Service execution method and apparatus, computer device, storage medium, and product

Abstract

The application relates to a service execution method and device, computer equipment, a storage medium and a product. The method is applied to a wireless access network intelligent layer, and comprises the following steps: in the case that a service request for a target artificial intelligence (AI) service sent by a user equipment is received, generating traffic perception information for the target AI service; according to the traffic perception information, determining data transmission information of data required for executing the target AI service and model transmission information of a required model; according to the data transmission information and the model transmission information, processing the target AI service into at least two transmission tasks, and determining resource requirement information of each transmission task; and sending available transmission resource information and the resource requirement information of each transmission task to a policy control face (PCF) network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task. The method can efficiently execute the AI service in real time, and improves network resource utilization.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wireless communication, and in particular to a service execution method and device, a computer device, a storage medium and a product. BACKGROUND

[0002] The rapid development of network intelligence will bring profound changes to the communication system, in which the traffic form will change from traditional communication traffic to AI (Artificial Intelligence) traffic. AI traffic is more data flow generated or driven by AI algorithms, including massive training data, model updates, input and output in inference processes, etc. The introduction of AI traffic means that the network will have self-learning and adaptive capabilities, and can handle complex and dynamic demand changes in real time.

[0003] However, the current AI system may continuously run in the background, frequently upload or download large-scale data (such as sensor data, training data sets, inference requests, etc.), and may interact data between different places and different devices, resulting in more complex and unpredictable distribution of AI traffic. Therefore, how to perform multi-element collaborative transmission for AI traffic and optimize resource allocation strategies to efficiently execute AI services becomes a difficult problem. SUMMARY

[0004] Therefore, it is necessary to provide a service execution method, device, computer device, storage medium and product to efficiently execute AI services in real time and improve network resource utilization.

[0005] In a first aspect, the present application provides a service execution method applied to a radio access network intelligent layer, comprising:

[0006] In the case of receiving a service request for a target artificial intelligence (AI) service sent by a user equipment, generating traffic perception information for the target AI service;

[0007] According to the traffic perception information, determining data transmission information of data required for executing the target AI service and model transmission information of a required model;

[0008] According to the data transmission information and the model transmission information, processing the target AI service into at least two transmission tasks, and determining resource requirement information of each transmission task;

[0009] Sending available transmission resource information and resource requirement information of each transmission task to a policy control face (PCF) network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0010] In one of the embodiments, the traffic awareness information at least includes service subscription information of the target AI service;

[0011] The determining of the data transmission information and the model transmission information required for executing the target AI service according to the traffic awareness information comprises:

[0012] The determining of model attribute information of a model required for executing the target AI service according to the service subscription information;

[0013] The determining of data attribute information of data required for executing the target AI service according to the model attribute information;

[0014] The determining of the model transmission information according to the model attribute information, and the determining of the data transmission information according to the data attribute information.

[0015] In one of the embodiments, the method further comprises:

[0016] Obtaining real-time network resource data and service quality (QoS) resource allocation strategy data;

[0017] The determining of the available transmission resource information according to the real-time network resource data and the QoS resource allocation strategy data.

[0018] In one of the embodiments, the data transmission information includes data identification information, data storage location, data storage format, data processing identification, data volume, data security identification, and data transmission strategy;

[0019] The model transmission information includes model authentication information, model storage location, model life cycle, model deployment location, and model transmission strategy.

[0020] In a second aspect, the application provides a service execution method applied to a PCF network element, comprising:

[0021] Receiving available transmission resource information and resource requirement information of each transmission task sent by a radio access network intelligent layer;

[0022] Determining a task execution strategy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task; wherein each task execution strategy includes a QoS strategy and a network access management strategy; each transmission task is obtained by processing the target AI service according to data transmission information of data required for executing the target AI service and model transmission information of a model required for executing the target AI service by the radio access network intelligent layer;

[0023] After receiving the policy obtaining request of the session management plane (SMF) network element, the task execution policies of the transmission tasks are transmitted to the SMF network element, so that the SMF network element executes each transmission task according to the task execution policy of each transmission task.

[0024] In one of the embodiments, the policy obtaining request is sent by the SMF network element after detecting that the network policy changes; or the policy obtaining request is sent by the SMF network element after receiving a session update request.

[0025] In one of the embodiments, the method further comprises:

[0026] The resource configuration completion response is sent to the radio access network intelligent layer, so that the radio access network intelligent layer sends a session update request to the SMF network element.

[0027] In a third aspect, the present application provides a service execution method, applied to an SMF network element, comprising:

[0028] The task execution policies of the transmission tasks sent by the PCF network element are received; wherein the task execution policies of the transmission tasks are obtained by the PCF network element according to available transmission resource information and resource requirement information of each transmission task, and each transmission task is obtained by processing a target AI service by the radio access network intelligent layer according to data transmission information of data required for executing the target AI service and model transmission information of a required model;

[0029] According to the task execution policies of each transmission task, the protocol data unit (PDU) session information corresponding to each transmission task is determined;

[0030] According to the PDU session information corresponding to each transmission task and the function information of the candidate user plane function (UPF) network element, a target UPF network element is selected from the candidate UPF network elements;

[0031] The task execution policies of each transmission task are sent to the target UPF network element, so that the target UPF network element executes each transmission task.

[0032] In a fourth aspect, the present application further provides a service execution device, configured in a radio access network intelligent layer, comprising:

[0033] The first obtaining module is configured to generate traffic sensing information for a target AI service if a service request for the target AI service sent by a user equipment is received;

[0034] The data determining module is configured to determine data transmission information of data required for executing the target AI service and model transmission information of a required model according to the traffic sensing information;

[0035] a task splitting module, configured to split the target AI service into at least two transmission tasks according to the data transmission information and the model transmission information, and determine resource requirement information of each transmission task;

[0036] a first execution module, configured to send the available transmission resource information and the resource requirement information of each transmission task to a PCF network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0037] In a fifth aspect, the present application further provides a service execution apparatus configured in a PCF network element, comprising:

[0038] a second acquisition module, configured to receive available transmission resource information and resource requirement information of each transmission task sent by a radio access network intelligent layer;

[0039] a policy determination module, configured to determine a task execution policy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task; wherein each task execution policy comprises a QoS policy and a network access management policy; each transmission task is obtained by processing the target AI service according to data transmission information of data required for executing the target AI service and model transmission information of a required model by the radio access network intelligent layer;

[0040] a second execution module, configured to transmit the task execution policy of each transmission task to an SMF network element after receiving a policy acquisition request of the SMF network element, so that the SMF network element executes each transmission task according to the task execution policy of each transmission task.

