Information transmission method, device and equipment

Abstract

The invention provides an information transmission method, device and equipment, and the method comprises the steps: receiving an AI (artificial intelligence) model request message sent by a terminal; sending an AI model strategy query request for the terminal to a second core network device according to the AI model request message; receiving an AI model strategy query response which is fed back by the second core network equipment and aims at the terminal; feeding back an AI model response message to the terminal according to the AI model strategy query response; the AI model response message carries AI model information. According to the scheme, the corresponding AI model information can be provided for the terminal according to the AI model request message, so that the AI model can be provided on the basis of user requirements; the problem that AI model provision based on user requirements cannot be supported in the prior art is well solved.

Description

Technical Field

[0001] This application relates to the field of communication technologies, and in particular, to an information transmission method, apparatus, and device. Background Art

[0002] In the prior art, although the communication system has added a Network Data Analytics Function (NWDAF) to provide Artificial Intelligence (AI) capabilities, model training only exists in the NWDAF with logical functions, and the NWDAF with analysis requirements is the only consumer of the AI model. The AI model can only be used for data analysis such as network internal network performance and congestion, lacking the network element function of providing AI model discovery and selection to mobile communication users, and the involved AI models are limited to data analysis types.

[0003] As described above, there are defects in the prior art that cannot support the implementation of providing an AI model based on user requirements. Summary of the Invention

[0004] The purpose of this application is to provide an information transmission method, apparatus, and device to solve the problem in the prior art that cannot support the implementation of providing an AI model based on user requirements.

[0005] To solve the above technical problem, an embodiment of this application provides an information transmission method, which is applied to a first core network device. The method includes:

[0006] Receiving an Artificial Intelligence (AI) model request message sent by a terminal;

[0007] Sending an AI model policy query request for the terminal to a second core network device according to the AI model request message;

[0008] Receiving an AI model policy query response for the terminal feedback by the second core network device;

[0009] Feeding back an AI model response message to the terminal according to the AI model policy query response; the AI model response message carries AI model information.

[0010] Optionally, before receiving the Artificial Intelligence (AI) model request message sent by the terminal, it further includes:

[0011] Receiving an AI model discovery message sent by the terminal; the AI model discovery message carries Artificial Intelligence Service Requirement Description (AIRD) information;

[0012] Determining AI service type information according to the AIRD information, and feeding back an AI model discovery response to the terminal.

[0013] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0014] The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0015] Determine, according to the AI ​​model policy query response and the identification information, whether the AI ​​service model applied for by the terminal is within the subscription model range corresponding to the terminal, to obtain a first determination result;

[0016] Acquire AI model information according to the first determination result;

[0017] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0018] Optionally, obtaining AI model information according to the first determination result includes:

[0019] When the first determination result indicates that the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0020] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0021] The information transmission method further includes:

[0022] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0023] The step of determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response, when the first determination result indicates that the AI ​​service model applied for by the terminal is not within the subscription model range corresponding to the terminal, includes:

[0024] When the first determination result indicates a first situation, or when the first determination result indicates the first situation and the overhead information does not satisfy the overhead limit information, determine the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response;

[0025] The first situation refers to a situation where the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal.

[0026] Optionally, the AI ​​model request message carries artificial intelligence reasoning result requirement AIRR information; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0027] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0028] Determine the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response;

[0029] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0030] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0031] The information transmission method further includes:

[0032] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0033] The determining the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response includes:

[0034] Determine the AI ​​model information according to the overhead information, the model strategy information carried by the AI ​​model strategy query response, and the candidate model information.

[0035] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0036] When the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device; the AI ​​model training request carries the AI ​​service type information and the candidate AI model information;

[0037] Receiving an AI model training response fed back by the third core network device;

[0038] Determining AI model information according to the AI ​​model training response;

[0039] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0040] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0041] The information transmission method further includes:

[0042] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0043] The determining the AI ​​model information according to the AI ​​model training response includes:

[0044] Determine AI model information based on the overhead information and the AI ​​model training response.

[0045] The embodiment of the present application further provides an information transmission method, which is applied to a second core network device, including:

[0046] Receiving an AI model policy query request for a terminal sent by a first core network device;

[0047] According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

[0048] Optionally, the AI ​​model strategy query request carries a terminal identifier of the terminal;

[0049] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0050] According to the terminal identifier, an AI model policy query response for the terminal is fed back to the first core network device.

[0051] Optionally, the AI ​​model strategy query request carries AIRR information and AI business type information;

[0052] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0053] Determine, according to the AIRR information and the AI ​​business type information, whether there is an AI historical policy corresponding to the AI ​​model policy query request, to obtain a second determination result;

[0054] Based on the second determination result, an AI model policy query response for the terminal is fed back to the first core network device.

[0055] Optionally, the feeding back an AI model policy query response for the terminal to the first core network device according to the second determination result includes:

[0056] If the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information;

[0057] When the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device.

[0058] The embodiment of the present application also provides an information transmission method, which is applied to a terminal, including:

[0059] Sending an AI model request message to the first core network device;

[0060] Receive an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

[0061] Optionally, before sending the AI ​​model request message to the first core network device, the method further includes:

[0062] Sending an AI model discovery message to the first core network device; the AI ​​model discovery message carries AIRD information;

[0063] Receive the AI ​​model discovery response fed back by the first core network device.

[0064] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0065] And / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information;

[0066] And / or, the AI ​​model request message carries AIRR information.

[0067] The embodiment of the present application also provides an information transmission method, which is applied to a third core network device, and the method includes:

[0068] Receive an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0069] According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device.

[0070] Optionally, feeding back an AI model training response to the first core network device according to the AI ​​model training request includes:

[0071] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed, and evaluation results are obtained;

[0072] Based on the evaluation result, an AI model training response is fed back to the first core network device.

[0073] Optionally, the AI ​​model training request also carries AIRR information;

[0074] The performing candidate AI model training according to the AI ​​business type information and the candidate AI model information, and obtaining an evaluation result, includes:

[0075] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed to obtain a training result;

[0076] Determining the weight of the evaluation index according to the AIRR information;

[0077] An evaluation result is obtained according to the weights and the training results.

[0078] The embodiment of the present application further provides an information transmission device, the information transmission device is a first core network device, and the device includes a memory, a transceiver, and a processor:

[0079] 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:

[0080] Receiving, through the transceiver, an artificial intelligence AI model request message sent by a terminal;

[0081] According to the AI ​​model request message, sending an AI model policy query request for the terminal to the second core network device through the transceiver;

[0082] Receiving, by the transceiver, an AI model policy query response for the terminal fed back by the second core network device;

[0083] According to the AI ​​model strategy query response, an AI model response message is fed back to the terminal; the AI ​​model response message carries AI model information.

[0084] Optionally, the operation further includes:

[0085] Before receiving the artificial intelligence AI model request message sent by the terminal, receiving the AI ​​model discovery message sent by the terminal through the transceiver; the AI ​​model discovery message carries the artificial intelligence service requirement description AIRD information;

[0086] According to the AIRD information, the AI ​​service type information is determined, and an AI model discovery response is fed back to the terminal through the transceiver.

[0087] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0088] The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0089] Determine, according to the AI ​​model policy query response and the identification information, whether the AI ​​service model applied for by the terminal is within the subscription model range corresponding to the terminal, to obtain a first determination result;

[0090] Acquire AI model information according to the first determination result;

[0091] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0092] Optionally, obtaining AI model information according to the first determination result includes:

[0093] When the first determination result indicates that the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0094] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0095] The operations also include:

[0096] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0097] The step of determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response, when the first determination result indicates that the AI ​​service model applied for by the terminal is not within the subscription model range corresponding to the terminal, includes:

[0098] When the first determination result indicates a first situation, or when the first determination result indicates the first situation and the overhead information does not satisfy the overhead limit information, determine the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response;

[0099] The first situation refers to a situation where the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal.

[0100] Optionally, the AI ​​model request message carries artificial intelligence reasoning result requirement AIRR information; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0101] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0102] Determine the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response;

[0103] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0104] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0105] The operations also include:

[0106] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0107] The determining the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response includes:

[0108] Determine the AI ​​model information according to the overhead information, the model strategy information carried by the AI ​​model strategy query response, and the candidate model information.

[0109] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0110] When the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device through the transceiver; the AI ​​model training request carries the AI ​​service type information and the candidate AI model information;

[0111] Receiving, through the transceiver, an AI model training response fed back by the third core network device;

[0112] Determining AI model information according to the AI ​​model training response;

[0113] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0114] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0115] The operations also include:

[0116] Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information;

[0117] The determining the AI ​​model information according to the AI ​​model training response includes:

[0118] Determine AI model information based on the overhead information and the AI ​​model training response.

[0119] The embodiment of the present application further provides an information transmission device, which is a second core network device and includes a memory, a transceiver, and a processor:

[0120] 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:

[0121] Receiving, by the transceiver, an AI model policy query request for the terminal sent by the first core network device;

[0122] According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

[0123] Optionally, the AI ​​model strategy query request carries a terminal identifier of the terminal;

[0124] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0125] According to the terminal identifier, an AI model policy query response for the terminal is fed back to the first core network device.

[0126] Optionally, the AI ​​model strategy query request carries AIRR information and AI business type information;

[0127] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0128] Determine, according to the AIRR information and the AI ​​business type information, whether there is an AI historical policy corresponding to the AI ​​model policy query request, to obtain a second determination result;

[0129] Based on the second determination result, an AI model policy query response for the terminal is fed back to the first core network device.

[0130] Optionally, the feeding back an AI model policy query response for the terminal to the first core network device according to the second determination result includes:

[0131] If the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information;

[0132] When the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device.

[0133] The embodiment of the present application further provides an information transmission device, which is a terminal and includes a memory, a transceiver, and a processor:

[0134] 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:

[0135] Sending an AI model request message to the first core network device through the transceiver;

[0136] An AI model response message fed back by the first core network device is received through the transceiver; the AI ​​model response message carries AI model information.

[0137] Optionally, the operation further includes:

[0138] Before sending the AI ​​model request message to the first core network device, sending an AI model discovery message to the first core network device through the transceiver; the AI ​​model discovery message carries AIRD information;

[0139] An AI model discovery response fed back by the first core network device is received through the transceiver.

[0140] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0141] And / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information;

[0142] And / or, the AI ​​model request message carries AIRR information.

[0143] The embodiment of the present application further provides an information transmission device, which is a third core network device, and includes a memory, a transceiver, and a processor:

[0144] 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:

[0145] Receiving, by the transceiver, an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0146] According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device through the transceiver.

[0147] Optionally, feeding back an AI model training response to the first core network device according to the AI ​​model training request includes:

[0148] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed, and evaluation results are obtained;

[0149] Based on the evaluation result, an AI model training response is fed back to the first core network device.

[0150] Optionally, the AI ​​model training request also carries AIRR information;

[0151] The performing candidate AI model training according to the AI ​​business type information and the candidate AI model information, and obtaining an evaluation result, includes:

[0152] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed to obtain a training result;

[0153] Determining the weight of the evaluation index according to the AIRR information;

[0154] An evaluation result is obtained according to the weights and the training results.

[0155] The embodiment of the present application further provides an information transmission device, which is applied to a first core network device, and the device includes:

[0156] A first receiving unit, configured to receive an artificial intelligence AI model request message sent by a terminal;

[0157] A first sending unit, configured to send an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message;

[0158] A second receiving unit, configured to receive an AI model policy query response for the terminal fed back by the second core network device;

[0159] The first feedback unit is used to feed back an AI model response message to the terminal according to the AI ​​model strategy query response; the AI ​​model response message carries AI model information.

[0160] Optionally, also include:

[0161] The third receiving unit is used to receive an AI model discovery message sent by the terminal before receiving the AI ​​model request message sent by the terminal; the AI ​​model discovery message carries AIRD information of the artificial intelligence service requirement description;

[0162] The first processing unit is used to determine the AI ​​service type information according to the AIRD information, and to feed back an AI model discovery response to the terminal.

