of the AI or ML model of the terminal and third target computing capability information, a ratio of the third capability information to the third target computing capability information.

[0471] According to the first model information of the AI or ML model of the terminal, and the first model information of the reference model or the first model information of the reference model, determine the ratio of the third capability information to the first model information of the reference model or the first model information of the reference model.

[0472] In some embodiments, the third capability information includes third storage capability information; the processing unit 402 is further configured to perform any one of the following:

[0473] According to the second model information of the AI or ML model of the terminal, determine the second model information of the AI or ML model of the terminal in the third capability information;

[0474] According to the second model information of the AI or ML model of the terminal and the third target storage capability information, determine the ratio of the third capability information to the third target storage capability information;

[0475] According to the second model information of the AI or ML model of the terminal, and the second model information of the reference model or the second model information of the reference model, determine the ratio of the third capability information to the second model information of the reference model or the second model information of the reference model.

[0476] In the embodiments of the present disclosure, the terminal can report the third capability information of the terminal, such as the absolute value of the third capability and / or the relative value of the third capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML model on the terminal side based on the third capability information, and further realize the management of the AI or ML model on the terminal side.

[0477] In some embodiments, the fourth capability information includes a relative value of the fourth capability occupied or required by the AI or ML function of the terminal.

[0478] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0479] The fourth computing capability information includes any one of the following: the first model information of at least one AI or ML model in the AI or ML function of the terminal, the ratio to the fourth target computing capability information, the ratio to the first model information of the reference model, or the ratio to the first model information of the reference model; the fourth storage capability information includes any one of the following: the second model information of at least one AI or ML model in the AI or ML function of the terminal, the ratio to the fourth target storage capability information, the ratio to the second model information of the reference model, or the ratio to the second model information of the reference model; or,

[0480] The fourth capability information is a sum of capability information occupied or required by activated AI or ML models in the AI or ML function of the terminal; or

[0481] The fourth capability information is a sum of capability information of AI or ML models running or performing inference operations in the AI or ML function of the terminal.

[0482] In some embodiments, the fourth capability information includes fourth computing capability information; and the processing unit 402 is further configured to perform any one of the following:

[0483] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, determining the first model information of at least one AI or ML model in the fourth capability information of the AI or ML function of the terminal;

[0484] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target computing capability information, determining a ratio of the fourth capability information to the fourth target computing capability information;

[0485] According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, and the first model information of the reference model or the first model information of the reference model, determining a ratio of the fourth capability information to the first model information of the reference model or a ratio of the fourth capability information to the first model information of the reference model.

[0486] In some embodiments, the fourth capability information includes fourth storage capability information; and the processing unit 402 is further configured to perform any one of the following:

[0487] According to the second model information of at least one AI or ML model in the AI or ML function of the terminal, determining the second model information of at least one AI or ML model in the fourth capability information of the AI or ML function of the terminal;

[0488] According to the second model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability information, determining a ratio of the fourth capability information to the fourth target storage capability information;

[0489] According to the second model information of at least one AI or ML model in the AI or ML function of the terminal, and the second model information of the reference model or the second model information of the reference model, determining a ratio of the fourth capability information to the second model information of the reference model or a ratio of the fourth capability information to the second model information of the reference model.

[0490] In the embodiments of the present disclosure, the terminal can report fourth capability information of the terminal, such as an absolute value of the fourth capability and / or a relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function on the terminal side based on the fourth capability information, and further implement management of the AI or ML function on the terminal side.

[0491] In some embodiments, the processing unit 402 is further configured to:

[0492] According to the time range or symbol of the third capability information occupied or required by each AI or ML model of the terminal when running or reasoning, or according to the time range or symbol of the fourth capability information occupied or required by the AI or ML function of the terminal when running or reasoning, the following operations are performed:

[0493] For any moment or symbol, if it is determined that the third capability information occupied or required by the AI or ML model running or performing reasoning operation of the terminal or the fourth capability information occupied or required by the AI or ML function running or performing reasoning operation of the terminal is greater than the first capability information of the terminal, a first AI or ML model and a second AI or ML model are determined from the activated AI or ML models or AI or ML functions of the terminal.

[0494] The first AI or ML model is an AI or ML model running or performing reasoning operation, and the second AI or ML model is an AI or ML model not running or not performing reasoning operation.

[0495] In the embodiments of the present disclosure, the terminal can ensure that the total AI or ML related capability occupied by the currently running AI or ML model or AI or ML function does not exceed the AI or ML related capability of the terminal by managing the running or reasoning of each AI or ML model of the terminal, or by managing the running or reasoning of each AI or ML function of the terminal.

[0496] In some embodiments, the fourth capability information is capability information occupied or required by a third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is an AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0497] In some embodiments, the second capability information includes a number of CSI processing units occupied.

[0498] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information and the calculation latency requirement information of the CSI report have a mapping relationship.

[0499] In some embodiments, the computation latency requirement information of the CSI report is associated with at least one of the following:

[0500] a complexity of the AI or ML model of the terminal or the AI or ML model in the AI or ML function;

[0501] a parameter amount of the AI or ML model of the terminal or the AI or ML model in the AI or ML function;

[0502] first capability information;

[0503] third capability information or fourth capability information;

[0504] content of the CSI report;

[0505] an output of the AI or ML model or an output of the AI or ML function.

[0506] In the embodiments of the present disclosure, after the AI or ML based CSI processing method, there can be an impact on the CSI processing unit and the CSI processing time, and the terminal can report related information of the CSI report, including at least one of the following: second capability information occupied by the CSI report, computation latency requirement information of the CSI report, capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the AI or ML based CSI processing of the terminal side based on the related information of the CSI report, and further realize the management of the AI or ML based CSI processing of the terminal side.

