Management method and device of AI / ML model or function and storage medium

Abstract

The embodiment of the invention provides an AI / ML model or function management method and device and a storage medium, and the method comprises the steps that a terminal receives auxiliary information corresponding to first network equipment, and the auxiliary information is used for the terminal to manage an AI / ML model or function.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a management method, device and storage medium for an AI / ML model or function. Background Art

[0002] The unilateral model deployed on the terminal (also called user equipment (UE)) side, or the part of the bilateral model that belongs to the UE side, may be trained and generated by the UE manufacturer, chip manufacturer or third-party service provider. The network side (NW) is not clear about how the artificial intelligence (AI) and machine learning (ML) models on the UE side are trained. The intrinsic relationship between the input and output learned by the AI / ML model comes from the dataset used in its training process. Different datasets represent different wireless environments. If the current NW environment is similar to the environment represented by the training dataset corresponding to the current AI / ML model or function of the UE, and the intrinsic relationship between the input and output is also similar, the performance of the AI / ML method is often better, and vice versa.

[0003] In order to improve performance, from the perspective of model life cycle management (LCM), if the UE deploys multiple AI / ML models and different models correspond to different training environments, it can select and switch to the most suitable model; or it can update and fine-tune the model deployed by the UE to make it more suitable for the current network environment and improve performance. However, there is currently no specific mechanism to assist the NW or UE in performing the more appropriate LCM operations mentioned above that are more conducive to improving performance. Summary of the invention

[0004] In response to the problems existing in the prior art, the embodiments of the present application provide a management method, device and storage medium for an AI / ML model or function.

[0005] In a first aspect, an embodiment of the present application provides a method for managing an AI / ML model or function, which is applied to a terminal, including:

[0006] Auxiliary information corresponding to the first network device is received, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

[0007] In some embodiments, the auxiliary information includes one or more of the following:

[0008] Signaling related information;

[0009] Information related to signal reception;

[0010] Information related to signal quality;

[0011] Distribution characteristics of model or function-related data;

[0012] Information about the physical space;

[0013] The identifier of the training dataset.

[0014] In some embodiments, the signal transmission related information includes one or more of the following:

[0015] Signaling configuration information;

[0016] Send antenna information.

[0017] In some embodiments, the signal reception related information includes one or more of the following:

[0018] Signal receiving configuration information;

[0019] Receive antenna information.

[0020] In some embodiments, the signal quality related information includes one or more of the following:

[0021] Signal quality metrics;

[0022] Signal quality characteristic value;

[0023] The source of the signal quality characteristic value.

[0024] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0025] signal information corresponding to the terminal;

[0026] Empirical signal information;

[0027] Signal information corresponding to the reference terminal.

[0028] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0029] Information about the distribution characteristics of model or function input data; and / or,

[0030] The distribution characteristics of the model or function output data.

[0031] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0032] a reference distribution of input data to the model or function;

[0033] A value representing the distribution characteristics of the input data to the model or function.

[0034] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0035] a reference distribution of the model or function output data;

[0036] The model or function outputs a value representing a distribution characteristic of the data.

[0037] In some embodiments, the physical space related information includes one or more of the following:

[0038] Network device identification;

[0039] The region where the network device is located;

[0040] Height of network equipment;

[0041] The coverage area of ​​the network equipment;

[0042] Physical space relationship information between multiple network devices;

[0043] Reference scenario for the area where the network equipment is located.

[0044] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0045] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0046] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0047] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0048] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0049] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0050] In a second aspect, an embodiment of the present application further provides a method for managing an AI / ML model or function, including:

[0051] Auxiliary information corresponding to the first network device is sent to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function.

[0052] In some embodiments, the auxiliary information includes one or more of the following:

[0053] Signaling related information;

[0054] Information related to signal reception;

[0055] Information related to signal quality;

[0056] Distribution characteristics of model or function-related data;

[0057] Information about the physical space;

[0058] The identifier of the training dataset.

[0059] In some embodiments, the signal transmission related information includes one or more of the following:

[0060] Signaling configuration information;

[0061] Send antenna information.

[0062] In some embodiments, the signal reception related information includes one or more of the following:

[0063] Signal receiving configuration information;

[0064] Receive antenna information.

[0065] In some embodiments, the signal quality related information includes one or more of the following:

[0066] Signal quality metrics;

[0067] Signal quality characteristic value;

[0068] The source of the signal quality characteristic value.

[0069] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0070] signal information corresponding to the terminal;

[0071] Empirical signal information;

[0072] Signal information corresponding to the reference terminal.

[0073] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0074] Information about the distribution characteristics of model or function input data; and / or,

[0075] The distribution characteristics of the model or function output data.

[0076] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0077] a reference distribution of input data to the model or function;

[0078] A value representing the distribution characteristics of the input data to the model or function.

[0079] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0080] a reference distribution of the model or function output data;

[0081] The model or function outputs a value representing a distribution characteristic of the data.

[0082] In some embodiments, the physical space related information includes one or more of the following:

[0083] Network device identification;

[0084] The region where the network device is located;

[0085] Height of network equipment;

[0086] The coverage area of ​​the network equipment;

[0087] Physical space relationship information between multiple network devices;

[0088] Reference scenario for the area where the network equipment is located.

[0089] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0090] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0091] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0092] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0093] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0094] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0095] In a third aspect, an embodiment of the present application further provides a terminal, including a memory, a transceiver, and a processor;

[0096] A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:

[0097] Auxiliary information corresponding to the first network device is received, where the auxiliary information is used for the terminal to manage an artificial intelligence or machine learning AI / ML model or function.

[0098] In some embodiments, the auxiliary information includes one or more of the following:

[0099] Signaling related information;

[0100] Information related to signal reception;

[0101] Information related to signal quality;

[0102] Distribution characteristics of model or function-related data;

[0103] Information about the physical space;

[0104] The identifier of the training dataset.

[0105] In some embodiments, the signal transmission related information includes one or more of the following:

[0106] Signaling configuration information;

[0107] Send antenna information.

[0108] In some embodiments, the signal reception related information includes one or more of the following:

[0109] Signal receiving configuration information;

[0110] Receive antenna information.

[0111] In some embodiments, the signal quality related information includes one or more of the following:

[0112] Signal quality metrics;

[0113] Signal quality characteristic value;

[0114] The source of the signal quality characteristic value.

[0115] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0116] signal information corresponding to the terminal;

[0117] Empirical signal information;

[0118] Signal information corresponding to the reference terminal.

[0119] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0120] Information about the distribution characteristics of model or function input data; and / or,

[0121] The distribution characteristics of the model or function output data.

[0122] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0123] a reference distribution of input data to the model or function;

[0124] A value representing the distribution characteristics of the input data to the model or function.

[0125] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0126] a reference distribution of the model or function output data;

[0127] The model or function outputs a value representing a distribution characteristic of the data.

[0128] In some embodiments, the physical space related information includes one or more of the following:

[0129] Network device identification;

[0130] The region where the network device is located;

[0131] Height of network equipment;

[0132] The coverage area of ​​the network equipment;

[0133] Physical space relationship information between multiple network devices;

[0134] Reference scenario for the area where the network equipment is located.

[0135] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0136] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0137] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0138] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0139] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0140] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0141] In a fourth aspect, an embodiment of the present application further provides a communication device, including a memory, a transceiver, and a processor;

[0142] A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:

[0143] Auxiliary information corresponding to the first network device is sent to the terminal, where the auxiliary information is used by the terminal to manage artificial intelligence or machine learning AI / ML models or functions.

[0144] In some embodiments, the auxiliary information includes one or more of the following:

[0145] Signaling related information;

[0146] Information related to signal reception;

[0147] Information related to signal quality;

[0148] Distribution characteristics of model or function-related data;

[0149] Information about the physical space;

[0150] The identifier of the training dataset.

[0151] In some embodiments, the signal transmission related information includes one or more of the following:

[0152] Signaling configuration information;

[0153] Send antenna information.

[0154] In some embodiments, the signal reception related information includes one or more of the following:

[0155] Signal receiving configuration information;

[0156] Receive antenna information.

[0157] In some embodiments, the signal quality related information includes one or more of the following:

[0158] Signal quality metrics;

[0159] Signal quality characteristic value;

[0160] The source of the signal quality characteristic value.

[0161] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0162] signal information corresponding to the terminal;

[0163] Empirical signal information;

[0164] Signal information corresponding to the reference terminal.

[0165] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0166] Information about the distribution characteristics of model or function input data; and / or,

[0167] The distribution characteristics of the model or function output data.

[0168] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0169] a reference distribution of input data to the model or function;

[0170] A value representing the distribution characteristics of the input data to the model or function.

[0171] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0172] a reference distribution of the model or function output data;

[0173] The model or function outputs a value representing a distribution characteristic of the data.

[0174] In some embodiments, the physical space related information includes one or more of the following:

[0175] Network device identification;

[0176] The region where the network device is located;

[0177] Height of network equipment;

[0178] The coverage area of ​​the network equipment;

[0179] Physical space relationship information between multiple network devices;

[0180] Reference scenario for the area where the network equipment is located.

[0181] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0182] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0183] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2K One of the information combinations; K is an integer greater than or equal to 1;

[0184] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0185] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0186] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0187] In a fifth aspect, an embodiment of the present application further provides a management device for an AI / ML model or function, including:

[0188] A receiving unit is used to receive auxiliary information corresponding to the first network device, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

[0189] In some embodiments, the auxiliary information includes one or more of the following:

[0190] Signaling related information;

[0191] Information related to signal reception;

[0192] Information related to signal quality;

[0193] Distribution characteristics of model or function-related data;

[0194] Information about the physical space;

[0195] The identifier of the training dataset.

[0196] In some embodiments, the signal transmission related information includes one or more of the following:

[0197] Signaling configuration information;

[0198] Send antenna information.

[0199] In some embodiments, the signal reception related information includes one or more of the following:

[0200] Signal receiving configuration information;

[0201] Receive antenna information.

[0202] In some embodiments, the signal quality related information includes one or more of the following:

[0203] Signal quality metrics;

[0204] Signal quality characteristic value;

[0205] The source of the signal quality characteristic value.

[0206] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0207] signal information corresponding to the terminal;

[0208] Empirical signal information;

[0209] Signal information corresponding to the reference terminal.

[0210] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0211] Information about the distribution characteristics of model or function input data; and / or,

[0212] The distribution characteristics of the model or function output data.

[0213] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0214] a reference distribution of input data to the model or function;

[0215] A value representing the distribution characteristics of the input data to the model or function.

[0216] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0217] a reference distribution of the model or function output data;

[0218] The model or function outputs a value representing a distribution characteristic of the data.

[0219] In some embodiments, the physical space related information includes one or more of the following:

[0220] Network device identification;

[0221] The region where the network device is located;

[0222] Height of network equipment;

[0223] The coverage area of ​​the network equipment;

[0224] Physical space relationship information between multiple network devices;

[0225] Reference scenario for the area where the network equipment is located.

[0226] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0227] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0228] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0229] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0230] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0231] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0232] In a sixth aspect, an embodiment of the present application further provides a management device for an AI / ML model or function, including:

[0233] A sending unit is used to send auxiliary information corresponding to the first network device to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function.

[0234] In some embodiments, the auxiliary information includes one or more of the following:

[0235] Signaling related information;

[0236] Information related to signal reception;

[0237] Information related to signal quality;

[0238] Distribution characteristics of model or function-related data;

[0239] Information about the physical space;

[0240] The identifier of the training dataset.

[0241] In some embodiments, the signal transmission related information includes one or more of the following:

[0242] Signaling configuration information;

[0243] Send antenna information.

[0244] In some embodiments, the signal reception related information includes one or more of the following:

[0245] Signal receiving configuration information;

[0246] Receive antenna information.

[0247] In some embodiments, the signal quality related information includes one or more of the following:

[0248] Signal quality metrics;

[0249] Signal quality characteristic value;

[0250] The source of the signal quality characteristic value.

[0251] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0252] signal information corresponding to the terminal;

[0253] Empirical signal information;

[0254] Signal information corresponding to the reference terminal.

[0255] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0256] Information about the distribution characteristics of model or function input data; and / or,

[0257] The distribution characteristics of the model or function output data.

[0258] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0259] a reference distribution of input data to the model or function;

[0260] A value representing the distribution characteristics of the input data to the model or function.

[0261] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0262] a reference distribution of the model or function output data;

[0263] The model or function outputs a value representing a distribution characteristic of the data.

[0264] In some embodiments, the physical space related information includes one or more of the following:

[0265] Network device identification;

[0266] The region where the network device is located;

[0267] Height of network equipment;

[0268] The coverage area of ​​the network equipment;

[0269] Physical space relationship information between multiple network devices;

[0270] Reference scenario for the area where the network equipment is located.

[0271] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation, administration, and maintenance (OAM) pre-configuration information.

[0272] In some embodiments, the training data set identifier includes at least one bit field, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0273] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0274] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0275] The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

[0276] In some embodiments, the training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0277] In the seventh aspect, an embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is used to enable a computer to execute the management method of the AI / ML model or function as described in the first aspect, or to execute the management method of the AI / ML model or function as described in the second aspect.

[0278] In an eighth aspect, an embodiment of the present application further provides a communication device, in which a computer program is stored, and the computer program is used to enable the communication device to execute the management method of the AI / ML model or function as described in the first aspect, or to execute the management method of the AI / ML model or function as described in the second aspect.

