Model management method, system, device, communication equipment and storage medium

Abstract

The present application discloses a model management method, system, device, communication equipment and storage medium, which relates to the field of mobile communications. The method is applied to a terminal device, and the terminal device includes: an AI model management module, and the method includes: the AI ​​model management module and the AI ​​model management network element execute a first interaction process; the AI ​​model management module is a functional module for AI model management on the terminal device side, and the AI ​​model management network element is a network element for AI model management on the data network side, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

Description

Technical Field

[0001] The present application relates to the field of mobile communications, and in particular to a model management method, system, device, communication equipment and storage medium. Background Art

[0002] With the development of artificial intelligence (AI) technology, big data analysis based on AI models has become a trend.

[0003] When terminal devices use AI models and participate in big data analysis, third-party application vendors of applications corresponding to the AI ​​models need to store and manage the AI ​​models used by the terminal devices.

[0004] Since a single AI model requires a large storage space and terminal devices use a large number of AI models, third-party application vendors incur high resource overhead when managing AI models. Summary of the invention

[0005] The embodiments of the present application provide a model management method, system, device, communication device and storage medium, which can manage AI models without occupying the local computer room resources of third-party application providers, thereby ensuring the use of enterprise resources. The technical solution is as follows.

[0006] According to one aspect of the present application, a model management method is provided, which is applied to a terminal device, wherein the terminal device includes: an AI model management module, and the method includes:

[0007] The AI ​​model management module and the AI ​​model management network element execute a first interaction process;

[0008] The AI ​​model management module is a functional module used by the terminal device to perform AI model management, and the AI ​​model management network element is a network element used by the data network to perform AI model management. The AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0009] According to one aspect of the present application, a model management method is provided, which is applied to a data network, wherein the data network includes: an AI model management network element, and the method includes:

[0010] The AI ​​model management network element and the AI ​​model management module execute a first interaction process;

[0011] Among them, the AI ​​model management module is a functional module used by the terminal device to manage AI models, the AI ​​model management network element is a network element used by the data network to manage AI models, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0012] According to one aspect of the present application, a model management system is provided, the model management system comprising: an AI model management module and an AI model management network element;

[0013] The AI ​​model management module is used to execute a first interaction process with the AI ​​model management network element;

[0014] Among them, the AI ​​model management module is a functional module used by the terminal device to manage AI models, the AI ​​model management network element is a network element used by the data network to manage AI models, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0015] According to one aspect of the present application, a model management device is provided, which is applied to a terminal device, and the device includes: an AI model management module;

[0016] The AI ​​model management module is used to execute a first interaction process with the AI ​​model management network element;

[0017] The AI ​​model management module is a functional module used by the terminal device to perform AI model management, and the AI ​​model management network element is a network element used by the data network to perform AI model management. The AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0018] According to one aspect of the present application, a model management device is provided, which is applied in a data network, and the device includes:

[0019] AI model management network element module;

[0020] The AI ​​model management network element module is used to execute a first interaction process with the AI ​​model management module;

[0021] Among them, the AI ​​model management module is a functional module used by the terminal device to manage AI models, the AI ​​model management network element module is a network element module used by the data network to manage AI models, and the AI ​​model management module supports interaction with the AI ​​model management network element module so as to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0022] According to one aspect of the present application, a terminal device is provided, comprising: a processor; a transceiver connected to the processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to load and execute the executable instructions to implement the model management method as described in the above aspects.

[0023] According to one aspect of the present application, a data network is provided, comprising: a processor; a transceiver connected to the processor; a memory for storing executable instructions of the processor; wherein the processor is configured to load and execute the executable instructions to implement the model management method as described in the above aspects.

[0024] According to one aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores executable instructions, and the executable instructions are loaded and executed by a processor to implement the model management method as described in the above aspects.

[0025] According to one aspect of the present application, a computer program product or a computer program is provided, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the model management method described in the above aspect.

[0026] The technical solution provided by the embodiments of the present application includes at least the following beneficial effects:

[0027] The task of AI model management is assigned to the AI ​​model management module and the AI ​​model management network element. Since the AI ​​model management module is a functional module used for AI model management on the terminal device side, and the AI ​​model management network element is a network element used for AI model management on the data network side, there is no need to occupy the local computer room resources of the third-party application provider to manage the AI ​​model, thereby ensuring the enterprise's resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0029] Figure 1 It is a schematic diagram of a scenario of performing big data analysis based on an AI model provided by an exemplary embodiment of the present application;

[0030] Figure 2 is a schematic diagram of a communication system provided by an exemplary embodiment of the present application;

[0031] Figure 3 is a flow chart of a model management method provided by an exemplary embodiment of the present application;

[0032] Figure 4 is a schematic diagram of a model management method provided by an exemplary embodiment of the present application;

[0033] Figure 5 is a schematic diagram of a model management system provided by an exemplary embodiment of the present application;

[0034] Figure 6 is a structural block diagram of a model management device provided by an exemplary embodiment of the present application;

[0035] Figure 7 is a structural block diagram of a model management device provided by an exemplary embodiment of the present application;

[0036] Figure 8 It is a structural diagram of a communication device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION

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

[0038] First, the nouns involved in the embodiments of the present application are briefly introduced:

[0039] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0040] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0041] Among them, the AI ​​model is a model corresponding to artificial intelligence technology.

[0042] In the scenario of big data analysis based on AI models, in order to improve the effect of big data analysis and user experience, you can consider adopting a multi-level AI model approach, that is, the network elements and terminal devices on the network side divide the work of big data analysis.