[0041] In a sixth aspect, the present application further provides a service execution apparatus configured in an SMF network element, comprising:

[0042] a third acquisition module, configured to receive the task execution policy of each transmission task sent by a PCF network element; wherein the task execution policy of each transmission task is obtained by the PCF network element according to available transmission resource information and resource requirement information of each transmission task, and each transmission task is obtained by processing the target AI service according to data transmission information of data required for executing the target AI service and model transmission information of a required model by a radio access network intelligent layer;

[0043] an information determination module, configured to determine PDU session information corresponding to each transmission task according to the task execution policy of each transmission task;

[0044] a function selection module, configured to select a target UPF network element from candidate UPF network elements according to the PDU session information corresponding to each transmission task and function information of the candidate UPF network elements;

[0045] The third execution module is configured to send the task execution strategy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task.

[0046] In a seventh aspect, the present application also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method in any one of the first aspect, the second aspect and the third aspect when executing the computer program.

[0047] In an eighth aspect, the present application also provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the method in any one of the first aspect, the second aspect and the third aspect when executed by a processor.

[0048] In a ninth aspect, the present application also provides a computer program product, comprising a computer program, and the computer program implements the steps of the method in any one of the first aspect, the second aspect and the third aspect when executed by a processor.

[0049] The service execution method, device, computer device, storage medium and product described above introduce the radio access network intelligent layer, and formulate the signaling interaction mode between the radio access network intelligent layer and the PCF network element and the SMF network element, to cooperatively execute the AI service, and improve the adaptation and processing capability of the existing network architecture to the AI traffic. Furthermore, after the radio access network intelligent layer generates the traffic perception information for the target AI service when perceiving the service request of the target AI service, the radio access network intelligent layer determines the data transmission information of the data required for executing the target AI service and the model transmission information of the model required for executing the target AI service according to the traffic perception information, further processes the target AI service into at least two transmission tasks according to the data transmission information and the model transmission information, and determines the resource requirement information of each transmission task, to realize the unified control and distribution of AI elements based on the radio access network intelligent layer. Finally, the available transmission resource information and the resource requirement information of each transmission task are sent to the PCF network element, to provide the downlink data transmission guarantee of the customized AI service for the user node, so that an end-to-end AI traffic transmission mechanism is established at the operator level. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0051] Figure 1 An application environment diagram of the service execution method in an embodiment;

[0052] Figure 2 a flowchart of a service execution method in an embodiment;

[0053] Figure 3 a flowchart of a method for determining data transmission information and model transmission information in an embodiment;

[0054] Figure 4 a flowchart of a method for obtaining available transmission resource information in an embodiment;

[0055] Figure 5 a flowchart of a service execution method in another embodiment;

[0056] Figure 6 a flowchart of a service execution method in yet another embodiment;

[0057] Figure 7 a signaling diagram of a service execution method in an embodiment;

[0058] Figure 8 a structural block diagram of a service execution apparatus in an embodiment;

[0059] Figure 9 a structural block diagram of a service execution apparatus in another embodiment;

[0060] Figure 10 a structural block diagram of a service execution apparatus in yet another embodiment;

[0061] Figure 11 an internal structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0062] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application.

[0063] The service execution method provided by the embodiments of the present application can be applied to, for example, Figure 1The application environment shown. Among them, the radio access network intelligent layer 101 is the AI traffic unified arrangement manager under the architecture of the 5G (5th Generation Mobile Communication) or 6G (6th Generation Mobile Communication) network, responsible for performing functions such as data transmission, user perception, multi-element collaborative control, model life cycle management, and the like related to AI traffic; the PCF (Policy Control Function) network element 102 is used to manage various behaviors and policies in the network; the SMF (Session Management Function) network element 103 is mainly responsible for functions such as session management, selection and control of UPF (User Plane Function) network elements, traffic control, policy execution, and QoS (Quality of Service) control. The user equipment 104 can be a device of an enterprise user, a personal user, or the like; the base station device 105 is a public mobile communication base station, wherein the user equipment can be referred to as UE (User Equipment) in short. Optionally, the radio access network intelligent layer 101 generates traffic perception information for the target AI service in a case where the service request for the target AI service is sent by the user equipment 104 through the base station device 105; determines data transmission information of data required for executing the target AI service and model transmission information of a required model according to the traffic perception information; processes the target AI service into at least two transmission tasks according to the data transmission information and the model transmission information, and determines resource requirement information of each transmission task; and sends available transmission resource information and the resource requirement information of each transmission task to the PCF network element 102, so that the PCF network element 102 executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0064] In one exemplary embodiment, as shown in Figure 2 A service execution method is provided, and the method is applied to the radio access network intelligent layer 101 in Figure 1 for example, and specifically includes the following steps:

[0065] S201, in a case where a service request for a target artificial intelligence (AI) service is sent by user equipment, generate traffic perception information for the target AI service.

[0066] It should be noted that in the embodiments of the present application, the process of executing the target AI service is implemented by transmitting corresponding AI traffic.

[0067] The target AI service can be any AI service required by the user, such as model training, model updating, and the like. The traffic awareness information is information generated by the radio access network intelligent layer when the service request is sensed. In the embodiments of the present application, the traffic awareness information includes but is not limited to user equipment identification, access control information (for example, the user equipment belongs to the UPF network element), AI service subscription information, and the like; wherein the AI service subscription information represents the basic information of the AI service subscribed by the UE, at least including AI service identification, AI service type, AI service validity time, AI service requirement, AI service input, and the like.

[0068] The AI service type represents the basic type of the AI service, such as remote medical diagnosis service, real-time text interaction service, automatic driving service, intelligent robot service, and the like; the AI service requirement represents the requirement of performance for executing the corresponding AI service, such as delay, throughput, energy consumption, and the like; the AI service input represents the user input required by the corresponding AI service, such as the prompt word (Prompt) or instruction (Instruction) required by the large language model. The AI service subscription information can correspond to multiple AI services, and the corresponding information can be arranged in order of AI service identification.

[0069] Optionally, the UE will send a service request for the target AI service to the network through the base station device when there is a demand for the target AI service, and further, the radio access network intelligent layer senses the AI traffic uploaded by the UE about the target AI service and generates traffic awareness information.

[0070] S202, according to the traffic awareness information, determining data transmission information required for executing the target AI service and model transmission information required for the model.

[0071] The model required for executing the target AI service can be an AI model, and the model transmission information represents the model-related information required for transmitting the AI model. In the embodiments of the present application, the model transmission information includes but is not limited to model authentication information, model storage location, model life cycle, model deployment location, and model transmission strategy of the required model.