[0163] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0164] The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0165] Determine, according to the AI ​​model policy query response and the identification information, whether the AI ​​service model applied for by the terminal is within the subscription model range corresponding to the terminal, to obtain a first determination result;

[0166] Acquire AI model information according to the first determination result;

[0167] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0168] Optionally, acquiring AI model information according to the first determination result includes:

[0169] When the first determination result indicates that the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0170] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0171] The information transmission device further includes:

[0172] A first determining unit, configured to determine, according to the overhead related information, overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of calculation time consumption information, calculation resource overhead information, and storage resource overhead information;

[0173] The step of determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response, when the first determination result indicates that the AI ​​service model applied for by the terminal is not within the subscription model range corresponding to the terminal, includes:

[0174] When the first determination result indicates a first situation, or when the first determination result indicates the first situation and the overhead information does not satisfy the overhead limit information, determine the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response;

[0175] The first situation refers to a situation where the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal.

[0176] Optionally, the AI ​​model request message carries artificial intelligence reasoning result requirement AIRR information; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0177] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0178] Determine the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response;

[0179] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0180] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0181] The information transmission device further includes:

[0182] A second determining unit is configured to determine, according to the overhead related information, overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of calculation time consumption information, calculation resource overhead information, and storage resource overhead information;

[0183] The determining the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response includes:

[0184] Determine the AI ​​model information according to the overhead information, the model strategy information carried by the AI ​​model strategy query response, and the candidate model information.

[0185] Optionally, the feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes:

[0186] When the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device; the AI ​​model training request carries the AI ​​service type information and the candidate AI model information;

[0187] Receiving an AI model training response fed back by the third core network device;

[0188] Determining AI model information according to the AI ​​model training response;

[0189] According to the AI ​​model information, an AI model response message is fed back to the terminal.

[0190] Optionally, the AI ​​model request message further carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information;

[0191] The information transmission device further includes:

[0192] A third determining unit is used to determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of calculation time consumption information, calculation resource overhead information and storage resource overhead information;

[0193] The determining the AI ​​model information according to the AI ​​model training response includes:

[0194] Determine AI model information based on the overhead information and the AI ​​model training response.

[0195] The embodiment of the present application further provides an information transmission device, which is applied to a second core network device, including:

[0196] A fourth receiving unit, configured to receive an AI model policy query request for a terminal sent by the first core network device;

[0197] The second feedback unit is used to feedback an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request.

[0198] Optionally, the AI ​​model strategy query request carries a terminal identifier of the terminal;

[0199] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0200] According to the terminal identifier, an AI model policy query response for the terminal is fed back to the first core network device.

[0201] Optionally, the AI ​​model strategy query request carries AIRR information and AI business type information;

[0202] The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes:

[0203] Determine, according to the AIRR information and the AI ​​business type information, whether there is an AI historical policy corresponding to the AI ​​model policy query request, to obtain a second determination result;

[0204] Based on the second determination result, an AI model policy query response for the terminal is fed back to the first core network device.

[0205] Optionally, the feeding back an AI model policy query response for the terminal to the first core network device according to the second determination result includes:

[0206] If the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information;

[0207] When the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device.

[0208] The embodiment of the present application further provides an information transmission device, applied to a terminal, including:

[0209] A second sending unit, configured to send an AI model request message to the first core network device;

[0210] The fifth receiving unit is used to receive the AI ​​model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

[0211] Optionally, also include:

[0212] A third sending unit is configured to send an AI model discovery message to the first core network device before sending the AI ​​model request message to the first core network device; the AI ​​model discovery message carries AIRD information;

[0213] The sixth receiving unit is used to receive the AI ​​model discovery response fed back by the first core network device.

[0214] Optionally, the AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal;

[0215] And / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information, and data dimension information;

[0216] And / or, the AI ​​model request message carries AIRR information.

[0217] The embodiment of the present application further provides an information transmission device, which is applied to a third core network device, and the device includes:

[0218] A seventh receiving unit, configured to receive an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0219] The third feedback unit is used to feedback an AI model training response to the first core network device according to the AI ​​model training request, and to send AI model training information corresponding to the AI ​​model training request to the second core network device.

[0220] Optionally, feeding back an AI model training response to the first core network device according to the AI ​​model training request includes:

[0221] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed, and evaluation results are obtained;

[0222] Based on the evaluation result, an AI model training response is fed back to the first core network device.

[0223] Optionally, the AI ​​model training request also carries AIRR information;

[0224] The performing candidate AI model training according to the AI ​​business type information and the candidate AI model information, and obtaining an evaluation result, includes:

[0225] According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed to obtain a training result;

[0226] Determining the weight of the evaluation index according to the AIRR information;

[0227] An evaluation result is obtained according to the weights and the training results.

[0228] An embodiment of the present application also provides a non-transitory readable storage medium, which stores a computer program, and the computer program is used to enable the processor to execute the above-mentioned method on the first core network device side, the second core network device side, the terminal side or the third core network device side.

[0229] The beneficial effects of the above technical solution of the present application are as follows:

[0230] In the above scheme, the information transmission method receives an artificial intelligence AI model request message sent by the terminal; sends an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message; receives an AI model policy query response for the terminal fed back by the second core network device; and feeds back an AI model response message to the terminal according to the AI ​​model policy query response; the AI ​​model response message carries AI model information; can support the provision of corresponding AI model information to the terminal according to the AI ​​model request message, and then support the provision of AI models based on user needs; it is a good solution to the problem that the prior art cannot support the provision of AI models based on user needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0231] Figure 1 A schematic diagram of the wireless communication system architecture of an embodiment of the present application;

[0232] Figure 2 Provide an architecture diagram for the machine learning model of the embodiment of the present application;

[0233] Figure 3 The information transmission method of the present application is shown in the following figure. Figure 1 ;

[0234] Figure 4 The information transmission method of the present application is shown in the following figure. Figure 2 ;

[0235] Figure 5 The information transmission method of the present application is shown in the following figure. Figure 3 ;

[0236] Figure 6 The information transmission method of the present application is shown in the following figure. Figure 4 ;

[0237] Figure 7 A schematic diagram of a network element architecture of an embodiment of the present application;

[0238] Figure 8 The information transmission method of the present application is specifically implemented in the following structure: Figure 1 ;

[0239] Fig. 9 The information transmission method of the present application is specifically implemented as follows: Figure 1 ;

[0240] Fig.10 The information transmission method of the present application is specifically implemented in the following structure: Figure 2 ;

[0241] Fig.11 The information transmission method of the present application is specifically implemented as follows: Figure 2 ;

[0242] Fig.12 The information transmission method of the present application is specifically implemented in the following structure: Figure 3 ;

[0243] Fig.13 The information transmission method of the present invention is specifically implemented as follows: Figure 3 ;

[0244] Fig.14 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 1 ;

[0245] Fig.15 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 2 ;

[0246] Fig.16 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 3 ;

[0247] Fig.17 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 4 ;

[0248] Fig.18 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 1 ;

[0249] Fig.19 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 2 ;

[0250] Fig. 20 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 3 ;

[0251] Fig.21 The structure of the information transmission device of the embodiment of the present application is shown as follows Figure 4 . DETAILED DESCRIPTION

[0252] 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.

[0253] In the embodiments of the present application, the term "and / or" describes the association relationship of the associated objects, indicating that there may be three relationships. 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.

[0254] In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0255] It is explained here that 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.

[0256] Figure 1 A block diagram of a wireless communication system applicable to the embodiments of the present application is shown. The wireless communication system includes a terminal device (referred to as a terminal) and a network device.

[0257] The terminal device 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 device may also be different. For example, in a 5G system, the terminal device may be called a user equipment (UE). The wireless terminal device can communicate with one or more core networks (CN) via a radio access network (RAN). The wireless terminal device can 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 can 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 (PDA) 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 in the embodiments of the present application.

[0258] The network device involved in the embodiments 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 (the next Generation Node B, 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.

[0259] Network devices and terminal devices 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.

[0260] The following first introduces the contents involved in the solution provided in the embodiment of the present application.

[0261] The organic combination of 6G network and artificial intelligence (AI) is the future development trend. In order to realize the vision of 6G, the assistance of AI is indispensable. Therefore, the 6G mobile communication network needs to consider deep coupling with AI in the design stage. 6G intelligence inherently requires the network to provide intelligent and inclusive AI capabilities. The AI ​​model provided should not only be used for the network itself, but also provide AI model discovery capabilities to provide AI services for mobile communication users. At the same time, the on-demand selection of AI models is very important, which determines the resource consumption and performance of AI models in solving AI services. At present, the AI ​​services of 5G communication systems are still limited to the network itself and AI models cannot be selected according to demand.

[0262] Among them, the 5G system AI capabilities and models are introduced as follows:

[0263] The current 5G mobile communication system network architecture is designed with connection and data transmission as the center. The 5G communication system has added a network data analysis function (NWDAF), which is used to collect data generated by 5G core network functions and provide AI data analysis capabilities about the network, including network service experience, network performance, slice load, network function load, terminal mobility, etc.

[0264] In terms of architecture, NWDAF can be decomposed into NWDAFs that include model training logical function (MTLF) and analytics logical function (AnLF). Analysis data can be shared, synthesized, and transferred between multiple NWDAFs.

[0265] Specific as Figure 2As shown in Figure 1, the 5G system architecture allows a NWDAF including an analysis logic function AnLF to use an AI model from another NWDAF including a model training logic function MTLF to provide AI services. The NWDAF including AnLF can use the Nnwdaf interface to request and select an AI model to provide services.

[0266] In addition, the 5G communication system does not consider the specific methods and processes for selecting AI models under different AI business requirements and business types; specifically, the existing 5G architecture does not consider the integration with AI and cannot provide flexible AI model selection for multiple types of AI services. The existing 6G intelligent endogenous architecture still fails to realize AI model discovery and make on-demand AI model selection decisions based on the data and needs provided by users.

[0267] Based on the above, the embodiments of the present application provide an information transmission method, device and equipment to solve the problem that the prior art cannot support the provision of AI models based on user needs. Among them, the method, device and equipment are based on the same application concept. Since the principles of solving the problems by the method, device and equipment are similar, the implementation of the method, device and equipment can refer to each other, and the repeated parts will not be repeated.

[0268] The information transmission method provided in the embodiment of the present application is applied to a first core network device, such as Figure 3 As shown, the method includes:

[0269] Step 31: Receive an artificial intelligence AI model request message sent by the terminal;

[0270] Step 32: Send an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message;

[0271] Step 33: Receive an AI model policy query response for the terminal fed back by the second core network device;

[0272] Step 34: Feedback an AI model response message to the terminal according to the AI ​​model strategy query response; the AI ​​model response message carries AI model information.

[0273] Among them, the first core network device can be implemented as an artificial intelligence evaluation and selection function AIESF device, and / or the AI ​​model request message can carry the terminal identification (UE ID) of the terminal, and / or the second core network device can be implemented as a policy control function PCF device, and / or the AI ​​model policy query request can carry the terminal identification (UE ID) of the terminal, and / or the AI ​​model information may include: at least one of: AI Model ID (AI model identification), AI model accuracy, estimated training time and other information, but is not limited to this.

[0274] The information transmission method provided in the embodiment of the present application receives an artificial intelligence (AI) model request message sent by a terminal; sends an AI model policy query request for the terminal to a second core network device according to the AI ​​model request message; receives an AI model policy query response for the terminal fed back by the second core network device; and feeds back an AI model response message to the terminal according to the AI ​​model policy query response; the AI ​​model response message carries AI model information; and can support the provision of corresponding AI model information to the terminal according to the AI ​​model request message, thereby supporting the provision of AI models based on user needs; and well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0275] Furthermore, before receiving the artificial intelligence AI model request message sent by the terminal, it also includes: receiving an AI model discovery message sent by the terminal; the AI ​​model discovery message carries artificial intelligence service demand description AIRD information; according to the AIRD information, determining the AI ​​service type information, and feeding back an AI model discovery response to the terminal.

[0276] In this way, the AI ​​service type information can be clarified first for subsequent use. Among them, the AI ​​model discovery response can carry the AI ​​service type information, but is not limited to this.