[0043] Figure 1 a) is a centralized scenario, that is, after all terminal devices report the required data, the big data analysis work is all performed on the server on the network side. Figure 1 b) in the figure is a fully distributed scenario, that is, different terminal devices analyze the collected data locally. Figure 1 c) is a hybrid scenario, that is, the terminal device analyzes part of the collected data locally and sends the results to the server on the network side for further calculation and analysis. In addition, in methods b) and c), data interaction between terminal devices may also be introduced to complete big data analysis or result sharing.

[0044] In the above scenario, the AI ​​model has the following characteristics:

[0045] 1) There are many types of AI models, and different third-party application providers and under different conditions use different AI models.

[0046] 2) Since many AI services are divided by region and time, terminal devices may need to use different AI models in different regions and times.

[0047] 3) For businesses such as machine learning and distributed AI computing, AI models need to be updated frequently.

[0048] 4) An AI model may require hundreds of megabytes or even more storage space. It is impossible for terminal devices to store all models locally, so it is necessary to be able to update AI models in real time and accurately.

[0049] From the above description of the characteristics of AI models, it can be seen that there are many types of AI models, the number is huge, and they are updated frequently. If enterprises use local computer room resources to manage AI models, the resource overhead required is relatively large.

[0050] Figure 2 A block diagram of a communication system provided by an exemplary embodiment of the present application is shown, and the communication system may include: a terminal device 12, a (radio) access network ((R)AN) 14, a core network 16 and a data network (Data Network, DN) 18.

[0051] The terminal device 12 may include various handheld devices with wireless communication functions, vehicle-mounted devices, wearable devices, computing devices or other processing devices connected to a wireless modem, as well as various forms of user equipment, mobile stations (MS), terminals, etc. For the convenience of description, the above-mentioned devices are collectively referred to as terminal devices.

[0052] The access network 14 includes several network devices. The network device may be a base station, which is a device deployed in the access network to provide wireless communication functions for terminals. Base stations may include various forms of macro base stations, micro base stations, relay stations, access points, and the like. In systems using different wireless access technologies, the names of devices with base station functions may be different. For example, in an LTE system, it is called eNodeB or eNB; in a 5G NR-U system, it is called gNodeB or gNB. With the evolution of communication technology, the description of "base station" may change. For the convenience of the embodiments of the present application, the above-mentioned devices that provide wireless communication functions for terminal devices are collectively referred to as network devices.

[0053] The core network 16 may include: User Plane Function (UPF) and control plane functions. The control plane functions may include: Access and Mobility Management Function (AMF), Session Management Function (SMF), Policy Control Function (PCF), Unified Data Manager (UDM), Application Function (AF), Network Slice Selection Function (NSSF), and Authentication Server Function (AUSF).

[0054] The wireless network may include an access network 14 and a core network 16. Different network elements in the wireless network may specifically correspond to base stations of the access network or various functional network elements of the core network (such as SMF, PCF, AMF, etc.).

[0055] The terminal device 12 communicates with the network devices in the access network 14 via some air interface technology, such as the Uu interface. The core network 16 can transmit data with the external data network 18 via the N6 interface and with the access network 14 via the N3 interface.

[0056] The technical solutions of the embodiments of the present application can be applied to various communication systems, such as: Global System of Mobile Communication (GSM) system, Code Division Multiple Access (CDMA) system, Wideband Code Division Multiple Access (WCDMA) system, General Packet Radio Service (GPRS), Long Term Evolution (LTE) system, LTE Frequency Division Duplex (FDD) system, LTE Time Division Duplex (TDD) system, Advanced Long Term Evolution (LTE-A) system, New Radio (NR) system, NR system evolution system, LTE on unlicensed spectrum (LTE-based access to Unlicensed spectrum, LTE-U) system, NR-U system, Universal Mobile Telecommunication System (UMTS), Worldwide Interoperability for Microwave Access (WiMAX) communication system, Wireless Local Area Network (WLAN) system. Networks, WLAN), Wireless Fidelity (WiFi), next generation communication systems or other communication systems, etc.

[0057] Generally speaking, the number of connections supported by traditional communication systems is limited and easy to implement. However, with the development of communication technology, mobile communication systems will not only support traditional communications, but also support, for example, device to device (Device to Device, D2D) communication, machine to machine (Machine to Machine, M2M) communication, machine type communication (Machine Type Communication, MTC), vehicle to vehicle (V2V) communication and vehicle to everything (V2X) system, etc. The embodiments of the present application can also be applied to these communication systems.

[0058] Figure 3A flowchart of a model management method provided by an exemplary embodiment of the present application is shown. Figure 2 The method is exemplarily described in the communication system shown in FIG.

[0059] Step 310: The AI ​​model management module and the AI ​​model management network element execute a first interaction process.

[0060] The terminal device includes several functional modules for realizing different functions, among which the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model.

[0061] The terminal device is installed with several applications that need to use AI models to execute corresponding AI services. The AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device. The wireless protocol layer is a protocol layer that supports processing 3GPP protocols. The embodiments of this application do not limit the specific implementation form of the wireless protocol layer.

[0062] The AI ​​model management network element is a network element used by the data network end to manage the AI ​​model. Optionally, the embodiment of the present application does not limit the specific implementation form of the AI ​​model management network element. The AI ​​model management network element can be an existing network element or a newly added network element for AI model management.

[0063] The first interaction process is an interaction process between the AI ​​model management module and the AI ​​model management network element. The first interaction process is used to download or upload the AI ​​model.

[0064] Optionally, the AI ​​model management module and the AI ​​model management network element both correspond to functional modules or network elements of the same operator, and the AI ​​model management module and the AI ​​model management network element are part of the operator's network.

