Business processing method and device, communication equipment and readable storage medium

Abstract

The invention discloses a service processing method and device, communication equipment and a readable storage medium, and belongs to the technical field of communication, and the service processing method comprises the steps that a first network element obtains an establishment request of a second network element requesting to establish an artificial intelligence (AI) service, and the AI service comprises at least one task; according to the establishment request, respectively determining an AI strategy for at least one task in the AI service; returning a response message of the establishment request to the second network element according to the AI strategy; the first network element is a core network element, and the second network element is a third-party server.

Description

Technical Field

[0001] The present application belongs to the field of communication technology, and specifically relates to a service processing method, apparatus, communication equipment and readable storage medium. Background Art

[0002] With the development of wireless mobile communication technology, artificial intelligence (AI) will become one of the pillar technologies of communication systems, and communication systems will provide end-to-end support for AI-related services and applications. In related technologies, communication systems can use AI to perform some operation optimizations, but AI is not provided as a general service to third-party servers, nor is AI services implemented according to the needs or purposes of third-party servers, such as multi-tasking AI services. Third-party servers are an important supplementary part of communication systems. In this case, how to implement a multi-tasking AI business service process based on third-party servers is an urgent problem to be solved. Summary of the invention

[0003] The embodiments of the present application provide a business processing method, apparatus, communication equipment and readable storage medium, which can solve the problem of how to implement a multi-tasking AI business service process.

[0004] In a first aspect, a service processing method is provided, which is performed by a first network element, and the method includes:

[0005] The first network element obtains a request from the second network element to establish an AI service, where the AI ​​service includes at least one task;

[0006] The first network element determines, according to the establishment request, an AI strategy for at least one task in the AI ​​service;

[0007] The first network element returns a response message of the establishment request to the second network element according to the AI ​​policy;

[0008] The first network element is a core network element, and the second network element is a third-party server.

[0009] In a second aspect, a service processing method is provided, which is performed by a second network element, and the method includes:

[0010] The second network element sends a request to the first network element to establish an AI service, where the AI ​​service includes at least one task;

[0011] The second network element receives a response message to the establishment request returned by the first network element;

[0012] The response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

[0013] According to a third aspect, a service processing method is provided, which is executed by a third network element, and the method includes:

[0014] The third network element sends a sixth message to the sixth network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service;

[0015] The third network element receives a response message to the sixth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

[0016] In a fourth aspect, a service processing method is provided, which is executed by a fourth network element, and the method includes:

[0017] The fourth network element receives the first data sent by the second network element;

[0018] The fourth network element performs an AI unit processing operation according to the first data;

[0019] The fourth network element sends second data to the second network element.

[0020] In a fifth aspect, a service processing device is provided, which is applied to a first network element, including:

[0021] An acquisition module, configured to acquire an establishment request from a second network element to establish an AI service, where the AI ​​service includes at least one task;

[0022] A determination module, configured to determine an AI strategy for at least one task in the AI ​​service according to the establishment request;

[0023] The first sending module is used to return a response message of the establishment request to the second network element according to the AI ​​strategy.

[0024] In a sixth aspect, a service processing device is provided, which is applied to a second network element, including:

[0025] A third sending module is configured to send a request to establish an AI service to the first network element, where the AI ​​service includes at least one task;

[0026] A first receiving module, used to receive a response message of the establishment request returned by the first network element;

[0027] The response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

[0028] In a seventh aspect, a service processing device is provided, which is applied to a third network element, including:

[0029] a fifth sending module, configured to send a fifth message to a sixth network element, where the fifth message is used to request an AI unit required for at least one task of the AI ​​service;

[0030] The third receiving module is used to receive a response message to the fifth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

[0031] In an eighth aspect, a service processing device is provided, which is applied to a fourth network element, including:

[0032] A fourth receiving module, used for receiving first data sent by the second network element;

[0033] A processing module, configured to perform an AI unit processing operation according to the first data;

[0034] The seventh sending module is used to send second data to the second network element.

[0035] In the ninth aspect, a communication device is provided, which network side device includes a processor and a memory, the memory storing a program or instruction that can be run on the processor, and the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect, or implements the steps of the method described in the second aspect, or implements the steps of the method described in the third aspect, or implements the steps of the method described in the fourth aspect.

[0036] In the tenth aspect, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented, or the steps of the method described in the second aspect are implemented, or the steps of the method described in the third aspect are implemented, or the steps of the method described in the fourth aspect are implemented.

[0037] In the eleventh aspect, a communication system is provided, including: a first network element, a second network element, a third network element and a fourth network element, wherein the first network element can be used to execute the steps of the method described in the first aspect, the second network element can be used to execute the steps of the method described in the second aspect, the third network element can be used to execute the steps of the method described in the third aspect, and the fourth network element can be used to execute the steps of the method described in the fourth aspect.

[0038] In the twelfth aspect, a chip is provided, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instructions to implement the steps of the method described in the first aspect, or the steps of the method described in the second aspect, or the steps of the method described in the third aspect, or the steps of the method described in the fourth aspect.

[0039] In the thirteenth aspect, a computer program / program product is provided, wherein the computer program / program product is stored in a storage medium, and the program / program product is executed by at least one processor to implement the steps of the method described in the first aspect, or the steps of the method described in the second aspect, or the steps of the method described in the third aspect, or the steps of the method described in the fourth aspect.

[0040] In a fourteenth aspect, an embodiment of the present application further provides a wireless communication system, including: a first network element, a second network element, a third network element, and a fourth network element, wherein:

[0041] The first network element is used to: obtain an establishment request from the second network element to establish an AI service, where the AI ​​service includes at least one task; determine an AI strategy for at least one task in the AI ​​service according to the establishment request; and return a response message of the establishment request to the second network element according to the AI ​​strategy;

[0042] The second network element is used to: send an establishment request to the first network element for requesting to establish an AI service; and receive a response message to the establishment request returned by the first network element;

[0043] The third network element is used to: send a sixth message to the sixth network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service; receive a response message to the sixth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task;

[0044] The fourth network element is used to: receive the first data sent by the second network element; perform the AI ​​unit processing operation according to the first data; and send the second data to the second network element;

[0045] Among them, the first network element, the third network element and the fourth network element are core network elements, and the second network element is a third-party server.

[0046] In an embodiment of the present application, the first network element can obtain an establishment request from the second network element (such as a third-party server) to establish an AI service, and the AI ​​service includes at least one task, so that the establishment of the AI ​​service can be initiated according to the needs and purposes of the third-party server; thereafter, according to the establishment request, an AI strategy can be determined for at least one task in the AI ​​service, and according to the AI ​​strategy, a response message can be returned to the second network element, thereby realizing a multi-task AI service process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a block diagram of a wireless communication system to which the embodiments of the present application can be applied;

[0048] Figure 2 It is a flow chart of a business processing method provided by an embodiment of the present application;

[0049] Figure 3 is a flow chart of another business processing method provided by an embodiment of the present application;

[0050] Figure 4 is a flow chart of another business processing method provided by an embodiment of the present application;

[0051] Figure 5 is a flow chart of another business processing method provided by an embodiment of the present application;

[0052] Fig. 6A This is one of the flowcharts of the multi-task AI business service process in Example 1 of the present application;

[0053] Figure 6B This is the second flowchart of the multi-task AI business service process in Example 1 of the present application;

[0054] Fig. 7A It is one of the flowcharts of the multi-task AI business service process in Embodiment 2 or 3 of the present application;

[0055] Figure 7B This is the second flowchart of the multi-task AI business service process in Example 2 of the present application;

[0056] Figure 8 This is the second flowchart of the multi-task AI business service process in Example 3 of the present application;

[0057] Fig.9A This is one of the flowcharts of the multi-task AI business service process in Example 4 of the present application;

[0058] Fig. 9B This is the second flowchart of the multi-task AI business service process in Example 4 of the present application;

[0059] Fig. 9CThis is the third flowchart of the multi-task AI business service process in Example 4 of the present application;

[0060] Fig.10 It is a structural diagram of a service processing device provided in an embodiment of the present application;

[0061] Fig.11 is a structural diagram of another service processing device provided in an embodiment of the present application;

[0062] Fig.12 is a structural diagram of another service processing device provided in an embodiment of the present application;

[0063] Fig.13 is a structural diagram of another service processing device provided in an embodiment of the present application;

[0064] Fig.14 It is a structural diagram of a communication device provided in an embodiment of the present application;

[0065] Fig.15 It is a structural diagram of a network side device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of this application.

[0067] The terms "first", "second", etc. of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of one type, and the number of objects is not limited, for example, the first object can be one or more. In addition, "or" in the present application represents at least one of the connected objects. For example, "A or B" covers three schemes, namely, Scheme 1: including A but not including B; Scheme 2: including B but not including A; Scheme 3: including both A and B. The character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0068] The term "indication" in this application can be a direct indication (or explicit indication) or an indirect indication (or implicit indication). A direct indication can be understood as the sender explicitly informing the receiver of specific information, operations to be performed, or request results in the sent indication; an indirect indication can be understood as the receiver determining the corresponding information according to the indication sent by the sender, or making a judgment and determining the operation to be performed or the request result according to the judgment result.

[0069] It is worth noting that the technology described in the embodiments of the present application is not limited to the Long Term Evolution (LTE) / LTE-Advanced (LTE-A) system, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA) or other systems. The terms "system" and "network" in the embodiments of the present application are often used interchangeably, and the described technology can be used for the systems and radio technologies mentioned above as well as for other systems and radio technologies. The following description describes a New Radio (NR) system for illustrative purposes, and NR terminology is used in most of the following descriptions, but these technologies can also be applied to systems other than NR systems, such as the 6th generation (6 th Generation, 6G) communication system.

[0070] Figure 1A block diagram of a wireless communication system applicable to an embodiment of the present application is shown. The wireless communication system includes a terminal 11 and a network side device 12. Among them, the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a handheld computer, a netbook, an ultra-mobile personal computer (Ultra-mobile Personal Computer, UMPC), a mobile Internet device (Mobile Internet Device, MID), an augmented reality (Augmented Reality, AR), a virtual reality (Virtual Reality, VR) device, a robot, a wearable device (Wearable Device), an aircraft (flight vehicle), a vehicle-mounted device (Vehicle User Equipment, VUE), a ship-mounted device, a pedestrian terminal (Pedestrian User Equipment, PUE), a smart home (home appliances with wireless communication functions, such as refrigerators, televisions, washing machines or furniture, etc.), a game console, a personal computer (Personal Computer, PC), a teller machine or a self-service machine and other terminal side devices. Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among them, the vehicle-mounted device can also be called a vehicle-mounted terminal, a vehicle-mounted controller, a vehicle-mounted module, a vehicle-mounted component, a vehicle-mounted chip or a vehicle-mounted unit, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application. The network side device 12 may include an access network device or a core network device, wherein the access network device may also be referred to as a radio access network (Radio Access Network, RAN) device, a radio access network function or a radio access network unit. The access network device may include a base station, a wireless local area network (Wireless Local Area Network, WLAN) access point (Access Point, AP) or a wireless fidelity (Wireless Fidelity, WiFi) node, etc.Among them, the base station may be referred to as a Node B (NB), an evolved Node B (eNB), a next generation Node B (gNB), a New Radio Node B (NR Node B), an access point, a Relay Base Station (RBS), a Serving Base Station (SBS), a Base Transceiver Station (BTS), a radio base station, a radio transceiver, a Basic Service Set (BSS), an Extended Service Set (ESS), a Home Node B (HNB), a Home Evolved Node B (home evolved Node B), a Transmission Reception Point (TRP) or other appropriate terms in the relevant field. As long as the same technical effect is achieved, the base station is not limited to specific technical vocabulary. It should be noted that in the embodiment of the present application, only the base station in the NR system is used as an example for introduction, and the specific type of the base station is not limited.

[0071] The core network equipment may include but is not limited to at least one of the following: core network node, core network function, network data analysis function (Network Data Analytics Function, NWDAF), mobility management entity (Mobility Management Entity, MME), access mobility management function (Access and Mobility Management Function, AMF), session management function (Session Management Function, SMF), user plane function (UserPlane Function, UPF), policy control function (Policy Control Function, PCF), policy and charging rules function unit (Policy and Charging Rules Function, PCRF), edge application service discovery function (Edge Application Server Discovery Function, EASDF), unified data management (Unified Data Management, UDM), unified data storage (Unified Data Repository, UDR), home subscriber server (Home Subscriber Server, HSS), centralized network configuration (CNC), network storage function (Network Repository Function, NRF), network exposure function (Network Exposure Function, NEF), local NEF (Local NEF, or L-NEF), Binding Support Function (BSF), Application Function (AF), AI business management function, collaborative control function, business analysis function, business management function, collaborative control function, AI resource node, computing power execution node, computing power execution function, computing power execution plane, AI model management function, AI model control function, AI algorithm control function, algorithm control function, AI resource management function, computing power control function, computing power control plane, etc.

[0072] In the embodiments of the present application, the AI ​​unit may also be referred to as one of the following: model, AI model, ML (machine learning) model, ML unit, AI structure, AI function, AI characteristics, machine learning model, neural network, neural network function, neural network function, etc.

[0073] Optionally, the AI ​​unit / AI model may refer to a processing unit that can implement specific AI-related algorithms, formulas, processing flows, capabilities, etc., or it may be a processing method, algorithm, function, module or unit for a specific data set, or it may be a processing method, algorithm, function, module or unit running on AI / ML related hardware such as a graphics processor (GPU), a neural processing unit (NPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), etc. This application does not make any specific limitations on this.

[0074] Optionally, the particular data set may include inputs and / or outputs of the AI ​​unit / AI model.

[0075] Optionally, the identifier of the AI ​​unit may include at least one of the following: an AI model identifier, an AI structure identifier, an AI algorithm identifier, an identifier of a specific data set associated with the AI ​​unit, an identifier of a specific scenario, environment, channel feature or device related to AI / ML, an identifier of a function, feature, capability or module related to AI / ML, etc. This application does not make any specific limitations on this.

[0076] In the embodiment of the present application, the third network element is used to coordinate various elements, and the first network element converts external demand into internal demand. The seventh network element is used for the selection and allocation of AI resource nodes, etc. The sixth network element is used for AI model management, model ID allocation, model acquisition, etc. The data plane management function is used for the selection and allocation of data resource nodes.