[0507] Therefore, the capability reporting apparatus provided by the present disclosure can ensure that the AI or ML model or AI or ML function of the terminal side activated or running at any time does not exceed the terminal capability (i.e., the capability of the UE), and make the network side aware of the activation or running of the AI or ML model or AI or ML function of the terminal side. At the same time, for the AI or ML based CSI reporting, the difference between the CSI processing unit occupied by the AI or ML based CSI reporting and the computation latency of the traditional (non-AI or ML) CSI reporting is considered, so that the CSI processing unit and the computation latency occupied by the AI or ML based CSI reporting of the UE can be determined according to the reported capability.

[0508] It should be noted that the capability reporting apparatus provided by the present disclosure can realize all the method steps realized by the terminal side performed capability reporting method embodiments, and achieve the same technical effects. Here, the same parts and beneficial effects in the method embodiments will not be described again.

[0509] Based on the same technical concept, the present disclosure also provides a capability determination apparatus, which can realize the functions of the network side (i.e., the network device) in the foregoing embodiments.

[0510] Referring to FIG. 5, a structural schematic diagram of a capability determination apparatus provided by an embodiment of the present disclosure is shown in FIG. 5. The capability determination apparatus provided by the embodiment is applied to a network device, and the capability determination apparatus provided by the embodiment includes a transceiver 500 configured to receive and send data under the control of a processor 510.

[0511] In FIG. 5, the bus architecture can include any number of interconnected buses and bridges, which link various circuits from one or more processors, represented by the processor 510, and memories, represented by the memory 520. The bus architecture can also link various other circuits, such as peripheral devices, voltage stabilizers, and power management circuits, which are well known in the art, and thus, are not further described herein. The bus interface provides an interface. The transceiver 500 can be multiple elements, i.e., including a transmitter and a receiver, which provide units for communicating with various other apparatuses on transmission media, including wireless channels, wired channels, optical cables, and the like. The processor 510 is responsible for managing the bus architecture and general processing, and the memory 520 can store data used by the processor 510 in performing operations.

[0512] The processor 510 can be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a complex programmable logic device (CPLD). The processor can also adopt a multi-core architecture.

[0513] In the embodiment, the memory 520 is configured to store a computer program, the transceiver 500 is configured to transceive data under the control of the processor, and the processor 510 is configured to read the computer program in the memory and perform the following operations:

[0514] receive AI or ML capability information of a terminal, wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

[0515] In the embodiments of the present disclosure, the network device can measure or quantify the AI or ML related capability of the terminal by receiving at least one of the following information reported by the terminal: the first capability information based on AI or ML, the second capability information occupied by the CSI report, the calculation time delay requirement information of the CSI report, the third capability information occupied or required by the AI or ML model of the terminal, or the fourth capability information occupied or required by the AI or ML function of the terminal, thereby achieving the management of the AI or ML technology based on the terminal.

[0516] In some embodiments, the first capability information of the terminal includes a relative value of the first capability information of the terminal.

[0517] In some embodiments, the first capability information is used to determine the first capability information, and the first capability information includes first calculation capability information and / or first storage capability information.

[0518] The first capability information includes any one of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum value of a processing reference model or a reference model; and / or, the first capability information includes any one of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum value of a storage reference model or a maximum value of a reference model.

[0519] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes the first capability information, such as an absolute value of the first capability and / or a relative value of the first capability, and the AI or ML related capability of the terminal can be measured or quantified based on the first capability information, thereby achieving the management of the AI or ML model or the AI or ML function of the terminal.

[0520] In some embodiments, the third capability information includes a relative value of the third capability information occupied or required by the AI or ML model of the terminal.

[0521] In some embodiments, the third capability information is used to determine the third capability information, and the third capability information includes third calculation capability information and / or third storage capability information.

[0522] The third capability information includes any one of the following: first model information of the AI or ML model of the terminal, a ratio relative to third target calculation capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and / or, the third capability information includes any one of the following: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

[0523] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes third capability information, such as an absolute value of the third capability and / or a relative value of the third capability, and the third capability information can be used to measure or quantify the related capability occupied by the AI or ML model on the terminal side, thereby realizing the management of the AI or ML model on the terminal side.

[0524] In some embodiments, the fourth capability information includes a relative value of the fourth capability information occupied or required by the AI or ML function of the terminal.

[0525] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0526] The fourth computing capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to the fourth target computing capability information, a ratio relative to the first model information of the reference model, or a ratio relative to the first model information of the reference model; and the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to the fourth target storage capability information, a ratio relative to the second model information of the reference model, or a ratio relative to the second model information of the reference model; or,

[0527] The fourth capability information is the sum of the capability information occupied or required by the activated AI or ML model in the AI or ML function of the terminal; or,

[0528] The fourth capability information is the sum of the capability information of the AI or ML model performing the inference operation in the AI or ML function of the terminal.

[0529] In some embodiments, the fourth capability information is the capability information occupied or required by the third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is the AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0530] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes fourth capability information of the terminal, such as an absolute value of the fourth capability and / or a relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function on the terminal side based on the fourth capability information, thereby realizing the management of the AI or ML function on the terminal side.

[0531] In some embodiments, the processor 510 is further configured to perform the following operations:

[0532] determine whether to activate the first target AI or ML model and / or the first target AI or ML function according to the capability information; and / or

[0533] determine whether to deactivate the second target AI or ML model and / or the second target AI or ML function according to the capability information;

[0534] The first target AI or ML model and the second target AI or ML model are AI or ML models in at least one AI or ML model of the terminal, and the first target AI or ML function and the second target AI or ML function are AI or ML functions in at least one AI or ML function of the terminal.