[0065] Optionally, in addition to performing the first interaction process to realize the download or upload of the AI ​​model, the AI ​​model management module and the AI ​​model management network element also support compression and decompression of the AI ​​model for transmission or storage of the AI ​​model content.

[0066] To sum up, the method provided in this embodiment entrusts the task of AI model management to the AI ​​model management module and the AI ​​model management network element. Since the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, and the AI ​​model management network element is a network element used by the data network to manage the AI ​​model, there is no need to occupy the local computer room resources of the third-party application provider to manage the AI ​​model, thereby ensuring the enterprise's resource utilization.

[0067] In all embodiments of the present application, the model management method of the AI ​​model can be used in conjunction with any embodiment of the present application, or can be used alone, and the embodiments of the present application do not limit this. However, in order to make the technical solution easier to understand, the following embodiments of the present application are illustrated in conjunction with the aforementioned embodiments.

[0068] Based on Figure 3 In an optional embodiment, the AI ​​models in the AI ​​model management module and the AI ​​model management network element may include but are not limited to: a first type of AI model and a second type of AI model.

[0069] Among them, the first type of AI model corresponds to the operator, and the second type of AI model corresponds to the third-party application (such as from the application or service provider). The operator is the operator who establishes and operates the network for the purpose of providing land mobile communication services to the public. The third-party application is an application developed for the terminal device. In other words, the first type of AI model is the operator's AI model, and the second type of AI model is the AI ​​model of the third-party application.

[0070] The AI ​​model corresponds to an AI model identifier. Optionally, the AI ​​model identifiers of the above two types of AI models correspond to the same or different model identifier formats. Among them, the model identifier format of the first type of AI model is the first type model identifier format, and the model identifier format of the second type of AI model is the second type model identifier format.

[0071] Optionally, the first type model identification format and the second type model identification format are different, which may mean that the two types of model identification formats contain different identification types; or, the positions of different identification types in the formats are different; or, the number of digits of different identification types is different.

[0072] Exemplarily, the first type of model identification format includes: an operator identification and a model identification.

[0073] The operator identifier may be recorded as PLMN ID, which is used to distinguish different operators. The model identifier may be recorded as Model Id, which is used to identify the model.

[0074] Exemplarily, the model identification format of the second type AI model is a second type model identification format, and the second type model identification format includes at least one of the following: an operator identification, an application identification, and a model identification.

[0075] In one implementation, the second type model identification format includes: an operator identification, an application identification, and a model identification; in another implementation, the second type model identification format includes: an application identification and a model identification.

[0076] The operator identifier may be recorded as PLMN ID, which is used to distinguish different operators. The application identifier may be recorded as Application Id, which is used to identify a third-party application. The model identifier may be recorded as Model Id, which is used to identify a model.

[0077] With respect to the operator identifier and the application identifier in the above two types of model identifier formats, the operator identifier may be determined by the operator or a third-party application, and the application identifier may be determined by the operator or a third-party application.

[0078] For the model identifiers in the above two types of model identifier formats, the model identifier may include at least one of a model type identifier (Type Id), a model structure parameter identifier (Structure Id), and a model weight parameter identifier (Weight Id).

[0079] Among them, the model type identifier is used to identify the type of AI model, and the type of AI model includes but is not limited to: at least one of a deep neural network (DNN) model, a convolutional neural network (CNN) model, and a recurrent neural network (RNN) model.

[0080] The model structure parameter identifier is used to identify the structure of the model. For example, for a DNN model, its internal neural network layers can be divided into three categories: input layer, hidden layer, and output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and the layers in between are all hidden layers.

[0081] The model weight parameter identifier is used to identify the weight value used by the model. For example, for a DNN model, each layer needs to use a specific algorithm, weight value, constant, etc. to calculate the sample parameters of the previous layer to obtain an output result, which is used as the input of the next layer to calculate the output of the next layer, and finally obtain the result of the output layer.

[0082] For the model identifiers in the above two types of model identifier formats, the value or partial value of the model identifier may include: a standardized model identifier value and a non-standardized model identifier value.

[0083] Among them, the standardized model identifier value is a value agreed upon through standardization, corresponding to the AI ​​model of the target type or the AI ​​model of the target business or the target AI model.

[0084] Through standardization, the values ​​or partial values ​​of one or more model identifiers can be agreed upon as standardized model identifier values. For example, for a model identifier with a value of 000010, it means that the AI ​​model is an AI model for processing the target business of video business; for a model identifier with a value of 000011, it means that the AI ​​model is an AI model corresponding to the target type of CNN; for a model identifier with a value of 000012, it means that the AI ​​model is a target AI model for the target business of autonomous driving.

[0085] Among them, the non-standardized model identifier value is a model identifier value determined by the operator or third-party application. In different operators or different third-party applications, the same non-standardized model value can be defined to correspond to different AI models. For example: 100001 is a non-standardized model identifier value. In operator A, the AI ​​model with a model identifier value of 100001 is a CNN model, while in operator B, the AI ​​model with a model identifier value of 100001 is an RNN model. Alternatively, the AI ​​model with a value of 100001 of operator A is used for autonomous driving services, and the AI ​​model with a value of 100001 of operator B is suitable for augmented reality (AR) / virtual reality (VR) services.

[0086] At the same time, based on Figure 3 In an optional embodiment, there is also the following interface for implementing AI model management. It can be understood that the interaction is carried out through the AI ​​model identification described in the above embodiment.

[0087] 1. The first interface.

[0088] The first interface is an interface between the AI ​​model management module and the AI ​​model management network element. The first interface can use an application layer protocol, such as HyperText Transfer Protocol (HTTP).