[0077] In the embodiment of the present application, for the definitions / examples of different network elements, please refer to the following Table 1:

[0078] Table 1

[0079]

[0080]

[0081] Optionally, each network element supports a service-based interface based on a service-based architecture (SBA) or an SBA-enhanced architecture, that is, messages exchanged between network elements can be implemented through a service-based interface protocol.

[0082] The following, in conjunction with the accompanying drawings, describes in detail the business processing method, apparatus, communication device, and readable storage medium provided in the embodiments of the present application through some embodiments and their application scenarios.

[0083] See also Figure 2 , Figure 2 is a flow chart of a service processing method provided in an embodiment of the present application, and the method is executed by a first network element. Figure 2 As shown, the method comprises the following steps:

[0084] Step 21: The first network element obtains a request from the second network element to establish an AI service, where the AI ​​service includes at least one task;

[0085] Step 22: The first network element determines an AI strategy for at least one task in the AI ​​service according to the establishment request;

[0086] Step 23: The first network element returns a response message of the establishment request to the second network element according to the AI ​​strategy.

[0087] In this embodiment, the first network element is a core network element, and the second network element is a third-party server.

[0088] The first network element can be understood as a central control node, which can be selected as but not limited to policy and charging functions, AI service management functions, external service management functions, service management functions, collaborative control functions or computing power service management functions. The policy and charging function PCF is mainly used for application and service data flow detection, policy (such as OTT server policy, UE policy, session policy, etc.) generation and provision, etc. The AI ​​service management function is mainly used for service management and service identity resolution, specifically, to coordinate different element control functions, such as AI resource management function, AI model management function and data management function; when the AI ​​service management function is used as a central control node, it will work with the policy and charging function to convert external AI service requests into AI policy functions that can be understood internally.

[0089] The second network element may be a third-party server such as an OTT server, whose content or service is built on the basic telecommunication service. The AI ​​service is a multi-task service, which may be a verification service, a reasoning service, a training service, etc.

[0090] Optionally, the first network element (i.e., the central control node) can convert the external service QoE requirements into executable internal policies, including AI policies, such as AI service processing delay, AI service processing accuracy, etc., according to the AI ​​service establishment request message. AI policies can include policies for multiple tasks at the same time. For example, the AI ​​service management function can decide on demand whether to interact with other AI network elements and obtain relevant configuration information according to the AI ​​policy; for example, if the second network element does not provide an AI unit, the AI ​​service management function interacts with the AI ​​model management function to obtain the AI ​​unit; if the AI ​​service establishment request contains an indication that the data source is the data plane, the AI ​​service management function interacts with the data plane to obtain the data source. The configuration information may include at least one of the following: AI unit information, AI unit load information (such as AI unit download address), data source information (such as data source address), AI service computing capacity (such as computing power size, type, etc.).

[0091] For example, when the first network element is a policy charging function, that is, the central control node is a policy charging function, the AI ​​policy can be directly generated by the policy charging function; or, when the first network element is an AI service management function, that is, the central control node is an AI service management function, the AI ​​service management function and the policy charging function can negotiate to generate the AI ​​policy.

[0092] It should be noted that the tasks of the AI ​​service may be determined based on actual needs, and may be obtained after the AI ​​service is disassembled in the network. The tasks of the AI ​​service may also be referred to as sub-services, sub-services, etc.

[0093] Therefore, the second network element (such as an OTT server) can initiate the establishment of AI services, thereby realizing a multi-tasking AI business service process.

[0094] Optionally, when the first network element is a policy charging function, the step 21 may include:

[0095] A first network element receives a first message sent by a second network element, where the first message is used to request to establish the AI ​​service for the second network element, and the first message includes at least one of the following:

[0096] The number of tasks for the AI ​​business;

[0097] The timing relationship between the multiple tasks of the AI ​​service; that is, the multiple tasks are sorted according to the chronological order, and the tasks executed first are arranged before the tasks executed later;

[0098] The dependency relationship between the data of multiple tasks of the AI ​​service; for example, the data required for the current task is obtained based on the previous task or the previous tasks;

[0099] The dependency relationship between the AI ​​units of multiple tasks of the AI ​​service; for example, the AI ​​unit required for the current task is an AI unit obtained based on the previous task or previous tasks;

[0100] Task identifiers of multiple tasks of the AI ​​service;

[0101] Task requirements of multiple tasks of the AI ​​business;

[0102] Demand information of the AI ​​service; for example, this demand information can be understood as target information, which may include but is not limited to AI service processing latency, AI service processing accuracy, AI service computing volume, data processing scale, data storage requirements, expected application scope, etc.;

[0103] The Quality of Experience (QoE) requirements of the AI ​​service.

[0104] Optionally, the task requirements may include but are not limited to at least one of the following:

[0105] Single task computing amount;

[0106] Data processing scale;

[0107] Data storage requirements;

[0108] AI unit application scope;

[0109] Data processing delay;

[0110] Data processing accuracy;

[0111] Expected AI unit size.

[0112] Optionally, the first message may further include at least one of the following:

[0113] Service type indication; for example, service type indication is environment reconstruction, face recognition, etc.;

[0114] AI business identification; for example, AI unit reasoning, AI unit deployment, AI unit training, and AI unit verification;

[0115] AI unit related parameters, such as whether the network internal AI unit indication is required, AI unit identification, etc.

[0116] An indication of whether network internal data and the source of the data are required.

[0117] In this embodiment, the first message may be an AI service establishment request message, etc.

[0118] In this embodiment, in addition to the above information, the first message may also include a data network name (Data Network Name, DNN), single network slice selection assistance information (Single Network Slice Selection Assistance information, S-NSSAI), a policy identifier ID, etc. For example, the policy ID may indicate the AI ​​service policy negotiated between the OTT server and the policy charging function.

[0119] Optionally, when the first network element is a policy charging function, the step 23 may include:

[0120] The first network element sends a second message to the second network element, the second message is used to respond to the first message, such as notifying the result of the AI ​​service establishment request, including AI resource node information; or, the second message is used to trigger data transmission or AI unit transmission related to the first task of the AI ​​service. The first task is specifically the task being executed by the AI ​​service.

[0121] Optionally, the second message includes: relevant information of the AI ​​resource node used to execute the first task, such as a node identifier used to identify the AI ​​resource node.

[0122] Here, the AI ​​resource node may also be referred to as a resource node, a computing power resource node, a business resource node, etc., and is used for business processing and implementation of business policy rules, etc.; specifically, the AI ​​business may be computed and processed according to the command of the seventh network element (e.g., the AI ​​resource management function), such as executing on the user plane, or updating the AI ​​resource status to the seventh network element (e.g., the AI ​​resource management function), etc.

[0123] In this embodiment, the second message may be an AI service establishment request response message, etc.

[0124] Optionally, when the first network element is a policy charging function, the service processing method further includes:

[0125] The first network element sends a third message to the third network element, where the third message is used to request allocation of AI resources or required AI units for the AI ​​service. For example, the third message can be used to request allocation of AI resources for the AI ​​service, can be used to request allocation of required AI units for the AI ​​service, or can be used to request allocation of AI resources and required AI units for the AI ​​service.

[0126] Here, the third network element may be selected as an AI service management function, an external service management function, a collaborative control function, a service management function, a computing power service management function or other network functions that can realize similar functions, and is used for service management and service identification resolution, etc.; specifically, it coordinates different element control functions, such as AI resource management function, AI model management function and data management function; when the AI ​​service management function acts as a central control node, it will work together with the policy billing function to convert external AI service requests into AI policy functions that can be understood internally.

[0127] Optionally, the third message may include at least one of the following: an identifier of the second network element (such as an OTT server identifier), an AI service identifier, AI unit-related parameters, an AI strategy, an application server AS address, etc.

[0128] In this embodiment, the third message may be an AI service establishment request message, such as a message sent by a policy charging function to an AI service management function.

[0129] Optionally, when the first network element is an AI service management function, the step 21 may include:

[0130] The first network element receives a fourth message, where the fourth message is used to request to establish an AI service for the second network element, and the fourth message includes at least one of the following:

[0131] an identifier of the second network element;

[0132] AI business logo;

[0133] Related parameters of the AI ​​unit for the AI ​​service, such as an AI unit identifier, an AI unit size, and indication information of whether the AI ​​unit needs to be obtained from the sixth network element;

[0134] an address of the first network element;

[0135] Demand information for AI services;

[0136] The task requirement of at least one task in the AI ​​business;

[0137] The number of tasks for the AI ​​business;

[0138] The timing relationship between multiple tasks of the AI ​​business;

[0139] The dependency relationship between the data of multiple tasks of the AI ​​service; for example, the data required for the current task is obtained based on the previous task or the previous tasks;

[0140] The dependency relationship between the AI ​​units of the multiple tasks of the AI ​​service; for example, the AI ​​unit required for the current task is an AI unit obtained based on the previous task or the previous tasks;

[0141] Task identifiers of multiple tasks of the AI ​​service;

[0142] Demand information for the AI ​​service;

[0143] The QoE requirements of the AI ​​service.

[0144] Optionally, the first network element may be selected by a fifth network element (such as a network open function), and receive the fourth message from the fifth network element.

[0145] Optionally, the task requirements may include but are not limited to at least one of the following:

[0146] Single task computing amount;

[0147] Data processing scale;

[0148] Data storage requirements;

[0149] AI unit application scope;

[0150] Data processing delay;

[0151] Data processing accuracy;

[0152] Expected AI unit size.

[0153] In this embodiment, the first network element and the third network element may be the same network element or jointly set up. When the first network element and the third network element are the same network element or jointly set up, the third message is implemented inside the network element. For example, in this embodiment, the first network element is a policy charging function, and the third network element is an AI service management function. The policy charging function and the AI ​​service management function may be the same network element or jointly set up.

[0154] See also Figure 3 , Figure 3 is a flow chart of a service processing method provided in an embodiment of the present application, the method is executed by the second network element, such as Figure 3 As shown, the method comprises the following steps:

[0155] Step 31: The second network element sends a request to the first network element to establish an AI service, where the AI ​​service includes at least one task;

[0156] Step 32: The second network element receives a response message of the establishment request returned by the first network element.

[0157] In this embodiment, the response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

[0158] In this embodiment, the first network element can be understood as a central control node, which can be selected as but not limited to a policy charging function, an AI service management function, an external service management function, a service management function, a collaborative control function, or a computing service management function. The second network element can be selected as a third-party server such as an OTT server, whose content or service is built on basic telecommunication services. The AI ​​service can be selected as a verification service, an inference service, a training service, etc.

[0159] It should be noted that the tasks of the AI ​​service may be determined based on actual needs, and may be obtained after the AI ​​service is disassembled in the network. The tasks of the AI ​​service may also be referred to as sub-services, sub-services, etc.

[0160] Therefore, the second network element (such as an OTT server) can initiate the establishment of AI services, thereby realizing a multi-tasking AI business service process.

[0161] Optionally, when the first network element is a policy charging function, the step 31 may include:

[0162] The second network element sends a first message to the first network element, where the first message is used to request to establish an AI service for the second network element, and the first message includes at least one of the following:

[0163] The number of tasks for the AI ​​business;

[0164] The timing relationship between the multiple tasks of the AI ​​service; that is, the multiple tasks are sorted according to the chronological order, and the tasks executed first are arranged before the tasks executed later;

[0165] The dependency relationship between the data of multiple tasks of the AI ​​service; for example, the data required for the current task is obtained based on the previous task or the previous tasks;

[0166] The dependency relationship between the AI ​​units of the multiple tasks of the AI ​​service; for example, the AI ​​unit required for the current task is an AI unit obtained based on the previous task or the previous tasks;

[0167] Task identifiers of multiple tasks of the AI ​​service;

[0168] Task requirements of multiple tasks of the AI ​​business;

[0169] Demand information for AI services; for example, this demand information can be understood as target information, which may include but is not limited to AI service processing latency, AI service processing accuracy, AI service computing volume, data processing scale, data storage requirements, expected application scope, etc.;

[0170] QoE requirements for AI services.

[0171] Optionally, the task requirements may include but are not limited to at least one of the following:

[0172] Single task computing amount;

[0173] Data processing scale;

[0174] Data storage requirements;

[0175] AI unit application scope;

[0176] Data processing delay;

[0177] Data processing accuracy;

[0178] Expected AI unit size.

[0179] Optionally, the first message may further include at least one of the following:

[0180] AI business identification; for example, identification of business as environment reconstruction, facial recognition, etc.

[0181] AI unit related parameters, such as whether the network internal AI unit indication is required, AI unit identification, etc.

[0182] An indication of whether network internal data and the source of the data are required.

[0183] In this embodiment, the first message may be an AI service establishment request message, etc.

[0184] Optionally, when the first network element is a policy charging function, the above step 32 may include:

[0185] The second network element receives a second message sent by the first network element; the second message is used to respond to the first message, such as notifying the result of the AI ​​service establishment request, including AI resource node information; or, the second message is used to trigger data transmission or AI unit transmission related to the first task of the AI ​​service; the first task is specifically the task being executed by the AI ​​service.

[0186] Optionally, the second message includes: relevant information of the AI ​​resource node used to execute the first task, such as a node identifier used to identify the AI ​​resource node.

[0187] In this embodiment, the second message may be an AI service establishment request response message, that is, a response message to the above-mentioned AI service establishment request message.

[0188] Optionally, when the first network element is an AI service management function, the step 31 may include:

[0189] The second network element sends a fourth message to the first network element through the fifth network element, where the fourth message is used to request to establish an AI service for the second network element, the first network element is selected by the fifth network element, and the fourth message includes at least one of the following:

[0190] an identifier of the second network element;

[0191] AI business logo;

[0192] Related parameters of the AI ​​unit for the AI ​​service, such as an AI unit identifier, an AI unit size, and indication information of whether the AI ​​unit needs to be obtained from the sixth network element;

[0193] an address of the first network element;

[0194] Demand information for AI services;

[0195] The task requirement of at least one task in the AI ​​business;

[0196] The number of tasks for the AI ​​business;

[0197] The timing relationship between multiple tasks of the AI ​​business;

[0198] The dependency relationship between the data of multiple tasks of the AI ​​service; for example, the data required for the current task is obtained based on the previous task or the previous tasks;

[0199] The dependency relationship between the AI ​​units of the multiple tasks of the AI ​​service; for example, the AI ​​unit required for the current task is an AI unit obtained based on the previous task or the previous tasks;

[0200] Task identifiers of multiple tasks of the AI ​​service;

[0201] Demand information for the AI ​​service;

[0202] The QoE requirements of the AI ​​service.