[0277] This solution may specifically include three situations: situation 1, the terminal specifies the AI ​​model; situation 2, the terminal does not specify the AI ​​model, but the second core network device stores the corresponding AI strategy (i.e., the model strategy information corresponding to the terminal); situation 3, the terminal does not specify the AI ​​model, and the second core network device does not store the corresponding AI strategy; the following are introduced respectively:

[0278] For situation one, the AI ​​model request message carries identification information of the AI ​​business model applied for by the terminal; and according to the AI ​​model policy query response, the AI ​​model response message is fed back to the terminal, including: determining whether the AI ​​business model applied for by the terminal is within the subscription model range corresponding to the terminal according to the AI ​​model policy query response and the identification information, and obtaining a first determination result; according to the first determination result, obtaining AI model information; and according to the AI ​​model information, feeding back an AI model response message to the terminal.

[0279] In this way, the AI ​​model response message can be accurately fed back to the terminal; wherein the identification information can correspond to the AI ​​business type information, and / or, the obtaining of the AI ​​model information according to the first determination result may include: when the first determination result indicates that the AI ​​business model applied for by the terminal is within the subscription model range corresponding to the terminal, directly obtaining the corresponding AI model information according to the identification information of the AI ​​business model applied for by the terminal; and / or, the AI ​​model request message carries AIRR information requiring artificial intelligence reasoning results (AIRR can be used as the model performance expected by the terminal, such as the expected model accuracy rate to reach more than 95%, even if the terminal specifies the AI ​​type in the case, there can also be performance requirements), wherein, "according to the AI ​​model policy query response and the identification information, determine The method of “determining whether the AI ​​business model applied for by the terminal is within the scope of the subscription model corresponding to the terminal, and obtaining a first determination result” may include: when it is determined according to the AI ​​model policy query response that the AI ​​business model applied for by the terminal satisfies the AIRR information, determining according to the AI ​​model policy query response and the identification information whether the AI ​​business model applied for by the terminal is within the scope of the subscription model corresponding to the terminal, and obtaining a first determination result; further, the present scheme may also include: when it is determined according to the AI ​​model policy query response that the AI ​​business model applied for by the terminal satisfies the AIRR information, the specific operation of “feeding back an AI model response message to the terminal according to the AI ​​model policy query response” corresponding to the following situation two or three may be performed, but is not limited thereto.

[0280] Among them, obtaining AI model information according to the first determination result includes: when the first determination result indicates that the AI ​​business model applied for by the terminal is not within the subscription model range corresponding to the terminal, determining the AI ​​model information corresponding to the terminal according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0281] This can support accurate acquisition of AI model information in the above situation. Among them, "stored AI model information" can be stored locally or stored in the third core network device, which is not limited here.

[0282] In an embodiment of the present application, the AI ​​model request message also carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission method also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead related information; the overhead information includes: at least one of calculation time information, calculation resource overhead information and storage resource overhead information; when the first determination result indicates that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal, determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response, including: when the first determination result indicates the first situation, or the first determination result indicates the first situation and the overhead information does not meet the overhead limit information, determining the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response; wherein the first situation refers to the situation that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal.

[0283] In this way, AI model information of an AI model that better meets the needs can be obtained.

[0284] For situations two and three, the AI ​​model request message carries AIRR information of artificial intelligence reasoning result requirements; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0285] This makes it easier to support the acquisition of AI models that better meet terminal needs.

[0286] For situation two, the feeding back an AI model response message to the terminal according to the AI ​​model policy query response includes: determining AI model information according to the model strategy information and candidate model information carried by the AI ​​model policy query response; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0287] This allows for accurate feedback of AI model response messages.

[0288] Furthermore, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of: data type information, data size information, and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission method also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message based on the overhead-related information; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information based on the model policy information and candidate model information carried by the AI ​​model policy query response includes: determining the AI ​​model information based on the overhead information, the model policy information and candidate model information carried by the AI ​​model policy query response.

[0289] In this way, AI model information of an AI model that better meets the requirements can be obtained. The method may further include: extracting data features based on the overhead related information; correspondingly, sending an AI model policy query request for the terminal to the second core network device based on the AI ​​model request message, including: sending an AI model policy query request for the terminal to the second core network device based on the AI ​​model request message and the data features; but it is not limited to this.

[0290] For situation three, the feeding back an AI model response message to the terminal based on the AI ​​model policy query response includes: when the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device; the AI ​​model training request carries the AI ​​service type information and candidate AI model information; receiving an AI model training response fed back by the third core network device; determining the AI ​​model information based on the AI ​​model training response; and feeding back an AI model response message to the terminal based on the AI ​​model information.

[0291] This can also support accurate feedback of AI model response messages. Among them, "the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal", which can be understood as: no AI model matching the AI ​​model request message is retrieved; and / or, the third core network device can be implemented as an artificial intelligence model storage function AIMSF device, and / or, the AI ​​model training request can also carry AIRR information, and the third core network device operates accordingly; and / or, the candidate AI model information can be obtained based on the correspondence between "AI business type-AI business model-evaluation index" (such as a relationship table), but is not limited to this.

[0292] Furthermore, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission method also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message based on the overhead-related information; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information based on the AI ​​model training response includes: determining the AI ​​model information based on the overhead information and the AI ​​model training response.

[0293] In this way, AI model information of an AI model that better meets the requirements can be obtained. Wherein, determining the AI ​​model information according to the overhead information and the AI ​​model training response may include: determining the AI ​​model information according to the overhead information, the AI ​​model training response and AIRR, but is not limited thereto.

[0294] The embodiment of the present application also provides an information transmission method, which is applied to a second core network device, such as Figure 4 As shown, including:

[0295] Step 41: Receive an AI model policy query request for the terminal sent by the first core network device;

[0296] Step 42: According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

[0297] Among them, the first core network device can be implemented as an artificial intelligence evaluation and selection function AIESF device, and / or, the second core network device can be implemented as a policy control function PCF device, and / or, the AI ​​model policy query request can carry the terminal identification (UE ID) of the terminal, but is not limited to this.

[0298] The information transmission method provided in the embodiment of the present application receives an AI model policy query request for a terminal sent by a first core network device; based on the AI ​​model policy query request, feeds back an AI model policy query response for the terminal to the first core network device; can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; and well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0299] This solution may specifically include three situations: situation 1, the terminal specifies the AI ​​model; situation 2, the terminal does not specify the AI ​​model, but the second core network device stores the corresponding AI strategy (i.e., the model strategy information corresponding to the terminal); situation 3, the terminal does not specify the AI ​​model, and the second core network device does not store the corresponding AI strategy; the following are introduced respectively:

[0300] For situation one, situation two and / or situation three, the AI ​​model policy query request carries the terminal identifier of the terminal; and according to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device includes: according to the terminal identifier, feeding back an AI model policy query response for the terminal to the first core network device.

[0301] This can accurately feedback the AI ​​model policy query response for the terminal.

[0302] For situations two and three, the AI ​​model policy query request carries AIRR information and AI service type information; based on the AI ​​model policy query request, the AI ​​model policy query response for the terminal is fed back to the first core network device, including: determining whether there is an AI historical policy corresponding to the AI ​​model policy query request based on the AIRR information and the AI ​​service type information, and obtaining a second determination result; based on the second determination result, the AI ​​model policy query response for the terminal is fed back to the first core network device.

[0303] In this way, even if the user has not specified an AI model, an AI model that meets the terminal requirements can be obtained, and accurate AI model strategy query response feedback can be performed.

[0304] Among them, the feeding back of the AI ​​model policy query response for the terminal to the first core network device based on the second determination result includes: (1) when the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back the AI ​​model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information; (2) when the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, feeding back to the first core network device an AI model policy query response indicating that there is no model policy information corresponding to the terminal.

[0305] In this way, the AI ​​model policy query response for the terminal can be accurately obtained for the above situation two or three. Among them, "indicating that there is no model policy information corresponding to the terminal" can be understood as: no AI model matching the AI ​​model request message initiated by the terminal is retrieved, but it is not limited to this. Among them, the model policy information may include the overhead information of the AI ​​model for each AI service, such as computing resource overhead information, storage resource overhead information, etc., which is not limited here.

[0306] The present application also provides an information transmission method, which is applied to a terminal, such as Figure 5 As shown, including:

[0307] Step 51: Send an AI model request message to the first core network device;

[0308] Step 52: Receive an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

[0309] Among them, the first core network device can be implemented as an artificial intelligence evaluation and selection function AIESF device, and / or the AI ​​model request message can carry the terminal identifier (UE ID) of the terminal, and / or the AI ​​model information may include: at least one of the following information: AI ModelID (AI model identifier), AI model accuracy, estimated training time, etc., and / or step 51 may include: sending an AI model request message to the first core network device according to the AI ​​service type information; and / or, after step 52, it may also include: according to the AI ​​model information, using the corresponding AI model to execute the generated AI service, but it is not limited to this.

[0310] The information transmission method provided in the embodiment of the present application sends an AI model request message to the first core network device; receives an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information; it can support the acquisition of AI model information corresponding to the AI ​​model request message, and then support the provision of AI models based on user needs; it is a good solution to the problem that the prior art cannot support the provision of AI models based on user needs.

[0311] Furthermore, before sending the AI ​​model request message to the first core network device, it also includes: sending an AI model discovery message to the first core network device; the AI ​​model discovery message carries AIRD information; and receiving an AI model discovery response fed back by the first core network device.

[0312] In this way, the first core network device can first clarify the AI ​​service type information for subsequent use. Among them, the AI ​​model discovery response can carry the AI ​​service type information, but is not limited to this.

[0313] Among them, the AI ​​model request message carries identification information of the AI ​​business model applied for by the terminal; and / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; and / or, the AI ​​model request message carries AIRR information.

[0314] This can support the terminal to obtain accurate AI model response messages. Among them, the "identification information of the AI ​​business model applied for by the terminal" can correspond to the AI ​​business type information, but is not limited to this.

[0315] The embodiment of the present application also provides an information transmission method, which is applied to a third core network device, such as Figure 6 As shown, the method includes:

[0316] Step 61: Receive an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0317] Step 62: According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device.

[0318] Among them, the first core network device can be implemented as an artificial intelligence evaluation and selection function AIESF device, and / or, the third core network device can be implemented as an artificial intelligence model storage function AIMSF device, and / or, the candidate AI model information can be obtained according to the correspondence between "AI business type-AI business model-evaluation index" (such as a relationship table), and / or, "AI model training information" can include: terminal identification, (corresponding to the AI ​​model request message) the optimal AI business model ID and (AI business model ID) corresponding to at least one of the hyperparameter group IDs, but is not limited to this.

[0319] The information transmission method provided in the embodiment of the present application receives an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; according to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device; it can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; it well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0320] Among them, the feeding back an AI model training response to the first core network device according to the AI ​​model training request includes: training the candidate AI model according to the AI ​​service type information and the candidate AI model information, and obtaining an evaluation result; and feeding back an AI model training response to the first core network device according to the evaluation result.

[0321] In this way, the AI ​​model training response can be obtained specifically and accurately.

[0322] In an embodiment of the present application, the AI ​​model training request also carries AIRR information; the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and an evaluation result is obtained, including: the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and the training result is obtained; the weight of the evaluation index is determined according to the AIRR information; and the evaluation result is obtained according to the weight and the training result.

[0323] In this way, the evaluation results corresponding to the candidate AI models can be accurately obtained. Among them, "according to the AI ​​business type information and the candidate AI model information, the candidate AI model is trained to obtain the training results; according to the AIRR information, the weights of the evaluation indicators are determined; according to the weights and the training results, the evaluation results are obtained", which may specifically include: training multiple candidate AI models in parallel (such as starting multiple training processes to train different AI models at the same time), determining the evaluation indicators corresponding to each candidate AI model according to the correspondence between "AI business type-AI business model-evaluation indicators" (such as a relationship table), and using the evaluation indicators to evaluate the output results of each candidate AI model to obtain the training results; determining the importance ranking of the evaluation indicators according to the AIRR information, assigning corresponding weights to each evaluation indicator, and obtaining the comprehensive training score (i.e., the evaluation result) according to the weights and the evaluation indicator results calculated by each candidate AI model (such as multiplying the evaluation indicator results calculated by each candidate AI model by the corresponding weights and adding them up); but it is not limited to this.

