Artificial intelligence service processing method and device

Abstract

The invention provides an artificial intelligence service processing method and device, and aims to solve the problem that an existing wireless access network architecture is difficult to meet 6G performance indexes. The method comprises the steps that AI service request information is acquired, the AI service request information comprises an identifier or a name of a target AI instance, and the target AI instance is used for realizing the AI service and corresponds to different model information in different application scenes; according to the AI service request information, triggering a process of applying a target model at a target network node, the target model being a model corresponding to the target AI instance in a target application scene, and the target application scene being an application scene of the AI service in the target network node; and obtaining a service result of the AI service according to an operation result of the target model in the target network node.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and more particularly to a method and device for processing artificial intelligence services. Background Art

[0002] In the process of 6G network evolution towards endogenous intelligence, the Radio Access Network (RAN) not only provides traditional communication functions, but also needs to introduce related functions such as intelligence, data and computing, so as to improve the overall performance of the wireless communication system through artificial intelligence / machine learning technology. Due to the inherent complex and changeable characteristics of wireless communication scenarios and the diversity of services, the existing wireless access network architecture is difficult to achieve multi-objective collaborative optimization of wireless communication scenarios (venues, urban areas, suburbs, etc.), wireless access network use cases (traffic management, load balancing, energy saving, etc.), and services, and thus it is difficult to meet the performance indicators of 6G. Summary of the invention

[0003] The purpose of this application is to provide a method and device for processing artificial intelligence services to solve the problem that the existing wireless access network architecture is difficult to meet the performance indicators of 6G.

[0004] In order to achieve the above-mentioned object, the present application provides a method for processing an artificial intelligence service, which is executed by a first network node, and the method includes:

[0005] Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model;

[0006] According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0007] Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

[0008] Optionally, the target network node includes the first network node.

[0009] Optionally, the method further comprises:

[0010] Deploy the target AI instance on the first network node.

[0011] Optionally, the first network node includes a first functional network layer and a second functional network layer;

[0012] Deploying the target AI instance on the first network node includes:

[0013] The first functional network layer sends a target AI instance deployment request to the second functional network layer, where the target AI instance deployment request includes link information of the target AI instance, where the link information is used to indicate storage location information of the target AI instance;

[0014] The second functional network layer obtains and runs the target AI instance according to the link information;

[0015] The second functional network layer returns a target AI instance deployment response message to the first functional network layer.

[0016] Optionally, before the first functional network layer sends the target AI instance deployment request to the second functional network layer, the method further includes:

[0017] The first functional network layer obtains the model included in the target AI instance according to the configuration file of the target AI instance;

[0018] The first functional network layer constructs the target AI instance according to the model.

[0019] Optionally, the method of the embodiment of the present application further includes:

[0020] The second functional network layer sends the performance management feedback data of the target AI instance to the first functional network layer;

[0021] The first functional network layer updates the target AI instance according to the performance management feedback data.

[0022] Optionally, the first network node includes a network node in an access network, a network node in a core network, or a terminal;

[0023] Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

[0024] Optionally, the AI ​​service includes at least one of the following:

[0025] Traffic management services;

[0026] Load balancing service;

[0027] Energy saving services.

[0028] The embodiment of the present application also provides a processing device for artificial intelligence services, including a memory, a transceiver, and a processor;

[0029] A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations:

[0030] Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model;

[0031] According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0032] Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

[0033] Optionally, the target network node includes the first network node.

[0034] Optionally, the processor further performs the following operations:

[0035] Deploy the target AI instance on the first network node.

[0036] Optionally, the first network node includes a first functional network layer and a second functional network layer;

[0037] The processor also performs the following operations:

[0038] Sending a target AI instance deployment request to the second functional network layer through the first functional network layer, wherein the target AI instance deployment request includes link information of the target AI instance, and the link information is used to indicate storage location information of the target AI instance;

[0039] According to the link information, obtain the target AI instance and run it;

[0040] A target AI instance deployment response message is returned to the first functional network layer through the second functional network layer.

[0041] Optionally, the processor further performs the following operations:

[0042] According to the configuration file of the target AI instance, obtain the model included in the target AI instance;

[0043] According to the model, the target AI instance is constructed.

[0044] Optionally, the processor further performs the following operations:

[0045] Sending the performance management feedback data of the target AI instance to the first functional network layer through the second functional network layer;

[0046] The target AI instance is updated according to the performance management feedback data.

[0047] Optionally, the first network node includes a network node in an access network, a network node in a core network, or a terminal;

[0048] Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

[0049] Optionally, the AI ​​service includes at least one of the following:

[0050] Traffic management services;

[0051] Load balancing service;

[0052] Energy saving services.

[0053] The present application also provides an artificial intelligence service processing device, including:

[0054] A first acquisition unit is used to acquire artificial intelligence (AI) service request information, wherein the AI ​​service request information includes an identifier or a name of a target AI instance, wherein the target AI instance is used to implement the AI ​​service, and wherein the target AI instance corresponds to different model information in different application scenarios; and the model information corresponding to each application scenario includes at least one model;

[0055] A processing unit, configured to trigger a process of applying a target model in a target network node according to the AI ​​service request information, wherein the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0056] The second acquisition unit is used to acquire the service result of the AI ​​service according to the operation result of the target model in the target network node.

[0057] An embodiment of the present application also provides a processor-readable storage medium, which stores a computer program, and the computer program is used to enable the processor to execute the steps of the processing method of the artificial intelligence service as described above.

[0058] The above technical solution of the present application has at least the following beneficial effects:

[0059] In an embodiment of the present application, AI service request information is obtained, and the AI ​​service request information includes the identifier or name of the target AI instance. The target AI instance is used to implement the AI ​​service, and the target AI instance corresponds to different model information in different application scenarios; according to the AI ​​service request information, a process of applying the target model in the target network node is triggered, and the target model is the model corresponding to the target AI instance in the target application scenario, and the target application scenario is the application scenario of the AI ​​service; according to the running result of the target model in the target network node, the service result of the AI ​​service is obtained. In this solution, different target models in the target AI instance are applied in the target network node for different application scenarios, so that the above-mentioned target AI instance can be applicable to various application scenarios in wireless communication scenarios (such as including business characteristics, traffic characteristics, load characteristics, etc.), so as to realize the coordinated optimization of various scenario characteristics in the above-mentioned wireless communication scenarios, and then meet the performance indicators of 6G. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A structural diagram showing a network system to which the embodiments of the present application can be applied;

[0061] Figure 2 A schematic diagram showing an intelligent framework of a wireless access network;

[0062] Figure 3 A flowchart showing a processing method of an artificial intelligence service according to an embodiment of the present application;

[0063] Figure 4 A schematic diagram showing a RAN architecture in an embodiment of the present application;

[0064] Figure 5 One of the schematic diagrams showing the relationship between consumers and various functional modules in the embodiment of the present application;