[0203] Optionally, the service processing method may further include:

[0204] The second network element sends first data to the fourth network element, where the first data includes at least one of the following:

[0205] Input data of an AI unit for the AI ​​service, such as model input data;

[0206] Label data used for AI unit processing operations; for example, the label data can be used for AI unit training or AI unit monitoring;

[0207] sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service;

[0208] The structure of the AI ​​unit used for the AI ​​service;

[0209] Parameters of the AI ​​unit used for the AI ​​service;

[0210] Description information of the AI ​​unit used for the AI ​​service;

[0211] Output of a segmentation layer in an AI unit for the AI ​​service; for example, the segmentation layer is used for data reasoning, etc.;

[0212] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0213] Therefore, the fourth network element can perform processing operations of the AI ​​unit according to the received first data, such as AI unit training, AI unit monitoring or AI unit verification.

[0214] Optionally, the first data may be interpretable data or privately processed data, such as encrypted data. The fourth network element is an AI resource node, a resource node, a computing resource node, a computing execution node, a computing execution plane, a business resource node, etc., which is used for business processing and business policy rule implementation, etc.; specifically, the AI ​​business can be calculated and processed according to the command of the seventh network element, such as execution on the user plane, or AI resource status can be updated to the seventh network element.

[0215] Optionally, the second network element may send the first data to the fourth network element through the fifth network element. The fifth network element is, for example, a network open function, which is located between the core network (such as the 6G core network) and an external third-party server (possibly also part of the internal AF), and is responsible for opening network capabilities and network internal information; all external applications that want to access the internal data of the core network must go through the network open function. The network open function provides corresponding security guarantees to ensure the security of external applications to the core network, and provides functions such as external application target / demand customization capability opening, mobility status event subscription, and AF request distribution.

[0216] Optionally, the service processing method may further include:

[0217] The second network element receives second data sent by the fourth network element, where the second data includes at least one of the following:

[0218] Input data of an AI unit for the AI ​​service, such as model input data;

[0219] Label data used for AI unit processing operations; for example, the label data can be used for AI unit training or AI unit monitoring;

[0220] Output data of the AI ​​unit used for the AI ​​service, such as model output data;

[0221] Input data of an AI unit for the AI ​​service generated based on reinforcement learning;

[0222] AI unit structure for the AI ​​service;

[0223] AI unit parameters for the AI ​​service;

[0224] AI unit description information for the AI ​​service;

[0225] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0226] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0227] Optionally, the second data may be interpretable data, or privately processed data, such as encrypted data.

[0228] Optionally, the second network element may perform processing operations of the AI ​​unit according to the received second data, such as AI unit verification, AI unit training, or AI unit monitoring. For example, in a vertical federated learning scenario, the second network element (such as an OTT server) and the fourth network element (such as an AI resource node) may jointly perform model training and respectively update model parameters.

[0229] See also Figure 4 , Figure 4 is a flow chart of a service processing method provided in an embodiment of the present application, the method is executed by a third network element, such as Figure 4 As shown, the method comprises the following steps:

[0230] Step 41: The third network element sends a fifth message to the sixth network element, where the fifth message is used to request an AI unit required for at least one task of the AI ​​service;

[0231] Step 42: The third network element receives a response message to the fifth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

[0232] In this embodiment, the third network element and the sixth network element may specifically be as described above.

[0233] In this embodiment, the fifth message may be an AI unit request message, a model request message, etc., which is used to determine the AI ​​unit / model information.

[0234] Therefore, with the help of the fifth message, the AI ​​unit required for the AI ​​service can be determined, thereby realizing a multi-tasking AI service process.

[0235] In this embodiment, the third network element (such as the AI ​​service management function) can request the AI ​​units required for multiple tasks (such as all tasks) at one time, or request the AI ​​units required for a single task at one time, such as requesting the required AI units for each task separately.

[0236] Optionally, when the fifth message is used to request AI units required for multiple tasks of an AI service, that is, requesting AI units required for multiple tasks at one time, the fifth message may include but is not limited to at least one of the following:

[0237] A first indication, where the first indication is used to indicate a task among the multiple tasks that requires obtaining an initial AI unit from a sixth network element; for example, indicating a task that requires obtaining an initial AI unit from an AI model management function;

[0238] a second indication, the second indication being used to indicate dependencies between the AI ​​units required for the plurality of tasks;

[0239] Use case identifiers corresponding to each of the multiple tasks;

[0240] The service types corresponding to the multiple tasks; for example, environment reconstruction, facial recognition, etc.;

[0241] Business identifiers of the multiple tasks; for example, AI unit training, AI unit reasoning, AI unit verification, etc.;

[0242] The expected accuracy of the multiple tasks;

[0243] data storage requirements for the multiple tasks;

[0244] The input data formats of the plurality of tasks;

[0245] Output data formats of the plurality of tasks;

[0246] Generalization metrics for the various tasks described.

[0247] Optionally, when the fifth message is used to request an AI unit required for a second task of the AI ​​service, such as requesting an AI unit required for a current task, the second task is a single task among multiple tasks of the AI ​​service, such as a currently executed task, and the fifth message may include but is not limited to at least one of the following:

[0248] The task identifier of the second task;

[0249] The business identifier corresponding to the second task; for example, AI unit training, AI unit reasoning, or AI unit verification;

[0250] The use case identifier corresponding to the second task;

[0251] The service type corresponding to the second task; for example, environment reconstruction, facial recognition, etc.

[0252] Expected accuracy of the second task;

[0253] Data storage requirements for the second task;

[0254] Data input format for the second task;

[0255] Data output format for the second task;

[0256] Generalization index of the second task;

[0257] An identifier of a third task, where the third task is a task before the second task and associated with the second task, that is, the third task is a previously trained task associated with the second task.

[0258] Optionally, the description is made from a business perspective, that is, when the multiplexed AI business only includes a single task, when the fifth message is used to request an AI unit required for a second task of the AI ​​business, such as requesting an AI unit required for a current task, the second task is a single task among multiple tasks of the AI ​​business, such as a currently executed task, and the fifth message includes but is not limited to at least one of the following:

[0259] Task identification;

[0260] Business identification; for example, AI unit training, AI unit reasoning, or AI unit verification;

[0261] Use case identification;

[0262] Service type; for example, environment reconstruction, facial recognition, etc.

[0263] Expected accuracy;

[0264] Data storage requirements;

[0265] Input data format;

[0266] Output data format;

[0267] Generalizability index;

[0268] An identifier of a fourth task, where the fourth task is a task before the second task and associated with the second task, that is, the fourth task is a previously trained task associated with the second task.

[0269] In this way, with the help of the content carried in the fifth message, the AI ​​units required for multiple tasks or a single task of the AI ​​service can be determined.

[0270] Optionally, the generalization index includes at least one of the following:

[0271] 1) The identifier of the first data set, the identifier of the second data set, and the difference in the index of the AI ​​unit under the second data set and the first data set, wherein the first data set is a data set for AI unit training, and the second data set is a data set for AI unit verification; for example, the data distribution characteristics of the first data set are different from the data distribution characteristics of the second data set, so as to verify the performance of the AI ​​unit trained according to the first data set under the second data set;

[0272] 2) A sixth indication, an indicator threshold of the AI ​​unit under a supportable data set, wherein the sixth indication is used to indicate a data set that the AI ​​unit can support; in this case, the indicator threshold can be used to verify the performance of the AI ​​unit under the supportable data set; the sixth indication can be used to characterize the scope of application of the model under the condition that the indicator threshold is met;

[0273] 3) The identifier of the first scenario, the identifier of the second scenario, and the difference in the index of the AI ​​unit in the second scenario and the first scenario, wherein the first scenario is the scenario when the AI ​​unit is trained, and the second scenario is the scenario when the AI ​​unit is verified; in this case, the scenario when the AI ​​unit is trained is different from the scenario when the AI ​​unit is verified, for example, the scenario when the AI ​​unit is trained is a sparse large rural scenario, while the scenario when the AI ​​unit is verified is a dense urban scenario, so as to verify the performance of the AI ​​unit trained in the first scenario in the second scenario;

[0274] 4) The seventh indication, the index threshold of the AI ​​unit in the supported scenario, the seventh indication is used to indicate the supported scenario of the AI ​​unit; in this case, with the help of the index threshold, the performance of the AI ​​unit in the supported scenario can be verified; with the help of the sixth indication, the scope of application of the model under the condition of meeting the index threshold can be represented;

[0275] 5) The identifier of the first field, the identifier of the second field, and the difference between the indicators of the AI ​​unit in the second field and the first field, wherein the first field is the field when the AI ​​unit is trained, and the second field is the field when the AI ​​unit is verified; in this case, the field when the AI ​​unit is trained is different from the field when the AI ​​unit is verified, for example, the field when the AI ​​unit is trained is speech recognition in the legal field, while the field when the AI ​​unit is verified is speech recognition in the fine arts field, so as to verify the performance of the AI ​​unit trained in the first field in the second field;

[0276] 6) An eighth indication, an indicator threshold of the AI ​​unit in a supportable field, wherein the eighth indication is used to indicate the supportable field of the AI ​​unit; in this case, with the help of the indicator threshold, the performance of the AI ​​unit in the supportable field can be verified; with the help of the sixth indication, the scope of application of the model can be represented when the indicator gate is met.

[0277] Here, the above 1) and 2) can be understood as a data set description method for generalization indicators, the above 3) and 4) can be understood as a scene description method for generalization indicators, and the above 5) and 6) can be understood as a domain description method for generalization indicators.

[0278] The above-mentioned indicator may be selected from but not limited to at least one of the following items of the AI ​​unit: accuracy, recall, normalized mean square error, normalized root mean square error, etc. For example, the above-mentioned indicator difference is the accuracy difference, and the above-mentioned indicator threshold is the accuracy threshold; or, the above-mentioned indicator difference is the recall difference, and the above-mentioned indicator threshold is the recall threshold; and so on.

[0279] Optionally, the service processing method further includes:

[0280] The third network element sends a sixth message to the seventh network element, where the sixth message is used to request AI resources for at least one task of the AI ​​service.

[0281] Here, the seventh network element can be selected as AI resource management function, resource management function, computing power control function, AI resource control function, computing power resource management function or other network functions that can realize similar functions, used for business resource management, etc.; specifically, the allocation, update and release of AI / computing power resources can be performed on demand; the status update and maintenance of AI / computing power resource nodes can be performed, etc.

[0282] In this embodiment, the sixth message may be an AI resource request message, which is used to request required AI resources.

[0283] In this embodiment, at least two (i.e., two or three) of the third network element, the sixth network element, and the seventh network element may be the same network element or may be jointly provided. For example, when the third network element in this embodiment is an AI service management function, the sixth network element is an AI model management function, and the seventh network element is an AI resource management function, at least two of the AI ​​service management function, the AI ​​model management function, and the AI ​​resource management function may be the same network element or may be jointly provided. At this point, the interaction between the corresponding network elements belongs to internal implementation.

[0284] In this embodiment, the third network element (such as the AI ​​service management function) can request AI resources for multiple tasks (such as all tasks) at one time, or request AI resources for a single task at one time, such as requesting the required AI resources for each task separately.

[0285] Optionally, when the sixth message is used to request AI resources for multiple tasks of the AI ​​service, that is, requesting AI resources for multiple tasks (such as all tasks) at one time, the sixth message may include but is not limited to at least one of the following:

[0286] AI unit information corresponding to the multiple tasks; for example, including the AI ​​unit download address, whether to use the AI ​​unit of the previous task, the associated task identifier, etc.;

[0287] The data source information of the multiple tasks, such as data source addresses;

[0288] The computational amount of the multiple tasks, such as computing power, type, etc.;

[0289] the maximum data storage requirements of the plurality of tasks;

[0290] the processing latency requirements of the multiple tasks;

[0291] The input data formats of the plurality of tasks;

[0292] The number of single-sample inferences for each of the multiple tasks, that is, the number of AI unit executions for each task.

[0293] Optionally, when the sixth message is used to request AI resources for the fifth task of the AI ​​service, such as requesting the required AI resources for the current task, the fifth task is a single task among multiple tasks of the AI ​​service, such as a currently executed task, and the sixth message may include but is not limited to at least one of the following:

[0294] The task identifier of the fifth task;

[0295] AI unit information corresponding to the fifth task; for example, including the AI ​​unit download address, whether to use the AI ​​unit of the previous task, and the associated task identifier;

[0296] The data source information of the fifth task, such as the data source address;

[0297] The computational amount of the fifth task, such as computing power, type, etc.

[0298] Data storage requirements for the fifth task;

[0299] The processing latency requirement of the fifth task;

[0300] The input data format for the fifth task;

[0301] The number of single-sample inferences for the fifth task, that is, the number of AI unit executions for the fifth task.

[0302] Optionally, the description is made from a business perspective, that is, when the multiplexed AI business only includes a single task, when the sixth message is used to request AI resources for the fifth task of the AI ​​business, such as requesting the required AI resources for the current task, the fifth task is a single task among multiple tasks of the AI ​​business, such as a currently executed task, and the sixth message may include but is not limited to at least one of the following:

[0303] Task identification;

[0304] AI unit information, such as the AI ​​unit download address, whether to use the AI ​​unit of the previous task, and the associated task identifier;

[0305] Data source information, such as the data source address;

[0306] Computational capacity, such as computing power, type, etc.

[0307] Data storage requirements;

[0308] Handling latency requirements;

[0309] Input data format;

[0310] The number of single-sample inferences, that is, the number of AI unit executions.

[0311] In this way, with the help of the content carried in the sixth message, the AI ​​resources required for multiple tasks or a single task of the AI ​​service can be determined.

[0312] Optionally, the service processing method further includes:

[0313] When completing the kth task of the AI ​​service, the third network element performs at least one of the following:

[0314] triggering the AI ​​resource of the next task of the k-th task;

[0315] For the next task of the k-th task, obtain a new initial AI unit; for example, the new initial AI unit may be obtained from an AI model management function, or may be an AI unit obtained from a previous task;

[0316] Among them, the kth task is not the last task of the AI ​​business.

[0317] See also Figure 5 , Figure 5 is a flow chart of a service processing method provided in an embodiment of the present application, the method is executed by a fourth network element, such as Figure 5 As shown, the method comprises the following steps:

[0318] Step 51: The fourth network element receives first data sent by the second network element;

[0319] Step 52: The fourth network element performs an AI unit processing operation according to the first data;

[0320] Step 53: The fourth network element sends the second data to the second network element.