[0324] It is noted here that the relevant contents of the above-mentioned methods on each side (such as the first core network device side and the terminal side methods) can refer to each other, and the repeated contents will not be repeated.

[0325] The information transmission method provided in the embodiment of the present application is illustrated below by taking an artificial intelligence evaluation and selection function AIESF device as an example of the first core network device, a policy control function PCF device as an example of the second core network device, and an artificial intelligence model storage function AIMSF device as an example of the third core network device.

[0326] In view of the above technical problems, and taking into account: In order to make the 6G network have the ability of endogenous intelligence, it should not only be able to provide AI model discovery and services to the network element functions within the network, but also be able to provide multiple AI types of model discovery to mobile communication user UE, and should be able to perform reasonable and efficient AI model selection according to AI service requirements; wherein, the above-mentioned network element functions and mobile communication user UE are users with AI needs; the user has registered with the network before performing AI model discovery and on-demand selection, and generated AI services that need to be processed. Based on this, an embodiment of the present application provides an information transmission method, which can be specifically implemented as a 6G intelligent endogenous AI model discovery and on-demand selection method, mainly involving: In order to achieve AI model discovery and on-demand selection, such as Figure 7 As shown in the figure, two new network functions are added to the network structure, namely AI Evaluation and Selection Function (AIESF) and AI Model Storage Function (AIMSF). AIESF and AIMSF are introduced below. Figure 7 PCF in it stands for Policy Control Function, UDR stands for Unified Data Repository, UDSF stands for Unstructured Data Storage Network Function, AMF stands for Access and Mobility Management Function, SMF stands for Session Management Function, NRF stands for Network Repository Function, RAN stands for Radio Access Network, UPF stands for User Plane Function, DN stands for Data Network, and N1, N2, N3, N4 and N6 stand for interfaces between corresponding network elements.

[0327] 1. AI Evaluation and Selection Function (AIESF):

[0328] 1) It can be deployed on the core network side;

[0329] 2) It can evaluate the computing resource and storage resource overhead and computing time;

[0330] 3) It can support AI model discovery function, receive AI model discovery requests from the RAN side or other network element functions of the core network, and feedback the corresponding discovered AI models;

[0331] 4) Ability to maintain (including adding, removing and updating) AI model information stored by AIMSF and synchronize AI model information with AIMSF; AI model information includes model name, identification ID, model description and other information;

[0332] 5) It can evaluate whether the submitted AI task has a matching training model (i.e., AI model), and can also evaluate information such as data type and data size. It can also weigh indicators such as computing time and model output accuracy to select the optimal AI model. At the same time, it can also combine the AI ​​strategy message provided by PCF (corresponding to the above-mentioned model strategy information) to make intelligent AI model selection decisions.

[0333] 2. AI Model Storage Function (AIMSF):

[0334] 1) Can be deployed on the core network side;

[0335] 2) It can store multiple types of AI models, as shown in Table 1-1 below. AIMSF can store pre-configured AI models and hyperparameters (corresponding to the AI ​​models); specifically, the search space of AIMSF can be composed of pre-configured AI models and hyperparameters (corresponding to the AI ​​models), but is not limited to this.

[0336] 3) It can support receiving AI model training requests and provide corresponding AI business model training and reasoning services based on the training requests;

[0337] 4) It can be responsible for maintaining AI models suitable for solving various problems (corresponding to service requests of various AI models) and corresponding hyperparameter data. It can be understood as storing relevant information after training the AI ​​model.

[0338] 5) It can intelligently update the stored AI models or obtain more suitable models according to the AI ​​business type; it can also support adding and / or deleting AI models. In addition, operation and maintenance personnel can also add or delete AI models through the management platform.

[0339] Table 1-1 Comparison table of AI business types, matching business models and evaluation indicators

[0340] AI Business Type AI Business Model Evaluation indicators Classification Cla SVM, SGD, KNN, RF Accuracy, precision, recall, F1-Score Prediction DT, LR, CNN, LTSM MSE, RMSE, MAE, R-squared Return to Reg SVM, SGD, RF MSE, RMSE, MAE, R-squared Clustering K-means SSE、SI Computer Vision CNN, YOLO Accuracy, precision, recall, F1-Score

[0341] Among them, SVM represents the Support Vector Machines model, SGD represents the Stochastic Gradient Descent method, KNN represents the K-Nearest Neighbor, RF represents the Random Forest, DT represents the Decision Tree, LR represents the Logistic Regression model, CNN represents the Convolutional Neural Network, LTSM represents the Long Short-Term Memory network, K-means represents the K-means clustering algorithm, and YOLO represents (You Only Look Once);

[0342] F1-Score stands for balanced F Score, which is the harmonic mean of precision and recall.

[0343] MSE stands for Mean Square Error, RMSE stands for Root Mean Square Error, MAE stands for Mean Absolute Error, R-squared stands for goodness of fit, SSE stands for Sum of Squares for Error, and SI stands for Silhouette Index.

[0344] In this solution, AI models can be pre-deployed in AIMSF, which stores various types of pre-trained models. Among them, AI model generation is related to AI model training, which will not be elaborated in detail here.

[0345] The following is a specific example to illustrate this solution.

[0346] Embodiment 0: The user specifies the AI ​​model;

[0347] This embodiment can adopt Figure 8 The architecture shown in the figure can be implemented as follows Fig. 9 As shown, the following steps are included:

[0348] Step 91. The user sends an AI model discovery message to AIESF (corresponding to the above-mentioned sending of an AI model discovery message to the first core network device), where the message content may include user information (UE Information) and AI service requirements description (AI Requirements Description, AIRD). When the user is a mobile communication user, the user information may include user ID (such as UE ID), application information that triggers the AI ​​service, manufacturer identification and other information. When the user is a network element function NF within the network, the user information may include NF Profile (network element function configuration) and other information. Among them, AIRD may include problem description, data source (whether data collection is requested) and other information.

[0349] Step 92. AIESF analyzes the AI ​​model discovery message and determines the AI ​​service type according to AIRD; corresponding to the above, the AI ​​service type information is determined according to the AIRD information.

[0350] AIESF responds to the user and sends an AI model discovery response to the user (corresponding to the above-mentioned feedback of the AI ​​model discovery response to the terminal), indicating that AIESF has received the AI ​​service demand (also understood as AI business demand) and classified the AI ​​type according to the demand description. The response content may include the AI ​​business type (corresponding to the above-mentioned AI business type information).

[0351] Step 93. The user sends an AI model request message (corresponding to the above-mentioned sending of an AI model request message to the first core network device), where the message content may include a user ID (such as UE ID), a data type (Data Type), a data size (DataSize), a data dimension (Data Dimension), an AI inference result requirement (AI Inference Result Requirements, AIRR), and relevant information of the applied AI business model (such as AI Model ID, which may correspond to the received AI business type information). Among them, the AI ​​business model may be provided based on whether the user has historical experience and knowledge of AI training; specifically, this embodiment considers that the user specifies the AI ​​model, and the AI ​​model sent in step 93 may be specified by the user based on historical experience information.

[0352] Step 94. AIESF queries the user AI model policy subscribed by the user to PCF (sends user AI policy subscription query information to) to determine whether the AI ​​service model applied by the user is within the model range subscribed by the user; corresponding to the above-mentioned AI model request message, sends an AI model policy query request for the terminal to the second core network device. Among them, the user AI policy subscription query information can carry UE ID.

[0353] Step 95. PCF responds to the user AI model policy query and returns a user AI model policy message (corresponding to the above-mentioned receiving the AI ​​model policy query request for the terminal sent by the first core network device; according to the AI ​​model policy query request, the AI ​​model policy query response for the terminal is fed back to the first core network device). The user AI model policy message may include at least one of the AI ​​model type (AI Type) subscribed by the user, AI service priority (Priority), historical AI model selection information and hyperparameters (corresponding to History in the figure), computing resources and storage resource consumption limits (corresponding to Limitation in the figure), model charges, transmission overhead and other information; the user AI model policy message may also include UEID.

[0354] AIESF evaluates the computation time and the overhead of computing resources, storage resources, etc. (may also include extracting data features) according to the information such as the data type and data size in the AI ​​model request message; corresponding to the above, the overhead information of the AI ​​service corresponding to the AI ​​model request message is determined according to the overhead related information; the overhead information includes: at least one of the computation time information, computing resource overhead information and storage resource overhead information. Among them, "AIESF evaluates the computation time and the overhead of computing resources, storage resources, etc. according to the data type and data size in the AI ​​model request message" can be performed when AIESF determines that the AI ​​service model applied by the terminal meets AIRR according to the AI ​​model policy message; further, when AIESF determines that the AI ​​service model applied by the terminal does not meet AIRR according to the AI ​​model policy message, it can enter the following embodiment 1 or 2 "AIESF evaluates the computation time and the overhead of computing resources, storage resources, etc. according to the data type and data size in the AI ​​model request message, and extracts data features" to continue to perform operations to complete the subsequent operation of feeding back the AI ​​model policy query response to the terminal (such as continuing to perform steps 116-118 in embodiment 1), but it is not limited to this.

[0355] Then proceed to step 96a or 96b:

[0356] Step 96a. If the AI ​​business model applied for by the user is within the model range subscribed by the user and meets the Limitation provided by PCF, AIESF sends an AI model feedback message (also called AI model response message) to the user. The message content may include AI Model ID (consistent with the AI ​​Model ID in step 93), AI model accuracy, estimated training time and other information; corresponding to the above-mentioned case where the first determination result indicates that the AI ​​business model applied for by the terminal is within the subscription model range corresponding to the terminal, the AI ​​model information is determined according to the identification information (the information may include: AI Model ID, AI model accuracy, estimated training time and other information).

[0357] Step 96b. If the AI ​​service model applied by the user is not within the model range subscribed by the user, the AIESF can make the best model selection decision based on the AI ​​model information stored by the maintained AIMSF, combined with the received user AI model request message and the AI ​​model policy message provided by the PCF; corresponding to the above-mentioned case where the first determination result indicates that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response. Then, the AIESF sends an AI model response message to the user (corresponding to the above-mentioned feedback of the AI ​​model response message to the terminal based on the AI ​​model information), indicating that the user failed to specify the AI ​​model (the content of the message may include the reason for the failure), and provides the user with recommended AI model information, which may include information such as AI Model ID, AI model accuracy, and estimated training time.

[0358] In this solution, the AI ​​model is stored in AIMSF. AIESF evaluates a series of factors and selects the most suitable one from AIMSF (AIESF is responsible for maintaining the AI ​​model information stored in AIMSF); but this is not limited to this.

[0359] In addition, the relevant contents of the AI ​​model request message mentioned above are introduced as follows:

[0360] a) Data Type Data Type, which may include: at least one of the following types: text TXT, picture IMG, audio AUD, video VID, data DAT, etc.;

[0361] b) AI business types AI Types, which may include at least one of the following types: Classification (Classification, Cla), Prediction (Pred), Regression (Regression, Reg), Clustering (Clustering, Clust), Computer Vision (Computer Vision, CV), etc.;

[0362] c) AI business model AI Model, which may include: support vector machine model SVM, gradient descent method SGD, K-nearest neighbor (KNN, K-NearestNeighbor), neural network model (Neural Network), logistic regression model (LogisticsRegression, LR), decision tree (Decision Tree, DT), convolutional neural network CNN, long short-term memory network LSTM, K-means clustering (K-means), random forest (Random Forest, RF), YOLO (You Only Look Once) and other models;

[0363] d) AI Inference Result Requirements (AIRR), which may include at least one of the following indicators: expected accuracy, mean square error, training time, generalization, consistency between loss function and optimization objective, and fairness.