[0065] Figure 6 The second schematic diagram showing the relationship between consumers and functional modules in the embodiment of the present application;

[0066] Figure 7 A schematic diagram showing the life cycle of an AI instance in an embodiment of the present application;

[0067] Figure 8 One of the interactive schematic diagrams showing the processing method of the artificial intelligence service in the embodiment of the present application;

[0068] Fig. 9 A second interactive diagram showing a processing method of an artificial intelligence service in an embodiment of the present application;

[0069] Fig.10 express Fig. 9A schematic diagram of the configuration of the AI ​​instance in the interaction process shown;

[0070] Fig.11 A schematic diagram showing the configuration of an AI instance in a base station and terminal collaboration use case;

[0071] Fig.12 A third interactive diagram showing a processing method of an artificial intelligence service in an embodiment of the present application;

[0072] Fig.13 A fourth interactive diagram showing a processing method of an artificial intelligence service in an embodiment of the present application;

[0073] Fig.14 One of the structural block diagrams showing the processing device of the artificial intelligence service according to the embodiment of the present application;

[0074] Fig.15 A second structural block diagram showing a processing device for an artificial intelligence service according to an embodiment of the present application;

[0075] Fig.16 A schematic diagram showing a module of a processing device for an artificial intelligence service according to an embodiment of the present application. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0077] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein, for example, are implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0078] In the embodiments of the present application, the term "and / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. In the embodiments of the present application, the term "plurality" refers to two or more than two, and other quantifiers are similar.

[0079] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0080] Figure 1 A block diagram of a wireless communication system applicable to an embodiment of the present application is shown. The wireless communication system includes a terminal device 11 and a network side device (or network device) 12. Among them, the terminal device 11 may also be referred to as a terminal or a user terminal (User Equipment, UE). It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application. The network side device 12 may be a base station or a core network. It should be noted that in the embodiment of the present application, only the base station in the NR system is taken as an example, but the specific type of the base station is not limited.

[0081] In order to enable those skilled in the art to better understand the embodiments of the present application, the following description is first made.

[0082] (1) Intelligent framework of Radio Access Network (RAN);

[0083] like Figure 2 As shown in the figure, the intelligent framework of the wireless access network includes:

[0084] Data collection provides input data for model training and model inference. Data preprocessing specific to AI / ML algorithms is not implemented in data collection.

[0085] Model Training performs artificial intelligence (AI) / machine learning (ML) model training, validation, and testing, and can generate model performance metrics during testing. If necessary, the model training function can be responsible for data preprocessing based on the training data provided by data collection.

[0086] Model Deployment / Update: used to initially deploy a trained, validated, and tested AI / ML model to the model inference module, or to transmit an updated model to the model inference module.

[0087] Model Inference provides AI / ML model inference output (e.g., predictions or decisions). Model Inference can provide model performance feedback to the Model Training function when appropriate. If necessary, the Model Inference module can also be responsible for data preprocessing based on the inference data provided by the Data Collection function.

[0088] The Actor receives the output from the model reasoning function and triggers the execution of corresponding actions, which can trigger actions directed to other entities or itself.

[0089] Feedback: Obtaining information that may be needed for training data, inference data, or monitoring the performance of AI / ML models and their impact on the network by updating key performance indicators (KPIs) and performance counters.

[0090] Taking the AI / ML-based load balancing use case as an example, there are two scenarios when mapping the functional framework to the specific network architecture:

[0091] Scenario 1: AI / ML model training is performed in the Operation Administration and Maintenance (OAM) node, and inference is performed in the Next Generation Radio Access Network / 5G Radio Access Network (NG-RAN) node. The specific signaling process includes:

[0092] Assume that NG-RAN node 2 (referred to as node 2) optionally has an AI / ML model and can provide input data, such as predicted resource status, to NG-RAN node 1 (referred to as node 1);

[0093] Node 1 configures the UE to provide measurement or location information and collects the information reported by the UE; Node 1 sends the UE measurement report and other input data to OAM; Node 2 sends input data to OAM; AI / ML model training is deployed on OAM, and OAM uses the data provided by NG-RAN Nodes to perform model training; OAM deploys / updates the AI / ML model to the NG-RAN node (which can also continue model training) The UE collects the requested measurement or location information and reports it to Node 1; Node 1 receives input information for inference from the adjacent Node 2; Node 1 performs model reasoning and generates load balancing predictions or decisions; and can send model performance feedback to OAM; Node 1 can perform load balancing actions and move the UE to Node 2; Node 1 and Node 2 send feedback information to OAM.

[0094] Scenario 2: AI / ML model training and reasoning are both in NG-RAN nodes. The specific signaling process includes:

[0095] Assume that NG-RAN node 2 (referred to as node 2) optionally has an AI / ML model and can provide input (input), such as predicted resource status, to NG-RAN node 1 (referred to as node 1);

[0096] Node 1 configures UE to provide measurement or location information and collects information reported by UE; Node 1 receives input information for training from neighboring node 2; Node 1 trains A1 / ML model using collected measurement and input data; Node 1 receives UE measurement or location information; Node 1 receives input information for load balancing reasoning n from neighboring node 2; Node 1 performs model reasoning and generates load balancing predictions or decisions; Node 1 can take load balancing actions to move UE to node 2; Node 2 sends feedback information to node 1.

[0097] Considering the separation architecture on the RAN side, there are two implementation methods:

[0098] AI / ML model training is done in OAM, and reasoning is done in the New Radio NodeB-Centralized Unit (gNB-CU).

[0099] AI / ML model training and inference are both on gNB-CU.

[0100] The following is a detailed description of the processing method of the artificial intelligence service provided by the embodiment of the present application through some embodiments and their application scenarios in conjunction with the accompanying drawings.

[0101] like Figure 3As shown, an embodiment of the present application provides a method for processing an artificial intelligence service, which is performed by a first network node. Optionally, the first network node includes a network node in an access network, a network node in a core network, or a terminal. The method includes:

[0102] Step 301: Obtain artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model.

[0103] It should be noted that the model included in the model information can be an AI model or a non-AI model, that is, the model in the model information includes at least one of an AI model and a non-AI model.

[0104] The above-mentioned AI service includes at least one of a traffic management service, a load balancing service and an energy-saving service.

[0105] Of course, the above-mentioned AI services may also be other services, which are not specifically limited here.

[0106] In the embodiments of the present application, different application scenarios correspond to different scenario characteristics, which include but are not limited to at least one of the following: business characteristics (such as extended reality (eXtended Reality, XR), vehicle-to-everything information exchange (Vehicle to everything, V2X)), scene location (venue, urban area, suburbs, etc.).

[0107] The above-mentioned AI service request information may be sent by a consumer to the first network node, or may be triggered by an instance management function in the first network node according to business requirements or performance feedback.