[0089] Optionally, step 310 in the above embodiment can be replaced by: the AI ​​model management module and the AI ​​model management network element execute the first interaction process through the first interface.

[0090] The first interaction process includes but is not limited to at least one of the following:

[0091] 1) AI model request process.

[0092] The AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element.

[0093] 2) AI model notification process.

[0094] The AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element.

[0095] 3) Capability negotiation process.

[0096] The capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading.

[0097] 4) AI model download process.

[0098] The AI ​​model download process is used to enable the AI ​​model management module to download the target AI model from the AI ​​model management network element.

[0099] 5) AI model upload process.

[0100] The AI ​​model upload process is used to enable the AI ​​model management module to upload the target AI model to the AI ​​model management network element.

[0101] Optionally, the first interface includes a first control plane and a first user plane; wherein, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

[0102] That is to say, the processes that can be executed by the first interface can be separated, with the first control plane executing control-related processes (such as capability negotiation processes), and the first user plane executing bearer-related functions (such as AI model download processes).

[0103] It is understandable that if Figure 4 As shown, the AI ​​model management module and the AI ​​model management network element can also be separated into two planes: the control plane and the user plane, separating the bearer and the control.

[0104] 2. The second interface.

[0105] The second interface is the interface between the AI ​​model management module and the third-party application in the terminal device.

[0106] Optionally, the AI ​​model management module and the third-party application perform a second interaction process through a second interface. Optionally, the second interaction process includes: the AI ​​model management module sends the AI ​​model to the third-party application, and the third-party application sends the AI ​​model to the AI ​​model management module.

[0107] Exemplarily, the AI ​​model management module downloads the AI ​​model from the AI ​​model management network element, and the application identifier Application ID in the AI ​​model identifier of the AI ​​model is 1, then the AI ​​model management module sends the downloaded AI model to Application-1.

[0108] Exemplarily, the second interface is used to apply a specific AI model to the AI ​​model management module so that the AI ​​model management module can save it or upload it to the AI ​​model management network element for unified storage / management.

[0109] 3. The third interface.

[0110] The third interface is the interface between the AI ​​model management network element and other servers / network elements in the data network.

[0111] Optionally, the AI ​​model management network element and the application server execute a third interaction process through a third interface; or, the AI ​​model management network element and a third-party network element execute a fourth interaction process through a third interface, and the third-party network element is a network element other than the AI ​​model management network element.

[0112] Among them, the application server may correspond to a third-party network element.

[0113] Optionally, the AI ​​model management network element can also interact with other network elements through a third interface. The other network elements and the AI ​​model management network element can correspond to the same operator or different operators.

[0114] Optionally, the third interaction process includes: storage, downloading, and uploading of AI models between the AI ​​model management network element and the application server; the fourth interaction process includes: storage, downloading, and uploading of AI models between the AI ​​model management network element and a third-party network element.

[0115] Exemplarily, the application server sends the AI ​​model of the third-party application to the AI ​​model management network element for safekeeping.

[0116] Exemplarily, the third-party network element is an AI model selection network element, which is used to determine the AI ​​model that the terminal device needs to use, and trigger the AI ​​model management network element through interface 3 to send the corresponding AI model to the terminal device for use.

[0117] 4. The fourth interface.

[0118] The fourth interface is the interface between the AI ​​model management network element and the wireless network.

[0119] The AI ​​model management network element and the network element in the wireless network perform the fifth interaction process through the fourth interface. The network element in the wireless network can have the function of the AI ​​model management module and interact with the AI ​​model management network element to perform corresponding functions.

[0120] Optionally, the purpose of the fifth interaction process may be any one of the following:

[0121] 1) The fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or a Quality of Service (QoS) data flow.

[0122] By triggering the wireless network to establish a dedicated bearer or QoS data stream, the security and efficiency of executing the first interactive process through the first interface are guaranteed.

[0123] 2) The fifth interaction process is used for the AI ​​model management network element to send the AI ​​model to the wireless network.

[0124] The wireless network can use the AI ​​model obtained from the AI ​​model management network element.

[0125] 3) The fifth interaction process is used by the AI ​​model management network element to obtain the AI ​​model from the wireless network and / or store the AI ​​model.

[0126] The AI ​​model management network element can also obtain the AI ​​model from the wireless network and store it.

[0127] 4) The fifth interaction process is used for the AI ​​model management network element to receive a request from a network element in the wireless network, where the request is used to request the AI ​​model management network element to send an AI model.

[0128] The AI ​​model management network element can send the AI ​​model to the terminal device accordingly based on the request of the network element in the wireless network.

[0129] Optionally, in addition to the above functions, the network elements in the wireless network can also negotiate capabilities with the AI ​​model management network elements to negotiate AI models that can be downloaded or uploaded.

[0130] Optionally, the AI ​​model management network element is connected to the capability exposure network element in the wireless network or directly connected to a network element in the wireless network. Among them, the capability exposure network element is a network element that provides capability exposure services. That is to say, the AI ​​model management network element can interact with the capability exposure network element in the wireless network through the fourth interface. Capability exposure network elements include but are not limited to: at least one of a network exposure function (NEF) network element, a service capability exposure function (SCEF) network element, and a policy control function (PCF) network element. NEF network elements and SCEF network elements can open service capabilities to third-party service providers through API interfaces, and PCF network elements are responsible for policy control.

[0131] To summarize, the method provided in this embodiment defines a first type of AI model and a second type of AI model and provides model identification formats for different types of AI models, thereby more accurately identifying the AI ​​model and facilitating the management of the AI ​​model.

[0132] The method provided in this embodiment defines different interfaces, so that operators can open relevant functions of AI management to other objects (such as third-party applications) to facilitate the transmission of AI model content.