[0364] Example 1: The user does not specify an AI model, but PCF has stored the corresponding AI strategy, historical experience and knowledge;

[0365] This embodiment can adopt Fig.10 The architecture shown in the figure can be implemented as follows Fig.11 As shown, the following steps are included:

[0366] Step 111. The user sends an AI model discovery message to AIESF (corresponding to the above-mentioned sending of an AI model discovery message to the first core network device), where the message content may include user information (UE Information) and AI service requirements description (AI Requirements Description, AIRD). When the user is a mobile communication user, the user information may include user ID (such as UE ID), application information that triggers the AI ​​service, manufacturer identification and other information. When the user is a network element function NF within the network, the user information may include information such as NF Profile. Among them, AIRD may include problem description, data source (whether data collection is requested) and other information.

[0367] Step 112. AIESF analyzes the AI ​​model discovery message and determines the AI ​​service type according to AIRD; corresponding to the above, determining the AI ​​service type information according to the AIRD information.

[0368] AIESF responds to the user and sends an AI model discovery response to the user (corresponding to the above-mentioned feedback of the AI ​​model discovery response to the terminal), indicating that AIESF has received the AI ​​service demand (also understood as AI business demand) and classified the AI ​​type according to the demand description. The response content may include the AI ​​business type (corresponding to the above-mentioned AI business type information).

[0369] Step 113. The user sends an AI model request message (corresponding to the above-mentioned sending of an AI model request message to the first core network device), wherein the message content may include a user ID (such as UE ID), a data type (Data Type), a data size (DataSize), a data dimension (Data Dimension), and AI inference result requirements (AI Inference Result Requirements, AIRR). Among them, the AI ​​business model may be provided based on whether the user has historical experience and knowledge of AI training.

[0370] The AI ​​model request message of step 113 does not carry: relevant information of the applied AI business model (such as AIModel ID, which may correspond to the received AI business type information); specifically, the difference between step 113 of this embodiment and step 93 of embodiment 0 is that: in this embodiment, the AI ​​model request message does not provide an AI Model ID, that is, the AI ​​model is not specified.

[0371] Step 114. AIESF queries the user AI model policy subscribed by the user to PCF (sends user AI policy subscription query information to); corresponding to the above-mentioned AI model request message, sends an AI model policy query request for the terminal to the second core network device. Among them, the user AI policy subscription query information can carry UE ID.

[0372] Step 115. PCF responds to the user AI model policy query and returns a user AI model policy message (corresponding to the above-mentioned receiving the AI ​​model policy query request for the terminal sent by the first core network device; according to the AI ​​model policy query request, feeding back the AI ​​model policy query response for the terminal to the first core network device). The user AI model policy message may include at least one of the AI ​​model type (AI Type) subscribed by the user, AI service priority (Priority), historical AI model selection information and hyperparameters (corresponding to History in the figure), computing resources and storage resource consumption limits (corresponding to Limitation in the figure), model charges, transmission overhead and other information; the user AI model policy message may also include UEID.

[0373] AIESF evaluates the computing time and the overhead of computing resources, storage resources, etc. according to the data type and data size in the AI ​​model request message (corresponding to the above-mentioned determination of the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead-related information; the overhead information includes: at least one of computing time information, computing resource overhead information and storage resource overhead information), and extracts data features; it can also be understood that AIESF evaluates the AI ​​model request information (corresponding to the AI ​​model request message) and generates information such as data features and overhead.

[0374] Step 116. AIESF sends an AI policy query request to PCF, which may include at least one of the following information: user ID (i.e., UE ID), data features, AI inference result requirements (AIRR), and AI service type; the request may be used to trigger PCF to execute: determine whether there is historical experience (AI model-related experience information) corresponding to the AI ​​service generated by the above user based on the above information.

[0375] Based on the information in the provided AI policy query request, the PCF retrieves historical experience and knowledge that matches the AIRR and AI business type, and determines whether there is a matched AI historical policy; corresponding to the above, based on the AIRR information and the AI ​​business type information, it determines whether there is an AI historical policy corresponding to the AI ​​model policy query request, and obtains a second determination result.

[0376] It is noted that in this embodiment, the request sent by AIESF to PCF in step 114 and the request sent by AIESF to PCF in step 116 may be included in the above-mentioned AI model policy query request for the terminal, that is, two sub-requests belonging to the request; this embodiment is carried out in a manner of sending these two sub-requests separately, of course, they can also be sent together, which is not limited here.

[0377] Step 117. If the PCF matches a matching AI policy, it returns the AI ​​policy (AI business model and corresponding hyperparameters) to the AIESF; corresponding to the above-mentioned feedback of the AI ​​model policy query response for the terminal to the first core network device according to the second determination result; the AI ​​model response message carries AI model information; specifically, corresponding to the case where the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, according to the AI ​​historical policy, the feedback of the AI ​​model policy query response for the terminal to the first core network device; the AI ​​historical policy includes: model policy information and candidate model information. Specifically, the PCF can provide one or more matching AI policy Profile Lists (configuration lists) to the AIESF according to the matching results. The Profile List may include recorded computing resource overhead, storage resource overhead, model charges, model performance, training time and other information. Among them, the information contained in each AI strategy carried by the Profile List can assist AIESF in selecting an AI model after receiving the Profile List; for example: after receiving the Profile List, AIESF selects an AI model in combination with the AIRR and other information provided by the terminal; for example, if the AIRR requires the time to be within 100 milliseconds, AIESF can exclude the AI ​​model provided by PCF accordingly (exclude some models), and finally select one of them. In addition, the information contained in each AI strategy carried by the Profile List can also assist AIESF in providing feedback to the terminal, such as feedback on the estimated training time.

[0378] After receiving the AI ​​policy query response from PCF, AIESF integrates the user AI policy message and AI policy query response provided by PCF, and combines the resource overhead information and other information evaluated by AIESF to make an AI model selection decision among the multiple AI models provided by PCF; corresponding to the above-mentioned model policy information and candidate model information carried by the AI ​​model policy query response, the AI ​​model information is determined. Among them, the AI ​​policy query response may carry at least one of the information such as UE ID, candidate AI Model ID, and hyperparameters corresponding to the candidate AI Model ID.

[0379] Step 118. AIESF sends an AI model feedback message to the user, which may include information such as AI Model ID, AI model accuracy, and estimated training time; corresponding to the above, based on the AI ​​model information, an AI model response message is fed back to the terminal.

[0380] Example 2: The user does not specify an AI model, PCF does not retrieve the corresponding AI strategy, and does not have historical experience and knowledge;

[0381] This embodiment can adopt Fig.12 The architecture shown in the figure can be implemented as follows Fig.13 As shown, the following steps are included:

[0382] Step 131. The user sends an AI model discovery message to AIESF (corresponding to the above-mentioned sending of an AI model discovery message to the first core network device), where the message content may include user information (UE Information) and AI service requirements description (AI Requirements Description, AIRD). When the user is a mobile communication user, the user information may include user ID (such as UE ID), application information that triggers the AI ​​service, manufacturer identification and other information. When the user is a network element function NF within the network, the user information may include information such as NF Profile. Among them, AIRD may include problem description, data source (whether data collection is requested) and other information.

[0383] Step 132. AIESF analyzes the AI ​​model discovery message and determines the AI ​​service type according to AIRD; corresponding to the above, determining the AI ​​service type information according to the AIRD information.

[0384] AIESF responds to the user and sends an AI model discovery response to the user (corresponding to the above-mentioned feedback of the AI ​​model discovery response to the terminal), indicating that AIESF has received the AI ​​service demand (also understood as AI business demand) and classified the AI ​​type according to the demand description. The response content may include the AI ​​business type (corresponding to the above-mentioned AI business type information).

[0385] Step 133. The user sends an AI model request message (corresponding to the above-mentioned sending of an AI model request message to the first core network device), wherein the message content may include a user ID (such as UE ID), a data type (Data Type), a data size (DataSize), a data dimension (Data Dimension), and AI inference result requirements (AI Inference Result Requirements, AIRR). Among them, the AI ​​business model may be provided based on whether the user has historical experience and knowledge of AI training.

[0386] The AI ​​model request message of step 133 does not carry: relevant information of the applied AI business model (such as AIModel ID, which may correspond to the received AI business type information); specifically, the difference between step 133 of this embodiment and step 93 of embodiment 0 is that: in this embodiment, the AI ​​model request message does not provide the AI ​​Model ID, and no matching AI model is retrieved subsequently based on historical experience, and a comparison of AI models is required, see the following steps for details.

[0387] Step 134. AIESF queries the user AI model policy subscribed by the user to PCF (sends user AI policy subscription query information to) and sends an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message. The user AI policy subscription query information may carry the UE ID.

[0388] Step 135. PCF responds to the user AI model policy query and returns a user AI model policy message (corresponding to the above-mentioned receiving the AI ​​model policy query request for the terminal sent by the first core network device; according to the AI ​​model policy query request, feeding back the AI ​​model policy query response for the terminal to the first core network device). The user AI model policy message may include at least one of the AI ​​model type (AI Type) subscribed by the user, AI service priority (Priority), historical AI model selection information and hyperparameters (corresponding to History in the figure), computing resources and storage resource consumption limits (corresponding to Limitation in the figure), model charges, transmission overhead and other information; the user AI model policy message may also include UEID.

[0389] AIESF evaluates the computing time and the overhead of computing resources, storage resources, etc. according to the data type and data size in the AI ​​model request message (corresponding to the above-mentioned determination of the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead-related information; the overhead information includes: computing time information and at least one of computing resource overhead information and storage resource overhead information), and extracts data features. It can also be understood that AIESF evaluates the AI ​​model request information (corresponding to the AI ​​model request message) and generates information such as data features and overhead.

[0390] Step 136. AIESF sends an AI policy query request to PCF, which may include at least one of the following information: user ID (i.e., UE ID), data features, AI inference result requirements (AIRR), and AI service type; the request may be used to trigger PCF to execute: determine whether there is historical experience (AI model-related experience information) corresponding to the AI ​​service generated by the above user based on the above information.

[0391] The PCF retrieves historical experience and knowledge that matches the AIRR and AI business type based on the information in the AI ​​policy query request provided in step 136, and determines whether there is a matched AI historical policy; corresponding to the above, based on the AIRR information and the AI ​​business type information, it determines whether there is an AI historical policy corresponding to the AI ​​model policy query request, and obtains a second determination result.

[0392] It is explained here that in this embodiment, the request sent by AIESF to PCF in step 134 and the request sent by AIESF to PCF in step 136 can be included in the above-mentioned AI model policy query request for the terminal, that is, two sub-requests belonging to the request; this embodiment is carried out in a manner of sending these two sub-requests separately, of course, they can also be sent together, which is not limited here.

[0393] Step 137. PCF does not match a matching AI policy, and returns a null value Void to AIESF, indicating that no AI model has been retrieved (i.e., an AI policy query response carrying a null value is returned); corresponding to the above-mentioned situation where the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device. The AI ​​policy query response may also carry a UE ID.

[0394] After AIESF receives the message that the AI ​​strategy is empty, it triggers the AI ​​model selection process.

[0395] Step 138. AIESF queries the corresponding "Table 1-1 Comparison Table of AI Business Types and Matching Business Models and Evaluation Indicators" according to the determined AI business type, and sends an AI model selection request to AIMSF, which may include the content in the AI ​​model request message in step 133, as well as the AI ​​business type and the corresponding multiple AI business model IDs; corresponding to the above-mentioned situation where the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, an AI model training request is sent to the third core network device; the AI ​​model training request carries the AI ​​business type information and candidate AI model information.

[0396] AIMSF trains multiple AI models in parallel, calculates weighted training scores, and generates a Profile List; specifically, it may include: trying multiple AI models in parallel, starting multiple training processes to train different AI models at the same time, and evaluating the output results of each AI model according to the evaluation indicators provided in Table 1-1; corresponding to the above, according to the AI ​​business type information and the candidate AI model information, the candidate AI model is trained to obtain the training results. Furthermore, the importance ranking of the evaluation indicators can be determined according to the requirements of the AI ​​reasoning results, and corresponding weights are assigned to each evaluation indicator. The evaluation indicator results calculated by each model are multiplied by the corresponding weights, and they are added up (that is, the indicators corresponding to each model multiplied by the weights are added up) to obtain the comprehensive training score (of each model) and generate the AI ​​model Profile List; corresponding to the above, the weights of the evaluation indicators are determined according to the AIRR information; the evaluation results are obtained according to the weights and training results.