[0108] Step 302: According to the AI ​​service request information, trigger the process of applying the target model in the target network node, where the target model is the model corresponding to the target AI instance in the target application scenario, and the target application scenario is the application scenario of the AI ​​service in the target network node.

[0109] Optionally, the target network node includes at least one network node. Optionally, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

[0110] In the embodiment of the present application, the at least one network node may include the first network node mentioned above, or may include a second network node other than the first network node mentioned above, that is, the target network node includes the first network node and / or at least one second network node.

[0111] As an optional implementation, the target network node includes the first network node, that is, the target network node may be the first network node itself. The first network node includes a network node in an access network, a network node in a core network, or a terminal.

[0112] As another optional implementation, the target network node includes the first network node and at least one second network node, or the target network node includes at least one second network node. The second network node is a network node different from the first network node, and the second network node includes a network node in an access network, a network node in a core network, or a terminal.

[0113] The above target AI instance includes a variety of model information in different application scenarios of the AI ​​service. Through the above target AI instance, unified description and flexible configuration of AI instance services in different application scenarios can be achieved.

[0114] Step 303: Obtain the service result of the AI ​​service according to the operation result of the target model in the target network node.

[0115] In an embodiment of the present application, AI service request information is obtained, and the AI ​​service request information includes the identifier or name of the target AI instance. The target AI instance is used to implement the AI ​​service, and the target AI instance corresponds to different model information in different application scenarios; according to the AI ​​service request information, a process of applying the target model in the target network node is triggered, and the target model is the model corresponding to the target AI instance in the target application scenario, and the target application scenario is the application scenario of the AI ​​service; according to the running result of the target model in the target network node, the service result of the AI ​​service is obtained. In this solution, different target models in the target AI instance are applied in the target network node for different application scenarios, so that the above-mentioned target AI instance can be applicable to various application scenarios in wireless communication scenarios (such as including business characteristics, traffic characteristics, load characteristics, etc.), so as to realize the coordinated optimization of various scenario characteristics in the above-mentioned wireless communication scenarios, and then meet the performance indicators of 6G.

[0116] As an implementation manner, applying the target model at the target network node includes a first model set applied at the first network node and a second model set applied at the second network node, and the first model set and the second model set each include at least one model.

[0117] At least some of the output parameters of at least some of the models in the second model set are input parameters of at least some of the models in the first model set, and the result of the AI ​​service is obtained based on the running results of the models in the first model set, or the result of the AI ​​service is obtained based on the running results of the first model set and the second model set.

[0118] Alternatively, at least some of the output parameters of at least some of the models in the first model set are input parameters of at least some of the models in the second model set, and the result of the AI ​​service is obtained based on the running results of the models in the second model set, or the result of the AI ​​service is obtained based on the running results of the first model set and the second model set.

[0119] In the case where the above-mentioned target network node includes the above-mentioned second network node, the second network node feeds back the running result of the target model to the first network node, and the first network node obtains the service result of the AI ​​service based on the running result of the target model in the first network node and the running result of the model in the second network node, or the first network node obtains the service result of the AI ​​service based on the running result of the model in the second network node.

[0120] In the embodiment of the present application, by applying the target model in the target AI instance in the first network node and the second network node, the function of implementing the AI ​​service through the cooperation of multiple network nodes in the wireless network can be realized. For example, the first network node and the second network node are both base stations; or the first network node is a base station and the second network node is a terminal.

[0121] Exemplarily, the first network node is base station 1 and the second network node is base station 2, the target model applied by the first network node is the first model and the second model, and the model applied by the second network node is the third model, wherein the first model, the second model or the third model respectively includes at least one model. At least part of the output parameters of the second model and at least part of the output parameters of the third model are the input parameters of the first model, and the output result of the first model is the service result of the AI ​​service, thereby realizing the function of realizing the AI ​​service through the cooperation of multiple network nodes in the wireless network.

[0122] The processing method of the artificial intelligence service in the embodiment of the present application is established on the basis of the RAN architecture supporting intelligence. Optionally, a RAN architecture such as Figure 4 shown.

[0123] Among them, OAM includes AI service management function, data management function and computing management function. The base station includes the central RAN layer, the medium RAN layer, and the low RAN layer, and the terminal includes the portal RAN layer.

[0124] The above AI Service Management Function can provide management and monitoring of functions and resources across nodes (such as between different base stations);

[0125] The above-mentioned Data Management Function can provide management and monitoring of functions and resources across nodes (such as between different base stations);

[0126] The Computing Management Function described above can provide management and monitoring of functions and resources across nodes (such as between different base stations).

[0127] The above-mentioned Central RAN Stratum can perform scenario recognition and AI instance generation, storage, execution and performance evaluation, can expose AI Instance, AI model, data or computing power to OAM, and can participate in the generation, storage, execution and performance evaluation of AI Instance across nodes (such as base stations) scheduled by OAM's AI Service Management Function. As a RAN functional module, Central RAN Stratum can be deployed inside or outside the CU with an optional strategy.

[0128] The Medium RAN Stratum layer can receive and deploy AI Instances provided by the Central RAN Stratum, and provide performance data to the Central RAN Stratum. It can realize scenario recognition and AI Instance storage and selection, and realize AI Instance performance evaluation and management. As a mid-layer function of the RAN, the Medium RAN Stratum can be deployed inside or outside the DU with an optional strategy.

[0129] The Low RAN Stratum layer can only deploy the AI ​​Instance provided by the Medium RAN Stratum layer and report performance data; as a low-layer function of the RAN, an optional strategy can be deployed inside or outside the RU. Optionally, the AI ​​Instance provided by the Medium RAN Stratum layer is transmitted from the Central RANStratum to the Medium RAN Stratum layer, that is, the Medium RAN Stratum layer transparently transmits the AI ​​Instance provided by the Central RANStratum.

[0130] The Portal RAN Stratum layer can realize scenario recognition, AI Instance storage and selection, AI Instance performance evaluation, and reporting of performance data. Portal RAN Stratum can be deployed on the terminal.

[0131] Furthermore, the above-mentioned Central RAN Stratum includes communication function (Communication Function), AI service subsystem function (AI Service Sub-system Function, AI3S), data subsystem function (Data Sub-system Function, DSS), and computing subsystem function (Computing Sub-system Function, CSS).

[0132] The AI3S can provide AI model management, AI model monitoring, AI Instance management, and AI Instance monitoring services. Taking into account different latency requirements, storage capabilities, security requirements, etc., AI3S can be tailored to only support AI model monitoring and AI Instance monitoring services, namely AI Proxy.

[0133] The DSS can provide services such as data pool, AI model library, and AI Instance library. Taking into account different latency requirements, storage capacity, security requirements, etc., the DSS can be tailored to only support data collection functions, namely data proxy.

[0134] The CSS can provide services such as AI model training, AI model reasoning, AI Instance generation, and AI Instance deployment and execution. Taking into account different latency requirements, storage capabilities, and security requirements, the CSS can be tailored to only support AI model reasoning and AI Instance deployment and execution services, namely, the Computing Proxy.