[0535] In some embodiments, the processor 510 is specifically configured to, when determining whether to activate the first target AI or ML model and / or the first target AI or ML function according to the capability information, specifically include:

[0536] (in the scenario that the third capability information of the terminal is occupied by the activated AI or ML model of the terminal or the fourth capability information of the terminal is occupied by the activated AI or ML function of the terminal), if the sum of the third capability information occupied or required by the activated AI or ML model and the fourth capability information occupied or required by the activated AI or ML function is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated according to the first capability information, it is determined that the first target AI or ML model and / or the first target AI or ML function is not activated.

[0537] In some embodiments, the processor 510 is specifically configured to, when determining whether to deactivate the second target AI or ML model and / or the second target AI or ML function according to the capability information, specifically include:

[0538] (in the scenario that the third capability information of the terminal is occupied or required by the activated AI or ML model of the terminal or the fourth capability information of the terminal is occupied or required by the activated AI or ML function of the terminal), if the sum of the third capability information occupied or required by the activated AI or ML model and the fourth capability information occupied or required by the activated AI or ML function is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated according to the capability information, it is determined that the second target AI or ML model and / or the second target AI or ML function is deactivated.

[0539] The second target AI or ML model and / or the second target AI or ML function are deactivated, and the first target AI or ML model and / or the first target AI or ML function are activated, and the sum of third capability information occupied or required by the activated AI or ML model in the terminal and fourth capability information occupied by the activated AI or ML function is less than or equal to the first capability information of the terminal.

[0540] In some embodiments, the second target AI or ML model and / or the second target AI or ML function are at least one of:

[0541] The second target AI or ML model and / or the second target AI or ML function are the AI or ML model and / or the AI or ML function activated first in time, or the second target AI or ML model and / or the second target AI or ML function are the AI or ML model and / or the AI or ML function with low activation priority.

[0542] In some embodiments, the second capability information includes the number of CSI processing units occupied.

[0543] In some embodiments, the calculation time delay requirement information of the CSI report is associated with at least one of:

[0544] The complexity of the AI or ML model in the AI or ML model or the AI or ML function of the terminal;

[0545] The number of parameters of the AI or ML model in the AI or ML model or the AI or ML function of the terminal;

[0546] The first capability information;

[0547] The third capability information or the fourth capability information;

[0548] The content of the CSI report;

[0549] The output of the AI or ML model, or the output of the AI or ML function.

[0550] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information and the calculation time delay requirement information of the CSI report have a mapping relationship.

[0551] In the embodiments of the present disclosure, after the CSI processing method based on AI or ML, there may be an impact on the CSI processing unit and the CSI processing time. The terminal can report related information of the CSI report, including at least one of the following: second capability information of the CSI report, calculation delay requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the terminal side based on AI or ML CSI processing based on the related information of the CSI report, and further realize the management of the terminal side based on AI or ML CSI processing.

[0552] Therefore, the capability determination apparatus provided by the present disclosure can ensure that the AI or ML model or AI or ML function activated or running at any time on the terminal side does not exceed the terminal capability (i.e., the capability of the UE), and enables the network side to know the activation or running situation of the AI or ML model or AI or ML function on the terminal side. At the same time, for AI or ML based CSI reporting, the difference between the CSI processing unit occupied by AI or ML based CSI reporting and the calculation delay of traditional (non-AI or ML) CSI reporting is considered, so that the CSI processing unit and calculation delay occupied by the AI or ML based CSI reporting of the UE can be determined according to the reported capability.

[0553] It should be noted that the capability determination apparatus provided by the present disclosure can realize all the method steps realized by the network side in the above-mentioned capability determination method embodiment, and can achieve the same technical effects. Therefore, the same parts and beneficial effects of the method embodiment in the present embodiment will not be described in detail.

[0554] Based on the same technical concept, the present disclosure also provides a capability determination apparatus, which can realize the functions of the network side (i.e., the network device) in the foregoing embodiments.

[0555] Referring to FIG. 6, FIG. 6 is a structural schematic diagram of a capability determination apparatus provided by an embodiment of the present disclosure. As shown in FIG. 6, the capability determination apparatus provided by the present embodiment is applied to a network device, and the capability determination apparatus comprises a receiving unit 601 and a processing unit 602.

[0556] The receiving unit 601 is configured to receive AI or ML capability information of a terminal, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of a CSI report, third capability information required or occupied by an AI or ML model of the terminal, or fourth capability information required or occupied by an AI or ML function of the terminal.

[0557] In the embodiments of the present disclosure, the network device can measure or quantify the AI or ML related capability of the terminal by receiving at least one of the following information reported by the terminal: the first capability information based on AI or ML, the second capability information occupied by the CSI report, the calculation time delay requirement information of the CSI report, the third capability information occupied or required by the AI or ML model of the terminal, or the fourth capability information occupied or required by the AI or ML function of the terminal, thereby achieving the management of the AI or ML technology based on the terminal.

[0558] In some embodiments, the first capability information of the terminal includes a relative value of the first capability information of the terminal.

[0559] In some embodiments, the first capability information is used to determine the first capability information, and the first capability information includes first calculation capability information and / or first storage capability information.

[0560] The first capability information includes any one of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum value of a processing reference model or a reference model; and / or, the first capability information includes any one of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum value of a storage reference model or a maximum value of a reference model.