[0133] It should be noted that the above method embodiments can be implemented separately or in combination, and this application does not limit this.

[0134] In each of the above-mentioned embodiments, the steps executed by the terminal device can be independently implemented as a model management method on the terminal device side, and the steps executed by the network device can be independently implemented as a model management method on the network device side.

[0135] Corresponding to the above method embodiment, the present application provides a model management system, the model management system includes: an artificial intelligence AI model management module and an AI model management network element;

[0136] An AI model management module, configured to execute a first interaction process with an AI model management network element;

[0137] Among them, the AI ​​model management module is a functional module used by the terminal device to manage AI models, the AI ​​model management network element is a network element used by the data network to manage AI models, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0138] Optionally, the model management system further includes: a first interface; an AI model management module, configured to execute a first interaction process with the AI ​​model management network element through the first interface; wherein the first interaction process includes at least one of the following:

[0139] AI model request process: The AI ​​model request process is used by the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element;

[0140] AI model notification process: The AI ​​model notification process is used by the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element;

[0141] Capability negotiation process: The capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading;

[0142] AI model download process;

[0143] AI model upload process.

[0144] Optionally, the first interface includes a first control plane and a first user plane; wherein, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

[0145] Optionally, the AI ​​model management system also includes: a second interface; the AI ​​model management module is used to execute a second interaction process with a third-party application through the second interface.

[0146] Optionally, the AI ​​model management system also includes: a third interface; an AI model management network element, used to execute a third interaction process with the application server through the third interface; or, an AI model management network element, used to execute a fourth interaction process with a third-party network element through the third interface, where the third-party network element is a network element other than the AI ​​model management network element.

[0147] Optionally, the AI ​​model management system further includes: a fourth interface; an AI model management network element, used to perform a fifth interaction process with a capability exposure network element in the wireless network through the fourth interface; wherein the fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or a QoS data flow;

[0148] Or, the fifth interaction process is used for the AI ​​model management network element to send the AI ​​model to the network element in the wireless network;

[0149] Or, the fifth interaction process is used for the AI ​​model management network element to obtain the AI ​​model from the wireless network, and / or store the AI ​​model;

[0150] Alternatively, the fifth interaction process is used for the AI ​​model management network element to receive a request from a network element in the wireless network, where the request is used to request the AI ​​model management network element to send an AI model.

[0151] Optionally, the network element in the wireless network includes: a capability open network element, and the capability open network element includes: at least one of a NEF network element, a SCEF network element and a PCF network element.

[0152] Optionally, the model identification format of the first type AI model is a first type model identification format, and the first type model identification format includes: an operator identification and a model identification;

[0153] The model identification format of the second type of AI model is the second type model identification format, and the second type model identification format includes: operator identification, model identification and application identification.

[0154] Optionally, the value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

[0155] Optionally, the model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

[0156] Optionally, the operator identifier is determined by the operator or a third-party application; the application identifier is determined by the operator or a third-party application.

[0157] Combined with reference Figure 5 , which shows a structural diagram of an AI model management system provided by an embodiment of the present application.

[0158] like Figure 5 As shown in (a), the AI ​​model management system includes: an AI model management module 501, and an AI model management network element 502.

[0159] An AI model management module 501 is introduced on the terminal device side, and an AI model management network element 502 is introduced on the network side. The AI ​​model management module 501 and the AI ​​model management network element 502 can be part of the operator network, and provide AI model management services for third-party applications through an application program interface (API).

[0160] Between the terminal device and the data network is the operator's wireless network (including base stations and core networks). The AI ​​model management module 501 and the AI ​​model management network element 502 can communicate through interface 1 (i.e., the first interface) in the user plane of the wireless network. The data transmitted by interface 1 can be transmitted through the protocol data unit (PDU) session of the wireless network.

[0161] Optionally, the AI ​​model management system also includes interface 2 (i.e., the second interface) and interface 3 (i.e., the third interface). The AI ​​model management module 501 can open relevant functions of AI management (including AI model storage, downloading, uploading, etc.) to third-party applications in the terminal device through interface 2. The AI ​​model management network element 502 can open relevant functions of AI management (including AI model storage, downloading, uploading, etc.) to third-party applications or other network elements through interface 3.

[0162] Optionally, the AI ​​model management system further includes interface 4 (ie, the fourth interface). The AI ​​model management network element 502 can interact with the capability open network element of the operator network through interface 4. The capability open network element includes but is not limited to: NEF network element, SCEF network element, PCF network element.

[0163] Alternatively, if Figure 5 As shown in (b), the AI ​​model management module 501 is connected to the wireless protocol layer 503, and the wireless protocol layer 503 can process the 3GPP protocol. Exemplarily, the AI ​​model management module 501 sends the AI ​​model used by the operator to the wireless protocol layer 503 for use. For example, it is used to optimize the wireless channel quality, mobility management optimization, session management optimization, UE strategy optimization, etc. of the wireless network.

[0164] To sum up, the embodiments of the present application provide a model management system that can utilize the operator's own cloud computing resources to implement AI model management services, and third-party application providers do not need to use local computer room resources to manage AI models.

[0165] Figure 6 6 is a block diagram of a model management device provided by an exemplary embodiment of the present application. The device is applied in a terminal device, or the device is implemented as a terminal device or a part of a terminal device. The device includes: an AI model management module 601;

[0166] The AI ​​model management module 601 is used to perform a first interaction process with the AI ​​model management network element;

[0167] The AI ​​model management module 601 is a functional module used by the terminal device to perform AI model management. The AI ​​model management network element is a network element used by the data network to perform AI model management. The AI ​​model management module 601 supports interaction with the AI ​​model management network element to download or upload AI models for the application layer and / or wireless protocol layer in the terminal device.