[0135] In addition, the consumers in the embodiments of the present application are consumers of the services mentioned above, and may be located inside the wireless access network node, outside the node, or outside the wireless access network node (such as the core network, terminal, and third-party server).

[0136] Please refer to Table 1 for the specific functions of the above modules.

[0137] Table 1

[0138]

[0139]

[0140] Optionally, as a first implementation, the relationship diagram between the consumer and the above-mentioned functional modules is as follows: Figure 5 As another implementation method, the relationship diagram between consumers and various functional modules is shown in Figure 6 As shown, Figure 6 The AI3S in can implement all the services in the first implementation method.

[0141] Optionally, the target network node includes the first network node, and the method further includes:

[0142] Deploy the target AI instance on the first network node.

[0143] Here, the target AI instance is deployed on the first network node so that the target model in the target AI instance can be applied based on the AI ​​service request information later.

[0144] Optionally, the first network node includes a first functional network layer and a second functional network layer;

[0145] Deploying the target AI instance on the first network node includes:

[0146] The first functional network layer sends a target AI instance deployment request to the second functional network layer, where the target AI instance deployment request includes link information of the target AI instance, where the link information is used to indicate storage location information of the target AI instance;

[0147] The second functional network layer obtains and runs the target AI instance according to the link information;

[0148] The second functional network layer returns a target AI instance deployment response message to the first functional network layer.

[0149] Optionally, the first functional network layer may be an instance management function in the AI3S of the first network node, and the second functional network layer may be an instance operation function in the computing agent of the first network node. Specifically, the instance operation function in the AI ​​agent of the first network node may send a target AI instance deployment request to the instance operation function in the computing agent. The instance operation function in the computing agent obtains the target AI instance by interacting with the instance storage function of the DSS based on the link information.

[0150] As an optional implementation, the first network node is a base station, the first functional network layer is a CU, and the second functional network layer is a DU.

[0151] Optionally, before the first functional network layer sends the target AI instance deployment request to the second functional network layer, the method further includes:

[0152] The first functional network layer obtains the model included in the target AI instance according to the configuration file of the target AI instance;

[0153] The first functional network layer constructs the target AI instance according to the model.

[0154] In the embodiment of the present application, the configuration file of the target instance is a description of the target instance, including relevant function or use case information, scenario information, model information and other resource information, etc. Specifically, the configuration file information (AIInstance Profile Information) of the AI ​​instance is shown in Table 2.

[0155] Table 2

[0156]

[0157]

[0158]

[0159]

[0160] In the embodiment of the present application, the life cycle of the AI ​​instance is as follows: Figure 7 As shown, where:

[0161] Instance Management: AI Instance (lifecycle) management function. Receive requests from consumers, drive AI Instance configuration, AI Instance access and access security management, AI Instance deployment / termination / activation / deactivation / selection / switching / rollback, and AI Instance monitoring and performance feedback and other related processes, obtain relevant process results, and feedback result response messages to consumers;

[0162] Instance Generation: AI Instance generation function. Receive configuration requests from the AI ​​Instance Management function, obtain the required model, parameters, AI and non-AI environment resources according to the AI ​​Instance Profile, and generate deployable AI Instances; publish or update the built AI Instance to the AI ​​Instance Storage function; and provide feedback on the AI ​​Instance configuration results to the AI ​​Instance Management function.

[0163] Instance Storage: AI Instance storage function. Complete AI Instance storage, as well as AI Instance registration, discovery, update, subscription, notification, unsubscription and other functions. Objects saved in instance storage are not limited to AI Instance (File / Image), but may also include configuration files (Profile), parameters, etc.;

[0164] Instance Execution: AI Instance execution function. Receive operation requests from the AI ​​Instance management function; (if necessary) obtain AI Instance from the AI ​​Instance storage function; perform deployment / termination / activation / deactivation / selection / switching / rollback of AIInstance and other related operations; feedback AI Instance operation results to the AI ​​Instance management function;

[0165] Instance Monitoring: AI Instance monitoring function. Receives monitoring instructions from the AI ​​Instance management function; performs monitoring data measurement and acquisition; returns monitoring results to the AI ​​Instance management function. Depending on the type of monitoring, it may also report monitoring results / data to the AI ​​Instance management function at regular intervals or when the relevant state changes.

[0166] In addition, in the case where the target network node includes a second network node, the process of triggering the application of the target model in the target network node according to the AI ​​service request information includes:

[0167] Send the AI ​​service request information to the second network node.

[0168] The processing flow after the second network node receives the AI ​​service request information is similar to the processing flow after the first network node receives the AI ​​service request information, and will not be repeated in the embodiments of the present application.

[0169] The deployment of the AI ​​instance in this application is described below in conjunction with Example 1.

[0170] Embodiment 1: In this embodiment, the model in the AI ​​instance is an AI model as an example for explanation, and of course, non-AI models may also be included:

[0171] like Figure 8 As shown, the deployment process includes:

[0172] Step 1: Consumer sends an AI Instance configuration request message to the Instance Management of AI3S, which carries the Instance ID or Instance name.

[0173] The AI ​​Instance request message is sent by the Consumer to the Instance Management of AI3S based on business needs or performance feedback.

[0174] In addition, the above configuration AI Instance request message can also be initiated by AI3S's Instance Management based on business needs or performance feedback, requiring the consumer to configure and implement the AI ​​Instance. Specifically, AI3S's Instance Management sends a configuration AI Instance request message to the consumer based on business needs or performance feedback, and the consumer replies with a configuration AI Instance confirmation message.

[0175] Step 2: "Instance Management" sends an AIInstance link request message to "Instance Storage" of DSS, which carries the Instance ID or Instance name;

[0176] Step 3: In this application, AI Instance is distributedly stored in each base station, so "InstanceStorage" queries the AI ​​Instance configuration Porfile and AI Instance link corresponding to the Instance ID / name first locally (the "Instance Storage" of the current base station) and then globally (for example, from the Data Management Function in OAM in the reference architecture). If the AI ​​Instance exists, its link may be located in the "Instance Storage" of the current base station or other base stations;

[0177] Step 4: "Instance Storage" returns a response message for obtaining the AI ​​Instance link to "InstanceManagement". If the AI ​​Instance exists, the response message carries the AI ​​Instance configuration profile and the AIInstance link. Otherwise, it only carries the AI ​​Instance configuration profile.

[0178] If the AI ​​Instance exists, go to step 16; otherwise, go to step 5.

[0179] Step 5: Because the AI ​​Instance has not been generated (first configuration), "Instance Management" sends an AI Instance configuration request message to "Instance Generation" of the CSS, which carries the AI ​​Instance configuration profile.