[0321] Here, the fourth network element and the second network element may specifically be as described above.

[0322] Optionally, the AI ​​unit processing operation may include at least one of the following:

[0323] AI unit aggregation;

[0324] Single AI unit training;

[0325] Multiple AI unit training;

[0326] Single AI unit reasoning;

[0327] Multiple AI unit reasoning.

[0328] Therefore, the fourth network element can perform AI unit processing operations, such as reasoning operations, based on the received first data, thereby realizing a multi-tasking AI business service process.

[0329] It should be noted that in some scenarios, the fourth network element (such as an AI resource node) may send the second data to the second network element (such as an OTT server), and then the second network element (such as an OTT server) may perform an AI unit processing operation based on the received second data. For example, in a vertical federated learning scenario, the fourth network element (such as an AI resource node) may send the second data to the second network element (such as an OTT server), and then the second network element (such as an OTT server) may update the AI ​​model parameters based on the received second data.

[0330] Optionally, the fourth network element may receive the first data sent by the second network element through a fifth network element (such as a network open function).

[0331] Optionally, the fourth network element may send the second data to the second network element through a fifth network element (such as a network open function).

[0332] Optionally, the first data may include at least one of the following provided by a second network element (such as an OTT server):

[0333] AI unit input data for the AI ​​service, such as model input data;

[0334] Label data used for AI unit processing operations; for example, the label data can be used for AI unit training or AI unit monitoring;

[0335] sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service;

[0336] The structure of the AI ​​unit used for the AI ​​service;

[0337] Parameters of the AI ​​unit used for the AI ​​service;

[0338] Description information of the AI ​​unit used for the AI ​​service;

[0339] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0340] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0341] Optionally, the second data may include at least one of the following provided by the network:

[0342] AI unit input data for the AI ​​service, such as input data for the model;

[0343] Label data used for AI unit processing operations; for example, the label data can be used for AI unit training or AI unit monitoring;

[0344] Output data of the AI ​​unit used for the AI ​​service, such as model output data;

[0345] Input data of an AI unit for the AI ​​service generated based on reinforcement learning;

[0346] The structure of the AI ​​unit used for the AI ​​service;

[0347] Parameters of the AI ​​unit used for the AI ​​service;

[0348] Description information of the AI ​​unit used for the AI ​​service;

[0349] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0350] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0351] Optionally, the service processing method may further include:

[0352] When completing the kth task of the AI ​​service, the fourth network element performs at least one of the following:

[0353] Do not release AI resources;

[0354] Obtaining AI resources for the next task of the k-th task;

[0355] For the next task of the k-th task, obtain a new initial AI unit, or use the AI ​​unit obtained after performing the k-th task;

[0356] Among them, the kth task is not the last task of the AI ​​business.

[0357] That is to say, when the AI ​​resource node completes the kth task (which is not the last task), the AI ​​resource will not be released, but the AI ​​resource allocation can be modified for the next task through an AI resource request or an AI resource modification request, that is, the AI ​​resource for the next task can be obtained; it can also determine whether to obtain a new initial AI unit from the AI ​​model management function or use the AI ​​unit obtained from the previous task based on the dependency relationship between the AI ​​units between tasks. For example, the AI ​​business management function can first obtain a new initial AI unit from the AI ​​model management function, then send it to the AI ​​resource management function, and then the AI ​​resource management function forwards it to the AI ​​resource node.

[0358] Optionally, the service processing method may further include:

[0359] The fourth network element sends first information to the sixth network element, where the first information includes at least one of the following:

[0360] Updated AI unit parameters;

[0361] Updated AI unit logo;

[0362] The task ID associated with the updated AI unit.

[0363] Here, the sixth network element may be an AI model / AI unit management function, a model management function or other network functions capable of implementing similar functions, for model / AI unit management. The AI ​​unit may be updated through the received first information.

[0364] For example, if the current task is a training task, after the training is completed, the fourth network element (such as an AI resource node) can send the updated model parameters / model identifier and the associated task identifier to the sixth network element (such as an AI model management function) through the seventh network element (such as an AI resource management function).

[0365] Optionally, the service processing method may further include:

[0366] The fourth network element receives configuration information sent by the seventh network element, where the configuration information is used to configure AI resources, and the configuration information includes at least one of the following:

[0367] A third indication, wherein the third indication is used to indicate an AI framework;

[0368] A fourth indication, wherein the fourth indication is used to indicate an AI library version;

[0369] A fifth indication, wherein the fifth indication is used to indicate the number of single sample inferences;

[0370] Data source information, such as data source address;

[0371] AI business detection information;

[0372] AI business processing rules; for example, if there are two business flows, the first business flow can be processed according to the first rule, and the second business flow can be processed according to the second rule.

[0373] Here, the seventh network element may specifically be as described above. The AI ​​framework is a set of standard interfaces, feature libraries, and toolkits for the design, training, and verification of AI algorithm models. It integrates the encapsulation of algorithms, the calling of data, and the use of computing resources. It also provides a development interface and an efficient execution platform for developers. It is an essential tool for the development of AI algorithms. Examples of AI frameworks include TensorFlow, PyTorch, Keras, Caffe, and MXNet. The AI ​​library may be a deep learning function library. Examples of AI library versions include versions of different libraries such as NumPy, Matplotlib, and Pandas.

[0374] The embodiments of the present application are described in detail below in conjunction with specific embodiments.

[0375] Example 1

[0376] In this embodiment 1, the policy and billing function acts as a central control node. Due to the need to establish a multi-task AI service, the OTT server requests the policy and billing function to establish a multi-task AI service; the policy and billing function allocates the corresponding AI policy according to the AI ​​service establishment request, and the AI ​​policy is used to allocate corresponding resources to the corresponding network element in the network; the policy and billing function shall comprehensively consider the end-to-end service requirements when allocating the policy; the policy and billing function initiates the AI ​​service establishment request process to the AI ​​service management function. Fig. 6A and Figure 6B As shown in the figure, the specific multi-task AI business service process includes:

[0377] Step 1. The second network element (e.g., an OTT server) initiates an AI service establishment request message to the fifth network element (e.g., a network open function), wherein the request message is used to request the network to establish a corresponding AI service for it; the request message is specifically the first message mentioned above, and the information included therein may be the content included in the first message mentioned above, and to avoid repetition, it will not be repeated here.

[0378] Step 2. Optionally, a fifth network element (eg, a network open function) authenticates the request message to determine whether the OTT server can request such a service.

[0379] Optionally, the fifth network element (e.g., a network open function) may request the OTT server's OTT-related data from the UDM / UDR, including but not limited to contract data, policy data, etc., to verify whether the AI ​​service-related contract information exists; if it exists, subsequent processes are executed; if it does not exist, the fifth network element (e.g., a network open function) rejects the OTT server's AI service establishment request.

[0380] Step 3. The fifth network element (eg, network open function) forwards the received AI service establishment request message to the first network element (eg, policy charging function).

[0381] Step 4. The first network element (e.g., policy and charging function) generates a corresponding AI policy according to the AI ​​service establishment request message; the AI ​​policy includes a multi-task AI policy. It should be noted that the first network element (e.g., policy and charging function) should comprehensively consider the end-to-end service requirements of the AI ​​service when allocating the AI ​​policy.

[0382] Step 5. The first network element (eg, policy charging function) selects a third network element (eg, AI service management function) according to the received AI service establishment request message.

[0383] Optionally, the first network element (e.g., a policy charging function) may comprehensively consider some factors and use NRF or local configuration information to discover a matching third network element (e.g., an AI service management function) instance, and the factors considered may include at least one of the following:

[0384] -Area of ​​interest;

[0385] -UE AI task-related contract information;

[0386] -Data network name DNN;

[0387] - Slice information S-NSSAI;

[0388] -AI business logo;

[0389] -Supported AI services;

[0390] -Third network element (e.g. AI service management function) instance capabilities, such as AI service processing latency, AI service processing accuracy, AI service computing capacity, etc.

[0391] Step 6. The first network element (eg, policy charging function) sends an AI service establishment request message to the selected third network element (eg, AI service management function) to request allocation of corresponding AI resources and required models for the AI ​​service.

[0392] Optionally, the AI ​​resource establishment request message includes at least one of the following: OTT server identifier, AI service identifier, model-related parameters, AI strategy, AS address, etc.

[0393] Optionally, the first network element (eg, policy charging function) and the third network element (eg, AI service management function) may be integrated. If integrated, step 6 is implemented inside the network element.

[0394] Step 7a. The third network element (e.g., AI service management function) sends a model request message to the sixth network element (e.g., AI model management function) to determine the model information. The model request message is specifically the fifth message mentioned above, and the method of requesting the model can be as follows:

[0395] Implementation method 1: Only request the model required for this task, that is, request the required model for each task;

[0396] Implementation method 2: Request the models required for all tasks, that is, request the models required for all tasks at one time.

[0397] Optionally, step 7a may be performed if at least one of the following conditions is met:

[0398] 1) The AI ​​service establishment request indicates that the OTT server does not contain model information;

[0399] 2) The AI ​​service establishment request contains indication information, indicating that interaction with the AI ​​model management function is required;

[0400] 3) The AI ​​business management function cannot obtain model information.

[0401] Optionally, the third network element (e.g., the AI ​​service management function) may select a corresponding sixth network element (e.g., the AI ​​model management function) according to at least one of the following:

[0402] -DNN;

[0403] -S-NSSAI;

[0404] -AI business logo;

[0405] -Supported AI services;

[0406] -Service area information;

[0407] -Supported model information.

[0408] Step 7b. The sixth network element (eg, the AI ​​model management function) sends a model request response to the third network element (eg, the AI ​​service management function) to indicate the model information.

[0409] Optionally, a third network element (eg, an AI service management function) may determine data source information based on AI model information; if the data source has been determined, steps 8a and 8b will not be executed.

[0410] Step 8a. Optionally, a third network element (such as an AI service management function) sends a data source request message to the data plane management function to determine the data source information; the specific implementation method may be as follows:

[0411] Implementation method 1: Only request the data source required for this task, that is, request the required data source for each task;

[0412] Implementation method 2: Request the data sources required by all tasks, that is, request the data sources required by all tasks at one time.

[0413] Optionally, step 8a may be performed if at least one of the following conditions is met:

[0414] (1) The AI ​​session establishment request indicates that the OTT server does not contain data source information;

[0415] (2) The AI ​​session establishment request includes indication information indicating that interaction with the data plane management function is required;

[0416] (3) The AI ​​business management function cannot obtain data source information.

[0417] Optionally, the third network element (e.g., an AI service management function) may select a corresponding data plane management function according to at least one of the following:

[0418] -DNN;

[0419] -S-NSSAI;

[0420] -AI business logo;

[0421] -Supported AI services;

[0422] -Service area information;

[0423] -Data source information.

[0424] Step 8b. The data plane management function sends a data source request response message to the third network element (eg, the AI ​​service management function) to indicate the data source information.

[0425] Step 9a. The third network element (e.g., the AI ​​service management function) sends an AI resource request message to the seventh network element (e.g., the AI ​​resource management function). The AI ​​resource request message is specifically the sixth message described above. The AI ​​resource request method may be as follows:

[0426] Implementation method 1: Request AI resources based on single task requirements, that is, request the required AI resources for each task;

[0427] Implementation method 2: Request AI resources based on the needs of all tasks, that is, request AI resources required by all tasks at once;

[0428] Implementation method 3: Request AI resources based on business needs (signaling), that is, request the required AI resources for each task.

[0429] Step 10. The seventh network element (e.g., the AI ​​resource management function) determines a suitable fourth network element (e.g., the AI ​​resource node) according to the AI ​​resource request information and the AI ​​policy, and sends AI resource configuration information to the fourth network element (e.g., the AI ​​resource node), where the configuration information may include but is not limited to at least one of the following:

[0430] -Data source information, such as data source address;

[0431] -AI business detection information;

[0432] -AI business processing rules;

[0433] -AI framework instructions;

[0434] -AI library version indication;

[0435] -Number of single-sample inferences.

[0436] Step 9b. The seventh network element (e.g., AI resource management function) sends an AI resource request response message to the third network element (e.g., AI service management function), which includes information of the fourth network element (e.g., AI resource node), and the AI ​​resource node information includes at least one of the following:

[0437] - IP address;

[0438] -ID information;

[0439] -Fully Qualified Domain Name (FQDN);

[0440] -Task identification;

[0441] -AI resource address;

[0442] - AI resource type;

[0443] -AI resource size.

[0444] Optionally, the third network element (e.g., AI service management function), the sixth network element (e.g., AI model management function) and the seventh network element (e.g., AI resource management function) can be combined; if combined, steps 7a, 7b, 9a and 9b are implemented internally in the network element.

[0445] Step 11. The third network element (eg, AI service management function) sends a task resource establishment notification message to the first network element (eg, policy charging function), which includes the task identifier of the current task, AI resource node information, etc.

[0446] Step 12. The first network element (e.g., policy charging function) sends a task resource establishment notification message to the OTT server to trigger data or model interaction related to task 1. Task 1 is, for example, a currently executed task. The response message is specifically the second message described above, and the information included therein may be the content included in the second message described above, and to avoid repetition, it is not repeated here.

[0447] In one implementation, the response message is forwarded by a fifth network element (eg, a network open function).

[0448] Step 13. The second network element (e.g., OTT server) sends the first data (i.e., AI service data) to the fourth network element (e.g., AI resource node). One way is to forward it through the fifth network element (e.g., network open function).

[0449] Step 14. Optionally, a fourth network element (eg, an AI resource node) obtains in-network data from a data source.

[0450] Step 15. The fourth network element (e.g., AI resource node) performs at least one of the following operations according to the AI ​​resource configuration:

[0451] - Model aggregation;

[0452] - Single model training;

[0453] -Multiple model training;

[0454] -Single model inference;

[0455] -Multiple model inferences.

[0456] Step 16. The fourth network element (eg, an AI resource node) sends the second data (ie, AI service data) to the second network element (eg, an OTT server). One way is to forward it through the fifth network element (eg, a network open function).

[0457] Optionally, the first data and the second data may be specifically as shown in the above embodiment, and will not be described again here to avoid repetition.

[0458] Step 17. If the task is a training task, the training is completed, that is, after the model converges, the fourth network element (e.g., AI resource node) sends the updated model parameter information and model identifier, as well as the associated task identifier, to the seventh network element (e.g., AI resource management function); thereafter, the seventh network element (e.g., AI resource management function) forwards the updated model parameter information and model identifier, as well as the associated task identifier, to the sixth network element (e.g., AI model management function). The sixth network element (e.g., AI model management function) performs model parameter update or model update based on the updated model parameter information and model identifier, which may also include model version update.