[0397] Step 139. AIMSF stores the AI ​​model and hyperparameter data of this training. AIMSF also reports the AI ​​strategy information to PCF (corresponding to the AI ​​model training information corresponding to the AI ​​model training request sent to the second core network device). The AI ​​strategy information may include the user ID (corresponding to the terminal identifier), the optimal AI business model ID (corresponding to the AI ​​model request message), and the corresponding hyperparameter group ID (AI business model ID), so as to be saved as historical experience and knowledge for quick retrieval later.

[0398] Step 1310. AIMSF selects the top three models with comprehensive training scores and sends the AI ​​model ProfileList to AIESF; corresponding to the above, based on the evaluation results, an AI model training response is fed back to the first core network device.

[0399] After AIESF receives the AI ​​model Profile List sent by AIMSF, it makes an AI model selection decision based on the Profile List, the user AI policy message provided by PCF, the overhead information evaluated by AIESF, AIRR and other information; corresponding to the above, the AI ​​model information is determined based on the overhead information and the AI ​​model training response.

[0400] Step 1311. AIESF sends an AI model feedback message to the user, which may include information such as AI Model ID, AI model accuracy, and estimated training time; corresponding to the above, based on the AI ​​model information, an AI model response message is fed back to the terminal.

[0401] From the above, Example 1 and Example 2 are two situations: Example 1 is to retrieve one or more AI models based on the historical experience obtained by PCF, and AIESF selects the model among them; Example 2 is that no matching historical experience is retrieved, and AIMSF is required to perform model comparison; the processes after step 7 of the two are different.

[0402] It is to be noted that the relevant contents of the above embodiments can be referenced to each other, and the repeated contents will not be repeated here.

[0403] From the above, the embodiment of the present application: in view of the 6G network intelligent endogenous architecture and the vision of intelligent inclusiveness, a method of AI model discovery and on-demand selection is proposed; specifically, by adding two network functions, AIESF and AIMSF, the proposed method can not only serve the network element functions within the network, but also provide AI services to mobile communication users UE to realize AI model discovery, and the proposed method can also realize on-demand selection of AI models in combination with information such as business needs. Based on this, the method provided in the embodiment of the present application is conducive to the realization of 6G network intelligent endogenous, and can flexibly respond to different business needs and scenarios to select a variety of AI types and models on demand.

[0404] In summary, the solution provided in the embodiment of the present application has the following advantages:

[0405] (1) For the future 6G intelligent endogenous architecture, it can support AI model discovery and make on-demand AI model selection decisions based on user-provided data and AI business needs.

[0406] (2) The AI ​​capabilities of the network have been expanded. The proposed AIESF and AIMSF with AI capabilities can not only serve the internal network but also provide AI services to the outside world.

[0407] (3) The flexibility and efficiency of AI model selection have been improved. A search table for different AI types has been established to store new AI training and reasoning strategies and corresponding hyperparameters. This allows for flexible and efficient on-demand selection of multiple AI types and models to meet different business needs and scenarios.

[0408] The embodiment of the present application also provides an information transmission device, wherein the information transmission device is a first core network device, such as Fig.14 As shown, the device includes a memory 141, a transceiver 142, and a processor 143:

[0409] The memory 141 is used to store computer programs; the transceiver 142 is used to send and receive data under the control of the processor 143; the processor 143 is used to read the computer program in the memory 141 and perform the following operations:

[0410] Receiving, through the transceiver 142, an artificial intelligence AI model request message sent by a terminal;

[0411] According to the AI ​​model request message, sending an AI model policy query request for the terminal to the second core network device through the transceiver 142;

[0412] Receiving, through the transceiver 142, an AI model policy query response for the terminal fed back by the second core network device;

[0413] According to the AI ​​model strategy query response, an AI model response message is fed back to the terminal; the AI ​​model response message carries AI model information.

[0414] The information transmission device provided in the embodiment of the present application receives an artificial intelligence (AI) model request message sent by a terminal; sends an AI model policy query request for the terminal to a second core network device according to the AI ​​model request message; receives an AI model policy query response for the terminal fed back by the second core network device; and feeds back an AI model response message to the terminal according to the AI ​​model policy query response; the AI ​​model response message carries AI model information; can support the provision of corresponding AI model information to the terminal according to the AI ​​model request message, and further support the provision of AI models based on user needs; and well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0415] Specifically, the transceiver 142 is used to receive and send data under the control of the processor 143.

[0416] Among them, Fig.14 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 143 and various circuits of memory represented by memory 141 are linked together. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 142 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 may include a wireless channel, a wired channel, an optical cable, and other transmission media. The processor 143 is responsible for managing the bus architecture and general processing, and the memory 141 may store data used by the processor 143 when performing operations.

[0417] The processor 143 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.

[0418] Furthermore, the operation also includes: before receiving the artificial intelligence AI model request message sent by the terminal, receiving the AI ​​model discovery message sent by the terminal through the transceiver; the AI ​​model discovery message carries the artificial intelligence service demand description AIRD information; according to the AIRD information, determining the AI ​​service type information, and feeding back the AI ​​model discovery response to the terminal through the transceiver.

[0419] Among them, the AI ​​model request message carries the identification information of the AI ​​business model applied for by the terminal; and the AI ​​model response message is fed back to the terminal according to the AI ​​model policy query response, including: determining whether the AI ​​business model applied for by the terminal is within the subscription model range corresponding to the terminal according to the AI ​​model policy query response and the identification information, and obtaining a first determination result; obtaining AI model information according to the first determination result; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0420] In an embodiment of the present application, obtaining AI model information according to the first determination result includes: when the first determination result indicates that the AI ​​business model applied for by the terminal is not within the subscription model range corresponding to the terminal, determining the AI ​​model information corresponding to the terminal according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0421] The AI ​​model request message also carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the operation also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead related information; the overhead information includes: at least one of calculation time information, calculation resource overhead information and storage resource overhead information; when the first determination result indicates that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal, determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response, including: when the first determination result indicates the first situation, or the first determination result indicates the first situation and the overhead information does not meet the overhead limit information, determining the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response; wherein the first situation refers to the situation that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal.

[0422] In an embodiment of the present application, the AI ​​model request message carries AIRR information requiring artificial intelligence reasoning results; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0423] Among them, the feeding back an AI model response message to the terminal according to the AI ​​model policy query response includes: determining AI model information according to model policy information and candidate model information carried by the AI ​​model policy query response; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0424] In an embodiment of the present application, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of: data type information, data size information, and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the operation also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message based on the overhead-related information; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information based on the model policy information and candidate model information carried by the AI ​​model policy query response includes: determining the AI ​​model information based on the overhead information, the model policy information and candidate model information carried by the AI ​​model policy query response.

[0425] Among them, the feeding back an AI model response message to the terminal according to the AI ​​model policy query response includes: when the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device through the transceiver; the AI ​​model training request carries the AI ​​service type information and candidate AI model information; receiving an AI model training response fed back by the third core network device through the transceiver; determining AI model information according to the AI ​​model training response; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0426] In an embodiment of the present application, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of data type information, data size information, and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the operation also includes: determining the overhead information of the AI ​​service corresponding to the AI ​​model request message based on the overhead-related information; the overhead information includes: at least one of computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information based on the AI ​​model training response includes: determining the AI ​​model information based on the overhead information and the AI ​​model training response.

[0427] 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 first core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0428] The embodiment of the present application also provides an information transmission device, wherein the information transmission device is a second core network device, such as Fig.15 As shown, it includes a memory 151, a transceiver 152, and a processor 153:

[0429] The memory 151 is used to store computer programs; the transceiver 152 is used to send and receive data under the control of the processor 153; the processor 153 is used to read the computer program in the memory 151 and perform the following operations:

[0430] Receiving, through the transceiver 152, an AI model policy query request for the terminal sent by the first core network device;

[0431] According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

[0432] The information transmission device provided in the embodiment of the present application receives an AI model policy query request for a terminal sent by a first core network device; based on the AI ​​model policy query request, it feeds back an AI model policy query response for the terminal to the first core network device; it can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; it well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0433] Specifically, the transceiver 152 is used to receive and send data under the control of the processor 153.

[0434] Among them, Fig.15 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 153 and various circuits of memory represented by memory 151 are linked together. The bus architecture may also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 152 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 may include a wireless channel, a wired channel, an optical cable, and other transmission media. The processor 153 is responsible for managing the bus architecture and general processing, and the memory 151 may store data used by the processor 153 when performing operations.

[0435] The processor 153 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.

[0436] Among them, the AI ​​model policy query request carries the terminal identifier of the terminal; and according to the AI ​​model policy query request, the AI ​​model policy query response for the terminal is fed back to the first core network device, including: according to the terminal identifier, the AI ​​model policy query response for the terminal is fed back to the first core network device.

[0437] In an embodiment of the present application, the AI ​​model policy query request carries AIRR information and AI service type information; the AI ​​model policy query response for the terminal is fed back to the first core network device based on the AI ​​model policy query request, including: determining whether there is an AI historical policy corresponding to the AI ​​model policy query request based on the AIRR information and the AI ​​service type information, and obtaining a second determination result; based on the second determination result, the AI ​​model policy query response for the terminal is fed back to the first core network device.

[0438] Among them, the feeding back of the AI ​​model policy query response for the terminal to the first core network device based on the second determination result includes: (1) when the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back the AI ​​model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information; (2) when the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, feeding back to the first core network device an AI model policy query response indicating that there is no model policy information corresponding to the terminal.

[0439] 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 second core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0440] The embodiment of the present application also provides an information transmission device, wherein the information transmission device is a terminal, such as Fig.16 As shown, it includes a memory 161, a transceiver 162, and a processor 163:

[0441] The memory 161 is used to store computer programs; the transceiver 162 is used to send and receive data under the control of the processor 163; the processor 163 is used to read the computer program in the memory 161 and perform the following operations:

[0442] Sending an AI model request message to the first core network device through the transceiver 162;

[0443] The transceiver 162 is used to receive an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

[0444] The information transmission device provided in the embodiment of the present application sends an AI model request message to the first core network device; receives an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information; it can support the acquisition of AI model information corresponding to the AI ​​model request message, and then support the provision of AI models based on user needs; it is a good solution to the problem that the prior art cannot support the provision of AI models based on user needs.

[0445] Specifically, the transceiver 162 is used to receive and send data under the control of the processor 163.

[0446] Among them, Fig.16 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 163 and various circuits of memory represented by memory 161 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 herein. The bus interface provides an interface. The transceiver 162 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, such as a wireless channel, a wired channel, an optical cable, and other transmission media. For different user devices, the user interface 164 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.

[0447] The processor 163 is responsible for managing the bus architecture and general processing, and the memory 161 can store data used by the processor 163 when performing operations.

[0448] Optionally, the processor 163 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), and the processor may also adopt a multi-core architecture.

[0449] The processor calls the computer program stored in the memory to execute any of the methods provided in the embodiments of the present application according to the obtained executable instructions. The processor and the memory can also be arranged physically separately.

[0450] Furthermore, the operation also includes: before sending an AI model request message to the first core network device, sending an AI model discovery message to the first core network device through the transceiver; the AI ​​model discovery message carries AIRD information; and receiving an AI model discovery response fed back by the first core network device through the transceiver.

[0451] Among them, the AI ​​model request message carries identification information of the AI ​​business model applied for by the terminal; and / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; and / or, the AI ​​model request message carries AIRR information.

[0452] 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 terminal side method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0453] The embodiment of the present application also provides an information transmission device, wherein the information transmission device is a third core network device, such as Fig.17 As shown, the device includes a memory 171, a transceiver 172, and a processor 173:

[0454] The memory 171 is used to store computer programs; the transceiver 172 is used to send and receive data under the control of the processor 173; the processor 173 is used to read the computer program in the memory 171 and perform the following operations:

[0455] Receiving, through the transceiver 172, an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0456] According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device through the transceiver 172.