[0180] Step 6: "Instance Generation" analyzes the AI ​​Instance configuration profile to obtain the model and environment information contained in the AI ​​Instance, and then iterates steps 7 to 12 to obtain all included AI models;

[0181] Step 7: "Instance Generation" sends a request message to obtain the AI ​​model link to "Model Management" of AI3S, which carries the model ID or model name;

[0182] Step 8: "Model Management" sends a request message to obtain the AI ​​model link to "Model Storage" of DSS, which carries the model ID or model name;

[0183] Step 9: In this application, the AI ​​model is distributedly stored in each base station, so the "Model Storage" of the DSS queries the model link corresponding to the model ID / name in a local (the "Model Storage" of the current base station) and then globally (for example: from the DataManagement Function in the OAM in Example 1). If the AI ​​model exists, its model link may be located in the "Model Storage" of the current base station or other base stations;

[0184] Step 10: "Model Storage" returns a response message of obtaining the AI ​​model link to "Model Management", which carries the model link.

[0185] Step 11: "Model Management" returns a response message of obtaining the AI ​​model link to "InstanceGeneration", which carries the model link;

[0186] Step 12: "Instance Generation" interacts with the "Model Storage" of DSS (which may be located in the current base station or in other base stations) to obtain the AI ​​model corresponding to the model link. This may be multiple interactions between "Instance Generation" and "Model Storage" for block transmission, or it may be transmitted through the file transfer protocol (FTP), depending on the specific implementation;

[0187] Step 13: "Instance Generation" generates an AI Instance image according to the AI ​​Instance configuration profile;

[0188] Step 14: "Instance Generation" interacts with "Instance Storage" of DSS to publish AIInstance. Here, "Instance Generation" and "Instance Storage" may interact multiple times to transfer AI Instance images in blocks, or they may be transferred via File Transfer Protocol (FTP), depending on the specific implementation. After the release is completed, the instance link is obtained;

[0189] Step 15: "Instance Generation" returns a configuration AI Instance response message to "Instance Management", which carries the Instance ID or Instance name, and the AI ​​Instance link;

[0190] Step 16-Step 17: "Instance Management" sends a deployment AI Instance request message to "Instance Execution" of Computing Proxy (CP) via "InstanceProxy" of AIProxy (AP) (possibly multi-level), and the message carries the AI ​​Instance link;

[0191] Step 18: The CP's "Instance Execution" interacts with the DSS's "Instance Storage" (which may be located in the current base station or in other base stations) to obtain the AI ​​Instance corresponding to the instance link. Here, the CP's "Instance Execution" and the DSS's "Instance Storage" may interact multiple times to transfer the AI ​​Instance image in blocks, or it may be transferred through the File Transfer Protocol (FTP), depending on the specific implementation;

[0192] Step 19: CP's "Instance Execution" loads and starts / executes the AI ​​Instance;

[0193] Step 20-Step 21: CP's "Instance Execution" returns a deployment AI Instance response message (i.e., deployment AI instance response) to "Instance Management" via the "Instance Proxy" of AIProxy (AP) (possibly multi-level), carrying the deployment result, and if failed, the failure reason;

[0194] Step 22: "Instance Management" returns a configuration AI Instance response message to "Consumer", which carries the configuration result. If it fails, it also carries the reason for the failure.

[0195] The application of the AI ​​example in this application is described below in conjunction with Example 2 and Example 3.

[0196] Example 2: Taking the generation process of multi-cell load balancing strategy across base stations as an example, the Consumer may be located in the Central Stratum of one of the base stations involved, or in the public function part outside the base station (such as OAM), which is not limited here. Taking two base stations as an example, base station 1 is used as the aggregated traffic prediction base station, and the process is as follows: Fig. 9 As shown:

[0197] in:

[0198] The Consumer first generates the AIInstance required for the load balancing strategy to the Medium Stratum configuration of each cell. Step 1, step 2, and step 3 can be executed in parallel.

[0199] Step 1: Consumer configures the AI ​​Instance required for traffic prediction to the "InstanceExecution" of the Computing Proxy of base station 1 through the "InstanceManagement" of the AI ​​Service Sub-system Function of base station 1 and the "Instance Proxy" of the AIProxy. For the specific process, refer to Example 2.

[0200] Step 2: Consumer configures the AI ​​Instance required for traffic prediction to the "InstanceExecution" of the Computing Proxy of base station 2 through the "InstanceManagement" of the AI ​​Service Sub-system Function of base station 2 and the "Instance Proxy" of AIProxy. For the specific process, refer to Example 2.

[0201] Step 3: Consumer configures the AI ​​Instance required for total traffic prediction to the "Instance Execution" of the Computing Sub-system Function through the "Instance Management" of the AI ​​Service Sub-system Function of base station 1. For the specific process, refer to Example 2.

[0202] Each cell makes traffic predictions separately and returns the traffic prediction results to the "Instance Execution" of the Computing Sub-system Function. Steps 4-5 and steps 6-7 can be executed in parallel.

[0203] Step 4: The "Instance Execution" of the Computing Proxy of base station 1 performs cell traffic prediction (which may include multiple cells);

[0204] Step 5: The "Instance Execution" of the Computing Proxy of base station 1 returns the cell traffic prediction result (which may include multiple cells) to the "Instance Execution" of the Computing Sub-system Function. The transmission process may pass through the "Instance Proxy" of the AIProxy of base station 1 and the "Instance Management" of the AI ​​Service Sub-system Function.

[0205] Step 6: The "Instance Execution" of the Computing Proxy of base station 2 performs cell traffic prediction (which may include multiple cells);

[0206] Step 7: The "Instance Execution" of the Computing Proxy of base station 2 returns the cell traffic prediction result (which may include multiple cells) to the "Instance Execution" of the Computing Sub-system Function. The transmission process may pass through the "Instance Proxy" of the AIProxy of base station 2 and the "Instance Management" of the AI ​​ServiceSub-system Function of base station 2 and base station 1;

[0207] Step 8: The "Instance Execution" of the Computing Sub-system Function of base station 1 performs total traffic prediction based on the traffic prediction results of each cell and applies the configured total traffic prediction AI Instance;

[0208] Step 9: The "Instance Execution" of the Computing Sub-system Function of base station 1 generates a load balancing strategy for each cell based on the total traffic prediction result;

[0209] Step 10: The “Instance Execution” of the Computing Sub-system Function of base station 1 returns the load balancing strategy to the Consumer.

[0210] The configuration of AI Instance in this embodiment is as follows Fig.10 As shown,

[0211] The context of AI Instance ID = 1 is stored in the AI ​​Service Sub-system Function of base station 1 (BS1), which contains three use cases:

[0212] The Use Case 1 model is used to configure traffic prediction in BS1AI proxy. Its input parameters are the measurement results collected locally, and its output parameters are the traffic prediction results sent to BS1 AI3S.

[0213] The Use Case 2 model is used to configure traffic prediction in the Base Station 2 (BS2) AI proxy. Its input parameters are the locally collected measurement results, and its output parameters are the traffic prediction results sent to the BS1 AI3S.