[0279] In the ninth aspect, an embodiment of the present application further provides a processor-readable storage medium, wherein the processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the management method of the AI / ML model or function as described in the first aspect, or to execute the management method of the AI / ML model or function as described in the second aspect.

[0280] In the tenth aspect, an embodiment of the present application further provides a chip product, in which a computer program is stored, and the computer program is used to enable the chip product to execute the management method of the AI / ML model or function described in the first aspect, or to execute the management method of the AI / ML model or function described in the second aspect.

[0281] The AI / ML model or function management method, device and storage medium provided in the embodiments of the present application enable the terminal to perform accurate and appropriate lifecycle management of the AI / ML model or function by receiving auxiliary information corresponding to the first network device, which helps to improve the transmission performance of the terminal and the overall performance of the network and avoid unnecessary AI / ML calculation and storage overhead. BRIEF DESCRIPTION OF THE DRAWINGS

[0282] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the following is a brief introduction to the drawings required for use in the embodiments or the related technical descriptions. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0283] Figure 1 It is one of the flowcharts of the method for managing the AI / ML model or function provided in the embodiment of the present application;

[0284] Figure 2 This is the second flowchart of the method for managing the AI / ML model or function provided in the embodiment of the present application;

[0285] Figure 3 This is an example diagram of an application scenario of the management method of the AI / ML model or function provided in an embodiment of the present application;

[0286] Figure 4 is a schematic diagram of the structure of a terminal provided in an embodiment of the present application;

[0287] Figure 5 is a schematic diagram of the structure of a communication device provided in an embodiment of the present application;

[0288] Figure 6 It is one of the structural schematic diagrams of the management device of the AI / ML model or function provided in the embodiment of the present application;

[0289] Figure 7 This is the second structural diagram of the management device of the AI / ML model or function provided in the embodiment of the present application. DETAILED DESCRIPTION

[0290] In the embodiments of the present application, the term "and / or" describes the association relationship of the associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0291] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0292] In the embodiments of the present application, the terms "first", "second", etc. are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0293] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0294] In order to facilitate a clearer understanding of the technical solutions of the embodiments of the present application, some technical contents related to the embodiments of the present application are first introduced.

[0295] AI / ML model refers to a data-driven algorithm that can infer and predict outputs based on certain inputs. Depending on the data collected during the training process, a model often has certain adaptation scenarios. UE can manage the AI / ML model on the UE side. If the AI / ML model on the UE side is visible to the NW, the NW can manage the model on the UE side.

[0296] AI / ML functionality refers to the AI / ML reasoning function of UE on certain functions (such as channel state information (CSI) compression, beam prediction, and positioning). This function is visible to NW, and NW can perform function-level management. However, NW does not know how many AI / ML models UE uses to implement AI / ML functionality. Therefore, model-level management is generally managed by UE itself.

[0297] If the manager of the AI / ML model or LCM of the function on the UE side is the UE, since the UE is not clear about the wireless environment characteristics and NW characteristics in the current cell, it is difficult for the UE to appropriately select / switch / update the model it manages.

[0298] Once the UE performs incorrect and inappropriate LCM management, such as activating an AI / ML model with poor performance, it may lead to a decrease in UE transmission performance and NW performance, and cause the UE to waste computing power and storage overhead to run AI / ML models / functions.

[0299] In order to solve the above problems, the present application embodiment proposes a management method for AI / ML models or functions, which has the following advantages:

[0300] (1) It can quantitatively characterize information such as the environment, site, and data set, making it possible to use them as auxiliary information.

[0301] (2) The NW can send auxiliary information to the UE to assist the UE in performing accurate and appropriate lifecycle management of the AI / ML model or function on the UE side according to the matching of the AI / ML model or function on the UE side with the current cell environment / configuration / conditions, thereby improving the transmission performance of the UE and the overall performance of the network and avoiding unnecessary AI / ML computing / storage overhead.

[0302] Figure 1 One of the flow charts of the method for managing the AI / ML model or function provided in the embodiment of the present application, the method is applied to the terminal, such as Figure 1 As shown, the method comprises the following steps:

[0303] Step 100: Receive auxiliary information corresponding to the first network device, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

[0304] Specifically, the AI / ML model or function in this application can be an AI / ML model or function on the terminal side or an AI / ML model or function deployed on other devices, which is not limited in this application. For ease of understanding, the following text takes the AI / ML model or function on the terminal side as an example, but it should be understood that AI / ML models or functions on other sides can also be applied here.

[0305] The terminal may be deployed with a unilateral model on the terminal side, or the terminal side part of the bilateral model. The model deployed in the terminal may be trained and generated by the terminal manufacturer, chip manufacturer or third-party service provider. The terminal may manage the LCM of its own AI / ML model or function. In order to ensure the performance of the AI / ML model or function on the terminal side in the current network, the terminal may receive auxiliary information corresponding to the first network device, and manage its own AI / ML model or function according to the auxiliary information.

[0306] The first network device may be any network device, and this application does not impose any limitation thereto. For example, the first network device may be a base station, a location management function (LMF) or other network device.

[0307] The auxiliary information corresponding to the first network device may be sent by the first network device to the terminal, or may be sent by other communication devices to the terminal. For example, a neighboring cell of the first network device or a base station close to the first network device may send the auxiliary information corresponding to the first network device to the terminal.

[0308] It is understandable that the first network device can send the auxiliary information corresponding to itself to the terminal, or can send the auxiliary information corresponding to other network devices to the terminal for other network devices. For example, the first network device can send the auxiliary information corresponding to its neighboring area to the terminal.

[0309] The auxiliary information corresponding to the first network device may be used to indicate one or more of the current network scenario (e.g., scenario and altitude of the network device), configuration (e.g., reference signal configuration information), and conditions (e.g., data distribution characteristics). The auxiliary information corresponding to the first network device may be the auxiliary information of the first network device itself, or the auxiliary information of a network device with a wireless network environment similar to that of the first network device, or the auxiliary information of a reference network device with a wireless network environment similar to that of the first network device.

[0310] In one implementation, the terminal may compare one or more applicable scenarios / configurations / conditions corresponding to its own AI / ML model or function with the scenarios / configurations / conditions indicated by the auxiliary information, and when all or part of the scenarios match, activate the AI / ML model or function, or initiate a suggestion or request to the first network device to activate the AI / ML model or function.

[0311] In one implementation, the terminal may compare one or more applicable scenarios / configurations / conditions corresponding to its own AI / ML model or function with the scenarios / configurations / conditions indicated by the auxiliary information. When the scenarios / configurations / conditions that "significantly affect generalization performance" match, the AI / ML model or function is activated; otherwise, the AI / ML model or function is deactivated, or a suggestion or request to the first network device to deactivate the AI / ML model or function is initiated.

[0312] In one implementation, the terminal may compare one or more applicable scenarios / configurations / conditions corresponding to its own AI / ML model or function with the scenario / configuration / condition indicated by the auxiliary information, select the AI / ML model or function that best matches the scenario / configuration / condition indicated by the auxiliary information for activation, or switch to the AI / ML model or function that best matches the current scenario / configuration / condition, or initiate a suggestion or request to the first network device to switch the AI / ML model or function.

[0313] In one embodiment, when the terminal collects training data for training / updating / fine-tuning an AI / ML model or function, the terminal classifies the collected data based on the auxiliary information. Different data sets collected from different network devices are associated with the auxiliary information corresponding to the different network devices. The terminal uses different data sets to train / update / fine-tune the AI / ML model or function. After completing the training / updating / fine-tuning, the terminal can activate the trained / updated / fine-tuned AI / ML model / function in different network devices associated with the data sets.

[0314] The management method of the AI / ML model or function provided in the embodiment of the present application can be applied to the fifth generation mobile communication system (the 5th generation mobile communication, 5G) new radio (New Radio, NR) system, and can also be applied to other systems that require AI / ML model-assisted information interaction.

[0315] The AI / ML model or function management method provided in the embodiment of the present application enables the terminal to perform accurate and appropriate lifecycle management of the AI / ML model or function by receiving auxiliary information corresponding to the first network device, which helps to improve the transmission performance of the terminal and the overall performance of the network and avoid unnecessary AI / ML calculation and storage overhead.

[0316] In some embodiments, the auxiliary information includes one or more of the following:

[0317] (1) Signal sending related information.

[0318] Specifically, the information related to information sending may be information such as scenarios, configurations or conditions related to signal sending, which is not specifically limited here.

[0319] (2) Information related to signal reception.

[0320] Specifically, the information related to information reception may be information such as scenarios, configurations or conditions related to signal reception, which is not specifically limited here.

[0321] (3) Information related to signal quality.

[0322] Specifically, the signal quality related information may be characteristic information related to the signal quality, such as a maximum value, a minimum value, a median value, etc. of the signal quality, which is not specifically limited here.

[0323] (4) Distribution characteristic information of model or function related data.

[0324] Specifically, the model or function related data may be data used as model or function input, data used as model or function output, or other model related data. The distribution characteristic information of the model or function related data may be any information used to characterize the distribution characteristics of the model or function related data, such as reference characteristics, or the maximum value, minimum value, average value, etc. of the data distribution.

[0325] In some embodiments, the distribution characteristic information of the model or function related data may include:

[0326] Information about the distribution characteristics of model or function input data; and / or,

[0327] The distribution characteristics of the model or function output data.

[0328] (5) Information related to the physical space.

[0329] Specifically, the physical space related information may be information related to the physical space where the communication device is located, such as the region where the communication device is located, the height of the communication device, etc.

[0330] (6) Training data set identification.

[0331] The training data set identifier corresponding to the first network device may also be expressed as the training data set identifier associated with the first network device.

[0332] The training data set identifier associated with the first network device may be that the first network device participates in the collection of the data set corresponding to the training data set identifier, such as participating in the signal transmission of the data sample, or participating in the signal reception of the data sample. Alternatively, although the first network device does not directly participate in the collection of the data set corresponding to the training data set identifier, the environment / configuration / condition of the network device participating in the collection of the data set is the same or similar to the environment / configuration / condition of the first network device, and the training data set identifier may also be considered to be the training data set identifier associated with the first network device.

[0333] The first network device may indicate the training data set identifier for itself, and may also indicate the training data set identifier for other network devices (eg, a neighboring cell of the first network device).

[0334] The first network device may indicate one or more training data set identifiers.

[0335] In some embodiments, the signal transmission related information includes one or more of the following:

[0336] (1) Signal transmission configuration information.

[0337] Specifically, when a network device sends a signal and a terminal receives or measures the signal, the signal sending configuration information may be reference signal (RS) configuration information of the network device when sending the signal.

[0338] The RS configuration information of the network device when sending a signal may include one or more of the following:

[0339] Signal type (e.g., Channel State Information Reference Signal (CSI-RS), Synchronization Signal Block (SSB), Positioning Reference Signal (PRS), etc.), number, time domain resources, frequency domain resources, number of ports, sequence, RS resource ID, RS resource set ID, and RS ID.

[0340] (2) Send antenna information.

[0341] Specifically, when the network device sends a signal and the terminal receives or measures the signal, the transmitting antenna information of the transmitting end device may be information related to the transmitting antenna of the network device.

[0342] The information related to the transmitting antenna of the network device may include one or more of the following:

[0343] The network device’s transmission (Tx) channel (panel), antenna array pattern, vertical angle of the transmission beam, horizontal angle of the transmission beam, transmission power, and relative power at each angle between multiple transmission beams.

[0344] After obtaining the above signal transmission related information, the terminal can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0345] A. Management through a simple comparison method, for example, only when the applicable configuration of the AI / ML model or function on the terminal side is the same as the configuration indicated by the network device, and / or the relevant information of the transmitting antenna is the same, decide to activate the AI / ML model or function on the terminal side, or switch to the AI / ML model or function with matching configuration / antenna, or initiate a suggestion or request to the first network device to activate the AI / ML model or function.

[0346] This approach is simple to implement, but may miss opportunities for the UE's AI / ML models / functions to be utilized.

[0347] B. Extract the wireless communication environment characteristics represented by the sending configuration and sending antenna indicated by the network device, and compare them with the wireless communication environment characteristics applicable to the terminal-side AI / ML model or function. When the environmental characteristics are the same / similar, decide to activate the terminal-side AI / ML model or function, or switch to the terminal-side AI / ML model or function with matching configuration / antenna, or initiate a suggestion or request to activate the AI / ML model or function to the first network device.

[0348] Environmental characteristics may be, for example, antenna horizontal coverage, vertical coverage, obstacle height and density, whether the base station is placed indoors or outdoors, the maximum coherence time or minimum coherence bandwidth of the detectable channel, etc.

[0349] For example, the terminal-side AI / ML model or function is used for mobility management. The terminal can determine the horizontal antenna coverage corresponding to the network device through the transmitting antenna-related information indicated by the network device, and compare it with the horizontal antenna coverage suitable for the terminal-side AI / ML model or function. If the two are found to be similar, even though the antenna configuration indicated by the network device is different from the antenna configuration of the network device during the training of the terminal-side AI / ML model or function, it is still decided to activate the terminal-side AI / ML model or function.

[0350] This method is more complicated to implement, but it can maximize the reasoning capabilities of the AI / ML model or function on the terminal side and improve network performance.

[0351] In some embodiments, the signal reception related information includes one or more of the following:

[0352] (1) Signal reception configuration information.

[0353] In the case where the terminal sends a signal and the network device receives or measures the signal, the signal reception configuration information may be RS configuration information of the network device when receiving or measuring the signal.