[0561] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes the first capability information, such as an absolute value of the first capability and / or a relative value of the first capability, and the AI or ML related capability of the terminal can be measured or quantified based on the first capability information, thereby achieving the management of the AI or ML model or the AI or ML function of the terminal.

[0562] In some embodiments, the third capability information includes a relative value of the third capability information occupied or required by the AI or ML model of the terminal.

[0563] In some embodiments, the third capability information is used to determine the third capability information, and the third capability information includes third calculation capability information and / or third storage capability information.

[0564] The third capability information includes any one of the following: first model information of the AI or ML model of the terminal, a ratio relative to third target calculation capability information, a ratio relative to first model information of a reference model or a ratio relative to first model information of a reference model; the third capability information includes any one of the following: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model or a ratio relative to second model information of a reference model.

[0565] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes third capability information, such as an absolute value of the third capability and / or a relative value of the third capability, and the third capability information can be used to measure or quantify the related capability occupied by the AI or ML model on the terminal side, thereby realizing the management of the AI or ML model on the terminal side.

[0566] In some embodiments, the fourth capability information includes a relative value of the fourth capability information occupied or required by the AI or ML function of the terminal.

[0567] In some embodiments, the fourth capability information includes fourth computing capability information and / or fourth storage capability information.

[0568] The fourth computing capability information includes any one of the following: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to the fourth target computing capability information, a ratio relative to the first model information of the reference model, or a ratio relative to the first model information of the reference model; the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to the fourth target storage capability information, a ratio relative to the second model information of the reference model, or a ratio relative to the second model information of the reference model; or,

[0569] The fourth capability information is the sum of the capability information occupied or required by the activated AI or ML model in the AI or ML function of the terminal; or,

[0570] The fourth capability information is the sum of the capability information of the AI or ML model performing inference operation in the AI or ML function of the terminal.

[0571] In some embodiments, the fourth capability information is the capability information occupied or required by the third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is the AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

[0572] In the embodiments of the present disclosure, the capability information of the terminal received by the network device includes fourth capability information of the terminal, such as an absolute value of the fourth capability and / or a relative value of the fourth capability, so that the receiving end can measure or quantify the related capability occupied by the AI or ML function on the terminal side based on the fourth capability information, thereby realizing the management of the AI or ML function on the terminal side.

[0573] In some embodiments, the processing unit 602 is configured to:

[0574] determine whether to activate the first target AI or ML model and / or the first target AI or ML function according to the capability information; and / or,

[0575] determine whether to deactivate the second target AI or ML model and / or the second target AI or ML function according to the capability information;

[0576] wherein the first target AI or ML model and the second target AI or ML model are AI or ML models in the at least one AI or ML model of the terminal, and the first target AI or ML function and the second target AI or ML function are AI or ML functions in the at least one AI or ML function of the terminal.

[0577] In some embodiments, the processing unit 602 is specifically configured to:

[0578] (in the scenario that the third capability information of the terminal is occupied by the AI or ML model of the terminal being activated or the fourth capability information of the terminal is occupied by the AI or ML function of the terminal being activated), according to the first capability information, if the sum of the third capability information occupied or required by the activated AI or ML model and the fourth capability information occupied or required by the activated AI or ML function in the terminal is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated, it is determined that the first target AI or ML model and / or the first target AI or ML function is not activated.

[0579] In some embodiments, the processing unit 602 is specifically configured to:

[0580] (in the scenario that the third capability information of the terminal is occupied or required by the AI or ML model of the terminal being activated or the fourth capability information of the terminal is occupied or required by the AI or ML function of the terminal being activated), according to the capability information, if the sum of the third capability information occupied or required by the activated AI or ML model and the fourth capability information occupied or required by the activated AI or ML function in the terminal is greater than the first capability information of the terminal after the first target AI or ML model and / or the first target AI or ML function is activated, it is determined that the second target AI or ML model and / or the second target AI or ML function is deactivated;

[0581] wherein after the second target AI or ML model and / or the second target AI or ML function is deactivated and the first target AI or ML model and / or the first target AI or ML function is activated, the sum of the third capability information occupied or required by the activated AI or ML model and the fourth capability information occupied by the activated AI or ML function in the terminal is less than or equal to the first capability information of the terminal.

[0582] In some embodiments, the second target AI or ML model and / or the second target AI or ML function is at least one of:

[0583] In some embodiments, the second target AI or ML model and / or the second target AI or ML function is an AI or ML model and / or an AI or ML function that is activated first in time; or the second target AI or ML model and / or the second target AI or ML function is an AI or ML model and / or an AI or ML function that has a low activation priority.

[0584] In some embodiments, the second capability information includes a number of occupied CSI processing units.

[0585] In some embodiments, the computation latency requirement information of the CSI report is associated with at least one of:

[0586] a complexity of an AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0587] a parameter amount of the AI or ML model in the AI or ML model or AI or ML function of the terminal;

[0588] first capability information;

[0589] third capability information or fourth capability information;

[0590] a content of the CSI report;

[0591] an output of the AI or ML model, or an output of the AI or ML function.

[0592] In some embodiments, the capability information further includes capability level information of the AI or ML based CSI report, and the capability level information has a mapping relationship with the computation latency requirement information of the CSI report.

[0593] In the embodiments of the present disclosure, after the AI or ML based CSI processing method, there may be an impact on the CSI processing unit and the CSI processing time, and the terminal can report related information of the CSI report, including at least one of the following: second capability information of the CSI report, computation latency requirement information of the CSI report, and capability level information of the CSI report, so that the receiving end can measure or quantify the related capability or information of the AI or ML based CSI processing of the terminal side based on the related information of the CSI report, and further realize the management of the AI or ML based CSI processing of the terminal side.