[0214] The Use Case 3 model is used to configure the load balancing (traffic steering) policy in BS1. Its input parameters are the traffic prediction results of BS1AI proxy and BS2AI proxy, and its output parameters are the traffic steering policy, which is used for the traffic steering configuration policy of BS1 and BS2.

[0215] Example 3: Base station and terminal collaborative use case, Channel State Information (CSI) feedback compression and decompression is a key technology, especially in wireless communications when efficient use of bandwidth and resources is required. This technology reduces the amount of CSI feedback data by using a bilateral AI model, thereby reducing communication overhead.

[0216] In this embodiment, the base station and the terminal collaborate to complete AI-enabled CSI feedback compression and decompression through AI Instance configuration. The AI ​​Instance includes the model information of the base station and the terminal, the time points at which the base station and the terminal start CSI feedback compression and decompression, etc., so as to achieve synchronous compression and decompression between the two parties.

[0217] The process includes:

[0218] Step 1: The UE triggers the configuration of the AI ​​Instance base station (BS) side by sending a request message for deploying the AI ​​Instance for CSI feedback compression / decompression to the "Instance Management" of AI3S (the message carries the Instance ID or Instance Name for CSI feedback compression / decompression and the activation time).

[0219] Step 2: AI3S's "Instance Management" queries the AIInstance information of the bilateral model in the Data Function. If there is an AI Instance that meets the CSI feedback compression / decompression requirements in this scenario, the base station will send the AI ​​Instance information to the UE. At the agreed time, the UE and the base station perform CSI feedback compression and decompression at the same time.

[0220] Step 3: If there is no suitable AI Instance in the database, the base station first trains the bilateral model to form a new AI Instance, and then sends the AI ​​Instance information to the UE. At the agreed time, the UE and the base station perform CSI feedback compression and decompression at the same time.

[0221] Fig.11 The AI ​​configuration information shown should be stored in the Data Function. Among them:

[0222] The context of AI Instance ID = 2 is stored in BS AI Service Sub-system Function, which contains 2 Use Cases:

[0223] The Use Case 1 model is used to configure the UE AI proxy for prediction. Its input parameter is the CSI feedbackMatrix, and its output parameter is the CSI compressed code stream, which is then sent to the BS.

[0224] Use Case 2 model is used to configure the BS AI proxy for prediction. Its input parameter is the CSI compressed code stream sent by the UE side, and its output parameter is the CSI feedback matrix.

[0225] Optionally, the method of the embodiment of the present application further includes:

[0226] The second functional network layer sends the performance management feedback data of the target AI instance to the first functional network layer;

[0227] The first functional network layer updates the target AI instance according to the performance management feedback data.

[0228] The above performance management feedback data includes the accuracy of AI Instance, the performance indicator requirements of AI Instance, etc.

[0229] By updating the target AI instance, the target AI instance can better implement AI services.

[0230] The update of the AI ​​instance is described below in conjunction with the fourth embodiment and the fifth embodiment.

[0231] Embodiment 4:

[0232] like Fig.12 As shown, including:

[0233] Step 1: The "Custumer" triggers the configuration of AIInstance by sending an AI Instance deployment request message (the message carries the Instance ID or Instance Name and activation time) to the "Instance Management" of AI3S.

[0234] Step 2: The “Computing Proxy” uses the data collected by the “Data Proxy” as input to execute the AIInstance.

[0235] Step 3: “Computing Proxy” sends the execution result to “Computing Sub-systemFunction” of AI3S.

[0236] Step 4: AI3S’s “Computing Sub-system Function” takes corresponding actions based on the execution results.

[0237] Step 5: Use “Data Collection” of “Data Proxy” to collect feedback data.

[0238] Step 6: “Data Collection” of “Data Proxy” sends the performance management feedback data to “Instance Management” of AI3S for evaluation and update of AI Instance.

[0239] Embodiment five:

[0240] like Fig.13 As shown, including:

[0241] Step 1: The "Custumer" triggers the configuration of the AI ​​Instance BS1 side by sending an AI Instance deployment request message (the message carries the Instance ID or Instance Name and activation time) to the "Instance Management" of AI3S.

[0242] Step 2: The "Custumer" triggers the configuration of the AI ​​Instance UE side by sending an AI Instance deployment request message (the message carries the Instance ID or Instance Name and activation time) to the "Instance Management" of AI3S.

[0243] Step 3: The “Computing Proxy” on the BS1 side uses the data collected by the “Data Proxy” as input to execute AIInstance.

[0244] Step 4: The "Computing Proxy" on BS1 side adopts corresponding strategies based on the execution results.

[0245] Step 5: The "Computing Proxy" on the UE side uses the data collected by the "Data Proxy" as input to execute the AIInstance.

[0246] Step 6: The "Computing Proxy" on the UE side adopts corresponding strategies based on the execution results.

[0247] Step 7: The UE hands over from BS1 to BS2 and transfers the context related to the AI ​​Instance.

[0248] Step 8: The "Custumer" triggers the configuration of the AI ​​Instance BS2 side by sending an AI Instance deployment request message (the message carries the Instance ID or Instance Name and activation time) to the "Instance Management" of AI3S.

[0249] Step 9: The "Custumer" triggers the configuration of the AI ​​Instance UE side by sending an AI Instance deployment request message (the message carries the Instance ID or Instance Name and activation time) to the "Instance Management" of AI3S.

[0250] Step 10: The "Computing Proxy" on the BS2 side uses the data collected by the "Data Proxy" as input to execute AIInstance.

[0251] Step 11: The "Computing Proxy" on BS2 side adopts corresponding strategies based on the execution results.

[0252] Step 12: The "Computing Proxy" on the UE side uses the data collected by the "Data Proxy" as input to execute the AIInstance.

[0253] Step 13: The "Computing Proxy" on the UE side adopts corresponding strategies based on the execution results.

[0254] This embodiment implements the update of the AI ​​instance corresponding to the terminal movement scenario.

[0255] For scenario-oriented AI instance applications, the demodulation function includes channel estimation, equalization, and demodulation submodules. In the face of different scenarios, such as UE speed, indoors and outdoors, rural suburbs or towns, the AI ​​models corresponding to the various submodules of the demodulation function may not be the same, and some submodules may even use non-AI models. In this way, for each specific scenario, the demodulation function may need to generate different AI Instance instances and store them in the database. When facing a specific scenario, the corresponding AI Instance can be called (through Instance management request or AI proxy request, etc.) to adapt to the scenario.

[0256] The solution of the embodiment of the present application can realize a unified description and flexible configuration mechanism of AI Instance services between and within each node (base station, terminal) of the wireless access network, improve the existing RAN intelligent architecture solution, and solve the multi-objective collaborative optimization problems faced by the existing intelligent architecture solution, including wireless communication scenarios (conference venues, urban areas, suburbs, etc.), wireless access network use cases (mobility, load balancing, energy saving, V2X, etc.), and services (XR, autonomous driving, etc.), as well as the collaborative optimization problems between nodes within the wireless access network.