[0168] In an optional embodiment, the AI ​​model management module 601 is used to perform a first interaction process with the AI ​​model management network element through a first interface; wherein the first interaction process includes at least one of the following:

[0169] AI model request process, the AI ​​model request process is used for the AI ​​model management module 601 to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element;

[0170] AI model notification process, the AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module 601 of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element;

[0171] Capability negotiation process, which is used for the AI ​​model management module 601 and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading;

[0172] AI model download process;

[0173] AI model upload process.

[0174] In an optional embodiment, the first interface includes a first control plane and a first user plane; wherein the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

[0175] In an optional embodiment, the AI ​​model management module 601 is used to execute a second interaction process with a third-party application through a second interface.

[0176] In an optional embodiment, the AI ​​model in the AI ​​model management module 601 includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application.

[0177] In an optional embodiment, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes at least one of the following: operator identification, application identification and model identification.

[0178] In an optional embodiment, the value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

[0179] In an optional embodiment, the model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

[0180] In an optional embodiment, the operator identifier is determined by the operator or a third-party application; the application identifier is determined by the operator or a third-party application.

[0181] Figure 7 701 is a block diagram of a model management device provided by an exemplary embodiment of the present application. The device is applied in a network device, or the device is implemented as a network device or a part of a network device. The device includes: an AI model management network element module 701;

[0182] The AI ​​model management network element module 701 is used to perform a first interaction process with the AI ​​model management module;

[0183] Among them, the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element module 701 is a network element module used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element module 701 to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device.

[0184] In an optional embodiment, the AI ​​model management network element module 701 is used to perform a first interaction process with the AI ​​model management module through a first interface; wherein the first interaction process includes at least one of the following:

[0185] AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element module 701 or upload the target AI model to the AI ​​model management network element;

[0186] AI model notification process, the AI ​​model notification process is used for the AI ​​model management network element module 701 to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element;

[0187] Capability negotiation process, which is used for the AI ​​model management module and the AI ​​model management network element module 701 to negotiate the target AI model that supports downloading or uploading;

[0188] AI model download process;

[0189] AI model upload process.

[0190] In an optional embodiment, the first interface includes a first control plane and a first user plane; wherein the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

[0191] In an optional embodiment, the AI ​​model management network element module 701 is used to execute a third interaction process with an application server through a third interface; or, the AI ​​model management network element module 701 is used to execute a fourth interaction process with a third-party network element through a third interface, and the third-party network element is a network element other than the AI ​​model management network element.

[0192] In an optional embodiment, the AI ​​model management network element module 701 is used to execute the fifth interaction process with the network elements in the wireless network through the fourth interface; wherein the fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or QoS data flow; or, the fifth interaction process is used for the AI ​​model management network element module 701 to send the AI ​​model to the network elements in the wireless network; or, the fifth interaction process is used for the AI ​​model management network element module 701 to obtain the AI ​​model from the wireless network, and / or store the AI ​​model; or, the fifth interaction process is used for the AI ​​model management network element module 701 to receive a request from the network element in the wireless network, and the request is used to request the AI ​​model management network element module 701 to send the AI ​​model.

[0193] In an optional embodiment, the network element in the wireless network includes: a capability opening network element, and the capability opening network element includes: at least one of a network opening function NEF network element, a service capability opening function SCEF network element and a policy control function PCF network element.

[0194] In an optional embodiment, the AI ​​model in the AI ​​model management network element module 701 includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application.

[0195] In an optional embodiment, the model identification format of the first type AI model is a first type model identification format, and the first type model identification format includes: an operator identification and a model identification;

[0196] The model identification format of the second type AI model is the second type model identification format, and the second type model identification format includes at least one of the following: operator identification, model identification and application identification.

[0197] In an optional embodiment, the value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a model identifier value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

[0198] In an optional embodiment, the model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

[0199] In an optional embodiment, the operator identifier is determined by the operator or a third-party application; the application identifier is determined by the operator or a third-party application.

[0200] Figure 8 A schematic diagram of the structure of a communication device provided by an exemplary embodiment of the present application is shown. The communication device includes: a processor 101, a receiver 102, a transmitter 103, a memory 104 and a bus 105.

[0201] The processor 101 includes one or more processing cores. The processor 101 executes various functional applications and information processing by running software programs and modules.

[0202] The receiver 102 and the transmitter 103 may be implemented as a communication component, which may be a communication chip.

[0203] The memory 104 is connected to the processor 101 via a bus 105 .

[0204] The memory 104 may be used to store at least one instruction, and the processor 101 may be used to execute the at least one instruction to implement each step in the above method embodiment.

[0205] In addition, the memory 104 can be implemented by any type of volatile or non-volatile storage device or a combination thereof. The volatile or non-volatile storage device includes but is not limited to: a magnetic disk or an optical disk, an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a static random access memory (SRAM), a read-only memory (ROM), a magnetic memory, a flash memory, and a programmable read-only memory (PROM).

[0206] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one instruction, at least one program, a code set or an instruction set is stored. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement the model management method provided by the above-mentioned various method embodiments and executed by a terminal device, or the model management method executed by a data network.

[0207] In an exemplary embodiment, a computer program product or a computer program is also provided, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the model management method described in the above aspects.

[0208] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0209] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A model management method, characterized in that: Applied in a terminal device, the terminal device includes: an artificial intelligence AI model management module, and the method includes: The AI ​​model management module and the AI ​​model management network element execute a first interaction process; The AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element is a network element used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device; The AI ​​model in the AI ​​model management module includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application; Among them, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes at least one of the following: the operator identification, the application identification and the model identification.