[0459] Step 18. The seventh network element (eg, AI resource management function) sends a task completion notification to the third network element (eg, AI service management function); the third network element (eg, AI service management function) feeds back ACK to the seventh network element (eg, AI resource management function).

[0460] Optionally, if the above step 7a adopts implementation method 1, that is, only the model required for this task is requested, and the model required for the next task is different from the model used or generated by the previous task, then Figure 6B , continue with the following steps:

[0461] Step 19a. The third network element (e.g., AI service management function) sends a model request message to the sixth network element (e.g., AI model management function) to determine the model information required for this task; please refer to the above step 7a for details.

[0462] Step 19b. The sixth network element (eg, AI model management function) sends a model request response to the third network element (eg, AI service management function) to indicate the model information.

[0463] Step 20a. If the above step 8b adopts implementation method 1, and the data source required to be used by the next task is different from the data source used or generated by the previous task, the third network element (such as the AI ​​service management function) sends a data source request message to the data plane management function to determine the data source information.

[0464] Step 20b. The data plane management function sends a data source request response message to the third network element (eg, the AI ​​service management function) to indicate the data source information.

[0465] Step 21a. The third network element (eg, the AI ​​service management function) sends an AI resource request message to the seventh network element (eg, the AI ​​resource management function), where the AI ​​resource request message is specifically the sixth message described above.

[0466] Step 22. The seventh network element (e.g., AI resource management function) determines a suitable AI resource node according to the AI ​​resource request information and the AI ​​policy, and sends AI resource configuration update information to the AI ​​resource node. The configuration update information may include but is not limited to at least one of the following:

[0467] -Data source information, such as data source address;

[0468] -AI business detection information;

[0469] -AI business processing rules;

[0470] -AI framework instructions;

[0471] -AI library version indication;

[0472] -Number of single-sample inferences.

[0473] Step 21b. The seventh network element (eg, the AI ​​resource management function) sends an AI resource request response message to the third network element (eg, the AI ​​service management function), including the AI ​​resource address, type, size, etc.

[0474] Step 23. The third network element (eg, AI service management function) sends a task resource establishment notification message to the first network element (eg, policy charging function), which includes the task identifier of the current task, etc.

[0475] Step 24. The first network element (e.g., policy charging function) sends a task resource establishment notification message to the second network element (e.g., OTT server) to trigger data or model interaction related to the current task. The response message is specifically the second message described above, and the information included therein may be the content included in the second message described above. To avoid repetition, it will not be repeated here.

[0476] Step 25. The second network element (e.g., OTT server) sends the first data (i.e., AI service data) to the fourth network element (e.g., AI resource node). One way is to forward it through the fifth network element (e.g., network open function).

[0477] Step 26. Optionally, a fourth network element (eg, an AI resource node) obtains in-network data from a data source.

[0478] Step 27. The fourth network element (eg, AI resource node) performs model reasoning according to the AI ​​resource configuration.

[0479] Step 28. The second network element (e.g., OTT server) sends the second data to the fourth network element (e.g., AI resource node), one way of forwarding it through the fifth network element (e.g., network open function). The second data is specifically AI service data, such as including multiple inference results.

[0480] Step 29. The second network element (e.g., an OTT server) sends the third data, such as a sorting result including multiple inference results, to the fourth network element (e.g., an AI resource node). One way is to forward it through a fifth network element (e.g., a network open function).

[0481] Step 30. The fourth network element (eg, AI resource node) performs model training.

[0482] Step 31. After the model converges, the fourth network element (e.g., AI resource node) sends updated model parameter information and model identifier, as well as the associated task identifier, to the seventh network element (e.g., AI resource management function); thereafter, the seventh network element (e.g., AI resource management function) forwards the updated model parameter information and model identifier, as well as the associated task identifier, to the sixth network element (e.g., AI model management function). The sixth network element (e.g., AI model management function) performs model parameter update or model update based on the updated model parameter information and model identifier, and may also include model version update.

[0483] Step 32. The seventh network element (eg, AI resource management function) sends a task completion notification to the third network element (eg, AI service management function); the third network element (eg, AI service management function) feeds back ACK to the seventh network element (eg, AI resource management function).

[0484] Step 33. When all tasks are completed, the seventh network element (eg, the AI ​​resource management function) sends a resource release indication to the fourth network element (eg, the AI ​​resource node) to release the AI ​​resources.

[0485] Example 2

[0486] In this embodiment 2, the AI ​​service management function serves as the central control node. Due to the need to establish an AI service, the OTT server requests the AI ​​service management function to establish a multi-task AI service; the AI ​​service management function receives the AI ​​service establishment request message and negotiates with the policy charging function to obtain the AI ​​policy. Fig. 7A and Figure 7B As shown in the figure, the specific multi-task AI business service process includes:

[0487] Step 1. The second network element (eg, an OTT server) initiates an AI service establishment request message to the fifth network element (eg, a network open function).

[0488] Step 2. Optionally, a fifth network element (eg, a network open function) authenticates the request message to determine whether the OTT server can request such a service.

[0489] Step 3. The fifth network element (eg, network opening function) selects the first network element (eg, AI service management function) according to the received request message.

[0490] Optionally, the fifth network element (eg, a network open function) may comprehensively consider some factors and use NRF or local configuration information to discover a matching first network element (eg, an AI service management function) instance, and the factors considered may include at least one of the following:

[0491] -Area of ​​interest;

[0492] -UE AI task-related contract information;

[0493] -Data network name DNN;

[0494] - Slice information S-NSSAI;

[0495] -AI business logo;

[0496] -Supported AI services;

[0497] - First network element (e.g. AI service management function) instance capabilities, such as AI service processing latency, AI service processing accuracy, AI service computing capacity, etc.

[0498] Step 4. The fifth network element (e.g., network open function) sends an AI service establishment request message to the first network element (e.g., AI service management function) to request the allocation of corresponding resources and models for the AI ​​service, including AI policies. The AI ​​policies include multi-task AI policies. The request message is specifically the fourth message mentioned above, and the information included therein may be the content included in the fourth message mentioned above. To avoid repetition, it will not be repeated here.

[0499] Optionally, the AI ​​service establishment request message includes at least one of the following: OTT server identifier, DNN, S-NSSAI, AI service identifier, model-related parameters, AI service management function address, AI service description information, etc.

[0500] Step 5. The first network element (eg, AI service management function) and the policy charging function negotiate a multi-task AI policy.

[0501] Optionally, the first network element (eg, an AI service management function) and a policy charging function may generate a corresponding AI policy based on information carried in an AI service establishment request message.

[0502] Optionally, the first network element (eg, AI service management function) and the policy charging function should comprehensively consider the end-to-end service requirements of the AI ​​service when allocating AI policies.

[0503] Step 6a. The third network element (eg, AI service management function) sends a model request message to the sixth network element (eg, AI model management function) to determine the model information.

[0504] Step 6b. The sixth network element (eg, the AI ​​model management function) sends a model request response to the third network element (eg, the AI ​​service management function) to indicate the model information.

[0505] Step 7a. Optionally, a third network element (eg, an AI service management function) sends a data source request message to the data plane management function to determine the data source information.

[0506] Step 7b. The data plane management function sends a data source request response message to the third network element (eg, the AI ​​service management function) to indicate the data source information.

[0507] Step 8a. The third network element (eg, the AI ​​service management function) sends an AI resource request message to the seventh network element (eg, the AI ​​resource management function), where the AI ​​resource request message is specifically the sixth message described above.

[0508] Step 9. The seventh network element (eg, the AI ​​resource management function) determines a suitable fourth network element (eg, an AI resource node) based on the AI ​​resource request information and the AI ​​policy, and sends the AI ​​resource configuration information to the fourth network element (eg, the AI ​​resource node).

[0509] Step 8b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes the AI ​​resource address, type, size, etc.

[0510] It should be noted that steps 6a to 9 in this embodiment 2 are similar to steps 7a to 10 in the above embodiment 1. For details, please refer to the above embodiment 1 and will not be repeated here.

[0511] Step 10. The first network element (e.g., AI service management function) sends a response message corresponding to the request message in step 1 to the OTT server, which is used to notify the result of the AI ​​service establishment request, including AI resource node information and may also include a task identifier; or, the first network element (e.g., AI service management function) sends a task 1 resource establishment notification message to the OTT server to trigger data or model interaction related to task 1. Task 1 is, for example, the currently executed task.

[0512] Steps 11-16 are similar to steps 13-18 in the above-mentioned embodiment 1 and will not be described in detail here.

[0513] Optionally, if the above step 6a adopts implementation method 1, that is, only the model required for this task is requested, and the model required to be used in the next task is different from the model used or generated in the previous task, then Figure 7B , continue with the following steps:

[0514] Steps 17a-20 are similar to steps 19a-22 in the above-mentioned embodiment 1 and will not be described in detail here.

[0515] Step 21. The first network element (e.g., AI service management function) sends a response message corresponding to the request message in step 1 to the OTT server, which is used to notify the result of the AI ​​service request, including the fourth network element (e.g., AI resource node) information and the task identifier; or, the first network element (e.g., AI service management function) sends a task resource establishment notification message to the OTT server, including the fourth network element (e.g., AI resource node) information and the task identifier, which is used to trigger task-related data or model interaction. One way is to forward through the network open function.

[0516] Steps 22-30 are similar to steps 23-33 in the above-mentioned embodiment 1 and will not be described in detail here.

[0517] Example 3

[0518] In this embodiment 3, the training and deployment of the corresponding model. The AI ​​service management function serves as the central control node (training + deployment). Due to the need to establish an AI service, the OTT server requests the AI ​​service management function to establish a multi-task AI service; the AI ​​service management function receives the AI ​​service establishment request message and negotiates with the policy billing function to obtain a multi-task AI policy. Fig. 7A and Figure 8 As shown in the figure, the specific multi-task AI business service process includes:

[0519] Step 1-16. Fig. 7A As shown, it is similar to steps 1-16 in the above embodiment 2 and will not be repeated here.

[0520] Optionally, if step 6a adopts implementation method 1, that is, only the model required for this task is requested, and the model required for the next task is different from the model used or generated by the previous task, then Figure 8 , continue with the following steps:

[0521] Steps 17a-29 are similar to steps 17-29 in the above-mentioned embodiment 2 and will not be described again here.

[0522] Step 30a. The third network element (eg, AI service management function) sends AI resource request information to the seventh network element (eg, AI resource management function), and the seventh network element (eg, AI resource management function) determines a suitable AI resource node based on the AI ​​resource request information and the AI ​​policy.

[0523] Optionally, the method of carrying content in the AI ​​resource request information can be implemented in a manner based on business requirements. The corresponding content can refer to the content included in the sixth message above. To avoid repetition, it will not be repeated here.

[0524] Step 31. The seventh network element (e.g., AI resource management function) determines a suitable fourth network element (e.g., AI resource node) based on the AI ​​resource request information and the AI ​​strategy, and sends AI resource configuration update information to the fourth network element (e.g., AI resource node). The content carried by the information can be found in the above embodiment.

[0525] Step 30b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes the AI ​​resource address, type, size, etc.

[0526] Step 32. The first network element (e.g., AI service management function) sends an AI service establishment request response message corresponding to the request message in step 1 to the second network element (e.g., OTT server) to notify the AI ​​unit that the training and deployment have been completed. One way is to forward it through the fifth network element (e.g., network open function).

[0527] Optionally, the AI ​​service establishment request response includes at least one of the following:

[0528] Model identification;

[0529] Type of service;

[0530] Model deployment address;

[0531] Verify accuracy.

[0532] Example 4

[0533] In this embodiment 4, the AI ​​service management function serves as the central control node. Due to the need to establish an AI service, the OTT server requests the AI ​​service management function to establish a multi-task AI service; the AI ​​service management function receives the AI ​​service establishment request message and negotiates with the policy billing function to obtain the AI ​​policy. If the multiple tasks involved include: Task 1, instruction learning; Task 2, scoring model training (supervised learning); Task 3, proxy strategy optimization (reinforcement learning); Task 4, model deployment; then Fig.9A , Fig. 9B and Fig. 9C As shown in the figure, the specific multi-task AI business service process includes:

[0534] Step 1. The OTT server (i.e., the second network element) initiates an AI service establishment request message to the fifth network element (e.g., the network open function).

[0535] Step 2. Optionally, a fifth network element (eg, a network open function) authenticates the request message to determine whether the OTT server can request such a service.

[0536] Step 3. The fifth network element (eg, network opening function) selects the first network element (eg, AI service management function) according to the received request message.

[0537] Step 4. The fifth network element (e.g., network open function) sends an AI service establishment request message to the first network element (e.g., AI service management function) to request the allocation of corresponding resources and models for the AI ​​service, including AI policies. The AI ​​policies include multi-task AI policies. The request message is used to request the network to establish a corresponding AI service for it; the request message is specifically the fourth message mentioned above, and the information included therein may be the content included in the fourth message mentioned above, and to avoid repetition, it will not be repeated here.

[0538] Optionally, the AI ​​service establishment request message includes at least one of the following: OTT server identifier, DNN, S-NSSAI, AI service identifier, model-related parameters, AI service management function address, AI service description information, etc.

[0539] Step 5. The first network element (eg, AI service management function) and the policy charging function negotiate a multi-task AI policy.

[0540] Step 6a. The third network element (e.g., AI service management function) sends a model request message to the sixth network element (e.g., AI model management function) to determine the model information. In this step, the model request message adopts implementation mode 2, that is, requesting the models required for all tasks, and the content included can be referred to the above embodiment.

[0541] Specifically, the model request message includes: the GPT initial large model for task 1 instruction learning, and the initial scoring model for task 2 scoring model training.

[0542] Step 6b. The sixth network element (e.g., AI model management function) sends a model request response to the third network element (e.g., AI service management function) to indicate model information. This step indicates that the initial training model to be used in Task 1 is the GPT initial large model.

[0543] Step 7a. Optionally, a third network element (eg, an AI service management function) sends a data source request message to the data plane management function to determine the data source information.

[0544] Step 7b. The data plane management function sends a data source request response message to the third network element (eg, the AI ​​service management function) to indicate the data source information.

[0545] Step 8a. The third network element (eg, the AI ​​service management function) sends an AI resource request message to the seventh network element (eg, the AI ​​resource management function), where the AI ​​resource request message is specifically the sixth message described above.