[0072] The model certification information includes but is not limited to a model identifier, model parameters and a model architecture; the model storage information includes a model storage location, a model storage mode, a model storage precision indication and the like, the model storage precision indication is a numerical precision format adopted when storing and calculating model parameters, is usually closely related to calculation efficiency, memory occupation and model performance, and is generally at least one of FP32 (floating-point 32), FP16 (floating-point 16), BF16 (Bfloat16), INT8 (8-bit Integer) or mixed precision.

[0073] The model life cycle includes a model life cycle management stage identifier and policy information corresponding to the model life cycle management stage identifier. The model life cycle management stage identifier is used to indicate that the AI model needs to be applied to the model training, model inference, model updating / fine-tuning, model monitoring and the like in the execution of the target AI service, and the policy information corresponding to the model life cycle management stage identifier indicates how to use the AI model to execute the target AI service in the current model life cycle management stage, including used hyperparameters, inference configurations and optimization algorithms. The hyperparameters refer to parameters preset in model training or inference, such as a learning rate, a batch size, an optimizer type and the like; the inference configuration includes specific settings in model inference, such as precision selection, inference framework selection (for example, TensorFlow, PyTorch), hardware accelerator selection and the like; and the optimization algorithm refers to a method for improving model efficiency or performance in the current stage, such as model pruning, quantization, distillation or dynamic inference scheduling and the like.

[0074] The model deployment location is used to indicate whether the AI model is deployed on the UE side (user side), the NW (Network) side or the double-side deployment mode (user side and network side), the UE side deployment needs to transmit the AI model to the UE side to execute the corresponding model life cycle management stage; the NW side deployment needs to transmit the AI model to an AI-related network element to execute the corresponding model life cycle management stage; and the double-side deployment mode indicates that the AI model needs to be deployed on both sides. It should be noted that one AI service can correspond to multiple AI models, each AI model can correspond to multiple model life cycle management stage identifiers, and the model transmission information needs to give information corresponding to each AI model.

[0075] The model transmission strategy includes but is not limited to whether the model is transmitted in segments, compressed transmission, model file transmission mode indication and the like.

[0076] The data required for executing the target AI service can be training data, prediction data, etc. In the embodiments of the present application, the data transmission information includes but is not limited to data identification information, data storage location, data storage format, data processing identification, data volume, data security identification, and data transmission strategy.

[0077] The data identification information refers to data ID (Identity), which is used to distinguish data according to geographical location, storage method, protocol, etc. The data storage location refers to the physical or logical storage location of the data, including edge nodes (such as base stations), central cloud servers, distributed databases, or user devices locally. The data storage format refers to the organization and encoding method of the data, i.e., how the data is stored in a file format or structure, such as Comma-Separated Values (CSV), JavaScript Object Notation (JSON), or 5G-specific data packaging format. The data processing identification is used to identify the indication of processing or preprocessing the data, indicating whether the data needs specific processing and the processing method, including data cleaning, denoising, normalization, feature extraction, encoding, fast Fourier transform, etc. The data volume refers to the data size or quantity used to execute the target AI service, usually represented by the number of data entries, file size, time range, network load. The data security identification is used to identify the data sensitivity or security requirements, indicating the protection level and access permission requirements of the data, including whether the data is encrypted, access control level, compliance requirements, whether it involves user privacy data or network configuration data. The data transmission strategy includes but is not limited to whether to use data verification mechanism, data encryption mechanism (based on data security identification), etc.

[0078] Optionally, AI service subscription information can be obtained from the traffic perception information, and a model mapping table of the AI model used to execute the target AI service is determined based on the AI service subscription information. The model mapping table corresponds to a specific AI service and is used to represent the relevant model information identified based on the AI service. Then, the model basic information of the required model can be obtained according to the mapping table, such as model authentication information, model storage location, model life cycle, model deployment location, and the model basic attribute information is packaged for further formulating a model transmission strategy for the transmission of the model, and the packaged model basic attribute information and model transmission strategy are used as model transmission information.

[0079] Further, the model mapping table is used to determine the data required for executing the target AI service, and a data mapping table is obtained; the data mapping table corresponds to the specific AI service and is used to represent the relevant data information identified based on the AI service. Data attribute information is obtained according to the data mapping table, such as data identification information, data storage location, data storage format, data processing identifier, data volume, data security identifier, etc., and the data attribute information is packaged, and further, a data transmission strategy is formulated for data transmission, and the packaged data attribute information and the data transmission strategy are used as data transmission information.

[0080] In S203, the target AI service is processed into at least two transmission tasks according to the data transmission information and the model transmission information, and resource requirement information of each transmission task is determined.

[0081] The transmission task is used to transmit part of the model and data required for executing the target AI service. The resource requirement information of each transmission task represents the network resources required for executing each transmission task, such as bandwidth, latency, transmission protocol support, etc.

[0082] Optionally, the radio access network intelligent layer formulates at least one transmission task according to the data transmission information, and formulates at least one transmission task based on the model transmission information, so as to map the target AI service into at least two transmission tasks; it should be noted that each transmission task corresponds to the transmission of one or more user plane data packets, and each transmission task has a transmission task identifier corresponding to the corresponding target AI service identifier. The transmission tasks have relevance in terms of work content, and the granularity of the work tasks is controlled by the radio access network intelligent layer.

[0083] It should be noted that the radio access network intelligent layer also generates transmission control information according to the data transmission information and the model transmission information, which is used to indicate the source address and the destination address of the transmission, the AI domain name system server address, the AI related network element address, the third party application information, the transmission data field meaning, etc. The transmission control information can be transmitted in series in the data transmission information and the model transmission information, or can be transmitted separately.

[0084] Further, resource requirements are formulated for each transmission task, and resource requirement information of each transmission task is generated. It should be noted that a resource mapping table can be formulated according to the transmission content and the resource requirement information of each transmission task, to record the transmission resource requirements corresponding to each transmission task. The specific form can be [transmission task 1: resource requirement 1; …; transmission task n: resource requirement n], where n is a positive integer greater than 1.

[0085] It should be noted that in the process of generating the transmission task, for the model transmission information, the deployment position of the model is different according to the model deployment position identifier in the model transmission information, that is, the transmission target of the model is different. Specifically, if deployed on the UE side, the model authentication information and the corresponding policy information corresponding to the model life cycle management stage identifier need to be transmitted to the UE, and the corresponding model file is obtained according to the model storage position. If deployed on the NW side, the UE needs to be fed back the result processed by the model life cycle management stage, such as model inference result, model update result, etc., and the corresponding description information. Among them, the result processed by the model life cycle management stage is the result obtained by the AI-related network element according to the determined model and data, and the corresponding policy information of the model life cycle management stage identifier is executed according to the model life cycle management process.