2. The method according to claim 1, characterized in that The AI ​​model management module and the AI ​​model management network element execute a first interaction process, including: The AI ​​model management module and the AI ​​model management network element execute the first interaction process through the first interface; The first interaction process includes at least one of the following: AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element; An AI model notification process, wherein the AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element; A capability negotiation process, wherein the capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading; AI model download process; AI model upload process.

3. The method according to claim 2, characterized in that The first interface includes a first control plane and a first user plane; Among them, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

4. The method according to claim 1, characterized in that: The method further comprises: The AI ​​model management module and the third-party application execute a second interaction process through a second interface.

5. The method according to claim 1, characterized in that The value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

6. The method according to claim 1, characterized in that The model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

7. The method according to claim 1, characterized in that The operator identifier is determined by the operator or the third-party application; The application identifier is determined by the operator or the third-party application.

8. A model management method, characterized in that: Applied in a data network, the data network includes: an artificial intelligence AI model management network element, the method includes: The AI ​​model management network element and the AI ​​model management module execute a first interaction process; Among them, the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element is a network element used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device; The AI ​​model in the AI ​​model management network element includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application; Among them, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes at least one of the following: the operator identification, the model identification and the application identification.

9. The method according to claim 8, characterized in that The AI ​​model management network element and the AI ​​model management module execute a first interaction process, including: The AI ​​model management network element and the AI ​​model management module execute a first interaction process through a first interface; The first interaction process includes at least one of the following: AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element; An AI model notification process, wherein the AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element; A capability negotiation process, wherein the capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading; AI model download process; AI model upload process.

10. The method according to claim 9, characterized in that The first interface includes a first control plane and a first user plane; Among them, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

11. The method according to claim 8, characterized in that The method further comprises: The AI ​​model management network element and the application server execute a third interaction process through a third interface; or, The AI ​​model management network element and the third-party network element perform a fourth interaction process through the third interface, and the third-party network element is a network element other than the AI ​​model management network element.

12. The method according to claim 8, characterized in that The method further comprises: The AI ​​model management network element and the network elements in the wireless network perform a fifth interaction process through the fourth interface; The fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or a quality of service QoS data flow; Or, the fifth interaction process is used by the AI ​​model management network element to send the AI ​​model to the network element in the wireless network; Or, the fifth interaction process is used by the AI ​​model management network element to obtain the AI ​​model from the wireless network, and / or store the AI ​​model; Alternatively, the fifth interaction process is used for the AI ​​model management network element to receive a request from a network element in the wireless network, and the request is used to request the AI ​​model management network element to send an AI model.

13. The method according to claim 12, characterized in that The network elements in the wireless network include: capability opening network elements, and the capability opening network elements include: at least one of a network opening function NEF network element, a service capability opening function SCEF network element and a policy control function PCF network element.

14. The method according to claim 8, characterized in that The value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a model identifier value agreed upon by standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

15. The method according to claim 8, characterized in that The model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

16. The method according to claim 8, characterized in that The operator identifier is determined by the operator or the third-party application; The application identifier is determined by the operator or the third-party application.

17. A model management system, characterized in that: The model management system includes: an artificial intelligence AI model management module and an AI model management network element; The AI ​​model management module is used to execute a first interaction process with the AI ​​model management network element; Among them, the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element is a network element used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device; The AI ​​model in the AI ​​model management system includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application; Among them, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes: the operator identification, the model identification and application identification.

18. The system according to claim 17, characterized in that The model management system further includes: a first interface; The AI ​​model management module is used to perform a first interaction process with the AI ​​model management network element, including: The AI ​​model management module is used to perform the first interaction process with the AI ​​model management network element through the first interface; The first interaction process includes at least one of the following: AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element; An AI model notification process, wherein the AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element; A capability negotiation process, wherein the capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading; AI model download process; AI model upload process.

19. The system of claim 18, wherein the first interface comprises a first control plane and a first user plane; in, The first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

20. The system according to claim 17, characterized in that The AI ​​model management system further includes: a second interface; The AI ​​model management module is used to execute a second interaction process with a third-party application through the second interface.

21. The system according to claim 17, characterized in that The AI ​​model management system further includes: a third interface; The AI ​​model management network element is used to perform a third interaction process with the application server through the third interface; or, The AI ​​model management network element is used to execute the fourth interaction process with a third-party network element through the third interface, and the third-party network element is a network element other than the AI ​​model management network element.

22. The system according to claim 17, characterized in that The AI ​​model management system further includes: a fourth interface; The AI ​​model management network element is used to perform a fifth interaction process with a network element in the wireless network through the fourth interface; The fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or a quality of service QoS data flow; Or, the fifth interaction process is used by the AI ​​model management network element to send the AI ​​model to the network element in the wireless network; Or, the fifth interaction process is used by the AI ​​model management network element to obtain the AI ​​model from the wireless network, and / or store the AI ​​model; Alternatively, the fifth interaction process is used for the AI ​​model management network element to receive a request from a network element in the wireless network, and the request is used to request the AI ​​model management network element to send an AI model.

23. The system according to claim 22, characterized in that The network elements in the wireless network include: capability opening network elements, and the capability opening network elements include: at least one of a network opening function NEF network element, a service capability opening function SCEF network element and a policy control function PCF network element.

24. The system according to claim 17, characterized in that The value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

25. The system according to claim 17, characterized in that The model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

26. The system according to claim 25, characterized in that The operator identifier is determined by the operator or the third-party application; The application identifier is determined by the operator or the third-party application.