[0354] The RS configuration information of the network device during signal reception or measurement may include one or more of the following:

[0355] Signal type (e.g. Sounding Reference Signal (SRS)), number, time domain resources, frequency domain resources, number of ports, sequence, RS resource ID, RS resource set ID and RS ID.

[0356] (2) Receiving antenna information.

[0357] When the terminal sends a signal and the network device receives or measures the signal, the receiving antenna information may be information related to the receiving antenna of the network device.

[0358] The information related to the receiving antenna of the network device may include one or more of the following:

[0359] The network device’s reception (Rx) transmission channel (panel), antenna array pattern, vertical angle of the receiving beam, horizontal angle of the receiving beam, and relative power at each angle between multiple receiving beams.

[0360] After obtaining the above signal reception related information, the terminal can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0361] A. Processing is performed through a simple comparison method. For example, only when the applicable configuration of the AI / ML model or function on the terminal side is the same as the configuration indicated by the network device, and / or the relevant information of the receiving antenna is the same, decide to activate the AI / ML model or function on the terminal side, or switch to the AI / ML model or function with matching configuration / antenna, or initiate a suggestion or request to the first network device to activate the AI / ML model or function.

[0362] This approach is simple to implement, but may miss opportunities for the UE's AI / ML models / functions to be utilized.

[0363] B. Extract the wireless communication environment characteristics represented by the receiving configuration and receiving antenna indicated by the network device, and compare them with the wireless communication environment characteristics applicable to the terminal-side AI / ML model or function. When the environmental characteristics are the same / similar, decide to activate the terminal-side AI / ML model or function, or switch to the terminal-side AI / ML model or function with matching configuration / antenna, or initiate a suggestion or request to activate the AI / ML model or function to the first network device.

[0364] Environmental characteristics may be, for example, antenna horizontal coverage, vertical coverage, obstacle height and density, whether the base station is placed indoors or outdoors, the maximum coherence time or minimum coherence bandwidth of the detectable channel, etc.

[0365] For example, the terminal-side AI / ML model or function is used for mobility management. The terminal can determine the horizontal antenna coverage corresponding to the network device through the receiving antenna-related information indicated by the network device, and compare it with the horizontal antenna coverage suitable for the terminal-side AI / ML model or function. If the two are found to be similar, even though the antenna configuration indicated by the network device is different from the antenna configuration of the network device when the terminal-side AI / ML model or function was trained, it is still decided to activate the terminal-side AI / ML model or function.

[0366] This method is more complicated to implement, but it can maximize the reasoning capabilities of the AI / ML model or function on the terminal side and improve network performance.

[0367] In some embodiments, the signal quality related information includes one or more of the following:

[0368] (1) Signal quality measurement.

[0369] Specifically, the signal quality metric may be one or more of the following:

[0370] Signal to Noise Ratio (SNR), Signal to Interference and Noise Ratio (SINR), Reference Signal Receiving Power (RSRP) and Reference Signal Receiving Quality (RSRQ). In addition, there may be other signal quality measurement methods, which are not specifically limited here.

[0371] (2) Signal quality characteristic value.

[0372] Specifically, the signal quality characteristic value may include a signal quality characteristic value corresponding to one or more signal quality metrics, and the characteristic value of the signal quality metric may be, for example, a reference value, a maximum value, a minimum value, a mode, a median, a mean, a standard deviation, a variance, and other characteristic values. In addition, there may be other signal quality characteristic values, which are not specifically limited here.

[0373] Among them, the reference value can be the most appropriate value of the signal quality metric, such as the most appropriate SINR value, the most appropriate RSRP value, or a value that enables the AI / ML model or function to achieve a specific performance indicator.

[0374] The maximum, minimum, mode, median, mean, standard deviation, and variance can be used to indicate the most suitable range, such as the most suitable SINR range, the most suitable RSRQ range, or the range that enables the AI / ML model or function to achieve a specific performance indicator.

[0375] When there are multiple signal quality metrics, the characteristic value (reference value, maximum value, minimum value, mode, median, mean, standard deviation and variance, etc.) of each signal quality metric may be indicated respectively.

[0376] (3) The source of signal quality characteristic values.

[0377] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0378] a. Signal information corresponding to the terminal.

[0379] The signal information corresponding to the terminal may be a signal of the terminal received by the network device, such as a physical random access channel (PRACH), a physical uplink shared channel (PUSCH), a physical uplink control channel (PUCCH), a demodulation reference signal (DMRS), SRS, etc. sent by the terminal.

[0380] b. Empirical signal information.

[0381] The empirical signal information may be feedback reported by other terminals in the network device after measuring the signal sent by the network device, or the quality of the signal sent by the network device from other terminals, or the signal quality received by the network device from multiple terminals and then a virtual terminal is created to represent the signal quality of multiple terminals.

[0382] c. Signal information corresponding to the reference terminal.

[0383] The signal information corresponding to the reference terminal may be a received signal quality report of a reference terminal in the network device, or the quality of a transmitted signal received by the network device from the reference terminal.

[0384] After obtaining the above signal quality related information, the terminal can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0385] A. Compare the reference value of the signal quality indicated by the network device with the typical value of the signal quality applicable to the terminal-side AI / ML model or function (such as the mean, mode, or expected value during network planning). When the difference between the two is less than a certain threshold value, decide to activate / switch to the terminal-side AI / ML model or function; otherwise, deactivate the terminal-side AI / ML model or function, or switch to a more appropriate terminal-side AI / ML model or function, or initiate a suggestion or request to the first network device to deactivate the AI / ML model or function.

[0386] B. Compare the range of signal quality indicated by the network device with the typical value of signal quality applicable to the AI / ML model or function on the terminal side. If the typical value of signal quality applicable to the AI / ML model or function is within the appropriate range indicated by the network device, decide to activate / switch to the AI / ML model or function on the terminal side; otherwise, deactivate the AI / ML model or function on the terminal side, or switch to a more appropriate AI / ML model or function on the terminal side.

[0387] C. Compare the range of signal quality indicated by the network device with the range of signal quality applicable to the terminal-side AI / ML model or function. If the difference between the two (such as the sum of the difference between the maximum value and the minimum value) is less than a certain threshold, decide to activate / switch to the terminal-side AI / ML model or function; otherwise, deactivate the terminal-side AI / ML model or function, or switch to a more appropriate terminal-side AI / ML model or function, or initiate a suggestion or request to the first network device to deactivate the AI / ML model or function.

[0388] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following, and the following takes the model input as an example:

[0389] (1) The reference distribution of the input data to the model or function.

[0390] Specifically, the reference distribution of the model or function input data can be one of the following predefined distributions: uniform distribution, Gaussian distribution (i.e., normal distribution), truncated Gaussian distribution, exponential distribution, or chi-square distribution. In addition, there can be other predefined distributions for reference, which are not specifically limited here.

[0391] For example, if the data distribution of the model input data is consistent with a certain reference distribution (such as the difference between the statistical distribution curve of the input data and the probability density distribution curve of a predefined distribution is less than a certain threshold value), then the reference distribution can be directly indicated, indicating that the model input data conforms to the reference distribution.

[0392] (2) A value representing the distribution characteristics of the input data to a model or function.

[0393] Specifically, the characterization value of the distribution characteristics of the model or function input data may be one or more of the following: reference value, characteristic value, maximum value, minimum value, mode, median, mean, standard deviation and variance. In addition, there may be other characterization values, which are not specifically limited here.

[0394] The reference value and / or characteristic value of the distribution characteristic of the model input data may be a special reference value in a certain distribution, such as an exponential value in an exponential distribution.

[0395] Maximum, minimum, mode, median, mean, standard deviation, and variance can be used to indicate the most appropriate range of model input data distribution characteristics, or the range of model input data distribution characteristics that enables the AI / ML model or function to achieve a specific performance indicator.

[0396] For example, when it is impossible to give a reference distribution of the model or function input data, the distribution characteristics of the input data can be described from various aspects through information such as reference value, characteristic value, maximum value, minimum value, etc.

[0397] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following, and the following takes the model output as an example:

[0398] (1) The reference distribution of model or function output data.

[0399] Specifically, the reference distribution of the model or function output data can be one of the following predefined distributions: uniform distribution, Gaussian distribution (i.e., normal distribution), truncated Gaussian distribution, exponential distribution, or chi-square distribution. In addition, there can be other predefined distributions for reference, which are not specifically limited here.

[0400] For example, if the data distribution of the model output data is consistent with a certain reference distribution (such as the difference between the statistical distribution curve of the output data and the probability density distribution curve of a predefined distribution is less than a certain threshold value), then the reference distribution can be directly indicated, indicating that the model output data conforms to the reference distribution.

[0401] (2) A value representing the distribution characteristics of the output data of a model or function.

[0402] Specifically, the characterization value of the distribution characteristics of the model or function output data may be one or more of the following: reference value, characteristic value, maximum value, minimum value, mode, median, mean, standard deviation and variance. In addition, there may be other characterization values, which are not specifically limited here.

[0403] The reference value and / or characteristic value of the distribution characteristic of the model output data may be a special reference value in a certain distribution, such as an exponential value in an exponential distribution.

[0404] The maximum, minimum, mode, median, mean, standard deviation, and variance can be used to indicate the most appropriate range of model output data distribution characteristics, or the range of model output data distribution characteristics that enables the AI / ML model or function to achieve a specific performance indicator.

[0405] For example, when it is impossible to give a reference distribution of the output data of a model or function, the distribution characteristics of the output data can be described from various aspects through information such as reference value, characteristic value, maximum value, minimum value, etc.

[0406] The management method of the AI / ML model or function provided in the embodiment of the present application solves the problem that the input and output distribution characteristics cannot be portrayed as auxiliary information by designing the distribution characteristic information of the model or function-related data as auxiliary information. When the distribution of the input or output data of the model or function is highly consistent with a specific predefined distribution (for example, the difference between the statistical distribution curve of the input or output data and the probability density distribution curve of a predefined distribution is less than a certain threshold value), the terminal can indicate a reference distribution to more accurately indicate the distribution of the input or output data. Moreover, even if the terminal cannot give a reference distribution of the input or output data, it can also describe the distribution characteristics of the input or output data from various aspects through information such as reference values, characteristic values, maximum values, and minimum values, so that the network device can have a quantitative understanding of the distribution of the input or output data of the terminal-side AI / ML model or function, making it possible to report auxiliary information of the data distribution.

[0407] After obtaining the distribution characteristic information of the above model or function-related data, the terminal can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0408] A. When the distribution of the input (or output) data of the AI / ML model or function on the terminal side is highly consistent with the reference distribution characteristics indicated by the network device (for example, the probability density distribution curve deviation between the data distribution statistically determined by the terminal and the reference distribution indicated by the network device is less than a certain threshold value), it is decided to activate or switch to the AI / ML model or function on the terminal side; otherwise, the AI / ML model or function on the terminal side is deactivated, or a suggestion or request to deactivate the AI / ML model or function is initiated to the first network device.

[0409] B. When the characteristics of the input (or output) data of the AI / ML model or function on the terminal side are highly consistent with the distribution characteristics of the data indicated by the network device (for example, the mean of the network device environment is between the maximum and minimum values ​​of the input data of the AI / ML model or function on the terminal side, or the mean of the input data of the AI / ML model or function on the terminal side is between the maximum and minimum values ​​of the network device environment), it is decided to activate or switch to the AI / ML model or function on the terminal side; otherwise, the AI / ML model or function on the terminal side is deactivated, or a suggestion or request to deactivate the AI / ML model or function is initiated to the first network device.

[0410] C. When the reference distribution and / or characteristics of the input or output data of the terminal-side AI / ML model or function match the input distribution and output distribution indicated by the network device, decide to activate or switch to the terminal-side AI / ML model or function; otherwise, deactivate the terminal-side AI / ML model or function, or initiate a suggestion or request to the first network device to deactivate the AI / ML model or function.

[0411] D. When only one of the reference distribution and / or characteristics of the input or output data of the terminal-side AI / ML model or function matches the input distribution and output distribution indicated by the network device, it can be decided to activate or switch to the terminal-side AI / ML model or function based on the one of the input data and output data that has a greater impact on the performance. Otherwise, the terminal-side AI / ML model or function is deactivated, or a suggestion or request to deactivate the AI / ML model or function is initiated to the first network device.

[0412] For example, the input data has a greater impact on performance. When the reference distribution and / or characteristics of the output data of the AI / ML model or function on the terminal side match the distribution characteristics of the output data indicated by the network device, the AI / ML model or function on the terminal side is deactivated; when the reference distribution and / or characteristics of the input data of the AI / ML model or function on the terminal side match the distribution characteristics of the input data indicated by the network device, it is decided to activate or switch to the AI / ML model or function on the terminal side.

[0413] In some embodiments, the physical space related information includes one or more of the following:

[0414] (1) Network device identification.

[0415] Specifically, the network device identifier may include one or more of the following:

[0416] a. Physical cell ID, which is carried and indicated by SSB and represents the ID of the cell visible at the physical layer.

[0417] b. Global cell ID, which can clearly indicate a cell in the global range.

[0418] c. An ID specially designed for AI / ML. Different physical cells or global cells can use the same AI / ML cell ID. The same AI / ML cell ID indicates that the terminal-side AI / ML model or function has the same or similar applicability or degree of applicability in these cells.

[0419] The training data collected from cells with the same AI / ML cell ID have similar distribution characteristics.

[0420] The AI / ML model or function trained with the training data collected from any one or more cells having the same AI / ML cell ID is suitable for use in any one of the cells having the same AI / ML cell ID.