[0086] Correspondingly, for the data transmission information, the corresponding data transmission target is different according to the different deployment positions of the model. If the model is deployed on the UE side, the required data is directly transmitted to the UE. If the model is deployed on the NW side, the required data is transmitted to the AI-related network element, and the feedback result is obtained. At this time, the required data does not need to be transmitted to the UE.

[0087] S204, the available transmission resource information and the resource requirement information of each transmission task are sent to a policy control plane PCF network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0088] Among them, the available transmission resource information represents the transmission resources that can be used for data transmission in the network, such as bandwidth, delay and transmission protocol support, etc.

[0089] It should be noted that the resource requirement information of each transmission task can also include the transmission resource type, allocation quantity, service time, service level agreement (Service Level Agreement, SLA) and shared resource identifier of the transmission task of the wireless access network intelligent layer. Among them, the SLA is determined by the wireless access network intelligent layer and the third party application.

[0090] Optionally, the available transmission resource information and the resource requirement information of each transmission task can be sent to the PCF network element, so that the PCF network element allocates available resources for each transmission task from the available transmission resource information according to the resource requirement information of each transmission task, and formulates the corresponding task execution strategy for each transmission task, so that each transmission task is executed based on the task execution strategy.

[0091] In the service execution method, the radio access network intelligent layer is introduced, and signaling interaction between the radio access network intelligent layer and the PCF network element and the SMF network element is formulated to cooperatively execute the AI service, and the adaptation and processing capability of the existing network architecture to the AI traffic is improved. Then, after the radio access network intelligent layer generates the traffic perception information for the target AI service when perceiving the service request of the target AI service, the data transmission information of the data required for executing the target AI service and the model transmission information of the model required for executing the target AI service are determined according to the traffic perception information, the target AI service is further processed into at least two transmission tasks according to the data transmission information and the model transmission information, and the resource requirement information of each transmission task is determined, so that the unified control and distribution of AI elements based on the radio access network intelligent layer are realized. Finally, the available transmission resource information and the resource requirement information of each transmission task are sent to the PCF network element, the downlink data transmission guarantee of the customized AI service for the user node is provided, and an end-to-end AI traffic transmission mechanism is established at the operator level.

[0092] Optionally, the traffic perception information at least includes service subscription information of the target AI service. On this basis, in order to accurately and quickly obtain the data transmission information and the model transmission information, in an exemplary embodiment, as shown in Figure 3 A method for determining data transmission information and model transmission information is provided to refine S202, specifically including the following steps:

[0093] S301, according to the service subscription information, determining model attribute information of a model required for executing the target AI service.

[0094] The model attribute information is information for describing and displaying the model required for executing the target AI service. In the embodiments of the present application, the model attribute information includes but is not limited to model type, model authentication information, model storage location, model life cycle, model deployment location and the like.

[0095] Optionally, the service type of the target AI service can be obtained from the service subscription information, and the AI model for executing the AI service of this type is obtained from the local storage as the model required for executing the target AI service, and the model authentication information, the model storage location, the model life cycle, the model deployment location and the like corresponding to the AI model are obtained from the local storage as the model attribute information.

[0096] S302, according to the model attribute information, determining data attribute information of data required for executing the target AI service.

[0097] The data attribute information is information used for describing and presenting data required for executing the target AI service. In the embodiments of the present application, the data attribute information includes, but is not limited to, data identification information, data storage location, data storage format, data processing identification, data volume, data security identification and the like.

[0098] Optionally, based on the model type in the model attribute information, the data corresponding to the model type is obtained from the locally stored database as the data required for executing the target AI service, and the data identification information, the data storage location, the data storage format, the data processing identification, the data volume, the data security identification and the like corresponding to the required data are obtained as the data attribute information of the data required for executing the target AI service.

[0099] S303, determining the model transmission information according to the model attribute information, and determining the data transmission information according to the data attribute information.

[0100] Optionally, the radio access network intelligent layer encapsulates the model attribute information into a data packet and formulates a corresponding model transmission strategy, including whether the model is transmitted in segments, whether the model is transmitted in compression, a model file transmission mode indication and the like, to form the model transmission information. Correspondingly, the data attribute information is also encapsulated into a data packet and a corresponding data transmission strategy is formulated, including whether a data verification mechanism is used, a data encryption mechanism (based on the data security identification) is used, to form the data transmission information.

[0101] In the embodiments, based on the service subscription information, the accuracy of the model attribute information of the required model obtained is ensured, and then the data required for executing the target AI service, i.e., the data required for using the required model, can be accurately determined according to the model attribute information, and the data attribute information can be accurately determined, and then the accuracy of the determined model transmission information and data transmission information is ensured.

[0102] Optionally, the available transmission resource information represents the transmission resources, such as bandwidth, delay and transmission protocol support, that can be used for transmitting data in the network. In order to obtain more accurate available transmission resource information, in an embodiment, as shown in Figure 4 an acquisition method of available transmission resource information is provided, which specifically includes the following steps:

[0103] S401, obtaining real-time network resource data and quality of service (QoS) resource allocation strategy data.

[0104] The real-time network resource data is real-time resource information in the network, such as network capacity, link status, node resource and health status and the like; the Quality of Service (QoS) is a bandwidth allocation for various services under limited bandwidth resources, and provides end-to-end service quality guarantee for the services. The QoS resource allocation strategy data is a QoS resource that can be allocated in the network.

[0105] Optionally, the radio access network intelligent layer can obtain a stored overall view of network resources from an Operation Administration and Maintenance (OAM) system, including network capacity, link status, node resource and health status and the like, to generate real-time network resource data; and directly obtain a QoS flow resource allocation strategy from a PCF network element.

[0106] S402, determining available transmission resource information according to the real-time network resource data and the QoS resource allocation strategy data.

[0107] Optionally, the transmission resource information can be determined based on intelligent analysis algorithms according to the real-time network resource data and the QoS resource allocation strategy data, for recording allocable transmission resource information at the current time. In the embodiment of the application, the granularity of the available transmission resource information can be UE level, cell level, network level or other granularity, which is controlled by the radio access network intelligent layer.

[0108] In the embodiment, by obtaining the real-time network resource data and the Quality of Service (QoS) resource allocation strategy data, the real-time available resource situation in the network can be accurately analyzed, and therefore, the available transmission resource information in the network can be accurately and efficiently determined.