27. A model management device, characterized in that: Applied in a terminal device, the device comprises: an artificial intelligence AI model management module; The AI ​​model management module is used to execute a first interaction process with the AI ​​model management network element; The AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element is a network element used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device; The AI ​​model in the AI ​​model management module includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application; Among them, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes at least one of the following: the operator identification, the application identification and the model identification.

28. The device according to claim 27, characterized in that The AI ​​model management module is used to perform the first interaction process with the AI ​​model management network element through the first interface; The first interaction process includes at least one of the following: AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element or upload the target AI model to the AI ​​model management network element; An AI model notification process, wherein the AI ​​model notification process is used for the AI ​​model management network element to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element; A capability negotiation process, wherein the capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element to negotiate the target AI model that supports downloading or uploading; AI model download process; AI model upload process.

29. The device according to claim 28, characterized in that The first interface includes a first control plane and a first user plane; Among them, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

30. The device according to claim 27, characterized in that The AI ​​model management module is used to execute a second interaction process with a third-party application through a second interface.

31. The device according to claim 27, characterized in that The value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a value agreed upon through standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

32. The device according to claim 27, characterized in that The model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

33. The device according to claim 27, characterized in that The operator identifier is determined by the operator or the third-party application; The application identifier is determined by the operator or the third-party application.

34. A model management device, characterized in that: Applied in a data network, the device comprises: an artificial intelligence AI model management network element module; The AI ​​model management network element module is used to execute a first interaction process with the AI ​​model management module; Among them, the AI ​​model management module is a functional module used by the terminal device to manage the AI ​​model, the AI ​​model management network element module is a network element module used by the data network to manage the AI ​​model, and the AI ​​model management module supports interaction with the AI ​​model management network element module to download or upload the AI ​​model for the application layer and / or wireless protocol layer in the terminal device; The AI ​​model in the AI ​​model management network element module includes at least one of the following: a first type AI model and a second type AI model; wherein the first type AI model corresponds to an operator, and the second type AI model corresponds to a third-party application; Among them, the model identification format of the first type of AI model is a first type of model identification format, and the first type of model identification format includes: operator identification and model identification; the model identification format of the second type of AI model is a second type of model identification format, and the second type of model identification format includes at least one of the following: the operator identification, the model identification and the application identification.

35. The device according to claim 34, characterized in that The AI ​​model management network element module is used to perform a first interaction process with the AI ​​model management module through a first interface; The first interaction process includes at least one of the following: AI model request process, the AI ​​model request process is used for the AI ​​model management module to request to download the target AI model from the AI ​​model management network element module or to upload the target AI model to the AI ​​model management network element; An AI model notification process, wherein the AI ​​model notification process is used for the AI ​​model management network element module to notify the AI ​​model management module of the target AI model that needs to be uploaded or the target AI model that needs to be sent to the AI ​​model management network element; A capability negotiation process, wherein the capability negotiation process is used for the AI ​​model management module and the AI ​​model management network element module to negotiate the target AI model that supports downloading or uploading; AI model download process; AI model upload process.

36. The device according to claim 35, characterized in that The first interface includes a first control plane and a first user plane; Among them, the first control plane is used to execute the AI ​​model request process, the AI ​​model notification process and the capability negotiation process, and the first user plane is used to execute the AI ​​model download process and the AI ​​model upload process.

37. The device according to claim 34, characterized in that The AI ​​model management network element module is used to perform a third interaction process with the application server through a third interface; or, The AI ​​model management network element module is used to execute the fourth interaction process with a third-party network element through the third interface, and the third-party network element is a network element other than the AI ​​model management network element.

38. The device according to claim 34, characterized in that The AI ​​model management network element module is used to perform a fifth interaction process with a network element in the wireless network through a fourth interface; The fifth interaction process is used to trigger the wireless network to establish a dedicated bearer or a quality of service QoS data flow; Or, the fifth interaction process is used by the AI ​​model management network element module to send the AI ​​model to the network element in the wireless network; Or, the fifth interaction process is used by the AI ​​model management network element module to obtain the AI ​​model from the wireless network, and / or store the AI ​​model; Alternatively, the fifth interaction process is used for the AI ​​model management network element module to receive a request from a network element in the wireless network, and the request is used to request the AI ​​model management network element module to send an AI model.

39. The device according to claim 38, characterized in that The network elements in the wireless network include: capability opening network elements, and the capability opening network elements include: at least one of a network opening function NEF network element, a service capability opening function SCEF network element and a policy control function PCF network element.

40. The device according to claim 34, characterized in that The value of the model identifier includes: a standardized model identifier value and a non-standardized model identifier value. The standardized model identifier value is a model identifier value agreed upon by standardization, corresponding to an AI model of a target type or an AI model of a target business or a target AI model.

41. The device according to claim 34, characterized in that The model identifier includes at least one of a model type identifier, a model structure parameter identifier, and a model weight parameter identifier.

42. The device according to claim 34, characterized in that The operator identifier is determined by the operator or the third-party application; The application identifier is determined by the operator or the third-party application.

43. A terminal device, characterized in that: The terminal device comprises: processor; a transceiver connected to the processor; a memory for storing executable instructions for the processor; The processor is configured to load and execute the executable instructions to implement the model management method as described in any one of claims 1 to 7.

44. A data network, characterized in that The data network includes: processor; a transceiver connected to the processor; a memory for storing executable instructions for the processor; The processor is configured to load and execute the executable instructions to implement the model management method as described in any one of claims 8 to 16.

45. A computer-readable storage medium, characterized in that The readable storage medium stores executable instructions, and the executable instructions are loaded and executed by a processor to implement the model management method as described in any one of claims 1 to 16.

Images