[0421] d. An ID that can be determined by other means, such as a cell ID determined by a PLMN number and a specific cellIdentity within the PLMN.

[0422] e. TRP ID.

[0423] (2) The area where the network equipment is located.

[0424] Specifically, the area where the network device is located can be one or more of the following:

[0425] a. Area of ​​predefined representation.

[0426] b. Areas divided by longitude and latitude (for example, each X longitude and Y latitude is divided into one area).

[0427] c. Areas divided by administrative means (country, province, city, district, county, township, street, building, etc.).

[0428] (3) The height of the network equipment. For example, the altitude of the network equipment or the height of the network equipment relative to a certain reference horizontal plane.

[0429] (4) Coverage area of ​​network equipment.

[0430] (5) Physical space relationship information between multiple network devices.

[0431] Specifically, the physical space relationship information between multiple network devices may be an average value, a maximum value, and a minimum value of the inter-station distances between the multiple network devices, an overlapping coverage area of ​​the multiple network devices, and the like.

[0432] (6) Reference scenarios for the areas where network devices are located.

[0433] Specifically, the reference scene of the area where the network device is located can be some predefined scenes, such as urban macro cells, urban micro cells, rural cells, and indoor factory cells, etc. The wireless signal propagation characteristics between different predefined scenes are different, such as different large-scale fading, and / or different small-scale channel characteristics, and / or different probabilities that the propagation path is a direct path, etc.

[0434] When the physical space related information of the network device includes a reference scene of the area where the network device is located, it can be indicated that the environment where the network device is located is the same as or similar to the environment of the reference scene.

[0435] The management method of AI / ML models or functions provided in the embodiments of the present application quantitatively characterizes the scenes of the physical space related to the network devices by using methods such as area division, quantitative description of the height of the network devices, quantitative description of the physical relationship between multiple network devices, and definition of reference scenes, thereby solving the problem that "scenes" are difficult to define in a standardized manner in the standards, and making it possible for network devices to indicate their own related scenes as auxiliary information for AI / ML models or functions.

[0436] After obtaining the above physical space-related information, the network device can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0437] A. When the physical space scene of the network device corresponding to the terminal-side AI / ML model or function is the same as or similar to the physical space scene indicated by the network device, decide to activate or switch to the terminal-side AI / ML model or function, or initiate a suggestion or request to the first network device to activate the AI / ML model or function.

[0438] B. When the physical space scene of the network device corresponding to the terminal-side AI / ML model or function only partially matches or coincides with the physical space scene indicated by the network device, and the amount of matching or coincident information is greater than a certain threshold value, decide to activate or switch to the terminal-side AI / ML model or function, or initiate a suggestion or request to activate the AI / ML model or function to the first network device.

[0439] This approach is relatively simple, but may miss opportunities for edge-side AI / ML models or functions to be used.

[0440] C. When the physical space scene of the network device corresponding to the terminal-side AI / ML model or function only partially matches or coincides with the physical space scene indicated by the network device, the terminal also decides whether to activate or switch to the terminal-side AI / ML model or function based on the impact of the generalization of the AI / ML model or function in the physical space of the network device.

[0441] For example, the positioning based on the AI / ML model has poor generalization for station spacing (that is, the best AI / ML performance can only be obtained when the station spacing matches), but has good generalization for longitude and latitude (that is, the difference in longitude and latitude of network devices has little impact on AI / ML performance). The terminal can decide to activate the AI / ML model when the station spacing of network devices matches (although the longitude and latitude of network devices do not match).

[0442] This method is relatively complex and requires the terminal to determine in which scenarios / configurations / conditions the AI / ML model or function has better generalization performance, but it may maximize the terminal-side AI / ML model or function and maximize network performance.

[0443] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operation administration and maintenance (OAM) pre-configuration information.

[0444] Specifically, the training data set identifier may be determined based on at least one of a predefined rule, a high-level signaling configuration, and OAM pre-configuration information.

[0445] In one implementation, when determining the identification of a training data set based on a predefined rule, it may be agreed in a manner specified by a protocol that a training data set that meets a specific scenario / configuration / condition has a specific identification.

[0446] For example, the ID of a dataset collected in an urban scene is 0, the ID of a dataset collected in a suburban scene is 1, the ID of a dataset collected in an indoor factory is 2, the ID of a mixed dataset collected in an urban + suburban scene is 3, and so on.

[0447] In addition, the training data set identifier can also be determined based on scene-related information that affects wireless communication performance, such as weather information in outdoor scenes. Specifically, the ID of data collected in urban areas + good weather is 11, the ID of data collected in urban areas + rainy weather is 12, the ID of data collected in suburban areas + good weather is 21, and so on.

[0448] In one implementation, when determining the training data set identifier based on high-level signaling configuration, the training data set identifier can be determined based on Radio Resource Control Protocol (RRC) signaling, Long-Term-Evolution Positioning Protocol (LPP) signaling, signaling from the core network, or signaling from the application layer.

[0449] In one implementation, when determining the training data set identifier based on OAM pre-configuration information, the terminal manufacturer and the network equipment manufacturer or operator may agree to use a specific data set ID to represent a specific scenario / configuration / condition, and determine the training data set ID manually.

[0450] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0451] Specifically, the training data set identifier may be determined based on one or more of the information related to signal transmission, the information related to signal reception, the information related to signal quality, the distribution characteristic information of model or function-related data, and the information related to physical space. The determined training data set identifier may include at least one bit field, each bit field being used to indicate one of the information related to signal transmission, the information related to signal reception, the information related to signal quality, the distribution characteristic information of model or function-related data, and the information related to physical space.

[0452] In one implementation, the training data set identifier includes M bit fields, corresponding to M information in signal transmission related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information, and each bit field occupies n1, n2, ...n respectively. M bits, and these M bits form the training data set identifier. The length of the training data set identifier is greater than or equal to n1+n2+…+n M bits, and the value of each bit field corresponds to candidate information in a corresponding indication field.

[0453] For example, five pieces of information including signal transmission related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information correspond to five bit fields respectively, and each bit field occupies n1, n2, ... n5 bits respectively. The five bit fields constitute the training data set identifier, and the identifier length N = n1 + n2 + ... + n M .

[0454] The SNR range in the signal quality related information is divided into 7 regions, so n3=ceil(log2(7))=3 bits are used to represent the 7 regions, and the value of the 3 bits is {000, 001, ...101, 110}, which is one of the 7 SNR regions.

[0455] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 KOne of the information combinations; K is an integer greater than or equal to 1;

[0456] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0457] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0458] Specifically, the training data set identifier can be determined based on different candidate combinations of L information including information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space.

[0459] According to the number of different candidate combinations of L information, the number of bits K contained in the training data set identifier can be determined. K The number of different candidate combinations greater than or equal to L information.

[0460] For example, the training data set identifier is determined according to three pieces of information {network device ID, terminal height, SINR} among the signal transmission related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information, and there are 60 possible candidate combinations of {network device ID, network device RS configuration, SINR} (3 network device IDs, 4 network device RS configurations, 5 SINRs), then K=6 (2 6 =64) bits of training data set identifier represents these 60 combinations. The values ​​of the training data set identifier are {000000, 000001, ..., 111011}, and each value represents one of the 60 combinations.

[0461] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal receiving related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0462] Specifically, the training data set identifier may also include multiple sub-identifiers, each of which may be used to indicate one of the following information: signal sending related information, signal receiving related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0463] For example, the sub-identifier of the training data set is determined according to the three information {network device ID, network device RS configuration, SINR}, and there are three sub-identifiers. The value of the first sub-identifier identifies the network device ID, the value of the second sub-identifier indicates the terminal height, and the value of the third sub-identifier indicates the SINR.

[0464] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0465] In some embodiments, the multiple sub-identifiers included in the training data set identifier may also be determined according to any one of the following methods:

[0466] (1) Determine multiple sub-identifiers included in the training data set identifier based on predefined rules.

[0467] For example, it can be agreed in a manner specified by the protocol that a training dataset that meets a specific scenario / configuration / condition has a specific sub-identifier.

[0468] (2) Determine multiple sub-identifiers included in the training data set identifier based on the high-level signaling configuration.

[0469] For example, the multiple sub-identifiers included in the training data set identifier may be determined according to RRC signaling, LPP signaling, signaling from a core network, or signaling from an application layer.

[0470] (3) Determine the training data set identifier based on the OAM pre-configuration information.

[0471] For example, the terminal manufacturer and the network equipment manufacturer or operator may agree to use a specific dataset sub-ID to represent a specific scenario / configuration / condition, and determine the training dataset sub-ID manually.

[0472] The management method of the AI / ML model or function provided in the embodiment of the present application establishes a one-to-one mapping relationship between the auxiliary information and the training data set identifier (dataset ID), so that the dataset ID can carry the auxiliary information, or through advance negotiation and consensus between the terminal and the network device, assisted by manual configuration or high-level signaling configuration, the terminal and the network device can align the applicable scenarios / configurations / conditions of the AI / ML model or function according to the dataset ID reported by the terminal, so that the dataset ID can characterize the characteristics of the training data in a quantitative manner, so that it becomes possible for the network device to indicate its associated dataset ID as auxiliary information.

[0473] After obtaining the training dataset identifier, the terminal can use this information to manage the terminal-side AI / ML model or function in one or more of the following ways:

[0474] A. Determine the scenario / configuration / condition corresponding to the training data of the terminal-side AI / ML model or function according to the dataset ID, and compare it with the scenario / configuration / condition of the current network device. When some or all of the scenarios / configurations / conditions match, determine that the terminal-side AI / ML model or function can be activated, or initiate a suggestion or request to the first network device to deactivate the AI / ML model or function.

[0475] B. Determine whether the training data set of the terminal-side AI / ML model or function includes a data set associated with the network device. When the training data set of the terminal-side AI / ML model or function includes a data set associated with the network device, determine that the terminal-side AI / ML model or function can be activated, or initiate a suggestion or request to the first network device to deactivate the AI / ML model or function.

[0476] C. The terminal itself saves or maintains relevant information about the "dataset ID corresponding to the applicable environment of the terminal-side AI / ML model or function", such as a dataset ID list or set (but the training of the terminal-side AI / ML model or function does not necessarily use every dataset therein). The list or set can be manually configured or come from high-level information. When the dataset ID indicated by the network device is in the list or set of dataset IDs saved or maintained by the terminal, it is determined that the terminal-side AI / ML model or function can be activated, or a suggestion or request to deactivate the AI / ML model or function is initiated to the first network device.

[0477] D. The terminal collects training data in the current network environment, and the ID of the collected training data set is the dataset ID indicated by the network device, or is a new ID determined based on the dataset ID indicated by the network device.

[0478] Figure 2 The second flowchart of the method for managing the AI / ML model or function provided in the embodiment of the present application is as follows: Figure 2 As shown, the method comprises the following steps:

[0479] Step 200: Send auxiliary information corresponding to the first network device to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function. The sending to the terminal here only means that the execution subject of the method sends the auxiliary information via a wireless signal, and the destination receiving device is the terminal, and the terminal obtains the auxiliary information by receiving the wireless signal.

[0480] Specifically, the execution subject of the method may be the first network device, or other communication devices other than the first network device.

[0481] The first network device may be any network device, and this application does not impose any limitation thereto. For example, the first network device may be a base station, a LMF, or other network device.

[0482] The auxiliary information corresponding to the first network device may be sent by the first network device to the terminal, or may be sent by other communication devices to the terminal. For example, a neighboring cell of the first network device or a base station close to the first network device may send the auxiliary information corresponding to the first network device to the terminal.

[0483] It is understandable that the first network device can send the auxiliary information corresponding to itself to the terminal, or can send the auxiliary information corresponding to other network devices to the terminal for other network devices. For example, the first network device can send the auxiliary information corresponding to its neighboring area to the terminal.

[0484] The auxiliary information corresponding to the first network device can be used to indicate one or more of the current network scenario (e.g., the scenario and height where the network device is located), configuration (e.g., reference signal configuration information), and conditions (e.g., data distribution characteristics). The auxiliary information corresponding to the first network device can be the auxiliary information of the first network device itself, or it can be the auxiliary information of a network device with a wireless network environment similar to that of the first network device, or it can be the auxiliary information of a reference network device with a wireless network environment similar to that of the first network device. The management method of AI / ML models or functions provided in an embodiment of the present application sends its own corresponding auxiliary information to the terminal, so that the terminal can perform accurate and appropriate lifecycle management of the AI / ML model or function, which helps to improve the transmission performance of the terminal and the overall performance of the network, and avoid unnecessary AI / ML calculation and storage overhead.

[0485] In some embodiments, the auxiliary information includes one or more of the following:

[0486] Signaling related information;

[0487] Information related to signal reception;

[0488] Information related to signal quality;

[0489] Distribution characteristics of model or function-related data;

[0490] Information about the physical space;

[0491] The identifier of the training dataset.

[0492] In some embodiments, the signal transmission related information includes one or more of the following:

[0493] Signaling configuration information;

[0494] Send antenna information.

[0495] In some embodiments, the signal reception related information includes one or more of the following:

[0496] Signal receiving configuration information;

[0497] Receive antenna information.

[0498] In some embodiments, the signal quality related information includes one or more of the following:

[0499] Signal quality metrics;

[0500] Signal quality characteristic value;

[0501] The source of the signal quality characteristic value.

[0502] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0503] Signal information corresponding to the terminal;

[0504] Empirical signal information;

[0505] Signal information corresponding to the reference terminal.