[0109] In an exemplary embodiment, as shown in Figure 5 , a service execution method is provided, which is applied to a PCF network element in Figure 1 for example, and specifically includes the following steps:

[0110] S501, receiving available transmission resource information and resource requirement information of each transmission task sent by a radio access network intelligent layer.

[0111] S502, determining a task execution strategy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0112] Each task execution strategy includes a QoS strategy and a network access management strategy; each transmission task is obtained by processing a target AI service by the radio access network intelligent layer according to data transmission information of data required for executing the target AI service and model transmission information of a model required for executing the target AI service.

[0113] Optionally, the PCF network element formulates the QoS policy and network access management policy used for each transmission task according to the available transmission resource information and the resource requirement information of each transmission task. Specifically, the PCF network element formulates a specific QoS policy for each work load based on the service requirement of the target AI service and the resource requirement information of each transmission task, including scheduling priority, etc. For example, a model file with low latency requirement can use a low QoS policy (transmission to the AI-related network element), a model inference result with high latency requirement needs high QoS guarantee, and when transmitting to the network edge user, the QoS guarantee needs to be formulated according to the allocation result of the transmission resource.

[0114] Optionally, the PCF network element also determines whether to allow the user to access the network according to the user's access control information and the current network state, and formulates the corresponding access strategy.

[0115] It should be noted that the PCF network element supports dynamic adjustment strategy. With the changes of network state, user behavior, user subscription service and AI service requirement, the PCF network element can dynamically adjust and optimize the above-mentioned strategies. The PCF network element supports formulating transmission strategy for multiple transmission tasks uniformly, binding multiple transmission tasks as one transmission task, and transmitting uniformly through aggregated QoS flow. Since the AI-related network element is deployed on the network side, the data transmitted to the AI-related network element does not involve the process of formulating QoS policy by the PCF network element, but is directly transmitted through the internal interface of the core network.

[0116] S503, after receiving the policy acquisition request of the session management plane (SMF) network element, the task execution strategy of each transmission task is transmitted to the SMF network element, so that the SMF network element executes each transmission task according to the task execution strategy of each transmission task.

[0117] Among them, the policy acquisition request is used to indicate that the PCF network element sends the task execution strategy of each transmission task to the SMF network element.

[0118] Optionally, the policy acquisition request is sent by the SMF network element after detecting that the network policy changes; or the policy acquisition request is sent by the SMF network element after receiving the session update request. That is, there are two ways for the SMF network element to send the policy acquisition request to the PCF network element. Further, the session update request is sent by the radio access network intelligent layer to the SMF after the PCF network element sends the resource configuration completion response to the radio access network intelligent layer.

[0119] Optionally, after receiving the policy obtaining request of the session management face (SMF) network element, the PCF network element transmits the task execution policy of each transmission task to the SMF network element, and then the SMF network element determines the protocol data unit (PDU) session information corresponding to each transmission task according to the task execution policy of each transmission task after receiving each transmission task; wherein the PDU session information includes but is not limited to at least one of QoS flow configuration, traffic type change (for example, Non-GBR traffic changes to GBR traffic), data network name, UE IP address allocation, routing policy, network slice information single network slice selection assistance information, AI authentication information, and security configuration. Wherein the QoS configuration refers to updating the QoS flow according to the QoS policy formulated by the PCF, including adjusting the QoS flow parameters (such as 5QI, priority, and bandwidth requirement); the AI authentication information includes at least one of AI service ID, AI service type, AI service validity time, and other AI related information.

[0120] Further, the SMF network element selects a suitable UPF network element from the candidate user plane function (UPF) network element as a target UPF network element for each transmission task. For example, considering the geographical location (whether it is at the network edge), bandwidth requirement, and delay requirement, the SMF selects a suitable UPF network element to carry user data.

[0121] Finally, the selected target UPF network element routes and forwards different QoS data flows, and provides different quality of service guarantees for QoS flows.

[0122] In this embodiment, after receiving the available transmission resource information sent by the radio access network intelligent layer and the resource requirement information of each transmission task, the PCF network element determines the task execution policy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task, wherein each task execution policy includes a QoS policy and a network access management policy; since each transmission task is obtained by processing the target AI service according to the data transmission information of the data required to execute the target AI service and the model transmission information of the required model by the radio access network intelligent layer, this scheme formulates a signaling transmission mode between the radio access network intelligent layer and the PCF network element; finally, after receiving the policy obtaining request of the session management face (SMF) network element, the task execution policy of each transmission task is transmitted to the SMF network element, so that the SMF network element executes each transmission task according to the task execution policy of each transmission task, and ensures that the target AI service can be efficiently executed.

[0123] In one exemplary embodiment, as shown in Figure 6 , a service execution method is provided, which is applied to the SMF network element in Figure 1 for example, and specifically includes the following steps:

[0124] S601, receiving a task execution strategy of each transmission task sent by a PCF network element.

[0125] The task execution strategy of each transmission task is obtained by the PCF network element according to available transmission resource information and resource requirement information of each transmission task, and each transmission task is obtained by processing a target AI service according to data transmission information of data required for executing the target AI service and model transmission information of a model required for executing the target AI service.

[0126] S602, determining protocol data unit (PDU) session information corresponding to each transmission task according to the task execution strategy of each transmission task.

[0127] Optionally, the SMF network element establishes or modifies a parameter configuration related to the PDU session corresponding to each transmission task according to the task execution strategy of each transmission task, including at least one of QoS flow configuration, traffic type change (for example, Non-GBR traffic changes to GBR traffic), data network name, UE IP (Internet Protocol) address allocation, routing policy, network slice information, single network slice selection assistance information, AI authentication information, and security configuration, to determine the PDU session information corresponding to each transmission task.

[0128] S603, selecting a target UPF network element from candidate UPF network elements according to the PDU session information corresponding to each transmission task and function information of the candidate UPF network elements.

[0129] The function information of the candidate UPF network elements includes, but is not limited to, processing capacity and payload of the UPF network element.

[0130] Optionally, the optimal UPF network element is selected as the target UPF network element from the candidate UPF network elements by comprehensively considering network access management strategies corresponding to each transmission task, processing capacity and payload of each candidate UPF network element, PDU session information (for example, data network name, traffic type, PDU session and service continuity, traffic routing destination, and the like), and UE context information (for example, device information of a user, network load and network topology, and the like).

[0131] S604, sending the task execution strategy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task.

[0132] Optionally, the task execution strategy of each transmission task is sent to the target UPF network element, so that the target UPF network element routes and forwards different QoS data flows of each transmission task and provides different quality of service guarantees for the QoS flows.