[0457] The information transmission device provided in the embodiment of the present application receives an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; according to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device; it can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; it well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0458] Specifically, the transceiver 172 is used to receive and send data under the control of the processor 173.

[0459] Among them, Fig.17 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 173 and various circuits of memory represented by memory 171 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 herein. The bus interface provides an interface. The transceiver 172 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 may include a wireless channel, a wired channel, an optical cable, and the like. The processor 173 is responsible for managing the bus architecture and general processing, and the memory 171 may store data used by the processor 173 when performing operations.

[0460] The processor 173 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.

[0461] Among them, the feeding back an AI model training response to the first core network device according to the AI ​​model training request includes: training the candidate AI model according to the AI ​​service type information and the candidate AI model information, and obtaining an evaluation result; and feeding back an AI model training response to the first core network device according to the evaluation result.

[0462] Furthermore, the AI ​​model training request also carries AIRR information; the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and an evaluation result is obtained, including: the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and a training result is obtained; the weight of the evaluation index is determined according to the AIRR information; and the evaluation result is obtained according to the weight and the training result.

[0463] 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 third core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0464] The embodiment of the present application also provides an information transmission device, which is applied to a first core network device, such as Fig.18 As shown, the device comprises:

[0465] The first receiving unit 181 is used to receive an artificial intelligence AI model request message sent by a terminal;

[0466] A first sending unit 182 is configured to send an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message;

[0467] The second receiving unit 183 is configured to receive an AI model policy query response for the terminal fed back by the second core network device;

[0468] The first feedback unit 184 is configured to feed back an AI model response message to the terminal according to the AI ​​model policy query response; the AI ​​model response message carries AI model information.

[0469] The information transmission device provided in the embodiment of the present application receives an artificial intelligence (AI) model request message sent by a terminal; sends an AI model policy query request for the terminal to a second core network device according to the AI ​​model request message; receives an AI model policy query response for the terminal fed back by the second core network device; and feeds back an AI model response message to the terminal according to the AI ​​model policy query response; the AI ​​model response message carries AI model information; and can support the provision of corresponding AI model information to the terminal according to the AI ​​model request message, thereby supporting the provision of AI models based on user needs; and well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0470] Furthermore, the information transmission device also includes: a third receiving unit, used to receive an AI model discovery message sent by the terminal before receiving the artificial intelligence AI model request message sent by the terminal; the AI ​​model discovery message carries artificial intelligence service demand description AIRD information; a first processing unit, used to determine the AI ​​service type information according to the AIRD information, and feedback the AI ​​model discovery response to the terminal.

[0471] Among them, the AI ​​model request message carries the identification information of the AI ​​business model applied for by the terminal; and the AI ​​model response message is fed back to the terminal according to the AI ​​model policy query response, including: determining whether the AI ​​business model applied for by the terminal is within the subscription model range corresponding to the terminal according to the AI ​​model policy query response and the identification information, and obtaining a first determination result; obtaining AI model information according to the first determination result; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0472] In an embodiment of the present application, obtaining AI model information according to the first determination result includes: when the first determination result indicates that the AI ​​business model applied for by the terminal is not within the subscription model range corresponding to the terminal, determining the AI ​​model information corresponding to the terminal according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

[0473] The AI ​​model request message also carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission device also includes: a first determination unit, used to determine the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead related information; the overhead information includes: at least one of calculation time information, calculation resource overhead information and storage resource overhead information; when the first determination result indicates that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal, the AI ​​model information is determined according to the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response, including: when the first determination result indicates the first situation, or the first determination result indicates the first situation and the overhead information does not meet the overhead limit information, the AI ​​model information is determined according to the overhead information, the AI ​​model request message, the locally stored AI model information and the model policy information carried by the AI ​​model policy query response; wherein the first situation refers to the situation that the AI ​​service model applied by the terminal is not within the subscription model range corresponding to the terminal.

[0474] In an embodiment of the present application, the AI ​​model request message carries AIRR information requiring artificial intelligence reasoning results; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

[0475] Among them, the feeding back an AI model response message to the terminal according to the AI ​​model policy query response includes: determining AI model information according to model policy information and candidate model information carried by the AI ​​model policy query response; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0476] In an embodiment of the present application, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of: data type information, data size information, and data dimension information; the model policy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission device also includes: a second determination unit, used to determine the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead-related information; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information according to the model policy information carried by the AI ​​model policy query response and the candidate model information includes: determining the AI ​​model information according to the overhead information, the model policy information carried by the AI ​​model policy query response, and the candidate model information.

[0477] Among them, the feeding back an AI model response message to the terminal according to the AI ​​model policy query response includes: when the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device; the AI ​​model training request carries the AI ​​service type information and candidate AI model information; receiving an AI model training response fed back by the third core network device; determining the AI ​​model information according to the AI ​​model training response; and feeding back an AI model response message to the terminal according to the AI ​​model information.

[0478] In an embodiment of the present application, the AI ​​model request message also carries overhead-related information; the overhead-related information includes: at least one of data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model policy query response includes: overhead limit information; the information transmission device also includes: a third determination unit, used to determine the overhead information of the AI ​​service corresponding to the AI ​​model request message according to the overhead-related information; the overhead information includes: at least one of computing time information, computing resource overhead information, and storage resource overhead information; determining the AI ​​model information based on the AI ​​model training response includes: determining the AI ​​model information based on the overhead information and the AI ​​model training response.

[0479] 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 by the above-mentioned first core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0480] The embodiment of the present application also provides an information transmission device, which is applied to a second core network device, such as Fig.19 As shown, including:

[0481] The fourth receiving unit 191 is used to receive an AI model policy query request for the terminal sent by the first core network device;

[0482] The second feedback unit 192 is used to feedback an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request.

[0483] The information transmission device provided in the embodiment of the present application receives an AI model policy query request for a terminal sent by a first core network device; based on the AI ​​model policy query request, it feeds back an AI model policy query response for the terminal to the first core network device; it can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; it well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0484] Among them, the AI ​​model policy query request carries the terminal identifier of the terminal; and according to the AI ​​model policy query request, the AI ​​model policy query response for the terminal is fed back to the first core network device, including: according to the terminal identifier, the AI ​​model policy query response for the terminal is fed back to the first core network device.

[0485] In an embodiment of the present application, the AI ​​model policy query request carries AIRR information and AI service type information; the AI ​​model policy query response for the terminal is fed back to the first core network device based on the AI ​​model policy query request, including: determining whether there is an AI historical policy corresponding to the AI ​​model policy query request based on the AIRR information and the AI ​​service type information, and obtaining a second determination result; based on the second determination result, the AI ​​model policy query response for the terminal is fed back to the first core network device.

[0486] Among them, the feeding back of the AI ​​model policy query response for the terminal to the first core network device based on the second determination result includes: (1) when the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back the AI ​​model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information; (2) when the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, feeding back to the first core network device an AI model policy query response indicating that there is no model policy information corresponding to the terminal.

[0487] 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 second core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0488] The present application also provides an information transmission device, which is applied to a terminal, such as Fig. 20 As shown, including:

[0489] The second sending unit 201 is used to send an AI model request message to the first core network device;

[0490] The fifth receiving unit 202 is used to receive the AI ​​model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

[0491] The information transmission device provided in the embodiment of the present application sends an AI model request message to the first core network device; receives an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information; it can support the acquisition of AI model information corresponding to the AI ​​model request message, and then support the provision of AI models based on user needs; it is a good solution to the problem that the prior art cannot support the provision of AI models based on user needs.

[0492] Furthermore, the information transmission device also includes: a third sending unit, used to send an AI model discovery message to the first core network device before sending an AI model request message to the first core network device; the AI ​​model discovery message carries AIRD information; and a sixth receiving unit, used to receive an AI model discovery response fed back by the first core network device.

[0493] Among them, the AI ​​model request message carries identification information of the AI ​​business model applied for by the terminal; and / or, the AI ​​model request message carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; and / or, the AI ​​model request message carries AIRR information.

[0494] 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 terminal side method embodiment, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0495] The embodiment of the present application also provides an information transmission device, which is applied to a third core network device, such as Fig.21As shown, the device comprises:

[0496] The seventh receiving unit 211 is used to receive an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information;

[0497] The third feedback unit 212 is used to feedback an AI model training response to the first core network device according to the AI ​​model training request, and to send AI model training information corresponding to the AI ​​model training request to the second core network device.

[0498] The information transmission device provided in the embodiment of the present application receives an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; according to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device; it can support the provision of AI model information corresponding to the AI ​​model request message to the terminal, and further support the provision of AI models based on user needs; it well solves the problem that the prior art cannot support the provision of AI models based on user needs.

[0499] Among them, the feeding back an AI model training response to the first core network device according to the AI ​​model training request includes: training the candidate AI model according to the AI ​​service type information and the candidate AI model information, and obtaining an evaluation result; and feeding back an AI model training response to the first core network device according to the evaluation result.

[0500] In an embodiment of the present application, the AI ​​model training request also carries AIRR information; the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and an evaluation result is obtained, including: the candidate AI model training is performed according to the AI ​​business type information and the candidate AI model information, and the training result is obtained; the weight of the evaluation index is determined according to the AIRR information; and the evaluation result is obtained according to the weight and the training result.

[0501] 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 third core network device side method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0502] 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.

[0503] 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.

[0504] An embodiment of the present application also provides a non-transitory readable storage medium, which stores a computer program, and the computer program is used to enable the processor to execute the above-mentioned method on the first core network device side, the second core network device side, the terminal side or the third core network device side.

[0505] The non-transitory readable storage medium can be any available medium or data storage device that can be accessed by the processor, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical (MO)), etc.), optical storage (such as compact discs (CD), digital video discs (DVD), Blu-ray discs (BD), high-definition versatile discs (HVD), etc.), and semiconductor memories (such as ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read only memory (EEPROM), non-volatile memory (NAND (Non-volatile Memory Device) FLASH), solid-state drives (SSD)), etc.

[0506] Among them, the implementation embodiments of the methods on the first core network device side, the second core network device side, the terminal side or the third core network device side are all applicable to the embodiments of the non-transitory readable storage medium, and can achieve the same technical effect.

[0507] 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.

[0508] 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 1A device that provides the functions specified in a block or multiple blocks.

[0509] 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.

[0510] 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.

[0511] 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 transmitting information, It is characterized in that Applied to a first core network device, the method includes: Receive the artificial intelligence AI model request message sent by the terminal; Sending an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message; Receiving an AI model policy query response for the terminal fed back by the second core network device; According to the AI ​​model strategy query response, an AI model response message is fed back to the terminal; the AI ​​model response message carries AI model information.

2. The information transmission method according to claim 1, It is characterized in that Before receiving the AI ​​model request message sent by the terminal, it also includes: Receiving an AI model discovery message sent by the terminal; the AI ​​model discovery message carries AIRD information of an artificial intelligence service requirement description; According to the AIRD information, the AI ​​service type information is determined, and an AI model discovery response is fed back to the terminal.

3. The information transmission method according to claim 1, It is characterized in that The AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal; The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: Determine, according to the AI ​​model policy query response and the identification information, whether the AI ​​service model applied for by the terminal is within the subscription model range corresponding to the terminal, to obtain a first determination result; Acquire AI model information according to the first determination result; According to the AI ​​model information, an AI model response message is fed back to the terminal.

4. The information transmission method according to claim 3, It is characterized in that The obtaining AI model information according to the first determination result includes: When the first determination result indicates that the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

5. The information transmission method according to claim 4, It is characterized in that The AI ​​model request message also carries overhead related information; The overhead related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The information transmission method further includes: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The step of determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response, when the first determination result indicates that the AI ​​service model applied for by the terminal is not within the subscription model range corresponding to the terminal, includes: When the first determination result indicates a first situation, or when the first determination result indicates the first situation and the overhead information does not satisfy the overhead limit information, determine the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response; The first situation refers to a situation where the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal.

6. The information transmission method according to claim 1 or 2, It is characterized in that The AI ​​model request message carries AIRR information of artificial intelligence reasoning result requirement; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

7. The information transmission method according to claim 6, It is characterized in that The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: Determine the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response; According to the AI ​​model information, an AI model response message is fed back to the terminal.