[0546] Step 9. The seventh network element (eg, the AI ​​resource management function) determines a suitable fourth network element (eg, an AI resource node) based on the AI ​​resource request information and the AI ​​policy, and sends the AI ​​resource configuration information to the fourth network element (eg, the AI ​​resource node).

[0547] Step 8b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes information about the fourth network element (eg, AI resource node).

[0548] Step 10. The first network element (e.g., AI service management function) sends a task 1 resource establishment notification message to the second network element (e.g., OTT server) to trigger data or model interaction related to task 1.

[0549] It should be noted that step 6b-10 in this embodiment 3 is similar to step 6b-10 in the above embodiment 2. For specific contents, please refer to the above embodiment 2 and will not be repeated here.

[0550] Step 11. The OTT server sends the first data to the fourth network element (e.g., AI resource node), and one way is to forward it through the fifth network element (e.g., network open function). The first data is a training data set for instruction learning, including input data and label data.

[0551] Step 12. Optionally, a fourth network element (eg, an AI resource node) obtains in-network data from a data source.

[0552] Step 13. The fourth network element (eg, AI resource node) executes instruction learning, which belongs to supervised learning.

[0553] Step 14. After the instruction learning is completed, the fourth network element (e.g., AI resource node) sends the updated model parameter information and model identifier, and the associated task identifier to the seventh network element (e.g., AI resource management function); thereafter, the seventh network element (e.g., AI resource management function) forwards the updated model parameter information and model identifier, and the associated task identifier to the sixth network element (e.g., AI model management function). The sixth network element (e.g., AI model management function) performs model parameter update or model update according to the updated model parameter information and model identifier, and may also include model version update. Step 15. The seventh network element (e.g., AI resource management function) sends a task end notification to the third network element (e.g., AI service management function); the third network element (e.g., AI service management function) feeds back ACK to the seventh network element (e.g., AI resource management function).

[0554] After the task 1 instruction learning is completed, Fig. 9B As shown, continue with the following steps:

[0555] Step 16a. The third network element (eg, AI service management function) sends an AI resource request (modification) message to the seventh network element (eg, AI resource management function). The request message is specifically the sixth message mentioned above, and reference may be made to the above content.

[0556] Step 17. The seventh network element (e.g., AI resource management function) determines a suitable fourth network element (e.g., AI resource node) based on the AI ​​resource request information and AI strategy, and sends AI resource configuration update information to the fourth network element (e.g., AI resource node). This step indicates that the inference model to be used in Task 2 is the instruction learning model trained in Task 1, and the initial model used for scoring model training comes from Step 6b.

[0557] Step 16b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes the AI ​​resource address, resource, size, etc.

[0558] Step 18. The first network element (e.g., AI service management function) sends a current task resource establishment notification message to the OTT server to trigger data or model interaction related to Task 2.

[0559] Step 19. The OTT server sends the first data (i.e., AI service data) to the fourth network element (e.g., AI resource node), one way of forwarding it through the fifth network element (e.g., network open function). The first data is one or more model inputs, i.e., input data for scoring model training.

[0560] Step 20. Optionally, a fourth network element (eg, an AI resource node) obtains in-network data from a data source.

[0561] Step 21. The fourth network element (e.g., an AI resource node) uses the first data as a model input, performs multiple AI model inferences using the model learned by the task 1 instruction, and generates multiple outputs.

[0562] Step 22. The fourth network element (e.g., AI resource node) sends the second data to the OTT server, one way of forwarding it through the fifth network element (e.g., network open function). Optionally, the second data may include at least one of the following:

[0563] Multiple outputs for one input, which corresponds to one training sample;

[0564] Multiple outputs for multiple inputs, where each input corresponds to multiple outputs, which corresponds to multiple training samples.

[0565] Step 23. After the OTT service sorts the different outputs in each training sample of the second data, the obtained third data is sent to the fourth network element (e.g., an AI resource node). One way is to forward it through the fifth network element (e.g., a network open function).

[0566] Optionally, the third data includes at least one of the following:

[0567] The sorting results of multiple outputs for one input correspond to one training sample;

[0568] Sorting results for multiple outputs, where each input corresponds to a set of sorted results for multiple outputs, which corresponds to multiple training samples.

[0569] Step 24. The fourth network element (such as an AI resource node) performs model training; specifically, the input and multiple outputs are used as inputs of the scoring model, and the sorting results of the multiple outputs are used as labels of the scoring model to perform supervised learning and update the parameters of the scoring model.

[0570] After that, steps 19-24 are iterated in a loop until the scoring model training converges.

[0571] Step 25. The fourth network element (e.g., AI resource node) sends updated model parameter information and model identifier, as well as the associated task identifier, to the seventh network element (e.g., AI resource management function); thereafter, the seventh network element (e.g., AI resource management function) forwards the updated model parameter information and model identifier, as well as the associated task identifier, to the sixth network element (e.g., AI model management function). The sixth network element (e.g., AI model management function) performs model parameter update or model update according to the updated model parameter information and model identifier, and may also include model version update.

[0572] Step 26. The seventh network element (eg, AI resource management function) sends a task completion notification to the third network element (eg, AI service management function); the third network element (eg, AI service management function) feeds back ACK to the seventh network element (eg, AI resource management function).

[0573] After the scoring model training of Task 2 is completed, Fig. 9C As shown, continue with the following steps:

[0574] Step 27a. The third network element (eg, AI service management function) sends an AI resource modification request message to the seventh network element (eg, AI resource management function). The request message is specifically the sixth message mentioned above, and reference may be made to the above content.

[0575] Step 28. The seventh network element (e.g., AI resource management function) determines a suitable fourth network element (e.g., AI resource node) based on the AI ​​resource request information and the AI ​​strategy, and sends AI resource configuration update information to the fourth network element (e.g., AI resource node). Among them, this step indicates that Task 3 needs to use the scoring model trained by Task 2 as feedback for the environment, and uses the model learned by Task 1 as the initial strategy generation model and the reference model for comparing distributions.

[0576] Step 27b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes the AI ​​resource address, resource, size, etc.

[0577] Step 29. The first network element (e.g., AI service management function) sends a current task resource establishment notification message to the OTT server to trigger data or model interaction related to task 3.

[0578] Step 30. The OTT server sends the first data (i.e., AI service data) to the fourth network element (e.g., AI resource node), one way of forwarding it through the fifth network element (e.g., network open function). The first data is input data of multiple samples, i.e., state information (model input) for reinforcement learning.

[0579] Step 31. Optionally, a fourth network element (eg, an AI resource node) obtains in-network data from a data source.

[0580] Step 32. The fourth network element (e.g., AI resource node) performs model training; specifically, using the first data as the state (model input for reinforcement learning), using the scoring model trained by Task 2 as feedback from the environment, using the model learned by Task 1 instructions as the initial policy generation model and a reference model for comparing distributions, performing reinforcement learning training, and updating the parameters of the policy generation model.

[0581] Optionally, if the input data is sent to the fourth network element (eg, an AI resource node) in multiple times, steps 30-32 are executed in an iterative loop.

[0582] Step 33. The fourth network element (e.g., AI resource node) sends updated model parameter information and model identifier, as well as the associated task identifier, to the seventh network element (e.g., AI resource management function); thereafter, the seventh network element (e.g., AI resource management function) forwards the updated model parameter information and model identifier, as well as the associated task identifier, to the sixth network element (e.g., AI model management function). The sixth network element (e.g., AI model management function) performs model parameter update or model update according to the updated model parameter information and model identifier, and may also include model version update.

[0583] Step 34. The seventh network element (eg, AI resource management function) sends a task completion notification to the third network element (eg, AI service management function); the third network element (eg, AI service management function) feeds back ACK to the seventh network element (eg, AI resource management function).

[0584] After completing Task 3 reinforcement learning, continue with the following steps:

[0585] Step 35a. The third network element (eg, AI service management function) sends AI resource request information to the seventh network element (eg, AI resource management function), and the AI ​​resource management function determines a suitable fourth network element (eg, AI resource node) based on the AI ​​resource request information and the AI ​​policy.

[0586] Step 36. The seventh network element (e.g., AI resource management function) determines a suitable fourth network element (e.g., AI resource node) according to the AI ​​resource request information and the AI ​​strategy, and sends AI resource configuration update information to the AI ​​resource node. The content carried can be referred to in the above embodiment.

[0587] Step 35b. The seventh network element (eg, AI resource management function) sends an AI resource request response message to the third network element (eg, AI service management function), which includes the AI ​​resource address, type, size, etc.

[0588] Step 37. The first network element (e.g., AI service management function) sends an AI service establishment request response message corresponding to the request message in step 1 to the OTT server, which is used to notify the AI ​​unit that the training and deployment have been completed, which may include model identification, service type, model deployment address, verification accuracy, etc. One way is to forward it through the fifth network element (e.g., network open function).

[0589] Based on the above embodiments, it can be seen that through the solution in this application, the operator network can not only provide AI business processing services for the OTT server and expand the operator's business scenarios, but also ensure the security of OTT server data flow.

[0590] The service processing method provided in the embodiment of the present application can be executed by a service processing device. In the embodiment of the present application, the service processing device provided in the embodiment of the present application is described by taking the service processing method executed by the service processing device as an example.

[0591] See also Fig.10 , Fig.10 is a structural diagram of a service processing device provided in an embodiment of the present application, the device is applied to a first network element, such as Fig.10 As shown, the service processing device 100 includes:

[0592] An acquisition module 101 is configured to acquire an establishment request from a second network element to establish an AI service, where the AI ​​service includes at least one task;

[0593] A determination module 102, configured to determine an AI strategy for at least one task in the AI ​​service according to the establishment request;

[0594] The first sending module 103 is used to return a response message of the establishment request to the second network element according to the AI ​​strategy.

[0595] Optionally, the first network element is a policy charging function, and the acquisition module 101 is specifically used to:

[0596] A first message sent by a second network element is received, where the first message is used to request to establish an AI service for the second network element, and the first message includes at least one of the following:

[0597] The number of tasks for the AI ​​business;

[0598] The timing relationship between multiple tasks of the AI ​​business;

[0599] The dependencies between the data of multiple tasks of the AI ​​business;

[0600] Dependencies between AI units of multiple tasks of the AI ​​service;

[0601] Task identifiers of multiple tasks of the AI ​​service;

[0602] Task requirements of multiple tasks of the AI ​​business;

[0603] Demand information for the AI ​​service;

[0604] Quality of experience (QoE) requirements for the AI ​​service.

[0605] Optionally, when the first network element is a policy charging function, the first sending module 103 is specifically used to:

[0606] Sending a second message to the second network element; the second message is used to respond to the first message, or the second message is used to trigger data transmission or AI unit transmission related to a first task of the AI ​​service, where the first task is a task being executed by the AI ​​service;

[0607] The second message includes: relevant information of the AI ​​resource node used to execute the first task.

[0608] Optionally, the first network element is a policy charging function, and the service processing device 100 further includes:

[0609] The second sending module is used to send a third message to a third network element, where the third message is used to request allocation of AI resources or required AI units for the AI ​​service.

[0610] Optionally, the first network element is an AI service management function, and the acquisition module 101 is specifically used to:

[0611] receiving a fourth message, where the fourth message is used to request to establish an AI service for the second network element, and the fourth message includes at least one of the following:

[0612] an identifier of the second network element;

[0613] AI business logo;

[0614] Relevant parameters of the AI ​​unit used for the AI ​​service;

[0615] an address of the first network element;

[0616] Demand information for AI services;

[0617] The task requirement of at least one task in the AI ​​business;

[0618] The number of tasks for the AI ​​business;

[0619] The timing relationship between multiple tasks of the AI ​​business;

[0620] The dependencies between the data of multiple tasks of the AI ​​business;

[0621] Dependencies between AI units of multiple tasks of the AI ​​service;

[0622] Task identifiers of multiple tasks of the AI ​​service;

[0623] Demand information for the AI ​​service;

[0624] The QoE requirements of the AI ​​service.

[0625] Optionally, the task requirements include at least one of the following:

[0626] Single task computing amount;

[0627] Data processing scale;

[0628] Data storage requirements;

[0629] AI unit application scope;

[0630] Data processing delay;

[0631] Data processing accuracy;

[0632] Expected AI unit size.

[0633] The service processing device 100 provided in the embodiment of the present application can realize Figure 2 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0634] See also Fig.11 , Fig.11 is a structural diagram of a service processing device provided in an embodiment of the present application, the device is applied to a second network element, such as Fig.11 As shown, the service processing device 110 includes:

[0635] A third sending module 111 is configured to send a request to establish an AI service to the first network element, where the AI ​​service includes at least one task;

[0636] The first receiving module 112 is used to receive a response message of the establishment request returned by the first network element; the response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

[0637] Optionally, the first network element is a policy charging function, and the third sending module 111 is specifically used for:

[0638] The second network element sends a first message to the first network element, wherein the first message is used to request to establish an AI service for the second network element, and the first message includes at least one of the following:

[0639] The number of tasks for the AI ​​business;

[0640] The timing relationship between multiple tasks of the AI ​​business;

[0641] The dependencies between the data of multiple tasks of the AI ​​business;

[0642] Dependencies between AI units of multiple tasks of the AI ​​service;

[0643] Task identifiers of multiple tasks of the AI ​​service;

[0644] Task requirements of multiple tasks of the AI ​​business;

[0645] Demand information for the AI ​​service;

[0646] The QoE requirements of the AI ​​service.

[0647] Optionally, the first receiving module 112 is specifically configured to:

[0648] Receive a second message sent by the first network element; the second message is used to respond to the first message, or the second message is used to trigger data transmission or AI unit transmission related to a first task of the AI ​​service, and the first task is a task being executed by the AI ​​service; the second message includes: relevant information of an AI resource node used to execute the first task.

[0649] Optionally, the first network element is an AI service management function, and the third sending module 111 is specifically used to:

[0650] A fourth message is sent to the first network element through the fifth network element, where the fourth message is used to request to establish an AI service for the second network element, the first network element is selected by the fifth network element, and the fourth message includes at least one of the following:

[0651] an identifier of the second network element;

[0652] AI business logo;

[0653] AI unit parameters;

[0654] an address of the first network element;

[0655] Demand information for AI services;

[0656] The task requirement of at least one task in the AI ​​business;

[0657] The number of tasks for the AI ​​business;

[0658] The timing relationship between multiple tasks of the AI ​​business;

[0659] The dependencies between the data of multiple tasks of the AI ​​business;

[0660] Dependencies between AI units of multiple tasks of the AI ​​service;

[0661] Task identifiers of multiple tasks of the AI ​​service;

[0662] Demand information for the AI ​​service;

[0663] The QoE requirements of the AI ​​service.