[0506] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0507] Information about the distribution characteristics of model or function input data; and / or,

[0508] The distribution characteristics of the model or function output data.

[0509] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0510] The reference distribution of the model or function input data;

[0511] A value that characterizes the distribution of input data to a model or function.

[0512] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0513] The reference distribution of model or function output data;

[0514] A value that characterizes the distribution of the output data of a model or function.

[0515] In some embodiments, the physical space related information includes one or more of the following:

[0516] Network device identification;

[0517] The region where the network device is located;

[0518] Height of network equipment;

[0519] The coverage area of ​​the network equipment;

[0520] Physical space relationship information between multiple network devices;

[0521] Reference scenario for the area where the network equipment is located.

[0522] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operations administration and maintenance (OAM) pre-configuration information.

[0523] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0524] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0525] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0526] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0527] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0528] The methods provided in the embodiments of the present application are based on the same application concept, so the implementation of each method can refer to each other, and the repeated parts will not be repeated.

[0529] The following describes the methods provided in the above embodiments of the present application by using examples of specific application scenarios.

[0530] Example 1: Applicable scenarios for management methods of AI / ML models or functions.

[0531] Figure 3 An example diagram of an application scenario of the management method of the AI / ML model or function provided in the embodiment of the present application, such as Figure 3 As shown, multiple UEs including UE1 and UE2 apply for wireless network connection services, and gNB provides wireless services for them. Data is exchanged and transmitted between gNB and UE1 and UE2 through wireless communication. In addition, the third-party server OTT server and gNB can be connected and exchange information through wired mode, or interact with gNB through UE.

[0532] The following embodiments take UE as an example, but can actually also be used for a third-party server (OTT server) or other entities that want to perform model recognition.

[0533] Example 2: NW sends auxiliary information related to the AI / ML model or function on the UE side to the UE.

[0534] The auxiliary information related to the AI / ML model or function on the UE side sent by the NW to the UE includes one or more of the following:

[0535] (1) Information related to NW signal transmission.

[0536] For NW transmission, UE reception or measurement, the information related to signal transmission may include one or more of the following:

[0537] a. NW sends configuration related information.

[0538] a1. RS configuration information, one or more of the following: signal type (e.g. CSI-RS, SSB, PRS, etc.), number, time domain resources, frequency domain resources, number of ports, sequence, RS resource ID, RS resource set ID, RS ID, etc.

[0539] b. Information related to the NW transmitting antenna, such as one or more of the following: the NW Tx transmitting channel (panel), antenna array pattern, vertical angle of the transmitting beam, horizontal angle of the transmitting beam, transmitting power, relative power of each angle between multiple transmitting beams, etc.

[0540] Through this information, the NW can instruct the NW side to send relevant information to assist the UE in performing appropriate AI / ML model or function management. After the UE obtains the auxiliary information, it uses the auxiliary information to manage one or more of the following models / functions:

[0541] A. Manage through a simple comparison method, for example, only when the applicable configuration of the AI / ML model or function on the terminal side is the same as the configuration indicated by the network device and / or the relevant information of the transmitting antenna is the same, decide to activate the AI / ML model or function on the terminal side, or switch to the AI / ML model or function with matching configuration / antenna.

[0542] This approach is simple to implement, but may miss opportunities for the UE's AI / ML models / functions to be utilized.

[0543] B. Extract the wireless communication environment characteristics represented by the sending configuration and sending antenna indicated by the network device, and compare them with the wireless communication environment characteristics applicable to the terminal-side AI / ML model or function. When the environmental characteristics are the same / similar, decide to activate the terminal-side AI / ML model or function, or switch to the terminal-side AI / ML model or function with matching configuration / antenna.

[0544] Environmental characteristics may be, for example, antenna horizontal coverage, vertical coverage, obstacle height and density, whether the base station is placed indoors or outdoors, the maximum coherence time or minimum coherence bandwidth of the detectable channel, etc.

[0545] For example, the terminal-side AI / ML model or function is used for mobility management. The terminal can determine the horizontal antenna coverage corresponding to the network device through the transmitting antenna-related information indicated by the network device, and compare it with the horizontal antenna coverage suitable for the terminal-side AI / ML model or function. If the two are found to be similar, even though the antenna configuration indicated by the network device is different from the antenna configuration of the network device during the training of the terminal-side AI / ML model or function, it is still decided to activate the terminal-side AI / ML model or function.

[0546] This method is more complicated to implement, but it can maximize the reasoning capabilities of the AI / ML model or function on the terminal side and improve network performance.

[0547] (2) Information related to NW signal reception.

[0548] For UE transmission, NW reception or measurement, the information related to signal reception may include one or more of the following:

[0549] a. NW receives configuration related information.

[0550] a1. RS configuration information, including one or more of the following: signal type (such as SRS, etc.), number, time domain resources, frequency domain resources, number of ports, sequence, RS resource ID, RS resource set ID, RS ID, etc.

[0551] b. Information related to the NW receiving antenna, such as one or more of the following: NW Rx receiving channel (panel), antenna array pattern, vertical angle of receiving beam, horizontal angle of receiving beam, receiving power, relative power of each angle between multiple receiving beams, etc.

[0552] Through this information, the NW can instruct the NW side to receive relevant information to assist the UE in performing appropriate AI / ML model or function management. After the UE obtains the auxiliary information, it uses the auxiliary information to manage one or more of the following models / functions:

[0553] A. Manage through simple comparison methods, for example, only when the applicable configuration of the AI / ML model or function on the terminal side is the same as the configuration indicated by the network device, and / or the relevant information of the receiving antenna is the same, decide to activate the AI / ML model or function on the terminal side, or switch to the AI / ML model or function with matching configuration / antenna.

[0554] This approach is simple to implement, but may miss opportunities for the UE's AI / ML models / functions to be utilized.

[0555] B. Extract the wireless communication environment characteristics represented by the receiving configuration and receiving antenna indicated by the network device, and compare them with the wireless communication environment characteristics applicable to the terminal-side AI / ML model or function. When the environmental characteristics are the same / similar, decide to activate the terminal-side AI / ML model or function, or switch to the terminal-side AI / ML model or function with matching configuration / antenna.

[0556] Environmental characteristics may be, for example, antenna horizontal coverage, vertical coverage, obstacle height and density, whether the base station is placed indoors or outdoors, the maximum coherence time or minimum coherence bandwidth of the detectable channel, etc.

[0557] For example, the terminal-side AI / ML model or function is used for mobility management. The terminal can determine the horizontal antenna coverage corresponding to the network device through the receiving antenna-related information indicated by the network device, and compare it with the horizontal antenna coverage suitable for the terminal-side AI / ML model or function. If the two are found to be similar, even though the antenna configuration indicated by the network device is different from the antenna configuration of the network device when the terminal-side AI / ML model or function was trained, it is still decided to activate the terminal-side AI / ML model or function.

[0558] This method is more complicated to implement, but it can maximize the reasoning capabilities of the AI / ML model or function on the terminal side and improve network performance.

[0559] (3) Information related to signal quality

[0560] a. Information related to the signal quality metric.

[0561] a1. The signal quality metric may be one or more of SNR, SINR, RSRP, and RSRQ. In addition, there may be other signal quality metric methods, which are not specifically limited here.

[0562] a2. The relevant information may be one or more of the following: reference value, maximum value, minimum value, mode, median, mean, standard deviation, variance, etc. In addition, there may be other characteristic values ​​of signal quality, which are not specifically limited here.

[0563] The reference value may be, for example, the most appropriate value, such as the most appropriate SINR, the most appropriate RSRP, or a value that enables the AI / ML model or function to achieve a specific performance indicator.

[0564] The maximum, minimum, mode, median, mean, standard deviation, and variance are used to indicate the most suitable range, such as the most suitable SINR range, the most suitable RSRQ range, or the range that enables the AI / ML model or function to achieve a specific performance indicator.

[0565] If a plurality of signal quality metrics are indicated, relevant information (reference value, maximum value, etc.) of each signal quality metric may be indicated separately.

[0566] b. The source of the information related to signal quality may be one or more of the following:

[0567] b1. NW receives the signal of the UE, such as PRACH, PUSCH, PUCCH, DMRS, SRS, etc. sent by the UE.

[0568] b2. NW's empirical signal information, such as the reporting feedback from other UEs in the NW after measuring the signal sent by the NW, or the quality of the signal sent by the NW from other UEs, or the signal quality of the virtual terminal after the NW receives the signal quality from multiple terminals, using the signal quality of the virtual terminal to represent the signal quality of multiple terminals.

[0569] b3. Reporting of the received signal quality of a signal of a reference UE in the NW, or the quality of a transmitted signal received by the NW from the reference UE.

[0570] Through this information, the NW can indicate information related to signal quality to assist the UE in managing appropriate AI / ML models or functions. After the UE obtains the auxiliary information, it uses the auxiliary information to manage one or more models or functions as follows:

[0571] A. Compare the reference value of the signal quality indicated by the NW with the typical value of the signal quality applicable to the UE's AI / ML model or function (such as the mean, mode, or expected value during network planning). When the difference between the two is less than a certain threshold, decide to activate / switch to the UE's AI / ML model or function; otherwise, deactivate the UE's AI / ML model or function, or switch to a more suitable UE AI / ML model or function.

[0572] B. Compare the appropriate range of signal quality indicated by the NW with the typical value of signal quality applicable to the AI / ML model or function of the UE. If the typical value of signal quality applicable to the AI / ML model or function is within the appropriate range indicated by the NW, decide to activate / switch to the AI / ML model or function of the UE; otherwise, deactivate the AI / ML model or function of the UE, or switch to a more suitable UE AI / ML model or function.

[0573] C. Compare the appropriate range of signal quality indicated by NW with the range of signal quality applicable to the AI / ML model or function of the UE. If the difference between the two (such as the sum of the difference between the maximum value and the minimum value) is less than a certain threshold, decide to activate / switch to the AI / ML model or function of the UE. Otherwise, deactivate the AI / ML model or function of the UE, or switch to a more suitable UE AI / ML model or function.

[0574] (4) Distribution characteristic information of input data or output data corresponding to AI / ML models or functions.

[0575] a. The distribution characteristic information of the input data includes one or more of the following:

[0576] a1. The reference distribution of the input data may be one of the following predefined distributions: uniform distribution, Gaussian distribution (ie, normal distribution), truncated Gaussian distribution, exponential distribution, chi-square distribution, etc. In addition, there may be other predefined distributions for reference, which are not specifically limited here.

[0577] a2. Information related to the distribution characteristics of the input data, which may be one or more of the following: reference value, characteristic value, maximum value, minimum value, mode, median, mean, standard deviation, variance, etc. In addition, there may be other characterization values, which are not specifically limited here.

[0578] Reference values ​​and characteristic values ​​can be special reference values ​​in a certain distribution, such as the exponential value in an exponential distribution.

[0579] Maximum, minimum, mode, median, mean, standard deviation, variance, etc. are used to indicate the most appropriate range or the range that enables the AI / ML model or function to achieve a specific performance indicator.

[0580] b. The distribution characteristic information of the output data includes one or more of the following:

[0581] b1. The reference distribution of the output data may be one of the following predefined distributions: uniform distribution, Gaussian distribution (ie, normal distribution), truncated Gaussian distribution, exponential distribution, chi-square distribution, etc. In addition, there may be other predefined distributions for reference, which are not specifically limited here.

[0582] b2. Information related to the distribution characteristics of the output data, which may be one or more of the following: reference value, characteristic value, maximum value, minimum value, mode, median, mean, standard deviation, variance, etc. In addition, there may be other characterization values, which are not specifically limited here.

[0583] Reference values ​​and characteristic values ​​can be special reference values ​​in a certain distribution, such as the exponential value in an exponential distribution.

[0584] Maximum, minimum, mode, median, mean, standard deviation, variance, etc. are used to indicate the most appropriate range or the range that enables the AI / ML model or function to achieve a specific performance indicator.

[0585] It should be emphasized that the design of the above-mentioned auxiliary information solves the problem that the input distribution characteristics cannot be characterized as auxiliary information. When the distribution of the input data of the model or function is highly consistent with a specific predefined distribution (such as the difference between the statistical distribution curve of the input data and the probability density distribution curve of a predefined distribution is less than a certain threshold value), the NW can indicate a reference distribution to more accurately indicate the distribution of the input data. Moreover, even if the NW cannot give a reference distribution, it can describe the distribution characteristics of the input data from various aspects through reference values, characteristic values, maximum values, minimum values ​​and other information, so that the UE can have a quantitative understanding of the distribution of the input or output data corresponding to the UE model / function, making it possible to report auxiliary information on the data distribution.

[0586] Through this information, the NW can indicate the distribution characteristics of the input and / or output of the training data of the UE's AI / ML model or function to assist the UE in performing appropriate AI / ML model or function management. After the NW obtains the auxiliary information, it uses the auxiliary information to manage one or more of the following models / functions:

[0587] A. When the reference distribution of the input (or output) data of the UE model or function is highly consistent with the reference distribution characteristics indicated by the NW (for example, the deviation of the probability density distribution curve between the data distribution statistically obtained by the UE and the reference distribution indicated by the NW is less than a certain threshold value), it is decided to activate / switch to the UE's AI / ML model or function, otherwise the UE's AI / ML model or function is deactivated.

[0588] B. When the data characteristics of the input (or output) of the UE model or function are highly consistent with the distribution characteristics of the data indicated by the NW (for example, the mean of the NW environment is between the maximum and minimum values ​​of the UE's input data, or the mean of the UE's input data is between the maximum and minimum values ​​of the NW environment), it is decided to activate / switch to the UE's AI / ML model or function, otherwise, deactivate the UE's AI / ML model or function.