[0133] It needs to be further explained that the UPF connects the base station device accessed by the UE for user service data transmission, and the base station device maps the QoS flow to the DRB (Dedicated Radio Bearer, dedicated radio bearer) and configures the SRB1 (Signaling Radio Bearer 1, signaling bearer mode 1) to transmit the corresponding control information. The control information can be transmitted to the UE through the radio resource control (Radio Resource Control, RRC) signaling, and the control information at least includes the transmission control information and the AI-related control information. Among them, the AI-related control information at least includes the model transmission strategy, the data transmission strategy, the transmission continuity indication, the transmission capacity indication, etc. The base station device can transmit the control information through the control plane SRB1 signaling bearer. In an embodiment, the control information is routed to the base station device through the AMF (Access and Mobility Management Function, access and mobility management function).

[0134] During the execution of each transmission task, the wireless access network intelligent layer receives the feedback information of the UE or the base station device in real time, and updates the available transmission resource information. The available transmission resource information can be exposed to the UE and the third-party application through the NEF (Network Exposure Function, network service presentation function) network element, so that the UE and the third-party application can request AI services from the network according to the service demand and the network state. The updated information fed back by the UE or the base station device is stored in the UDR (Unified Data Repository, unified data repository) network element after being processed by the wireless access network intelligent layer.

[0135] In this embodiment, the SMF network element receives the task execution strategy of each transmission task sent by the PCF network element; since the task execution strategy of each transmission task is obtained by the PCF network element according to the available transmission resource information and the resource requirement information of each transmission task, and each transmission task is obtained by the wireless access network intelligent layer by processing the target AI service according to the data transmission information of the data required to execute the target AI service and the model transmission information of the model required to execute the target AI service, it is equivalent to a signaling transmission mode between the wireless access network intelligent layer, the PCF network element and the SMF network element; further, according to the task execution strategy of each transmission task, the protocol data unit PDU session information corresponding to each transmission task is determined, and according to the PDU session information corresponding to each transmission task and the function information of the candidate user plane function UPF network element, the target UPF network element is selected from the candidate UPF network element; the task execution strategy of each transmission task is sent to the target UPF network element, so that the target UPF network element executes each transmission task, so that the target AI service can be efficiently executed, and the resource utilization rate in the network is improved.

[0136] Figure 7 For another embodiment of the signaling diagram of the service execution method, on the basis of the above-mentioned embodiments, the present embodiment provides an optional example of the service execution method. In combination with Figure 7 , the specific implementation process is as follows:

[0137] S701, the radio access network intelligent layer generates traffic awareness information for the target AI service in the case of receiving the service request for the target AI service sent by the user equipment through the base station equipment.

[0138] S702, the radio access network intelligent layer determines the model attribute information of the model required for executing the target AI service according to the service subscription information in the traffic awareness information.

[0139] S703, the radio access network intelligent layer determines the data attribute information of the data required for executing the target AI service according to the model attribute information.

[0140] S704, the radio access network intelligent layer determines the model transmission information according to the model attribute information, and determines the data transmission information according to the data attribute information.

[0141] Among them, the data transmission information includes data identification information, data storage location, data storage format, data processing identification, data volume, data security identification and data transmission strategy; the model transmission information includes model authentication information, model storage location, model life cycle, model deployment location and model transmission strategy.

[0142] S705, the radio access network intelligent layer processes the target AI service into at least two transmission tasks according to the data transmission information and the model transmission information, and determines the resource requirement information of each transmission task.

[0143] S706, the radio access network intelligent layer sends the available transmission resource information and the resource requirement information of each transmission task to the PCF network element.

[0144] Optionally, the radio access network intelligent layer acquires real-time network resource data and quality of service (QoS) resource allocation strategy data; determines the available transmission resource information according to the real-time network resource data and the QoS resource allocation strategy data.

[0145] S707, the PCF network element receives the available transmission resource information and the resource requirement information of each transmission task sent by the radio access network intelligent layer.

[0146] S708, the PCF network element determines the task execution strategy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0147] S709, after receiving the policy obtaining request from the SMF network element, the PCF network element transmits the task execution policy of each transmission task to the SMF network element.

[0148] Optionally, the policy obtaining request is sent by the SMF network element after detecting that the network policy changes; or the policy obtaining request is sent by the SMF network element after receiving the session update request.

[0149] Optionally, a resource configuration completion response is sent to the radio access network intelligent layer, so that the radio access network intelligent layer sends a session update request to the SMF network element.

[0150] S710, the SMF network element receives the task execution policy of each transmission task sent by the PCF network element.

[0151] S711, the SMF network element determines the PDU session information corresponding to each transmission task according to the task execution policy of each transmission task.

[0152] S712, the SMF network element selects a target UPF network element from the candidate UPF network element according to the PDU session information corresponding to each transmission task and the function information of the candidate UPF network element.

[0153] S713, the SMF network element sends the task execution policy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task.

[0154] The specific process of S701-S713 can refer to the description of the above method embodiments, and the implementation principle and technical effect are similar, which will not be repeated here.

[0155] It should be understood that although each step in the flowchart involved in each of the above embodiments is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise stated herein, the execution of these steps has no strict sequence limitation, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each of the above embodiments can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with other steps or steps or stages in other steps.

[0156] Based on the same inventive concept, the embodiments of the present application also provide a service execution apparatus for implementing the service execution method described above. The apparatus provides a solution to the implementation scheme as described in the above method, and therefore the specific limitations in one or more service execution apparatus embodiments provided below can refer to the limitations of the service execution method described above, which will not be repeated here.

[0157] In one exemplary embodiment, as shown in Figure 8 A service execution apparatus 800 is provided, configured in a wireless access network intelligent layer, comprising: a first acquisition module 810, a data determination module 820, a task splitting module 830 and a first execution module 840, wherein:

[0158] The first acquisition module 810 is configured to generate traffic awareness information for a target AI service in the case of receiving a service request for the target AI service sent by a user equipment.

[0159] The data determination module 820 is configured to determine data transmission information of data required for executing the target AI service and model transmission information of a required model according to the traffic awareness information.

[0160] The task splitting module 830 is configured to process the target AI service into at least two transmission tasks according to the data transmission information and the model transmission information, and determine resource requirement information of each transmission task.

[0161] The first execution module 840 is configured to send available transmission resource information and the resource requirement information of each transmission task to a PCF network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

[0162] In one embodiment, the traffic awareness information at least includes service subscription information of the target AI service; and the data determination module 820 is specifically configured to:

[0163] determine model attribute information of the required model for executing the target AI service according to the service subscription information; determine data attribute information of the required data for executing the target AI service according to the model attribute information; determine the model transmission information according to the model attribute information, and determine the data transmission information according to the data attribute information.