[0664] Optionally, the task requirements include at least one of the following:

[0665] Single task computing amount;

[0666] Data processing scale;

[0667] Data storage requirements;

[0668] AI unit application scope;

[0669] Data processing delay;

[0670] Data processing accuracy;

[0671] Expected AI unit size.

[0672] Optionally, the service processing device 110 further includes:

[0673] The fourth sending module is configured to send first data to a fourth network element, where the first data includes at least one of the following:

[0674] Input data of the AI ​​unit for the AI ​​service;

[0675] Label data for AI unit processing operations;

[0676] sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service;

[0677] The structure of the AI ​​unit used for the AI ​​service;

[0678] Parameters of the AI ​​unit used for the AI ​​service;

[0679] Description information of the AI ​​unit used for the AI ​​service;

[0680] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0681] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0682] Optionally, the fourth sending module is specifically used to: send the first data to the fourth network element through a fifth network element.

[0683] Optionally, the service processing device 110 further includes:

[0684] The second receiving module is configured to receive second data sent by the fourth network element, where the second data includes at least one of the following:

[0685] Input data of the AI ​​unit for the AI ​​service;

[0686] Label data for AI unit processing operations;

[0687] Output data of the AI ​​unit for the AI ​​service;

[0688] Input data of an AI unit for the AI ​​service generated based on reinforcement learning;

[0689] The structure of the AI ​​unit used for the AI ​​service;

[0690] Parameters of the AI ​​unit used for the AI ​​service;

[0691] Description information of the AI ​​unit used for the AI ​​service;

[0692] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0693] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0694] The service processing device 110 provided in the embodiment of the present application can realize Figure 3 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0695] See also Fig.12 , Fig.12 is a schematic diagram of the structure of a service processing device provided in an embodiment of the present application, the device is applied to a third network element, such as Fig.12 As shown, the service processing device 120 includes:

[0696] A fifth sending module 121, configured to send a fifth message to a sixth network element, wherein the fifth message is used to request an AI unit required for at least one task of the AI ​​service;

[0697] The third receiving module 122 is used to receive a response message to the fifth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

[0698] Optionally, when the fifth message is used to request AI units required for multiple tasks of the AI ​​service, the fifth message includes at least one of the following:

[0699] a first indication, where the first indication is used to indicate a task among the multiple tasks that requires obtaining an initial AI unit from a sixth network element;

[0700] a second indication, the second indication being used to indicate dependencies between the AI ​​units required for the plurality of tasks;

[0701] Use case identifiers corresponding to each of the multiple tasks;

[0702] The service type corresponding to each of the multiple tasks;

[0703] business identifiers of the plurality of tasks;

[0704] The expected accuracy of the multiple tasks;

[0705] Data storage requirements for the multiple tasks;

[0706] The input data formats of the plurality of tasks;

[0707] Output data formats of the plurality of tasks;

[0708] Generalization metrics for the multiple tasks described;

[0709] Alternatively, when the fifth message is used to request the AI ​​unit required by the second task of the AI ​​service, and the second task is a single task among the multiple tasks, the fifth message includes at least one of the following:

[0710] a task identifier of the second task;

[0711] A service identifier corresponding to the second task;

[0712] A use case identifier corresponding to the second task;

[0713] The service type corresponding to the second task;

[0714] the expected accuracy of the second task;

[0715] data storage requirements of the second task;

[0716] the data input format of the second task;

[0717] The data output format of the second task;

[0718] A generalization indicator of the second task;

[0719] an identifier of a third task, the third task being a task before the second task and associated with the second task;

[0720] Alternatively, when the fifth message is used to request the AI ​​unit required by the second task of the AI ​​service, and the second task is a single task among the multiple tasks, the fifth message includes at least one of the following:

[0721] Task identification;

[0722] Business logo;

[0723] Use case identification;

[0724] Type of service;

[0725] Expected accuracy;

[0726] Data storage requirements;

[0727] Input data format;

[0728] Output data format;

[0729] Generalizability index;

[0730] An identifier of a fourth task, where the fourth task is a task that precedes the second task and is associated with the second task.

[0731] Optionally, the service processing device 120 includes:

[0732] The sixth sending module is used to send a sixth message to the seventh network element, where the sixth message is used to request AI resources for at least one task of the AI ​​service.

[0733] Optionally, when the sixth message is used to request AI resources for multiple tasks of the AI ​​service, the sixth message includes at least one of the following:

[0734] AI unit information corresponding to each of the multiple tasks;

[0735] Data source information of the multiple tasks;

[0736] the computational effort of the plurality of tasks;

[0737] the maximum data storage requirements of the plurality of tasks;

[0738] the processing latency requirements of the multiple tasks;

[0739] The input data formats of the plurality of tasks;

[0740] The number of single-shot inferences for each of the multiple tasks.

[0741] Optionally, when the sixth message is used to request AI resources for the fifth task of the AI ​​service, the sixth message includes at least one of the following:

[0742] The task identifier of the fifth task;

[0743] AI unit information corresponding to the fifth task;

[0744] Data source information of the fifth task;

[0745] The computational amount of the fifth task;

[0746] The data storage requirements of the fifth task;

[0747] The processing delay requirement of the fifth task;

[0748] The input data format of the fifth task;

[0749] The number of single-sample inferences for the fifth task.

[0750] Optionally, when the sixth message is used to request AI resources for the fifth task of the AI ​​service, the sixth message includes at least one of the following:

[0751] Task identification;

[0752] AI unit information;

[0753] Data source information;

[0754] Calculation amount;

[0755] Data storage requirements;

[0756] Handling latency requirements;

[0757] Input data format;

[0758] Number of single-sample inferences.

[0759] Optionally, the service processing device 120 further includes:

[0760] The first execution module is configured to execute at least one of the following when completing the kth task of the AI ​​service:

[0761] triggering the AI ​​resource of the next task of the k-th task;

[0762] For the next task of the k-th task, obtain a new initial AI unit;

[0763] Among them, the kth task is not the last task of the AI ​​business.

[0764] The service processing device 120 provided in the embodiment of the present application can realize Figure 4 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0765] See also Fig.13 , Fig.13 is a structural diagram of a service processing device provided in an embodiment of the present application, the device is applied to a fourth network element, such as Fig.13 As shown, the service processing device 130 includes:

[0766] The fourth receiving module 131 is used to receive the first data sent by the second network element;

[0767] A processing module 132, configured to perform an AI unit processing operation according to the first data;

[0768] The seventh sending module 133 is used to send second data to the second network element.

[0769] Optionally, the first data includes at least one of the following:

[0770] Input data of the AI ​​unit for the AI ​​service;

[0771] Label data for AI unit processing operations;

[0772] sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service;

[0773] The structure of the AI ​​unit used for the AI ​​service;

[0774] Parameters of the AI ​​unit used for the AI ​​service;

[0775] Description information of the AI ​​unit used for the AI ​​service;

[0776] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0777] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0778] Alternatively, the second data includes at least one of the following:

[0779] Input data of the AI ​​unit for the AI ​​service;

[0780] Label data for AI unit processing operations;

[0781] Output data of the AI ​​unit for the AI ​​service;

[0782] Input data of an AI unit for the AI ​​service generated based on reinforcement learning;

[0783] The structure of the AI ​​unit used for the AI ​​service;

[0784] Parameters of the AI ​​unit used for the AI ​​service;

[0785] Description information of the AI ​​unit used for the AI ​​service;

[0786] Output of the segmentation layer in the AI ​​unit for the AI ​​service;

[0787] The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

[0788] Optionally, the AI ​​unit processing operation includes at least one of the following:

[0789] AI unit aggregation;

[0790] Single AI unit training;

[0791] Multiple AI unit training;

[0792] Single AI unit reasoning;

[0793] Multiple AI unit reasoning.

[0794] Optionally, the service processing device 130 further includes:

[0795] The second execution module is configured to execute at least one of the following when completing the kth task of the AI ​​service:

[0796] Do not release AI resources;

[0797] Obtaining AI resources for the next task of the k-th task;

[0798] For the next task of the k-th task, obtain a new initial AI unit, or use the AI ​​unit obtained after performing the k-th task;

[0799] Among them, the kth task is not the last task of the AI ​​business.

[0800] Optionally, the service processing device 130 further includes:

[0801] An eighth sending module is configured to send first information to the sixth network element, where the first information includes at least one of the following:

[0802] Updated AI unit parameters;

[0803] Updated AI unit logo;

[0804] The task ID associated with the updated AI unit.

[0805] Optionally, the service processing device 130 further includes:

[0806] A fifth receiving module is configured to receive configuration information sent by the seventh network element, where the configuration information is used to configure AI resources, and the configuration information includes at least one of the following:

[0807] A third indication, wherein the third indication is used to indicate an AI framework;

[0808] A fourth indication, wherein the fourth indication is used to indicate an AI library version;

[0809] A fifth indication, wherein the fifth indication is used to indicate the number of single sample inferences;

[0810] Data source information;

[0811] AI business detection information;

[0812] AI business processing rules.

[0813] The service processing device 130 provided in the embodiment of the present application can realize Figure 5 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0814] Optionally, an embodiment of the present application further provides a communication system, including: a first network element, a second network element, a third network element and a fourth network element, wherein:

[0815] The first network element is used to: obtain an establishment request from the second network element to establish an AI service, where the AI ​​service includes at least one task; determine an AI strategy for at least one task in the AI ​​service according to the establishment request; and return a response message of the establishment request to the second network element according to the AI ​​strategy;

[0816] The second network element is used to: send an establishment request to the first network element for requesting to establish an AI service; and receive a response message to the establishment request returned by the first network element;

[0817] The third network element is used to: send a sixth message to the sixth network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service; receive a response message to the sixth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task;

[0818] The fourth network element is used to: receive the first data sent by the second network element; perform the AI ​​unit processing operation according to the first data; and send the second data to the second network element;

[0819] Among them, the first network element, the third network element and the fourth network element are core network elements, and the second network element is a third-party server.

[0820] Optionally, the communication system further includes a fifth network element, wherein:

[0821] The fourth network element is further configured to: receive first data sent by the second network element through the fifth network element, and send second data to the second network element through the fifth network element.

[0822] Optionally, the communication system further includes a sixth network element, wherein:

[0823] The sixth network element is used to: receive a sixth message sent by the third network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service; and send a response message to the sixth message to the third network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

[0824] Optionally, the communication system further includes a seventh network element, wherein:

[0825] The third network element is further used to: send a sixth message to the seventh network element, where the sixth message is used to request AI resources for at least one task of the AI ​​service;

[0826] The fourth network element is further used to: receive configuration information sent by the seventh network element, where the configuration information is used to configure AI resources.

[0827] like Fig.14As shown, an embodiment of the present application also provides a communication device 140, including a processor 141 and a memory 142, and the memory 142 stores programs or instructions that can be executed on the processor 141. When the program or instruction is executed by the processor 141, the various steps of the above-mentioned business processing method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0828] The embodiment of the present application also provides a communication device, including a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run a program or instruction to implement the steps of the above-mentioned service processing method embodiment. The device embodiment corresponds to the above-mentioned method embodiments, and each implementation process and implementation method of the above-mentioned method embodiments can be applied to the communication device embodiment and can achieve the same technical effect.

[0829] Specifically, the embodiment of the present application also provides a network side device. Fig.15 As shown, the network side device 150 includes: a processor 151, a network interface 152 and a memory 153. The network interface 152 is, for example, a common public radio interface (CPRI).

[0830] Specifically, the network side device 150 of the embodiment of the present application also includes: instructions or programs stored in the memory 153 and executable on the processor 151. The processor 151 calls the instructions or programs in the memory 153 to execute the above-mentioned business processing method and achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0831] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, each process of the above-mentioned business processing method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0832] The processor is the processor in the terminal described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk. In some examples, the readable storage medium may be a non-transient readable storage medium.

[0833] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned business processing method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0834] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0835] The embodiments of the present application further provide a computer program / program product, which is stored in a storage medium and is executed by at least one processor to implement the various processes of the above-mentioned business processing method embodiment and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0836] The embodiment of the present application also provides a communication system, including: a first network element, a second network element, a third network element and a fourth network element, wherein the first network element can be used to perform the following steps: Figure 2 The steps of the method described above, the second network element can be used to perform Figure 3 The steps of the method described above, the third network element can be used to perform Figure 4 The steps of the method described above, the fourth network element can be used to perform Figure 5 The steps of the method are described.

[0837] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0838] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of a computer software product plus a necessary general hardware platform, and of course, can also be implemented by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, disk, CD, etc.), including several instructions to enable a terminal or a network-side device to execute the methods described in each embodiment of the present application.

[0839] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms of implementation methods without departing from the purpose of the present application and the scope of protection of the claims, and these implementation methods are all within the protection of the present application.

Claims

1. A business processing method, characterized in that: include: The first network element obtains a request from the second network element to establish an artificial intelligence AI service, where the AI ​​service includes at least one task; The first network element determines, according to the establishment request, an AI strategy for at least one task in the AI ​​service; The first network element returns a response message of the establishment request to the second network element according to the AI ​​policy; The first network element is a core network element, and the second network element is a third-party server.

2. The method according to claim 1, characterized in that The first network element is a policy charging function, and the first network element obtains a request from the second network element to establish an artificial intelligence AI service, including: The first network element receives a first message sent by the second network element, wherein the first message is used to request to establish an AI service for the second network element, and the first message includes at least one of the following: The number of tasks for the AI ​​business; The timing relationship between multiple tasks of the AI ​​business; The dependencies between the data of multiple tasks of the AI ​​business; Dependencies between AI units of multiple tasks of the AI ​​service; Task identifiers of multiple tasks of the AI ​​service; Task requirements of multiple tasks of the AI ​​business; Demand information for the AI ​​service; Quality of experience (QoE) requirements for the AI ​​service.

3. The method according to claim 1 or 2, characterized in that: The returning a response message of the establishment request to the second network element includes: The first network element sends a second message to the second network element; The second message is used to respond to the first message, or the second message is used to trigger data transmission or AI unit transmission related to a first task of the AI ​​service, and the first task is a task being executed by the AI ​​service; The second message includes: relevant information of the AI ​​resource node used to execute the first task.