[0594] Therefore, the capability determination apparatus provided by the present disclosure can ensure that the AI or ML model or AI or ML function activated or running at the terminal side at any time does not exceed the terminal capability (i.e., the capability of the UE), and enable the network side to know the activation or running of the AI or ML model or AI or ML function at the terminal side. At the same time, for AI or ML based CSI reporting, the difference between the AI or ML based CSI reporting and the traditional (non-AI or ML) CSI reporting in terms of the CSI processing unit occupied and the CSI calculation delay is considered, so that the CSI processing unit and the calculation delay occupied by the AI or ML based CSI reporting can be determined according to the capability reported by the UE.

[0595] It should be noted that the capability determination apparatus provided by the present disclosure can implement all the method steps implemented by the network side capability determination method embodiment described above, and achieve the same technical effects. Therefore, the same parts and beneficial effects of the method embodiment will not be described in detail.

[0596] The technical solutions provided by the embodiments of the present disclosure can be applied to various systems. For example, the applicable systems can be a long term evolution (LTE) system, an LTE frequency division duplex (FDD) system, an LTE time division duplex (TDD) system, a long term evolution advanced (LTE-A) system, a universal mobile system (UMTS), a worldwide interoperability for microwave access (WiMAX) system, a 5G new radio (NR) system and its evolution communication system, a 6G (sixth generation mobile communication technology) system, etc. The various systems can include terminal devices and network devices. The system can also include a core network part, such as an evolved packet system (EPC), a 5G core network (5GC), etc.

[0597] The terminal device involved in the embodiments of the present disclosure can refer to a device that provides voice and / or data connectivity for a user, a handheld device with wireless connection function, or other processing devices connected to a wireless modem, etc. In different systems, the name of the terminal device can also be different, for example, in the 5G system or the 6G system, the terminal device can be called user equipment (User Equipment, UE). The wireless terminal device can be a USB storage device, other personal computer memory devices and a dongle, and can also communicate with one or more core networks (Core Network, CN) through a radio access network (Radio Access Network, RAN). The wireless terminal device can be a mobile terminal device, such as a mobile phone (or called “cellular” phone) and a computer with a mobile terminal device, for example, it can be a portable, pocket, handheld, computer built-in or vehicle-mounted mobile device, which exchanges language and / or data with the radio access network. For example, personal communication service (Personal Communication Service, PCS) phones, cordless phones, session initiation protocol (Session Initiated Protocol, SIP) phones, wireless local loop (Wireless Local Loop, WLL) stations, personal digital assistants (Personal Digital Assistant, PDA), personal computers, tablet computers, machine type communication (Machine-type Communication, MTC) terminal devices, etc. The wireless terminal device can also be called a system, a subscriber unit, a subscriber station, a mobile station, a mobile, a remote station, an access point, a remote terminal, an access terminal, a user terminal, a user agent, a user device, and a wireless access device and a router / modem that meet the limitations of the present definition, etc. The embodiments of the present disclosure are not limited.

[0598] The network device related to the embodiments of the present disclosure can be a base station, which can include a plurality of cells serving terminals. According to different application scenarios, the base station can also be referred to as an access point, or can be a device in an access network that communicates with wireless terminal devices through one or more sectors over an air interface, or other names. The network device can be used to exchange received air frames and Internet Protocol (IP) packets as a router between the wireless terminal device and the rest of the access network, which can include an Internet Protocol (IP) communication network. The network device can also coordinate the management of the properties of the air interface. For example, the network device related to the embodiments of the present disclosure can be an evolved network device (eNB or e-NodeB) in a long term evolution (LTE) system, a 5G base station (gNB) in a 5G network architecture, and the like. It can also be a home evolved base station (HeNB), a relay node, a femto, a pico, a network test device, and the like, which is not limited in the embodiments of the present disclosure. In some network structures, the network device can include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit can also be arranged geographically apart.

[0599] It should be noted that the division of units in the embodiments of the present disclosure is illustrative, and is only a logical functional division. In actual implementation, another division manner can be used. In addition, each functional unit in each embodiment of the present disclosure can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0600] When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solutions of the present disclosure, essentially or the part that makes contributions to the related art, or all or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to perform all or part of the steps of the methods described in the various embodiments of the present disclosure.

[0601] The embodiment of the present disclosure further provides a non-transitory readable storage medium. The non-transitory readable storage medium stores a computer program, and the computer program is used for causing a processor to execute any one of the method embodiments.

[0602] The non-transitory readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to a magnetic storage (such as a floppy disk, a hard disk, a magnetic tape, a magneto-optical disk (MO), etc.), an optical storage (such as a CD, a DVD, a BD, a HVD, etc.), and a semiconductor storage (such as a ROM, an EPROM, an EEPROM, a NAND FLASH, a solid-state disk (SSD)), etc.

[0603] The embodiment of the present disclosure further provides a processor-readable storage medium. The processor-readable storage medium stores a computer program, and the computer program is used for causing a processor to execute any one of the method embodiments.

[0604] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to a magnetic storage (such as a floppy disk, a hard disk, a magnetic tape, a magneto-optical disk (MO), etc.), an optical storage (such as a CD, a DVD, a BD, a HVD, etc.), and a semiconductor storage (such as a ROM, an EPROM, an EEPROM, a NAND FLASH, a solid-state disk (SSD)), etc.

[0605] Those skilled in the art should understand that the embodiments of the present disclosure can be provided as a method, a system, or a computer program product. Therefore, the present disclosure can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present disclosure can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to a magnetic disk storage and an optical storage, etc.) containing computer-usable program code.