[0164] In one embodiment, the service execution apparatus 800 further comprises a resource information determination module, which is configured to:

[0165] acquire real-time network resource data and QoS resource allocation strategy data; and determine the available transmission resource information according to the real-time network resource data and the QoS resource allocation strategy data.

[0166] In one embodiment, the data transmission information includes data identification information, data storage location, data storage format, data processing identification, data volume, data security identification, and data transmission strategy; and the model transmission information includes model authentication information, model storage location, model life cycle, model deployment location, and model transmission strategy.

[0167] In one example embodiment, as shown in FIG. 9, a service execution apparatus 900 is provided, configured in a PCF network element, comprising a second acquisition module 910, a policy determination module 920, and a second execution module 930, wherein: Figure 9

[0168] The second acquisition module 910 is configured to receive available transmission resource information and resource requirement information of each transmission task sent by the radio access network intelligent layer.

[0169] The policy determination module 920 is configured to determine a task execution policy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task; wherein each task execution policy includes a QoS policy and a network access management policy; and each transmission task is obtained by the radio access network intelligent layer processing a target AI service according to data transmission information of data required for executing the target AI service and model transmission information of a required model.

[0170] The second execution module 930 is configured to, after receiving a policy acquisition request of the SMF network element, transmit the task execution policy of each transmission task to the SMF network element, so that the SMF network element executes each transmission task according to the task execution policy of each transmission task.

[0171] In one embodiment, the policy acquisition request is sent by the SMF network element after detecting a change in network policy; or the policy acquisition request is sent by the SMF network element after receiving a session update request.

[0172] In one embodiment, the service execution apparatus 900 further comprises a response sending module, which is further configured to:

[0173] send a resource configuration completion response to the radio access network intelligent layer, so that the radio access network intelligent layer sends a session update request to the SMF network element.

[0174] In one example embodiment, as shown in FIG. 10, a service execution apparatus 1000 is provided, configured in an SMF network element, comprising a third acquisition module 1010, an information determination module 1020, a function selection module 1030, and a third execution module 1040, wherein: Figure 10

[0175] ​​The third obtaining module 1010 is configured to receive the task execution strategy of each transmission task sent by the PCF network element; wherein the task execution strategy of each transmission task is obtained by the PCF network element according to the available transmission resource information and the resource requirement information of each transmission task, and each transmission task is obtained by the intelligent layer of the radio access network by processing the target AI service according to the data transmission information of the data required for executing the target AI service and the model transmission information of the model required for executing the target AI service.

[0176] The information determining module 1020 is configured to determine the PDU session information corresponding to each transmission task according to the task execution strategy of each transmission task.

[0177] The function selecting module 1030 is configured to select the target UPF network element from the candidate UPF network elements according to the PDU session information corresponding to each transmission task and the function information of the candidate UPF network elements.

[0178] The third execution module 1040 is configured to send the task execution strategy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task.

[0179] Each module in the above service execution device can be realized by software, hardware and a combination thereof in whole or in part. Each module can be embedded in or independent of the processor in the device in hardware form, or can be stored in the memory in the device in the form of software, so as to call and execute the operation corresponding to each module by the processor.

[0180] In one exemplary embodiment, a computer device is provided, which can be a device integrating the intelligent layer of the radio access network, the PCF network element or the SMF network element, and the internal structure diagram thereof can be as shown in Figure 11 The device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the device is configured to provide computing and control capabilities. The memory of the device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The input / output interface of the device is configured to exchange information between the processor and external devices. The communication interface of the device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a service execution method.

[0181] Those skilled in the art can understand that, Figure 11The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the equipment to which the scheme of the present application is applied. The specific equipment can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0182] In an exemplary embodiment, a computer device is provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of any of the service execution methods involved in the above embodiments when executing the computer program.

[0183] In an embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program implementing the steps of any of the service execution methods involved in the above embodiments when executed by a processor.

[0184] In an embodiment, a computer program product is provided, comprising a computer program, and the computer program implementing the steps of any of the service execution methods involved in the above embodiments when executed by a processor.

[0185] It should be noted that the data involved in the present application (including but not limited to data transmission information and model transmission information of the required model for executing the target AI service) are all information and data fully authorized by all parties, and the collection, use and processing of the relevant data need to comply with the relevant regulations.

[0186] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0187] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0188] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A service execution method characterized by, The method is applied to a radio access network intelligent layer, and comprises the following steps: In a case where a service request for a target artificial intelligence (AI) service is received from a user equipment, traffic awareness information for the target AI service is generated, wherein the traffic awareness information comprises a user equipment identifier, access control information, and service subscription information of the target AI service, and the service subscription information comprises an AI service identifier, an AI service type, an AI service validity time, AI service requirements, and AI service input; According to the traffic awareness information, data transmission information of data required for executing the target AI service and model transmission information of a required model are determined; At least one transmission task is formulated according to the data transmission information, and at least one transmission task is formulated according to the model transmission information, so as to map the target AI service to at least two transmission tasks, and resource requirement information of each transmission task is determined; Real-time network resource data and quality of service (QoS) resource allocation strategy data are acquired; According to the real-time network resource data and the QoS resource allocation strategy data, available transmission resource information is determined; The available transmission resource information and the resource requirement information of each transmission task are sent to a policy control function (PCF) network element, so that the PCF network element executes each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

2. The method of claim 1, wherein, The method comprises the following steps: According to the service subscription information, model attribute information of a model required for executing the target AI service is determined; According to the model attribute information, data attribute information of data required for executing the target AI service is determined; According to the model attribute information, the model transmission information is determined, and according to the data attribute information, the data transmission information is determined.

3. The method of claim 1, wherein, The data transmission information comprises data identifier information, data storage location, data storage format, data processing identifier, data volume, data security identifier, and data transmission strategy; The model transmission information comprises model authentication information, model storage location, model life cycle, model deployment location, and model transmission strategy.