[0257] like Fig.14 As shown, an embodiment of the present application provides a processing device for an artificial intelligence service, which is applied to a first network node, where the first network node is a terminal. Specifically, the terminal includes a memory 1420, a transceiver 1400, and a processor 1410;

[0258] The memory 1420 is used to store computer programs; the transceiver 1400 is used to send and receive data under the control of the processor 1410; the processor 1410 is used to read the computer program in the memory 1420 and perform the following operations:

[0259] Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model;

[0260] According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0261] Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

[0262] Among them, Fig.14 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 1410 and various circuits of memory represented by memory 1420 are linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 1400 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, including wireless channels, wired channels, optical cables, and other transmission media. For different user devices, the user interface 1430 may also be an interface capable of externally and internally connecting required devices, and the connected devices include but are not limited to a keypad, a display, a speaker, a microphone, a joystick, and the like.

[0263] The processor 1410 is responsible for managing the bus architecture and general processing, and the memory 1420 can store data used by the processor 1410 when performing operations.

[0264] Optionally, the processor 1410 may be a CPU (central processing unit), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array) or a CPLD (Complex Programmable Logic Device), and the processor may also adopt a multi-core architecture.

[0265] The processor calls the computer program stored in the memory to execute any of the methods provided in the embodiments of the present application according to the obtained executable instructions. The processor and the memory can also be arranged physically separately.

[0266] Optionally, the target network node includes the first network node.

[0267] Optionally, the processor further performs the following operations:

[0268] Deploy the target AI instance on the first network node.

[0269] Optionally, the first network node includes a first functional network layer and a second functional network layer;

[0270] The processor also performs the following operations:

[0271] Sending a target AI instance deployment request to the second functional network layer through the first functional network layer, wherein the target AI instance deployment request includes link information of the target AI instance, and the link information is used to indicate storage location information of the target AI instance;

[0272] According to the link information, obtain the target AI instance and run it;

[0273] The target AI instance deployment response message is returned to the first functional network layer through the second functional network layer.

[0274] Optionally, the processor further performs the following operations:

[0275] According to the configuration file of the target AI instance, obtain the model included in the target AI instance;

[0276] According to the model, the target AI instance is constructed.

[0277] Optionally, the processor further performs the following operations:

[0278] Sending performance management feedback data of the target AI instance to the first functional network layer;

[0279] The target AI instance is updated according to the performance management feedback data.

[0280] Optionally, the first network node includes a network node in an access network, a network node in a core network, or a terminal;

[0281] Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

[0282] Optionally, the AI ​​service includes at least one of the following:

[0283] Traffic management services;

[0284] Load balancing service;

[0285] Energy saving services.

[0286] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned processing method embodiment of the artificial intelligence service applied to the first network node, and can achieve the same technical effect. The parts and beneficial effects that are the same as those in the method embodiment in this embodiment will not be specifically repeated here.

[0287] like Fig.15 As shown, the embodiment of the present application further provides a processing device for an artificial intelligence service, which is applied to a first network node, the first network node is a base station, and the base station includes a memory 1520, a transceiver 1500, and a processor 1510;

[0288] The memory 1520 is used to store computer programs; the transceiver 1500 is used to send and receive data under the control of the processor; the processor 1510 is used to read the computer program in the memory and perform the following operations:

[0289] Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model;

[0290] According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0291] Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

[0292] Among them, Fig.15 In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically one or more processors represented by processor 1510 and various circuits of memory represented by memory 1520 are linked together. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 1500 may be a plurality of components, namely, a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium, which transmission medium may include a wireless channel, a wired channel, an optical cable, and other transmission media. The processor 1510 is responsible for managing the bus architecture and general processing, and the memory 1520 may store data used by the processor 1510 when performing operations.

[0293] The processor 1510 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or a complex programmable logic device (CPLD). The processor may also adopt a multi-core architecture.

[0294] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned artificial intelligence service processing method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0295] like Fig.16 As shown, the embodiment of the present application also provides a processing device for artificial intelligence services, which is applied to a first network node, and the device includes:

[0296] The first acquisition unit 1601 is used to acquire artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model;

[0297] Processing unit 1602 is used to trigger a process of applying a target model in a target network node according to the AI ​​service request information, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node;

[0298] The second acquisition unit 1603 is used to acquire the service result of the AI ​​service according to the operation result of the target model in the target network node.

[0299] Optionally, the target network node includes the first network node.

[0300] Optionally, the device of the embodiment of the present application further includes:

[0301] A deployment unit, configured to deploy the target AI instance on the first network node.

[0302] Optionally, the first network node includes a first functional network layer and a second functional network layer;

[0303] The deployment unit is used for:

[0304] Sending a target AI instance deployment request to the second functional network layer through the first functional network layer, wherein the target AI instance deployment request includes link information of the target AI instance, and the link information is used to indicate storage location information of the target AI instance;

[0305] According to the link information, obtain the target AI instance and run it;

[0306] The target AI instance deployment response message is returned to the first functional network layer through the second functional network layer.

[0307] Optionally, the device further comprises:

[0308] A third acquisition unit is used to acquire a model included in the target AI instance according to a configuration file of the target AI instance before the first functional network layer sends a target AI instance deployment request to the second functional network layer;

[0309] A construction unit is used to construct the target AI instance according to the model.

[0310] Optionally, the device further comprises:

[0311] A sending unit, configured to send the performance management feedback data of the target AI instance to the first functional network layer;

[0312] An updating unit is used to update the target AI instance according to the performance management feedback data.

[0313] Optionally, the first network node includes a network node in an access network, a network node in a core network, or a terminal;

[0314] Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

[0315] Optionally, the AI ​​service includes at least one of the following:

[0316] Traffic management services;

[0317] Load balancing service;

[0318] Energy saving services.

[0319] It should be noted here that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned artificial intelligence service processing method embodiment, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0320] It should be noted that the division of units in the embodiments of the present application is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional unit in each embodiment of the present application may be integrated into a processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0321] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program codes.

[0322] In some embodiments of the present application, a processor-readable storage medium is further provided, wherein the processor-readable storage medium stores program instructions, wherein the program instructions are used to enable the processor to execute all steps implemented in the processing method embodiment for implementing the above-mentioned artificial intelligence service, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as those of the method embodiment will not be specifically described herein.