[0606] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer executable instructions. These computer executable instructions can be provided to a general purpose computer, a special purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce the functions specified in one or more flows in the flowcharts and / or one or more blocks in the block diagrams.

[0607] These processor-executable instructions can also be loaded into the processor-readable memory of the computer or other programmable data processing devices, so that the computer or other programmable data processing devices can work in a specific way, so that the instructions stored in the processor-readable memory produce a manufactured product including instruction devices, which realize the functions specified in one or more flows of the flow chart and / or one or more blocks of the block diagram.

[0608] Obviously, those skilled in the art can make various modifications and variations to the present disclosure without departing from the spirit and scope of the present disclosure. Thus, if these modifications and variations of the present disclosure fall within the scope of the claims of the present disclosure and their equivalent technologies, the present disclosure also intends to include these modifications and variations.

Claims

1. A capability reporting method, wherein, Applied to a terminal, the method comprises: sending artificial intelligence AI or machine learning ML capability information, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of channel state information CSI report occupation, CSI report calculation latency requirement information, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

2. The method of claim 1, wherein, The first capability information comprises a relative value of the first capability of the terminal.

3. The method of claim 1, wherein, The first capability information comprises first calculation capability information and / or first storage capability information. The first calculation capability information comprises any of the following: a first calculation capability size, a ratio relative to first target calculation capability information, a maximum number of processing reference models or reference models; and the first storage capability information comprises any of the following: a first storage capability size, a ratio relative to first target storage capability information, a maximum number of storing reference models or reference models.

4. The method according to any one of claims 1 to 3, wherein, The first capability information comprises first calculation capability information, and the method further comprises any of the following: determining a first calculation capability size in the first capability information according to a first calculation capability size of the terminal; determining a ratio relative to first target calculation capability information in the first capability information according to the first calculation capability size and the first target calculation capability information; determining a maximum number of processing reference models or reference models in the first capability information according to the first calculation capability size, first model information of a reference model, or first model information of a reference model.

5. The method of claim 2 or 3, wherein, The first capability information comprises first storage capability information, and the method further comprises any of the following: determining a first storage capability size in the first capability information according to a first storage capability size of the terminal; determining a ratio relative to first target storage capability information in the first capability information according to the first storage capability size and the first target storage capability information; determining a maximum number of storing reference models or reference models in the first capability information according to the first storage capability size, second model information of a reference model, or second model information of a reference model.

6. The method of claim 1, wherein, The third capability information comprises a relative value of the third capability of the AI or ML model occupation or requirement of the terminal.

7. The method of claim 1, wherein, The third capability information comprises third calculation capability information and / or third storage capability information. The third calculation capability information comprises any of the following: first model information of the AI or ML model of the terminal, a ratio relative to third target calculation capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and the third storage capability information comprises any of the following: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

8. The method of claim 6 or 7, wherein, The third capability information comprises third calculation capability information; and the method further comprises any of the following: According to the first model information of the AI or ML model of the terminal, determine the first model information of the AI or ML model of the terminal in the third capability information; According to the first model information of the AI or ML model of the terminal and the third target computing capability, determine the ratio of the relative third target computing capability information in the third capability information; According to the first model information of the AI or ML model of the terminal, and the first model information of the reference model or the first model information of the reference model, determine the ratio of the relative first model information of the reference model or the ratio of the first model information of the reference model in the third capability information.

9. The method of claim 6 or 7, wherein, The third capability information includes third storage capability information; the method further comprises any one of the following: According to the second model information of the AI or ML model of the terminal, determine the second model information of the AI or ML model of the terminal in the third capability information; According to the second model information of the AI or ML model of the terminal and the third target storage capability, determine the ratio of the relative third target storage capability information in the third capability information; According to the second model information of the AI or ML model of the terminal, and the second model information of the reference model or the second model information of the reference model, determine the ratio of the relative second model information of the reference model or the ratio of the second model information of the reference model in the third capability information.

10. The method of claim 1, wherein, The fourth capability information includes the relative value of the fourth capability occupied or required by the AI or ML function of the terminal.

11. The method of claim 1, wherein, The fourth capability information includes fourth computing capability information and / or fourth storage capability information; Among them, the fourth computing capability information includes any one of the following: the first model information of at least one AI or ML model in the AI or ML function of the terminal, the ratio of the relative fourth target computing capability information, the ratio of the relative first model information of the reference model or the first model information of the reference model; the fourth storage capability information includes any one of the following: the second model information of at least one AI or ML model in the AI or ML function of the terminal, the ratio of the relative fourth target storage capability information, the ratio of the relative second model information of the reference model or the second model information of the reference model; or, The fourth capability information is the sum of the capability information occupied or required by the activated AI or ML model in the AI or ML function of the terminal; or, The fourth capability information is the sum of the capability information of the AI or ML model running or performing inference operation in the AI or ML function of the terminal.

12. The method of claim 10 or 11, wherein, The fourth capability information includes fourth computing capability information; the method further comprises any one of the following: According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, determine the first model information of at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information; According to the first model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target computing capability, determine the ratio of the relative fourth target computing capability information in the fourth capability information; According to the first model information of at least one AI or ML model in the AI or ML function of the terminal, and the first model information of the reference model or the first model information of the reference model, determine the ratio of the fourth capability information to the first model information of the reference model or the first model information of the reference model.