4. A service execution method characterized by, The method is applied to a PCF network element, and comprises the following steps: receive available transmission resource information and resource requirement information of each transmission task sent by a radio access network intelligent layer, wherein the available transmission resource information is determined by the radio access network intelligent layer according to real-time network resource data and quality of service (QoS) resource allocation strategy data obtained by the radio access network intelligent layer, each transmission task is obtained by the radio access network intelligent layer according to traffic sensing information, data transmission information required for data transmission of data required for executing a target AI service, and model transmission information required for model transmission of a required model, at least one transmission task is formulated according to the data transmission information, at least one transmission task is formulated according to the model transmission information, and the resource requirement information of each transmission task is determined by the radio access network intelligent layer according to the data transmission information and the model transmission information, the traffic sensing information is generated by the radio access network intelligent layer when a service request for the target AI service is received from a user equipment, and the traffic sensing information includes a user equipment identifier, access control information, and service subscription information of the target AI service, the service subscription information includes an AI service identifier, an AI service type, an AI service valid time, AI service requirements, and AI service input; determine a task execution strategy for executing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task, wherein each task execution strategy includes a QoS strategy and a network access management strategy; after receiving a policy obtaining request of a session management function (SMF) network element, transmit the task execution strategy of each transmission task to the SMF network element, so that the SMF network element executes each transmission task according to the task execution strategy of each transmission task.

5. The method of claim 4, wherein, The policy obtaining request is sent by the SMF network element after detecting a change in network policy, or the policy obtaining request is sent by the SMF network element after receiving a session update request.

6. The method of claim 5, wherein, The method further includes: sending a resource configuration completion response to the radio access network intelligent layer, so that the radio access network intelligent layer sends a session update request to the SMF network element.

7. A service execution method characterized by, The method applied to the SMF network element includes: receive a task execution strategy of each transmission task sent by a PCF network element; wherein the task execution strategy of each transmission task is obtained by the PCF network element according to available transmission resource information and resource requirement information of each transmission task, the available transmission resource information is determined by a radio access network intelligent layer according to obtained real-time network resource data and quality of service (QoS) resource allocation strategy data, each transmission task is obtained by the radio access network intelligent layer according to data transmission information and model transmission information required for executing data and models of a target AI service, the data transmission information is formulated into at least one transmission task, the model transmission information is formulated into at least one transmission task, and the resource requirement information of each transmission task is determined by the radio access network intelligent layer according to the data transmission information and the model transmission information, the traffic sensing information is generated by the radio access network intelligent layer in a case where a service request for the target AI service sent by a user equipment is received, the traffic sensing information includes a user equipment identifier, access control information, and service subscription information of the target AI service, and the service subscription information includes an AI service identifier, an AI service type, an AI service validity time, AI service requirements, and AI service inputs; determine protocol data unit (PDU) session information corresponding to each transmission task according to the task execution strategy of each transmission task; select a target user plane function (UPF) network element from candidate UPF network elements according to the PDU session information corresponding to each transmission task and function information of the candidate UPF network elements; send the task execution strategy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task.

8. A service execution apparatus characterized by comprising: The device is configured in a radio access network intelligent layer, and the device includes: a first obtaining module configured to generate traffic sensing information for a target AI service in a case where a service request for the target AI service sent by a user equipment is received; wherein the traffic sensing information includes a user equipment identifier, access control information, and service subscription information of the target AI service, and the service subscription information includes an AI service identifier, an AI service type, an AI service validity time, AI service requirements, and AI service inputs; a data determining module configured to determine data transmission information and model transmission information required for executing data and models of the target AI service according to the traffic sensing information; a task splitting module configured to formulate at least one transmission task according to the data transmission information, formulate at least one transmission task according to the model transmission information, map the target AI service to at least two transmission tasks, and determine resource requirement information of each transmission task; a resource information determining module configured to obtain real-time network resource data and quality of service (QoS) resource allocation strategy data, and determine available transmission resource information according to the real-time network resource data and the QoS resource allocation strategy data. The first execution module is configured to send the available transmission resource information and the resource requirement information of each transmission task to a PCF network element, so that the PCF network element performs each transmission task according to the available transmission resource information and the resource requirement information of each transmission task.

9. A service execution apparatus characterized by comprising: The device is configured in a PCF network element, and the device comprises: The second acquisition module is configured to receive available transmission resource information and resource requirement information of each transmission task sent by a radio access network intelligent layer; wherein the available transmission resource information is determined by the radio access network intelligent layer according to acquired real-time network resource data and quality of service (QoS) resource allocation strategy data, each transmission task is obtained after the radio access network intelligent layer formulates at least one transmission task according to data transmission information and at least one transmission task according to model transmission information according to data transmission information and model transmission information determined by the radio access network intelligent layer according to traffic sensing information, the resource requirement information of each transmission task is determined by the radio access network intelligent layer according to the data transmission information and the model transmission information, the traffic sensing information is generated by the radio access network intelligent layer in the case of receiving a service request for a target AI service sent by a user equipment, the traffic sensing information comprises a user equipment identifier, access control information and service subscription information of the target AI service, the service subscription information comprises an AI service identifier, an AI service type, an AI service valid time, AI service requirements and AI service input; The policy determination module is configured to determine a task execution policy for performing each transmission task according to the available transmission resource information and the resource requirement information of each transmission task; wherein each task execution policy comprises a QoS policy and a network access management policy; The second execution module is configured to transmit the task execution policy of each transmission task to an SMF network element after receiving a policy acquisition request of the SMF network element, so that the SMF network element performs each transmission task according to the task execution policy of each transmission task.

10. A service execution apparatus characterized by comprising: The device is configured in an SMF network element, and the device comprises: The third obtaining module is configured to receive a task execution strategy of each transmission task sent by a PCF network element; wherein the task execution strategy of each transmission task is obtained by the PCF network element according to available transmission resource information and resource requirement information of each transmission task, the available transmission resource information is determined by a radio access network intelligent layer according to obtained real-time network resource data and quality of service (QoS) resource allocation strategy data, each transmission task is obtained by the radio access network intelligent layer after determining data transmission information required for executing data of a target AI service and model transmission information required for executing a model according to traffic sensing information, and is obtained by the radio access network intelligent layer after formulating at least one transmission task according to the data transmission information and formulating at least one transmission task according to the model transmission information, the resource requirement information of each transmission task is determined by the radio access network intelligent layer according to the data transmission information and the model transmission information, the traffic sensing information is generated by the radio access network intelligent layer in a case where a service request for the target AI service sent by a user equipment is received, and includes a user equipment identifier, access control information, and service subscription information of the target AI service, the service subscription information includes an AI service identifier, an AI service type, an AI service valid time, AI service requirements, and AI service input; The information determining module is configured to determine PDU session information corresponding to each transmission task according to the task execution strategy of each transmission task; The function selecting module is configured to select a target UPF network element from candidate UPF network elements according to the PDU session information corresponding to each transmission task and function information of the candidate UPF network elements. The third execution module is configured to send the task execution strategy of each transmission task to the target UPF network element, so that the target UPF network element executes each transmission task. 11.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-10 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the method in any one of claims 1 to 7.

12. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

13. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 7.

Images