4. The method according to any one of claims 1 to 3, characterized in that: The first network element is a policy charging function, and the method further includes: The first network element sends a third message to the third network element, where the third message is used to request allocation of AI resources or required AI units for the AI ​​service.

5. The method according to claim 1, characterized in that The first network element is an AI service management function, and the first network element obtains a request from the second network element to establish an artificial intelligence AI service, including: The first network element receives a fourth message, where the fourth message is used to request to establish an AI service for the second network element, and the fourth message includes at least one of the following: an identifier of the second network element; AI business logo; Relevant parameters of the AI ​​unit used for the AI ​​service; an address of the first network element; Demand information for AI services; The task requirement of at least one task in the AI ​​business; The number of tasks for the AI ​​business; The timing relationship between multiple tasks of the AI ​​business; The dependencies between the data of multiple tasks of the AI ​​business; Dependencies between AI units of multiple tasks of the AI ​​service; Task identifiers of multiple tasks of the AI ​​service; Demand information for the AI ​​service; The QoE requirements of the AI ​​service.

6. The method according to claim 2 or 5, characterized in that: The task requirements include at least one of the following: Single task computing amount; Data processing scale; Data storage requirements; AI unit application scope; Data processing delay; Data processing accuracy; Expected AI unit size.

7. A business processing method, characterized in that: include: The second network element sends a request to the first network element to establish an AI service, where the AI ​​service includes at least one task; The second network element receives a response message to the establishment request returned by the first network element; The response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

8. The method according to claim 7, characterized in that The first network element is a policy charging function, and the second network element sends a request to the first network element to establish an AI service, including: The second network element sends a first message to the first network element, wherein the first message is used to request to establish an AI service for the second network element, and the first message includes at least one of the following: The number of tasks for the AI ​​business; The timing relationship between multiple tasks of the AI ​​business; The dependencies between the data of multiple tasks of the AI ​​business; Dependencies between AI units of multiple tasks of the AI ​​service; Task identifiers of multiple tasks of the AI ​​service; Task requirements of multiple tasks of the AI ​​business; Demand information for the AI ​​service; The QoE requirements of the AI ​​service.

9. The method according to claim 7 or 8, characterized in that: The receiving a response message of the establishment request returned by the first network element includes: The second network element receives a second message sent by the first network element; The second message is used to respond to the first message, or the second message is used to trigger data transmission or AI unit transmission related to a first task of the AI ​​service, and the first task is a task being executed by the AI ​​service; The second message includes: relevant information of the AI ​​resource node used to execute the first task.

10. The method according to claim 7, characterized in that The first network element is an AI service management function, and the second network element sends a request to the first network element to establish an AI service, including: The second network element sends a fourth message to the first network element through the fifth network element, where the fourth message is used to request to establish an AI service for the second network element, the first network element is selected by the fifth network element, and the fourth message includes at least one of the following: an identifier of the second network element; AI business logo; Related parameters of the AI ​​unit used to execute the AI ​​service; an address of the first network element; Demand information for AI services; The task requirement of at least one task in the AI ​​business; The number of tasks for the AI ​​business; The timing relationship between multiple tasks of the AI ​​business; The dependencies between the data of multiple tasks of the AI ​​business; Dependencies between AI units of multiple tasks of the AI ​​service; Task identifiers of multiple tasks of the AI ​​service; Demand information for the AI ​​service; The QoE requirements of the AI ​​service.

11. The method according to claim 8 or 10, characterized in that: The task requirements include at least one of the following: Single task computing amount; Data processing scale; Data storage requirements; AI unit application scope; Data processing delay; Data processing accuracy; Expected AI unit size.

12. The method according to any one of claims 7 to 11, characterized in that The method further comprises: The second network element sends first data to the fourth network element, wherein the first data includes at least one of the following: Input data of the AI ​​unit for the AI ​​service; Label data for AI unit processing operations; sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service; The structure of the AI ​​unit used for the AI ​​service; Parameters of the AI ​​unit used for the AI ​​service; Description information of the AI ​​unit used for the AI ​​service; Output of the segmentation layer in the AI ​​unit for the AI ​​service; The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

13. The method according to claim 12, characterized in that The second network element sending the first data to the fourth network element includes: The second network element sends the first data to the fourth network element through the fifth network element.

14. The method according to claim 12 or 13, characterized in that The method further comprises: The second network element receives second data sent by the fourth network element, wherein the second data includes at least one of the following: Input data of the AI ​​unit for the AI ​​service; Label data for AI unit processing operations; Output data of the AI ​​unit for the AI ​​service; Input data of an AI unit for the AI ​​service generated based on reinforcement learning; The structure of the AI ​​unit used for the AI ​​service; Parameters of the AI ​​unit used for the AI ​​service; Description information of the AI ​​unit used for the AI ​​service; Output of the segmentation layer in the AI ​​unit for the AI ​​service; The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

15. A business processing method, characterized in that: include: The third network element sends a fifth message to the sixth network element, wherein the fifth message is used to request an AI unit required for at least one task of the AI ​​service; The third network element receives a response message to the fifth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

16. The method according to claim 15, characterized in that When the fifth message is used to request the AI ​​units required for multiple tasks of the AI ​​service, the fifth message includes at least one of the following: a first indication, where the first indication is used to indicate a task among the multiple tasks that requires obtaining an initial AI unit from a sixth network element; a second indication, the second indication being used to indicate dependencies between the AI ​​units required for the plurality of tasks; Use case identifiers corresponding to each of the multiple tasks; The service type corresponding to each of the multiple tasks; business identifiers of the plurality of tasks; The expected accuracy of the multiple tasks; data storage requirements for the multiple tasks; The input data formats of the plurality of tasks; Output data formats of the plurality of tasks; Generalization metrics for the multiple tasks described; or, When the fifth message is used to request an AI unit required by a second task of the AI ​​service, and the second task is a single task among the multiple tasks, the fifth message includes at least one of the following: a task identifier of the second task; A service identifier corresponding to the second task; A use case identifier corresponding to the second task; The service type corresponding to the second task; the expected accuracy of the second task; data storage requirements of the second task; the data input format of the second task; The data output format of the second task; A generalization indicator of the second task; an identifier of a third task, the third task being a task before the second task and associated with the second task; or, When the fifth message is used to request an AI unit required by a second task of the AI ​​service, and the second task is a single task among the multiple tasks, the fifth message includes at least one of the following: Task identification; Business logo; Use case identification; Type of service; Expected accuracy; Data storage requirements; Input data format; Output data format; Generalizability index; An identifier of a fourth task, where the fourth task is a task that precedes the second task and is associated with the second task.

17. The method according to claim 15 or 16, characterized in that The method further comprises: The third network element sends a sixth message to the seventh network element, wherein the sixth message is used to request AI resources for at least one task of the AI ​​service.

18. The method according to claim 17, characterized in that When the sixth message is used to request AI resources for multiple tasks of the AI ​​service, the sixth message includes at least one of the following: AI unit information corresponding to each of the multiple tasks; Data source information of the multiple tasks; the computational effort of the plurality of tasks; the maximum data storage requirements of the plurality of tasks; the processing latency requirements of the multiple tasks; The input data formats of the plurality of tasks; the number of single-sample inferences for each of the multiple tasks; or, When the sixth message is used to request AI resources for a fifth task of the AI ​​service, and the fifth task is a single task among the multiple tasks, the sixth message includes at least one of the following: The task identifier of the fifth task; AI unit information corresponding to each of the fifth tasks; Data source information of the fifth task; The computational amount of the fifth task; The data storage requirements of the fifth task; The processing delay requirement of the fifth task; The input data format of the fifth task; The number of single-sample inferences of the fifth task; or, When the sixth message is used to request AI resources for a fifth task of the AI ​​service, and the fifth task is a single task among the multiple tasks, the sixth message includes at least one of the following: Task identification; AI unit information; Data source information; Calculation amount; Data storage requirements; Handling latency requirements; Input data format; Number of single-sample inferences.

19. The method according to any one of claims 15 to 18, characterized in that The method further comprises: When completing the kth task of the AI ​​service, the third network element performs at least one of the following: triggering the AI ​​resource of the next task of the k-th task; For the next task of the k-th task, obtain a new initial AI unit; Among them, the kth task is not the last task of the AI ​​business.

20. A business processing method, characterized in that: include: The fourth network element receives the first data sent by the second network element; The fourth network element performs an AI unit processing operation according to the first data; The fourth network element sends second data to the second network element.

21. The method according to claim 20, characterized in that The first data includes at least one of the following: Input data of the AI ​​unit for the AI ​​service; Label data for AI unit processing operations; sorting information for a plurality of inference results corresponding to a single input data of an AI unit for the AI ​​service; The structure of the AI ​​unit used for the AI ​​service; Parameters of the AI ​​unit used for the AI ​​service; Description information of the AI ​​unit used for the AI ​​service; Output of the segmentation layer in the AI ​​unit for the AI ​​service; A gradient of a segmentation layer in an AI unit for the AI ​​service; or, The second data includes at least one of the following provided by the network: Input data of the AI ​​unit for the AI ​​service; Label data for AI unit processing operations; Output data of the AI ​​unit for the AI ​​service; Input data of an AI unit for the AI ​​service generated based on reinforcement learning; The structure of the AI ​​unit used for the AI ​​service; Parameters of the AI ​​unit used for the AI ​​service; Description information of the AI ​​unit used for the AI ​​service; Output of the segmentation layer in the AI ​​unit for the AI ​​service; The gradient of the segmentation layer in the AI ​​unit for the AI ​​service.

22. The method according to claim 20 or 21, characterized in that The AI ​​unit processing operation includes at least one of the following: AI unit aggregation; Single AI unit training; Multiple AI unit training; Single AI unit reasoning; Multiple AI unit reasoning.

23. The method according to any one of claims 20 to 22, characterized in that The method further comprises: When completing the kth task of the AI ​​service, the fourth network element performs at least one of the following: Do not release AI resources; Obtaining AI resources for the next task of the k-th task; For the next task of the k-th task, obtain a new initial AI unit, or use the AI ​​unit obtained after performing the k-th task; Among them, the kth task is not the last task of the AI ​​business.

24. The method according to any one of claims 20 to 23, characterized in that The method further comprises: The fourth network element sends first information to the sixth network element, wherein the first information includes at least one of the following: Updated AI unit parameters; Updated AI unit logo; The task ID associated with the updated AI unit.

25. The method according to any one of claims 20 to 24, characterized in that The method further comprises: The fourth network element receives configuration information sent by the seventh network element, wherein the configuration information is used to configure AI resources, and the configuration information includes at least one of the following: A third indication, wherein the third indication is used to indicate an AI framework; A fourth indication, wherein the fourth indication is used to indicate an AI library version; A fifth indication, wherein the fifth indication is used to indicate the number of single sample inferences; Data source information; AI business detection information; AI business processing rules.

26. A service processing device, characterized in that: include: An acquisition module, configured to acquire an establishment request from a second network element to establish an AI service, where the AI ​​service includes at least one task; A determination module, configured to determine an AI strategy for at least one task in the AI ​​service according to the establishment request; The first sending module is used to return a response message of the establishment request to the second network element according to the AI ​​strategy.

27. A service processing device, characterized in that: include: A third sending module is configured to send a request to establish an AI service to the first network element, where the AI ​​service includes at least one task; A first receiving module, used to receive a response message of the establishment request returned by the first network element; The response message is returned according to the AI ​​policy, and the AI ​​policy is determined for at least one task in the AI ​​service according to the establishment request; the first network element is a core network element, and the second network element is a third-party server.

28. A service processing device, characterized in that: include: a fifth sending module, configured to send a fifth message to a sixth network element, where the fifth message is used to request an AI unit required for at least one task of the AI ​​service; The third receiving module is used to receive a response message to the fifth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

29. A service processing device, characterized in that: include: A fourth receiving module, used for receiving first data sent by the second network element; A processing module, configured to perform an AI unit processing operation according to the first data; The seventh sending module is used to send second data to the second network element.

30. A communication system, characterized in that: include: A first network element, a second network element, a third network element and a fourth network element, wherein: The first network element is used to: obtain an establishment request from the second network element to establish an AI service, where the AI ​​service includes at least one task; determine an AI strategy for at least one task in the AI ​​service according to the establishment request; and return a response message of the establishment request to the second network element according to the AI ​​strategy; The second network element is used to: send an establishment request to the first network element for requesting to establish an AI service; and receive a response message to the establishment request returned by the first network element; The third network element is used to: send a sixth message to the sixth network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service; receive a response message to the sixth message returned by the sixth network element, where the response message is used to indicate the AI ​​unit required for the at least one task; The fourth network element is used to: receive the first data sent by the second network element; perform the AI ​​unit processing operation according to the first data; and send the second data to the second network element; Among them, the first network element, the third network element and the fourth network element are core network elements, and the second network element is a third-party server.

31. The communication system according to claim 30, characterized in that The communication system further includes a fifth network element, wherein: The fourth network element is further configured to: receive first data sent by the second network element through the fifth network element, and send second data to the second network element through the fifth network element.

32. The communication system according to claim 30, characterized in that The communication system further includes a sixth network element, wherein: The sixth network element is used to: receive a sixth message sent by the third network element, where the sixth message is used to request an AI unit required for at least one task of the AI ​​service; and send a response message to the sixth message to the third network element, where the response message is used to indicate the AI ​​unit required for the at least one task.

33. The communication system according to claim 30, characterized in that The communication system further includes a seventh network element, wherein: The third network element is further used to: send a sixth message to the seventh network element, where the sixth message is used to request AI resources for at least one task of the AI ​​service; The fourth network element is further used to: receive configuration information sent by the seventh network element, where the configuration information is used to configure AI resources.

34. A communication device, characterized in that: It includes a processor and a memory, the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, it implements the steps of the business processing method as described in any one of claims 1 to 6, or implements the steps of the business processing method as described in any one of claims 7 to 14, or implements the steps of the business processing method as described in any one of claims 15 to 19, or implements the steps of the business processing method as described in any one of claims 20 to 25.

35. A readable storage medium, characterized in that: The readable storage medium stores programs or instructions, and when the programs or instructions are executed by the processor, they implement the steps of the business processing method as described in any one of claims 1 to 6, or implement the steps of the business processing method as described in any one of claims 7 to 14, or implement the steps of the business processing method as described in any one of claims 15 to 19, or implement the steps of the business processing method as described in any one of claims 20 to 25.

Images