13. The method of claim 10 or 11, wherein, The fourth capability information includes fourth storage capability information; the method further includes any one of the following: According to the second model information of at least one AI or ML model in the AI or ML function of the terminal, determine the second model information of at least one AI or ML model in the AI or ML function of the terminal in the fourth capability information; According to the second model information of at least one AI or ML model in the AI or ML function of the terminal and the fourth target storage capability, determine the ratio of the fourth capability information to the fourth target storage capability information; According to the second model information of at least one AI or ML model in the AI or ML function of the terminal, and the second model information of the reference model or the second model information of the reference model, determine the ratio of the fourth capability information to the second model information of the reference model or the second model information of the reference model.

14. The method of claim 1, wherein, The method further includes: According to the time range or symbol of the third capability information occupied or required by each AI or ML model of the terminal when running or reasoning, or according to the time range or symbol of the fourth capability information occupied or required by the AI or ML function of the terminal when running or reasoning, perform the following operations: For any time or symbol, if it is determined that the third capability information occupied or required by the AI or ML model of the terminal running or performing reasoning operation or the fourth capability information occupied or required by the AI or ML function of the terminal running or performing reasoning operation is greater than the first capability information of the terminal, determine the first AI or ML model and the second AI or ML model from the activated AI or ML model or AI or ML function of the terminal.

15. The method of claim 10 or 11, wherein, The fourth capability information is the capability information occupied or required by the third AI or ML model in the AI or ML function of the terminal, and the third AI or ML model is the AI or ML model with the largest occupied or required capability in the AI or ML function of the terminal.

16. The method of claim 1, wherein, The second capability information includes the number of occupied CSI processing units.

17. The method of claim 1, wherein, The capability information further includes AI or ML based CSI report capability level information, and the capability level information and the calculation latency requirement information of the CSI report have a mapping relationship.

18. The method of claim 1, wherein, The calculation latency requirement information of the CSI report is associated with at least one of the following: The complexity of the AI or ML model or the AI or ML model in the AI or ML function of the terminal; The parameter amount of the AI or ML model or the AI or ML model in the AI or ML function of the terminal; The first capability information; The third capability information or the fourth capability information; The content of the CSI report; The output of the AI or ML model or the output of the AI or ML function.

19. A capability determination method, wherein, Applied to a network device, the method includes: receiving capability information of AI or ML of the terminal, wherein the capability information comprises at least one of: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

20. The method of claim 19, wherein, The first capability information of the terminal comprises a relative value of a first capability of the terminal.

21. The method of claim 19, wherein, The first capability information comprises first computing capability information and / or first storage capability information. The first computing capability information comprises any one of: a first computing capability size, a ratio relative to a first target computing capability, a maximum number of processing reference models or reference models; and the first storage capability information comprises any one of: a first storage capability size, a ratio relative to a first target storage capability, a maximum number of storage reference models or reference models.

22. The method of claim 19, wherein, The third capability information comprises a relative value of a third capability of AI or ML model occupation or requirement of the terminal.

23. The method of claim 19, wherein, The third capability information comprises third computing capability information and / or third storage capability information. The third computing capability information comprises any one of: first model information of the AI or ML model of the terminal, a ratio relative to third target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and the third storage capability information comprises any one of: second model information of the AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model.

24. The method of claim 19, wherein, The fourth capability information comprises a relative value of a fourth capability of AI or ML function occupation or requirement of the terminal.

25. The method of claim 19, wherein, The fourth capability information comprises fourth computing capability information and / or fourth storage capability information. The fourth computing capability information comprises any one of: first model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target computing capability information, a ratio relative to first model information of a reference model, or a ratio relative to first model information of a reference model; and the fourth storage capability information comprises any one of: second model information of at least one AI or ML model in the AI or ML function of the terminal, a ratio relative to fourth target storage capability information, a ratio relative to second model information of a reference model, or a ratio relative to second model information of a reference model; or The fourth capability information is a sum of capability information of activated AI or ML models in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information of AI or ML models performing inference operation in the AI or ML function of the terminal.

26. The method of claim 24 or 25, wherein, The fourth capability information is capability information of a third AI or ML model in the AI or ML function of the terminal, the third AI or ML model being an AI or ML model with the largest occupation or requirement of capability in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information occupied or required by activated AI or ML models in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information of AI or ML models running or performing inference operations in the AI or ML function of the terminal.

27. The method of claim 19, wherein, The method further comprises: determining, according to the capability information, whether to activate a first target AI or ML model and / or a first target AI or ML function; and / or determining, according to the capability information, whether to deactivate a second target AI or ML model and / or a second target AI or ML function; The first target AI or ML model and the second target AI or ML model are AI or ML models in at least one AI or ML model of the terminal, and the first target AI or ML function and the second target AI or ML function are AI or ML functions in at least one AI or ML function of the terminal.

28. The method of claim 27, wherein, The determining, according to the capability information, whether to activate the first target AI or ML model and / or the first target AI or ML function comprises: If, according to the first capability information, a sum of the third capability information occupied or required by activated AI or ML models and the fourth capability information occupied or required by activated AI or ML functions in the terminal is greater than the first capability information of the terminal after activating the first target AI or ML model and / or the first target AI or ML function, it is determined that the first target AI or ML model and / or the first target AI or ML function is not activated.

29. The method of claim 27, wherein, The determining, according to the capability information, whether to deactivate the second target AI or ML model and / or the second target AI or ML function comprises: If, according to the capability information, a sum of the third capability information occupied or required by activated AI or ML models and the fourth capability information occupied or required by activated AI or ML functions in the terminal is greater than the first capability information of the terminal after activating the first target AI or ML model and / or the first target AI or ML function, it is determined that the second target AI or ML model and / or the second target AI or ML function is deactivated. After the second target AI or ML model and / or the second target AI or ML function is deactivated and the first target AI or ML model and / or the first target AI or ML function is activated, a sum of the third capability information occupied or required by activated AI or ML models and the fourth capability information occupied by activated AI or ML functions in the terminal is less than or equal to the first capability information of the terminal.