8. The information transmission method according to claim 7, It is characterized in that The AI ​​model request message also carries overhead related information; The overhead related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The information transmission method further includes: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The determining the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response includes: Determine the AI ​​model information according to the overhead information, the model strategy information carried by the AI ​​model strategy query response, and the candidate model information.

9. The information transmission method according to claim 6, It is characterized in that The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: When the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device; the AI ​​model training request carries the AI ​​service type information and the candidate AI model information; Receiving an AI model training response fed back by the third core network device; Determining AI model information according to the AI ​​model training response; According to the AI ​​model information, an AI model response message is fed back to the terminal.

10. The information transmission method according to claim 9, It is characterized in that The AI ​​model request message also carries overhead related information; The overhead related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The information transmission method further includes: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The determining the AI ​​model information according to the AI ​​model training response includes: Determine AI model information based on the overhead information and the AI ​​model training response.

11. An information transmission method, applied to a second core network device, It is characterized in that include: Receiving an AI model policy query request for a terminal sent by a first core network device; According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

12. The information transmission method according to claim 11, It is characterized in that The AI ​​model strategy query request carries the terminal identifier of the terminal; The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes: According to the terminal identifier, an AI model policy query response for the terminal is fed back to the first core network device.

13. The information transmission method according to claim 11, It is characterized in that The AI ​​model strategy query request carries AIRR information and AI service type information; The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes: Determine, according to the AIRR information and the AI ​​business type information, whether there is an AI historical policy corresponding to the AI ​​model policy query request, to obtain a second determination result; Based on the second determination result, an AI model policy query response for the terminal is fed back to the first core network device.

14. The information transmission method according to claim 13, It is characterized in that The feeding back an AI model policy query response for the terminal to the first core network device according to the second determination result includes: If the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information; When the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device.

15. An information transmission method, applied to a terminal, It is characterized in that include: Sending an AI model request message to the first core network device; Receive an AI model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

16. The information transmission method according to claim 15, It is characterized in that Before sending the AI ​​model request message to the first core network device, the method further includes: Sending an AI model discovery message to the first core network device; the AI ​​model discovery message carries AIRD information; Receive the AI ​​model discovery response fed back by the first core network device.

17. The information transmission method according to claim 15, It is characterized in that The AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal; And / or, the AI ​​model request message carries overhead related information; The overhead related information includes: at least one of data type information, data size information, and data dimension information; And / or, the AI ​​model request message carries AIRR information.

18. A method for transmitting information, It is characterized in that Applied to a third core network device, the method includes: Receive an AI model training request sent by a first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device.

19. The information transmission method according to claim 18, It is characterized in that The feeding back an AI model training response to the first core network device according to the AI ​​model training request includes: According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed, and evaluation results are obtained; Based on the evaluation result, an AI model training response is fed back to the first core network device.

20. The information transmission method according to claim 19, It is characterized in that The AI ​​model training request also carries AIRR information; The performing candidate AI model training according to the AI ​​business type information and the candidate AI model information, and obtaining an evaluation result, includes: According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed to obtain a training result; Determining the weight of the evaluation index according to the AIRR information; An evaluation result is obtained according to the weights and the training results.

21. An information transmission device, It is characterized in that The information transmission device is a first core network device, and the device includes a memory, a transceiver, and a 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: Receiving, through the transceiver, an artificial intelligence AI model request message sent by a terminal; According to the AI ​​model request message, sending an AI model policy query request for the terminal to the second core network device through the transceiver; Receiving, by the transceiver, an AI model policy query response for the terminal fed back by the second core network device; According to the AI ​​model strategy query response, an AI model response message is fed back to the terminal; the AI ​​model response message carries AI model information.

22. The information transmission device according to claim 21, It is characterized in that The operations also include: Before receiving the artificial intelligence AI model request message sent by the terminal, receiving the AI ​​model discovery message sent by the terminal through the transceiver; the AI ​​model discovery message carries the artificial intelligence service requirement description AIRD information; According to the AIRD information, the AI ​​service type information is determined, and an AI model discovery response is fed back to the terminal through the transceiver.

23. The information transmission device according to claim 21, It is characterized in that The AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal; The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: Determine, according to the AI ​​model policy query response and the identification information, whether the AI ​​service model applied for by the terminal is within the subscription model range corresponding to the terminal, to obtain a first determination result; Acquire AI model information according to the first determination result; According to the AI ​​model information, an AI model response message is fed back to the terminal.

24. The information transmission device according to claim 23, It is characterized in that The obtaining AI model information according to the first determination result includes: When the first determination result indicates that the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal, the AI ​​model information corresponding to the terminal is determined according to the AI ​​model request message, the stored AI model information and the model policy information carried by the AI ​​model policy query response.

25. The information transmission device according to claim 24, It is characterized in that The AI ​​model request message also carries overhead related information; The overhead related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The operations also include: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The step of determining the AI ​​model information according to the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response, when the first determination result indicates that the AI ​​service model applied for by the terminal is not within the subscription model range corresponding to the terminal, includes: When the first determination result indicates a first situation, or when the first determination result indicates the first situation and the overhead information does not satisfy the overhead limit information, determine the AI ​​model information according to the overhead information, the AI ​​model request message, the locally stored AI model information, and the model policy information carried by the AI ​​model policy query response; The first situation refers to a situation where the AI ​​service model applied for by the terminal is not within the scope of the subscription model corresponding to the terminal.

26. The information transmission device according to claim 21 or 22, It is characterized in that The AI ​​model request message carries AIRR information of artificial intelligence reasoning result requirement; and the AI ​​model strategy query request carries the AIRR information and AI business type information.

27. The information transmission device according to claim 26, It is characterized in that The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: Determine the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response; According to the AI ​​model information, an AI model response message is fed back to the terminal.

28. The information transmission device according to claim 27, It is characterized in that The AI ​​model request message also carries overhead related information; the overhead related information includes: at least one of data type information, data size information and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The operations also include: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The determining the AI ​​model information according to the model strategy information and the candidate model information carried by the AI ​​model strategy query response includes: Determine the AI ​​model information according to the overhead information, the model strategy information carried by the AI ​​model strategy query response, and the candidate model information.

29. The information transmission device according to claim 26, It is characterized in that The feeding back an AI model response message to the terminal according to the AI ​​model strategy query response includes: When the AI ​​model policy query response indicates that there is no model policy information corresponding to the terminal, sending an AI model training request to a third core network device through the transceiver; the AI ​​model training request carries the AI ​​service type information and the candidate AI model information; Receiving, through the transceiver, an AI model training response fed back by the third core network device; Determining AI model information according to the AI ​​model training response; According to the AI ​​model information, an AI model response message is fed back to the terminal.

30. The information transmission device according to claim 29, It is characterized in that The AI ​​model request message also carries overhead related information; The overhead related information includes: at least one of: data type information, data size information, and data dimension information; the model strategy information carried by the AI ​​model strategy query response includes: overhead limit information; The operations also include: Determine, according to the overhead related information, the overhead information of the AI ​​service corresponding to the AI ​​model request message; the overhead information includes: at least one of: computing time information, computing resource overhead information, and storage resource overhead information; The determining the AI ​​model information according to the AI ​​model training response includes: Determine AI model information based on the overhead information and the AI ​​model training response.

31. An information transmission device, wherein the information transmission device is a second core network device, It is 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: Receiving, by the transceiver, an AI model policy query request for the terminal sent by the first core network device; According to the AI ​​model policy query request, an AI model policy query response for the terminal is fed back to the first core network device.

32. The information transmission device according to claim 31, It is characterized in that The AI ​​model strategy query request carries the terminal identifier of the terminal; The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes: According to the terminal identifier, an AI model policy query response for the terminal is fed back to the first core network device.

33. The information transmission device according to claim 31, It is characterized in that The AI ​​model strategy query request carries AIRR information and AI service type information; The feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request includes: Determine, according to the AIRR information and the AI ​​business type information, whether there is an AI historical policy corresponding to the AI ​​model policy query request, to obtain a second determination result; Based on the second determination result, an AI model policy query response for the terminal is fed back to the first core network device.

34. The information transmission device according to claim 33, It is characterized in that The feeding back an AI model policy query response for the terminal to the first core network device according to the second determination result includes: If the second determination result indicates that there is an AI historical policy corresponding to the AI ​​model policy query request, feeding back an AI model policy query response for the terminal to the first core network device according to the AI ​​historical policy; the AI ​​historical policy includes: model policy information and candidate model information; When the second determination result indicates that there is no AI historical policy corresponding to the AI ​​model policy query request, an AI model policy query response indicating that there is no model policy information corresponding to the terminal is fed back to the first core network device.

35. An information transmission device, the information transmission device being a terminal, It is 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: Sending an AI model request message to the first core network device through the transceiver; An AI model response message fed back by the first core network device is received through the transceiver; the AI ​​model response message carries AI model information.

36. The information transmission device according to claim 35, It is characterized in that The operations also include: Before sending the AI ​​model request message to the first core network device, sending an AI model discovery message to the first core network device through the transceiver; the AI ​​model discovery message carries AIRD information; An AI model discovery response fed back by the first core network device is received through the transceiver.

37. The information transmission device according to claim 35, It is characterized in that The AI ​​model request message carries identification information of the AI ​​service model applied for by the terminal; And / or, the AI ​​model request message carries overhead related information; The overhead related information includes: at least one of data type information, data size information, and data dimension information; And / or, the AI ​​model request message carries AIRR information.

38. An information transmission device, It is characterized in that The information transmission device is a third core network device, and the device includes a memory, a transceiver, and a 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: Receiving, by the transceiver, an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; According to the AI ​​model training request, an AI model training response is fed back to the first core network device, and AI model training information corresponding to the AI ​​model training request is sent to the second core network device through the transceiver.

39. The information transmission device according to claim 38, It is characterized in that The feeding back an AI model training response to the first core network device according to the AI ​​model training request includes: According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed, and evaluation results are obtained; Based on the evaluation result, an AI model training response is fed back to the first core network device.

40. The information transmission device according to claim 39, It is characterized in that The AI ​​model training request also carries AIRR information; The performing candidate AI model training according to the AI ​​business type information and the candidate AI model information, and obtaining an evaluation result, includes: According to the AI ​​business type information and the candidate AI model information, candidate AI model training is performed to obtain a training result; Determining the weight of the evaluation index according to the AIRR information; An evaluation result is obtained according to the weights and the training results.

41. An information transmission device, It is characterized in that Applied to a first core network device, the apparatus includes: A first receiving unit, configured to receive an artificial intelligence AI model request message sent by a terminal; A first sending unit, configured to send an AI model policy query request for the terminal to the second core network device according to the AI ​​model request message; A second receiving unit, configured to receive an AI model policy query response for the terminal fed back by the second core network device; The first feedback unit is used to feed back an AI model response message to the terminal according to the AI ​​model strategy query response; the AI ​​model response message carries AI model information.

42. An information transmission device, applied to a second core network device, It is characterized in that include: A fourth receiving unit, configured to receive an AI model policy query request for a terminal sent by the first core network device; The second feedback unit is used to feedback an AI model policy query response for the terminal to the first core network device according to the AI ​​model policy query request.

43. An information transmission device, applied to a terminal, It is characterized in that include: A second sending unit, configured to send an AI model request message to the first core network device; The fifth receiving unit is used to receive the AI ​​model response message fed back by the first core network device; the AI ​​model response message carries AI model information.

44. An information transmission device, It is characterized in that Applied to a third core network device, the device includes: A seventh receiving unit, configured to receive an AI model training request sent by the first core network device; the AI ​​model training request carries AI service type information and candidate AI model information; The third feedback unit is used to feedback an AI model training response to the first core network device according to the AI ​​model training request, and to send AI model training information corresponding to the AI ​​model training request to the second core network device.

45. A non-transitory readable storage medium, It is characterized in that The non-transitory readable storage medium stores a computer program, and the computer program is used to enable a processor to execute the method according to any one of claims 1 to 20.

Images