[0589] C. Based on management mode A or B, when the input data reference distribution and output data reference distribution of the UE match the input distribution and output distribution indicated by the NW, decide to activate / switch to the AI / ML model or function of the UE, otherwise deactivate the AI / ML model or function of the UE.

[0590] D. Based on management mode A or B, when only one of the input data reference distribution or the output data reference distribution of the UE can match the input distribution and output distribution indicated by the NW, decide to activate / switch to the AI / ML model or function of the UE according to the one that has a greater impact on performance, otherwise deactivate the AI / ML model or function of the UE.

[0591] (5) Information related to the physical space corresponding to the AI / ML model or function.

[0592] a. Base station ID. The base station ID can be one or more of the following:

[0593] a1. Physical cell ID, which is carried and indicated by SSB and represents the ID of the cell visible at the physical layer.

[0594] a2. Global cell ID, which can clearly indicate a cell in the global range.

[0595] a3. ID specially designed for AI / ML. Different physical cells / global cells can use the same AI / ML cell ID. The same AI / ML cell ID indicates that the AI / ML model or function of the UE has the same or similar applicability or degree of applicability in these cells.

[0596] The training data collected from cells with the same AI / ML cell ID have similar distribution characteristics.

[0597] The AI / ML model / function trained with the training data collected from any one or more cells having the same AI / ML cell ID is suitable for use in any one of the cells having the same AI / ML cell ID.

[0598] a4. ID that can be determined by other methods, such as the cell ID determined by the PLMN number and the cellIdentity specified in the PLMN.

[0599] a5. TRP ID.

[0600] b. Information related to the physical space of the base station.

[0601] b1. The area where the base station is located can be one or more of the following:

[0602] Areas of predefined representation;

[0603] Areas divided by longitude and latitude (e.g., each X longitude and Y latitude is divided into one area);

[0604] An area divided by administrative means (country, province, city, district, county, township, street, building, etc.).

[0605] b2. The height of the base station.

[0606] b3. The physical space related information between multiple base stations may be one or more of the following:

[0607] The average, maximum, and minimum values ​​of the inter-station distances of multiple stations;

[0608] The coverage area of ​​the plurality of stations (the coverage area of ​​each of the plurality of stations and / or the joint coverage area of ​​the plurality of stations).

[0609] b4. Reference scene of the area where NW is located.

[0610] The reference scenario may be some predefined scenarios, such as urban macrocell, urban microcell, rural cell, indoor factory cell. The wireless signal propagation characteristics between different predefined scenarios are different, such as different large-scale fading, and / or different small-scale channel characteristics, and / or different probabilities that the propagation path is a direct path, etc.

[0611] If the information indicates a reference scene, it means that the environment where the NW is located is the same as or similar to the environment of the reference scene.

[0612] It should be emphasized that the above-mentioned auxiliary information quantitatively describes the scenes of the physical space related to the base station by using methods such as regional division, quantitative description of the height of the base station, quantitative description of the physical relationship between multiple base stations, and definition of reference scenes. This solves the problem that "scenes" are difficult to define in a standardized manner in the standard, making it possible for NW to indicate its own related scenes as auxiliary information for AI / ML models or functions.

[0613] Through this information, information related to the physical space of the NW can be indicated to assist the UE in performing appropriate AI / ML model or function management.

[0614] A. When the NW physical space scene corresponding to the UE's AI / ML model or function is the same as or similar to the physical space scene indicated by the NW, decide to activate or switch to the UE's AI / ML model or function.

[0615] B. When the NW physical space scene corresponding to the UE's AI / ML model or function only partially matches or coincides with the physical space scene indicated by the NW, and the amount of matching or coincident information is greater than a certain threshold value, it is decided to activate or switch to the terminal-side AI / ML model or function.

[0616] This approach is relatively simple, but may miss opportunities for edge-side AI / ML models or functions to be used.

[0617] C. When the NW physical space scene corresponding to the UE's AI / ML model or function only partially matches or coincides with the physical space scene indicated by the NW, the UE also decides whether to activate or switch to the terminal-side AI / ML model or function based on the impact of the generalization of the AI / ML model or function in the NW physical space.

[0618] For example, the positioning based on the AI / ML model has poor generalization for station spacing (that is, the best AI / ML performance can only be obtained when the station spacing matches), but has good generalization for longitude and latitude (that is, the longitude and latitude differences of NW have little impact on AI / ML performance). The terminal can decide to activate the AI / ML model when the NW station spacing matches (although the NW longitude and latitude do not match).

[0619] This method is relatively complex and requires the UE to determine in which scenarios / configurations / conditions the AI / ML model or function has better generalization performance, but it may maximize the terminal-side AI / ML model or function and maximize network performance.

[0620] In this example, the NW provides the UE with one or more of the above-mentioned auxiliary information to assist the UE in determining whether the AI / ML model or function on the UE side is applicable in the current scenario / configuration / condition.

[0621] After receiving the auxiliary information sent by the NW, the UE can reasonably manage the AI / ML model / functionality on the UE side based on the auxiliary information. For example, the UE can:

[0622] A. Compare one or more applicable scenarios / configurations / conditions corresponding to the UE's AI / ML model or function with the scenarios / configurations / conditions indicated by the NW. When all or part of the scenarios match, activate the UE's AI / ML model or function.

[0623] B. Compare one or more applicable scenarios / configurations / conditions corresponding to the UE's AI / ML model or function with the scenarios / configurations / conditions indicated by the NW. When the scenarios / configurations / conditions that "significantly affect generalization performance" can be matched, activate the UE's AI / ML model or function; otherwise, deactivate the UE's AI / ML model or function.

[0624] C. Compare one or more applicable scenarios / configurations / conditions corresponding to multiple AI / ML models or functions of the UE, select the AI / ML model or function that best matches the scenario / configuration / condition indicated by the NW for activation, or switch to the AI / ML model or function that best matches the current configuration.

[0625] Example 3: The auxiliary information sent by the NW to the UE is the ID of one or more training data sets associated with the NW.

[0626] NW indicates the ID of the training data set, indicating that an NW is associated with the training data set.

[0627] The association relationship between the NW and the training data set may be that the NW participated in the collection of the data set, such as the NW participated in the signal transmission of the data sample, or participated in the signal reception of the data sample. Alternatively, although the NW did not directly participate in the collection of the data set, the environment / configuration / condition of the NW that participated in the collection of the data set is the same or similar to the environment / configuration / condition of the NW.

[0628] It should be noted that in the above method, by defining the association relationship between NW and training data set (and training data set ID), NW and training data set do not need to be tightly coupled, that is, not only the training data set collected in the current NW is associated with the NW, as long as the NW environment / configuration / conditions when the data is collected are the same or similar, the NW can be considered to be associated with the training data set. This design allows the AI / ML model / function of the UE trained with one training data set to be applicable in multiple associated NWs, effectively expanding the scope of use of the UE's AI / ML model / function.

[0629] One NW may indicate a data set ID associated with the NW, or may indicate a data set ID associated with another NW. For example, a cell may indicate a data set ID associated with a neighboring cell.

[0630] NW may indicate the ID of one training dataset, or may indicate the IDs of multiple training datasets.

[0631] For each indicated training dataset:

[0632] (1) The ID of a training data set can be an ID, which can be determined as follows:

[0633] An ID determined in a predefined manner, or an ID configured by high-level signaling, or an ID configured manually (OAM mode).

[0634] a. ID confirmed in a predefined manner. For example, it can be agreed in a manner specified by the protocol that a training data set that meets a specific scenario / configuration / condition has a specific ID, such as the ID of a data set collected in an urban scenario is 0, the ID of a data set collected in a suburban scenario is 1, the ID of a data set collected in an indoor factory is 2, the ID of a mixed data set collected in an urban + suburban area is 3, and so on.

[0635] In addition, the ID can also be determined based on scene-related information that affects wireless communication performance, such as weather information in outdoor scenes. Specifically, the ID of data collected in urban areas + good weather is 11, the ID of data collected in urban areas + rainy weather is 12, the ID of data collected in suburban areas + good weather is 21, and so on.

[0636] b. ID configured by high-level signaling, such as RRC signaling, LPP signaling, or signaling from the core network, or ID configured by signaling from the application layer.

[0637] c. ID configured manually (OAM mode): For example, the UE manufacturer and the NW manufacturer / operator may agree to use a specific data set ID to represent a specific scenario / configuration / condition, and the ID may be manually configured.

[0638] d. A data set ID determined based on one or more of information including signal transmission related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0639] d1, L information among the information related to transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space, respectively correspond to L bit fields, and each bit field occupies k1, k2, ..., k L bits, the L bit field constitutes the ID of the training data set, and the ID length K is greater than or equal to k1+k2+…+k L Bit; the value of each bit field corresponds to candidate information of a corresponding indication field.

[0640] For example, five pieces of information, including information related to transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function-related data, and information related to physical space, correspond to five bit fields, each of which occupies k1, k2, ..., k5 bits. The five bit fields constitute the training data set ID, and the ID length is K = k1, k2, ..., k5. Among them, the SNR range in the signal quality-related information is divided into seven regions, so k3 = ceil (log2 (7)) = 3 bits are used to represent the seven regions, and the 3-bit value {000, 001, ... 101, 110} represents one of the seven SNR regions.

[0641] d2. The combination of L information among the sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information is mapped into an ID of length K, where K satisfies 2 K The minimum integer value of the number of candidates for a combination of L information that is greater than or equal to L; the value of this ID corresponds to a candidate for a combination of information.

[0642] For example, there are three pieces of information {base station ID, base station RS configuration, SINR} in the transmission related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information to determine the data set ID, and there are 60 possible candidate combinations of {base station ID, base station RS configuration, SINR}, so K=6 (2 6 =64) bits of ID represent the 60 combinations, and the value of ID {000000, 000001, ..., 111011} represents one of the 60 combinations.

[0643] (2) The ID of a training data set is multiple (sub)IDs, and the multiple (sub)IDs are determined as follows:

[0644] a. Each (sub) ID corresponds to one of the information related to sending, related to signal reception, related to signal quality, distribution characteristic information of model or function-related data, and related to physical space. L (sub) IDs correspond to L information, respectively indicating a candidate item of one of the information related to sending, related to signal reception, related to signal quality, distribution characteristic information of model or function-related data, and related to physical space.

[0645] For example, there are three pieces of information {base station ID, UE height, SINR} in the information related to transmission, signal reception, signal quality, distribution characteristics of model or function related data, and physical space related information to determine the data set sub-ID. Then the data set has three (sub) IDs. The value of the first (sub) ID represents the base station ID, the value of the second (sub) ID represents the UE height, and the value of the third (sub) ID represents the SINR

[0646] b. Multiple IDs determined in a predefined manner, or multiple IDs configured by high-level signaling, or multiple IDs configured manually (OAM mode).

[0647] It should be emphasized that in this embodiment, by establishing a one-to-one mapping relationship between the auxiliary information and the dataset ID, so that the dataset ID carries the auxiliary information, or by pre-negotiation between the UE and the NW, assisted by manual configuration / high-level signaling configuration, the NW and the UE can align the understanding of the applicable scenarios / configurations / conditions of the AI / ML model or function according to the dataset ID indicated by the NW, so that the dataset ID can characterize the characteristics of the training data in a quantitative expression, so it becomes possible for the NW to indicate its associated dataset ID as auxiliary information.

[0648] In this example, the UE can determine the matching between the characteristics of the training data set of the UE's AI / ML model or function and the NW's associated data set based on the NW-associated training data set ID indicated by the NW, thereby reasonably managing the AI / ML model or function on the UE side. After the UE obtains the auxiliary information, it uses the auxiliary information to manage one or more of the following models / functions:

[0649] A. According to the dataset ID, determine the scenario / configuration / condition corresponding to the training data of the UE's AI / ML model or function, and compare it with the current NW's scenario / configuration / condition. When one or more or all of the scenarios / configurations / conditions match, determine that the UE's AI / ML model or function can be activated.

[0650] B. Determine whether the training data set of the UE's AI / ML model or function includes NW-related data collection. When the training data set of the UE's AI / ML model or function includes NW-related data collection, determine that the UE's AI / ML model or function can be activated.

[0651] C. The UE itself saves / maintains relevant information of "dataset IDs suitable for AI / ML models or functions used in the current NW", such as a list or set (but the training of the UE's AI / ML model or function does not necessarily use every dataset in it). The list or set can be manually configured or come from high-level information; when the dataset ID indicated by the NW is in the list or set of dataset IDs saved / maintained by the UE, it is determined that the UE's AI / ML model or function can be activated.

[0652] D. The terminal collects training data in the current NW environment, and the ID of the collected training data set is the dataset ID indicated by the NW, or a new ID determined based on the dataset ID indicated by the NW.

[0653] Example 4: NW sends auxiliary information and training dataset ID related to the AI / ML model or function on the UE side to the UE.

[0654] The training data set ID carries / indicates K pieces of information including sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information, but does not carry / indicate the other J pieces of information; the NW can send the "training data set ID" and "the other J pieces of auxiliary information" to the NW, so that the UE can determine the match between the characteristics of the training data set of the UE's AI / ML model or function and the environment / configuration / conditions of the current cell based on the "training data set ID and the other J pieces of auxiliary information", thereby reasonably managing the AI / ML model or function on the UE side.