[0323] The terminal device involved in the embodiment of the present application may be a device that provides voice and / or data connectivity to a user, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem. In different systems, the name of the terminal device may also be different. For example, in a 5G system, the terminal device may be called a user equipment (UE). A wireless terminal device can communicate with one or more core networks (CN) via a radio access network (RAN). The wireless terminal device may be a mobile terminal device, such as a mobile phone (or a "cellular" phone) and a computer with a mobile terminal device. For example, it may be a portable, pocket-sized, handheld, computer-built-in or vehicle-mounted mobile device that exchanges language and / or data with a wireless access network. For example, personal communication service (PCS) phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs) and other devices. The wireless terminal device may also be referred to as a system, a subscriber unit, a subscriber station, a mobile station, a mobile station, a remote station, an access point, a remote terminal device, an access terminal device, a user terminal device, a user agent, and a user device, but is not limited in the embodiments of the present application.

[0324] The network device (or network side device) involved in the embodiment of the present application may be a base station, which may include multiple cells providing services for the terminal. Depending on the specific application scenario, the base station may also be called an access point, or may be a device in the access network that communicates with the wireless terminal device through one or more sectors on the air interface, or other names. The network device may be used to interchange received air frames with Internet Protocol (IP) packets, and serve as a router between the wireless terminal device and the rest of the access network, wherein the rest of the access network may include an Internet Protocol (IP) communication network. The network device may also coordinate the attribute management of the air interface. For example, the network device involved in the embodiments of the present application may be a network device (Base TransceiverStation, BTS) in the Global System for Mobile communications (Global System for Mobile communications, GSM) or Code Division Multiple Access (Code Division Multiple Access, CDMA), or a network device (NodeB) in Wide-band Code Division Multiple Access (WCDMA), or an evolutionary network device (evolutional Node B, eNB or e-NodeB) in the Long Term Evolution (Long Term Evolution, LTE) system, a 5G base station (gNB) in the 5G network architecture (next generation system), or a home evolved Node B (Home evolved Node B, HeNB), a relay node, a home base station (femto), a pico base station (pico), etc., which is not limited in the embodiments of the present application. In some network structures, the network device may include a centralized unit (CU) node and a distributed unit (DU) node, and the centralized unit and the distributed unit may also be arranged geographically separately.

[0325] Network devices and terminal devices can each use one or more antennas for multiple-input multiple-output (MIMO) transmission. MIMO transmission can be single-user MIMO (SU-MIMO) or multi-user MIMO (MU-MIMO). Depending on the form and number of antenna combinations, MIMO transmission can be 2D-MIMO, 3D-MIMO, FD-MIMO or massive-MIMO, or it can be diversity transmission, precoded transmission or beamforming transmission, etc.

[0326] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0327] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer executable instructions. These computer executable instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0328] These processor executable instructions may also be stored in a processor readable memory that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the processor readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0329] These processor-executable instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0330] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for processing an artificial intelligence service, characterized in that: Executed by a first network node, the method includes: Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model; According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node; Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

2. The method according to claim 1, characterized in that The target network node includes the first network node.

3. The method according to claim 2, characterized in that The method further comprises: Deploy the target AI instance on the first network node.

4. The method according to claim 3, characterized in that The first network node includes a first functional network layer and a second functional network layer; Deploying the target AI instance on the first network node includes: The first functional network layer sends a target AI instance deployment request to the second functional network layer, where the target AI instance deployment request includes link information of the target AI instance, where the link information is used to indicate storage location information of the target AI instance; The second functional network layer obtains and runs the target AI instance according to the link information; The second functional network layer returns a target AI instance deployment response message to the first functional network layer.

5. The method according to claim 4, characterized in that Before the first functional network layer sends a target AI instance deployment request to the second functional network layer, the method further includes: The first functional network layer obtains the model included in the target AI instance according to the configuration file of the target AI instance; The first functional network layer constructs the target AI instance according to the model.

6. The method according to claim 4, characterized in that Also includes: The second functional network layer sends the performance management feedback data of the target AI instance to the first functional network layer; The first functional network layer updates the target AI instance according to the performance management feedback data.

7. The method according to claim 2, characterized in that The first network node includes a network node in an access network, a network node in a core network, or a terminal; Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

8. The method according to claim 1, characterized in that The AI ​​service includes at least one of the following: Traffic management services; Load balancing service; Energy saving services.

9. A processing device for an artificial intelligence service, characterized in that: Including memory, transceiver, processor; A memory for storing a computer program; a transceiver for transmitting and receiving data under the control of the processor; and a processor for reading the computer program in the memory and performing the following operations: Obtaining artificial intelligence (AI) service request information, where the AI ​​service request information includes an identifier or name of a target AI instance, where the target AI instance is used to implement the AI ​​service, and where the target AI instance corresponds to different model information in different application scenarios; the model information corresponding to each application scenario includes at least one model; According to the AI ​​service request information, trigger a process of applying a target model in a target network node, where the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node; Obtain a service result of the AI ​​service according to the running result of the target model in the target network node.

10. The device according to claim 9, characterized in that The target network node comprises a first network node.

11. The device according to claim 10, characterized in that The processor also performs the following operations: Deploy the target AI instance on the first network node.

12. The device according to claim 11, characterized in that The first network node includes a first functional network layer and a second functional network layer; The processor also performs the following operations: Sending a target AI instance deployment request to the second functional network layer through the first functional network layer, wherein the target AI instance deployment request includes link information of the target AI instance, and the link information is used to indicate storage location information of the target AI instance; According to the link information, obtain the target AI instance and run it; A target AI instance deployment response message is returned to the first functional network layer through the second functional network layer.

13. The device according to claim 12, characterized in that The processor also performs the following operations: According to the configuration file of the target AI instance, obtain the model included in the target AI instance; According to the model, the target AI instance is constructed.

14. The device according to claim 12, characterized in that The processor also performs the following operations: Sending the performance management feedback data of the target AI instance to the first functional network layer through the second functional network layer; The target AI instance is updated according to the performance management feedback data.

15. The device according to claim 10, characterized in that The first network node includes a network node in an access network, a network node in a core network, or a terminal; Alternatively, the target network node includes a network node in an access network, a network node in a core network, or a terminal.

16. The device according to claim 9, characterized in that The AI ​​service includes at least one of the following: Traffic management services; Load balancing service; Energy saving services.

17. A processing device for an artificial intelligence service, characterized in that: include: A first acquisition unit is used to acquire artificial intelligence (AI) service request information, wherein the AI ​​service request information includes an identifier or a name of a target AI instance, wherein the target AI instance is used to implement the AI ​​service, and wherein the target AI instance corresponds to different model information in different application scenarios; and the model information corresponding to each application scenario includes at least one model; A processing unit, configured to trigger a process of applying a target model in a target network node according to the AI ​​service request information, wherein the target model is a model corresponding to the target AI instance in a target application scenario, and the target application scenario is an application scenario of the AI ​​service in the target network node; The second acquisition unit is used to acquire the service result of the AI ​​service according to the operation result of the target model in the target network node.

18. A processor-readable storage medium, characterized in that: The processor-readable storage medium stores a computer program, and the computer program is used to enable the processor to execute the steps of the processing method of the artificial intelligence service according to any one of claims 1 to 8.

Images