30. The method of claim 27, wherein, The second target AI or ML model and / or the second target AI or ML function is at least one. The second target AI or ML model and / or the second target AI or ML function is an AI or ML model and / or an AI or ML function activated first in time; or the second target AI or ML model and / or the second target AI or ML function is an AI or ML model and / or an AI or ML function with a low activation priority.

31. The method of any one of claims 19-25, wherein, The second capability information comprises a number of CSI processing units occupied.

32. The method of any one of claims 19-25, wherein, The computation latency requirement information of the CSI report is associated with at least one of the following: a complexity of an AI or ML model of the terminal or in an AI or ML function of the terminal; a parameter amount of the AI or ML model of the terminal or in the AI or ML function of the terminal; first capability information; third capability information or fourth capability information; content of a CSI report; an output of an AI or ML model or an output of an AI or ML function.

33. The method of any one of claims 19-25, wherein, The capability information further includes capability level information of the AI or ML based CSI report, and the capability level information is in a mapping relationship with the computation latency requirement information of the CSI report.

34. A capability reporting apparatus, comprising: The apparatus is applied to a terminal, and the apparatus includes: a sending unit configured to send capability information of AI or ML, wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by a CSI report, computation latency requirement information of the CSI report, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal.

35. A capability determining apparatus, wherein, The apparatus is applied to a network device, and the apparatus includes: a receiving unit configured to receive capability information of AI or ML of a terminal, wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by a CSI report, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal.

36. A capability reporting apparatus, comprising: The apparatus is applied to a terminal, and the apparatus includes a memory, a transceiver, and a processor: a memory configured to store a computer program; a transceiver configured to transceive data under control of the processor; a processor configured to read the computer program in the memory and perform the following operations:

37. The apparatus of claim 36, wherein, send capability information of artificial intelligence (AI) or machine learning (ML), wherein the capability information includes at least one of the following: first capability information of the terminal, second capability information occupied by channel state information (CSI) report, computation latency requirement information of the CSI report, third capability information occupied or required by an AI or ML model of the terminal, or fourth capability information occupied or required by an AI or ML function of the terminal. The first capability information includes first computation capability information and / or first storage capability information; 38. The apparatus of claim 36, wherein, The first computation capability information includes any one of the following: a first computation capability size, a ratio relative to a first target computation capability, a maximum number of processing reference models or reference models; and the first storage capability information includes any one of the following: a first storage capability size, a ratio relative to a first target storage capability, a maximum number of storage reference models or a maximum number of reference models. The third capability information includes third computation capability information and / or third storage capability information; The third computing capability information includes any one of the following: first model information of an AI or ML model of the terminal, a ratio relative to third target computing capability information, a ratio relative to first model information of a reference model, or a ratio of first model information of a reference model; and the third storage capability information includes any one of the following: second model information of an AI or ML model of the terminal, a ratio relative to third target storage capability information, a ratio relative to second model information of a reference model, or a ratio of second model information of a reference model.

39. The device of claim 36, wherein, The fourth capability information includes fourth computing capability information and / or fourth storage capability information. The fourth computing capability information includes any one of the following: first model information of at least one AI or ML model in an AI or ML function of the terminal, a ratio relative to fourth target computing capability information, a ratio relative to first model information of a reference model, or a ratio of first model information of a reference model; the fourth storage capability information includes any one of the following: second model information of at least one AI or ML model in an AI or ML function of the terminal, a ratio relative to fourth target storage capability information, a ratio relative to second model information of a reference model, or a ratio of second model information of a reference model; or The fourth capability information is a sum of capability information occupied or required by activated AI or ML models in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information of AI or ML models performing inference operations in the AI or ML function of the terminal.

40. The apparatus of claim 39, wherein, The fourth capability information is capability information occupied or required by a third AI or ML model in the AI or ML function of the terminal, the third AI or ML model being an AI or ML model occupying or requiring the most capability in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information occupied or required by activated AI or ML models in the AI or ML function of the terminal; or The fourth capability information is a sum of capability information of AI or ML models performing inference operations in the AI or ML function of the terminal.

41. The device of any one of claims 36-40, wherein, The second capability information includes a number of CSI processing units occupied.

42. The device of any one of claims 36-40, wherein, The capability information further includes capability level information of the AI or ML based CSI report, the capability level information being in a mapping relationship with computing latency requirement information of the CSI report.

43. The device of any one of claims 36-40, wherein, The computing latency requirement information of the CSI report is associated with at least one of the following: Complexity of an AI or ML model of the terminal or an AI or ML model in an AI or ML function of the terminal; Parameter quantity of an AI or ML model of the terminal or an AI or ML model in an AI or ML function of the terminal; First capability information; Third capability information or fourth capability information; Content of a CSI report; Output of an AI or ML model, or output of an AI or ML function.

44. A capability determining apparatus, wherein, The apparatus is applied to a network device, and the apparatus includes a memory, a transceiver, and a processor: a memory for storing a computer program; a transceiver for transceiving data under control of the processor; a processor for reading the computer program in the memory and performing the following operations: receiving capability information of AI or ML of a terminal, wherein the capability information comprises at least one of the following: first capability information of the terminal, second capability information of CSI report occupation, third capability information of AI or ML model occupation or requirement of the terminal, or fourth capability information of AI or ML function occupation or requirement of the terminal.

45. A processor-readable storage medium, wherein, The processor readable storage medium stores a computer program for causing the processor to execute the method of any one of claims 1 to 33.

Images