[0655] For example, the training data set ID indicates / carries sending-related information and signal reception-related information, but does not carry / indicate information related to signal quality, distribution characteristic information of model or function-related data, and information related to physical space. In this case, the NW can send "information related to the training data set ID and information related to signal quality, distribution characteristic information of model or function-related data, and information related to physical space" to the UE.

[0656] The management method of the AI / ML model or function provided in the embodiment of the present application can quantitatively characterize information such as the environment, site, and data set, making it possible to use them as auxiliary information. The NW can send auxiliary information to the UE to assist the UE in accurately and appropriately managing the lifecycle of the AI / ML model or function on the UE side according to the matching of the AI / ML model or function on the UE side with the current cell environment / configuration / conditions, thereby helping to improve the transmission performance of the UE and the overall performance of the network, and avoiding unnecessary AI / ML computing / storage overhead.

[0657] The methods and devices provided in the various embodiments of the present application are based on the same application concept. Since the methods and devices solve problems based on similar principles, the implementation of the devices and methods can refer to each other, and the repeated parts will not be repeated.

[0658] Figure 4 A schematic diagram of the structure of a terminal provided in an embodiment of the present application, such as Figure 4 As shown, the terminal includes a memory 420, a transceiver 410 and a processor 400; wherein the processor 400 and the memory 420 may also be arranged physically separately.

[0659] The memory 420 is used to store computer programs; the transceiver 410 is used to send and receive data under the control of the processor 400.

[0660] Specifically, the transceiver 410 is used to receive and send data under the control of the processor 400 .

[0661] Among them, Figure 4 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 400 and memory represented by memory 420. The bus architecture may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this application. The bus interface provides an interface. The transceiver 410 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which may include a wireless channel, a wired channel, an optical cable, and other transmission media.

[0662] The processor 400 is responsible for managing the bus architecture and general processing, and the memory 420 can store data used by the processor 400 when performing operations.

[0663] The processor 400 may 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 may also adopt a multi-core architecture.

[0664] The processor 400 calls the computer program stored in the memory 420 to execute any of the methods provided in the embodiments of the present application according to the obtained executable instructions, for example: receiving auxiliary information corresponding to the first network device, and the auxiliary information is used for the terminal to manage the artificial intelligence or machine learning AI / ML model or function.

[0665] In some embodiments, the auxiliary information includes one or more of the following:

[0666] Signaling related information;

[0667] Information related to signal reception;

[0668] Information related to signal quality;

[0669] Distribution characteristics of model or function-related data;

[0670] Information about the physical space;

[0671] The identifier of the training dataset.

[0672] In some embodiments, the signal transmission related information includes one or more of the following:

[0673] Signaling configuration information;

[0674] Send antenna information.

[0675] In some embodiments, the signal reception related information includes one or more of the following:

[0676] Signal receiving configuration information;

[0677] Receive antenna information.

[0678] In some embodiments, the signal quality related information includes one or more of the following:

[0679] Signal quality metrics;

[0680] Signal quality characteristic value;

[0681] The source of the signal quality characteristic value.

[0682] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0683] Signal information corresponding to the terminal;

[0684] Empirical signal information;

[0685] Signal information corresponding to the reference terminal.

[0686] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0687] Information about the distribution characteristics of model or function input data; and / or,

[0688] The distribution characteristics of the model or function output data.

[0689] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0690] The reference distribution of the model or function input data;

[0691] A value that characterizes the distribution of input data to a model or function.

[0692] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0693] The reference distribution of model or function output data;

[0694] A value that characterizes the distribution of the output data of a model or function.

[0695] In some embodiments, the physical space related information includes one or more of the following:

[0696] Network device identification;

[0697] The region where the network device is located;

[0698] Height of network equipment;

[0699] The coverage area of ​​the network equipment;

[0700] Physical space relationship information between multiple network devices;

[0701] Reference scenario for the area where the network equipment is located.

[0702] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operations administration and maintenance (OAM) pre-configuration information.

[0703] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0704] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0705] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0706] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0707] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0708] Figure 5 A schematic diagram of the structure of a communication device provided in an embodiment of the present application, such as Figure 5 As shown, the communication device includes a memory 520, a transceiver 510 and a processor 500; wherein the processor 500 and the memory 520 may also be arranged physically separately.

[0709] The memory 520 is used to store computer programs; the transceiver 510 is used to send and receive data under the control of the processor 500.

[0710] Specifically, the transceiver 510 is used to receive and send data under the control of the processor 500 .

[0711] Among them, Figure 5 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 500 and various circuits of memory represented by memory 520 are linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described in this application. The bus interface provides an interface. The transceiver 510 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which transmission medium includes a wireless channel, a wired channel, an optical cable, and other transmission media. For different user devices, the user interface 530 may also be an interface capable of externally and internally connecting required devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.

[0712] The processor 500 is responsible for managing the bus architecture and general processing, and the memory 520 can store data used by the processor 500 when performing operations.

[0713] The processor 500 may be a CPU, an ASIC, an FPGA or a CPLD, and the processor may also adopt a multi-core architecture.

[0714] The processor 500 calls the computer program stored in the memory 520 to execute any of the methods provided in the embodiments of the present application according to the obtained executable instructions, for example: sending auxiliary information corresponding to the first network device to the terminal, and the auxiliary information is used for the terminal to manage artificial intelligence or machine learning AI / ML models or functions.

[0715] In some embodiments, the auxiliary information includes one or more of the following:

[0716] Signaling related information;

[0717] Information related to signal reception;

[0718] Information related to signal quality;

[0719] Distribution characteristics of model or function-related data;

[0720] Information about the physical space;

[0721] The identifier of the training dataset.

[0722] In some embodiments, the signal transmission related information includes one or more of the following:

[0723] Signaling configuration information;

[0724] Send antenna information.

[0725] In some embodiments, the signal reception related information includes one or more of the following:

[0726] Signal receiving configuration information;

[0727] Receive antenna information.

[0728] In some embodiments, the signal quality related information includes one or more of the following:

[0729] Signal quality metrics;

[0730] Signal quality characteristic value;

[0731] The source of the signal quality characteristic value.

[0732] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0733] Signal information corresponding to the terminal;

[0734] Empirical signal information;

[0735] Signal information corresponding to the reference terminal.

[0736] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0737] Information about the distribution characteristics of model or function input data; and / or,

[0738] The distribution characteristics of the model or function output data.

[0739] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0740] The reference distribution of the model or function input data;

[0741] A value that characterizes the distribution of input data to a model or function.

[0742] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0743] The reference distribution of model or function output data;

[0744] A value that characterizes the distribution of the output data of a model or function.

[0745] In some embodiments, the physical space related information includes one or more of the following:

[0746] Network device identification;

[0747] The region where the network device is located;

[0748] Height of network equipment;

[0749] The coverage area of ​​the network equipment;

[0750] Physical space relationship information between multiple network devices;

[0751] Reference scenario for the area where the network equipment is located.

[0752] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operations administration and maintenance (OAM) pre-configuration information.

[0753] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0754] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0755] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0756] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0757] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0758] It should be noted here that the above-mentioned terminal and communication equipment provided in the embodiments of the present application can implement all the method steps implemented in the above-mentioned method embodiments, and can achieve the same technical effects. The parts and beneficial effects of this embodiment that are the same as those in the method embodiments will not be described in detail here.

[0759] Figure 6 A schematic diagram of the structure of a management device for an AI / ML model or function provided in an embodiment of the present application, such as Figure 6 As shown, the device comprises:

[0760] The receiving unit 600 is used to receive auxiliary information corresponding to the first network device, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

[0761] In some embodiments, the auxiliary information includes one or more of the following:

[0762] Signaling related information;

[0763] Information related to signal reception;

[0764] Information related to signal quality;

[0765] Distribution characteristics of model or function-related data;

[0766] Information about the physical space;

[0767] The identifier of the training dataset.

[0768] In some embodiments, the signal transmission related information includes one or more of the following:

[0769] Signaling configuration information;

[0770] Send antenna information.

[0771] In some embodiments, the signal reception related information includes one or more of the following:

[0772] Signal receiving configuration information;

[0773] Receive antenna information.

[0774] In some embodiments, the signal quality related information includes one or more of the following:

[0775] Signal quality metrics;

[0776] Signal quality characteristic value;

[0777] The source of the signal quality characteristic value.

[0778] In some embodiments, the source of the signal quality characteristic value includes one or more of the following:

[0779] Signal information corresponding to the terminal;

[0780] Empirical signal information;

[0781] Signal information corresponding to the reference terminal.

[0782] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0783] Information about the distribution characteristics of model or function input data; and / or,

[0784] The distribution characteristics of the model or function output data.

[0785] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0786] The reference distribution of the model or function input data;

[0787] A value that characterizes the distribution of input data to a model or function.

[0788] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0789] The reference distribution of model or function output data;

[0790] A value that characterizes the distribution of the output data of a model or function.

[0791] In some embodiments, the physical space related information includes one or more of the following:

[0792] Network device identification;

[0793] The region where the network device is located;

[0794] Height of network equipment;

[0795] The coverage area of ​​the network equipment;

[0796] Physical space relationship information between multiple network devices;

[0797] Reference scenario for the area where the network equipment is located.

[0798] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operations administration and maintenance (OAM) pre-configuration information.

[0799] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0800] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0801] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0802] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0803] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0804] Figure 7 A schematic diagram of the structure of a management device for an AI / ML model or function provided in an embodiment of the present application, such as Figure 7 As shown, the device comprises:

[0805] The sending unit 700 is used to send auxiliary information corresponding to the first network device to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function.

[0806] In some embodiments, the auxiliary information includes one or more of the following:

[0807] Signaling related information;

[0808] Information related to signal reception;

[0809] Information related to signal quality;

[0810] Distribution characteristics of model or function-related data;

[0811] Information about the physical space;

[0812] The identifier of the training dataset.

[0813] In some embodiments, the signal transmission related information includes one or more of the following:

[0814] Signaling configuration information;

[0815] Send antenna information.

[0816] In some embodiments, the signal reception related information includes one or more of the following:

[0817] Signal receiving configuration information;

[0818] Receive antenna information.

[0819] In some embodiments, the signal quality related information includes one or more of the following:

[0820] Signal quality metrics;

[0821] Signal quality characteristic value;

[0822] The source of the signal quality characteristic value.

[0823] In some embodiments, the source of the signal quality characteristic includes one or more of the following:

[0824] Signal information corresponding to the terminal;

[0825] Empirical signal information;

[0826] Signal information corresponding to the reference terminal.

[0827] In some embodiments, the distribution characteristic information of the model or function related data includes:

[0828] Information about the distribution characteristics of model or function input data; and / or,

[0829] The distribution characteristics of the model or function output data.

[0830] In some embodiments, the distribution characteristic information of the model or function input data includes one or more of the following:

[0831] The reference distribution of the model or function input data;

[0832] A value that characterizes the distribution of input data to a model or function.

[0833] In some embodiments, the distribution characteristic information of the model or function output data includes one or more of the following:

[0834] The reference distribution of model or function output data;

[0835] A value that characterizes the distribution of the output data of a model or function.

[0836] In some embodiments, the physical space related information includes one or more of the following:

[0837] Network device identification;

[0838] The region where the network device is located;

[0839] Height of network equipment;

[0840] The coverage area of ​​the network equipment;

[0841] Physical space relationship information between multiple network devices;

[0842] Reference scenario for the area where the network equipment is located.

[0843] In some embodiments, the training data set identifier is a training data set identifier determined based on at least one of a predefined rule, a high-layer signaling configuration, and operations administration and maintenance (OAM) pre-configuration information.

[0844] In some embodiments, the training data set identifier includes at least one bit field, each bit field is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0845] In some embodiments, the training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1;

[0846] The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space;

[0847] The number of different candidate combinations of L information is less than or equal to 2 K , L is an integer greater than or equal to 1.

[0848] In some embodiments, the training data set identifier includes multiple sub-identifiers, each sub-identifier is used to indicate one of the following information: signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

[0849] It should be noted that the division of units in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0850] If the integrated unit is implemented 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 solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0851] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.

[0852] On the other hand, an embodiment of the present application also provides a computer-readable storage medium, which stores a computer program, and the computer program is used to enable a computer to execute the management method of the AI / ML model or function provided in the above embodiments.

[0853] It should be noted here that the computer-readable storage medium provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those in the method embodiment will not be described in detail here.

[0854] The computer-readable storage medium can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.

[0855] The technical solution provided in the embodiment of the present application can be applicable to a variety of systems, especially 5G systems. For example, the applicable system can be a global system of mobile communication (GSM) system, a code division multiple access (CDMA) system, a wideband code division multiple access (WCDMA) general packet radio service (GPRS) system, a long term evolution (LTE) system, a LTE frequency division duplex (FDD) system, a LTE time division duplex (TDD) system, an advanced long term evolution (LTE-A) system, a universal mobile telecommunication system (UMTS), a world-wide interoperability for microwave access (WiMAX) system, a 5G new air interface (NR) system, etc. These various systems include terminal equipment and network equipment. The system may also include a core network part, such as an evolved packet system (EPS), a 5G system (5GS), etc.

[0856] The terminal involved in the embodiment of the present application may be a device that provides voice and / or data connectivity to a user, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem. In different systems, the name of the terminal may also be different. For example, in a 5G system, the terminal may be called a user equipment (UE). A wireless terminal device can communicate with one or more core networks (CN) via a radio access network (RAN). The wireless terminal device may be a mobile terminal device, such as a mobile phone (or a "cellular" phone) and a computer with a mobile terminal device. For example, it may be a portable, pocket-sized, handheld, computer-built-in or vehicle-mounted mobile device that exchanges language and / or data with a wireless access network. For example, personal communication service (PCS) phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs) and other devices. The wireless terminal device may also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile station, a remote station, an access point, a remote terminal device, an access terminal device, a user terminal device, a user agent, and a user device, but is not limited thereto in the embodiments of the present application.

[0857] The network device involved in the embodiment of the present application may be a base station, which may include multiple cells providing services for the terminal. Depending on the specific application scenario, the base station may also be called an access point, or may be a device in the access network that communicates with the wireless terminal device through one or more sectors on the air interface, or other names. The network device may be used to interchange received air frames with Internet Protocol (IP) packets, and serve as a router between the wireless terminal device and the rest of the access network, wherein the rest of the access network may include an Internet Protocol (IP) communication network. The network device may also coordinate the attribute management of the air interface. For example, the network device involved in the embodiments of the present application may be a network device (Base Transceiver Station, BTS) in the Global System for Mobile communications (Global System for Mobile communications, GSM) or Code Division Multiple Access (Code Division Multiple Access, CDMA), or a network device (NodeB) in Wide-band Code Division Multiple Access (WCDMA), or an evolutionary network device (evolutional Node B, eNB or e-NodeB) in the long term evolution (long term evolution, LTE) system, a 5G base station (gNB) in the 5G network architecture (next generation system), or a home evolved Node B (Home evolved Node B, HeNB), a relay node, a home base station (femto), a pico base station (pico), etc., which is not limited in the embodiments of the present application. In some network structures, the network device may include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit may also be arranged geographically separately.

[0858] Network devices and terminals can each use one or more antennas for multiple-input multiple-output (MIMO) transmission. MIMO transmission can be single-user MIMO (SU-MIMO) or multi-user MIMO (MU-MIMO). Depending on the form and number of antenna combinations, MIMO transmission can be 2D-MIMO, 3D-MIMO, FD-MIMO or massive-MIMO, or it can be diversity transmission, precoded transmission or beamforming transmission, etc.

[0859] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0860] The present application 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 application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0861] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0862] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0863] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for managing artificial intelligence or machine learning AI / ML models or functions, characterized in that: Applied to terminals, including: Auxiliary information corresponding to the first network device is received, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

2. The method for managing AI / ML models or functions according to claim 1, characterized in that: The auxiliary information includes one or more of the following: Signaling related information; Information related to signal reception; Information related to signal quality; Distribution characteristics of model or function-related data; Information about the physical space; The identifier of the training dataset.

3. The method for managing AI / ML models or functions according to claim 2, characterized in that: The signal transmission related information includes one or more of the following: Signaling configuration information; Send antenna information.

4. The method for managing AI / ML models or functions according to claim 2, characterized in that: The signal reception related information includes one or more of the following: Signal receiving configuration information; Receive antenna information.

5. The method for managing AI / ML models or functions according to claim 2, characterized in that: The signal quality related information includes one or more of the following: Signal quality metrics; Signal quality characteristic value; The source of the signal quality characteristic value.

6. The method for managing AI / ML models or functions according to claim 5, characterized in that: The source of the signal quality characteristic value includes one or more of the following: signal information corresponding to the terminal; Empirical signal information; Signal information corresponding to the reference terminal.

7. The method for managing AI / ML models or functions according to claim 2, characterized in that: The distribution characteristic information of the model or function-related data includes: Information about the distribution characteristics of model or function input data; and / or, The distribution characteristics of the model or function output data.

8. The method for managing AI / ML models or functions according to claim 7, characterized in that: The distribution characteristic information of the model or function input data includes one or more of the following: a reference distribution of input data to the model or function; A value representing the distribution characteristics of the input data to the model or function.

9. The method for managing AI / ML models or functions according to claim 7, characterized in that: The distribution characteristic information of the model or function output data includes one or more of the following: a reference distribution of the model or function output data; The model or function outputs a value representing a distribution characteristic of the data.

10. The method for managing AI / ML models or functions according to claim 2, characterized in that: The physical space related information includes one or more of the following: Network device identification; The region where the network device is located; Height of network equipment; The coverage area of ​​the network equipment; Physical space relationship information between multiple network devices; Reference scenario for the area where the network equipment is located.

11. The method for managing AI / ML models or functions according to claim 2, characterized in that: The training data set identifier is a training data set identifier determined based on at least one of predefined rules, high-layer signaling configuration, and operation, administration and maintenance (OAM) pre-configuration information.

12. The method for managing an AI / ML model or function according to any one of claims 2 to 11, characterized in that: The training data set identifier includes at least one bit field, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

13. The method for managing an AI / ML model or function according to any one of claims 2 to 11, characterized in that: The training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1; The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space; The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

14. The method for managing an AI / ML model or function according to any one of claims 2 to 11, characterized in that: The training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

15. A method for managing artificial intelligence or machine learning AI / ML models or functions, characterized in that: include: Auxiliary information corresponding to the first network device is sent to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function.

16. The method for managing AI / ML models or functions according to claim 15, characterized in that: The auxiliary information includes one or more of the following: Signaling related information; Information related to signal reception; Information related to signal quality; Distribution characteristics of model or function-related data; Information about the physical space; The identifier of the training dataset.

17. The method for managing AI / ML models or functions according to claim 16, characterized in that: The signal transmission related information includes one or more of the following: Signaling configuration information; Send antenna information.

18. The method for managing AI / ML models or functions according to claim 16, characterized in that: The signal reception related information includes one or more of the following: Signal receiving configuration information; Receive antenna information.

19. The method for managing AI / ML models or functions according to claim 16, characterized in that: The signal quality related information includes one or more of the following: Signal quality metrics; Signal quality characteristic value; The source of the signal quality characteristic value.

20. The method for managing AI / ML models or functions according to claim 19, characterized in that: The sources of the signal quality characteristics include one or more of the following: signal information corresponding to the terminal; Empirical signal information; Signal information corresponding to the reference terminal.

21. The method for managing AI / ML models or functions according to claim 16, characterized in that: The distribution characteristic information of the model or function-related data includes: Information about the distribution characteristics of model or function input data; and / or, The distribution characteristics of the model or function output data.

22. The method for managing AI / ML models or functions according to claim 21, characterized in that: The distribution characteristic information of the model or function input data includes one or more of the following: a reference distribution of input data to the model or function; A value representing the distribution characteristics of the input data to the model or function.

23. The method for managing AI / ML models or functions according to claim 21, characterized in that: The distribution characteristic information of the model or function output data includes one or more of the following: a reference distribution of the model or function output data; The model or function outputs a value representing a distribution characteristic of the data.

24. The method for managing AI / ML models or functions according to claim 16, characterized in that: The physical space related information includes one or more of the following: Network device identification; The region where the network device is located; Height of network equipment; The coverage area of ​​the network equipment; Physical space relationship information between multiple network devices; Reference scenario for the area where the network equipment is located.

25. The method for managing AI / ML models or functions according to claim 16, characterized in that: The training data set identifier is a training data set identifier determined based on at least one of predefined rules, high-layer signaling configuration, and operation, administration and maintenance (OAM) pre-configuration information.

26. A method for managing an AI / ML model or function according to any one of claims 16 to 25, characterized in that: The training data set identifier includes at least one bit field, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

27. A method for managing an AI / ML model or function according to any one of claims 16 to 25, characterized in that: The training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1; The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space; The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

28. A method for managing an AI / ML model or function according to any one of claims 16 to 25, characterized in that: The training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

29. A terminal, characterized in that: Including memory, transceiver, processor; A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations: Auxiliary information corresponding to the first network device is received, where the auxiliary information is used for the terminal to manage an artificial intelligence or machine learning AI / ML model or function.

30. The terminal according to claim 29, characterized in that The auxiliary information includes one or more of the following: Signaling related information; Information related to signal reception; Information related to signal quality; Distribution characteristics of model or function-related data; Information about the physical space; The identifier of the training dataset.

31. The terminal according to claim 30, characterized in that The signal transmission related information includes one or more of the following: Signaling configuration information; Send antenna information.

32. The terminal according to claim 30, characterized in that The signal reception related information includes one or more of the following: Signal receiving configuration information; Receive antenna information.

33. The terminal according to claim 30, characterized in that The signal quality related information includes one or more of the following: Signal quality metrics; Signal quality characteristic value; The source of the signal quality characteristic value.

34. The terminal according to claim 33, characterized in that The source of the signal quality characteristic value includes one or more of the following: signal information corresponding to the terminal; Empirical signal information; Signal information corresponding to the reference terminal.

35. The terminal according to claim 30, characterized in that The distribution characteristic information of the model or function-related data includes: Information about the distribution characteristics of model or function input data; and / or, The distribution characteristics of the model or function output data.

36. The terminal according to claim 35, characterized in that The distribution characteristic information of the model or function input data includes one or more of the following: a reference distribution of input data to the model or function; A value representing the distribution characteristics of the input data to the model or function.

37. The terminal according to claim 35, characterized in that The distribution characteristic information of the model or function output data includes one or more of the following: a reference distribution of the model or function output data; The model or function outputs a value representing a distribution characteristic of the data.

38. The terminal according to claim 30, characterized in that The physical space related information includes one or more of the following: Network device identification; The region where the network device is located; Height of network equipment; The coverage area of ​​the network equipment; Physical space relationship information between multiple network devices; Reference scenario for the area where the network equipment is located.

39. The terminal according to claim 30, characterized in that The training data set identifier is a training data set identifier determined based on at least one of predefined rules, high-layer signaling configuration, and operation, administration and maintenance (OAM) pre-configuration information.

40. The terminal according to any one of claims 30 to 39, characterized in that: The training data set identifier includes at least one bit field, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

41. The terminal according to any one of claims 30 to 39, characterized in that: The training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1; The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space; The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

42. The terminal according to any one of claims 30 to 39, characterized in that: The training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

43. A communication device, characterized in that: Including memory, transceiver, processor; A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations: Auxiliary information corresponding to the first network device is sent to the terminal, where the auxiliary information is used by the terminal to manage artificial intelligence or machine learning AI / ML models or functions.

44. The communication device according to claim 43, characterized in that The auxiliary information includes one or more of the following: Signaling related information; Information related to signal reception; Information related to signal quality; Distribution characteristics of model or function-related data; Information about the physical space; The identifier of the training dataset.

45. The communication device according to claim 44, characterized in that The signal transmission related information includes one or more of the following: Signaling configuration information; Send antenna information.

46. ​​The communication device according to claim 44, characterized in that The signal reception related information includes one or more of the following: Signal receiving configuration information; Receive antenna information.

47. The communication device according to claim 44, characterized in that The signal quality related information includes one or more of the following: Signal quality metrics; Signal quality characteristic value; The source of the signal quality characteristic value.

48. The communication device according to claim 47, characterized in that The sources of the signal quality characteristics include one or more of the following: signal information corresponding to the terminal; Empirical signal information; Signal information corresponding to the reference terminal.

49. The communication device according to claim 44, characterized in that The distribution characteristic information of the model or function-related data includes: Information about the distribution characteristics of model or function input data; and / or, The distribution characteristics of the model or function output data.

50. The communication device according to claim 49, characterized in that The distribution characteristic information of the model or function input data includes one or more of the following: a reference distribution of input data to the model or function; A value representing the distribution characteristics of the input data to the model or function.

51. The communication device according to claim 49, characterized in that The distribution characteristic information of the model or function output data includes one or more of the following: a reference distribution of the model or function output data; The model or function outputs a value representing a distribution characteristic of the data.

52. The communication device according to claim 44, characterized in that The physical space related information includes one or more of the following: Network device identification; The region where the network device is located; Height of network equipment; The coverage area of ​​the network equipment; Physical space relationship information between multiple network devices; Reference scenario for the area where the network equipment is located.

53. The communication device according to claim 44, characterized in that The training data set identifier is a training data set identifier determined based on at least one of predefined rules, high-layer signaling configuration, and operation, administration and maintenance (OAM) pre-configuration information.

54. The communication device according to any one of claims 44 to 53, characterized in that: The training data set identifier includes at least one bit field, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

55. The communication device according to any one of claims 44 to 53, characterized in that: The training data set identifier includes K bits, and the K bits are used to indicate at most 2 K One of the information combinations; K is an integer greater than or equal to 1; The information combination is different candidate combinations of L information among information related to signal transmission, information related to signal reception, information related to signal quality, distribution characteristic information of model or function related data, and information related to physical space; The number of different candidate combinations of the L information is less than or equal to 2 K , where L is an integer greater than or equal to 1.

56. The communication device according to any one of claims 44 to 53, characterized in that: The training data set identifier includes multiple sub-identifiers, each of which is used to indicate one of the information selected from the group consisting of signal sending related information, signal reception related information, signal quality related information, distribution characteristic information of model or function related data, and physical space related information.

57. A management device for artificial intelligence or machine learning AI / ML models or functions, characterized in that: include: A receiving unit is used to receive auxiliary information corresponding to the first network device, where the auxiliary information is used for the terminal to manage the AI / ML model or function.

58. A management device for artificial intelligence or machine learning AI / ML models or functions, characterized in that: include: A sending unit is used to send auxiliary information corresponding to the first network device to the terminal, where the auxiliary information is used by the terminal to manage the AI / ML model or function.

59. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program is used to make a computer execute the method according to any one of claims 1 to 14.

60. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program is used to cause a computer to execute the method according to any one of claims 15 to 28.

Images